Original Research

The US Department of Veterans Affairs (VA) Home Loan Program, administered by the Veterans Benefits Administration (VBA), is a unique benefit for veterans, active-duty service members, National Guard and Reserve members, and eligible surviving spouses. Established in 1944, the program aims to help these individuals achieve homeownership by leveraging a third-party guarantee, typically from a government agency, to enhance access to credit and improve loan terms for borrowers who may not meet conventional loan qualifications.¹ Since its inception, the VA has guaranteed > 28.5 million loans, enabling millions of veterans to buy, build, repair, retain, or adapt homes for personal occupancy.² The program is designed to support veterans and eligible individuals to become homeowners, recognizing homeownership as a pathway to financial stability and community integration. VA home loans are provided by private lenders (eg, banks, mortgage companies) with a portion guaranteed by the VA, which reduces the risk for lenders and enables them to offer competitive terms, such as no down payment and lower interest rates, making homeownership more accessible to veterans.²

Eligibility criteria for the VA Home Loan Program include military service criteria such as active-duty service members with ≥ 90 continuous days of service; veterans with an honorable discharge meeting minimum service requirements; individuals who served in the National Guard/Reserve for ≥ 90 days of active service or 6 years of service with an honorable discharge; and surviving spouses of veterans who died in service or from a service-connected disability, were designated as missing in action/ prisoner of war, and the spouse is receiving Dependency and Indemnity Compensation. Financial criteria also apply: borrowers must meet lender requirements for credit and income (although VA loans are more flexible than conventional loans) and the home must be for personal occupancy rather than an investment property.³

A June 2025 PubMed literature search did not reveal any prior research on the VA Home Loan Program, although a limited number of studies tackled a wide range of issues related to federal and private home loans.^4-12 To our knowledge, there is no prior published examination of the VA Home Loan Program. Understanding VA Home Loan Program usage among Veterans Health Administration (VHA) users can inform the future direction of the program. The VHA operates the largest integrated US health care system, serving > 9 million enrolled veterans annually at 1321 facilities, including 172 medical centers and 1138 outpatient clinics, providing primary and specialized health care, and related medical and social support services for enrolled veterans, including those who are experiencing housing instability or homelessness.¹³ Specialized VHA programs for homeless veterans include housing, employment, health care, justice, and re-entryrelated services in collaboration with federal and community partners.¹⁴ Housing instability has been defined as the state of being at risk of losing housing due to challenges such as difficulties paying rent, overcrowding, frequent relocation, and a substantial proportion of income spent on housing.^15,16 Homelessness is a severe manifestation of housing instability that has been defined as the lack of stable, safe, and functioning housing.^17,18

Health care and social services, including those that address housing instability and homelessness, are major priorities for the VHA and VBA.¹⁹ The VA Home Loan Program may represent an important resource to help veterans achieve long-term housing stability through home ownership. There has been wide public concern about housing affordability and the ability of many Americans, including veterans, to achieve home ownership.²⁰ Homeownership is considered an important part of developing financial assets and achieving financial stability. Lowincome veterans, in particular, may benefit from this program as a national study found that 8.0% of low-income veterans and 13.9% of veterans with a history of homelessness have previously experienced a home foreclosure. ²¹ A greater understanding of who applies for and receives assistance from the VA Home Loan Program would inform homelessness prevention services and future planning for this program.

We conducted a quality improvement (QI) project on behalf of the VHA Homeless Programs Office and in partnership with the VBA. Our goals were to: (1) describe the annual number of applicants and recipients of the VA Home Loan Program by age group, sex, race/ethnicity, presence of any diagnosed substance use and/or mental health disorder, and history of homelessness; and (2) compare demographic, clinical, and homelessness characteristics among individuals who apply and are granted a loan through this program, individuals who apply and are denied a loan through this program, and individuals who do not apply for a loan through this program.

Methods

This project involved linked VA administrative national databases and was undertaken by the VHA Homeless Programs Office in partnership with the VBA. Specifically, VHA and VBA databases were linked together using veteran identifiers and all data were managed and analyzed on secure VA servers. The project followed VA’s Program Guide 1200.21 for nonresearch activities and institutional review board approval was waived through sponsorship by the VA Homeless Programs Office. The VHA Corporate Data Warehouse (CDW) was accessed to obtain data from the Homeless Operations Management and Evaluation System (HOMES) and other clinical data systems used by VHA clinicians and administrators that capture diagnoses, workload, and other health care data.^22,23 HOMES collects intake, progress, and outcome data on homeless veterans within its care system that enables the VA to assess the effectiveness of programs and strategically allocate resources to prevent homelessness.^24,25

A list of veterans who filed disability compensation and pension claims was obtained from the VBA Office of Performance Analysis and Integrity, including Social Security number, name, city and state, date of claim submission, grant or increase in benefits, homeless status, VA home loan approval, and homeless aid for dependent children from fiscal year (FY) 2022 through FY 2024. VBA data were linked to VHA CDW electronic health record data from veterans who sought VA health care services and HOMES data on veteran participation in homeless programs who were also experiencing homelessness. VHA data included demographic characteristics (eg, sex, age, race, marital status, combat service) at an index date (earliest visit to the VHA between October 1, 2021, and September 30, 2024); military sexual trauma; clinical characteristics within 12 months prior to the index date (VHA disability rating, substance use disorder [SUD] diagnosis, mental health disorder diagnosis, Charlson Comorbidity Index [CCI] score), and homelessness experience ≤ 5 years prior to the index date.

History of homelessness ≤ 5 years prior to the index date was determined using an operational definition of homelessness based on multiple indicators, including International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnostic code Z59.0; clinic stop codes or HOMES records indicating VA homeless programs clinical encounters; or a positive screen on an annual homelessness screener.¹⁶ US Department of Housing and Urban Development-VA Supportive Housing enrollees were excluded because they are considered to no longer be experiencing homelessness, and Veterans Justice Program enrollees were excluded because the program primarily focuses on serving criminal justice-involved veterans. The CCI predicts the risk of death ≤ 1 year by assessing the number and severity of a patient’s coexisting health conditions and is a valuable tool for understanding a patient’s overall health burden, aiding in clinical decision-making and evaluation research studies.^26-29 Diagnoses based on ICD-10-CM codes were used to determine SUDs, mental health disorders, and CCI score, using methods that have been described in other publications.³⁰

Population

The VBA cohort of veterans requesting benefits was further restricted to those who met the following eligibility criteria: (1) requested VA benefits FYs 2022 to 2024; (2) sought VHA services ≥ 1 time between FY 2022 and 2024; (3) had matching VBA/VHA records; (4) had no missing data on claim status and/ or demographic, clinical, and homelessness characteristics; and (5) had known home loan status FYs 2022 to 2024. The original VBA dataset consisted of 4,219,755 records and the original VHA dataset consisted of 7,170,199 records (Figure 1). The final linked VBA/VHA dataset after excluding 29 records with missing data on sex, 7 with missing data on age, 6 with missing data on marital status, and an additional 143,444 with unknown VBA claim status, consisted of 3,089,295 records corresponding to 2,260,851 unique veterans. Specifically, 251,796 records corresponded to veterans who had applied and received a loan, 84,751 to veterans who had applied and were nonrecipients of a loan, and 2,752,748 to veterans who did not apply for a loan.

Statistical Analysis

All statistical analyses were performed using SAS Enterprise Guide, an application that provides a point-and-click interface for data access, analysis, and management, accommodating both code-based and visual programming. ³¹ First, we relied on the final analytic sample to calculate the annual proportions of veterans who applied for and/or received a loan through the VA Home Loan Program. We also generated descriptive statistics stratified by age group, sex, race/ethnicity, SUD, mental health disorder, and homelessness, overall and within each FY. Pearson χ² and Cochran-Armitage trend tests were applied to examine differences in application and receipt of a home loan by baseline characteristics and FY, respectively. Second, we conducted bivariate and multivariable analyses to compare demographic, clinical, and homelessness characteristics between 3 groups of veterans as they pertain to the VA Home Loan Program. Veterans who applied and were nonrecipients of a loan (group 1), veterans who applied and were recipients of a loan (group 2), and veterans who did not apply for a loan (group 3). Similar analyses compared VA Home Loan Program applicants who were recipients of a home loan vs VA Home Loan Program applicants who were nonrecipients of a home loan. Multinomial and binary logistic regression models were constructed to estimate the relative risk ratio (RR) and odds ratio (OR) with 95% CIs for comparisons between these distinct groups on demographic, clinical, and homelessness characteristics. Two-sided statistical tests were evaluated at α = 0.05.

Results

Tables 1 and 2 present the number of VBA applicants, including those who applied for and received benefits through the VA Home Loan Program, by age group, sex, race/ethnicity, as well as histories of SUDs, mental health disorders, and homelessness, overall, and by FY. As shown in Figure 2, 336,547 of 3,089,295 VBA applications (10.9%) pertained to the VA Home Loan Program, with a statistically significant decline in application rates, from 12.2% in FY 2022 to 9.9% in FY 2024 (P < .001 for trend). Among 336,547 veterans who applied for the VA Home Loan Program, 251,796 (74.8%) received a home loan during FYs 2022 to 2024, ranging between 73.8% for FY 2024 and 75.5% for FY 2023 (P < .001 for trend).

Multinomial logistic regression models for demographic, clinical, and homelessness characteristics as predictors of VA Home Loan Program status are provided in Appendix 1. Based on the fully adjusted model, compared with veterans who did not apply to the VA Home Loan Program, those who applied for a home loan were less likely to be aged ≥ 50 years, unmarried, Hispanic ethnicity, mixed race, or other race, diagnosed with a SUD, or history of homelessness. Veterans with higher VA service-connected disability ratings were more frequently recipients of VA home loans, whereas those who self-identified as non-Hispanic Black and those with higher CCI scores were less frequently recipients of VA home loans. Finally, those with mental health disorders were more likely than their counterparts to be applicants (recipients or nonrecipients) of VA home loans.

Binary logistic regression models for demographic, clinical, and homelessness characteristics as predictors of receipt status among applicants to the VA Home Loan Program are provided in Appendix 2. Among applicants, those who were granted a VA home loan were less likely to be aged ≥ 50 years; have a CCI score > 0; have experienced combat service and/or military sexual trauma; be diagnosed with a SUD and/or mental health disorder; or to have a history of homelessness compared with those denied a VA home loan. Applicants granted a VA home loan were also more likely to be female, non-Hispanic White, single or never married, and/or have a VA service-connected disability ratings > 0%.

Discussion

The VA Home Loan Program is a unique benefit and resource for eligible veterans that may be increasingly important in a time of growing concern about the affordability of housing for many Americans. Research on other federally-supported home loan programs as well as private home mortgage programs has been mostly conducted in the economic realm, and studies focused on understanding these programs from a health care system perspective have been sparse.^32,33 However, there is a large body of literature documenting the importance of stable, safe, and secure housing on health and well-being.^34-37 This study did not focus on evaluating the effects of the VA Home Loan Program, because we wanted to first examine the characteristics of veterans who benefited from the program and how they differed from veterans who did not apply or did apply but had a denied application.

Our findings suggest that several thousands of veterans benefit from the VA Home Loan Program each year. For historical context, the time period examined was one of economic downturn with rising costs of living, including housing, and steady increases in homelessness as reported in the annual point-in-time count of sheltered and unsheltered people experiencing homelessness on a single night as mandated by the US Department of Housing and Urban Development.^38-40 The Sergeant First Class Heath Robinson Honoring Our Promise to Address Comprehensive Toxics (PACT) Act of 2022 expanded health care and benefits for veterans exposed to burn pits, Agent Orange, and other toxic substances, resulting in more VA disability benefit claims, including large retroactive payments.^41-43 Anecdotally, the VBA has noted that the PACT Act helped some homeless veterans with funds and stability to exit homelessness and enroll in the VA Home Loan Program.

Our analysis suggests that beneficiaries of the VA Home Loan Program were frequently aged < 50 years, female, of non-Hispanic White race, and did not have histories of psychiatric disorders or homelessness. Most of these demographic and clinical characteristics were not surprising given the composition of the veteran population, although in-depth analyses are needed to examine sex differences that may have led to more females than males benefiting from the VA Home Loan Program. In addition, it was notable that many younger and non-Hispanic Black veterans had applied. While relatively few veterans with SUDs benefited from the VA Home Loan Program, few had applied. Research is warranted into why veterans with SUDs are less likely to apply for home loans. Quite surprisingly, a sizable proportion of veterans with histories of homelessness reported they had applied to the VA Home Loan Program, although they were less likely than veterans who had not experienced homelessness to be granted a loan.

The examination of differences between veterans who did not apply, were granted, and denied a loan through the VA Home Loan Program revealed several key predictors of application outcomes in multivariable models. Specifically, veterans who applied for home loans were less likely to be aged ≥ 50 years, unmarried, of Hispanic, mixed, or other race/ethnicity, diagnosed with an SUD, or have a history of homelessness. Veterans with higher disability ratings were less frequently denied and more frequently approved, while non-Hispanic Black veterans and those with higher CCI scores were more frequently denied and less frequently approved. VBA applicants with mental health disorders were also more likely to apply for a home loan. Conversely, those granted a home loan were more likely than those denied a home loan to be female, non-Hispanic White, single/unmarried, or to have > 0% VA service-connected disability rating, but less likely to be aged ≥ 50 years, have CCI score > 0, be diagnosed with psychiatric disorders, or have a history of homelessness.

Limitations

This analysis was restricted to a subset of FY 2022 to FY 2024 linked VBA/VHA databases (ie, to veterans who had both VBA and VHA records and met prespecified eligibility criteria). Despite the large number of linked records, a small percentage of these records corresponded to veterans who were applicants or recipients of the VA Home Loan Program. Future studies should expand the time frame to examine variations in application outcomes over time and by background characteristics of veterans enrolled in VHA care who applied for VBA benefits. In addition, we relied on data and ICD-10-CM diagnostic codes from existing electronic health records and claims data to define histories of homelessness, comorbidities, SUDs, and mental health disorders. Given the time-varying nature of these conditions, the temporal sequence of events was difficult to ascertain. Third, it is worth noting that these findings can only be generalized to veterans who applied for VBA benefits and met eligibility criteria, and that these veterans may differ in terms of their demographic and clinical characteristics from those who did not apply for these benefits.

Conclusions

This study analyzed data from 251,796 individuals who applied for and received a VA home loan, 84,751 who were denied a VA home loan, and 2,752,748 veterans who did not apply for a VA home loan from FY 2022 to FY 2024. Accordingly, 11% of applications pertained to the VA Home Loan Program, and 75% of VA Home Loan Program applicants received a home loan. Distinct demographic and clinical characteristics were observed for applicants and recipients of the VA Home Loan Program, which can set the stage for future planning and evaluation of the program. Despite the broad accessibility of veterans to the VA Home Loan Program, there were differences in approval rates among applicants based on sociodemographic and clinical characteristics. Further evaluation, perhaps using qualitative methods, is needed to better understand opportunities and challenges to achieving a VA home loan, especially among underserved veteran populations. Investigation and research can guide future recommendations for any development or corrective actions that can help increase access to veterans who can benefit from the program. Future analyses are also needed to compare veterans enrolled and not enrolled in the VA Home Loan Program on health care-related outcomes.

The US Department of Veterans Affairs (VA) Home Loan Program, administered by the Veterans Benefits Administration (VBA), is a unique benefit for veterans, active-duty service members, National Guard and Reserve members, and eligible surviving spouses. Established in 1944, the program aims to help these individuals achieve homeownership by leveraging a third-party guarantee, typically from a government agency, to enhance access to credit and improve loan terms for borrowers who may not meet conventional loan qualifications.¹ Since its inception, the VA has guaranteed > 28.5 million loans, enabling millions of veterans to buy, build, repair, retain, or adapt homes for personal occupancy.² The program is designed to support veterans and eligible individuals to become homeowners, recognizing homeownership as a pathway to financial stability and community integration. VA home loans are provided by private lenders (eg, banks, mortgage companies) with a portion guaranteed by the VA, which reduces the risk for lenders and enables them to offer competitive terms, such as no down payment and lower interest rates, making homeownership more accessible to veterans.²

Eligibility criteria for the VA Home Loan Program include military service criteria such as active-duty service members with ≥ 90 continuous days of service; veterans with an honorable discharge meeting minimum service requirements; individuals who served in the National Guard/Reserve for ≥ 90 days of active service or 6 years of service with an honorable discharge; and surviving spouses of veterans who died in service or from a service-connected disability, were designated as missing in action/ prisoner of war, and the spouse is receiving Dependency and Indemnity Compensation. Financial criteria also apply: borrowers must meet lender requirements for credit and income (although VA loans are more flexible than conventional loans) and the home must be for personal occupancy rather than an investment property.³

A June 2025 PubMed literature search did not reveal any prior research on the VA Home Loan Program, although a limited number of studies tackled a wide range of issues related to federal and private home loans.^4-12 To our knowledge, there is no prior published examination of the VA Home Loan Program. Understanding VA Home Loan Program usage among Veterans Health Administration (VHA) users can inform the future direction of the program. The VHA operates the largest integrated US health care system, serving > 9 million enrolled veterans annually at 1321 facilities, including 172 medical centers and 1138 outpatient clinics, providing primary and specialized health care, and related medical and social support services for enrolled veterans, including those who are experiencing housing instability or homelessness.¹³ Specialized VHA programs for homeless veterans include housing, employment, health care, justice, and re-entryrelated services in collaboration with federal and community partners.¹⁴ Housing instability has been defined as the state of being at risk of losing housing due to challenges such as difficulties paying rent, overcrowding, frequent relocation, and a substantial proportion of income spent on housing.^15,16 Homelessness is a severe manifestation of housing instability that has been defined as the lack of stable, safe, and functioning housing.^17,18

Health care and social services, including those that address housing instability and homelessness, are major priorities for the VHA and VBA.¹⁹ The VA Home Loan Program may represent an important resource to help veterans achieve long-term housing stability through home ownership. There has been wide public concern about housing affordability and the ability of many Americans, including veterans, to achieve home ownership.²⁰ Homeownership is considered an important part of developing financial assets and achieving financial stability. Lowincome veterans, in particular, may benefit from this program as a national study found that 8.0% of low-income veterans and 13.9% of veterans with a history of homelessness have previously experienced a home foreclosure. ²¹ A greater understanding of who applies for and receives assistance from the VA Home Loan Program would inform homelessness prevention services and future planning for this program.

We conducted a quality improvement (QI) project on behalf of the VHA Homeless Programs Office and in partnership with the VBA. Our goals were to: (1) describe the annual number of applicants and recipients of the VA Home Loan Program by age group, sex, race/ethnicity, presence of any diagnosed substance use and/or mental health disorder, and history of homelessness; and (2) compare demographic, clinical, and homelessness characteristics among individuals who apply and are granted a loan through this program, individuals who apply and are denied a loan through this program, and individuals who do not apply for a loan through this program.

Methods

This project involved linked VA administrative national databases and was undertaken by the VHA Homeless Programs Office in partnership with the VBA. Specifically, VHA and VBA databases were linked together using veteran identifiers and all data were managed and analyzed on secure VA servers. The project followed VA’s Program Guide 1200.21 for nonresearch activities and institutional review board approval was waived through sponsorship by the VA Homeless Programs Office. The VHA Corporate Data Warehouse (CDW) was accessed to obtain data from the Homeless Operations Management and Evaluation System (HOMES) and other clinical data systems used by VHA clinicians and administrators that capture diagnoses, workload, and other health care data.^22,23 HOMES collects intake, progress, and outcome data on homeless veterans within its care system that enables the VA to assess the effectiveness of programs and strategically allocate resources to prevent homelessness.^24,25

A list of veterans who filed disability compensation and pension claims was obtained from the VBA Office of Performance Analysis and Integrity, including Social Security number, name, city and state, date of claim submission, grant or increase in benefits, homeless status, VA home loan approval, and homeless aid for dependent children from fiscal year (FY) 2022 through FY 2024. VBA data were linked to VHA CDW electronic health record data from veterans who sought VA health care services and HOMES data on veteran participation in homeless programs who were also experiencing homelessness. VHA data included demographic characteristics (eg, sex, age, race, marital status, combat service) at an index date (earliest visit to the VHA between October 1, 2021, and September 30, 2024); military sexual trauma; clinical characteristics within 12 months prior to the index date (VHA disability rating, substance use disorder [SUD] diagnosis, mental health disorder diagnosis, Charlson Comorbidity Index [CCI] score), and homelessness experience ≤ 5 years prior to the index date.

History of homelessness ≤ 5 years prior to the index date was determined using an operational definition of homelessness based on multiple indicators, including International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnostic code Z59.0; clinic stop codes or HOMES records indicating VA homeless programs clinical encounters; or a positive screen on an annual homelessness screener.¹⁶ US Department of Housing and Urban Development-VA Supportive Housing enrollees were excluded because they are considered to no longer be experiencing homelessness, and Veterans Justice Program enrollees were excluded because the program primarily focuses on serving criminal justice-involved veterans. The CCI predicts the risk of death ≤ 1 year by assessing the number and severity of a patient’s coexisting health conditions and is a valuable tool for understanding a patient’s overall health burden, aiding in clinical decision-making and evaluation research studies.^26-29 Diagnoses based on ICD-10-CM codes were used to determine SUDs, mental health disorders, and CCI score, using methods that have been described in other publications.³⁰

Population

The VBA cohort of veterans requesting benefits was further restricted to those who met the following eligibility criteria: (1) requested VA benefits FYs 2022 to 2024; (2) sought VHA services ≥ 1 time between FY 2022 and 2024; (3) had matching VBA/VHA records; (4) had no missing data on claim status and/ or demographic, clinical, and homelessness characteristics; and (5) had known home loan status FYs 2022 to 2024. The original VBA dataset consisted of 4,219,755 records and the original VHA dataset consisted of 7,170,199 records (Figure 1). The final linked VBA/VHA dataset after excluding 29 records with missing data on sex, 7 with missing data on age, 6 with missing data on marital status, and an additional 143,444 with unknown VBA claim status, consisted of 3,089,295 records corresponding to 2,260,851 unique veterans. Specifically, 251,796 records corresponded to veterans who had applied and received a loan, 84,751 to veterans who had applied and were nonrecipients of a loan, and 2,752,748 to veterans who did not apply for a loan.

Statistical Analysis

All statistical analyses were performed using SAS Enterprise Guide, an application that provides a point-and-click interface for data access, analysis, and management, accommodating both code-based and visual programming. ³¹ First, we relied on the final analytic sample to calculate the annual proportions of veterans who applied for and/or received a loan through the VA Home Loan Program. We also generated descriptive statistics stratified by age group, sex, race/ethnicity, SUD, mental health disorder, and homelessness, overall and within each FY. Pearson χ² and Cochran-Armitage trend tests were applied to examine differences in application and receipt of a home loan by baseline characteristics and FY, respectively. Second, we conducted bivariate and multivariable analyses to compare demographic, clinical, and homelessness characteristics between 3 groups of veterans as they pertain to the VA Home Loan Program. Veterans who applied and were nonrecipients of a loan (group 1), veterans who applied and were recipients of a loan (group 2), and veterans who did not apply for a loan (group 3). Similar analyses compared VA Home Loan Program applicants who were recipients of a home loan vs VA Home Loan Program applicants who were nonrecipients of a home loan. Multinomial and binary logistic regression models were constructed to estimate the relative risk ratio (RR) and odds ratio (OR) with 95% CIs for comparisons between these distinct groups on demographic, clinical, and homelessness characteristics. Two-sided statistical tests were evaluated at α = 0.05.

Results

Tables 1 and 2 present the number of VBA applicants, including those who applied for and received benefits through the VA Home Loan Program, by age group, sex, race/ethnicity, as well as histories of SUDs, mental health disorders, and homelessness, overall, and by FY. As shown in Figure 2, 336,547 of 3,089,295 VBA applications (10.9%) pertained to the VA Home Loan Program, with a statistically significant decline in application rates, from 12.2% in FY 2022 to 9.9% in FY 2024 (P < .001 for trend). Among 336,547 veterans who applied for the VA Home Loan Program, 251,796 (74.8%) received a home loan during FYs 2022 to 2024, ranging between 73.8% for FY 2024 and 75.5% for FY 2023 (P < .001 for trend).

Multinomial logistic regression models for demographic, clinical, and homelessness characteristics as predictors of VA Home Loan Program status are provided in Appendix 1. Based on the fully adjusted model, compared with veterans who did not apply to the VA Home Loan Program, those who applied for a home loan were less likely to be aged ≥ 50 years, unmarried, Hispanic ethnicity, mixed race, or other race, diagnosed with a SUD, or history of homelessness. Veterans with higher VA service-connected disability ratings were more frequently recipients of VA home loans, whereas those who self-identified as non-Hispanic Black and those with higher CCI scores were less frequently recipients of VA home loans. Finally, those with mental health disorders were more likely than their counterparts to be applicants (recipients or nonrecipients) of VA home loans.

Binary logistic regression models for demographic, clinical, and homelessness characteristics as predictors of receipt status among applicants to the VA Home Loan Program are provided in Appendix 2. Among applicants, those who were granted a VA home loan were less likely to be aged ≥ 50 years; have a CCI score > 0; have experienced combat service and/or military sexual trauma; be diagnosed with a SUD and/or mental health disorder; or to have a history of homelessness compared with those denied a VA home loan. Applicants granted a VA home loan were also more likely to be female, non-Hispanic White, single or never married, and/or have a VA service-connected disability ratings > 0%.

Discussion

The VA Home Loan Program is a unique benefit and resource for eligible veterans that may be increasingly important in a time of growing concern about the affordability of housing for many Americans. Research on other federally-supported home loan programs as well as private home mortgage programs has been mostly conducted in the economic realm, and studies focused on understanding these programs from a health care system perspective have been sparse.^32,33 However, there is a large body of literature documenting the importance of stable, safe, and secure housing on health and well-being.^34-37 This study did not focus on evaluating the effects of the VA Home Loan Program, because we wanted to first examine the characteristics of veterans who benefited from the program and how they differed from veterans who did not apply or did apply but had a denied application.

Our findings suggest that several thousands of veterans benefit from the VA Home Loan Program each year. For historical context, the time period examined was one of economic downturn with rising costs of living, including housing, and steady increases in homelessness as reported in the annual point-in-time count of sheltered and unsheltered people experiencing homelessness on a single night as mandated by the US Department of Housing and Urban Development.^38-40 The Sergeant First Class Heath Robinson Honoring Our Promise to Address Comprehensive Toxics (PACT) Act of 2022 expanded health care and benefits for veterans exposed to burn pits, Agent Orange, and other toxic substances, resulting in more VA disability benefit claims, including large retroactive payments.^41-43 Anecdotally, the VBA has noted that the PACT Act helped some homeless veterans with funds and stability to exit homelessness and enroll in the VA Home Loan Program.

Our analysis suggests that beneficiaries of the VA Home Loan Program were frequently aged < 50 years, female, of non-Hispanic White race, and did not have histories of psychiatric disorders or homelessness. Most of these demographic and clinical characteristics were not surprising given the composition of the veteran population, although in-depth analyses are needed to examine sex differences that may have led to more females than males benefiting from the VA Home Loan Program. In addition, it was notable that many younger and non-Hispanic Black veterans had applied. While relatively few veterans with SUDs benefited from the VA Home Loan Program, few had applied. Research is warranted into why veterans with SUDs are less likely to apply for home loans. Quite surprisingly, a sizable proportion of veterans with histories of homelessness reported they had applied to the VA Home Loan Program, although they were less likely than veterans who had not experienced homelessness to be granted a loan.

The examination of differences between veterans who did not apply, were granted, and denied a loan through the VA Home Loan Program revealed several key predictors of application outcomes in multivariable models. Specifically, veterans who applied for home loans were less likely to be aged ≥ 50 years, unmarried, of Hispanic, mixed, or other race/ethnicity, diagnosed with an SUD, or have a history of homelessness. Veterans with higher disability ratings were less frequently denied and more frequently approved, while non-Hispanic Black veterans and those with higher CCI scores were more frequently denied and less frequently approved. VBA applicants with mental health disorders were also more likely to apply for a home loan. Conversely, those granted a home loan were more likely than those denied a home loan to be female, non-Hispanic White, single/unmarried, or to have > 0% VA service-connected disability rating, but less likely to be aged ≥ 50 years, have CCI score > 0, be diagnosed with psychiatric disorders, or have a history of homelessness.

Limitations

This analysis was restricted to a subset of FY 2022 to FY 2024 linked VBA/VHA databases (ie, to veterans who had both VBA and VHA records and met prespecified eligibility criteria). Despite the large number of linked records, a small percentage of these records corresponded to veterans who were applicants or recipients of the VA Home Loan Program. Future studies should expand the time frame to examine variations in application outcomes over time and by background characteristics of veterans enrolled in VHA care who applied for VBA benefits. In addition, we relied on data and ICD-10-CM diagnostic codes from existing electronic health records and claims data to define histories of homelessness, comorbidities, SUDs, and mental health disorders. Given the time-varying nature of these conditions, the temporal sequence of events was difficult to ascertain. Third, it is worth noting that these findings can only be generalized to veterans who applied for VBA benefits and met eligibility criteria, and that these veterans may differ in terms of their demographic and clinical characteristics from those who did not apply for these benefits.

Conclusions

This study analyzed data from 251,796 individuals who applied for and received a VA home loan, 84,751 who were denied a VA home loan, and 2,752,748 veterans who did not apply for a VA home loan from FY 2022 to FY 2024. Accordingly, 11% of applications pertained to the VA Home Loan Program, and 75% of VA Home Loan Program applicants received a home loan. Distinct demographic and clinical characteristics were observed for applicants and recipients of the VA Home Loan Program, which can set the stage for future planning and evaluation of the program. Despite the broad accessibility of veterans to the VA Home Loan Program, there were differences in approval rates among applicants based on sociodemographic and clinical characteristics. Further evaluation, perhaps using qualitative methods, is needed to better understand opportunities and challenges to achieving a VA home loan, especially among underserved veteran populations. Investigation and research can guide future recommendations for any development or corrective actions that can help increase access to veterans who can benefit from the program. Future analyses are also needed to compare veterans enrolled and not enrolled in the VA Home Loan Program on health care-related outcomes.

The US Department of Veterans Affairs (VA) Home Loan Program, administered by the Veterans Benefits Administration (VBA), is a unique benefit for veterans, active-duty service members, National Guard and Reserve members, and eligible surviving spouses. Established in 1944, the program aims to help these individuals achieve homeownership by leveraging a third-party guarantee, typically from a government agency, to enhance access to credit and improve loan terms for borrowers who may not meet conventional loan qualifications.¹ Since its inception, the VA has guaranteed > 28.5 million loans, enabling millions of veterans to buy, build, repair, retain, or adapt homes for personal occupancy.² The program is designed to support veterans and eligible individuals to become homeowners, recognizing homeownership as a pathway to financial stability and community integration. VA home loans are provided by private lenders (eg, banks, mortgage companies) with a portion guaranteed by the VA, which reduces the risk for lenders and enables them to offer competitive terms, such as no down payment and lower interest rates, making homeownership more accessible to veterans.²

Eligibility criteria for the VA Home Loan Program include military service criteria such as active-duty service members with ≥ 90 continuous days of service; veterans with an honorable discharge meeting minimum service requirements; individuals who served in the National Guard/Reserve for ≥ 90 days of active service or 6 years of service with an honorable discharge; and surviving spouses of veterans who died in service or from a service-connected disability, were designated as missing in action/ prisoner of war, and the spouse is receiving Dependency and Indemnity Compensation. Financial criteria also apply: borrowers must meet lender requirements for credit and income (although VA loans are more flexible than conventional loans) and the home must be for personal occupancy rather than an investment property.³

A June 2025 PubMed literature search did not reveal any prior research on the VA Home Loan Program, although a limited number of studies tackled a wide range of issues related to federal and private home loans.^4-12 To our knowledge, there is no prior published examination of the VA Home Loan Program. Understanding VA Home Loan Program usage among Veterans Health Administration (VHA) users can inform the future direction of the program. The VHA operates the largest integrated US health care system, serving > 9 million enrolled veterans annually at 1321 facilities, including 172 medical centers and 1138 outpatient clinics, providing primary and specialized health care, and related medical and social support services for enrolled veterans, including those who are experiencing housing instability or homelessness.¹³ Specialized VHA programs for homeless veterans include housing, employment, health care, justice, and re-entryrelated services in collaboration with federal and community partners.¹⁴ Housing instability has been defined as the state of being at risk of losing housing due to challenges such as difficulties paying rent, overcrowding, frequent relocation, and a substantial proportion of income spent on housing.^15,16 Homelessness is a severe manifestation of housing instability that has been defined as the lack of stable, safe, and functioning housing.^17,18

Health care and social services, including those that address housing instability and homelessness, are major priorities for the VHA and VBA.¹⁹ The VA Home Loan Program may represent an important resource to help veterans achieve long-term housing stability through home ownership. There has been wide public concern about housing affordability and the ability of many Americans, including veterans, to achieve home ownership.²⁰ Homeownership is considered an important part of developing financial assets and achieving financial stability. Lowincome veterans, in particular, may benefit from this program as a national study found that 8.0% of low-income veterans and 13.9% of veterans with a history of homelessness have previously experienced a home foreclosure. ²¹ A greater understanding of who applies for and receives assistance from the VA Home Loan Program would inform homelessness prevention services and future planning for this program.

We conducted a quality improvement (QI) project on behalf of the VHA Homeless Programs Office and in partnership with the VBA. Our goals were to: (1) describe the annual number of applicants and recipients of the VA Home Loan Program by age group, sex, race/ethnicity, presence of any diagnosed substance use and/or mental health disorder, and history of homelessness; and (2) compare demographic, clinical, and homelessness characteristics among individuals who apply and are granted a loan through this program, individuals who apply and are denied a loan through this program, and individuals who do not apply for a loan through this program.

Methods

This project involved linked VA administrative national databases and was undertaken by the VHA Homeless Programs Office in partnership with the VBA. Specifically, VHA and VBA databases were linked together using veteran identifiers and all data were managed and analyzed on secure VA servers. The project followed VA’s Program Guide 1200.21 for nonresearch activities and institutional review board approval was waived through sponsorship by the VA Homeless Programs Office. The VHA Corporate Data Warehouse (CDW) was accessed to obtain data from the Homeless Operations Management and Evaluation System (HOMES) and other clinical data systems used by VHA clinicians and administrators that capture diagnoses, workload, and other health care data.^22,23 HOMES collects intake, progress, and outcome data on homeless veterans within its care system that enables the VA to assess the effectiveness of programs and strategically allocate resources to prevent homelessness.^24,25

A list of veterans who filed disability compensation and pension claims was obtained from the VBA Office of Performance Analysis and Integrity, including Social Security number, name, city and state, date of claim submission, grant or increase in benefits, homeless status, VA home loan approval, and homeless aid for dependent children from fiscal year (FY) 2022 through FY 2024. VBA data were linked to VHA CDW electronic health record data from veterans who sought VA health care services and HOMES data on veteran participation in homeless programs who were also experiencing homelessness. VHA data included demographic characteristics (eg, sex, age, race, marital status, combat service) at an index date (earliest visit to the VHA between October 1, 2021, and September 30, 2024); military sexual trauma; clinical characteristics within 12 months prior to the index date (VHA disability rating, substance use disorder [SUD] diagnosis, mental health disorder diagnosis, Charlson Comorbidity Index [CCI] score), and homelessness experience ≤ 5 years prior to the index date.

History of homelessness ≤ 5 years prior to the index date was determined using an operational definition of homelessness based on multiple indicators, including International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnostic code Z59.0; clinic stop codes or HOMES records indicating VA homeless programs clinical encounters; or a positive screen on an annual homelessness screener.¹⁶ US Department of Housing and Urban Development-VA Supportive Housing enrollees were excluded because they are considered to no longer be experiencing homelessness, and Veterans Justice Program enrollees were excluded because the program primarily focuses on serving criminal justice-involved veterans. The CCI predicts the risk of death ≤ 1 year by assessing the number and severity of a patient’s coexisting health conditions and is a valuable tool for understanding a patient’s overall health burden, aiding in clinical decision-making and evaluation research studies.^26-29 Diagnoses based on ICD-10-CM codes were used to determine SUDs, mental health disorders, and CCI score, using methods that have been described in other publications.³⁰

Population

The VBA cohort of veterans requesting benefits was further restricted to those who met the following eligibility criteria: (1) requested VA benefits FYs 2022 to 2024; (2) sought VHA services ≥ 1 time between FY 2022 and 2024; (3) had matching VBA/VHA records; (4) had no missing data on claim status and/ or demographic, clinical, and homelessness characteristics; and (5) had known home loan status FYs 2022 to 2024. The original VBA dataset consisted of 4,219,755 records and the original VHA dataset consisted of 7,170,199 records (Figure 1). The final linked VBA/VHA dataset after excluding 29 records with missing data on sex, 7 with missing data on age, 6 with missing data on marital status, and an additional 143,444 with unknown VBA claim status, consisted of 3,089,295 records corresponding to 2,260,851 unique veterans. Specifically, 251,796 records corresponded to veterans who had applied and received a loan, 84,751 to veterans who had applied and were nonrecipients of a loan, and 2,752,748 to veterans who did not apply for a loan.

Statistical Analysis

All statistical analyses were performed using SAS Enterprise Guide, an application that provides a point-and-click interface for data access, analysis, and management, accommodating both code-based and visual programming. ³¹ First, we relied on the final analytic sample to calculate the annual proportions of veterans who applied for and/or received a loan through the VA Home Loan Program. We also generated descriptive statistics stratified by age group, sex, race/ethnicity, SUD, mental health disorder, and homelessness, overall and within each FY. Pearson χ² and Cochran-Armitage trend tests were applied to examine differences in application and receipt of a home loan by baseline characteristics and FY, respectively. Second, we conducted bivariate and multivariable analyses to compare demographic, clinical, and homelessness characteristics between 3 groups of veterans as they pertain to the VA Home Loan Program. Veterans who applied and were nonrecipients of a loan (group 1), veterans who applied and were recipients of a loan (group 2), and veterans who did not apply for a loan (group 3). Similar analyses compared VA Home Loan Program applicants who were recipients of a home loan vs VA Home Loan Program applicants who were nonrecipients of a home loan. Multinomial and binary logistic regression models were constructed to estimate the relative risk ratio (RR) and odds ratio (OR) with 95% CIs for comparisons between these distinct groups on demographic, clinical, and homelessness characteristics. Two-sided statistical tests were evaluated at α = 0.05.

Results

Tables 1 and 2 present the number of VBA applicants, including those who applied for and received benefits through the VA Home Loan Program, by age group, sex, race/ethnicity, as well as histories of SUDs, mental health disorders, and homelessness, overall, and by FY. As shown in Figure 2, 336,547 of 3,089,295 VBA applications (10.9%) pertained to the VA Home Loan Program, with a statistically significant decline in application rates, from 12.2% in FY 2022 to 9.9% in FY 2024 (P < .001 for trend). Among 336,547 veterans who applied for the VA Home Loan Program, 251,796 (74.8%) received a home loan during FYs 2022 to 2024, ranging between 73.8% for FY 2024 and 75.5% for FY 2023 (P < .001 for trend).

Multinomial logistic regression models for demographic, clinical, and homelessness characteristics as predictors of VA Home Loan Program status are provided in Appendix 1. Based on the fully adjusted model, compared with veterans who did not apply to the VA Home Loan Program, those who applied for a home loan were less likely to be aged ≥ 50 years, unmarried, Hispanic ethnicity, mixed race, or other race, diagnosed with a SUD, or history of homelessness. Veterans with higher VA service-connected disability ratings were more frequently recipients of VA home loans, whereas those who self-identified as non-Hispanic Black and those with higher CCI scores were less frequently recipients of VA home loans. Finally, those with mental health disorders were more likely than their counterparts to be applicants (recipients or nonrecipients) of VA home loans.

Binary logistic regression models for demographic, clinical, and homelessness characteristics as predictors of receipt status among applicants to the VA Home Loan Program are provided in Appendix 2. Among applicants, those who were granted a VA home loan were less likely to be aged ≥ 50 years; have a CCI score > 0; have experienced combat service and/or military sexual trauma; be diagnosed with a SUD and/or mental health disorder; or to have a history of homelessness compared with those denied a VA home loan. Applicants granted a VA home loan were also more likely to be female, non-Hispanic White, single or never married, and/or have a VA service-connected disability ratings > 0%.

Discussion

The VA Home Loan Program is a unique benefit and resource for eligible veterans that may be increasingly important in a time of growing concern about the affordability of housing for many Americans. Research on other federally-supported home loan programs as well as private home mortgage programs has been mostly conducted in the economic realm, and studies focused on understanding these programs from a health care system perspective have been sparse.^32,33 However, there is a large body of literature documenting the importance of stable, safe, and secure housing on health and well-being.^34-37 This study did not focus on evaluating the effects of the VA Home Loan Program, because we wanted to first examine the characteristics of veterans who benefited from the program and how they differed from veterans who did not apply or did apply but had a denied application.

Our findings suggest that several thousands of veterans benefit from the VA Home Loan Program each year. For historical context, the time period examined was one of economic downturn with rising costs of living, including housing, and steady increases in homelessness as reported in the annual point-in-time count of sheltered and unsheltered people experiencing homelessness on a single night as mandated by the US Department of Housing and Urban Development.^38-40 The Sergeant First Class Heath Robinson Honoring Our Promise to Address Comprehensive Toxics (PACT) Act of 2022 expanded health care and benefits for veterans exposed to burn pits, Agent Orange, and other toxic substances, resulting in more VA disability benefit claims, including large retroactive payments.^41-43 Anecdotally, the VBA has noted that the PACT Act helped some homeless veterans with funds and stability to exit homelessness and enroll in the VA Home Loan Program.

Our analysis suggests that beneficiaries of the VA Home Loan Program were frequently aged < 50 years, female, of non-Hispanic White race, and did not have histories of psychiatric disorders or homelessness. Most of these demographic and clinical characteristics were not surprising given the composition of the veteran population, although in-depth analyses are needed to examine sex differences that may have led to more females than males benefiting from the VA Home Loan Program. In addition, it was notable that many younger and non-Hispanic Black veterans had applied. While relatively few veterans with SUDs benefited from the VA Home Loan Program, few had applied. Research is warranted into why veterans with SUDs are less likely to apply for home loans. Quite surprisingly, a sizable proportion of veterans with histories of homelessness reported they had applied to the VA Home Loan Program, although they were less likely than veterans who had not experienced homelessness to be granted a loan.

The examination of differences between veterans who did not apply, were granted, and denied a loan through the VA Home Loan Program revealed several key predictors of application outcomes in multivariable models. Specifically, veterans who applied for home loans were less likely to be aged ≥ 50 years, unmarried, of Hispanic, mixed, or other race/ethnicity, diagnosed with an SUD, or have a history of homelessness. Veterans with higher disability ratings were less frequently denied and more frequently approved, while non-Hispanic Black veterans and those with higher CCI scores were more frequently denied and less frequently approved. VBA applicants with mental health disorders were also more likely to apply for a home loan. Conversely, those granted a home loan were more likely than those denied a home loan to be female, non-Hispanic White, single/unmarried, or to have > 0% VA service-connected disability rating, but less likely to be aged ≥ 50 years, have CCI score > 0, be diagnosed with psychiatric disorders, or have a history of homelessness.

Limitations

This analysis was restricted to a subset of FY 2022 to FY 2024 linked VBA/VHA databases (ie, to veterans who had both VBA and VHA records and met prespecified eligibility criteria). Despite the large number of linked records, a small percentage of these records corresponded to veterans who were applicants or recipients of the VA Home Loan Program. Future studies should expand the time frame to examine variations in application outcomes over time and by background characteristics of veterans enrolled in VHA care who applied for VBA benefits. In addition, we relied on data and ICD-10-CM diagnostic codes from existing electronic health records and claims data to define histories of homelessness, comorbidities, SUDs, and mental health disorders. Given the time-varying nature of these conditions, the temporal sequence of events was difficult to ascertain. Third, it is worth noting that these findings can only be generalized to veterans who applied for VBA benefits and met eligibility criteria, and that these veterans may differ in terms of their demographic and clinical characteristics from those who did not apply for these benefits.

Conclusions

This study analyzed data from 251,796 individuals who applied for and received a VA home loan, 84,751 who were denied a VA home loan, and 2,752,748 veterans who did not apply for a VA home loan from FY 2022 to FY 2024. Accordingly, 11% of applications pertained to the VA Home Loan Program, and 75% of VA Home Loan Program applicants received a home loan. Distinct demographic and clinical characteristics were observed for applicants and recipients of the VA Home Loan Program, which can set the stage for future planning and evaluation of the program. Despite the broad accessibility of veterans to the VA Home Loan Program, there were differences in approval rates among applicants based on sociodemographic and clinical characteristics. Further evaluation, perhaps using qualitative methods, is needed to better understand opportunities and challenges to achieving a VA home loan, especially among underserved veteran populations. Investigation and research can guide future recommendations for any development or corrective actions that can help increase access to veterans who can benefit from the program. Future analyses are also needed to compare veterans enrolled and not enrolled in the VA Home Loan Program on health care-related outcomes.

Traumatic brain injury (TBI) is the signature injury of post-9/11 military operations, impacting > 441,000 combat veterans from 2001 to 2021 and 87% diagnosed with mild TBI (mTBI).^1,2 The most common cause of mTBI during these operations was blast exposures stemming from improvised explosive devices, rocket-propelled grenades, or land mines. mTBI was once thought to be self-limiting, lasting hours or days postinjury, but is now recognized as a complex focal and diffuse injury causing a cascade of molecular and biochemical responses with significant physiologic effects lasting for a longer duration. A significant number of combat veterans with mTBI (23%-48%) experience long-standing postconcussive symptoms (PCSs) for many years postinjury.^3-5

Developing and implementing strategies to reduce persistent symptoms associated with mTBI is of critical importance. Veterans diagnosed with mTBI and experiencing PCSs present ongoing treatment challenges to the health care system due to limited or suboptimal treatment options.^6,7 According to the 2021 US Department of Veterans Affairs (VA) and US Department of Defense (DoD) clinical guidelines for postacute mTBI, treatment for PCSs should be symptom focused. ^8,9 For instance, veterans with migraine headaches associated with mTBIs are often treated with abortive agents (eg, triptans) and preventive medications (eg, anticonvulsants and tricyclics).¹⁰ Cognitive dysfunction and insomnia are treated with cognitive rehabilitation programs, cognitive behaviorial therapy, occupational therapy, and medications (eg, hypnotics for insomnia).^11,12 The 2021 VA/DoD guidelines note that veteran and military focus groups described greater success with nonpharmacologic treatments than with pharmacologic treatments.⁸ The VA launched an enterprise-wide Whole Health Service program with the requirement that complementary and integrative health approaches must be available to veterans.¹³ As a nonpharmacologic, integrative, and noninvasive modality, neurofeedback (NFB) supports the VA Whole Health initiative and veterans’ preferences for integrative treatments.¹⁴

Neurofeedback

Rather than a symptom management approach, Defina et al described the possibilities of brain repair in TBI by treatments to enhance neuroplasticity, thereby establishing a more normalized or stable brain environment and enabling the brain to reorganize itself and function more normally.¹⁵ NFB has been shown to influence neuroplasticity,¹⁶ as evident in microstructural changes in white and gray matter¹⁷ and its ability to contribute to functional rehabilitation by restoring connectivity in specific areas of the brain that may have been impaired.¹⁸ The benefits of neuroenhancement strategies include potentially reduced pain for patients with mTBI and improved quality of life (QOL).¹⁹

NFB assists individuals by helping them become more aware of and self-regulate their physiology.^20,21 Because there are several types of NFB (eg, quantitative electroencephalography, Z-scored, α-θ) that differ in terms of equipment, mechanism of action, focus, and patient and clinician procedures, it is important to note that this study used a novel technologically advanced form of NFB, referred to as infra-low frequency (ILF) NFB. It works by reflecting a person’s brain wave activity via conventional electroencephalography back to the person through the visual cortex, thus providing relevant information to which the brain responds to improve core state regulation.²²

In 2006, ILF NFB developers sought to extend NFB capability into the slow cortical potential domain (< 0.1 Hz) and then gradually extended to lower frequencies on the basis of favorable clinical responses.^22,23 In 2017, the technology reached an ILF capacity that appeared to be helpful for several clinical issues. These developments depended on instrumentation capable of low noise signal detection down to the lowest frequency of interest. Instrumentation was developed for the purpose (eg, Bee Medic Cygnet NFB).

Although mTBI has been a clinical focus in NFB since the 1980s, there are few published studies demonstrating the efficacy of ILF NFB relating to the PCSs of interest in this study, and 2 suggested ILF NFB positively affected change in PCS severity.^24,25 Other studies found that ILF NFB decreased incidence of migraines and tension type headaches.^26,27 However, the findings of these studies had limited generalizability due to methodologic limitations, such as selection bias and small sample sizes.^24-27 Of importance to this article, there are also several publications on the efficacy of ILF NFB in clinical settings.^28-33

This article presents the second analysis of data from veterans who completed ILF NFB intervention and control group procedures during a 5-year randomized controlled trial (RCT). The RCT included veterans who experienced an mTBI while participating in post-9/11 military operations to evaluate the impact of ILF NFB on chronic PCSs, including headache, insomnia, and attention dysfunction. Initial results of this trial demonstrated significant differences between the intervention and control groups with strong effect sizes on all outcome measures at the end of treatment.³⁴

Methods

Participants included male and nonpregnant female veterans with a diagnosed mTBI during post-9/11 military operations; aged 18 to 65 years; reports of persistent (ie, > 3 months in duration) headaches, insomnia, and attention difficulties; and able to read and write English, comprehend what is read, and follow directions. mTBI diagnosis was verified for each veteran via the electronic health record. Patients were excluded if they had a severe TBI diagnosis or impaired decision-making capacity; were unable to comply with study visit schedule; or endorsed active suicidal intent on the Columbia-Suicide Severity Rating Scale.³⁵

Recruitment efforts included: (1) letters sent to eligible veterans with mTBI who were identified by clinical informatics data after waiver of Health Insurance Portability and Accountability Act was obtained; veterans could contact the research team directly or the research team would call the veteran 2 weeks after the letter was sent; (2) veterans could be referred by a clinician; and (3) veterans could self-refer based on flyers and other study marketing materials.

The study was conducted from 2019 to 2024 at Spark M. Matsunaga VA Medical Center, in Honolulu, Hawaii. Four private research spaces in compliance with human research standards were used for consent, treatment, and assessment.

Consenting Procedure and Randomization

The privacy rights of potential participants were observed, and interested veterans who met the eligibility criteria underwent an informed consent procedure and were administered the Columbia-Suicide Severity Rating Scale.³⁵ Those veterans not indicating active suicidal intent were randomized into the intervention or control group. Once randomized, the participant was enrolled and scheduled for baseline assessment.

All procedures of this study were performed in adherence with relevant laws and institutional guidelines. The study was reviewed and approved by the VA Pacific Islands Health Care System Institutional Review Board (#2019-06-JC/Promise 0003).

Outcome Measures

The outcome measures were administered at baseline, midpoint (3-7 weeks), end of treatment (6-12 weeks), and at a 2-month follow-up appointment with the research assistant or project coordinator.

The primary outcome measures include the Headache Impact Test (HIT-6), TBIQOL Headache Pain item short form, Insomnia Severity Index (ISI), Quality of Life in Neurological Disorders (Neuro-QOL) Sleep Disturbance short form, and attention measure: QIKtest Continuous Performance Test (QIKtest) (Table 1).^36-44

Secondary outcome measures included QOL After Brain Injury (QOLIBRI), Neuro- QOL Satisfaction With Roles/Activities short form (Neuro-QOL Satisfaction), Neuro-QOL Ability to Participate in Roles/Activities short form (Neuro-QOL Participate), Depression Anxiety Stress Scales (DASS-21), Patient Health Questionnaire-9 (PHQ-9), Posttraumatic Stress Disorder (PTSD) Checklist for DSM-5 (PCL-5), and the General Symptom Inventory (eAppendix 1).^39,42,45-52

Sample

Seventy-two participants (36 in each group) were needed to have adequate statistical power for the analysis. Presuming attrition, the goal was to recruit 100 veterans. Literature on NFB studies of patients with mTBI have reported dropout rates ranging from 10% to 30%.^53,54 Assuming a dropout rate of 28% and a moderate autocorrelation of 0.6 among repeated measures, this sample size ensured the detection of an average difference of at least 0.49 SDs with a power of 80% in the NFB intervention group compared with the control group using a 2-tailed significance level of 0.05.

Control Group

Following baseline assessment, control group participants received 8 phone calls (1 call/wk) from 1 of 4 clinical investigators over 8 to 10 weeks. During each 15-minute call, 1 of the following health topics was discussed: sleep hygiene, basic nutritional concepts, beverage choices, positive thinking, thought reframing, fitness, daily calming activity, and enhancement of focus strategies. A script for each topic was used to guide each call.

Intervention Group

Following baseline assessment, intervention group participants completed 20 half-hour ILF NFB sessions, typically receiving 3 sessions per week over an 8- to 10-week period. ILF NFB treatments were administered by 1 of 4 licensed health care employees who had received substantial ILF NFB training and achieved a skill reliability index score of 0.95, ensuring the skill level of the ILF NFB providers was equal. A script was used by the ILF NFB providers during the ILF NFB sessions to keep the interaction approach consistent with all participants.

All procedures were explained in advance to participants and voluntary participation affirmed. At the first session, participants filled out a clinical symptom checklist of 5 symptoms (eAppendix 1).^39,42,45-49 The initial rating on the symptom checklist was reflective of their experience over the past month, while in each subsequent session, participants indicated their experience of those symptoms that day. ILF NFB providers were never privy to participants’ primary or secondary outcome measures data during the study, so these recurring clinical symptom checklist ratings, as well as other feedback provided by participants on their experience within and between sessions, were the clinical data used to make decisions about ILF NFB treatment protocol.

The Othmer Optimal Response Frequency (ORF) protocol was used for participants in this study.⁵⁵ Through an iterative process, ORF protocol establishes the specific frequency point along the 0.000001 mHz to 0.1 Hz continuum, which is optimal to diminish symptoms experienced in real-time during the session (eg, tension or pain in shoulders; racing thoughts).

During each ILF NFB session, participants were seated comfortably and encouraged to look at the feedback screen. The moving images on the game screen provided almost instantaneous feedback (within 500 ms) to participants about their brain functioning, as ascertained by electrodes placed on the scalp as dictated by study protocol.⁵⁶ A standardized protocol for site placement was used beginning with T3-T4, followed by the weekly addition of a site as tolerated in the following sequence: T4-P4, FP2-T4, and FP1-T4. More information about the ILF NFB procedures are outlined in the report of the pilot study and RCT initial results.^22,34

Statistical Analysis

Eighty-seven participants were randomized, with 43 assigned to the intervention group and 44 to the control group to achieve the enrollment goal of ≥ 36 participants in each group. This report is the second analysis of data from this RCT that employed a per-protocol approach, analyzing a subset of participants who fully adhered to the study protocol and completed all study procedures. Outcome scores at baseline, midpoint, end of treatment, and 2-month follow-up were summarized as means with corresponding 95% CIs. Group comparisons at the end of treatment and 2-month follow-up time points were conducted using 2-sample t tests. All statistical tests were 2-sided with a significance level of .05 (Type I error rate). SAS software version 9.4 Maintenance 8 was used for statistical analysis. Cohen d analyses were used for effect sizes.

Results

Seventy-four participants fully adhered to the study protocol and were included in the present analyses, with 38 in the control group and 36 in the intervention group. eAppendix 2 depicts the flow of participants through this study. There were no adverse events related to treatment, and the 13 participants who withdrew typically reported difficulty with scheduling or transportation as the primary reason. This study also took place during the COVID-19 pandemic, which likely had some impact on enrollment; participants were differentially impacted by changes in employment and moves to the continental United States.

Participants were aged 30 to 60 years (mean [SD], 45.4 [8.0]). Most participants (90.5%) were male, and multiracial and White were the most common racial identities (Table 2). Participant characteristics were largely balanced across randomized groups. Similarly, test scores on the primary variables of interest in this study and secondary clinical variables assessed were comparable across participants (Table 3).

Primary Variables of Interest Analyses

This study’s hypothesis was that those who completed ILF NFB treatment per protocol would demonstrate statistically significant improvement in symptoms related to headaches, sleep disturbance, and difficulty with attention when compared with veterans in the control group. This hypothesis was partially supported. A 2-sample t test showed that veterans in the intervention group demonstrated significant improvement in headache symptoms compared with veterans in the control group on the HIT-6 at the end-of-treatment (P < .001, d = 1.53) and 2-month follow-up assessment (P < .001, d = 1.14). This pattern also was consistent with the TBI-QOL Headache Pain item short form, with veterans in the intervention group showing improvement beyond those in the control group at the end-of-treatment (P < .001, d = 0.89) and 2-month follow-up assessment (P < .001, d = 0.83). Two-sample t tests also demonstrated significant improvement in subjective reports of sleep; those in the intervention group had significantly lower scores on the ISI at the end-of-study (P < .001, d = 1.53) and 2-month follow-up assessment (P < .001, d = 0.97). This pattern also held true for the Neuro-QOL Sleep Disturbance short form subtest, which demonstrated significantly more improvement in the intervention group compared with the control group at the end-of-study (P < .001, d = 0.97) and 2-month follow- up assessment (P < .001, d = 0.92). improvement in attention was not supported by the present results. A 2-sample t test found no significant difference between performance on the QIKtest for veterans in the intervention group vs the control group at the end-of-study (P = .40, d = 0.19) or the 2-month follow-up (P = .43, d = 0.20) (eAppendix 3).

Secondary Variables of Interest Analysis

Secondary variables examined differences in QOL, PTSD, depressive symptoms, and general symptoms reported between veterans in the intervention and control groups. Results demonstrated that veterans in the intervention group showed improvement above and beyond those in the control group on all measures. In regard to QOL, veterans in the intervention group had significantly higher scores on the Neuro-QOL Participate subtest than those in the control group at the end-of-study (P = .01, d = 0.89) and 2-month follow-up assessment (P < .001, d = 0.62). A similar pattern was found for the Neuro-QOL Satisfaction subtest, with veterans in the intervention group showing significantly higher scores than those in the control group at the end-of-study (P = .001, d = 0.95) and 2-month follow-up assessment (P < .001, d = 0.62). This also held true on the QOLIBRI, with veterans in the intervention group demonstrating significantly higher scores than those in the control group at the end-of-study (P = .001, d = 0.92) and 2-month follow-up assessment (P < .001, d = 0.66).

Veterans in the intervention group had significantly lower scores on the PCL-5 than those in the control group at the end-of- study (P = .003, d = 0.78) and 2-month follow-up assessment (P = .001, d = 0.72). Veterans in the intervention group also had significantly lower scores on the PHQ-9 than those in the control group at the end-of-study (P < .001, d = 0.98) and 2-month follow-up assessment (P < .001, d = 0.83). Veterans in the intervention group had significantly lower scores on the DASS- 21 than those in the control group at the end-of-study (P = .002, d = 0.80) and 2-month follow-up assessment (P = .001, d = 0.77). They also had significantly lower scores on the General Symptom Inventory than those in the control group at the end-of-study (P = .02, d = 0.75) and 2-month follow-up assessment (P = .002, d = 0.57). A clinically significant shift of score occurred for each of the measures except DASS-21 (eAppendix 3). eAppendix 4 depicts the change in scores for the intervention group at the end of treatment and the clinically significant shift score of each measure.

Discussion

The results of this RCT revealed a promising impact of ILF NFB on the commonly experienced persistent PCSs of headaches and disrupted sleep. Veterans in the intervention group demonstrated statistically significant improvement in headache symptoms compared with veterans in the control group when assessed at the end of treatment and during a 2-month follow-up. The statistical significance of these improvements was also supported by large or very large effect sizes. In addition to these primary variables of interest, veterans in the intervention group notably demonstrated significant improvement compared with those in the control group in a number of secondary clinical measures, including QOL, traumatic stress-related symptoms, depressive symptoms, and general symptom report. The clinical impact was further supported by the clinically relevant shift in scores in the intervention group.

The data did not support the hypothesis that attention concerns would show significant improvement following ILF NFB. Performance on an attention measure did not differ significantly between groups at either the end-of-treatment or 2-month follow up assessment. The QIKtest, a continuous performance test used to measure attention, was a go/no-go task and calculated based on a combination of various types of errors and outlier responses. The stimulus for this task is a series of computerized, blinking lights, for which participants are tasked with discriminating targets and nontargets under time pressure. However, the order of the stimuli are consistent across administrations, rather than being randomized, introducing a potential confound of practice effects on this task since patients were administered the QIKtest 3 times in a 2-month period and again 2 months later. Veterans in the control group notably improved in their average performance of this task from baseline to the endpoint of their treatment participation and demonstrated further improvement at the 2-month follow-up assessment; this pattern would be consistent with potential practice effects and warrants caution in its interpretation for both groups.

Previously published ILF NFB clinical studies that used the QIKtest and found positive results were mostly conducted among children and teen populations across longer treatment periods. This research may indicate the QIKtest is not an appropriate measure to assess adults who have specialized training in responding to stimuli (ie, trained military personnel). This suggests the concept of attention dysfunction experienced by veterans and the best method to measure it may need to be explored further. This construct may not be related to the focus and skill in prolonged attention needed in selecting go/ no-go tasks, but rather related to a broader conceptual basis involving memory, recall, clarity of rational thought, and decision making impacted by the mTBI. For instance, a study among combat veterans with mTBI and PTSD found that performance on objective cognitive measures did not significantly correlate with their subjective reports of cognitive difficulties.⁵⁷ This reflects the pattern of the present study, in which subjective reports of attention improved over time on the clinical symptom checklist filled out by participants at each session, but the objective measure did not. The mean attention dysfunction score was 6 at session 1 and 1 to 2 at session 20 (lower scores are better on a 10-point scale).

Strengths and Limitations

This study presents results stemming from the first RCT examining clinical effectiveness of ILF NFB in a VA setting for veterans with diagnoses of mTBI. The study design shows promising external validity. Veterans were able to participate in a treatment consisting of 20 sessions over a period of typically 8 to 10 weeks, entailing 2 to 3 sessions per week, with an attrition of only 18% over the course of the study. Notably, attrition rates may have been impacted by the time course of the study, which was recruiting and running participants throughout the COVID-19 pandemic (March 2020 to May 2023). No attrition was due to the intervention itself, and no adverse reactions to ILF NFB were reported during the course of the study. Other strengths of the study include the ethnically and racially diverse participants, representative of the population of veterans in Hawaii. Additionally, all ILF NFB providers underwent supervised ILF NFB training and achieved a skill reliability index score of 0.95 prior to providing ILF NFB to the intervention group.

This study was not blinded. Neither veterans nor ILF NFB providers were blinded and were therefore aware of the randomly assigned groups. Research assistants administering the periodic assessments were meant to be blinded to condition by design; however, as the study progressed, a research assistant became unintentionally aware of each study participant's condition based on required documentation in the veteran’s health records; more notes were present for those in the intervention group (20 specialist notes) than the control group (8 notes). While the presence of a control group represents a strength relative to much of the existing ILF NFB literature, the control group in this case did not account for the total time spent with the researchers. Participants in the intervention group met with researchers for 20 total sessions as opposed to 8 telephone calls. Therefore, the study design cannot fully rule out the differential impact of demand characteristics between the 2 groups, nor can it fully address or rule out the impact of differential motivation and expectations between groups. There is also evidence that technological innovation can influence the expectations of research participants, meaning that the intervention group may have been unduly influenced by the novelty of the ILF NFB technology, to which the control group did not have exposure.⁵⁸

A second attention measure for this study would have been beneficial, perhaps in identifying true change in attention ability or providing more insight into finding better methods to assess attention among veterans with mTBI. ILF NFB demonstrated significant impact across multiple outcome measures of clinical relevance for veterans diagnosed with mTBI, including the primary outcome variables of headache and sleep. The strength of the improvements seen in these areas, supported by large practical effects as well as veterans’ subjective reports, indicates much promise. Follow-up studies may also focus on the potential effectiveness of ILF NFB as a treatment of the secondary concerns measured in this study, including traumatic stress-related and depressive symptoms, and may explore the added benefit, if any, of ILF NFB alongside other evidence-based treatments for traumatic stress-related and mood disorders (eg, cognitive behavioral therapy). Using functional magnetic resonance imaging before and after assessments to determine actual brain enhancement with ILF NFB for certain disorders in which a brain signature exists (ie, migraine) should be explored. Further examination of ILF NFB as an intervention for attention may also be warranted, using more effective measures of attention in the population of veterans with mTBI, given the concerns noted earlier. Future research on this topic will need to clearly define attention in relation to the veteran experience and use relevant measures.

Conclusions

This study supports ILF NFB as a safe, noninvasive, nonpharmacologic treatment that may be effective in addressing the complex clinical concerns of veterans diagnosed with mTBI, a population for whom effective treatments have been difficult to identify. This intervention can provide veterans with a desirable and effective nonpharmacologic alternative in their care.

Traumatic brain injury (TBI) is the signature injury of post-9/11 military operations, impacting > 441,000 combat veterans from 2001 to 2021 and 87% diagnosed with mild TBI (mTBI).^1,2 The most common cause of mTBI during these operations was blast exposures stemming from improvised explosive devices, rocket-propelled grenades, or land mines. mTBI was once thought to be self-limiting, lasting hours or days postinjury, but is now recognized as a complex focal and diffuse injury causing a cascade of molecular and biochemical responses with significant physiologic effects lasting for a longer duration. A significant number of combat veterans with mTBI (23%-48%) experience long-standing postconcussive symptoms (PCSs) for many years postinjury.^3-5

Developing and implementing strategies to reduce persistent symptoms associated with mTBI is of critical importance. Veterans diagnosed with mTBI and experiencing PCSs present ongoing treatment challenges to the health care system due to limited or suboptimal treatment options.^6,7 According to the 2021 US Department of Veterans Affairs (VA) and US Department of Defense (DoD) clinical guidelines for postacute mTBI, treatment for PCSs should be symptom focused. ^8,9 For instance, veterans with migraine headaches associated with mTBIs are often treated with abortive agents (eg, triptans) and preventive medications (eg, anticonvulsants and tricyclics).¹⁰ Cognitive dysfunction and insomnia are treated with cognitive rehabilitation programs, cognitive behaviorial therapy, occupational therapy, and medications (eg, hypnotics for insomnia).^11,12 The 2021 VA/DoD guidelines note that veteran and military focus groups described greater success with nonpharmacologic treatments than with pharmacologic treatments.⁸ The VA launched an enterprise-wide Whole Health Service program with the requirement that complementary and integrative health approaches must be available to veterans.¹³ As a nonpharmacologic, integrative, and noninvasive modality, neurofeedback (NFB) supports the VA Whole Health initiative and veterans’ preferences for integrative treatments.¹⁴

Neurofeedback

Rather than a symptom management approach, Defina et al described the possibilities of brain repair in TBI by treatments to enhance neuroplasticity, thereby establishing a more normalized or stable brain environment and enabling the brain to reorganize itself and function more normally.¹⁵ NFB has been shown to influence neuroplasticity,¹⁶ as evident in microstructural changes in white and gray matter¹⁷ and its ability to contribute to functional rehabilitation by restoring connectivity in specific areas of the brain that may have been impaired.¹⁸ The benefits of neuroenhancement strategies include potentially reduced pain for patients with mTBI and improved quality of life (QOL).¹⁹

NFB assists individuals by helping them become more aware of and self-regulate their physiology.^20,21 Because there are several types of NFB (eg, quantitative electroencephalography, Z-scored, α-θ) that differ in terms of equipment, mechanism of action, focus, and patient and clinician procedures, it is important to note that this study used a novel technologically advanced form of NFB, referred to as infra-low frequency (ILF) NFB. It works by reflecting a person’s brain wave activity via conventional electroencephalography back to the person through the visual cortex, thus providing relevant information to which the brain responds to improve core state regulation.²²

In 2006, ILF NFB developers sought to extend NFB capability into the slow cortical potential domain (< 0.1 Hz) and then gradually extended to lower frequencies on the basis of favorable clinical responses.^22,23 In 2017, the technology reached an ILF capacity that appeared to be helpful for several clinical issues. These developments depended on instrumentation capable of low noise signal detection down to the lowest frequency of interest. Instrumentation was developed for the purpose (eg, Bee Medic Cygnet NFB).

Although mTBI has been a clinical focus in NFB since the 1980s, there are few published studies demonstrating the efficacy of ILF NFB relating to the PCSs of interest in this study, and 2 suggested ILF NFB positively affected change in PCS severity.^24,25 Other studies found that ILF NFB decreased incidence of migraines and tension type headaches.^26,27 However, the findings of these studies had limited generalizability due to methodologic limitations, such as selection bias and small sample sizes.^24-27 Of importance to this article, there are also several publications on the efficacy of ILF NFB in clinical settings.^28-33

This article presents the second analysis of data from veterans who completed ILF NFB intervention and control group procedures during a 5-year randomized controlled trial (RCT). The RCT included veterans who experienced an mTBI while participating in post-9/11 military operations to evaluate the impact of ILF NFB on chronic PCSs, including headache, insomnia, and attention dysfunction. Initial results of this trial demonstrated significant differences between the intervention and control groups with strong effect sizes on all outcome measures at the end of treatment.³⁴

Methods

Participants included male and nonpregnant female veterans with a diagnosed mTBI during post-9/11 military operations; aged 18 to 65 years; reports of persistent (ie, > 3 months in duration) headaches, insomnia, and attention difficulties; and able to read and write English, comprehend what is read, and follow directions. mTBI diagnosis was verified for each veteran via the electronic health record. Patients were excluded if they had a severe TBI diagnosis or impaired decision-making capacity; were unable to comply with study visit schedule; or endorsed active suicidal intent on the Columbia-Suicide Severity Rating Scale.³⁵

Recruitment efforts included: (1) letters sent to eligible veterans with mTBI who were identified by clinical informatics data after waiver of Health Insurance Portability and Accountability Act was obtained; veterans could contact the research team directly or the research team would call the veteran 2 weeks after the letter was sent; (2) veterans could be referred by a clinician; and (3) veterans could self-refer based on flyers and other study marketing materials.

The study was conducted from 2019 to 2024 at Spark M. Matsunaga VA Medical Center, in Honolulu, Hawaii. Four private research spaces in compliance with human research standards were used for consent, treatment, and assessment.

Consenting Procedure and Randomization

The privacy rights of potential participants were observed, and interested veterans who met the eligibility criteria underwent an informed consent procedure and were administered the Columbia-Suicide Severity Rating Scale.³⁵ Those veterans not indicating active suicidal intent were randomized into the intervention or control group. Once randomized, the participant was enrolled and scheduled for baseline assessment.

All procedures of this study were performed in adherence with relevant laws and institutional guidelines. The study was reviewed and approved by the VA Pacific Islands Health Care System Institutional Review Board (#2019-06-JC/Promise 0003).

Outcome Measures

The outcome measures were administered at baseline, midpoint (3-7 weeks), end of treatment (6-12 weeks), and at a 2-month follow-up appointment with the research assistant or project coordinator.

The primary outcome measures include the Headache Impact Test (HIT-6), TBIQOL Headache Pain item short form, Insomnia Severity Index (ISI), Quality of Life in Neurological Disorders (Neuro-QOL) Sleep Disturbance short form, and attention measure: QIKtest Continuous Performance Test (QIKtest) (Table 1).^36-44

Secondary outcome measures included QOL After Brain Injury (QOLIBRI), Neuro- QOL Satisfaction With Roles/Activities short form (Neuro-QOL Satisfaction), Neuro-QOL Ability to Participate in Roles/Activities short form (Neuro-QOL Participate), Depression Anxiety Stress Scales (DASS-21), Patient Health Questionnaire-9 (PHQ-9), Posttraumatic Stress Disorder (PTSD) Checklist for DSM-5 (PCL-5), and the General Symptom Inventory (eAppendix 1).^39,42,45-52

Sample

Seventy-two participants (36 in each group) were needed to have adequate statistical power for the analysis. Presuming attrition, the goal was to recruit 100 veterans. Literature on NFB studies of patients with mTBI have reported dropout rates ranging from 10% to 30%.^53,54 Assuming a dropout rate of 28% and a moderate autocorrelation of 0.6 among repeated measures, this sample size ensured the detection of an average difference of at least 0.49 SDs with a power of 80% in the NFB intervention group compared with the control group using a 2-tailed significance level of 0.05.

Control Group

Following baseline assessment, control group participants received 8 phone calls (1 call/wk) from 1 of 4 clinical investigators over 8 to 10 weeks. During each 15-minute call, 1 of the following health topics was discussed: sleep hygiene, basic nutritional concepts, beverage choices, positive thinking, thought reframing, fitness, daily calming activity, and enhancement of focus strategies. A script for each topic was used to guide each call.

Intervention Group

Following baseline assessment, intervention group participants completed 20 half-hour ILF NFB sessions, typically receiving 3 sessions per week over an 8- to 10-week period. ILF NFB treatments were administered by 1 of 4 licensed health care employees who had received substantial ILF NFB training and achieved a skill reliability index score of 0.95, ensuring the skill level of the ILF NFB providers was equal. A script was used by the ILF NFB providers during the ILF NFB sessions to keep the interaction approach consistent with all participants.

All procedures were explained in advance to participants and voluntary participation affirmed. At the first session, participants filled out a clinical symptom checklist of 5 symptoms (eAppendix 1).^39,42,45-49 The initial rating on the symptom checklist was reflective of their experience over the past month, while in each subsequent session, participants indicated their experience of those symptoms that day. ILF NFB providers were never privy to participants’ primary or secondary outcome measures data during the study, so these recurring clinical symptom checklist ratings, as well as other feedback provided by participants on their experience within and between sessions, were the clinical data used to make decisions about ILF NFB treatment protocol.

The Othmer Optimal Response Frequency (ORF) protocol was used for participants in this study.⁵⁵ Through an iterative process, ORF protocol establishes the specific frequency point along the 0.000001 mHz to 0.1 Hz continuum, which is optimal to diminish symptoms experienced in real-time during the session (eg, tension or pain in shoulders; racing thoughts).

During each ILF NFB session, participants were seated comfortably and encouraged to look at the feedback screen. The moving images on the game screen provided almost instantaneous feedback (within 500 ms) to participants about their brain functioning, as ascertained by electrodes placed on the scalp as dictated by study protocol.⁵⁶ A standardized protocol for site placement was used beginning with T3-T4, followed by the weekly addition of a site as tolerated in the following sequence: T4-P4, FP2-T4, and FP1-T4. More information about the ILF NFB procedures are outlined in the report of the pilot study and RCT initial results.^22,34

Statistical Analysis

Eighty-seven participants were randomized, with 43 assigned to the intervention group and 44 to the control group to achieve the enrollment goal of ≥ 36 participants in each group. This report is the second analysis of data from this RCT that employed a per-protocol approach, analyzing a subset of participants who fully adhered to the study protocol and completed all study procedures. Outcome scores at baseline, midpoint, end of treatment, and 2-month follow-up were summarized as means with corresponding 95% CIs. Group comparisons at the end of treatment and 2-month follow-up time points were conducted using 2-sample t tests. All statistical tests were 2-sided with a significance level of .05 (Type I error rate). SAS software version 9.4 Maintenance 8 was used for statistical analysis. Cohen d analyses were used for effect sizes.

Results

Seventy-four participants fully adhered to the study protocol and were included in the present analyses, with 38 in the control group and 36 in the intervention group. eAppendix 2 depicts the flow of participants through this study. There were no adverse events related to treatment, and the 13 participants who withdrew typically reported difficulty with scheduling or transportation as the primary reason. This study also took place during the COVID-19 pandemic, which likely had some impact on enrollment; participants were differentially impacted by changes in employment and moves to the continental United States.

Participants were aged 30 to 60 years (mean [SD], 45.4 [8.0]). Most participants (90.5%) were male, and multiracial and White were the most common racial identities (Table 2). Participant characteristics were largely balanced across randomized groups. Similarly, test scores on the primary variables of interest in this study and secondary clinical variables assessed were comparable across participants (Table 3).

Primary Variables of Interest Analyses

This study’s hypothesis was that those who completed ILF NFB treatment per protocol would demonstrate statistically significant improvement in symptoms related to headaches, sleep disturbance, and difficulty with attention when compared with veterans in the control group. This hypothesis was partially supported. A 2-sample t test showed that veterans in the intervention group demonstrated significant improvement in headache symptoms compared with veterans in the control group on the HIT-6 at the end-of-treatment (P < .001, d = 1.53) and 2-month follow-up assessment (P < .001, d = 1.14). This pattern also was consistent with the TBI-QOL Headache Pain item short form, with veterans in the intervention group showing improvement beyond those in the control group at the end-of-treatment (P < .001, d = 0.89) and 2-month follow-up assessment (P < .001, d = 0.83). Two-sample t tests also demonstrated significant improvement in subjective reports of sleep; those in the intervention group had significantly lower scores on the ISI at the end-of-study (P < .001, d = 1.53) and 2-month follow-up assessment (P < .001, d = 0.97). This pattern also held true for the Neuro-QOL Sleep Disturbance short form subtest, which demonstrated significantly more improvement in the intervention group compared with the control group at the end-of-study (P < .001, d = 0.97) and 2-month follow- up assessment (P < .001, d = 0.92). improvement in attention was not supported by the present results. A 2-sample t test found no significant difference between performance on the QIKtest for veterans in the intervention group vs the control group at the end-of-study (P = .40, d = 0.19) or the 2-month follow-up (P = .43, d = 0.20) (eAppendix 3).

Secondary Variables of Interest Analysis

Secondary variables examined differences in QOL, PTSD, depressive symptoms, and general symptoms reported between veterans in the intervention and control groups. Results demonstrated that veterans in the intervention group showed improvement above and beyond those in the control group on all measures. In regard to QOL, veterans in the intervention group had significantly higher scores on the Neuro-QOL Participate subtest than those in the control group at the end-of-study (P = .01, d = 0.89) and 2-month follow-up assessment (P < .001, d = 0.62). A similar pattern was found for the Neuro-QOL Satisfaction subtest, with veterans in the intervention group showing significantly higher scores than those in the control group at the end-of-study (P = .001, d = 0.95) and 2-month follow-up assessment (P < .001, d = 0.62). This also held true on the QOLIBRI, with veterans in the intervention group demonstrating significantly higher scores than those in the control group at the end-of-study (P = .001, d = 0.92) and 2-month follow-up assessment (P < .001, d = 0.66).

Veterans in the intervention group had significantly lower scores on the PCL-5 than those in the control group at the end-of- study (P = .003, d = 0.78) and 2-month follow-up assessment (P = .001, d = 0.72). Veterans in the intervention group also had significantly lower scores on the PHQ-9 than those in the control group at the end-of-study (P < .001, d = 0.98) and 2-month follow-up assessment (P < .001, d = 0.83). Veterans in the intervention group had significantly lower scores on the DASS- 21 than those in the control group at the end-of-study (P = .002, d = 0.80) and 2-month follow-up assessment (P = .001, d = 0.77). They also had significantly lower scores on the General Symptom Inventory than those in the control group at the end-of-study (P = .02, d = 0.75) and 2-month follow-up assessment (P = .002, d = 0.57). A clinically significant shift of score occurred for each of the measures except DASS-21 (eAppendix 3). eAppendix 4 depicts the change in scores for the intervention group at the end of treatment and the clinically significant shift score of each measure.

Discussion

The results of this RCT revealed a promising impact of ILF NFB on the commonly experienced persistent PCSs of headaches and disrupted sleep. Veterans in the intervention group demonstrated statistically significant improvement in headache symptoms compared with veterans in the control group when assessed at the end of treatment and during a 2-month follow-up. The statistical significance of these improvements was also supported by large or very large effect sizes. In addition to these primary variables of interest, veterans in the intervention group notably demonstrated significant improvement compared with those in the control group in a number of secondary clinical measures, including QOL, traumatic stress-related symptoms, depressive symptoms, and general symptom report. The clinical impact was further supported by the clinically relevant shift in scores in the intervention group.

The data did not support the hypothesis that attention concerns would show significant improvement following ILF NFB. Performance on an attention measure did not differ significantly between groups at either the end-of-treatment or 2-month follow up assessment. The QIKtest, a continuous performance test used to measure attention, was a go/no-go task and calculated based on a combination of various types of errors and outlier responses. The stimulus for this task is a series of computerized, blinking lights, for which participants are tasked with discriminating targets and nontargets under time pressure. However, the order of the stimuli are consistent across administrations, rather than being randomized, introducing a potential confound of practice effects on this task since patients were administered the QIKtest 3 times in a 2-month period and again 2 months later. Veterans in the control group notably improved in their average performance of this task from baseline to the endpoint of their treatment participation and demonstrated further improvement at the 2-month follow-up assessment; this pattern would be consistent with potential practice effects and warrants caution in its interpretation for both groups.

Previously published ILF NFB clinical studies that used the QIKtest and found positive results were mostly conducted among children and teen populations across longer treatment periods. This research may indicate the QIKtest is not an appropriate measure to assess adults who have specialized training in responding to stimuli (ie, trained military personnel). This suggests the concept of attention dysfunction experienced by veterans and the best method to measure it may need to be explored further. This construct may not be related to the focus and skill in prolonged attention needed in selecting go/ no-go tasks, but rather related to a broader conceptual basis involving memory, recall, clarity of rational thought, and decision making impacted by the mTBI. For instance, a study among combat veterans with mTBI and PTSD found that performance on objective cognitive measures did not significantly correlate with their subjective reports of cognitive difficulties.⁵⁷ This reflects the pattern of the present study, in which subjective reports of attention improved over time on the clinical symptom checklist filled out by participants at each session, but the objective measure did not. The mean attention dysfunction score was 6 at session 1 and 1 to 2 at session 20 (lower scores are better on a 10-point scale).

Strengths and Limitations

This study presents results stemming from the first RCT examining clinical effectiveness of ILF NFB in a VA setting for veterans with diagnoses of mTBI. The study design shows promising external validity. Veterans were able to participate in a treatment consisting of 20 sessions over a period of typically 8 to 10 weeks, entailing 2 to 3 sessions per week, with an attrition of only 18% over the course of the study. Notably, attrition rates may have been impacted by the time course of the study, which was recruiting and running participants throughout the COVID-19 pandemic (March 2020 to May 2023). No attrition was due to the intervention itself, and no adverse reactions to ILF NFB were reported during the course of the study. Other strengths of the study include the ethnically and racially diverse participants, representative of the population of veterans in Hawaii. Additionally, all ILF NFB providers underwent supervised ILF NFB training and achieved a skill reliability index score of 0.95 prior to providing ILF NFB to the intervention group.

This study was not blinded. Neither veterans nor ILF NFB providers were blinded and were therefore aware of the randomly assigned groups. Research assistants administering the periodic assessments were meant to be blinded to condition by design; however, as the study progressed, a research assistant became unintentionally aware of each study participant's condition based on required documentation in the veteran’s health records; more notes were present for those in the intervention group (20 specialist notes) than the control group (8 notes). While the presence of a control group represents a strength relative to much of the existing ILF NFB literature, the control group in this case did not account for the total time spent with the researchers. Participants in the intervention group met with researchers for 20 total sessions as opposed to 8 telephone calls. Therefore, the study design cannot fully rule out the differential impact of demand characteristics between the 2 groups, nor can it fully address or rule out the impact of differential motivation and expectations between groups. There is also evidence that technological innovation can influence the expectations of research participants, meaning that the intervention group may have been unduly influenced by the novelty of the ILF NFB technology, to which the control group did not have exposure.⁵⁸

A second attention measure for this study would have been beneficial, perhaps in identifying true change in attention ability or providing more insight into finding better methods to assess attention among veterans with mTBI. ILF NFB demonstrated significant impact across multiple outcome measures of clinical relevance for veterans diagnosed with mTBI, including the primary outcome variables of headache and sleep. The strength of the improvements seen in these areas, supported by large practical effects as well as veterans’ subjective reports, indicates much promise. Follow-up studies may also focus on the potential effectiveness of ILF NFB as a treatment of the secondary concerns measured in this study, including traumatic stress-related and depressive symptoms, and may explore the added benefit, if any, of ILF NFB alongside other evidence-based treatments for traumatic stress-related and mood disorders (eg, cognitive behavioral therapy). Using functional magnetic resonance imaging before and after assessments to determine actual brain enhancement with ILF NFB for certain disorders in which a brain signature exists (ie, migraine) should be explored. Further examination of ILF NFB as an intervention for attention may also be warranted, using more effective measures of attention in the population of veterans with mTBI, given the concerns noted earlier. Future research on this topic will need to clearly define attention in relation to the veteran experience and use relevant measures.

Conclusions

This study supports ILF NFB as a safe, noninvasive, nonpharmacologic treatment that may be effective in addressing the complex clinical concerns of veterans diagnosed with mTBI, a population for whom effective treatments have been difficult to identify. This intervention can provide veterans with a desirable and effective nonpharmacologic alternative in their care.

Traumatic brain injury (TBI) is the signature injury of post-9/11 military operations, impacting > 441,000 combat veterans from 2001 to 2021 and 87% diagnosed with mild TBI (mTBI).^1,2 The most common cause of mTBI during these operations was blast exposures stemming from improvised explosive devices, rocket-propelled grenades, or land mines. mTBI was once thought to be self-limiting, lasting hours or days postinjury, but is now recognized as a complex focal and diffuse injury causing a cascade of molecular and biochemical responses with significant physiologic effects lasting for a longer duration. A significant number of combat veterans with mTBI (23%-48%) experience long-standing postconcussive symptoms (PCSs) for many years postinjury.^3-5

Developing and implementing strategies to reduce persistent symptoms associated with mTBI is of critical importance. Veterans diagnosed with mTBI and experiencing PCSs present ongoing treatment challenges to the health care system due to limited or suboptimal treatment options.^6,7 According to the 2021 US Department of Veterans Affairs (VA) and US Department of Defense (DoD) clinical guidelines for postacute mTBI, treatment for PCSs should be symptom focused. ^8,9 For instance, veterans with migraine headaches associated with mTBIs are often treated with abortive agents (eg, triptans) and preventive medications (eg, anticonvulsants and tricyclics).¹⁰ Cognitive dysfunction and insomnia are treated with cognitive rehabilitation programs, cognitive behaviorial therapy, occupational therapy, and medications (eg, hypnotics for insomnia).^11,12 The 2021 VA/DoD guidelines note that veteran and military focus groups described greater success with nonpharmacologic treatments than with pharmacologic treatments.⁸ The VA launched an enterprise-wide Whole Health Service program with the requirement that complementary and integrative health approaches must be available to veterans.¹³ As a nonpharmacologic, integrative, and noninvasive modality, neurofeedback (NFB) supports the VA Whole Health initiative and veterans’ preferences for integrative treatments.¹⁴

Neurofeedback

Rather than a symptom management approach, Defina et al described the possibilities of brain repair in TBI by treatments to enhance neuroplasticity, thereby establishing a more normalized or stable brain environment and enabling the brain to reorganize itself and function more normally.¹⁵ NFB has been shown to influence neuroplasticity,¹⁶ as evident in microstructural changes in white and gray matter¹⁷ and its ability to contribute to functional rehabilitation by restoring connectivity in specific areas of the brain that may have been impaired.¹⁸ The benefits of neuroenhancement strategies include potentially reduced pain for patients with mTBI and improved quality of life (QOL).¹⁹

NFB assists individuals by helping them become more aware of and self-regulate their physiology.^20,21 Because there are several types of NFB (eg, quantitative electroencephalography, Z-scored, α-θ) that differ in terms of equipment, mechanism of action, focus, and patient and clinician procedures, it is important to note that this study used a novel technologically advanced form of NFB, referred to as infra-low frequency (ILF) NFB. It works by reflecting a person’s brain wave activity via conventional electroencephalography back to the person through the visual cortex, thus providing relevant information to which the brain responds to improve core state regulation.²²

In 2006, ILF NFB developers sought to extend NFB capability into the slow cortical potential domain (< 0.1 Hz) and then gradually extended to lower frequencies on the basis of favorable clinical responses.^22,23 In 2017, the technology reached an ILF capacity that appeared to be helpful for several clinical issues. These developments depended on instrumentation capable of low noise signal detection down to the lowest frequency of interest. Instrumentation was developed for the purpose (eg, Bee Medic Cygnet NFB).

Although mTBI has been a clinical focus in NFB since the 1980s, there are few published studies demonstrating the efficacy of ILF NFB relating to the PCSs of interest in this study, and 2 suggested ILF NFB positively affected change in PCS severity.^24,25 Other studies found that ILF NFB decreased incidence of migraines and tension type headaches.^26,27 However, the findings of these studies had limited generalizability due to methodologic limitations, such as selection bias and small sample sizes.^24-27 Of importance to this article, there are also several publications on the efficacy of ILF NFB in clinical settings.^28-33

This article presents the second analysis of data from veterans who completed ILF NFB intervention and control group procedures during a 5-year randomized controlled trial (RCT). The RCT included veterans who experienced an mTBI while participating in post-9/11 military operations to evaluate the impact of ILF NFB on chronic PCSs, including headache, insomnia, and attention dysfunction. Initial results of this trial demonstrated significant differences between the intervention and control groups with strong effect sizes on all outcome measures at the end of treatment.³⁴

Methods

Participants included male and nonpregnant female veterans with a diagnosed mTBI during post-9/11 military operations; aged 18 to 65 years; reports of persistent (ie, > 3 months in duration) headaches, insomnia, and attention difficulties; and able to read and write English, comprehend what is read, and follow directions. mTBI diagnosis was verified for each veteran via the electronic health record. Patients were excluded if they had a severe TBI diagnosis or impaired decision-making capacity; were unable to comply with study visit schedule; or endorsed active suicidal intent on the Columbia-Suicide Severity Rating Scale.³⁵

Recruitment efforts included: (1) letters sent to eligible veterans with mTBI who were identified by clinical informatics data after waiver of Health Insurance Portability and Accountability Act was obtained; veterans could contact the research team directly or the research team would call the veteran 2 weeks after the letter was sent; (2) veterans could be referred by a clinician; and (3) veterans could self-refer based on flyers and other study marketing materials.

The study was conducted from 2019 to 2024 at Spark M. Matsunaga VA Medical Center, in Honolulu, Hawaii. Four private research spaces in compliance with human research standards were used for consent, treatment, and assessment.

Consenting Procedure and Randomization

The privacy rights of potential participants were observed, and interested veterans who met the eligibility criteria underwent an informed consent procedure and were administered the Columbia-Suicide Severity Rating Scale.³⁵ Those veterans not indicating active suicidal intent were randomized into the intervention or control group. Once randomized, the participant was enrolled and scheduled for baseline assessment.

All procedures of this study were performed in adherence with relevant laws and institutional guidelines. The study was reviewed and approved by the VA Pacific Islands Health Care System Institutional Review Board (#2019-06-JC/Promise 0003).

Outcome Measures

The outcome measures were administered at baseline, midpoint (3-7 weeks), end of treatment (6-12 weeks), and at a 2-month follow-up appointment with the research assistant or project coordinator.

The primary outcome measures include the Headache Impact Test (HIT-6), TBIQOL Headache Pain item short form, Insomnia Severity Index (ISI), Quality of Life in Neurological Disorders (Neuro-QOL) Sleep Disturbance short form, and attention measure: QIKtest Continuous Performance Test (QIKtest) (Table 1).^36-44

Secondary outcome measures included QOL After Brain Injury (QOLIBRI), Neuro- QOL Satisfaction With Roles/Activities short form (Neuro-QOL Satisfaction), Neuro-QOL Ability to Participate in Roles/Activities short form (Neuro-QOL Participate), Depression Anxiety Stress Scales (DASS-21), Patient Health Questionnaire-9 (PHQ-9), Posttraumatic Stress Disorder (PTSD) Checklist for DSM-5 (PCL-5), and the General Symptom Inventory (eAppendix 1).^39,42,45-52

Sample

Seventy-two participants (36 in each group) were needed to have adequate statistical power for the analysis. Presuming attrition, the goal was to recruit 100 veterans. Literature on NFB studies of patients with mTBI have reported dropout rates ranging from 10% to 30%.^53,54 Assuming a dropout rate of 28% and a moderate autocorrelation of 0.6 among repeated measures, this sample size ensured the detection of an average difference of at least 0.49 SDs with a power of 80% in the NFB intervention group compared with the control group using a 2-tailed significance level of 0.05.

Control Group

Following baseline assessment, control group participants received 8 phone calls (1 call/wk) from 1 of 4 clinical investigators over 8 to 10 weeks. During each 15-minute call, 1 of the following health topics was discussed: sleep hygiene, basic nutritional concepts, beverage choices, positive thinking, thought reframing, fitness, daily calming activity, and enhancement of focus strategies. A script for each topic was used to guide each call.

Intervention Group

Following baseline assessment, intervention group participants completed 20 half-hour ILF NFB sessions, typically receiving 3 sessions per week over an 8- to 10-week period. ILF NFB treatments were administered by 1 of 4 licensed health care employees who had received substantial ILF NFB training and achieved a skill reliability index score of 0.95, ensuring the skill level of the ILF NFB providers was equal. A script was used by the ILF NFB providers during the ILF NFB sessions to keep the interaction approach consistent with all participants.

All procedures were explained in advance to participants and voluntary participation affirmed. At the first session, participants filled out a clinical symptom checklist of 5 symptoms (eAppendix 1).^39,42,45-49 The initial rating on the symptom checklist was reflective of their experience over the past month, while in each subsequent session, participants indicated their experience of those symptoms that day. ILF NFB providers were never privy to participants’ primary or secondary outcome measures data during the study, so these recurring clinical symptom checklist ratings, as well as other feedback provided by participants on their experience within and between sessions, were the clinical data used to make decisions about ILF NFB treatment protocol.

The Othmer Optimal Response Frequency (ORF) protocol was used for participants in this study.⁵⁵ Through an iterative process, ORF protocol establishes the specific frequency point along the 0.000001 mHz to 0.1 Hz continuum, which is optimal to diminish symptoms experienced in real-time during the session (eg, tension or pain in shoulders; racing thoughts).

During each ILF NFB session, participants were seated comfortably and encouraged to look at the feedback screen. The moving images on the game screen provided almost instantaneous feedback (within 500 ms) to participants about their brain functioning, as ascertained by electrodes placed on the scalp as dictated by study protocol.⁵⁶ A standardized protocol for site placement was used beginning with T3-T4, followed by the weekly addition of a site as tolerated in the following sequence: T4-P4, FP2-T4, and FP1-T4. More information about the ILF NFB procedures are outlined in the report of the pilot study and RCT initial results.^22,34

Statistical Analysis

Eighty-seven participants were randomized, with 43 assigned to the intervention group and 44 to the control group to achieve the enrollment goal of ≥ 36 participants in each group. This report is the second analysis of data from this RCT that employed a per-protocol approach, analyzing a subset of participants who fully adhered to the study protocol and completed all study procedures. Outcome scores at baseline, midpoint, end of treatment, and 2-month follow-up were summarized as means with corresponding 95% CIs. Group comparisons at the end of treatment and 2-month follow-up time points were conducted using 2-sample t tests. All statistical tests were 2-sided with a significance level of .05 (Type I error rate). SAS software version 9.4 Maintenance 8 was used for statistical analysis. Cohen d analyses were used for effect sizes.

Results

Seventy-four participants fully adhered to the study protocol and were included in the present analyses, with 38 in the control group and 36 in the intervention group. eAppendix 2 depicts the flow of participants through this study. There were no adverse events related to treatment, and the 13 participants who withdrew typically reported difficulty with scheduling or transportation as the primary reason. This study also took place during the COVID-19 pandemic, which likely had some impact on enrollment; participants were differentially impacted by changes in employment and moves to the continental United States.

Participants were aged 30 to 60 years (mean [SD], 45.4 [8.0]). Most participants (90.5%) were male, and multiracial and White were the most common racial identities (Table 2). Participant characteristics were largely balanced across randomized groups. Similarly, test scores on the primary variables of interest in this study and secondary clinical variables assessed were comparable across participants (Table 3).

Primary Variables of Interest Analyses

This study’s hypothesis was that those who completed ILF NFB treatment per protocol would demonstrate statistically significant improvement in symptoms related to headaches, sleep disturbance, and difficulty with attention when compared with veterans in the control group. This hypothesis was partially supported. A 2-sample t test showed that veterans in the intervention group demonstrated significant improvement in headache symptoms compared with veterans in the control group on the HIT-6 at the end-of-treatment (P < .001, d = 1.53) and 2-month follow-up assessment (P < .001, d = 1.14). This pattern also was consistent with the TBI-QOL Headache Pain item short form, with veterans in the intervention group showing improvement beyond those in the control group at the end-of-treatment (P < .001, d = 0.89) and 2-month follow-up assessment (P < .001, d = 0.83). Two-sample t tests also demonstrated significant improvement in subjective reports of sleep; those in the intervention group had significantly lower scores on the ISI at the end-of-study (P < .001, d = 1.53) and 2-month follow-up assessment (P < .001, d = 0.97). This pattern also held true for the Neuro-QOL Sleep Disturbance short form subtest, which demonstrated significantly more improvement in the intervention group compared with the control group at the end-of-study (P < .001, d = 0.97) and 2-month follow- up assessment (P < .001, d = 0.92). improvement in attention was not supported by the present results. A 2-sample t test found no significant difference between performance on the QIKtest for veterans in the intervention group vs the control group at the end-of-study (P = .40, d = 0.19) or the 2-month follow-up (P = .43, d = 0.20) (eAppendix 3).

Secondary Variables of Interest Analysis

Secondary variables examined differences in QOL, PTSD, depressive symptoms, and general symptoms reported between veterans in the intervention and control groups. Results demonstrated that veterans in the intervention group showed improvement above and beyond those in the control group on all measures. In regard to QOL, veterans in the intervention group had significantly higher scores on the Neuro-QOL Participate subtest than those in the control group at the end-of-study (P = .01, d = 0.89) and 2-month follow-up assessment (P < .001, d = 0.62). A similar pattern was found for the Neuro-QOL Satisfaction subtest, with veterans in the intervention group showing significantly higher scores than those in the control group at the end-of-study (P = .001, d = 0.95) and 2-month follow-up assessment (P < .001, d = 0.62). This also held true on the QOLIBRI, with veterans in the intervention group demonstrating significantly higher scores than those in the control group at the end-of-study (P = .001, d = 0.92) and 2-month follow-up assessment (P < .001, d = 0.66).

Veterans in the intervention group had significantly lower scores on the PCL-5 than those in the control group at the end-of- study (P = .003, d = 0.78) and 2-month follow-up assessment (P = .001, d = 0.72). Veterans in the intervention group also had significantly lower scores on the PHQ-9 than those in the control group at the end-of-study (P < .001, d = 0.98) and 2-month follow-up assessment (P < .001, d = 0.83). Veterans in the intervention group had significantly lower scores on the DASS- 21 than those in the control group at the end-of-study (P = .002, d = 0.80) and 2-month follow-up assessment (P = .001, d = 0.77). They also had significantly lower scores on the General Symptom Inventory than those in the control group at the end-of-study (P = .02, d = 0.75) and 2-month follow-up assessment (P = .002, d = 0.57). A clinically significant shift of score occurred for each of the measures except DASS-21 (eAppendix 3). eAppendix 4 depicts the change in scores for the intervention group at the end of treatment and the clinically significant shift score of each measure.

Discussion

The results of this RCT revealed a promising impact of ILF NFB on the commonly experienced persistent PCSs of headaches and disrupted sleep. Veterans in the intervention group demonstrated statistically significant improvement in headache symptoms compared with veterans in the control group when assessed at the end of treatment and during a 2-month follow-up. The statistical significance of these improvements was also supported by large or very large effect sizes. In addition to these primary variables of interest, veterans in the intervention group notably demonstrated significant improvement compared with those in the control group in a number of secondary clinical measures, including QOL, traumatic stress-related symptoms, depressive symptoms, and general symptom report. The clinical impact was further supported by the clinically relevant shift in scores in the intervention group.

The data did not support the hypothesis that attention concerns would show significant improvement following ILF NFB. Performance on an attention measure did not differ significantly between groups at either the end-of-treatment or 2-month follow up assessment. The QIKtest, a continuous performance test used to measure attention, was a go/no-go task and calculated based on a combination of various types of errors and outlier responses. The stimulus for this task is a series of computerized, blinking lights, for which participants are tasked with discriminating targets and nontargets under time pressure. However, the order of the stimuli are consistent across administrations, rather than being randomized, introducing a potential confound of practice effects on this task since patients were administered the QIKtest 3 times in a 2-month period and again 2 months later. Veterans in the control group notably improved in their average performance of this task from baseline to the endpoint of their treatment participation and demonstrated further improvement at the 2-month follow-up assessment; this pattern would be consistent with potential practice effects and warrants caution in its interpretation for both groups.

Previously published ILF NFB clinical studies that used the QIKtest and found positive results were mostly conducted among children and teen populations across longer treatment periods. This research may indicate the QIKtest is not an appropriate measure to assess adults who have specialized training in responding to stimuli (ie, trained military personnel). This suggests the concept of attention dysfunction experienced by veterans and the best method to measure it may need to be explored further. This construct may not be related to the focus and skill in prolonged attention needed in selecting go/ no-go tasks, but rather related to a broader conceptual basis involving memory, recall, clarity of rational thought, and decision making impacted by the mTBI. For instance, a study among combat veterans with mTBI and PTSD found that performance on objective cognitive measures did not significantly correlate with their subjective reports of cognitive difficulties.⁵⁷ This reflects the pattern of the present study, in which subjective reports of attention improved over time on the clinical symptom checklist filled out by participants at each session, but the objective measure did not. The mean attention dysfunction score was 6 at session 1 and 1 to 2 at session 20 (lower scores are better on a 10-point scale).

Strengths and Limitations

This study presents results stemming from the first RCT examining clinical effectiveness of ILF NFB in a VA setting for veterans with diagnoses of mTBI. The study design shows promising external validity. Veterans were able to participate in a treatment consisting of 20 sessions over a period of typically 8 to 10 weeks, entailing 2 to 3 sessions per week, with an attrition of only 18% over the course of the study. Notably, attrition rates may have been impacted by the time course of the study, which was recruiting and running participants throughout the COVID-19 pandemic (March 2020 to May 2023). No attrition was due to the intervention itself, and no adverse reactions to ILF NFB were reported during the course of the study. Other strengths of the study include the ethnically and racially diverse participants, representative of the population of veterans in Hawaii. Additionally, all ILF NFB providers underwent supervised ILF NFB training and achieved a skill reliability index score of 0.95 prior to providing ILF NFB to the intervention group.

This study was not blinded. Neither veterans nor ILF NFB providers were blinded and were therefore aware of the randomly assigned groups. Research assistants administering the periodic assessments were meant to be blinded to condition by design; however, as the study progressed, a research assistant became unintentionally aware of each study participant's condition based on required documentation in the veteran’s health records; more notes were present for those in the intervention group (20 specialist notes) than the control group (8 notes). While the presence of a control group represents a strength relative to much of the existing ILF NFB literature, the control group in this case did not account for the total time spent with the researchers. Participants in the intervention group met with researchers for 20 total sessions as opposed to 8 telephone calls. Therefore, the study design cannot fully rule out the differential impact of demand characteristics between the 2 groups, nor can it fully address or rule out the impact of differential motivation and expectations between groups. There is also evidence that technological innovation can influence the expectations of research participants, meaning that the intervention group may have been unduly influenced by the novelty of the ILF NFB technology, to which the control group did not have exposure.⁵⁸

A second attention measure for this study would have been beneficial, perhaps in identifying true change in attention ability or providing more insight into finding better methods to assess attention among veterans with mTBI. ILF NFB demonstrated significant impact across multiple outcome measures of clinical relevance for veterans diagnosed with mTBI, including the primary outcome variables of headache and sleep. The strength of the improvements seen in these areas, supported by large practical effects as well as veterans’ subjective reports, indicates much promise. Follow-up studies may also focus on the potential effectiveness of ILF NFB as a treatment of the secondary concerns measured in this study, including traumatic stress-related and depressive symptoms, and may explore the added benefit, if any, of ILF NFB alongside other evidence-based treatments for traumatic stress-related and mood disorders (eg, cognitive behavioral therapy). Using functional magnetic resonance imaging before and after assessments to determine actual brain enhancement with ILF NFB for certain disorders in which a brain signature exists (ie, migraine) should be explored. Further examination of ILF NFB as an intervention for attention may also be warranted, using more effective measures of attention in the population of veterans with mTBI, given the concerns noted earlier. Future research on this topic will need to clearly define attention in relation to the veteran experience and use relevant measures.

Conclusions

This study supports ILF NFB as a safe, noninvasive, nonpharmacologic treatment that may be effective in addressing the complex clinical concerns of veterans diagnosed with mTBI, a population for whom effective treatments have been difficult to identify. This intervention can provide veterans with a desirable and effective nonpharmacologic alternative in their care.

The Veterans Health Administration (VHA) is now in the sixth year of its enterprise-wide transformation into a high reliability organization (HRO). This effort began with a 2016 pilot project and is now implemented in 170 VHA medical centers.^1-4 This transformation reflects a commitment to implementing standardized and reliable health care practices.

The VHA HRO implementation strategy includes a multifaceted approach to engage leadership through education, training, leader coaching, and change management initiatives.² Despite the diversity of facilities in terms of cultures, geographies, and complexities, US Department of Veterans Affairs (VA) medical centers (VAMCs) have increasingly embraced standardized HRO practices. These changes are evident in improvements in VHA All Employee Survey scores, which assess 4 key patient safety culture dimensions: risk identification and just culture, error transparency and mitigation, supervisor communication and trust, and team cohesion and engagement.⁵ Positive trends in these dimensions highlight a cultural shift toward greater reliability, even amid challenges introduced by the COVID-19 pandemic.

However, this progress has encountered some challenges. Leadership turnover, budgetary constraints, and extensive educational demands for implementing and sustaining HRO practices have created obstacles, particularly for frontline health care practitioners.⁶ Additionally, there are pockets of resistance similar to what the airline industry faced when implementing crew resource management (CRM). Specifically, senior pilots and legacy leaders were reluctant to abandon their high-status, autocratic management styles and embrace CRM, despite its proven benefits for enhancing commercial airline safety.⁷ Similarly, some VHA staff have expressed resistance to foundational HRO practices, which include safety huddles, safety forums, leader rounding, and visual management systems.^6,8

The training requirements for HRO practices range from a 25-minute introductory course (HRO 101) to a 7.5-hour team training session facilitated by the VHA National Center for Patient Safety (NCPS).⁹ While some supervisors view these requirements as burdensome, others have demonstrated strong enthusiasm for the process.⁶ Understanding the perspectives of unit and departmental managers regarding factors that enhance or hinder the adoption of HRO practices is critical for continuous improvement.^10-12 Research has suggested that fostering psychological safety can create an environment where new ideas are shared openly, helping organizations navigate resistance to change.^13-16

A 2024 quality improvement study, drawing on the perceptions of HRO leads, identified key facilitators, including training, coaching, leader approachability, and psychological safety, as well as barriers such as inadequate training and lack of accountability among managers.¹⁷ Building on this work, the current study focused on frontline supervisors, who are directly involved in integrating HRO practices into patient care activities. By addressing both barriers and facilitators, this expanded approach aims to provide a more comprehensive understanding of the challenges influencing HRO implementation in day-to-day operations.

Methods

This quality improvement initiative examined facilitators and barriers to establishing just culture and implementing high reliability practices, focusing on frontline supervisors overseeing clinical care teams (eg, emergency department, critical care) or patient-support functions (eg, dietary services). A questionnaire was sent to a randomized sample of VHA facility supervisors.

A qualitative grounded theory approach was employed to provide a deeper understanding of nuanced phenomena that cannot be captured through numerical data alone. This method enables systematic analysis using open, axial, and thematic coding, ensuring that emerging themes achieve saturation.^18,19 It is particularly suited for this study, given the limited prior data on frontline supervisors. Additionally, qualitative methods help mitigate biases common in Likert-style scales, where respondents may lean toward agreement, potentially skewing results.²⁰

Inclusion Criteria

Participants were required to have served as a frontline supervisor for ≥ 6 months. Frontline supervisors are assigned responsibility for supporting staff who deliver services to VHA patients, including clinical care, dietary support, and other functions. These staff must complete baseline HRO cultural training as well as NCPS team training and are responsible for supporting quality, safety, and patient experience. Potential participants were identified from a list of frontline supervisors provided by VHA management. A subset was chosen through random sampling across geographically distributed VHA hospital facilities that vary in size and complexity. Invitations to participate in completing the questionnaire were sent via email, explaining the quality improvement initiative’s purpose, and encouraging voluntary participation. Of 2000 frontline supervisors invited to participate in the initiative, 97 completed the questionnaire. Participants represented a mix of VHA sites in terms of geography, size, and complexity.

Procedures

The authors used a qualitative approach and administered a confidential online survey. Demographic data were collected within the survey to understand characteristics of the participant population, including length of time as a frontline supervisor, facility complexity level, and professional background (clinical vs nonclinical). Survey questions were developed to elicit responses to specific areas of interest based on existing literature related to HRO and just culture.

Facilitators were defined as factors that increase the likelihood of establishing or sustaining high reliability practices and/or culture. Barriers were defined as factors that decrease or inhibit the likelihood of establishing or sustaining high reliability practices and/or culture. The questionnaire consisted of open-ended questions asking frontline supervisors to describe HRO practices and just culture at their individual facility and within their role. Participants also were asked to identify facilitators and barriers that helped or hindered their efforts to establish and maintain high reliability practices and just culture. The questionnaire solicited recommendations for additional support, training, resources, or leadership interventions to strengthen high reliability practices and just culture from each participant.

Analysis

Participant characteristics were analyzed using descriptive statistics. Responses to the 7 open-ended questions were coded and analyzed using ATLAS.ti v.24 qualitative data analysis software by an experienced researcher and coauthor. Grounded theory methodology allowed themes to emerge from the data and although the response rate was limited, the themes reached a saturation point.^18,19

Ethical Considerations

Institutional review board (IRB) review and approval were not required for this quality improvement initiative. Formal IRB review and approval of a quality improvement initiative are not required by VHA. Participation was confidential and voluntary, and participants could withdraw at any time without consequences. Completion of the survey indicated consent, and facility names and participant identifiers were not used. Unique numbers were assigned to each participant and all responses were kept confidential and nonattributional. Frequency coding was used to identify the facilitators and barriers to high reliability practices implementation and just culture among frontline supervisors until thematic saturation was obtained.

Results

A total of 2000 frontline supervisors were invited to participate, of whom 97 completed the questionnaire (response rate, 4.9%). Participants were first asked to describe just culture and high reliability practices in their own words. The consensus was that a just culture emphasizes a nonpunitive environment where employees can report errors or incidents without fear of retaliation. It encourages accountability at the systems level, focusing on learning from mistakes to improve processes. In response to a question asking respondents to describe HRO practices and just culture in their own words, participants noted that organizations with a just culture promote open communication, allowing staff to discuss safety issues and concerns without fear of personal blame. Additionally, participants agreed that HRO practices were defined as a set of principles and practices aimed at minimizing errors and promoting safety, especially within complex and high-risk environments. Participants responded that key characteristics include a preoccupation with failure, sensitivity to operations, reluctance to simplify, and a commitment to resilience. HRO practices entail proactively identifying and mitigating risks through open communication and collaboration among team members, they added.

Overall, participants were aligned with their view of the role a frontline supervisor plays in supporting just culture and HRO principles at their facility by fostering open communication and psychological safety, encouraging continuous learning and improvement, and promoting team collaboration and shared accountability. Among frontline supervisors, 93 (96%) identified their role as being critical to creating a safe space and reinforcing just culture and HRO principles at their facility, while 4 (4%) failed to identify a single duty.

Identified Themes

Table 1 summarizes 6 key themes identified from participants’ responses, highlighting the most frequently cited facilitators and barriers to implementing and sustaining high reliability practices and a just culture. Table 2 shows the classification of several themes in relation to facility complexity, emphasizing leadership commitment and support as a pivotal facilitator, while listing resistance to change and entrenched attitudes as a prominent barrier.

Role of Leadership

Facilitators. Leadership commitment and support were the most frequently identified facilitator, accounting for 44 mentions (45%). This aligns with participants’ descriptions of leadership involvement as crucial, particularly in setting standards and fostering accountability throughout the organization. For example, a frontline supervisor with < 5 years of experience from a nonclinical background at a 1B facility remarked, “Facility leadership are involved, which trickles down to lower-level leads and supervisors, which keeps everyone accountable and holds everyone to the same standards.” Participants frequently identified that leaders setting the standard and communicating expectations as paramount facilitators for strengthening high reliability practices and just culture at their facility.

Barriers. A lack of leadership commitment and support was a significant barrier, cited in 17 responses (18%). Participants described this barrier as a deficiency in follow-through, transparency, and presence, which undermines efforts to sustain just culture and high reliability principles. Notably, the lack of leadership commitment and support stood out as a distinct and recurring theme, underscoring its critical role as an independent challenge to achieving organizational goals. “Many leaders are not yet fully bought in,” a respondent explained. “They take the training and pass it off and go right back to their units and focus on blaming or chastis[ing] people for speaking up.” This theme frequently intersected with mentions of insufficient resources and entrenched attitudes, amplifying other challenges.

Open Communication and Transparency

Facilitators. Open communication and transparency were identified as facilitators in 12 responses (12%). Participants emphasized the importance of mechanisms such as HRO meetings and the sharing of “real” examples of positive outcomes from applying HRO principles. Transparent communication fosters psychological safety, allowing staff to report concerns without fear of reprisal. One participant with < 5 years of experience from a clinical background at a 1A facility encapsulated this theme by saying, “Quarterly ‘fireside chats’ are helpful, [this] creates open dialogue about where the next safety issue may occur, what staff need to do their job safely, while also imparting more of the philosophy of HRO that staff may not be aware of.”

Barriers. While communication serves as a facilitator, participants also highlighted barriers such as siloed communication and fear of reprisal. These reflect challenges in creating open and transparent feedback loops essential to high reliability. For example, a participant concluded, “Leadership does not communicate problem-solving efforts and resolution down the chain, they do not see the problems.” Another participant added, “[HRO principles] are not discussed that much.” While this theme presented as a barrier, it was noted less frequently.

Education and Training

Facilitators. Education and training were noted as facilitators in 10 responses (10%), underscoring their role in establishing high reliability practices. Participants suggested tailored training, simulation-based exercises, and mentorship to enhance practical application. However, they noted the importance of linking training to real change and ensuring leadership enforcement of learned behaviors. This theme is best represented by a participant who concluded, “Trainings have helped, but I think as a supervisor, being involved and interacting with your staff, observing, doing the work they do to help identify potential problems areas, especially when new systems are introduced are key. Being hands-on is the only way to successfully manage your team.”

Barriers. Insufficient resources, including time and staffing constraints, were identified as barriers to education and training, accounting for 24 responses (25%). Participants observed that mandatory training without mentorship or application diminishes its effectiveness.

Insufficient Resources and Funding

Barriers. Resource constraints, including low staffing levels and budget cuts, accounted for 24 responses (25%). Participants reported these limitations prevented staff from attending training and affected the overall implementation of just culture and HRO principles. “Low staffing in supporting services as well as in my own service line have created barriers,” a participant reported. Another participant responded that barriers to HRO were primarily “…financial, as the focus is how to curb costs and bring in more funding rather than taking the time to review and apply the concepts of high reliability.”

Resistance to Change and Entrenched Attitudes

Barriers. Resistance to change was the most frequently identified barrier, with 31 responses (32%). One participant described a persistent “gotcha” culture, where blame and punishment hinder progress toward just culture. This entrenched mindset creates significant obstacles to adopting HRO practices and requires active leadership and supervisor intervention to overcome. This theme is best captured by a respondent who noted that “culture change is difficult, especially among staff with such long tenure. It’s a long game.”

Synthesis and Integration of Findings

The data in Table 1 and Table 2 reinforce the themes identified in the qualitative analysis. Leadership commitment and support are pivotal, both as a facilitator and barrier. Open communication and education and training, while recognized as facilitators, were less frequently mentioned, but still critical. Resistance to change and insufficient resources were the most prominent barriers, indicating where organizational efforts should focus to further foster a culture of high reliability.

By addressing these barriers, particularly resistance to change and resource constraints, and leveraging facilitators like leadership engagement and transparent communication, organizations can enhance their implementation of just culture and high reliability practices. These efforts require deliberate strategies, including effective training, mentorship, and the active presence of leadership.

Discussion

This quality improvement initiative builds on prior research by examining the implementation of HRO practices from the perspective of frontline supervisors. Unlike earlier research focused on HRO leads, this study explores the critical role of supervisors who integrate HRO principles into clinical and administrative operations.¹⁷ By analyzing their experiences, this study offers practical insights into facilitating HRO implementation across organizational levels.

The findings highlight broad agreement on the value of just culture and HRO principles in fostering safe, accountable health care environments. Participants described just culture as promoting system—level accountability rather than individual blame, encouraging error reporting and learning for continuous improvement. Similarly, HRO practices—emphasizing a preoccupation with failure, operational sensitivity, and resilience— were seen as vital for patient safety in complex settings.

Frontline supervisors play a pivotal role, with 96% of respondents identifying their influence on fostering open communication, psychological safety, and shared accountability. Key facilitators included leadership commitment, open communication, and mentorship. Active leadership involvement was particularly valued, as it trickles down to reinforce standards across all organizational levels. HRO meetings using real-world examples were seen as instrumental in demonstrating the tangible benefits of these principles, helping embed them into daily practices.

Despite these facilitators, several barriers to implementation were noted. Resistance to change and entrenched attitudes, and a persistent gotcha culture undermined efforts to establish just culture. Resource constraints, including staffing shortages and budget limitations, further hindered the adoption of HRO practices. The lack of consistent leadership engagement, marked by limited visibility, follow-through, and transparency, exacerbated these challenges.

HRO leads are important for promoting education and embedding HRO principles into daily operations. These individuals provide vital support to frontline supervisors, translating HRO concepts into actionable practices. However, organizational challenges such as staff turnover and redirected funding have weakened the infrastructure supporting HRO initiatives. The elimination of HRO lead roles due to budgetary pressures at several facilities reflects a short-term focus on operational demands at the expense of long-term cultural transformation.

Additional barriers included siloed communication, fear of reprisal, bureaucratic obstacles, and outdated technology. These challenges limit progress toward high reliability and diminish the effectiveness of HRO principles.

Participants proposed strategies focused on education, training, and leadership engagement. Simulation-based training tailored to specific roles was identified as an effective tool for preparing staff to apply HRO principles in real-world scenarios. Enhanced communication, such as regular leadership rounding and transparent updates on safety concerns, was also emphasized. Participants stressed the importance of showing staff how their feedback influences organizational decisions to build trust and accountability. Finally, standardizing procedures and protocols across facilities was seen as critical for aligning practices and reducing variability in safety processes.

This study underscores the need for sustained leadership commitment and infrastructure to ensure the long-term success of HRO implementation. Addressing the identified barriers and leveraging the proposed mitigation strategies can foster a culture of safety and reliability across the organization.

Limitations

This quality improvement initiative used qualitative grounded theory methods and sampled a relatively small group of experienced leaders specifically involved in implementing HRO within the VHA. In addition, while saturation of themes was reached, the number of responses represents a small sample of VHA frontline supervisors. As such, the findings may not be fully representative of the perspectives of all unit and departmental leaders across the VHA or other health care systems. A previous qualitative quality improvement initiative focused on the perceptions of HRO leads regarding facilitators and barriers to just culture.¹⁷ This quality improvement initiative broadened that focus by examining the perspectives of frontline supervisors in the operational environment, who may not be HRO experts but work to implement HRO principles with the guidance of HRO leads (HRO subject matter experts).

There remains an opportunity to address a critical gap by assessing facilitators and barriers beyond the facility level, incorporating both the Veterans Integrated Service Networks (VISN) and VHA Central Office (VHACO). While qualitative methods, such as those used in this study, provide deep insights and detailed understanding, they are limited in their ability to identify system-wide trends and variations at a more strategic VISN and VHACO level. Addressing this could enhance the broader applicability of HRO principles across the VHA.

Conclusions

Successful implementation of the recommendations reported in this study will require sustained focus and continued commitment from all stakeholders across the VHA. As the VHA enters its seventh year on the HRO journey, the risk of organizational drift remains an ongoing concern. Progress has been made, as evidenced by incremental improvements in All Employee Survey scores and increased reporting of adverse events and near misses, but the challenge will be to maintain focus and continue to build upon progress amid the current climate of budgetary constraints.

This study builds on previous quality improvement efforts and provides valuable insights into the barriers and facilitators that can either hinder or support the VHA’s ongoing pursuit of high reliability. The findings offer a model for understanding the complexities of this journey—one that requires continuous effort and adaptation, as there is no definitive endpoint in the quest for high reliability.

Since completion of this study in 2024, the VHA has entered a period of organizational transition and restructuring. Such transitions are often accompanied by increased operational demands and organizational strain. These include realignments, personnel changes, staffing adjustments, workforce reductions, and continued implementation of a new electronic health record system. In this context, maintaining attention to culture, communication, frontline engagement, and mechanisms that provide visibility into organizational climate is essential to sustain momentum in high-reliability efforts.

The Veterans Health Administration (VHA) is now in the sixth year of its enterprise-wide transformation into a high reliability organization (HRO). This effort began with a 2016 pilot project and is now implemented in 170 VHA medical centers.^1-4 This transformation reflects a commitment to implementing standardized and reliable health care practices.

The VHA HRO implementation strategy includes a multifaceted approach to engage leadership through education, training, leader coaching, and change management initiatives.² Despite the diversity of facilities in terms of cultures, geographies, and complexities, US Department of Veterans Affairs (VA) medical centers (VAMCs) have increasingly embraced standardized HRO practices. These changes are evident in improvements in VHA All Employee Survey scores, which assess 4 key patient safety culture dimensions: risk identification and just culture, error transparency and mitigation, supervisor communication and trust, and team cohesion and engagement.⁵ Positive trends in these dimensions highlight a cultural shift toward greater reliability, even amid challenges introduced by the COVID-19 pandemic.

However, this progress has encountered some challenges. Leadership turnover, budgetary constraints, and extensive educational demands for implementing and sustaining HRO practices have created obstacles, particularly for frontline health care practitioners.⁶ Additionally, there are pockets of resistance similar to what the airline industry faced when implementing crew resource management (CRM). Specifically, senior pilots and legacy leaders were reluctant to abandon their high-status, autocratic management styles and embrace CRM, despite its proven benefits for enhancing commercial airline safety.⁷ Similarly, some VHA staff have expressed resistance to foundational HRO practices, which include safety huddles, safety forums, leader rounding, and visual management systems.^6,8

The training requirements for HRO practices range from a 25-minute introductory course (HRO 101) to a 7.5-hour team training session facilitated by the VHA National Center for Patient Safety (NCPS).⁹ While some supervisors view these requirements as burdensome, others have demonstrated strong enthusiasm for the process.⁶ Understanding the perspectives of unit and departmental managers regarding factors that enhance or hinder the adoption of HRO practices is critical for continuous improvement.^10-12 Research has suggested that fostering psychological safety can create an environment where new ideas are shared openly, helping organizations navigate resistance to change.^13-16

A 2024 quality improvement study, drawing on the perceptions of HRO leads, identified key facilitators, including training, coaching, leader approachability, and psychological safety, as well as barriers such as inadequate training and lack of accountability among managers.¹⁷ Building on this work, the current study focused on frontline supervisors, who are directly involved in integrating HRO practices into patient care activities. By addressing both barriers and facilitators, this expanded approach aims to provide a more comprehensive understanding of the challenges influencing HRO implementation in day-to-day operations.

Methods

This quality improvement initiative examined facilitators and barriers to establishing just culture and implementing high reliability practices, focusing on frontline supervisors overseeing clinical care teams (eg, emergency department, critical care) or patient-support functions (eg, dietary services). A questionnaire was sent to a randomized sample of VHA facility supervisors.

A qualitative grounded theory approach was employed to provide a deeper understanding of nuanced phenomena that cannot be captured through numerical data alone. This method enables systematic analysis using open, axial, and thematic coding, ensuring that emerging themes achieve saturation.^18,19 It is particularly suited for this study, given the limited prior data on frontline supervisors. Additionally, qualitative methods help mitigate biases common in Likert-style scales, where respondents may lean toward agreement, potentially skewing results.²⁰

Inclusion Criteria

Participants were required to have served as a frontline supervisor for ≥ 6 months. Frontline supervisors are assigned responsibility for supporting staff who deliver services to VHA patients, including clinical care, dietary support, and other functions. These staff must complete baseline HRO cultural training as well as NCPS team training and are responsible for supporting quality, safety, and patient experience. Potential participants were identified from a list of frontline supervisors provided by VHA management. A subset was chosen through random sampling across geographically distributed VHA hospital facilities that vary in size and complexity. Invitations to participate in completing the questionnaire were sent via email, explaining the quality improvement initiative’s purpose, and encouraging voluntary participation. Of 2000 frontline supervisors invited to participate in the initiative, 97 completed the questionnaire. Participants represented a mix of VHA sites in terms of geography, size, and complexity.

Procedures

The authors used a qualitative approach and administered a confidential online survey. Demographic data were collected within the survey to understand characteristics of the participant population, including length of time as a frontline supervisor, facility complexity level, and professional background (clinical vs nonclinical). Survey questions were developed to elicit responses to specific areas of interest based on existing literature related to HRO and just culture.

Facilitators were defined as factors that increase the likelihood of establishing or sustaining high reliability practices and/or culture. Barriers were defined as factors that decrease or inhibit the likelihood of establishing or sustaining high reliability practices and/or culture. The questionnaire consisted of open-ended questions asking frontline supervisors to describe HRO practices and just culture at their individual facility and within their role. Participants also were asked to identify facilitators and barriers that helped or hindered their efforts to establish and maintain high reliability practices and just culture. The questionnaire solicited recommendations for additional support, training, resources, or leadership interventions to strengthen high reliability practices and just culture from each participant.

Analysis

Participant characteristics were analyzed using descriptive statistics. Responses to the 7 open-ended questions were coded and analyzed using ATLAS.ti v.24 qualitative data analysis software by an experienced researcher and coauthor. Grounded theory methodology allowed themes to emerge from the data and although the response rate was limited, the themes reached a saturation point.^18,19

Ethical Considerations

Institutional review board (IRB) review and approval were not required for this quality improvement initiative. Formal IRB review and approval of a quality improvement initiative are not required by VHA. Participation was confidential and voluntary, and participants could withdraw at any time without consequences. Completion of the survey indicated consent, and facility names and participant identifiers were not used. Unique numbers were assigned to each participant and all responses were kept confidential and nonattributional. Frequency coding was used to identify the facilitators and barriers to high reliability practices implementation and just culture among frontline supervisors until thematic saturation was obtained.

Results

A total of 2000 frontline supervisors were invited to participate, of whom 97 completed the questionnaire (response rate, 4.9%). Participants were first asked to describe just culture and high reliability practices in their own words. The consensus was that a just culture emphasizes a nonpunitive environment where employees can report errors or incidents without fear of retaliation. It encourages accountability at the systems level, focusing on learning from mistakes to improve processes. In response to a question asking respondents to describe HRO practices and just culture in their own words, participants noted that organizations with a just culture promote open communication, allowing staff to discuss safety issues and concerns without fear of personal blame. Additionally, participants agreed that HRO practices were defined as a set of principles and practices aimed at minimizing errors and promoting safety, especially within complex and high-risk environments. Participants responded that key characteristics include a preoccupation with failure, sensitivity to operations, reluctance to simplify, and a commitment to resilience. HRO practices entail proactively identifying and mitigating risks through open communication and collaboration among team members, they added.

Overall, participants were aligned with their view of the role a frontline supervisor plays in supporting just culture and HRO principles at their facility by fostering open communication and psychological safety, encouraging continuous learning and improvement, and promoting team collaboration and shared accountability. Among frontline supervisors, 93 (96%) identified their role as being critical to creating a safe space and reinforcing just culture and HRO principles at their facility, while 4 (4%) failed to identify a single duty.

Identified Themes

Table 1 summarizes 6 key themes identified from participants’ responses, highlighting the most frequently cited facilitators and barriers to implementing and sustaining high reliability practices and a just culture. Table 2 shows the classification of several themes in relation to facility complexity, emphasizing leadership commitment and support as a pivotal facilitator, while listing resistance to change and entrenched attitudes as a prominent barrier.

Role of Leadership

Facilitators. Leadership commitment and support were the most frequently identified facilitator, accounting for 44 mentions (45%). This aligns with participants’ descriptions of leadership involvement as crucial, particularly in setting standards and fostering accountability throughout the organization. For example, a frontline supervisor with < 5 years of experience from a nonclinical background at a 1B facility remarked, “Facility leadership are involved, which trickles down to lower-level leads and supervisors, which keeps everyone accountable and holds everyone to the same standards.” Participants frequently identified that leaders setting the standard and communicating expectations as paramount facilitators for strengthening high reliability practices and just culture at their facility.

Barriers. A lack of leadership commitment and support was a significant barrier, cited in 17 responses (18%). Participants described this barrier as a deficiency in follow-through, transparency, and presence, which undermines efforts to sustain just culture and high reliability principles. Notably, the lack of leadership commitment and support stood out as a distinct and recurring theme, underscoring its critical role as an independent challenge to achieving organizational goals. “Many leaders are not yet fully bought in,” a respondent explained. “They take the training and pass it off and go right back to their units and focus on blaming or chastis[ing] people for speaking up.” This theme frequently intersected with mentions of insufficient resources and entrenched attitudes, amplifying other challenges.

Open Communication and Transparency

Facilitators. Open communication and transparency were identified as facilitators in 12 responses (12%). Participants emphasized the importance of mechanisms such as HRO meetings and the sharing of “real” examples of positive outcomes from applying HRO principles. Transparent communication fosters psychological safety, allowing staff to report concerns without fear of reprisal. One participant with < 5 years of experience from a clinical background at a 1A facility encapsulated this theme by saying, “Quarterly ‘fireside chats’ are helpful, [this] creates open dialogue about where the next safety issue may occur, what staff need to do their job safely, while also imparting more of the philosophy of HRO that staff may not be aware of.”

Barriers. While communication serves as a facilitator, participants also highlighted barriers such as siloed communication and fear of reprisal. These reflect challenges in creating open and transparent feedback loops essential to high reliability. For example, a participant concluded, “Leadership does not communicate problem-solving efforts and resolution down the chain, they do not see the problems.” Another participant added, “[HRO principles] are not discussed that much.” While this theme presented as a barrier, it was noted less frequently.

Education and Training

Facilitators. Education and training were noted as facilitators in 10 responses (10%), underscoring their role in establishing high reliability practices. Participants suggested tailored training, simulation-based exercises, and mentorship to enhance practical application. However, they noted the importance of linking training to real change and ensuring leadership enforcement of learned behaviors. This theme is best represented by a participant who concluded, “Trainings have helped, but I think as a supervisor, being involved and interacting with your staff, observing, doing the work they do to help identify potential problems areas, especially when new systems are introduced are key. Being hands-on is the only way to successfully manage your team.”

Barriers. Insufficient resources, including time and staffing constraints, were identified as barriers to education and training, accounting for 24 responses (25%). Participants observed that mandatory training without mentorship or application diminishes its effectiveness.

Insufficient Resources and Funding

Barriers. Resource constraints, including low staffing levels and budget cuts, accounted for 24 responses (25%). Participants reported these limitations prevented staff from attending training and affected the overall implementation of just culture and HRO principles. “Low staffing in supporting services as well as in my own service line have created barriers,” a participant reported. Another participant responded that barriers to HRO were primarily “…financial, as the focus is how to curb costs and bring in more funding rather than taking the time to review and apply the concepts of high reliability.”

Resistance to Change and Entrenched Attitudes

Barriers. Resistance to change was the most frequently identified barrier, with 31 responses (32%). One participant described a persistent “gotcha” culture, where blame and punishment hinder progress toward just culture. This entrenched mindset creates significant obstacles to adopting HRO practices and requires active leadership and supervisor intervention to overcome. This theme is best captured by a respondent who noted that “culture change is difficult, especially among staff with such long tenure. It’s a long game.”

Synthesis and Integration of Findings

The data in Table 1 and Table 2 reinforce the themes identified in the qualitative analysis. Leadership commitment and support are pivotal, both as a facilitator and barrier. Open communication and education and training, while recognized as facilitators, were less frequently mentioned, but still critical. Resistance to change and insufficient resources were the most prominent barriers, indicating where organizational efforts should focus to further foster a culture of high reliability.

By addressing these barriers, particularly resistance to change and resource constraints, and leveraging facilitators like leadership engagement and transparent communication, organizations can enhance their implementation of just culture and high reliability practices. These efforts require deliberate strategies, including effective training, mentorship, and the active presence of leadership.

Discussion

This quality improvement initiative builds on prior research by examining the implementation of HRO practices from the perspective of frontline supervisors. Unlike earlier research focused on HRO leads, this study explores the critical role of supervisors who integrate HRO principles into clinical and administrative operations.¹⁷ By analyzing their experiences, this study offers practical insights into facilitating HRO implementation across organizational levels.

The findings highlight broad agreement on the value of just culture and HRO principles in fostering safe, accountable health care environments. Participants described just culture as promoting system—level accountability rather than individual blame, encouraging error reporting and learning for continuous improvement. Similarly, HRO practices—emphasizing a preoccupation with failure, operational sensitivity, and resilience— were seen as vital for patient safety in complex settings.

Frontline supervisors play a pivotal role, with 96% of respondents identifying their influence on fostering open communication, psychological safety, and shared accountability. Key facilitators included leadership commitment, open communication, and mentorship. Active leadership involvement was particularly valued, as it trickles down to reinforce standards across all organizational levels. HRO meetings using real-world examples were seen as instrumental in demonstrating the tangible benefits of these principles, helping embed them into daily practices.

Despite these facilitators, several barriers to implementation were noted. Resistance to change and entrenched attitudes, and a persistent gotcha culture undermined efforts to establish just culture. Resource constraints, including staffing shortages and budget limitations, further hindered the adoption of HRO practices. The lack of consistent leadership engagement, marked by limited visibility, follow-through, and transparency, exacerbated these challenges.

HRO leads are important for promoting education and embedding HRO principles into daily operations. These individuals provide vital support to frontline supervisors, translating HRO concepts into actionable practices. However, organizational challenges such as staff turnover and redirected funding have weakened the infrastructure supporting HRO initiatives. The elimination of HRO lead roles due to budgetary pressures at several facilities reflects a short-term focus on operational demands at the expense of long-term cultural transformation.

Additional barriers included siloed communication, fear of reprisal, bureaucratic obstacles, and outdated technology. These challenges limit progress toward high reliability and diminish the effectiveness of HRO principles.

Participants proposed strategies focused on education, training, and leadership engagement. Simulation-based training tailored to specific roles was identified as an effective tool for preparing staff to apply HRO principles in real-world scenarios. Enhanced communication, such as regular leadership rounding and transparent updates on safety concerns, was also emphasized. Participants stressed the importance of showing staff how their feedback influences organizational decisions to build trust and accountability. Finally, standardizing procedures and protocols across facilities was seen as critical for aligning practices and reducing variability in safety processes.

This study underscores the need for sustained leadership commitment and infrastructure to ensure the long-term success of HRO implementation. Addressing the identified barriers and leveraging the proposed mitigation strategies can foster a culture of safety and reliability across the organization.

Limitations

This quality improvement initiative used qualitative grounded theory methods and sampled a relatively small group of experienced leaders specifically involved in implementing HRO within the VHA. In addition, while saturation of themes was reached, the number of responses represents a small sample of VHA frontline supervisors. As such, the findings may not be fully representative of the perspectives of all unit and departmental leaders across the VHA or other health care systems. A previous qualitative quality improvement initiative focused on the perceptions of HRO leads regarding facilitators and barriers to just culture.¹⁷ This quality improvement initiative broadened that focus by examining the perspectives of frontline supervisors in the operational environment, who may not be HRO experts but work to implement HRO principles with the guidance of HRO leads (HRO subject matter experts).

There remains an opportunity to address a critical gap by assessing facilitators and barriers beyond the facility level, incorporating both the Veterans Integrated Service Networks (VISN) and VHA Central Office (VHACO). While qualitative methods, such as those used in this study, provide deep insights and detailed understanding, they are limited in their ability to identify system-wide trends and variations at a more strategic VISN and VHACO level. Addressing this could enhance the broader applicability of HRO principles across the VHA.

Conclusions

Successful implementation of the recommendations reported in this study will require sustained focus and continued commitment from all stakeholders across the VHA. As the VHA enters its seventh year on the HRO journey, the risk of organizational drift remains an ongoing concern. Progress has been made, as evidenced by incremental improvements in All Employee Survey scores and increased reporting of adverse events and near misses, but the challenge will be to maintain focus and continue to build upon progress amid the current climate of budgetary constraints.

This study builds on previous quality improvement efforts and provides valuable insights into the barriers and facilitators that can either hinder or support the VHA’s ongoing pursuit of high reliability. The findings offer a model for understanding the complexities of this journey—one that requires continuous effort and adaptation, as there is no definitive endpoint in the quest for high reliability.

Since completion of this study in 2024, the VHA has entered a period of organizational transition and restructuring. Such transitions are often accompanied by increased operational demands and organizational strain. These include realignments, personnel changes, staffing adjustments, workforce reductions, and continued implementation of a new electronic health record system. In this context, maintaining attention to culture, communication, frontline engagement, and mechanisms that provide visibility into organizational climate is essential to sustain momentum in high-reliability efforts.

The Veterans Health Administration (VHA) is now in the sixth year of its enterprise-wide transformation into a high reliability organization (HRO). This effort began with a 2016 pilot project and is now implemented in 170 VHA medical centers.^1-4 This transformation reflects a commitment to implementing standardized and reliable health care practices.

The VHA HRO implementation strategy includes a multifaceted approach to engage leadership through education, training, leader coaching, and change management initiatives.² Despite the diversity of facilities in terms of cultures, geographies, and complexities, US Department of Veterans Affairs (VA) medical centers (VAMCs) have increasingly embraced standardized HRO practices. These changes are evident in improvements in VHA All Employee Survey scores, which assess 4 key patient safety culture dimensions: risk identification and just culture, error transparency and mitigation, supervisor communication and trust, and team cohesion and engagement.⁵ Positive trends in these dimensions highlight a cultural shift toward greater reliability, even amid challenges introduced by the COVID-19 pandemic.

However, this progress has encountered some challenges. Leadership turnover, budgetary constraints, and extensive educational demands for implementing and sustaining HRO practices have created obstacles, particularly for frontline health care practitioners.⁶ Additionally, there are pockets of resistance similar to what the airline industry faced when implementing crew resource management (CRM). Specifically, senior pilots and legacy leaders were reluctant to abandon their high-status, autocratic management styles and embrace CRM, despite its proven benefits for enhancing commercial airline safety.⁷ Similarly, some VHA staff have expressed resistance to foundational HRO practices, which include safety huddles, safety forums, leader rounding, and visual management systems.^6,8

The training requirements for HRO practices range from a 25-minute introductory course (HRO 101) to a 7.5-hour team training session facilitated by the VHA National Center for Patient Safety (NCPS).⁹ While some supervisors view these requirements as burdensome, others have demonstrated strong enthusiasm for the process.⁶ Understanding the perspectives of unit and departmental managers regarding factors that enhance or hinder the adoption of HRO practices is critical for continuous improvement.^10-12 Research has suggested that fostering psychological safety can create an environment where new ideas are shared openly, helping organizations navigate resistance to change.^13-16

A 2024 quality improvement study, drawing on the perceptions of HRO leads, identified key facilitators, including training, coaching, leader approachability, and psychological safety, as well as barriers such as inadequate training and lack of accountability among managers.¹⁷ Building on this work, the current study focused on frontline supervisors, who are directly involved in integrating HRO practices into patient care activities. By addressing both barriers and facilitators, this expanded approach aims to provide a more comprehensive understanding of the challenges influencing HRO implementation in day-to-day operations.

Methods

This quality improvement initiative examined facilitators and barriers to establishing just culture and implementing high reliability practices, focusing on frontline supervisors overseeing clinical care teams (eg, emergency department, critical care) or patient-support functions (eg, dietary services). A questionnaire was sent to a randomized sample of VHA facility supervisors.

A qualitative grounded theory approach was employed to provide a deeper understanding of nuanced phenomena that cannot be captured through numerical data alone. This method enables systematic analysis using open, axial, and thematic coding, ensuring that emerging themes achieve saturation.^18,19 It is particularly suited for this study, given the limited prior data on frontline supervisors. Additionally, qualitative methods help mitigate biases common in Likert-style scales, where respondents may lean toward agreement, potentially skewing results.²⁰

Inclusion Criteria

Participants were required to have served as a frontline supervisor for ≥ 6 months. Frontline supervisors are assigned responsibility for supporting staff who deliver services to VHA patients, including clinical care, dietary support, and other functions. These staff must complete baseline HRO cultural training as well as NCPS team training and are responsible for supporting quality, safety, and patient experience. Potential participants were identified from a list of frontline supervisors provided by VHA management. A subset was chosen through random sampling across geographically distributed VHA hospital facilities that vary in size and complexity. Invitations to participate in completing the questionnaire were sent via email, explaining the quality improvement initiative’s purpose, and encouraging voluntary participation. Of 2000 frontline supervisors invited to participate in the initiative, 97 completed the questionnaire. Participants represented a mix of VHA sites in terms of geography, size, and complexity.

Procedures

The authors used a qualitative approach and administered a confidential online survey. Demographic data were collected within the survey to understand characteristics of the participant population, including length of time as a frontline supervisor, facility complexity level, and professional background (clinical vs nonclinical). Survey questions were developed to elicit responses to specific areas of interest based on existing literature related to HRO and just culture.

Facilitators were defined as factors that increase the likelihood of establishing or sustaining high reliability practices and/or culture. Barriers were defined as factors that decrease or inhibit the likelihood of establishing or sustaining high reliability practices and/or culture. The questionnaire consisted of open-ended questions asking frontline supervisors to describe HRO practices and just culture at their individual facility and within their role. Participants also were asked to identify facilitators and barriers that helped or hindered their efforts to establish and maintain high reliability practices and just culture. The questionnaire solicited recommendations for additional support, training, resources, or leadership interventions to strengthen high reliability practices and just culture from each participant.

Analysis

Participant characteristics were analyzed using descriptive statistics. Responses to the 7 open-ended questions were coded and analyzed using ATLAS.ti v.24 qualitative data analysis software by an experienced researcher and coauthor. Grounded theory methodology allowed themes to emerge from the data and although the response rate was limited, the themes reached a saturation point.^18,19

Ethical Considerations

Institutional review board (IRB) review and approval were not required for this quality improvement initiative. Formal IRB review and approval of a quality improvement initiative are not required by VHA. Participation was confidential and voluntary, and participants could withdraw at any time without consequences. Completion of the survey indicated consent, and facility names and participant identifiers were not used. Unique numbers were assigned to each participant and all responses were kept confidential and nonattributional. Frequency coding was used to identify the facilitators and barriers to high reliability practices implementation and just culture among frontline supervisors until thematic saturation was obtained.

Results

A total of 2000 frontline supervisors were invited to participate, of whom 97 completed the questionnaire (response rate, 4.9%). Participants were first asked to describe just culture and high reliability practices in their own words. The consensus was that a just culture emphasizes a nonpunitive environment where employees can report errors or incidents without fear of retaliation. It encourages accountability at the systems level, focusing on learning from mistakes to improve processes. In response to a question asking respondents to describe HRO practices and just culture in their own words, participants noted that organizations with a just culture promote open communication, allowing staff to discuss safety issues and concerns without fear of personal blame. Additionally, participants agreed that HRO practices were defined as a set of principles and practices aimed at minimizing errors and promoting safety, especially within complex and high-risk environments. Participants responded that key characteristics include a preoccupation with failure, sensitivity to operations, reluctance to simplify, and a commitment to resilience. HRO practices entail proactively identifying and mitigating risks through open communication and collaboration among team members, they added.

Overall, participants were aligned with their view of the role a frontline supervisor plays in supporting just culture and HRO principles at their facility by fostering open communication and psychological safety, encouraging continuous learning and improvement, and promoting team collaboration and shared accountability. Among frontline supervisors, 93 (96%) identified their role as being critical to creating a safe space and reinforcing just culture and HRO principles at their facility, while 4 (4%) failed to identify a single duty.

Identified Themes

Table 1 summarizes 6 key themes identified from participants’ responses, highlighting the most frequently cited facilitators and barriers to implementing and sustaining high reliability practices and a just culture. Table 2 shows the classification of several themes in relation to facility complexity, emphasizing leadership commitment and support as a pivotal facilitator, while listing resistance to change and entrenched attitudes as a prominent barrier.

Role of Leadership

Facilitators. Leadership commitment and support were the most frequently identified facilitator, accounting for 44 mentions (45%). This aligns with participants’ descriptions of leadership involvement as crucial, particularly in setting standards and fostering accountability throughout the organization. For example, a frontline supervisor with < 5 years of experience from a nonclinical background at a 1B facility remarked, “Facility leadership are involved, which trickles down to lower-level leads and supervisors, which keeps everyone accountable and holds everyone to the same standards.” Participants frequently identified that leaders setting the standard and communicating expectations as paramount facilitators for strengthening high reliability practices and just culture at their facility.

Barriers. A lack of leadership commitment and support was a significant barrier, cited in 17 responses (18%). Participants described this barrier as a deficiency in follow-through, transparency, and presence, which undermines efforts to sustain just culture and high reliability principles. Notably, the lack of leadership commitment and support stood out as a distinct and recurring theme, underscoring its critical role as an independent challenge to achieving organizational goals. “Many leaders are not yet fully bought in,” a respondent explained. “They take the training and pass it off and go right back to their units and focus on blaming or chastis[ing] people for speaking up.” This theme frequently intersected with mentions of insufficient resources and entrenched attitudes, amplifying other challenges.

Open Communication and Transparency

Facilitators. Open communication and transparency were identified as facilitators in 12 responses (12%). Participants emphasized the importance of mechanisms such as HRO meetings and the sharing of “real” examples of positive outcomes from applying HRO principles. Transparent communication fosters psychological safety, allowing staff to report concerns without fear of reprisal. One participant with < 5 years of experience from a clinical background at a 1A facility encapsulated this theme by saying, “Quarterly ‘fireside chats’ are helpful, [this] creates open dialogue about where the next safety issue may occur, what staff need to do their job safely, while also imparting more of the philosophy of HRO that staff may not be aware of.”

Barriers. While communication serves as a facilitator, participants also highlighted barriers such as siloed communication and fear of reprisal. These reflect challenges in creating open and transparent feedback loops essential to high reliability. For example, a participant concluded, “Leadership does not communicate problem-solving efforts and resolution down the chain, they do not see the problems.” Another participant added, “[HRO principles] are not discussed that much.” While this theme presented as a barrier, it was noted less frequently.

Education and Training

Facilitators. Education and training were noted as facilitators in 10 responses (10%), underscoring their role in establishing high reliability practices. Participants suggested tailored training, simulation-based exercises, and mentorship to enhance practical application. However, they noted the importance of linking training to real change and ensuring leadership enforcement of learned behaviors. This theme is best represented by a participant who concluded, “Trainings have helped, but I think as a supervisor, being involved and interacting with your staff, observing, doing the work they do to help identify potential problems areas, especially when new systems are introduced are key. Being hands-on is the only way to successfully manage your team.”

Barriers. Insufficient resources, including time and staffing constraints, were identified as barriers to education and training, accounting for 24 responses (25%). Participants observed that mandatory training without mentorship or application diminishes its effectiveness.

Insufficient Resources and Funding

Barriers. Resource constraints, including low staffing levels and budget cuts, accounted for 24 responses (25%). Participants reported these limitations prevented staff from attending training and affected the overall implementation of just culture and HRO principles. “Low staffing in supporting services as well as in my own service line have created barriers,” a participant reported. Another participant responded that barriers to HRO were primarily “…financial, as the focus is how to curb costs and bring in more funding rather than taking the time to review and apply the concepts of high reliability.”

Resistance to Change and Entrenched Attitudes

Barriers. Resistance to change was the most frequently identified barrier, with 31 responses (32%). One participant described a persistent “gotcha” culture, where blame and punishment hinder progress toward just culture. This entrenched mindset creates significant obstacles to adopting HRO practices and requires active leadership and supervisor intervention to overcome. This theme is best captured by a respondent who noted that “culture change is difficult, especially among staff with such long tenure. It’s a long game.”

Synthesis and Integration of Findings

The data in Table 1 and Table 2 reinforce the themes identified in the qualitative analysis. Leadership commitment and support are pivotal, both as a facilitator and barrier. Open communication and education and training, while recognized as facilitators, were less frequently mentioned, but still critical. Resistance to change and insufficient resources were the most prominent barriers, indicating where organizational efforts should focus to further foster a culture of high reliability.

By addressing these barriers, particularly resistance to change and resource constraints, and leveraging facilitators like leadership engagement and transparent communication, organizations can enhance their implementation of just culture and high reliability practices. These efforts require deliberate strategies, including effective training, mentorship, and the active presence of leadership.

Discussion

This quality improvement initiative builds on prior research by examining the implementation of HRO practices from the perspective of frontline supervisors. Unlike earlier research focused on HRO leads, this study explores the critical role of supervisors who integrate HRO principles into clinical and administrative operations.¹⁷ By analyzing their experiences, this study offers practical insights into facilitating HRO implementation across organizational levels.

The findings highlight broad agreement on the value of just culture and HRO principles in fostering safe, accountable health care environments. Participants described just culture as promoting system—level accountability rather than individual blame, encouraging error reporting and learning for continuous improvement. Similarly, HRO practices—emphasizing a preoccupation with failure, operational sensitivity, and resilience— were seen as vital for patient safety in complex settings.

Frontline supervisors play a pivotal role, with 96% of respondents identifying their influence on fostering open communication, psychological safety, and shared accountability. Key facilitators included leadership commitment, open communication, and mentorship. Active leadership involvement was particularly valued, as it trickles down to reinforce standards across all organizational levels. HRO meetings using real-world examples were seen as instrumental in demonstrating the tangible benefits of these principles, helping embed them into daily practices.

Despite these facilitators, several barriers to implementation were noted. Resistance to change and entrenched attitudes, and a persistent gotcha culture undermined efforts to establish just culture. Resource constraints, including staffing shortages and budget limitations, further hindered the adoption of HRO practices. The lack of consistent leadership engagement, marked by limited visibility, follow-through, and transparency, exacerbated these challenges.

HRO leads are important for promoting education and embedding HRO principles into daily operations. These individuals provide vital support to frontline supervisors, translating HRO concepts into actionable practices. However, organizational challenges such as staff turnover and redirected funding have weakened the infrastructure supporting HRO initiatives. The elimination of HRO lead roles due to budgetary pressures at several facilities reflects a short-term focus on operational demands at the expense of long-term cultural transformation.

Additional barriers included siloed communication, fear of reprisal, bureaucratic obstacles, and outdated technology. These challenges limit progress toward high reliability and diminish the effectiveness of HRO principles.

Participants proposed strategies focused on education, training, and leadership engagement. Simulation-based training tailored to specific roles was identified as an effective tool for preparing staff to apply HRO principles in real-world scenarios. Enhanced communication, such as regular leadership rounding and transparent updates on safety concerns, was also emphasized. Participants stressed the importance of showing staff how their feedback influences organizational decisions to build trust and accountability. Finally, standardizing procedures and protocols across facilities was seen as critical for aligning practices and reducing variability in safety processes.

This study underscores the need for sustained leadership commitment and infrastructure to ensure the long-term success of HRO implementation. Addressing the identified barriers and leveraging the proposed mitigation strategies can foster a culture of safety and reliability across the organization.

Limitations

This quality improvement initiative used qualitative grounded theory methods and sampled a relatively small group of experienced leaders specifically involved in implementing HRO within the VHA. In addition, while saturation of themes was reached, the number of responses represents a small sample of VHA frontline supervisors. As such, the findings may not be fully representative of the perspectives of all unit and departmental leaders across the VHA or other health care systems. A previous qualitative quality improvement initiative focused on the perceptions of HRO leads regarding facilitators and barriers to just culture.¹⁷ This quality improvement initiative broadened that focus by examining the perspectives of frontline supervisors in the operational environment, who may not be HRO experts but work to implement HRO principles with the guidance of HRO leads (HRO subject matter experts).

There remains an opportunity to address a critical gap by assessing facilitators and barriers beyond the facility level, incorporating both the Veterans Integrated Service Networks (VISN) and VHA Central Office (VHACO). While qualitative methods, such as those used in this study, provide deep insights and detailed understanding, they are limited in their ability to identify system-wide trends and variations at a more strategic VISN and VHACO level. Addressing this could enhance the broader applicability of HRO principles across the VHA.

Conclusions

Successful implementation of the recommendations reported in this study will require sustained focus and continued commitment from all stakeholders across the VHA. As the VHA enters its seventh year on the HRO journey, the risk of organizational drift remains an ongoing concern. Progress has been made, as evidenced by incremental improvements in All Employee Survey scores and increased reporting of adverse events and near misses, but the challenge will be to maintain focus and continue to build upon progress amid the current climate of budgetary constraints.

This study builds on previous quality improvement efforts and provides valuable insights into the barriers and facilitators that can either hinder or support the VHA’s ongoing pursuit of high reliability. The findings offer a model for understanding the complexities of this journey—one that requires continuous effort and adaptation, as there is no definitive endpoint in the quest for high reliability.

Since completion of this study in 2024, the VHA has entered a period of organizational transition and restructuring. Such transitions are often accompanied by increased operational demands and organizational strain. These include realignments, personnel changes, staffing adjustments, workforce reductions, and continued implementation of a new electronic health record system. In this context, maintaining attention to culture, communication, frontline engagement, and mechanisms that provide visibility into organizational climate is essential to sustain momentum in high-reliability efforts.

The Veterans Health Administration (VHA) is transitioning from its native electronic health record (EHR) Vista/Computerized Patient Record System to the commercial Cerner/Oracle Health EHR. Though this process was temporarily discontinued in April 2023 due to patient safety, usability, and reliability concerns, it resumed in April 2026. It was originally projected to cost $50 billion to implement. ^1-3 As of March 9, 2024, 6 sites had transitioned to the new EHR.² The transition is the largest of its kind in the US, offering an unparalleled opportunity to examine the effects of EHR transitions on an often overlooked part of the workforce: health professions trainees (HPTs).

HPTs serve a central role in VHA. About one-third of patients receive care directly from HPTs who make up about one-third of the VHA workforce. VHA trains > 60 clinical disciplines, comprising > 122,000 trainees annually.^4,5 A paucity of literature exists exploring the experiences of HPTs during EHR transitions, and many studies are often limited to single-site or small populations. HPTs face distinct challenges and needs during EHR transitions and are particularly vulnerable to their negative impacts on retention, clinical training, and efficiency and confidence in EHR use.^6-10 HPTs at VHA sites that have already transitioned to the Cerner/Oracle Health EHR identified many challenges, including significant delays in gaining EHR access, pervasive perceptions of poor training, concerns that EHR functionality issues limited patient care, and decreased ability to track clinical skill acquisition.⁶ These challenges may impact some HPTs more than others (eg, students on short rotations are affected more acutely by delayed EHR access and usage).

This quality improvement project evaluated HPT EHR transition experiences at the Captain James A. Lovell Federal Health Care Center (FHCC). This article contributes to the limited literature on HPT transition experiences, identifies opportunities to support HPTs, and informs broader efforts in teaching HPTs new technologies.

Methods

FHCC is jointly operated by the US Department of Defense and US Department of Veterans Affairs (VA). It treats 80,000 inpatient and outpatients annually. FHCC was the sixth VA facility to transition to the new EHR, which went live on March 9, 2024.^2,11 About 700 HPTs rotate through FHCC annually. HPTs were eligible for inclusion if they were present during the March 9 transition according to a VA Office of Academic Affiliations database. A total of 216 HPTs were identified for inclusion.

Preparations for the transition included scaling down operations (ie, blocking clinician schedules, not scheduling future appointments that may conflict with the transition, making decisions on new facility- and service-line workflows, required EHR training, and speaking with support staff, including VHA National EHR Modernization Supplemental Staffing Unit [NESSU]). This evaluation was designated nonresearch/quality improvement by the VA Bedford Healthcare System Institutional Review Board.

Surveys

Forty-seven interviews were conducted with HPTs, site leaders, and supervisors from January 2024 to June 2024 (Table 1). Participants were identified by service leads and recruited via email; snowball sampling identified additional participants.

The evaluation team developed semistructured interview guides using grounded probes based on a pilot evaluation and existing research on EHR transitions.¹² Questions focused on participant experiences preparing for the EHR transition, learning and using the site’s EHR, and the impact the transition had on clinical training experiences. Interviews were conducted at different times to capture the range of user experiences: 1 month prelaunch, 2 to 6 weeks postlaunch, and 2 months postlaunch. Interviewees were informed of participant rights and provided verbal consent.

HPTs present at FHCC at each survey’s release were emailed invitations and 2 reminders. The anonymous surveys took about 10 minutes to complete. Survey items queried HPTs about their experiences preparing to use the new EHR, perceptions of the current EHR (adapted from the System Usability Scale), satisfaction with VHA training, impact on clinical training, ability to work with preceptors and patients, and experiences with the VHA clinical learning environment (adapted from the VHA Learners Preceptor Survey).^13-15 Survey questions used a 5-point Likert response scale.

Analysis

Interviewers completed postinterview summaries for team debriefing and consensus building. Interviews were coded using a priori (from piloting evaluations and relevant literature) and emergent (refined and developed from data) codes. Deductive and inductive content analyses were conducted. ¹⁶ Deductive analysis used a priori categories (eg, care coordination, EHR training). Inductive content analysis consisted of open and unstructured coding, capturing data outside a priori categories. Emergent codes captured unidentified categories. Qualitative researchers met weekly to discuss data and reach consensus on interpretation.

Descriptive analysis was conducted using top-2 box scoring (proportion responding within the 2 most favorable responses [agree/ strongly agree]). Survey data were analyzed in SAS.¹⁷ The analysis used a merging approach on simultaneously collected qualitative and quantitative data to reach findings consensus.¹⁸

Researcher and research team decisions may shape the data collected due to prior assumptions and experience.¹⁹ This analysis attempted to integrate reflexivity practices to enhance awareness of the researchers’ assumptions and positionality, including by integrating intent collaborative conversing and memorandum writing into the processes.^20,21

Results

This analysis created a survey and fielded responses from HPTs present at FHCC across 3 time points (6 months prelaunch, 1 month prelaunch, and 2 months postlaunch), resulting in a total of 103 responses and an average response rate of 19.0% (Table 2). Six key findings were identified in analysis of responses: (1) critiques of transition management; (2) concerns with training; (3) hope about the EHR; (4) at-the-elbow support was essential; (5) HPTs adjusted to, and later preferred, the new EHR; and (6) transition impacted clinical training, but not overall career plans for HPTs. Findings are presented in this section, with illustrative quantitative data and qualitative data quotes available in the eAppendix.

Critiques of the Transition’s Management

While participants were aware of the transition to the new EHR, most felt they did not have enough information or time to prepare for it, indicating it was “too little, too late.” HPTs felt necessary workflow processes for Cerner/Oracle Health were not determined with enough time to learn them prior to transition. Supervisors shared that important workflow and onboarding decisions remained undecided mere weeks before the transition. Some service lines did not decrease patient loads until right before the transition, making it difficult to manage their schedules and resulting in insufficient time to learn the new EHR.

EHR Training Concerns

Overall, HPTs expressed low satisfaction with computer-based Training Management System (TMS) EHR training, believing it did not prepare them for the new EHR. The percentage of HPTs satisfied or very satisfied with the quality of TMS training was lower than that of instructor-based training pre- and posttransition, with 50% of 36 prelaunch respondents, and 43% of 29 postlaunch respondents expressing satisfaction with computer-based trainings (Figure 1). HPTs were dissatisfied with the training content. They felt it was too general and failed to teach basic tasks in the workflow for their service areas and roles, such as writing a note or order. Furthermore, poor content was exacerbated by poor and unengaging instruction, and HPTs were dissatisfied with the practice EHR used in training, which glitched frequently.

EHR Transition Optimism

Even though the transition was stressful, most HPTs hoped it would be a temporary disruption and that they would quickly adjust to the new EHR. Many participants expected that once they switched to the new EHR, they would pick it up quickly. In addition, many anticipated Cerner/Oracle Health would be better and easier to use in the long run.

At-The-Elbow Support Essential

VHA peer support with NESSU was highly valued among HPTs. NESSU staff were highly knowledgeable and could provide both broad and service-line-specific support. NESSU provided prompt answers to EHR questions. This was particularly critical as other forms of in-person support were often inaccessible or absent during the transition.

HPTs found facility support helpful: 85% of 36 respondents reported being satisfied/ very satisfied with support from supervisors and preceptors, and 84% of 36 respondents were satisfied/very satisfied with technical support from facility informatics staff pretransition (n = 36) (Figure 2). NESSU and supervisor support with daily workflows were particularly helpful, as pretransition training only provided a general introduction to the EHR.

HPTs Adjusted to and Later Preferred the New EHR

The EHR learning experience was intense but short, with many HPTs feeling able to use it only 2 to 4 weeks posttransition. Confidence grew as HPTs came to view Cerner/Oracle Health as a more integrated and intuitive system than the previous EHR. Most participants preferred the new EHR, even if they criticized some features (eg, no group documentation capabilities). Survey participants frequently rated Cerner/Oracle Health usability higher than the original. A total of 32% of 29 posttransition respondents agreed or strongly agreed that Cerner/Oracle Health helps prevent situations that can lead to patient safety risks—higher than pretransition rates. Additionally, fewer respondents found the new EHR unnecessarily complex or thought it contained too many alerts and flags compared to the original EHR (Figure 3).

Impact on Clinical Training, Not Career Plans

The extensive time and energy the transition demanded of HPTs caused stress and affected their clinical training. Many believed they would have learned more if their training had happened outside the transition.

Concerns that the transition affected learning were most acutely felt pretransition. HPTs reporting that EHR implementation positively affected their clinical education fell from 38% of 36 respondents 6 months pretransition to 19% of 29 respondents 1 month pretransition, but returned to 37% posttransition (Figure 4). However, some HPTs believed there was a silver lining: it provided a learning experience they otherwise would not have had.

HPTs who believed the transition positively impacted their likelihood of pursuing future career opportunities within the VHA rose to 33% of 29 respondents posttransition. Overall, Cerner/Oracle Health was characterized as a tool: something used in training, but not something that precluded wanting VHA careers or having meaningful experiences, such as caring for patients.

Discussion

This evaluation addressed an underexplored aspect of EHR transitions: their impact on HPTs. It identified HPT challenges, including dissatisfaction with poor transition preparation and EHR training experiences. Promising findings include positive experiences with transition support, EHR uptake, and overall positive educational experiences despite the transition’s disruption.

When EHR users, including HPTs, are dissatisfied with transition preparations, consequent stress can lead to undesired effects, including increased burnout, inappropriate EHR use, and low work satisfaction.^22-24 Negative EHR transition experiences shape HPTs’ subsequent EHR adoption, user satisfaction, as well as confidence and career intent.^3,25,26 Health systems have strong incentives to implement effective transition change management.

HPTs at previous VHA EHR transition sites reported significantly more disruption to their clinical training compared with HPTs at FHCC. Academic programs were shut down at the first transition site, and HPTs expressed decreased interest in VHA careers at another, even a year posttransition.^6,27 These findings are consistent with the limited literature on the adverse impacts that EHR transitions have on HPTs.^7,28

HPT retention is critical. VA is mandated to prepare the next generation of HPTs for its needs, and those of the nation. The VA relies heavily on HPT retention to recruit clinicians: > 65% of VHA physicians nationwide participated in VHA training programs prior to recruitment into staff positions.^5,29

VHA should invest in transition change management with demonstrated, positive impacts on HPTs, such as in-house support from clinicians. Previous research found that lack of support was a major source of stress and negative outcomes.^6,27 Consequently, supporting HPTs through EHR transitions directly contributes to the VHA’s ability to attract high-quality staff from its HPTs. The challenges and promising practices described in this analysis underscore the necessity of understanding how all EHR users are affected by transitions. What happens to them has direct implications for the VA mission to provide safe, efficient care, and its mandate to provide quality clinical training to HPTs.

These findings hold hopeful implications for supporting HPT EHR use, both during and outside EHR transitions. HPTs expressing that an EHR is only 1 part of their clinical training experience suggests that change management can improve EHR transitions. HPT learning can enhance known factors that are important for HPTs in clinical training, including the health care organization’s mission, caring for patients, and personal development.

Further investigations may engage HPTs at future VHA sites making the transition to the new EHR. One focus would involve applying a learning health systems framework to examine the nature of change management efforts—and their effects on HPT transition experiences—iteratively across transition sites to evaluate the effect of the efforts. Another focus may be longitudinal engagement with HPTs at health care systems and sites transitioning to new EHRs. Research has found that disruptions to EHR usability, satisfaction, and care provision can persist for 2 years and beyond following an EHR transition.³⁰ Evaluating the long-term effects of transitions on HPTs is of interest, given their distinct characteristics and differences from employees.

Limitations

Study data came from voluntary participants at 1 highly engaged site, raising the possibility of self-selection bias. HPT experiences at other VA and non-VA sites may differ. Employees and HPTs were engaged during a high-stress event; snowballing recruitment reach was limited by high workloads and limited time for engagement. Statistical data were descriptive and should not be interpreted as causal. Results may reflect, in part, temporal effects, and respondents include HPTs at different stages of training and with different levels of VA experience. Survey sample sizes may limit generalizability; however, merging data streams strengthened the reliability of findings.

Conclusions

The results of this analysis of FHCC HPTs were notably more positive than those of HPTs at previous VHA EHR transition sites. VHA is one of many health care systems that provide clinical training for HPTs and relies on this population to provide patient care. By highlighting challenges and positive experiences of HPTs during an EHR transition, this evaluation produces actionable insights that can inform the actions of health care systems seeking to support HPTs during disruptive EHR transitions.

The Veterans Health Administration (VHA) is transitioning from its native electronic health record (EHR) Vista/Computerized Patient Record System to the commercial Cerner/Oracle Health EHR. Though this process was temporarily discontinued in April 2023 due to patient safety, usability, and reliability concerns, it resumed in April 2026. It was originally projected to cost $50 billion to implement. ^1-3 As of March 9, 2024, 6 sites had transitioned to the new EHR.² The transition is the largest of its kind in the US, offering an unparalleled opportunity to examine the effects of EHR transitions on an often overlooked part of the workforce: health professions trainees (HPTs).

HPTs serve a central role in VHA. About one-third of patients receive care directly from HPTs who make up about one-third of the VHA workforce. VHA trains > 60 clinical disciplines, comprising > 122,000 trainees annually.^4,5 A paucity of literature exists exploring the experiences of HPTs during EHR transitions, and many studies are often limited to single-site or small populations. HPTs face distinct challenges and needs during EHR transitions and are particularly vulnerable to their negative impacts on retention, clinical training, and efficiency and confidence in EHR use.^6-10 HPTs at VHA sites that have already transitioned to the Cerner/Oracle Health EHR identified many challenges, including significant delays in gaining EHR access, pervasive perceptions of poor training, concerns that EHR functionality issues limited patient care, and decreased ability to track clinical skill acquisition.⁶ These challenges may impact some HPTs more than others (eg, students on short rotations are affected more acutely by delayed EHR access and usage).

This quality improvement project evaluated HPT EHR transition experiences at the Captain James A. Lovell Federal Health Care Center (FHCC). This article contributes to the limited literature on HPT transition experiences, identifies opportunities to support HPTs, and informs broader efforts in teaching HPTs new technologies.

Methods

FHCC is jointly operated by the US Department of Defense and US Department of Veterans Affairs (VA). It treats 80,000 inpatient and outpatients annually. FHCC was the sixth VA facility to transition to the new EHR, which went live on March 9, 2024.^2,11 About 700 HPTs rotate through FHCC annually. HPTs were eligible for inclusion if they were present during the March 9 transition according to a VA Office of Academic Affiliations database. A total of 216 HPTs were identified for inclusion.

Preparations for the transition included scaling down operations (ie, blocking clinician schedules, not scheduling future appointments that may conflict with the transition, making decisions on new facility- and service-line workflows, required EHR training, and speaking with support staff, including VHA National EHR Modernization Supplemental Staffing Unit [NESSU]). This evaluation was designated nonresearch/quality improvement by the VA Bedford Healthcare System Institutional Review Board.

Surveys

Forty-seven interviews were conducted with HPTs, site leaders, and supervisors from January 2024 to June 2024 (Table 1). Participants were identified by service leads and recruited via email; snowball sampling identified additional participants.

The evaluation team developed semistructured interview guides using grounded probes based on a pilot evaluation and existing research on EHR transitions.¹² Questions focused on participant experiences preparing for the EHR transition, learning and using the site’s EHR, and the impact the transition had on clinical training experiences. Interviews were conducted at different times to capture the range of user experiences: 1 month prelaunch, 2 to 6 weeks postlaunch, and 2 months postlaunch. Interviewees were informed of participant rights and provided verbal consent.

HPTs present at FHCC at each survey’s release were emailed invitations and 2 reminders. The anonymous surveys took about 10 minutes to complete. Survey items queried HPTs about their experiences preparing to use the new EHR, perceptions of the current EHR (adapted from the System Usability Scale), satisfaction with VHA training, impact on clinical training, ability to work with preceptors and patients, and experiences with the VHA clinical learning environment (adapted from the VHA Learners Preceptor Survey).^13-15 Survey questions used a 5-point Likert response scale.

Analysis

Interviewers completed postinterview summaries for team debriefing and consensus building. Interviews were coded using a priori (from piloting evaluations and relevant literature) and emergent (refined and developed from data) codes. Deductive and inductive content analyses were conducted. ¹⁶ Deductive analysis used a priori categories (eg, care coordination, EHR training). Inductive content analysis consisted of open and unstructured coding, capturing data outside a priori categories. Emergent codes captured unidentified categories. Qualitative researchers met weekly to discuss data and reach consensus on interpretation.

Descriptive analysis was conducted using top-2 box scoring (proportion responding within the 2 most favorable responses [agree/ strongly agree]). Survey data were analyzed in SAS.¹⁷ The analysis used a merging approach on simultaneously collected qualitative and quantitative data to reach findings consensus.¹⁸

Researcher and research team decisions may shape the data collected due to prior assumptions and experience.¹⁹ This analysis attempted to integrate reflexivity practices to enhance awareness of the researchers’ assumptions and positionality, including by integrating intent collaborative conversing and memorandum writing into the processes.^20,21

Results

This analysis created a survey and fielded responses from HPTs present at FHCC across 3 time points (6 months prelaunch, 1 month prelaunch, and 2 months postlaunch), resulting in a total of 103 responses and an average response rate of 19.0% (Table 2). Six key findings were identified in analysis of responses: (1) critiques of transition management; (2) concerns with training; (3) hope about the EHR; (4) at-the-elbow support was essential; (5) HPTs adjusted to, and later preferred, the new EHR; and (6) transition impacted clinical training, but not overall career plans for HPTs. Findings are presented in this section, with illustrative quantitative data and qualitative data quotes available in the eAppendix.

Critiques of the Transition’s Management

While participants were aware of the transition to the new EHR, most felt they did not have enough information or time to prepare for it, indicating it was “too little, too late.” HPTs felt necessary workflow processes for Cerner/Oracle Health were not determined with enough time to learn them prior to transition. Supervisors shared that important workflow and onboarding decisions remained undecided mere weeks before the transition. Some service lines did not decrease patient loads until right before the transition, making it difficult to manage their schedules and resulting in insufficient time to learn the new EHR.

EHR Training Concerns

Overall, HPTs expressed low satisfaction with computer-based Training Management System (TMS) EHR training, believing it did not prepare them for the new EHR. The percentage of HPTs satisfied or very satisfied with the quality of TMS training was lower than that of instructor-based training pre- and posttransition, with 50% of 36 prelaunch respondents, and 43% of 29 postlaunch respondents expressing satisfaction with computer-based trainings (Figure 1). HPTs were dissatisfied with the training content. They felt it was too general and failed to teach basic tasks in the workflow for their service areas and roles, such as writing a note or order. Furthermore, poor content was exacerbated by poor and unengaging instruction, and HPTs were dissatisfied with the practice EHR used in training, which glitched frequently.

EHR Transition Optimism

Even though the transition was stressful, most HPTs hoped it would be a temporary disruption and that they would quickly adjust to the new EHR. Many participants expected that once they switched to the new EHR, they would pick it up quickly. In addition, many anticipated Cerner/Oracle Health would be better and easier to use in the long run.

At-The-Elbow Support Essential

VHA peer support with NESSU was highly valued among HPTs. NESSU staff were highly knowledgeable and could provide both broad and service-line-specific support. NESSU provided prompt answers to EHR questions. This was particularly critical as other forms of in-person support were often inaccessible or absent during the transition.

HPTs found facility support helpful: 85% of 36 respondents reported being satisfied/ very satisfied with support from supervisors and preceptors, and 84% of 36 respondents were satisfied/very satisfied with technical support from facility informatics staff pretransition (n = 36) (Figure 2). NESSU and supervisor support with daily workflows were particularly helpful, as pretransition training only provided a general introduction to the EHR.

HPTs Adjusted to and Later Preferred the New EHR

The EHR learning experience was intense but short, with many HPTs feeling able to use it only 2 to 4 weeks posttransition. Confidence grew as HPTs came to view Cerner/Oracle Health as a more integrated and intuitive system than the previous EHR. Most participants preferred the new EHR, even if they criticized some features (eg, no group documentation capabilities). Survey participants frequently rated Cerner/Oracle Health usability higher than the original. A total of 32% of 29 posttransition respondents agreed or strongly agreed that Cerner/Oracle Health helps prevent situations that can lead to patient safety risks—higher than pretransition rates. Additionally, fewer respondents found the new EHR unnecessarily complex or thought it contained too many alerts and flags compared to the original EHR (Figure 3).

Impact on Clinical Training, Not Career Plans

The extensive time and energy the transition demanded of HPTs caused stress and affected their clinical training. Many believed they would have learned more if their training had happened outside the transition.

Concerns that the transition affected learning were most acutely felt pretransition. HPTs reporting that EHR implementation positively affected their clinical education fell from 38% of 36 respondents 6 months pretransition to 19% of 29 respondents 1 month pretransition, but returned to 37% posttransition (Figure 4). However, some HPTs believed there was a silver lining: it provided a learning experience they otherwise would not have had.

HPTs who believed the transition positively impacted their likelihood of pursuing future career opportunities within the VHA rose to 33% of 29 respondents posttransition. Overall, Cerner/Oracle Health was characterized as a tool: something used in training, but not something that precluded wanting VHA careers or having meaningful experiences, such as caring for patients.

Discussion

This evaluation addressed an underexplored aspect of EHR transitions: their impact on HPTs. It identified HPT challenges, including dissatisfaction with poor transition preparation and EHR training experiences. Promising findings include positive experiences with transition support, EHR uptake, and overall positive educational experiences despite the transition’s disruption.

When EHR users, including HPTs, are dissatisfied with transition preparations, consequent stress can lead to undesired effects, including increased burnout, inappropriate EHR use, and low work satisfaction.^22-24 Negative EHR transition experiences shape HPTs’ subsequent EHR adoption, user satisfaction, as well as confidence and career intent.^3,25,26 Health systems have strong incentives to implement effective transition change management.

HPTs at previous VHA EHR transition sites reported significantly more disruption to their clinical training compared with HPTs at FHCC. Academic programs were shut down at the first transition site, and HPTs expressed decreased interest in VHA careers at another, even a year posttransition.^6,27 These findings are consistent with the limited literature on the adverse impacts that EHR transitions have on HPTs.^7,28

HPT retention is critical. VA is mandated to prepare the next generation of HPTs for its needs, and those of the nation. The VA relies heavily on HPT retention to recruit clinicians: > 65% of VHA physicians nationwide participated in VHA training programs prior to recruitment into staff positions.^5,29

VHA should invest in transition change management with demonstrated, positive impacts on HPTs, such as in-house support from clinicians. Previous research found that lack of support was a major source of stress and negative outcomes.^6,27 Consequently, supporting HPTs through EHR transitions directly contributes to the VHA’s ability to attract high-quality staff from its HPTs. The challenges and promising practices described in this analysis underscore the necessity of understanding how all EHR users are affected by transitions. What happens to them has direct implications for the VA mission to provide safe, efficient care, and its mandate to provide quality clinical training to HPTs.

These findings hold hopeful implications for supporting HPT EHR use, both during and outside EHR transitions. HPTs expressing that an EHR is only 1 part of their clinical training experience suggests that change management can improve EHR transitions. HPT learning can enhance known factors that are important for HPTs in clinical training, including the health care organization’s mission, caring for patients, and personal development.

Further investigations may engage HPTs at future VHA sites making the transition to the new EHR. One focus would involve applying a learning health systems framework to examine the nature of change management efforts—and their effects on HPT transition experiences—iteratively across transition sites to evaluate the effect of the efforts. Another focus may be longitudinal engagement with HPTs at health care systems and sites transitioning to new EHRs. Research has found that disruptions to EHR usability, satisfaction, and care provision can persist for 2 years and beyond following an EHR transition.³⁰ Evaluating the long-term effects of transitions on HPTs is of interest, given their distinct characteristics and differences from employees.

Limitations

Study data came from voluntary participants at 1 highly engaged site, raising the possibility of self-selection bias. HPT experiences at other VA and non-VA sites may differ. Employees and HPTs were engaged during a high-stress event; snowballing recruitment reach was limited by high workloads and limited time for engagement. Statistical data were descriptive and should not be interpreted as causal. Results may reflect, in part, temporal effects, and respondents include HPTs at different stages of training and with different levels of VA experience. Survey sample sizes may limit generalizability; however, merging data streams strengthened the reliability of findings.

Conclusions

The results of this analysis of FHCC HPTs were notably more positive than those of HPTs at previous VHA EHR transition sites. VHA is one of many health care systems that provide clinical training for HPTs and relies on this population to provide patient care. By highlighting challenges and positive experiences of HPTs during an EHR transition, this evaluation produces actionable insights that can inform the actions of health care systems seeking to support HPTs during disruptive EHR transitions.

The Veterans Health Administration (VHA) is transitioning from its native electronic health record (EHR) Vista/Computerized Patient Record System to the commercial Cerner/Oracle Health EHR. Though this process was temporarily discontinued in April 2023 due to patient safety, usability, and reliability concerns, it resumed in April 2026. It was originally projected to cost $50 billion to implement. ^1-3 As of March 9, 2024, 6 sites had transitioned to the new EHR.² The transition is the largest of its kind in the US, offering an unparalleled opportunity to examine the effects of EHR transitions on an often overlooked part of the workforce: health professions trainees (HPTs).

HPTs serve a central role in VHA. About one-third of patients receive care directly from HPTs who make up about one-third of the VHA workforce. VHA trains > 60 clinical disciplines, comprising > 122,000 trainees annually.^4,5 A paucity of literature exists exploring the experiences of HPTs during EHR transitions, and many studies are often limited to single-site or small populations. HPTs face distinct challenges and needs during EHR transitions and are particularly vulnerable to their negative impacts on retention, clinical training, and efficiency and confidence in EHR use.^6-10 HPTs at VHA sites that have already transitioned to the Cerner/Oracle Health EHR identified many challenges, including significant delays in gaining EHR access, pervasive perceptions of poor training, concerns that EHR functionality issues limited patient care, and decreased ability to track clinical skill acquisition.⁶ These challenges may impact some HPTs more than others (eg, students on short rotations are affected more acutely by delayed EHR access and usage).

This quality improvement project evaluated HPT EHR transition experiences at the Captain James A. Lovell Federal Health Care Center (FHCC). This article contributes to the limited literature on HPT transition experiences, identifies opportunities to support HPTs, and informs broader efforts in teaching HPTs new technologies.

Methods

FHCC is jointly operated by the US Department of Defense and US Department of Veterans Affairs (VA). It treats 80,000 inpatient and outpatients annually. FHCC was the sixth VA facility to transition to the new EHR, which went live on March 9, 2024.^2,11 About 700 HPTs rotate through FHCC annually. HPTs were eligible for inclusion if they were present during the March 9 transition according to a VA Office of Academic Affiliations database. A total of 216 HPTs were identified for inclusion.

Preparations for the transition included scaling down operations (ie, blocking clinician schedules, not scheduling future appointments that may conflict with the transition, making decisions on new facility- and service-line workflows, required EHR training, and speaking with support staff, including VHA National EHR Modernization Supplemental Staffing Unit [NESSU]). This evaluation was designated nonresearch/quality improvement by the VA Bedford Healthcare System Institutional Review Board.

Surveys

Forty-seven interviews were conducted with HPTs, site leaders, and supervisors from January 2024 to June 2024 (Table 1). Participants were identified by service leads and recruited via email; snowball sampling identified additional participants.

The evaluation team developed semistructured interview guides using grounded probes based on a pilot evaluation and existing research on EHR transitions.¹² Questions focused on participant experiences preparing for the EHR transition, learning and using the site’s EHR, and the impact the transition had on clinical training experiences. Interviews were conducted at different times to capture the range of user experiences: 1 month prelaunch, 2 to 6 weeks postlaunch, and 2 months postlaunch. Interviewees were informed of participant rights and provided verbal consent.

HPTs present at FHCC at each survey’s release were emailed invitations and 2 reminders. The anonymous surveys took about 10 minutes to complete. Survey items queried HPTs about their experiences preparing to use the new EHR, perceptions of the current EHR (adapted from the System Usability Scale), satisfaction with VHA training, impact on clinical training, ability to work with preceptors and patients, and experiences with the VHA clinical learning environment (adapted from the VHA Learners Preceptor Survey).^13-15 Survey questions used a 5-point Likert response scale.

Analysis

Interviewers completed postinterview summaries for team debriefing and consensus building. Interviews were coded using a priori (from piloting evaluations and relevant literature) and emergent (refined and developed from data) codes. Deductive and inductive content analyses were conducted. ¹⁶ Deductive analysis used a priori categories (eg, care coordination, EHR training). Inductive content analysis consisted of open and unstructured coding, capturing data outside a priori categories. Emergent codes captured unidentified categories. Qualitative researchers met weekly to discuss data and reach consensus on interpretation.

Descriptive analysis was conducted using top-2 box scoring (proportion responding within the 2 most favorable responses [agree/ strongly agree]). Survey data were analyzed in SAS.¹⁷ The analysis used a merging approach on simultaneously collected qualitative and quantitative data to reach findings consensus.¹⁸

Researcher and research team decisions may shape the data collected due to prior assumptions and experience.¹⁹ This analysis attempted to integrate reflexivity practices to enhance awareness of the researchers’ assumptions and positionality, including by integrating intent collaborative conversing and memorandum writing into the processes.^20,21

Results

This analysis created a survey and fielded responses from HPTs present at FHCC across 3 time points (6 months prelaunch, 1 month prelaunch, and 2 months postlaunch), resulting in a total of 103 responses and an average response rate of 19.0% (Table 2). Six key findings were identified in analysis of responses: (1) critiques of transition management; (2) concerns with training; (3) hope about the EHR; (4) at-the-elbow support was essential; (5) HPTs adjusted to, and later preferred, the new EHR; and (6) transition impacted clinical training, but not overall career plans for HPTs. Findings are presented in this section, with illustrative quantitative data and qualitative data quotes available in the eAppendix.

Critiques of the Transition’s Management

While participants were aware of the transition to the new EHR, most felt they did not have enough information or time to prepare for it, indicating it was “too little, too late.” HPTs felt necessary workflow processes for Cerner/Oracle Health were not determined with enough time to learn them prior to transition. Supervisors shared that important workflow and onboarding decisions remained undecided mere weeks before the transition. Some service lines did not decrease patient loads until right before the transition, making it difficult to manage their schedules and resulting in insufficient time to learn the new EHR.

EHR Training Concerns

Overall, HPTs expressed low satisfaction with computer-based Training Management System (TMS) EHR training, believing it did not prepare them for the new EHR. The percentage of HPTs satisfied or very satisfied with the quality of TMS training was lower than that of instructor-based training pre- and posttransition, with 50% of 36 prelaunch respondents, and 43% of 29 postlaunch respondents expressing satisfaction with computer-based trainings (Figure 1). HPTs were dissatisfied with the training content. They felt it was too general and failed to teach basic tasks in the workflow for their service areas and roles, such as writing a note or order. Furthermore, poor content was exacerbated by poor and unengaging instruction, and HPTs were dissatisfied with the practice EHR used in training, which glitched frequently.

EHR Transition Optimism

Even though the transition was stressful, most HPTs hoped it would be a temporary disruption and that they would quickly adjust to the new EHR. Many participants expected that once they switched to the new EHR, they would pick it up quickly. In addition, many anticipated Cerner/Oracle Health would be better and easier to use in the long run.

At-The-Elbow Support Essential

VHA peer support with NESSU was highly valued among HPTs. NESSU staff were highly knowledgeable and could provide both broad and service-line-specific support. NESSU provided prompt answers to EHR questions. This was particularly critical as other forms of in-person support were often inaccessible or absent during the transition.

HPTs found facility support helpful: 85% of 36 respondents reported being satisfied/ very satisfied with support from supervisors and preceptors, and 84% of 36 respondents were satisfied/very satisfied with technical support from facility informatics staff pretransition (n = 36) (Figure 2). NESSU and supervisor support with daily workflows were particularly helpful, as pretransition training only provided a general introduction to the EHR.

HPTs Adjusted to and Later Preferred the New EHR

The EHR learning experience was intense but short, with many HPTs feeling able to use it only 2 to 4 weeks posttransition. Confidence grew as HPTs came to view Cerner/Oracle Health as a more integrated and intuitive system than the previous EHR. Most participants preferred the new EHR, even if they criticized some features (eg, no group documentation capabilities). Survey participants frequently rated Cerner/Oracle Health usability higher than the original. A total of 32% of 29 posttransition respondents agreed or strongly agreed that Cerner/Oracle Health helps prevent situations that can lead to patient safety risks—higher than pretransition rates. Additionally, fewer respondents found the new EHR unnecessarily complex or thought it contained too many alerts and flags compared to the original EHR (Figure 3).

Impact on Clinical Training, Not Career Plans

The extensive time and energy the transition demanded of HPTs caused stress and affected their clinical training. Many believed they would have learned more if their training had happened outside the transition.

Concerns that the transition affected learning were most acutely felt pretransition. HPTs reporting that EHR implementation positively affected their clinical education fell from 38% of 36 respondents 6 months pretransition to 19% of 29 respondents 1 month pretransition, but returned to 37% posttransition (Figure 4). However, some HPTs believed there was a silver lining: it provided a learning experience they otherwise would not have had.

HPTs who believed the transition positively impacted their likelihood of pursuing future career opportunities within the VHA rose to 33% of 29 respondents posttransition. Overall, Cerner/Oracle Health was characterized as a tool: something used in training, but not something that precluded wanting VHA careers or having meaningful experiences, such as caring for patients.

Discussion

This evaluation addressed an underexplored aspect of EHR transitions: their impact on HPTs. It identified HPT challenges, including dissatisfaction with poor transition preparation and EHR training experiences. Promising findings include positive experiences with transition support, EHR uptake, and overall positive educational experiences despite the transition’s disruption.

When EHR users, including HPTs, are dissatisfied with transition preparations, consequent stress can lead to undesired effects, including increased burnout, inappropriate EHR use, and low work satisfaction.^22-24 Negative EHR transition experiences shape HPTs’ subsequent EHR adoption, user satisfaction, as well as confidence and career intent.^3,25,26 Health systems have strong incentives to implement effective transition change management.

HPTs at previous VHA EHR transition sites reported significantly more disruption to their clinical training compared with HPTs at FHCC. Academic programs were shut down at the first transition site, and HPTs expressed decreased interest in VHA careers at another, even a year posttransition.^6,27 These findings are consistent with the limited literature on the adverse impacts that EHR transitions have on HPTs.^7,28

HPT retention is critical. VA is mandated to prepare the next generation of HPTs for its needs, and those of the nation. The VA relies heavily on HPT retention to recruit clinicians: > 65% of VHA physicians nationwide participated in VHA training programs prior to recruitment into staff positions.^5,29

VHA should invest in transition change management with demonstrated, positive impacts on HPTs, such as in-house support from clinicians. Previous research found that lack of support was a major source of stress and negative outcomes.^6,27 Consequently, supporting HPTs through EHR transitions directly contributes to the VHA’s ability to attract high-quality staff from its HPTs. The challenges and promising practices described in this analysis underscore the necessity of understanding how all EHR users are affected by transitions. What happens to them has direct implications for the VA mission to provide safe, efficient care, and its mandate to provide quality clinical training to HPTs.

These findings hold hopeful implications for supporting HPT EHR use, both during and outside EHR transitions. HPTs expressing that an EHR is only 1 part of their clinical training experience suggests that change management can improve EHR transitions. HPT learning can enhance known factors that are important for HPTs in clinical training, including the health care organization’s mission, caring for patients, and personal development.

Further investigations may engage HPTs at future VHA sites making the transition to the new EHR. One focus would involve applying a learning health systems framework to examine the nature of change management efforts—and their effects on HPT transition experiences—iteratively across transition sites to evaluate the effect of the efforts. Another focus may be longitudinal engagement with HPTs at health care systems and sites transitioning to new EHRs. Research has found that disruptions to EHR usability, satisfaction, and care provision can persist for 2 years and beyond following an EHR transition.³⁰ Evaluating the long-term effects of transitions on HPTs is of interest, given their distinct characteristics and differences from employees.

Limitations

Study data came from voluntary participants at 1 highly engaged site, raising the possibility of self-selection bias. HPT experiences at other VA and non-VA sites may differ. Employees and HPTs were engaged during a high-stress event; snowballing recruitment reach was limited by high workloads and limited time for engagement. Statistical data were descriptive and should not be interpreted as causal. Results may reflect, in part, temporal effects, and respondents include HPTs at different stages of training and with different levels of VA experience. Survey sample sizes may limit generalizability; however, merging data streams strengthened the reliability of findings.

Conclusions

The results of this analysis of FHCC HPTs were notably more positive than those of HPTs at previous VHA EHR transition sites. VHA is one of many health care systems that provide clinical training for HPTs and relies on this population to provide patient care. By highlighting challenges and positive experiences of HPTs during an EHR transition, this evaluation produces actionable insights that can inform the actions of health care systems seeking to support HPTs during disruptive EHR transitions.

Droperidol is a butyrophenone antipsychotic approved by the US Food and Drug Administration (FDA) for use in postoperative nausea and vomiting (PONV). Off-label, it has also been utilized for its sedative, anxiolytic, and analgesic properties.¹ While its exact mechanism of action remains elusive, it is believed that binding to postsynaptic γ-aminobutyric acid receptors induces anxiolysis and sedation, while dopaminergic activity in the chemoreceptor trigger zone contributes to its antiemetic effects.² Since the introduction of droperidol in 1967, it has been widely used by emergency physicians, psychiatrists, and anesthesiologists globally.¹

Despite its therapeutic efficacy, use of droperidol has been tempered by concerns regarding its cardiovascular safety profile, specifically its potential to prolong the QT interval and precipitate cardiac arrhythmias. In 2001, the FDA placed a boxed warning on droperidol that mandated electrocardiogram (EKG) monitoring before and after treatment. This requirement has led to a widespread decrease in use, and the FDA decision sparked significant controversy among clinicians, with many organizations arguing that the evidence did not support this mandate.¹

Further review of the cases cited by the FDA revealed that there were 277 reported cases of droperidol-related adverse events (AEs), but many of these cases were duplicates and occurred outside the US.³ Additionally, the doses of droperidol used in these cases were significantly higher than the typical doses used in the emergency department (ED), ranging from 25 to 250 mg.⁴ Typical doses for PONV range from 0.625 to 2.5 mg intravenous (IV) or intramuscular (IM). Recommended doses for agitation typically range from 2.5 to 10 mg IV and 5 to 10 mg IM.⁵

There has been growing interest in reevaluating the risk-benefit profile of droperidol in the ED. Since the original decision by the FDA, multiple publications have challenged the idea that droperidol has significantly higher risks associated with its use. The 2014 review by the Clinical Guidelines Committee of the American Academy of Emergency Medicine did not find evidence that low-dose droperidol (< 2.5 is unsafe for use in the ED.⁶ A retrospective cohort study from 2020 found no fatalities in 5784 patients. Furthermore, a prospective observational study of 1009 patients at 6 EDs who received high-dose droperidol (≤ 20.0 mg) found no evidence of increased risk for QT prolongation.⁷ The evidence supports the safety of droperidol for use in prehospital and hospital settings as well as in pediatric, adult, and geriatric populations.^8-14 Droperidol was eventually reintroduced in 2019, which led to increased use.

The US Department of Veterans Affairs (VA) formulary has limited options (eg, haloperidol and olanzapine) that have robust evidence supporting their use to treat aggression or psychosis-related agitation. Ziprasidone injections are not on the formulary and require authorization for use, which may delay patient care and pose a safety risk. In 2021, VA Greater Los Angeles Healthcare System (VAGLAHS) received Pharmacy and Therapeutics Committee approval to use droperidol in the ED for agitation or nausea and vomiting. The purpose of this study was to evaluate safety outcomes for patients prescribed droperidol and the need for rescue medications (ie, effectiveness) in the VAGLAHS ED.

Methods

This retrospective chart review analyzed patients administered droperidol in the VAGLAHS ED from February 1, 2021, through April 30, 2023. A list of patients who had droperidol ordered in the VAGLAHS ED was obtained from the Veterans Health Information Systems and Technology Architecture. Charts were reviewed using the Computerized Patient Record System to confirm droperidol administration. Nurse documentation was reviewed to confirm the time, dose, and route of administration. In addition, droperidol dosages were categorized as < 5 mg, 5 to 10 mg, and > 10 mg to review outcomes based on the total amount administered to each patient.

Patients included in the study received droperidol in the ED within the study period, were aged ≥ 18 years, and received droperidol for acute agitation or antiemesis. Patients were excluded if they received droperidol for an indication other than agitation or antiemesis.

The study team reviewed the list of patients and audited the collected data. Reviewers were trained on the study protocols and variables identified. The following data were collected: patient demographics (age, sex, race, height, weight, allergies), Charlson Comorbidity Index (CCI) conditions, cardiac comorbidities, laboratory values at admission, basic metabolic panels, liver function tests, droperidol use (doses, indications, and documentation of safety), concomitant medications ordered with the initial droperidol order, AEs (arrhythmias, extrapyramidal symptoms [EPS], respiratory depression, mortality), medications used within 60 minutes of droperidol administration (rescue medications), other medications used within 24 hours after droperidol administration, and EKG/QTc (corrected QT interval) intervals. The data reviewed and recorded were from the date of the initial patient ED visit.

Outcomes

The primary outcome was all-cause mortality within 24 hours after droperidol administration. This outcome was measured in all patients included in this study. Secondary outcomes included rescue medications needed after droperidol administration, incidence of QT prolongation, incidence of EPS (defined as akathisia, dystonia, parkinsonism, or tardive dyskinesia), and incidence of respiratory depression. Clinically significant QTc was defined as an interval of ≥ 500 ms with incidence of arrhythmias, code blues, or intubations. Baseline risk factors for QTc prolongation were taken into consideration including electrolyte abnormalities, concomitant QT-prolonging medications, CCI score, and cardiac comorbidities. Incidence of EPS was counted if patients received medications such as diphenhydramine or benztropine after droperidol administration in addition to documentation of EPS signs and symptoms. Incidences of EPS findings were reviewed by emergency department physicians to confirm the diagnosis.

Safety was assessed by quantifying mortality rates 24 hours after droperidol administration along with incidence of AEs associated with droperidol use including QT prolongation, EPS, and respiratory depression.

The necessity of rescue medication use was assessed by nursing documentation, additional medications ordered, and/or no additional medications required for agitation within 60 minutes of droperidol administration. Sixty minutes was the chosen timeframe given that the onset of droperidol action is between 3 and 10 minutes and peaks in about 30 minutes. Medications that were considered rescue medications included diphenhydramine < 25 mg, diphenhydramine 25 to 50 mg, lorazepam < 1 mg, lorazepam 1 to 2 mg, diphenhydramine < 25 mg and lorazepam < 1 mg, diphenhydramine < 25 mg and lorazepam 1 to 2 mg, diphenhydramine 25 to 50 mg and lorazepam 1 to 2 mg, and other medications, the names and doses of which were manually documented by investigators.

Statistical Analysis

For all variables in the study, descriptive analysis was used to categorize findings. Microsoft Excel was used to calculate means, frequency counts, percentages, and categorize data.

Results

Between February 1, 2021, and April 16, 2023, 214 patients received droperidol in the VAGLAHS ED, and 207 patients were included in the study. Seven patients did not receive droperidol for the indications included (acute agitation or antiemesis). Most of the study population (89.4%) was male, and the mean age was 51.0 years. The mean CCI was 1.6. In the study, 183 (88.4%) patients received droperidol for agitation and 24 (11.6%) for nausea and vomiting (Table 1).

Primary Outcome

No deaths were observed in a 24-hour period after droperidol administration among the 207 patients included in the study. There were also no arrhythmias, code blues, or intubations observed with the administration of droperidol (Table 2).

Secondary Outcomes

A total of 144 patients (69.6%) received droperidol alone to resolve agitation or nausea and vomiting. In the remaining population, 63 (30.4%) patients were given medications concomitantly with droperidol.

Fifteen patients (7.2%) required rescue medications that were administered within 60 minutes of droperidol administration. Rescue medications were required for 7 patients (4.9%) who initially received droperidol alone compared with 8 patients (12.7%) who were administered concomitant medications with droperidol (Figure).

Extrapyramidal Symptoms

EPS occurred in 2 patients (1.0%). There was 1 incidence of tardive dyskinesia (TD) in which the patient received droperidol 2.5 mg IM for emesis. TD was resolved with diphenhydramine 50 mg. A second patient who experienced dystonia received droperidol 10 mg IM for agitation. Dystonia was resolved with benztropine 2 mg. Both patients had a CCI of 0, no cardiac comorbidities, and laboratory test results were within reference ranges. The second patient received olanzapine within 24 hours of droperidol administration; however, it was after the EPS event.

QTc Prolongation

Baseline EKGs (within 6 months prior to ED visit) were available for 102 patients (49.3%). Nine patients (8.8%) had a reported baseline QTc of ≥ 500 ms (Table 3). Of these patients, 6 had a repeat EKG and 5 had a repeat QTc < 500 ms. One patient had a baseline and repeated QTc of 512 ms with essentially no change after droperidol administration. Only 1 patient was on a potentially QTc-prolonging medication at home. None of the patients with baseline QTc > 500 ms experienced arrhythmias after droperidol administration.

We found that 59 patients (28.5%) had EKGs performed within 24 hours after droperidol administration. Five patients had documented QTc ≥ 500 ms, but no arrhythmias were observed in a 24-hour period. Table 4 describes the additional medications administered after the 60-minute window but within 24 hours after droperidol administration. Quetiapine 300 mg and metoclopramide 5 mg were the only medications documented that can potentially increase QTc. Patient adherence to home medications and the timing of the last dose prior to ED visit were unknown. However, no arrhythmias were noted in these patients with QT^c changes. No patients experienced respiratory depression within 24 hours of droperidol administration.

Older Adult Patients

Thirty-eight patients were aged ≥ 65 years with a mean age of 74.2 years. Thirty-four patients (89.5%) received droperidol for agitation and 4 (10.6%) for nausea and vomiting. Only 21 patients had a baseline EKG, and 4 had QTc ≤ 500 ms. At 24 hours, EKGs were performed for 18 patients and 3 had a QTc ≤ 500 ms. No mortality or arrhythmias were reported and there were no incidences of rescue medications, EPS, or respiratory depression.

Discussion

The study included 207 patients who received droperidol for either agitation or nausea/vomiting in the VAGLAHS ED. No mortality occurred within 24 hours of droperidol administration, which is consistent with recent studies.^8-14

Furthermore, 59 patients (28.5%) had an EKG performed within 24 hours of droperidol administration; 5 patients had documented QTc ≥ 500 ms. Only 3 of the patients with prolonged QTc had baseline readings for comparison. Only 2 patients had an increase in QTc interval. No arrhythmias were observed; however, the effects of observing QTc prolongation were limited due to the lack of post-EKG readings following droperidol administration. Because of the retrospective nature of the study, neither standardization of EKG at baseline nor 24-hour postadministration were possible. The study found that droperidol was effective with only 15 patients (7.3%) requiring rescue medications. In the patients who were given medications concomitantly with droperidol, it was not possible to conclude whether the patients would have required rescue medications to resolve their agitation or nausea/vomiting. Administration of concomitant medications with droperidol may be attributed to practice patterns associated with haloperidol use, which is frequently administered with concomitant medications such as diphenhydramine and/or a benzodiazepine.

AEs were rare with no documentation of respiratory depression and 2 cases (1.0%) of EPS. Both incidences of EPS resolved with diphenhydramine or benztropine. However, given the reliance on nursing documentation to capture AEs, the number of events may have been underreported.

Limitations

Standardization of dosing was a limiting factor that could affect the need for rescue medications. Another limitation was reliance on nursing reports of resolution of symptoms and comfort with agitated patients. Given the retrospective design and small sample size, this study may not have captured all potential AEs. However, the doses administered within this study population were consistent with what was expected based on other studies.^8-14

Conclusions

Droperidol, an antipsychotic, is currently approved for PONV, but is also used off-label for agitation. This study found no fatalities among patients who received droperidol in the ED. The findings suggest that droperidol used for agitation and as an antiemetic, despite its FDA boxed warning, appears to be safe and showed no evidence of mortality, arrhythmias, code blues, or intubations despite the lack of postdose EKG monitoring. Among the 38 patients aged ≥ 65 years, the use of droperidol revealed no increased risks. It should be noted that droperidol appeared safe and few patients required rescue medications within this study population.

Droperidol is a butyrophenone antipsychotic approved by the US Food and Drug Administration (FDA) for use in postoperative nausea and vomiting (PONV). Off-label, it has also been utilized for its sedative, anxiolytic, and analgesic properties.¹ While its exact mechanism of action remains elusive, it is believed that binding to postsynaptic γ-aminobutyric acid receptors induces anxiolysis and sedation, while dopaminergic activity in the chemoreceptor trigger zone contributes to its antiemetic effects.² Since the introduction of droperidol in 1967, it has been widely used by emergency physicians, psychiatrists, and anesthesiologists globally.¹

Despite its therapeutic efficacy, use of droperidol has been tempered by concerns regarding its cardiovascular safety profile, specifically its potential to prolong the QT interval and precipitate cardiac arrhythmias. In 2001, the FDA placed a boxed warning on droperidol that mandated electrocardiogram (EKG) monitoring before and after treatment. This requirement has led to a widespread decrease in use, and the FDA decision sparked significant controversy among clinicians, with many organizations arguing that the evidence did not support this mandate.¹

Further review of the cases cited by the FDA revealed that there were 277 reported cases of droperidol-related adverse events (AEs), but many of these cases were duplicates and occurred outside the US.³ Additionally, the doses of droperidol used in these cases were significantly higher than the typical doses used in the emergency department (ED), ranging from 25 to 250 mg.⁴ Typical doses for PONV range from 0.625 to 2.5 mg intravenous (IV) or intramuscular (IM). Recommended doses for agitation typically range from 2.5 to 10 mg IV and 5 to 10 mg IM.⁵

There has been growing interest in reevaluating the risk-benefit profile of droperidol in the ED. Since the original decision by the FDA, multiple publications have challenged the idea that droperidol has significantly higher risks associated with its use. The 2014 review by the Clinical Guidelines Committee of the American Academy of Emergency Medicine did not find evidence that low-dose droperidol (< 2.5 is unsafe for use in the ED.⁶ A retrospective cohort study from 2020 found no fatalities in 5784 patients. Furthermore, a prospective observational study of 1009 patients at 6 EDs who received high-dose droperidol (≤ 20.0 mg) found no evidence of increased risk for QT prolongation.⁷ The evidence supports the safety of droperidol for use in prehospital and hospital settings as well as in pediatric, adult, and geriatric populations.^8-14 Droperidol was eventually reintroduced in 2019, which led to increased use.

The US Department of Veterans Affairs (VA) formulary has limited options (eg, haloperidol and olanzapine) that have robust evidence supporting their use to treat aggression or psychosis-related agitation. Ziprasidone injections are not on the formulary and require authorization for use, which may delay patient care and pose a safety risk. In 2021, VA Greater Los Angeles Healthcare System (VAGLAHS) received Pharmacy and Therapeutics Committee approval to use droperidol in the ED for agitation or nausea and vomiting. The purpose of this study was to evaluate safety outcomes for patients prescribed droperidol and the need for rescue medications (ie, effectiveness) in the VAGLAHS ED.

Methods

This retrospective chart review analyzed patients administered droperidol in the VAGLAHS ED from February 1, 2021, through April 30, 2023. A list of patients who had droperidol ordered in the VAGLAHS ED was obtained from the Veterans Health Information Systems and Technology Architecture. Charts were reviewed using the Computerized Patient Record System to confirm droperidol administration. Nurse documentation was reviewed to confirm the time, dose, and route of administration. In addition, droperidol dosages were categorized as < 5 mg, 5 to 10 mg, and > 10 mg to review outcomes based on the total amount administered to each patient.

Patients included in the study received droperidol in the ED within the study period, were aged ≥ 18 years, and received droperidol for acute agitation or antiemesis. Patients were excluded if they received droperidol for an indication other than agitation or antiemesis.

The study team reviewed the list of patients and audited the collected data. Reviewers were trained on the study protocols and variables identified. The following data were collected: patient demographics (age, sex, race, height, weight, allergies), Charlson Comorbidity Index (CCI) conditions, cardiac comorbidities, laboratory values at admission, basic metabolic panels, liver function tests, droperidol use (doses, indications, and documentation of safety), concomitant medications ordered with the initial droperidol order, AEs (arrhythmias, extrapyramidal symptoms [EPS], respiratory depression, mortality), medications used within 60 minutes of droperidol administration (rescue medications), other medications used within 24 hours after droperidol administration, and EKG/QTc (corrected QT interval) intervals. The data reviewed and recorded were from the date of the initial patient ED visit.

Outcomes

The primary outcome was all-cause mortality within 24 hours after droperidol administration. This outcome was measured in all patients included in this study. Secondary outcomes included rescue medications needed after droperidol administration, incidence of QT prolongation, incidence of EPS (defined as akathisia, dystonia, parkinsonism, or tardive dyskinesia), and incidence of respiratory depression. Clinically significant QTc was defined as an interval of ≥ 500 ms with incidence of arrhythmias, code blues, or intubations. Baseline risk factors for QTc prolongation were taken into consideration including electrolyte abnormalities, concomitant QT-prolonging medications, CCI score, and cardiac comorbidities. Incidence of EPS was counted if patients received medications such as diphenhydramine or benztropine after droperidol administration in addition to documentation of EPS signs and symptoms. Incidences of EPS findings were reviewed by emergency department physicians to confirm the diagnosis.

Safety was assessed by quantifying mortality rates 24 hours after droperidol administration along with incidence of AEs associated with droperidol use including QT prolongation, EPS, and respiratory depression.

The necessity of rescue medication use was assessed by nursing documentation, additional medications ordered, and/or no additional medications required for agitation within 60 minutes of droperidol administration. Sixty minutes was the chosen timeframe given that the onset of droperidol action is between 3 and 10 minutes and peaks in about 30 minutes. Medications that were considered rescue medications included diphenhydramine < 25 mg, diphenhydramine 25 to 50 mg, lorazepam < 1 mg, lorazepam 1 to 2 mg, diphenhydramine < 25 mg and lorazepam < 1 mg, diphenhydramine < 25 mg and lorazepam 1 to 2 mg, diphenhydramine 25 to 50 mg and lorazepam 1 to 2 mg, and other medications, the names and doses of which were manually documented by investigators.

Statistical Analysis

For all variables in the study, descriptive analysis was used to categorize findings. Microsoft Excel was used to calculate means, frequency counts, percentages, and categorize data.

Results

Between February 1, 2021, and April 16, 2023, 214 patients received droperidol in the VAGLAHS ED, and 207 patients were included in the study. Seven patients did not receive droperidol for the indications included (acute agitation or antiemesis). Most of the study population (89.4%) was male, and the mean age was 51.0 years. The mean CCI was 1.6. In the study, 183 (88.4%) patients received droperidol for agitation and 24 (11.6%) for nausea and vomiting (Table 1).

Primary Outcome

No deaths were observed in a 24-hour period after droperidol administration among the 207 patients included in the study. There were also no arrhythmias, code blues, or intubations observed with the administration of droperidol (Table 2).

Secondary Outcomes

A total of 144 patients (69.6%) received droperidol alone to resolve agitation or nausea and vomiting. In the remaining population, 63 (30.4%) patients were given medications concomitantly with droperidol.

Fifteen patients (7.2%) required rescue medications that were administered within 60 minutes of droperidol administration. Rescue medications were required for 7 patients (4.9%) who initially received droperidol alone compared with 8 patients (12.7%) who were administered concomitant medications with droperidol (Figure).

Extrapyramidal Symptoms

EPS occurred in 2 patients (1.0%). There was 1 incidence of tardive dyskinesia (TD) in which the patient received droperidol 2.5 mg IM for emesis. TD was resolved with diphenhydramine 50 mg. A second patient who experienced dystonia received droperidol 10 mg IM for agitation. Dystonia was resolved with benztropine 2 mg. Both patients had a CCI of 0, no cardiac comorbidities, and laboratory test results were within reference ranges. The second patient received olanzapine within 24 hours of droperidol administration; however, it was after the EPS event.

QTc Prolongation

Baseline EKGs (within 6 months prior to ED visit) were available for 102 patients (49.3%). Nine patients (8.8%) had a reported baseline QTc of ≥ 500 ms (Table 3). Of these patients, 6 had a repeat EKG and 5 had a repeat QTc < 500 ms. One patient had a baseline and repeated QTc of 512 ms with essentially no change after droperidol administration. Only 1 patient was on a potentially QTc-prolonging medication at home. None of the patients with baseline QTc > 500 ms experienced arrhythmias after droperidol administration.

We found that 59 patients (28.5%) had EKGs performed within 24 hours after droperidol administration. Five patients had documented QTc ≥ 500 ms, but no arrhythmias were observed in a 24-hour period. Table 4 describes the additional medications administered after the 60-minute window but within 24 hours after droperidol administration. Quetiapine 300 mg and metoclopramide 5 mg were the only medications documented that can potentially increase QTc. Patient adherence to home medications and the timing of the last dose prior to ED visit were unknown. However, no arrhythmias were noted in these patients with QT^c changes. No patients experienced respiratory depression within 24 hours of droperidol administration.

Older Adult Patients

Thirty-eight patients were aged ≥ 65 years with a mean age of 74.2 years. Thirty-four patients (89.5%) received droperidol for agitation and 4 (10.6%) for nausea and vomiting. Only 21 patients had a baseline EKG, and 4 had QTc ≤ 500 ms. At 24 hours, EKGs were performed for 18 patients and 3 had a QTc ≤ 500 ms. No mortality or arrhythmias were reported and there were no incidences of rescue medications, EPS, or respiratory depression.

Discussion

The study included 207 patients who received droperidol for either agitation or nausea/vomiting in the VAGLAHS ED. No mortality occurred within 24 hours of droperidol administration, which is consistent with recent studies.^8-14

Furthermore, 59 patients (28.5%) had an EKG performed within 24 hours of droperidol administration; 5 patients had documented QTc ≥ 500 ms. Only 3 of the patients with prolonged QTc had baseline readings for comparison. Only 2 patients had an increase in QTc interval. No arrhythmias were observed; however, the effects of observing QTc prolongation were limited due to the lack of post-EKG readings following droperidol administration. Because of the retrospective nature of the study, neither standardization of EKG at baseline nor 24-hour postadministration were possible. The study found that droperidol was effective with only 15 patients (7.3%) requiring rescue medications. In the patients who were given medications concomitantly with droperidol, it was not possible to conclude whether the patients would have required rescue medications to resolve their agitation or nausea/vomiting. Administration of concomitant medications with droperidol may be attributed to practice patterns associated with haloperidol use, which is frequently administered with concomitant medications such as diphenhydramine and/or a benzodiazepine.

AEs were rare with no documentation of respiratory depression and 2 cases (1.0%) of EPS. Both incidences of EPS resolved with diphenhydramine or benztropine. However, given the reliance on nursing documentation to capture AEs, the number of events may have been underreported.

Limitations

Standardization of dosing was a limiting factor that could affect the need for rescue medications. Another limitation was reliance on nursing reports of resolution of symptoms and comfort with agitated patients. Given the retrospective design and small sample size, this study may not have captured all potential AEs. However, the doses administered within this study population were consistent with what was expected based on other studies.^8-14

Conclusions

Droperidol, an antipsychotic, is currently approved for PONV, but is also used off-label for agitation. This study found no fatalities among patients who received droperidol in the ED. The findings suggest that droperidol used for agitation and as an antiemetic, despite its FDA boxed warning, appears to be safe and showed no evidence of mortality, arrhythmias, code blues, or intubations despite the lack of postdose EKG monitoring. Among the 38 patients aged ≥ 65 years, the use of droperidol revealed no increased risks. It should be noted that droperidol appeared safe and few patients required rescue medications within this study population.

Droperidol is a butyrophenone antipsychotic approved by the US Food and Drug Administration (FDA) for use in postoperative nausea and vomiting (PONV). Off-label, it has also been utilized for its sedative, anxiolytic, and analgesic properties.¹ While its exact mechanism of action remains elusive, it is believed that binding to postsynaptic γ-aminobutyric acid receptors induces anxiolysis and sedation, while dopaminergic activity in the chemoreceptor trigger zone contributes to its antiemetic effects.² Since the introduction of droperidol in 1967, it has been widely used by emergency physicians, psychiatrists, and anesthesiologists globally.¹

Despite its therapeutic efficacy, use of droperidol has been tempered by concerns regarding its cardiovascular safety profile, specifically its potential to prolong the QT interval and precipitate cardiac arrhythmias. In 2001, the FDA placed a boxed warning on droperidol that mandated electrocardiogram (EKG) monitoring before and after treatment. This requirement has led to a widespread decrease in use, and the FDA decision sparked significant controversy among clinicians, with many organizations arguing that the evidence did not support this mandate.¹

Further review of the cases cited by the FDA revealed that there were 277 reported cases of droperidol-related adverse events (AEs), but many of these cases were duplicates and occurred outside the US.³ Additionally, the doses of droperidol used in these cases were significantly higher than the typical doses used in the emergency department (ED), ranging from 25 to 250 mg.⁴ Typical doses for PONV range from 0.625 to 2.5 mg intravenous (IV) or intramuscular (IM). Recommended doses for agitation typically range from 2.5 to 10 mg IV and 5 to 10 mg IM.⁵

There has been growing interest in reevaluating the risk-benefit profile of droperidol in the ED. Since the original decision by the FDA, multiple publications have challenged the idea that droperidol has significantly higher risks associated with its use. The 2014 review by the Clinical Guidelines Committee of the American Academy of Emergency Medicine did not find evidence that low-dose droperidol (< 2.5 is unsafe for use in the ED.⁶ A retrospective cohort study from 2020 found no fatalities in 5784 patients. Furthermore, a prospective observational study of 1009 patients at 6 EDs who received high-dose droperidol (≤ 20.0 mg) found no evidence of increased risk for QT prolongation.⁷ The evidence supports the safety of droperidol for use in prehospital and hospital settings as well as in pediatric, adult, and geriatric populations.^8-14 Droperidol was eventually reintroduced in 2019, which led to increased use.

The US Department of Veterans Affairs (VA) formulary has limited options (eg, haloperidol and olanzapine) that have robust evidence supporting their use to treat aggression or psychosis-related agitation. Ziprasidone injections are not on the formulary and require authorization for use, which may delay patient care and pose a safety risk. In 2021, VA Greater Los Angeles Healthcare System (VAGLAHS) received Pharmacy and Therapeutics Committee approval to use droperidol in the ED for agitation or nausea and vomiting. The purpose of this study was to evaluate safety outcomes for patients prescribed droperidol and the need for rescue medications (ie, effectiveness) in the VAGLAHS ED.

Methods

This retrospective chart review analyzed patients administered droperidol in the VAGLAHS ED from February 1, 2021, through April 30, 2023. A list of patients who had droperidol ordered in the VAGLAHS ED was obtained from the Veterans Health Information Systems and Technology Architecture. Charts were reviewed using the Computerized Patient Record System to confirm droperidol administration. Nurse documentation was reviewed to confirm the time, dose, and route of administration. In addition, droperidol dosages were categorized as < 5 mg, 5 to 10 mg, and > 10 mg to review outcomes based on the total amount administered to each patient.

Patients included in the study received droperidol in the ED within the study period, were aged ≥ 18 years, and received droperidol for acute agitation or antiemesis. Patients were excluded if they received droperidol for an indication other than agitation or antiemesis.

The study team reviewed the list of patients and audited the collected data. Reviewers were trained on the study protocols and variables identified. The following data were collected: patient demographics (age, sex, race, height, weight, allergies), Charlson Comorbidity Index (CCI) conditions, cardiac comorbidities, laboratory values at admission, basic metabolic panels, liver function tests, droperidol use (doses, indications, and documentation of safety), concomitant medications ordered with the initial droperidol order, AEs (arrhythmias, extrapyramidal symptoms [EPS], respiratory depression, mortality), medications used within 60 minutes of droperidol administration (rescue medications), other medications used within 24 hours after droperidol administration, and EKG/QTc (corrected QT interval) intervals. The data reviewed and recorded were from the date of the initial patient ED visit.

Outcomes

The primary outcome was all-cause mortality within 24 hours after droperidol administration. This outcome was measured in all patients included in this study. Secondary outcomes included rescue medications needed after droperidol administration, incidence of QT prolongation, incidence of EPS (defined as akathisia, dystonia, parkinsonism, or tardive dyskinesia), and incidence of respiratory depression. Clinically significant QTc was defined as an interval of ≥ 500 ms with incidence of arrhythmias, code blues, or intubations. Baseline risk factors for QTc prolongation were taken into consideration including electrolyte abnormalities, concomitant QT-prolonging medications, CCI score, and cardiac comorbidities. Incidence of EPS was counted if patients received medications such as diphenhydramine or benztropine after droperidol administration in addition to documentation of EPS signs and symptoms. Incidences of EPS findings were reviewed by emergency department physicians to confirm the diagnosis.

Safety was assessed by quantifying mortality rates 24 hours after droperidol administration along with incidence of AEs associated with droperidol use including QT prolongation, EPS, and respiratory depression.

The necessity of rescue medication use was assessed by nursing documentation, additional medications ordered, and/or no additional medications required for agitation within 60 minutes of droperidol administration. Sixty minutes was the chosen timeframe given that the onset of droperidol action is between 3 and 10 minutes and peaks in about 30 minutes. Medications that were considered rescue medications included diphenhydramine < 25 mg, diphenhydramine 25 to 50 mg, lorazepam < 1 mg, lorazepam 1 to 2 mg, diphenhydramine < 25 mg and lorazepam < 1 mg, diphenhydramine < 25 mg and lorazepam 1 to 2 mg, diphenhydramine 25 to 50 mg and lorazepam 1 to 2 mg, and other medications, the names and doses of which were manually documented by investigators.

Statistical Analysis

For all variables in the study, descriptive analysis was used to categorize findings. Microsoft Excel was used to calculate means, frequency counts, percentages, and categorize data.

Results

Between February 1, 2021, and April 16, 2023, 214 patients received droperidol in the VAGLAHS ED, and 207 patients were included in the study. Seven patients did not receive droperidol for the indications included (acute agitation or antiemesis). Most of the study population (89.4%) was male, and the mean age was 51.0 years. The mean CCI was 1.6. In the study, 183 (88.4%) patients received droperidol for agitation and 24 (11.6%) for nausea and vomiting (Table 1).

Primary Outcome

No deaths were observed in a 24-hour period after droperidol administration among the 207 patients included in the study. There were also no arrhythmias, code blues, or intubations observed with the administration of droperidol (Table 2).

Secondary Outcomes

A total of 144 patients (69.6%) received droperidol alone to resolve agitation or nausea and vomiting. In the remaining population, 63 (30.4%) patients were given medications concomitantly with droperidol.

Fifteen patients (7.2%) required rescue medications that were administered within 60 minutes of droperidol administration. Rescue medications were required for 7 patients (4.9%) who initially received droperidol alone compared with 8 patients (12.7%) who were administered concomitant medications with droperidol (Figure).

Extrapyramidal Symptoms

EPS occurred in 2 patients (1.0%). There was 1 incidence of tardive dyskinesia (TD) in which the patient received droperidol 2.5 mg IM for emesis. TD was resolved with diphenhydramine 50 mg. A second patient who experienced dystonia received droperidol 10 mg IM for agitation. Dystonia was resolved with benztropine 2 mg. Both patients had a CCI of 0, no cardiac comorbidities, and laboratory test results were within reference ranges. The second patient received olanzapine within 24 hours of droperidol administration; however, it was after the EPS event.

QTc Prolongation

Baseline EKGs (within 6 months prior to ED visit) were available for 102 patients (49.3%). Nine patients (8.8%) had a reported baseline QTc of ≥ 500 ms (Table 3). Of these patients, 6 had a repeat EKG and 5 had a repeat QTc < 500 ms. One patient had a baseline and repeated QTc of 512 ms with essentially no change after droperidol administration. Only 1 patient was on a potentially QTc-prolonging medication at home. None of the patients with baseline QTc > 500 ms experienced arrhythmias after droperidol administration.

We found that 59 patients (28.5%) had EKGs performed within 24 hours after droperidol administration. Five patients had documented QTc ≥ 500 ms, but no arrhythmias were observed in a 24-hour period. Table 4 describes the additional medications administered after the 60-minute window but within 24 hours after droperidol administration. Quetiapine 300 mg and metoclopramide 5 mg were the only medications documented that can potentially increase QTc. Patient adherence to home medications and the timing of the last dose prior to ED visit were unknown. However, no arrhythmias were noted in these patients with QT^c changes. No patients experienced respiratory depression within 24 hours of droperidol administration.

Older Adult Patients

Thirty-eight patients were aged ≥ 65 years with a mean age of 74.2 years. Thirty-four patients (89.5%) received droperidol for agitation and 4 (10.6%) for nausea and vomiting. Only 21 patients had a baseline EKG, and 4 had QTc ≤ 500 ms. At 24 hours, EKGs were performed for 18 patients and 3 had a QTc ≤ 500 ms. No mortality or arrhythmias were reported and there were no incidences of rescue medications, EPS, or respiratory depression.

Discussion

The study included 207 patients who received droperidol for either agitation or nausea/vomiting in the VAGLAHS ED. No mortality occurred within 24 hours of droperidol administration, which is consistent with recent studies.^8-14

Furthermore, 59 patients (28.5%) had an EKG performed within 24 hours of droperidol administration; 5 patients had documented QTc ≥ 500 ms. Only 3 of the patients with prolonged QTc had baseline readings for comparison. Only 2 patients had an increase in QTc interval. No arrhythmias were observed; however, the effects of observing QTc prolongation were limited due to the lack of post-EKG readings following droperidol administration. Because of the retrospective nature of the study, neither standardization of EKG at baseline nor 24-hour postadministration were possible. The study found that droperidol was effective with only 15 patients (7.3%) requiring rescue medications. In the patients who were given medications concomitantly with droperidol, it was not possible to conclude whether the patients would have required rescue medications to resolve their agitation or nausea/vomiting. Administration of concomitant medications with droperidol may be attributed to practice patterns associated with haloperidol use, which is frequently administered with concomitant medications such as diphenhydramine and/or a benzodiazepine.

AEs were rare with no documentation of respiratory depression and 2 cases (1.0%) of EPS. Both incidences of EPS resolved with diphenhydramine or benztropine. However, given the reliance on nursing documentation to capture AEs, the number of events may have been underreported.

Limitations

Standardization of dosing was a limiting factor that could affect the need for rescue medications. Another limitation was reliance on nursing reports of resolution of symptoms and comfort with agitated patients. Given the retrospective design and small sample size, this study may not have captured all potential AEs. However, the doses administered within this study population were consistent with what was expected based on other studies.^8-14

Conclusions

Droperidol, an antipsychotic, is currently approved for PONV, but is also used off-label for agitation. This study found no fatalities among patients who received droperidol in the ED. The findings suggest that droperidol used for agitation and as an antiemetic, despite its FDA boxed warning, appears to be safe and showed no evidence of mortality, arrhythmias, code blues, or intubations despite the lack of postdose EKG monitoring. Among the 38 patients aged ≥ 65 years, the use of droperidol revealed no increased risks. It should be noted that droperidol appeared safe and few patients required rescue medications within this study population.

Cannabis has a long history of use for medicinal and recreational purposes. Research illustrates the potential benefits and increased prevalence of cannabis use in patients with cancer.¹ Cannabis products have been shown to possess antineoplastic and palliative activity, improving nociceptive and neuropathic pain in addition to chemotherapy-related nausea and vomiting.^2-5 Despite these developments and changing social attitudes toward cannabis, there remains a lack of comprehensive data on patient perspectives regarding its use, especially in regions where cannabis remains illegal. This knowledge gap is notable among veterans undergoing cancer treatment in states where cannabis is prohibited. Up to 57% of veterans report lifetime marijuana use, making it crucial to understand this population’s cannabis use patterns and potential interactions with cancer treatments.⁶

This observational study sought to determine the prevalence of cannabis use among patients undergoing cancer treatment at the US Department of Veterans Affairs (VA) Memphis Healthcare System and evaluate the potential risks associated with combining cannabis products with anticancer therapies.

METHODS

This prospective observational study identified cannabis use among veterans receiving antineoplastic therapy at the Lt. Col. Luke Weathers Jr. VA Medical Center (WJVAMC) and analyzed potential interactions between cannabis products and their cancer treatments. Participants included adults aged > 18 years undergoing antineoplastic therapy at WJVAMC who consented to the study. Data collection involved a written survey approved by the WJVAMC Institutional Review Board and verbal consent from participants. The survey asked participants about their cannabis use in the previous 90 days, including details on quantity, frequency, and method of consumption (eg, inhalation, oral, topical). No incentives were offered for participation.

Surveys from 50 patients who used cannabis were analyzed and their electronic health records were reviewed for sex, age, diagnosis, and antineoplastic regimen. This information was securely stored. A literature review was conducted using PubMed and the Cochrane Library to explore potential interactions between cannabis and the antineoplastic agents that were prescribed to patients in the study, focusing on toxicity, efficacy, or synergistic effects.

Patients were categorized into 4 groups based on treatment: cytotoxic chemotherapy, immunotherapy, endocrine therapy, and targeted therapy. Patients undergoing multiple types of therapies were included in each applicable category.

RESULTS

A total of 132 patients agreed to participate. Fifty patients (38%) acknowledged using cannabis products within 90 days. The patients that used cannabis products within 90 days of the survey reported the following malignancies: 8 patients (16%) had prostate cancer, 3 patients (6%) had hepatocellular carcinoma, 7 patients (14%) had pancreatic carcinoma, 5 patients (10%) had multiple myeloma, 3 patients (6%) had chronic lymphocytic leukemia, 9 patients (18%) had non-small cell lung cancer, 3 patients (6%) had breast cancer, 3 (6%) patients had bladder cancer, 2 patients (4%) had renal cell carcinoma, 1 (2%) patient had chronic myeloid leukemia, 1 (2%) patient had renal amyloid, 1 patient (2%) had supraglottic squamous cell carcinoma, 1 patient (2%) had esophageal carcinoma, 1 (2%) patient had small cell lung cancer, 1 (2%) patient had gastric cancer, and 1 patient (2%) had follicular lymphoma.

Five (10%) of the cannabis users were female, and 45 (90%) were male. Twenty-nine patients (58%) were aged 66 to 75 years, 16 (32%) were aged 56 to 65 years, 3 (6%) were aged 46 to 55 years, and 2 (4%) were aged 76 to 85 years.

Thirty-five patients (70%) inhaled cannabis as opposed to using it via other formulations or a combination (eg, inhalation and topical). Thirty-eight percent of patients used cannabis once daily, 24% used < 1 daily, and 28% used it ≥ 2 times daily. Five patients (10%) did not report the frequency of their cannabis use. Among the patients who reported cannabis use, 21 (42%) were undergoing cytotoxic chemotherapy, 19 (38%) were undergoing immunotherapy, 12 (24%) were undergoing targeted therapy, and 10 (20%) were undergoing endocrine therapy. Some patients were treated with multiple types of antineoplastic agents and were counted in multiple categories (Table 1).

Following a literature review of cannabis and antineoplastic agents, patients were evaluated for the potential effects of cannabis on their treatment. The literature review revealed that 31% of cytotoxic chemotherapy agents received by patients in this study might have increased toxicity, and 19% could have reduced efficacy when combined with cannabis. Among immunotherapy agents received by patients in this study, 70% might have decreased efficacy when combined with cannabis use. For targeted therapies, 35% could have increased toxicity, and 70% of endocrine agents could potentially have decreased efficacy (Table 2).

DISCUSSION

This prospective study corroborates previous research by demonstrating that more than one-third of patients receiving oncology care at WJVAMC use cannabis, most often inhaled. Cannabis use was observed among patients undergoing various cancer therapies, including cytotoxic chemotherapy, immunotherapy, targeted therapy, and endocrine therapy. The most common malignancies among cannabis users at WJVAMC include patients with lung cancer, prostate cancer, pancreatic cancer, and multiple myeloma. Cannabis use in patients with pancreatic cancer and multiple myeloma was significantly out of proportion to their prevalence at WJVAMC. This could potentially be due to their drastic effect on quality of life.

Cannabis use increased the risk of toxicity in patients treated with cytotoxic chemotherapy and targeted therapy. Cannabis use potentially decreased efficacy for patients treated with cytotoxic chemotherapy and/or immunotherapy. Cannabis use did not increase the risk of toxicity or efficacy in patients treated with endocrine therapy.

Antineoplastics/Cannabis Interactions

The potential interactions between cannabis and antineoplastic therapies administered at WJVAMC are worth exploring. While this review aims to shed light on possible interactions, it is important to acknowledge that much of the data is preliminary and derived from in vitro studies. The interactions should be interpreted as potential risks rather than established facts. Additional research is needed to confirm these interactions and effectively guide clinical practices. Understanding these dynamics is essential to optimize patient care and manage the complex interplay between cannabis use and cancer treatment.

Originating from Central Asia, the cannabis plant contains > 400 medicinally relevant compounds, of which about 100 are cannabinoids (CBs). Key CBs are cannabidiol (CBD), a nonpsychoactive compound, and ?-9-tetrahydrocannabinol (THC), a psychoactive compound. THC can make up 20% to 30% of the dry weight of female cannabis flowers.⁷

CBs act through the endocannabinoid system, involving CB1 and CB2 receptors, endogenous CBs like anandamide (AEA) and 2-arachidonoylglycerol, and various enzymes. These endogenous CBs, derived from arachidonic acid, play roles in cell growth and proliferation.⁸ In some studies, AEA has induced apoptosis in neuroblastoma cells and inhibited proliferation in breast cancer cells. However, other research suggests AEA may block apoptosis under certain conditions.⁹

CB receptors are transmembrane proteins that interact with CBs differently depending on tissue type and CB structure. Synthetic CBs are designed to target specific receptors, while natural CBs may act as both agonists and antagonists.¹⁰

Cytochrome P450 Metabolism

The human cytochrome P450 (CYP) 3A subfamily affects the metabolism of many therapeutic drugs, including cancer therapeutics.¹¹ The various compositions of cannabis are primarily metabolized by the CYP450 pathway, the same as many cancer-directed pharmacologic treatments. CBs act as both CYP inducers and inhibitors. THC, for example, is a CYP inducer whereas CBD is a CYP inhibitor; both are found in the various compounds available for consumption.^12,13 Pharmacology research has suggested potential interactions and effects on established adverse symptoms, but clinical data are lacking, and current research revealing interactions are only recognized in vitro.¹⁴

The Antineoplastic Activity of Cannabis

CBs can affect various cancer-related pathways such as PKB, AMPK, CAMKK-ß, mTOR, PDHK, HIF-1 a, and PPAR-γ. Δ-9-THC can selectively induce apoptosis in tumor cells without harming normal cells, though the exact mechanism remains unclear. Promising results from early mouse studies led to a 2006 human study where intracranial Δ-9-THC in patients with recurrent glioma yielded a median survival of 24 weeks, with 2 patients surviving > 1 year.¹⁵

In a 2022 review article, Cherkasova et al highlighted potential clinical benefits of cannabis across various cancers. They found that upregulated CB1 receptors in colon cancer might enhance the effect of 5-fluorouracil. However, many studies are preliminary and therefore not definitive.¹⁰

Additional research is needed to refine these findings. Challenges include variability in cannabis formulations, the complex tumor microenvironment, and the legal and psychoactive issues surrounding cannabis use. These factors complicate the design of multicenter randomized studies and may deter patients from disclosing cannabis use, thereby hindering efforts to fully understand its therapeutic potential.

Cannabis/Cytotoxic Chemotherapy Interactions

The chemotherapy agents used in this study included carboplatin, paclitaxel, 5-fluorouracil, etoposide, irinotecan, oxaliplatin, pemetrexed, docetaxel, cabazitaxel, T-DM1, gemcitabine, and cyclophosphamide. There is a paucity of research regarding the interactions between cytotoxic chemotherapy and cannabis. Most studies focused on CBD due to its inhibition of the CYP450 pathway, which is used for metabolizing cytotoxic chemotherapies. Through this mechanism, CBD could potentially increase the concentrations of chemotherapeutic agents, enhancing their toxicity.

When combined with irinotecan, cannabis can pose risks. Δ-9-THC undergoes first-pass metabolism in the liver, mediated by the CYP450 system and CYP3A4. The glucuronidation of irinotecan is mediated by uridine diphosphate glycosyltransferase, leading to its recirculation within the hepatic system and potentially increased toxicity due to prolonged drug presence. Cannabis may also compete with drug binding to albumin, altering the plasma concentrations of irinotecan and its conversion to the metabolite SN38.¹⁶

Cannabis products can affect chemotherapy levels by interacting with cellular transporters. The MRP1 transporter family, encoded by the ABCC gene family, is expressed mainly in the lung, kidney, skeletal muscle, and hematopoietic stem cells. A 2018 study investigating the effects of THC, CBD, and CBN on MRP1 transporters found that the presence of a cannabis component increased the concentration of vincristine 3-fold. Additional studies suggest the interaction with the CB1 receptor may lead to changes in the expression of MRP1 transporters.¹⁷

CBD inhibits the BCRP transporter, which functions as an efflux pump for methotrexate. Consequently, CBD can increase methotrexate levels, potentially enhancing efficacy but also worsening adverse effects.¹⁸

In pancreatic cancer, CBD specifically interacts with gemcitabine. CB1 and CB2 receptors are upregulated, and CBD inhibits the GPR55 receptor. These interactions may enhance the antineoplastic effect of gemcitabine, reducing cell cycle progression and growth.¹⁹

CBD also interacts with temozolomide (TMZ) by affecting extracellular vesicles used by cells for pro-oncogenic signaling and immune system evasion. Experiments on patient-derived glioblastoma cells, both chemotherapy-resistant and chemotherapy-sensitive, found that CBD increases the formation of extracellular vesicles with reduced levels of miR21 (pro-oncogenic) and elevated levels of miR126 (antioncogenic).²⁰ CBD has also been found to decrease prohibitin levels, a protein associated with TMZ resistance.

In patients with glioblastoma, CBD combined with chemotherapeutic agents like TMZ, carmustine, doxorubicin, and cisplatin has shown increased sensitivity and improved tumor response. CBD is also known to inhibit NF-kB, a pathway that sustains tumor viability despite chemotherapy.²¹ Additionally, CBD inhibits the P-glycoprotein system, affecting chemotherapy efflux from neoplastic cells.¹⁴ In vitro studies have found that CBD is synergistic with bortezomib in inhibiting cancer cell viability. In another glioblastoma model, CBD enhanced the antiproliferative effects of both TMZ and carmustine.¹⁴

Different cannabis formulations may vary in how they interact with various cytotoxic chemotherapeutic agents. Some may potentiate the effects of chemotherapy and act synergistically to inhibit tumor growth, while others may lead to increased toxicity.¹⁰ More research is needed to determine which formulations, in combination with specific agents and doses, may have significant interactions that warrant adjustments in chemotherapy dosing.

Cannabis/Immunotherapy Interactions

Cannabis is an immunosuppressant. Data suggest the use of cannabis during immunotherapy worsens treatment outcomes in patients with cancer.²² Exogenous (THC) and endogenous (AEA) CBs negatively affect antitumor immunity by impairing the function of tumor-specific T cells via CB2 and by inhibiting the Jak1-STATs signaling in T cells through CNR2. Xiong et al found that THC reduces the therapeutic effect of anti-PD-1 therapy.²²

In a prospective observational clinical study, Bar-Sela et al analyzed 102 patients with advanced cancer—of which 68 were cannabis users—that were started on immune checkpoint inhibitor therapy. The study found that cannabis users on anti-PD-1 (nivolumab, pembrolizumab), anti-CTLA-4 (ipilimumab), and anti-PD-L1 (durvalumab, atezolizumab) had a significant decrease in time to treatment progression and overall survival vs cannabis non-users.²³ However, a 2023 study by Waissengrin et al found that concomitant use of medical cannabis with pembrolizumab had no harmful effect in advanced non-small cell lung cancer.²⁴ Time to treatment progression of cannabis users did not differ from cannabis nonusers.²⁵

Cannabis/Endocrine Therapy Interactions

In addition to having direct antineoplastic activity on tumor cells, data exist that show how cannabis affects the endocrine system. In animal models, cannabis has been found to suppress the whole hypothalamic-pituitary-adrenal axis as well as other hormones like thyroid, prolactin, and growth hormone. In breast cancer, cannabis competes with estrogen for the estrogen receptor and suppresses growth.²⁶

The endocrine agents used by patients with cancer in this study were antiandrogens like abiraterone, enzalutamide, tamoxifen and anastrozole. Abiraterone is metabolized by CYP450 isoenzymes and uridine diphosphate glycosyltransferases. Cannabis inhibits both processes and therefore may lead to increased toxicities.²⁷ Conversely, enzalutamide is a strong CYP3A inducer, and cannabis use during enzalutamide therapy may significantly increase the toxic effects of cannabis.

There is evidence that molecular pathways involving CB receptors and estrogens overlap, which may lead to interactions when antiestrogens are used in cannabis users with hormone receptor-positive breast cancer.²⁶ In preclinical studies, tamoxifen has been shown to act as an inverse agonist on CB1 and CB2 receptors, though the significance of this finding is unclear. There is no research evaluating the effects of CBs on tamoxifen treatment. However, CBD has been found to potentiate the effectiveness of anastrozole or exemestane in breast cancer cell lines.²⁸ Dobovišek et al demonstrated no inhibitory effect of CBD on the activity of tamoxifen, fulvestrant, or palbociclib in breast cancer cell lines.²⁹ The interactions between hormone receptor-positive breast cancer and cannabinoids are complex, and the clinical significance of these interactions remains difficult to identify.

Cannabis/Targeted Therapy Interactions

The targeted therapies used by patients in this study included zanubrutinib, ibrutinib, sorafenib, acalabrutinib, dabrafenib, trametinib, trastuzumab, bevacizumab, daratumumab, and imatinib. Compared to other classes of cancer treatments, most studies have not demonstrated decreased efficacy or increased toxicity of targeted anticancer drugs when used concomitantly with CBD.²⁹

Trastuzumab is a recombinant humanized monoclonal antibody that targets the proto-oncogene HER2/neu. It is used to treat select patients with metastatic breast cancer. Studies have shown that cannabis use does not attenuate the effectiveness of trastuzumab in HER2-positive and triple-negative breast cancer subtypes.²⁹ One study found that CBD, in combination with chemotherapeutics and Bruton tyrosine kinase inhibitors, such as ibrutinib and zanubrutinib, has synergistic potential for treating diffuse large B-cell lymphoma and mantle cell lymphoma cell lines. This synergy is attributed to the CB1 antagonist activity of cannabis against diffuse large B-cell lymphoma and mantle cell lymphoma cell lines.^30,31

Moreover, combining cannabinoids with bevacizumab (a monoclonal anti-VEGF antibody) has been shown to decrease tumor growth and intratumoral hypoxia in clinically relevant human glioblastoma models. This effect is mediated through the downregulation of HIF-1α.³² Long-term studies evaluating the potential harmful or synergistic potential of CBD on targeted anticancer therapy are needed.

CONCLUSIONS

This exploratory study of patients receiving cancer therapy at WJVAMC found a significant prevalence of concurrent cannabis use among patients undergoing antineoplastic treatments. Given that many antineoplastic agents are metabolized by the CYP450 enzyme system, the findings of this study suggest that concurrent cannabis use may pose risks of suboptimal therapeutic outcomes due to potential interactions affecting drug metabolism. These interactions could impact the efficacy and toxicity of the antineoplastic therapies, potentially leading to diminished therapeutic effects or exacerbated adverse reactions.

Patients should be informed regarding the potential decreased efficacy of immunotherapy with concurrent use of cannabis products. They should also be aware of the possibility of increased toxicity with other treatment modalities, though the exact impact on efficacy remains unclear. This highlights the necessity of caution when combining cannabis with prescribed cancer treatments.

While this study identified possible interactions, its data are preliminary and highlight the need for more rigorous research. Future studies should include larger, well-designed cohorts to compare outcomes between cannabis users and nonusers. Such research is essential to fully elucidate the clinical implications of cannabis use during cancer treatment, address the high prevalence of cannabis use among patients with cancer, and mitigate potential risks associated with combining cannabis products with antineoplastic therapies. This will ensure that treatment strategies are optimized for safety and efficacy in this complex patient population.

Cannabis has a long history of use for medicinal and recreational purposes. Research illustrates the potential benefits and increased prevalence of cannabis use in patients with cancer.¹ Cannabis products have been shown to possess antineoplastic and palliative activity, improving nociceptive and neuropathic pain in addition to chemotherapy-related nausea and vomiting.^2-5 Despite these developments and changing social attitudes toward cannabis, there remains a lack of comprehensive data on patient perspectives regarding its use, especially in regions where cannabis remains illegal. This knowledge gap is notable among veterans undergoing cancer treatment in states where cannabis is prohibited. Up to 57% of veterans report lifetime marijuana use, making it crucial to understand this population’s cannabis use patterns and potential interactions with cancer treatments.⁶

This observational study sought to determine the prevalence of cannabis use among patients undergoing cancer treatment at the US Department of Veterans Affairs (VA) Memphis Healthcare System and evaluate the potential risks associated with combining cannabis products with anticancer therapies.

METHODS

This prospective observational study identified cannabis use among veterans receiving antineoplastic therapy at the Lt. Col. Luke Weathers Jr. VA Medical Center (WJVAMC) and analyzed potential interactions between cannabis products and their cancer treatments. Participants included adults aged > 18 years undergoing antineoplastic therapy at WJVAMC who consented to the study. Data collection involved a written survey approved by the WJVAMC Institutional Review Board and verbal consent from participants. The survey asked participants about their cannabis use in the previous 90 days, including details on quantity, frequency, and method of consumption (eg, inhalation, oral, topical). No incentives were offered for participation.

Surveys from 50 patients who used cannabis were analyzed and their electronic health records were reviewed for sex, age, diagnosis, and antineoplastic regimen. This information was securely stored. A literature review was conducted using PubMed and the Cochrane Library to explore potential interactions between cannabis and the antineoplastic agents that were prescribed to patients in the study, focusing on toxicity, efficacy, or synergistic effects.

Patients were categorized into 4 groups based on treatment: cytotoxic chemotherapy, immunotherapy, endocrine therapy, and targeted therapy. Patients undergoing multiple types of therapies were included in each applicable category.

RESULTS

A total of 132 patients agreed to participate. Fifty patients (38%) acknowledged using cannabis products within 90 days. The patients that used cannabis products within 90 days of the survey reported the following malignancies: 8 patients (16%) had prostate cancer, 3 patients (6%) had hepatocellular carcinoma, 7 patients (14%) had pancreatic carcinoma, 5 patients (10%) had multiple myeloma, 3 patients (6%) had chronic lymphocytic leukemia, 9 patients (18%) had non-small cell lung cancer, 3 patients (6%) had breast cancer, 3 (6%) patients had bladder cancer, 2 patients (4%) had renal cell carcinoma, 1 (2%) patient had chronic myeloid leukemia, 1 (2%) patient had renal amyloid, 1 patient (2%) had supraglottic squamous cell carcinoma, 1 patient (2%) had esophageal carcinoma, 1 (2%) patient had small cell lung cancer, 1 (2%) patient had gastric cancer, and 1 patient (2%) had follicular lymphoma.

Five (10%) of the cannabis users were female, and 45 (90%) were male. Twenty-nine patients (58%) were aged 66 to 75 years, 16 (32%) were aged 56 to 65 years, 3 (6%) were aged 46 to 55 years, and 2 (4%) were aged 76 to 85 years.

Thirty-five patients (70%) inhaled cannabis as opposed to using it via other formulations or a combination (eg, inhalation and topical). Thirty-eight percent of patients used cannabis once daily, 24% used < 1 daily, and 28% used it ≥ 2 times daily. Five patients (10%) did not report the frequency of their cannabis use. Among the patients who reported cannabis use, 21 (42%) were undergoing cytotoxic chemotherapy, 19 (38%) were undergoing immunotherapy, 12 (24%) were undergoing targeted therapy, and 10 (20%) were undergoing endocrine therapy. Some patients were treated with multiple types of antineoplastic agents and were counted in multiple categories (Table 1).

Following a literature review of cannabis and antineoplastic agents, patients were evaluated for the potential effects of cannabis on their treatment. The literature review revealed that 31% of cytotoxic chemotherapy agents received by patients in this study might have increased toxicity, and 19% could have reduced efficacy when combined with cannabis. Among immunotherapy agents received by patients in this study, 70% might have decreased efficacy when combined with cannabis use. For targeted therapies, 35% could have increased toxicity, and 70% of endocrine agents could potentially have decreased efficacy (Table 2).

DISCUSSION

This prospective study corroborates previous research by demonstrating that more than one-third of patients receiving oncology care at WJVAMC use cannabis, most often inhaled. Cannabis use was observed among patients undergoing various cancer therapies, including cytotoxic chemotherapy, immunotherapy, targeted therapy, and endocrine therapy. The most common malignancies among cannabis users at WJVAMC include patients with lung cancer, prostate cancer, pancreatic cancer, and multiple myeloma. Cannabis use in patients with pancreatic cancer and multiple myeloma was significantly out of proportion to their prevalence at WJVAMC. This could potentially be due to their drastic effect on quality of life.

Cannabis use increased the risk of toxicity in patients treated with cytotoxic chemotherapy and targeted therapy. Cannabis use potentially decreased efficacy for patients treated with cytotoxic chemotherapy and/or immunotherapy. Cannabis use did not increase the risk of toxicity or efficacy in patients treated with endocrine therapy.

Antineoplastics/Cannabis Interactions

The potential interactions between cannabis and antineoplastic therapies administered at WJVAMC are worth exploring. While this review aims to shed light on possible interactions, it is important to acknowledge that much of the data is preliminary and derived from in vitro studies. The interactions should be interpreted as potential risks rather than established facts. Additional research is needed to confirm these interactions and effectively guide clinical practices. Understanding these dynamics is essential to optimize patient care and manage the complex interplay between cannabis use and cancer treatment.

Originating from Central Asia, the cannabis plant contains > 400 medicinally relevant compounds, of which about 100 are cannabinoids (CBs). Key CBs are cannabidiol (CBD), a nonpsychoactive compound, and ?-9-tetrahydrocannabinol (THC), a psychoactive compound. THC can make up 20% to 30% of the dry weight of female cannabis flowers.⁷

CBs act through the endocannabinoid system, involving CB1 and CB2 receptors, endogenous CBs like anandamide (AEA) and 2-arachidonoylglycerol, and various enzymes. These endogenous CBs, derived from arachidonic acid, play roles in cell growth and proliferation.⁸ In some studies, AEA has induced apoptosis in neuroblastoma cells and inhibited proliferation in breast cancer cells. However, other research suggests AEA may block apoptosis under certain conditions.⁹

CB receptors are transmembrane proteins that interact with CBs differently depending on tissue type and CB structure. Synthetic CBs are designed to target specific receptors, while natural CBs may act as both agonists and antagonists.¹⁰

Cytochrome P450 Metabolism

The human cytochrome P450 (CYP) 3A subfamily affects the metabolism of many therapeutic drugs, including cancer therapeutics.¹¹ The various compositions of cannabis are primarily metabolized by the CYP450 pathway, the same as many cancer-directed pharmacologic treatments. CBs act as both CYP inducers and inhibitors. THC, for example, is a CYP inducer whereas CBD is a CYP inhibitor; both are found in the various compounds available for consumption.^12,13 Pharmacology research has suggested potential interactions and effects on established adverse symptoms, but clinical data are lacking, and current research revealing interactions are only recognized in vitro.¹⁴

The Antineoplastic Activity of Cannabis

CBs can affect various cancer-related pathways such as PKB, AMPK, CAMKK-ß, mTOR, PDHK, HIF-1 a, and PPAR-γ. Δ-9-THC can selectively induce apoptosis in tumor cells without harming normal cells, though the exact mechanism remains unclear. Promising results from early mouse studies led to a 2006 human study where intracranial Δ-9-THC in patients with recurrent glioma yielded a median survival of 24 weeks, with 2 patients surviving > 1 year.¹⁵

In a 2022 review article, Cherkasova et al highlighted potential clinical benefits of cannabis across various cancers. They found that upregulated CB1 receptors in colon cancer might enhance the effect of 5-fluorouracil. However, many studies are preliminary and therefore not definitive.¹⁰

Additional research is needed to refine these findings. Challenges include variability in cannabis formulations, the complex tumor microenvironment, and the legal and psychoactive issues surrounding cannabis use. These factors complicate the design of multicenter randomized studies and may deter patients from disclosing cannabis use, thereby hindering efforts to fully understand its therapeutic potential.

Cannabis/Cytotoxic Chemotherapy Interactions

The chemotherapy agents used in this study included carboplatin, paclitaxel, 5-fluorouracil, etoposide, irinotecan, oxaliplatin, pemetrexed, docetaxel, cabazitaxel, T-DM1, gemcitabine, and cyclophosphamide. There is a paucity of research regarding the interactions between cytotoxic chemotherapy and cannabis. Most studies focused on CBD due to its inhibition of the CYP450 pathway, which is used for metabolizing cytotoxic chemotherapies. Through this mechanism, CBD could potentially increase the concentrations of chemotherapeutic agents, enhancing their toxicity.

When combined with irinotecan, cannabis can pose risks. Δ-9-THC undergoes first-pass metabolism in the liver, mediated by the CYP450 system and CYP3A4. The glucuronidation of irinotecan is mediated by uridine diphosphate glycosyltransferase, leading to its recirculation within the hepatic system and potentially increased toxicity due to prolonged drug presence. Cannabis may also compete with drug binding to albumin, altering the plasma concentrations of irinotecan and its conversion to the metabolite SN38.¹⁶

Cannabis products can affect chemotherapy levels by interacting with cellular transporters. The MRP1 transporter family, encoded by the ABCC gene family, is expressed mainly in the lung, kidney, skeletal muscle, and hematopoietic stem cells. A 2018 study investigating the effects of THC, CBD, and CBN on MRP1 transporters found that the presence of a cannabis component increased the concentration of vincristine 3-fold. Additional studies suggest the interaction with the CB1 receptor may lead to changes in the expression of MRP1 transporters.¹⁷

CBD inhibits the BCRP transporter, which functions as an efflux pump for methotrexate. Consequently, CBD can increase methotrexate levels, potentially enhancing efficacy but also worsening adverse effects.¹⁸

In pancreatic cancer, CBD specifically interacts with gemcitabine. CB1 and CB2 receptors are upregulated, and CBD inhibits the GPR55 receptor. These interactions may enhance the antineoplastic effect of gemcitabine, reducing cell cycle progression and growth.¹⁹

CBD also interacts with temozolomide (TMZ) by affecting extracellular vesicles used by cells for pro-oncogenic signaling and immune system evasion. Experiments on patient-derived glioblastoma cells, both chemotherapy-resistant and chemotherapy-sensitive, found that CBD increases the formation of extracellular vesicles with reduced levels of miR21 (pro-oncogenic) and elevated levels of miR126 (antioncogenic).²⁰ CBD has also been found to decrease prohibitin levels, a protein associated with TMZ resistance.

In patients with glioblastoma, CBD combined with chemotherapeutic agents like TMZ, carmustine, doxorubicin, and cisplatin has shown increased sensitivity and improved tumor response. CBD is also known to inhibit NF-kB, a pathway that sustains tumor viability despite chemotherapy.²¹ Additionally, CBD inhibits the P-glycoprotein system, affecting chemotherapy efflux from neoplastic cells.¹⁴ In vitro studies have found that CBD is synergistic with bortezomib in inhibiting cancer cell viability. In another glioblastoma model, CBD enhanced the antiproliferative effects of both TMZ and carmustine.¹⁴

Different cannabis formulations may vary in how they interact with various cytotoxic chemotherapeutic agents. Some may potentiate the effects of chemotherapy and act synergistically to inhibit tumor growth, while others may lead to increased toxicity.¹⁰ More research is needed to determine which formulations, in combination with specific agents and doses, may have significant interactions that warrant adjustments in chemotherapy dosing.

Cannabis/Immunotherapy Interactions

Cannabis is an immunosuppressant. Data suggest the use of cannabis during immunotherapy worsens treatment outcomes in patients with cancer.²² Exogenous (THC) and endogenous (AEA) CBs negatively affect antitumor immunity by impairing the function of tumor-specific T cells via CB2 and by inhibiting the Jak1-STATs signaling in T cells through CNR2. Xiong et al found that THC reduces the therapeutic effect of anti-PD-1 therapy.²²

In a prospective observational clinical study, Bar-Sela et al analyzed 102 patients with advanced cancer—of which 68 were cannabis users—that were started on immune checkpoint inhibitor therapy. The study found that cannabis users on anti-PD-1 (nivolumab, pembrolizumab), anti-CTLA-4 (ipilimumab), and anti-PD-L1 (durvalumab, atezolizumab) had a significant decrease in time to treatment progression and overall survival vs cannabis non-users.²³ However, a 2023 study by Waissengrin et al found that concomitant use of medical cannabis with pembrolizumab had no harmful effect in advanced non-small cell lung cancer.²⁴ Time to treatment progression of cannabis users did not differ from cannabis nonusers.²⁵

Cannabis/Endocrine Therapy Interactions

In addition to having direct antineoplastic activity on tumor cells, data exist that show how cannabis affects the endocrine system. In animal models, cannabis has been found to suppress the whole hypothalamic-pituitary-adrenal axis as well as other hormones like thyroid, prolactin, and growth hormone. In breast cancer, cannabis competes with estrogen for the estrogen receptor and suppresses growth.²⁶

The endocrine agents used by patients with cancer in this study were antiandrogens like abiraterone, enzalutamide, tamoxifen and anastrozole. Abiraterone is metabolized by CYP450 isoenzymes and uridine diphosphate glycosyltransferases. Cannabis inhibits both processes and therefore may lead to increased toxicities.²⁷ Conversely, enzalutamide is a strong CYP3A inducer, and cannabis use during enzalutamide therapy may significantly increase the toxic effects of cannabis.

There is evidence that molecular pathways involving CB receptors and estrogens overlap, which may lead to interactions when antiestrogens are used in cannabis users with hormone receptor-positive breast cancer.²⁶ In preclinical studies, tamoxifen has been shown to act as an inverse agonist on CB1 and CB2 receptors, though the significance of this finding is unclear. There is no research evaluating the effects of CBs on tamoxifen treatment. However, CBD has been found to potentiate the effectiveness of anastrozole or exemestane in breast cancer cell lines.²⁸ Dobovišek et al demonstrated no inhibitory effect of CBD on the activity of tamoxifen, fulvestrant, or palbociclib in breast cancer cell lines.²⁹ The interactions between hormone receptor-positive breast cancer and cannabinoids are complex, and the clinical significance of these interactions remains difficult to identify.

Cannabis/Targeted Therapy Interactions

The targeted therapies used by patients in this study included zanubrutinib, ibrutinib, sorafenib, acalabrutinib, dabrafenib, trametinib, trastuzumab, bevacizumab, daratumumab, and imatinib. Compared to other classes of cancer treatments, most studies have not demonstrated decreased efficacy or increased toxicity of targeted anticancer drugs when used concomitantly with CBD.²⁹

Trastuzumab is a recombinant humanized monoclonal antibody that targets the proto-oncogene HER2/neu. It is used to treat select patients with metastatic breast cancer. Studies have shown that cannabis use does not attenuate the effectiveness of trastuzumab in HER2-positive and triple-negative breast cancer subtypes.²⁹ One study found that CBD, in combination with chemotherapeutics and Bruton tyrosine kinase inhibitors, such as ibrutinib and zanubrutinib, has synergistic potential for treating diffuse large B-cell lymphoma and mantle cell lymphoma cell lines. This synergy is attributed to the CB1 antagonist activity of cannabis against diffuse large B-cell lymphoma and mantle cell lymphoma cell lines.^30,31

Moreover, combining cannabinoids with bevacizumab (a monoclonal anti-VEGF antibody) has been shown to decrease tumor growth and intratumoral hypoxia in clinically relevant human glioblastoma models. This effect is mediated through the downregulation of HIF-1α.³² Long-term studies evaluating the potential harmful or synergistic potential of CBD on targeted anticancer therapy are needed.

CONCLUSIONS

This exploratory study of patients receiving cancer therapy at WJVAMC found a significant prevalence of concurrent cannabis use among patients undergoing antineoplastic treatments. Given that many antineoplastic agents are metabolized by the CYP450 enzyme system, the findings of this study suggest that concurrent cannabis use may pose risks of suboptimal therapeutic outcomes due to potential interactions affecting drug metabolism. These interactions could impact the efficacy and toxicity of the antineoplastic therapies, potentially leading to diminished therapeutic effects or exacerbated adverse reactions.

Patients should be informed regarding the potential decreased efficacy of immunotherapy with concurrent use of cannabis products. They should also be aware of the possibility of increased toxicity with other treatment modalities, though the exact impact on efficacy remains unclear. This highlights the necessity of caution when combining cannabis with prescribed cancer treatments.

While this study identified possible interactions, its data are preliminary and highlight the need for more rigorous research. Future studies should include larger, well-designed cohorts to compare outcomes between cannabis users and nonusers. Such research is essential to fully elucidate the clinical implications of cannabis use during cancer treatment, address the high prevalence of cannabis use among patients with cancer, and mitigate potential risks associated with combining cannabis products with antineoplastic therapies. This will ensure that treatment strategies are optimized for safety and efficacy in this complex patient population.

Cannabis has a long history of use for medicinal and recreational purposes. Research illustrates the potential benefits and increased prevalence of cannabis use in patients with cancer.¹ Cannabis products have been shown to possess antineoplastic and palliative activity, improving nociceptive and neuropathic pain in addition to chemotherapy-related nausea and vomiting.^2-5 Despite these developments and changing social attitudes toward cannabis, there remains a lack of comprehensive data on patient perspectives regarding its use, especially in regions where cannabis remains illegal. This knowledge gap is notable among veterans undergoing cancer treatment in states where cannabis is prohibited. Up to 57% of veterans report lifetime marijuana use, making it crucial to understand this population’s cannabis use patterns and potential interactions with cancer treatments.⁶

This observational study sought to determine the prevalence of cannabis use among patients undergoing cancer treatment at the US Department of Veterans Affairs (VA) Memphis Healthcare System and evaluate the potential risks associated with combining cannabis products with anticancer therapies.

METHODS

This prospective observational study identified cannabis use among veterans receiving antineoplastic therapy at the Lt. Col. Luke Weathers Jr. VA Medical Center (WJVAMC) and analyzed potential interactions between cannabis products and their cancer treatments. Participants included adults aged > 18 years undergoing antineoplastic therapy at WJVAMC who consented to the study. Data collection involved a written survey approved by the WJVAMC Institutional Review Board and verbal consent from participants. The survey asked participants about their cannabis use in the previous 90 days, including details on quantity, frequency, and method of consumption (eg, inhalation, oral, topical). No incentives were offered for participation.

Surveys from 50 patients who used cannabis were analyzed and their electronic health records were reviewed for sex, age, diagnosis, and antineoplastic regimen. This information was securely stored. A literature review was conducted using PubMed and the Cochrane Library to explore potential interactions between cannabis and the antineoplastic agents that were prescribed to patients in the study, focusing on toxicity, efficacy, or synergistic effects.

Patients were categorized into 4 groups based on treatment: cytotoxic chemotherapy, immunotherapy, endocrine therapy, and targeted therapy. Patients undergoing multiple types of therapies were included in each applicable category.

RESULTS

A total of 132 patients agreed to participate. Fifty patients (38%) acknowledged using cannabis products within 90 days. The patients that used cannabis products within 90 days of the survey reported the following malignancies: 8 patients (16%) had prostate cancer, 3 patients (6%) had hepatocellular carcinoma, 7 patients (14%) had pancreatic carcinoma, 5 patients (10%) had multiple myeloma, 3 patients (6%) had chronic lymphocytic leukemia, 9 patients (18%) had non-small cell lung cancer, 3 patients (6%) had breast cancer, 3 (6%) patients had bladder cancer, 2 patients (4%) had renal cell carcinoma, 1 (2%) patient had chronic myeloid leukemia, 1 (2%) patient had renal amyloid, 1 patient (2%) had supraglottic squamous cell carcinoma, 1 patient (2%) had esophageal carcinoma, 1 (2%) patient had small cell lung cancer, 1 (2%) patient had gastric cancer, and 1 patient (2%) had follicular lymphoma.

Five (10%) of the cannabis users were female, and 45 (90%) were male. Twenty-nine patients (58%) were aged 66 to 75 years, 16 (32%) were aged 56 to 65 years, 3 (6%) were aged 46 to 55 years, and 2 (4%) were aged 76 to 85 years.

Thirty-five patients (70%) inhaled cannabis as opposed to using it via other formulations or a combination (eg, inhalation and topical). Thirty-eight percent of patients used cannabis once daily, 24% used < 1 daily, and 28% used it ≥ 2 times daily. Five patients (10%) did not report the frequency of their cannabis use. Among the patients who reported cannabis use, 21 (42%) were undergoing cytotoxic chemotherapy, 19 (38%) were undergoing immunotherapy, 12 (24%) were undergoing targeted therapy, and 10 (20%) were undergoing endocrine therapy. Some patients were treated with multiple types of antineoplastic agents and were counted in multiple categories (Table 1).

Following a literature review of cannabis and antineoplastic agents, patients were evaluated for the potential effects of cannabis on their treatment. The literature review revealed that 31% of cytotoxic chemotherapy agents received by patients in this study might have increased toxicity, and 19% could have reduced efficacy when combined with cannabis. Among immunotherapy agents received by patients in this study, 70% might have decreased efficacy when combined with cannabis use. For targeted therapies, 35% could have increased toxicity, and 70% of endocrine agents could potentially have decreased efficacy (Table 2).

DISCUSSION

This prospective study corroborates previous research by demonstrating that more than one-third of patients receiving oncology care at WJVAMC use cannabis, most often inhaled. Cannabis use was observed among patients undergoing various cancer therapies, including cytotoxic chemotherapy, immunotherapy, targeted therapy, and endocrine therapy. The most common malignancies among cannabis users at WJVAMC include patients with lung cancer, prostate cancer, pancreatic cancer, and multiple myeloma. Cannabis use in patients with pancreatic cancer and multiple myeloma was significantly out of proportion to their prevalence at WJVAMC. This could potentially be due to their drastic effect on quality of life.

Cannabis use increased the risk of toxicity in patients treated with cytotoxic chemotherapy and targeted therapy. Cannabis use potentially decreased efficacy for patients treated with cytotoxic chemotherapy and/or immunotherapy. Cannabis use did not increase the risk of toxicity or efficacy in patients treated with endocrine therapy.

Antineoplastics/Cannabis Interactions

The potential interactions between cannabis and antineoplastic therapies administered at WJVAMC are worth exploring. While this review aims to shed light on possible interactions, it is important to acknowledge that much of the data is preliminary and derived from in vitro studies. The interactions should be interpreted as potential risks rather than established facts. Additional research is needed to confirm these interactions and effectively guide clinical practices. Understanding these dynamics is essential to optimize patient care and manage the complex interplay between cannabis use and cancer treatment.

Originating from Central Asia, the cannabis plant contains > 400 medicinally relevant compounds, of which about 100 are cannabinoids (CBs). Key CBs are cannabidiol (CBD), a nonpsychoactive compound, and ?-9-tetrahydrocannabinol (THC), a psychoactive compound. THC can make up 20% to 30% of the dry weight of female cannabis flowers.⁷

CBs act through the endocannabinoid system, involving CB1 and CB2 receptors, endogenous CBs like anandamide (AEA) and 2-arachidonoylglycerol, and various enzymes. These endogenous CBs, derived from arachidonic acid, play roles in cell growth and proliferation.⁸ In some studies, AEA has induced apoptosis in neuroblastoma cells and inhibited proliferation in breast cancer cells. However, other research suggests AEA may block apoptosis under certain conditions.⁹

CB receptors are transmembrane proteins that interact with CBs differently depending on tissue type and CB structure. Synthetic CBs are designed to target specific receptors, while natural CBs may act as both agonists and antagonists.¹⁰

Cytochrome P450 Metabolism

The human cytochrome P450 (CYP) 3A subfamily affects the metabolism of many therapeutic drugs, including cancer therapeutics.¹¹ The various compositions of cannabis are primarily metabolized by the CYP450 pathway, the same as many cancer-directed pharmacologic treatments. CBs act as both CYP inducers and inhibitors. THC, for example, is a CYP inducer whereas CBD is a CYP inhibitor; both are found in the various compounds available for consumption.^12,13 Pharmacology research has suggested potential interactions and effects on established adverse symptoms, but clinical data are lacking, and current research revealing interactions are only recognized in vitro.¹⁴

The Antineoplastic Activity of Cannabis

CBs can affect various cancer-related pathways such as PKB, AMPK, CAMKK-ß, mTOR, PDHK, HIF-1 a, and PPAR-γ. Δ-9-THC can selectively induce apoptosis in tumor cells without harming normal cells, though the exact mechanism remains unclear. Promising results from early mouse studies led to a 2006 human study where intracranial Δ-9-THC in patients with recurrent glioma yielded a median survival of 24 weeks, with 2 patients surviving > 1 year.¹⁵

In a 2022 review article, Cherkasova et al highlighted potential clinical benefits of cannabis across various cancers. They found that upregulated CB1 receptors in colon cancer might enhance the effect of 5-fluorouracil. However, many studies are preliminary and therefore not definitive.¹⁰

Additional research is needed to refine these findings. Challenges include variability in cannabis formulations, the complex tumor microenvironment, and the legal and psychoactive issues surrounding cannabis use. These factors complicate the design of multicenter randomized studies and may deter patients from disclosing cannabis use, thereby hindering efforts to fully understand its therapeutic potential.

Cannabis/Cytotoxic Chemotherapy Interactions

The chemotherapy agents used in this study included carboplatin, paclitaxel, 5-fluorouracil, etoposide, irinotecan, oxaliplatin, pemetrexed, docetaxel, cabazitaxel, T-DM1, gemcitabine, and cyclophosphamide. There is a paucity of research regarding the interactions between cytotoxic chemotherapy and cannabis. Most studies focused on CBD due to its inhibition of the CYP450 pathway, which is used for metabolizing cytotoxic chemotherapies. Through this mechanism, CBD could potentially increase the concentrations of chemotherapeutic agents, enhancing their toxicity.

When combined with irinotecan, cannabis can pose risks. Δ-9-THC undergoes first-pass metabolism in the liver, mediated by the CYP450 system and CYP3A4. The glucuronidation of irinotecan is mediated by uridine diphosphate glycosyltransferase, leading to its recirculation within the hepatic system and potentially increased toxicity due to prolonged drug presence. Cannabis may also compete with drug binding to albumin, altering the plasma concentrations of irinotecan and its conversion to the metabolite SN38.¹⁶

Cannabis products can affect chemotherapy levels by interacting with cellular transporters. The MRP1 transporter family, encoded by the ABCC gene family, is expressed mainly in the lung, kidney, skeletal muscle, and hematopoietic stem cells. A 2018 study investigating the effects of THC, CBD, and CBN on MRP1 transporters found that the presence of a cannabis component increased the concentration of vincristine 3-fold. Additional studies suggest the interaction with the CB1 receptor may lead to changes in the expression of MRP1 transporters.¹⁷

CBD inhibits the BCRP transporter, which functions as an efflux pump for methotrexate. Consequently, CBD can increase methotrexate levels, potentially enhancing efficacy but also worsening adverse effects.¹⁸

In pancreatic cancer, CBD specifically interacts with gemcitabine. CB1 and CB2 receptors are upregulated, and CBD inhibits the GPR55 receptor. These interactions may enhance the antineoplastic effect of gemcitabine, reducing cell cycle progression and growth.¹⁹

CBD also interacts with temozolomide (TMZ) by affecting extracellular vesicles used by cells for pro-oncogenic signaling and immune system evasion. Experiments on patient-derived glioblastoma cells, both chemotherapy-resistant and chemotherapy-sensitive, found that CBD increases the formation of extracellular vesicles with reduced levels of miR21 (pro-oncogenic) and elevated levels of miR126 (antioncogenic).²⁰ CBD has also been found to decrease prohibitin levels, a protein associated with TMZ resistance.

In patients with glioblastoma, CBD combined with chemotherapeutic agents like TMZ, carmustine, doxorubicin, and cisplatin has shown increased sensitivity and improved tumor response. CBD is also known to inhibit NF-kB, a pathway that sustains tumor viability despite chemotherapy.²¹ Additionally, CBD inhibits the P-glycoprotein system, affecting chemotherapy efflux from neoplastic cells.¹⁴ In vitro studies have found that CBD is synergistic with bortezomib in inhibiting cancer cell viability. In another glioblastoma model, CBD enhanced the antiproliferative effects of both TMZ and carmustine.¹⁴

Different cannabis formulations may vary in how they interact with various cytotoxic chemotherapeutic agents. Some may potentiate the effects of chemotherapy and act synergistically to inhibit tumor growth, while others may lead to increased toxicity.¹⁰ More research is needed to determine which formulations, in combination with specific agents and doses, may have significant interactions that warrant adjustments in chemotherapy dosing.

Cannabis/Immunotherapy Interactions

Cannabis is an immunosuppressant. Data suggest the use of cannabis during immunotherapy worsens treatment outcomes in patients with cancer.²² Exogenous (THC) and endogenous (AEA) CBs negatively affect antitumor immunity by impairing the function of tumor-specific T cells via CB2 and by inhibiting the Jak1-STATs signaling in T cells through CNR2. Xiong et al found that THC reduces the therapeutic effect of anti-PD-1 therapy.²²

In a prospective observational clinical study, Bar-Sela et al analyzed 102 patients with advanced cancer—of which 68 were cannabis users—that were started on immune checkpoint inhibitor therapy. The study found that cannabis users on anti-PD-1 (nivolumab, pembrolizumab), anti-CTLA-4 (ipilimumab), and anti-PD-L1 (durvalumab, atezolizumab) had a significant decrease in time to treatment progression and overall survival vs cannabis non-users.²³ However, a 2023 study by Waissengrin et al found that concomitant use of medical cannabis with pembrolizumab had no harmful effect in advanced non-small cell lung cancer.²⁴ Time to treatment progression of cannabis users did not differ from cannabis nonusers.²⁵

Cannabis/Endocrine Therapy Interactions

In addition to having direct antineoplastic activity on tumor cells, data exist that show how cannabis affects the endocrine system. In animal models, cannabis has been found to suppress the whole hypothalamic-pituitary-adrenal axis as well as other hormones like thyroid, prolactin, and growth hormone. In breast cancer, cannabis competes with estrogen for the estrogen receptor and suppresses growth.²⁶

The endocrine agents used by patients with cancer in this study were antiandrogens like abiraterone, enzalutamide, tamoxifen and anastrozole. Abiraterone is metabolized by CYP450 isoenzymes and uridine diphosphate glycosyltransferases. Cannabis inhibits both processes and therefore may lead to increased toxicities.²⁷ Conversely, enzalutamide is a strong CYP3A inducer, and cannabis use during enzalutamide therapy may significantly increase the toxic effects of cannabis.

There is evidence that molecular pathways involving CB receptors and estrogens overlap, which may lead to interactions when antiestrogens are used in cannabis users with hormone receptor-positive breast cancer.²⁶ In preclinical studies, tamoxifen has been shown to act as an inverse agonist on CB1 and CB2 receptors, though the significance of this finding is unclear. There is no research evaluating the effects of CBs on tamoxifen treatment. However, CBD has been found to potentiate the effectiveness of anastrozole or exemestane in breast cancer cell lines.²⁸ Dobovišek et al demonstrated no inhibitory effect of CBD on the activity of tamoxifen, fulvestrant, or palbociclib in breast cancer cell lines.²⁹ The interactions between hormone receptor-positive breast cancer and cannabinoids are complex, and the clinical significance of these interactions remains difficult to identify.

Cannabis/Targeted Therapy Interactions

The targeted therapies used by patients in this study included zanubrutinib, ibrutinib, sorafenib, acalabrutinib, dabrafenib, trametinib, trastuzumab, bevacizumab, daratumumab, and imatinib. Compared to other classes of cancer treatments, most studies have not demonstrated decreased efficacy or increased toxicity of targeted anticancer drugs when used concomitantly with CBD.²⁹

Trastuzumab is a recombinant humanized monoclonal antibody that targets the proto-oncogene HER2/neu. It is used to treat select patients with metastatic breast cancer. Studies have shown that cannabis use does not attenuate the effectiveness of trastuzumab in HER2-positive and triple-negative breast cancer subtypes.²⁹ One study found that CBD, in combination with chemotherapeutics and Bruton tyrosine kinase inhibitors, such as ibrutinib and zanubrutinib, has synergistic potential for treating diffuse large B-cell lymphoma and mantle cell lymphoma cell lines. This synergy is attributed to the CB1 antagonist activity of cannabis against diffuse large B-cell lymphoma and mantle cell lymphoma cell lines.^30,31

Moreover, combining cannabinoids with bevacizumab (a monoclonal anti-VEGF antibody) has been shown to decrease tumor growth and intratumoral hypoxia in clinically relevant human glioblastoma models. This effect is mediated through the downregulation of HIF-1α.³² Long-term studies evaluating the potential harmful or synergistic potential of CBD on targeted anticancer therapy are needed.

CONCLUSIONS

This exploratory study of patients receiving cancer therapy at WJVAMC found a significant prevalence of concurrent cannabis use among patients undergoing antineoplastic treatments. Given that many antineoplastic agents are metabolized by the CYP450 enzyme system, the findings of this study suggest that concurrent cannabis use may pose risks of suboptimal therapeutic outcomes due to potential interactions affecting drug metabolism. These interactions could impact the efficacy and toxicity of the antineoplastic therapies, potentially leading to diminished therapeutic effects or exacerbated adverse reactions.

Patients should be informed regarding the potential decreased efficacy of immunotherapy with concurrent use of cannabis products. They should also be aware of the possibility of increased toxicity with other treatment modalities, though the exact impact on efficacy remains unclear. This highlights the necessity of caution when combining cannabis with prescribed cancer treatments.

While this study identified possible interactions, its data are preliminary and highlight the need for more rigorous research. Future studies should include larger, well-designed cohorts to compare outcomes between cannabis users and nonusers. Such research is essential to fully elucidate the clinical implications of cannabis use during cancer treatment, address the high prevalence of cannabis use among patients with cancer, and mitigate potential risks associated with combining cannabis products with antineoplastic therapies. This will ensure that treatment strategies are optimized for safety and efficacy in this complex patient population.

Prostate cancer is the most common cancer in US males, with an estimated 313,780 new cases and 35,770 deaths in 2025.¹ Several treatment options are available for localized prostate cancer that have similar outcomes, including active surveillance for low-risk cancers, surgery, or radiotherapy.^2,3 Conventional fractionation radiotherapy (CFRT) with 40 to 45 fractions over 8 to 9 weeks has been used for decades. Over the past 2 decades, moderate hypofractionation schedules with 2.4 to 3.4 Gy per fraction over 20 to 28 fractions have become standard, as many noninferiority randomized clinical trials (RCTs) such as CHHiP (UK),⁴ PROFIT (Canada and Europe),⁵ NRG Oncology RTOG 0415 (US),⁶ HYPRO (Netherlands),^7,8 and HYPO-RT-PC (Sweden and Denmark),⁹ have shown the noninferiority of moderately hypofractionated radiotherapy compared with CFRT. Notably, most of these noninferiority studies primarily included patients with low- or intermediate-risk prostate cancer, except for the HYPO-RT-PC trial,⁹ which also included patients with intermediate- and high-risk prostate cancer.

These noninferiority studies, along with technological advances in radiotherapy, such as intensity-modulated radiotherapy (IMRT), volumetric modulated arc therapy (VMAT), and image-guided radiotherapy (IGRT), paved the path to ultrahypofractionated stereotactic body radiotherapy (SBRT) that is delivered in 5 fractions of ≥ 6 Gy. This high dose per fraction may have a radiobiologic advantage over conventional fractionation. The relatively low a/ß ratio of prostate cancer, estimated to be between 1 and 2, suggests that tumor cells may be particularly sensitive to the high doses per fraction delivered in SBRT.^10-13 Compared with CFRT, SBRT-induced tumor cell death may also be mediated through different pathways; this pathway appears to be generated in a dose-dependent manner, particularly with doses > 8 Gy per fraction.^14,15 Additionally, the higher a/ß ratio for the surrounding organs at risk, such as the bladder and rectum, theoretically allows for an improved therapeutic ratio window that maximizes tumor control while minimizing damage to healthy tissues.

A substantial body of evidence from prospective studies and meta-analyses supports the use of SBRT for localized prostate cancer. HYPO-RT-PC, a significant phase 3 noninferiority study, enrolled 1200 patients with intermediate (89%) and high-risk (11%) prostate cancer randomized between 2 arms, including CFRT to 78 Gy in 39 fractions and SBRT to 42.7 Gy in 7 fractions, treated 3 days weekly. After a median follow-up of 60 months, the estimated 5-year biochemical relapse-free survival rate was 84% in both groups.⁹ This trial was notable because it was the first randomized study to demonstrate that SBRT was noninferior to CFRT in intermediate- and high-risk prostate cancer patients. Another pivotal phase 3 trial, the PACE-B study, enrolled 874 patients to compare SBRT (36.25 Gy to the prostate gland, with a secondary dose of 40 Gy to the gross tumor volume where applicable, in 5 fractions) with CFRT (78 Gy in 39 fractions) and moderately hypofractionated radiotherapy (HFRT) (62 Gy in 20 fractions) in patients with low- or intermediate-risk prostate cancer. With a 74-month median follow-up, the study reported 5-year biochemical free rates of 94.6% for CFRT and 95.8% for SBRT, confirming the noninferiority of SBRT to CFRT.¹⁵

SBRT offers short, effective, and convenient treatment to many patients with localized prostate cancer. While previous guidelines were more restrictive, the March 2026 National Comprehensive Cancer Network (NCCN) guidelines now list SBRT as a preferred treatment modality for high-risk prostate cancer.¹⁶

Given the growing body of evidence supporting the efficacy and safety of SBRT, we implemented an SBRT program in 2014 at a tertiary care center for veterans. This retrospective study was undertaken to evaluate the early efficacy and toxicity of SBRT in patients with localized prostate cancer treated at our institution, including patients across all risk stratifications.

METHODS

We identified 242 patients diagnosed with prostate cancer who underwent SBRT treatment between November 2014 and October 2024 at Overland Park Veterans Affairs Radiation Oncology Clinic. For the final analysis, 46 patients with < 2 years of follow-up and 22 patients who died from causes other than prostate cancer were excluded, resulting in a cohort of 174 patients with ≥ 24-month follow-up.

Treatment

Patients eligible for staging underwent imaging according to NCCN guidelines, including computed tomography (CT) of the abdomen and pelvis, bone scintigraphy, or, in recent years, prostate-specific membrane antigen positron emission tomography, primarily used for unfavorable intermediate-risk (UIR) and high-risk (HR) cancers. Patients with a negative staging work-up for nodal or skeletal disease were included. Prior to planning the CT simulation, patients were given bowel preparation instructions, including a low-fiber and low-gas-producing diet, simethicone, and enemas, the night before and morning of the simulation. Patients were instructed to arrive with a comfortably full bladder, having not voided for 2 to 3 hours prior to the procedure. At Kansas City Veterans Affairs Medical Center (KCVAMC), SBRT treatment was generally restricted to patients with a baseline American Urological Association symptom score of 15 to 20 out of 35 and a prostate gland size < 80 mL to minimize the risk of acute urinary toxicity. We did not use intraprostatic fiducials, hydrogel rectal spacers, or intravenous contrast agents for planning CT simulation.

Patients were placed in a supine position, and a vacuum bag was used for immobilization. Following the CT simulation, the images were transferred to the Eclipse treatment planning system. The clinical target volume (CTV) encompassed the prostate and the proximal 1.0 cm of the seminal vesicles for Gleason score (GS) 1 to 2, and the entire seminal vesicle was included for GS 3 to 5, which is consistent with KCVAMC practice and established safety protocols. The planning target volume (PTV) was created by uniformly expanding the CTV by 5 to 7 mm, except for the posterior margin, which was limited to 3 to 5 mm. When elective nodal radiotherapy was planned for HR prostate cancer, the pelvic field for CT simulation started at the L-2 upper border, with the lower border extending to the lesser trochanter. The pelvic nodes were delineated per Radiation Therapy Oncology Group (RTOG) guidelines.¹⁷ The CTV nodes (CTVn), including common iliac, external and internal iliac nodes, obturator, and presacral nodes, were created by uniformly expanding the CTVn by 2 to 3 mm. Slice-by-slice corrections were made to avoid bowel overlap in these patients.

The use of androgen deprivation therapy (ADT) for a duration of 6 to 24 months was prescribed for patients with UIR or HR prostate cancer per NCCN guidelines.¹⁶ The prescribed dose to the PTV was 36.25 to 40 Gy (40 Gy was mostly used as a boost to the dominant lesion) in 5 fractions, with each fraction ranging from 7.25 to 8 Gy. For elective nodal radiotherapy in patients at HR, the prescribed dose was 25 Gy in 5 fractions. All patients were planned for VMAT, which aims to deliver ≥ 95% of the prescription dose to 95% of the PTV. Once the physician approved the treatment plan and physics quality assessment was completed, treatments commenced on an every-other-day schedule. Patients received the same bowel preparation instructions for each treatment as for the planning CT simulation. Daily treatment accuracy was confirmed via daily 3-dimensional cone-beam CT (CBCT) for IGRT. No fiducials or hydrogel rectal spacers were used.

Follow-up Schedule and Toxicity Assessment

Follow-up assessments were conducted 4 to 6 weeks after radiation therapy and then repeated every 6 months for 2 to 5 years, and annually thereafter. At each follow-up visit, patients were evaluated for genitourinary (GU) and gastrointestinal (GI) toxicity, according to RTOG toxicity criteria. Prostate-specific antigen (PSA) levels were monitored; in patients receiving ADT, testosterone levels were also checked.

Statistical Analysis

Biochemical failure was defined using the Phoenix definition (nadir PSA + 2 ng/mL). Differences between dose cohorts were assessed using the log-rank test for survival outcomes and X² testing for categorical variables. GU and GI toxicities were summarized as cumulative incidences of RTOG grade ≥ II events. Statistical significance was set at P < .05.

RESULTS

One hundred seventy-four patients were included in the retrospective review. Patients had a median follow-up of 45 months (range, 24-111) (Figure). The median age at treatment was 74 years (range, 51-88), and the median pretreatment PSA level was 11.9 ng/mL (range, 0.6-69.5). Twenty-six patients (14.9%) had a GS 1, 77 (44.3%) had GS 2, 41 (23.6%) had GS 3, 18 (10.3%) had GS 4, and 12 (6.9%) had GS 5. Fifty-one patients (29.3%) received elective pelvic nodal radiotherapy, and 93 patients (53.4%) received ADT (Table 1).

At 24 months follow-up, 6 patients (3.4%) had biochemical failures. One patient died from metastatic prostate cancer, and 5 patients are living with biochemical failure (Table 2). The actuarial 5-year overall survival (OS) rate was 99.4%, and the 5-year disease-free survival (DFS) rate was 96.6%. We performed a subanalysis comparing outcomes of the 36.25 Gy vs 40 Gy SBRT cohorts. There was no statistically significant difference in DFS, OS, or the cumulative incidence of grade II/III toxicity between patients treated with 40 Gy vs 36.25 Gy. Outcomes stratified by NCCN risk groups (low, intermediate, high/very high) are detailed in Table 3. As expected, DFS was slightly lower in the high-risk group, but overall disease control remained high across all stratifications.

The cumulative incidence of RTOG grade II and higher GU toxicity was 28.2% (Table 4). This included 46 patients (26.4%) with grade II GU toxicity and 2 patients (1.2%) who developed grade III GU complications (1 requiring self-catheterization and another a suprapubic catheter for urinary retention). One patient (0.6%) treated with a 40 Gy dose regimen experienced a grade IV GU complication in the form of a rectovesical fistula necessitating surgical intervention.

The cumulative incidence of RTOG grade II or higher GI toxicity was 3.4%, and no grade III or IV gastrointestinal toxicities were observed during the follow-up period. Importantly, intraprostatic fiducials, hydrogel rectal spacers, or intravenous contrast were not routinely used in this cohort of patients.

The high rates of actuarial 5-year DFS and OS observed suggest a favorable initial response to the SBRT regimen employed at KCVAMC. However, given the potential for late recurrence in patients with prostate cancer, longer follow-up is essential to determine the durability of these outcomes. The observed GU toxicity rate of 28.2% for grade II and higher events warrants careful consideration and compares with other published data on SBRT for prostate cancer.¹⁵ The occurrence of a grade IV rectovesical fistula, although rare, is a notable adverse event that warrants discussion in the context of the treatment approach. The low incidence of grade II or higher GI toxicity is an encouraging finding, particularly given that hydrogel rectal spacers are not routinely used to minimize rectal exposure.

DISCUSSION

The primary objective of this retrospective study was to evaluate the outcomes of SBRT for patients with localized prostate cancer treated at KCVAMC and to compare these results with those reported in the literature. Our findings demonstrate promising intermediate-term efficacy, with an estimated 5-year DFS of 96.6% and OS of 99.4% at a median follow-up of 45 months. Furthermore, the observed toxicity profile appears acceptable, with a cumulative grade II and higher GU toxicity rate of 28.2% and a grade II or higher GI toxicity rate of 3.4%. Notably, these outcomes were achieved without the routine use of intraprostatic fiducials or hydrogel rectal spacers.

Two pivotal randomized phase 3 trials have established the noninferiority of ultrahypofractionated radiotherapy (UHRT) with SBRT over conventional fractionation. The HYPO-RT-PC trial compared SBRT (42.7 Gy in 7 fractions) with conventional fractionation (78 Gy in 39 fractions) in intermediate- and high-risk patients with prostate cancer and reported a 5-year biochemical relapse-free survival of 84% in both arms.⁹ The PACE-B trial, which included patients at low- and intermediate-risk, compared SBRT (36.25 Gy in 5 fractions) with conventional or moderate HFRT and reported a 5-year biochemical control rate of 95.8% in the SBRT arm and 94.6% in the control arm.¹⁵

A comprehensive review and meta-analysis of 7 phase 3 studies involving 6795 patients compared different radiotherapy regimens, namely, UHRT, HFRT, and CFRT, and reported that after 5 years, the DFS rates were 85.1% for CFRT, 86% for HFRT, and 85% for UHRT, with no significant difference in toxicity among the 3 different treatment approaches.¹⁸ This suggests that shorter, more intense radiotherapy schedules (UHRT and HFRT) may be as effective and safe as traditional, longer courses of radiation.

There are multiple published nonrandomized prospective trials in which thousands of patients with extreme hypofractionation have been treated with different doses, fractions, and techniques. While heterogeneity and limited long-term follow-up in the existing evidence are acknowledged, these data suggest that prostate SBRT provides appropriate biochemical control with few high-grade toxicities, supporting its ongoing global use and justifying further prospective investigations. Comparative data are shown in Table 5. Several ongoing studies are evaluating noninferiority, superiority, and cost-effectiveness using different methodologies (Table 6).^9,15,19-24

This study’s efficacy outcomes, particularly the high DFS rate, are consistent with the findings from these landmark trials, suggesting that the SBRT regimen used at KCVAMC is effective in achieving early disease control despite 17.2% of patients having high-risk disease. The GU toxicity observed in this study, with a 28.2% rate of grade II or higher events, is also comparable with the 26.9% reported in the 5-fraction SBRT arm of the PACE-B trial, which had a longer median follow-up of 74 months.¹⁵ It is important to note that a portion of these grade II events occurred in patients who were already on a blockers for pre-existing lower urinary tract symptoms before starting radiotherapy, which may inflate the observed cumulative acute toxicity score.

A critical comparison is how SBRT toxicity aligns with moderate hypofractionation (eg, 60 Gy in 20 fractions or 70 Gy in 28 fractions as reported by others).^4,6 Our observed grade III and higher GU toxicity rate (1.7%) and grade III and higher GI toxicity rate (0%) are highly favorable when compared with historical moderate hypofractionation data, which typically report grade III GU toxicity in the range of 2% to 3% and grade III GI toxicity around 1% to 2%. This suggests that despite the higher dose per fraction, SBRT does not necessarily lead to increased severe acute toxicity, potentially offering a superior therapeutic ratio for GI and GU sparing.

However, the occurrence of a grade IV rectovesical fistula in 1 patient (0.6%)—who received the 40 Gy dose—was a serious complication that warrants careful consideration. This rare, but severe, complication in the higher dose cohort underscores the potential for increased organ-at-risk toxicity, particularly in the absence of a hydrogel rectal spacer, which is designed to mitigate high-dose rectal exposure. While the overall rate of significant GU toxicity remains low, this event highlights the potential risks associated with SBRT. Hydrogel rectal spacers are designed to increase the distance between the prostate and the rectum, which can reduce the rectal radiation dose and potentially mitigate the risk of such fistulas. The low rate of grade II or worse GI toxicity (3.4%) in our study is noteworthy, especially considering that hydrogel spacers were not routinely used. This finding aligns with the 2.5% GI toxicity rate reported in the SBRT arm of the PACE-B trial, suggesting that careful treatment planning and delivery techniques, such as VMAT-IMRT and daily CBCT for IGRT, may contribute to minimizing GI toxicity even without the use of rectal spacers.¹⁵ The exclusive use of 3-dimensional CBCT for IGRT in our study, without the use of fiducial markers, suggests that accurate target localization can be achieved with this approach, contributing to the observed efficacy and reduced toxicity.

Strengths and Limitations

This study’s retrospective, single-center design may have introduced selection bias. The median follow-up of 45 months, while substantial, is still relatively short for assessing very late toxicities and long-term oncologic outcomes in prostate cancer, which is known for late recurrences. Additionally, the lack of a direct comparison group within KCVAMC limits the ability to definitively attribute the observed outcomes solely to SBRT treatment. However, the strengths of this study include the inclusion of a consecutive series of veteran patients with localized prostate cancer across all risk categories, providing a real-world perspective on SBRT outcomes in a diverse patient population. Furthermore, the detailed assessment of efficacy and toxicity via standardized RTOG criteria enhances the comparability of our findings with those of other published prospective studies, despite the retrospective nature of the data.

CONCLUSIONS

This single-institution retrospective analysis revealed that short-term SBRT (36.25 to 40 Gy in 5 fractions), with a minimum follow-up of 24 months and a median follow-up of 45 months, for localized prostate cancer, including patients at HR, is associated with promising early efficacy and acceptable toxicity, even in the absence of routine fiducial or hydrogel spacer use. The favorable actuarial 5-year DFS and OS rates, coupled with a manageable toxicity profile, suggest that SBRT is a safe and convenient treatment option for many patients with localized prostate cancer. However, a longer follow-up is necessary to confirm these findings and fully characterize the long-term efficacy and toxicity of this SBRT regimen. Nevertheless, the results contribute to the growing body of evidence suggesting that SBRT is a safe and convenient treatment option for many patients with localized prostate cancer.

Prostate cancer is the most common cancer in US males, with an estimated 313,780 new cases and 35,770 deaths in 2025.¹ Several treatment options are available for localized prostate cancer that have similar outcomes, including active surveillance for low-risk cancers, surgery, or radiotherapy.^2,3 Conventional fractionation radiotherapy (CFRT) with 40 to 45 fractions over 8 to 9 weeks has been used for decades. Over the past 2 decades, moderate hypofractionation schedules with 2.4 to 3.4 Gy per fraction over 20 to 28 fractions have become standard, as many noninferiority randomized clinical trials (RCTs) such as CHHiP (UK),⁴ PROFIT (Canada and Europe),⁵ NRG Oncology RTOG 0415 (US),⁶ HYPRO (Netherlands),^7,8 and HYPO-RT-PC (Sweden and Denmark),⁹ have shown the noninferiority of moderately hypofractionated radiotherapy compared with CFRT. Notably, most of these noninferiority studies primarily included patients with low- or intermediate-risk prostate cancer, except for the HYPO-RT-PC trial,⁹ which also included patients with intermediate- and high-risk prostate cancer.

These noninferiority studies, along with technological advances in radiotherapy, such as intensity-modulated radiotherapy (IMRT), volumetric modulated arc therapy (VMAT), and image-guided radiotherapy (IGRT), paved the path to ultrahypofractionated stereotactic body radiotherapy (SBRT) that is delivered in 5 fractions of ≥ 6 Gy. This high dose per fraction may have a radiobiologic advantage over conventional fractionation. The relatively low a/ß ratio of prostate cancer, estimated to be between 1 and 2, suggests that tumor cells may be particularly sensitive to the high doses per fraction delivered in SBRT.^10-13 Compared with CFRT, SBRT-induced tumor cell death may also be mediated through different pathways; this pathway appears to be generated in a dose-dependent manner, particularly with doses > 8 Gy per fraction.^14,15 Additionally, the higher a/ß ratio for the surrounding organs at risk, such as the bladder and rectum, theoretically allows for an improved therapeutic ratio window that maximizes tumor control while minimizing damage to healthy tissues.

A substantial body of evidence from prospective studies and meta-analyses supports the use of SBRT for localized prostate cancer. HYPO-RT-PC, a significant phase 3 noninferiority study, enrolled 1200 patients with intermediate (89%) and high-risk (11%) prostate cancer randomized between 2 arms, including CFRT to 78 Gy in 39 fractions and SBRT to 42.7 Gy in 7 fractions, treated 3 days weekly. After a median follow-up of 60 months, the estimated 5-year biochemical relapse-free survival rate was 84% in both groups.⁹ This trial was notable because it was the first randomized study to demonstrate that SBRT was noninferior to CFRT in intermediate- and high-risk prostate cancer patients. Another pivotal phase 3 trial, the PACE-B study, enrolled 874 patients to compare SBRT (36.25 Gy to the prostate gland, with a secondary dose of 40 Gy to the gross tumor volume where applicable, in 5 fractions) with CFRT (78 Gy in 39 fractions) and moderately hypofractionated radiotherapy (HFRT) (62 Gy in 20 fractions) in patients with low- or intermediate-risk prostate cancer. With a 74-month median follow-up, the study reported 5-year biochemical free rates of 94.6% for CFRT and 95.8% for SBRT, confirming the noninferiority of SBRT to CFRT.¹⁵

SBRT offers short, effective, and convenient treatment to many patients with localized prostate cancer. While previous guidelines were more restrictive, the March 2026 National Comprehensive Cancer Network (NCCN) guidelines now list SBRT as a preferred treatment modality for high-risk prostate cancer.¹⁶

Given the growing body of evidence supporting the efficacy and safety of SBRT, we implemented an SBRT program in 2014 at a tertiary care center for veterans. This retrospective study was undertaken to evaluate the early efficacy and toxicity of SBRT in patients with localized prostate cancer treated at our institution, including patients across all risk stratifications.

METHODS

We identified 242 patients diagnosed with prostate cancer who underwent SBRT treatment between November 2014 and October 2024 at Overland Park Veterans Affairs Radiation Oncology Clinic. For the final analysis, 46 patients with < 2 years of follow-up and 22 patients who died from causes other than prostate cancer were excluded, resulting in a cohort of 174 patients with ≥ 24-month follow-up.

Treatment

Patients eligible for staging underwent imaging according to NCCN guidelines, including computed tomography (CT) of the abdomen and pelvis, bone scintigraphy, or, in recent years, prostate-specific membrane antigen positron emission tomography, primarily used for unfavorable intermediate-risk (UIR) and high-risk (HR) cancers. Patients with a negative staging work-up for nodal or skeletal disease were included. Prior to planning the CT simulation, patients were given bowel preparation instructions, including a low-fiber and low-gas-producing diet, simethicone, and enemas, the night before and morning of the simulation. Patients were instructed to arrive with a comfortably full bladder, having not voided for 2 to 3 hours prior to the procedure. At Kansas City Veterans Affairs Medical Center (KCVAMC), SBRT treatment was generally restricted to patients with a baseline American Urological Association symptom score of 15 to 20 out of 35 and a prostate gland size < 80 mL to minimize the risk of acute urinary toxicity. We did not use intraprostatic fiducials, hydrogel rectal spacers, or intravenous contrast agents for planning CT simulation.

Patients were placed in a supine position, and a vacuum bag was used for immobilization. Following the CT simulation, the images were transferred to the Eclipse treatment planning system. The clinical target volume (CTV) encompassed the prostate and the proximal 1.0 cm of the seminal vesicles for Gleason score (GS) 1 to 2, and the entire seminal vesicle was included for GS 3 to 5, which is consistent with KCVAMC practice and established safety protocols. The planning target volume (PTV) was created by uniformly expanding the CTV by 5 to 7 mm, except for the posterior margin, which was limited to 3 to 5 mm. When elective nodal radiotherapy was planned for HR prostate cancer, the pelvic field for CT simulation started at the L-2 upper border, with the lower border extending to the lesser trochanter. The pelvic nodes were delineated per Radiation Therapy Oncology Group (RTOG) guidelines.¹⁷ The CTV nodes (CTVn), including common iliac, external and internal iliac nodes, obturator, and presacral nodes, were created by uniformly expanding the CTVn by 2 to 3 mm. Slice-by-slice corrections were made to avoid bowel overlap in these patients.

The use of androgen deprivation therapy (ADT) for a duration of 6 to 24 months was prescribed for patients with UIR or HR prostate cancer per NCCN guidelines.¹⁶ The prescribed dose to the PTV was 36.25 to 40 Gy (40 Gy was mostly used as a boost to the dominant lesion) in 5 fractions, with each fraction ranging from 7.25 to 8 Gy. For elective nodal radiotherapy in patients at HR, the prescribed dose was 25 Gy in 5 fractions. All patients were planned for VMAT, which aims to deliver ≥ 95% of the prescription dose to 95% of the PTV. Once the physician approved the treatment plan and physics quality assessment was completed, treatments commenced on an every-other-day schedule. Patients received the same bowel preparation instructions for each treatment as for the planning CT simulation. Daily treatment accuracy was confirmed via daily 3-dimensional cone-beam CT (CBCT) for IGRT. No fiducials or hydrogel rectal spacers were used.

Follow-up Schedule and Toxicity Assessment

Follow-up assessments were conducted 4 to 6 weeks after radiation therapy and then repeated every 6 months for 2 to 5 years, and annually thereafter. At each follow-up visit, patients were evaluated for genitourinary (GU) and gastrointestinal (GI) toxicity, according to RTOG toxicity criteria. Prostate-specific antigen (PSA) levels were monitored; in patients receiving ADT, testosterone levels were also checked.

Statistical Analysis

Biochemical failure was defined using the Phoenix definition (nadir PSA + 2 ng/mL). Differences between dose cohorts were assessed using the log-rank test for survival outcomes and X² testing for categorical variables. GU and GI toxicities were summarized as cumulative incidences of RTOG grade ≥ II events. Statistical significance was set at P < .05.

RESULTS

One hundred seventy-four patients were included in the retrospective review. Patients had a median follow-up of 45 months (range, 24-111) (Figure). The median age at treatment was 74 years (range, 51-88), and the median pretreatment PSA level was 11.9 ng/mL (range, 0.6-69.5). Twenty-six patients (14.9%) had a GS 1, 77 (44.3%) had GS 2, 41 (23.6%) had GS 3, 18 (10.3%) had GS 4, and 12 (6.9%) had GS 5. Fifty-one patients (29.3%) received elective pelvic nodal radiotherapy, and 93 patients (53.4%) received ADT (Table 1).

At 24 months follow-up, 6 patients (3.4%) had biochemical failures. One patient died from metastatic prostate cancer, and 5 patients are living with biochemical failure (Table 2). The actuarial 5-year overall survival (OS) rate was 99.4%, and the 5-year disease-free survival (DFS) rate was 96.6%. We performed a subanalysis comparing outcomes of the 36.25 Gy vs 40 Gy SBRT cohorts. There was no statistically significant difference in DFS, OS, or the cumulative incidence of grade II/III toxicity between patients treated with 40 Gy vs 36.25 Gy. Outcomes stratified by NCCN risk groups (low, intermediate, high/very high) are detailed in Table 3. As expected, DFS was slightly lower in the high-risk group, but overall disease control remained high across all stratifications.

The cumulative incidence of RTOG grade II and higher GU toxicity was 28.2% (Table 4). This included 46 patients (26.4%) with grade II GU toxicity and 2 patients (1.2%) who developed grade III GU complications (1 requiring self-catheterization and another a suprapubic catheter for urinary retention). One patient (0.6%) treated with a 40 Gy dose regimen experienced a grade IV GU complication in the form of a rectovesical fistula necessitating surgical intervention.

The cumulative incidence of RTOG grade II or higher GI toxicity was 3.4%, and no grade III or IV gastrointestinal toxicities were observed during the follow-up period. Importantly, intraprostatic fiducials, hydrogel rectal spacers, or intravenous contrast were not routinely used in this cohort of patients.

The high rates of actuarial 5-year DFS and OS observed suggest a favorable initial response to the SBRT regimen employed at KCVAMC. However, given the potential for late recurrence in patients with prostate cancer, longer follow-up is essential to determine the durability of these outcomes. The observed GU toxicity rate of 28.2% for grade II and higher events warrants careful consideration and compares with other published data on SBRT for prostate cancer.¹⁵ The occurrence of a grade IV rectovesical fistula, although rare, is a notable adverse event that warrants discussion in the context of the treatment approach. The low incidence of grade II or higher GI toxicity is an encouraging finding, particularly given that hydrogel rectal spacers are not routinely used to minimize rectal exposure.

DISCUSSION

The primary objective of this retrospective study was to evaluate the outcomes of SBRT for patients with localized prostate cancer treated at KCVAMC and to compare these results with those reported in the literature. Our findings demonstrate promising intermediate-term efficacy, with an estimated 5-year DFS of 96.6% and OS of 99.4% at a median follow-up of 45 months. Furthermore, the observed toxicity profile appears acceptable, with a cumulative grade II and higher GU toxicity rate of 28.2% and a grade II or higher GI toxicity rate of 3.4%. Notably, these outcomes were achieved without the routine use of intraprostatic fiducials or hydrogel rectal spacers.

Two pivotal randomized phase 3 trials have established the noninferiority of ultrahypofractionated radiotherapy (UHRT) with SBRT over conventional fractionation. The HYPO-RT-PC trial compared SBRT (42.7 Gy in 7 fractions) with conventional fractionation (78 Gy in 39 fractions) in intermediate- and high-risk patients with prostate cancer and reported a 5-year biochemical relapse-free survival of 84% in both arms.⁹ The PACE-B trial, which included patients at low- and intermediate-risk, compared SBRT (36.25 Gy in 5 fractions) with conventional or moderate HFRT and reported a 5-year biochemical control rate of 95.8% in the SBRT arm and 94.6% in the control arm.¹⁵

A comprehensive review and meta-analysis of 7 phase 3 studies involving 6795 patients compared different radiotherapy regimens, namely, UHRT, HFRT, and CFRT, and reported that after 5 years, the DFS rates were 85.1% for CFRT, 86% for HFRT, and 85% for UHRT, with no significant difference in toxicity among the 3 different treatment approaches.¹⁸ This suggests that shorter, more intense radiotherapy schedules (UHRT and HFRT) may be as effective and safe as traditional, longer courses of radiation.

There are multiple published nonrandomized prospective trials in which thousands of patients with extreme hypofractionation have been treated with different doses, fractions, and techniques. While heterogeneity and limited long-term follow-up in the existing evidence are acknowledged, these data suggest that prostate SBRT provides appropriate biochemical control with few high-grade toxicities, supporting its ongoing global use and justifying further prospective investigations. Comparative data are shown in Table 5. Several ongoing studies are evaluating noninferiority, superiority, and cost-effectiveness using different methodologies (Table 6).^9,15,19-24

This study’s efficacy outcomes, particularly the high DFS rate, are consistent with the findings from these landmark trials, suggesting that the SBRT regimen used at KCVAMC is effective in achieving early disease control despite 17.2% of patients having high-risk disease. The GU toxicity observed in this study, with a 28.2% rate of grade II or higher events, is also comparable with the 26.9% reported in the 5-fraction SBRT arm of the PACE-B trial, which had a longer median follow-up of 74 months.¹⁵ It is important to note that a portion of these grade II events occurred in patients who were already on a blockers for pre-existing lower urinary tract symptoms before starting radiotherapy, which may inflate the observed cumulative acute toxicity score.

A critical comparison is how SBRT toxicity aligns with moderate hypofractionation (eg, 60 Gy in 20 fractions or 70 Gy in 28 fractions as reported by others).^4,6 Our observed grade III and higher GU toxicity rate (1.7%) and grade III and higher GI toxicity rate (0%) are highly favorable when compared with historical moderate hypofractionation data, which typically report grade III GU toxicity in the range of 2% to 3% and grade III GI toxicity around 1% to 2%. This suggests that despite the higher dose per fraction, SBRT does not necessarily lead to increased severe acute toxicity, potentially offering a superior therapeutic ratio for GI and GU sparing.

However, the occurrence of a grade IV rectovesical fistula in 1 patient (0.6%)—who received the 40 Gy dose—was a serious complication that warrants careful consideration. This rare, but severe, complication in the higher dose cohort underscores the potential for increased organ-at-risk toxicity, particularly in the absence of a hydrogel rectal spacer, which is designed to mitigate high-dose rectal exposure. While the overall rate of significant GU toxicity remains low, this event highlights the potential risks associated with SBRT. Hydrogel rectal spacers are designed to increase the distance between the prostate and the rectum, which can reduce the rectal radiation dose and potentially mitigate the risk of such fistulas. The low rate of grade II or worse GI toxicity (3.4%) in our study is noteworthy, especially considering that hydrogel spacers were not routinely used. This finding aligns with the 2.5% GI toxicity rate reported in the SBRT arm of the PACE-B trial, suggesting that careful treatment planning and delivery techniques, such as VMAT-IMRT and daily CBCT for IGRT, may contribute to minimizing GI toxicity even without the use of rectal spacers.¹⁵ The exclusive use of 3-dimensional CBCT for IGRT in our study, without the use of fiducial markers, suggests that accurate target localization can be achieved with this approach, contributing to the observed efficacy and reduced toxicity.

Strengths and Limitations

This study’s retrospective, single-center design may have introduced selection bias. The median follow-up of 45 months, while substantial, is still relatively short for assessing very late toxicities and long-term oncologic outcomes in prostate cancer, which is known for late recurrences. Additionally, the lack of a direct comparison group within KCVAMC limits the ability to definitively attribute the observed outcomes solely to SBRT treatment. However, the strengths of this study include the inclusion of a consecutive series of veteran patients with localized prostate cancer across all risk categories, providing a real-world perspective on SBRT outcomes in a diverse patient population. Furthermore, the detailed assessment of efficacy and toxicity via standardized RTOG criteria enhances the comparability of our findings with those of other published prospective studies, despite the retrospective nature of the data.

CONCLUSIONS

This single-institution retrospective analysis revealed that short-term SBRT (36.25 to 40 Gy in 5 fractions), with a minimum follow-up of 24 months and a median follow-up of 45 months, for localized prostate cancer, including patients at HR, is associated with promising early efficacy and acceptable toxicity, even in the absence of routine fiducial or hydrogel spacer use. The favorable actuarial 5-year DFS and OS rates, coupled with a manageable toxicity profile, suggest that SBRT is a safe and convenient treatment option for many patients with localized prostate cancer. However, a longer follow-up is necessary to confirm these findings and fully characterize the long-term efficacy and toxicity of this SBRT regimen. Nevertheless, the results contribute to the growing body of evidence suggesting that SBRT is a safe and convenient treatment option for many patients with localized prostate cancer.

Prostate cancer is the most common cancer in US males, with an estimated 313,780 new cases and 35,770 deaths in 2025.¹ Several treatment options are available for localized prostate cancer that have similar outcomes, including active surveillance for low-risk cancers, surgery, or radiotherapy.^2,3 Conventional fractionation radiotherapy (CFRT) with 40 to 45 fractions over 8 to 9 weeks has been used for decades. Over the past 2 decades, moderate hypofractionation schedules with 2.4 to 3.4 Gy per fraction over 20 to 28 fractions have become standard, as many noninferiority randomized clinical trials (RCTs) such as CHHiP (UK),⁴ PROFIT (Canada and Europe),⁵ NRG Oncology RTOG 0415 (US),⁶ HYPRO (Netherlands),^7,8 and HYPO-RT-PC (Sweden and Denmark),⁹ have shown the noninferiority of moderately hypofractionated radiotherapy compared with CFRT. Notably, most of these noninferiority studies primarily included patients with low- or intermediate-risk prostate cancer, except for the HYPO-RT-PC trial,⁹ which also included patients with intermediate- and high-risk prostate cancer.

These noninferiority studies, along with technological advances in radiotherapy, such as intensity-modulated radiotherapy (IMRT), volumetric modulated arc therapy (VMAT), and image-guided radiotherapy (IGRT), paved the path to ultrahypofractionated stereotactic body radiotherapy (SBRT) that is delivered in 5 fractions of ≥ 6 Gy. This high dose per fraction may have a radiobiologic advantage over conventional fractionation. The relatively low a/ß ratio of prostate cancer, estimated to be between 1 and 2, suggests that tumor cells may be particularly sensitive to the high doses per fraction delivered in SBRT.^10-13 Compared with CFRT, SBRT-induced tumor cell death may also be mediated through different pathways; this pathway appears to be generated in a dose-dependent manner, particularly with doses > 8 Gy per fraction.^14,15 Additionally, the higher a/ß ratio for the surrounding organs at risk, such as the bladder and rectum, theoretically allows for an improved therapeutic ratio window that maximizes tumor control while minimizing damage to healthy tissues.

A substantial body of evidence from prospective studies and meta-analyses supports the use of SBRT for localized prostate cancer. HYPO-RT-PC, a significant phase 3 noninferiority study, enrolled 1200 patients with intermediate (89%) and high-risk (11%) prostate cancer randomized between 2 arms, including CFRT to 78 Gy in 39 fractions and SBRT to 42.7 Gy in 7 fractions, treated 3 days weekly. After a median follow-up of 60 months, the estimated 5-year biochemical relapse-free survival rate was 84% in both groups.⁹ This trial was notable because it was the first randomized study to demonstrate that SBRT was noninferior to CFRT in intermediate- and high-risk prostate cancer patients. Another pivotal phase 3 trial, the PACE-B study, enrolled 874 patients to compare SBRT (36.25 Gy to the prostate gland, with a secondary dose of 40 Gy to the gross tumor volume where applicable, in 5 fractions) with CFRT (78 Gy in 39 fractions) and moderately hypofractionated radiotherapy (HFRT) (62 Gy in 20 fractions) in patients with low- or intermediate-risk prostate cancer. With a 74-month median follow-up, the study reported 5-year biochemical free rates of 94.6% for CFRT and 95.8% for SBRT, confirming the noninferiority of SBRT to CFRT.¹⁵

SBRT offers short, effective, and convenient treatment to many patients with localized prostate cancer. While previous guidelines were more restrictive, the March 2026 National Comprehensive Cancer Network (NCCN) guidelines now list SBRT as a preferred treatment modality for high-risk prostate cancer.¹⁶

Given the growing body of evidence supporting the efficacy and safety of SBRT, we implemented an SBRT program in 2014 at a tertiary care center for veterans. This retrospective study was undertaken to evaluate the early efficacy and toxicity of SBRT in patients with localized prostate cancer treated at our institution, including patients across all risk stratifications.

METHODS

We identified 242 patients diagnosed with prostate cancer who underwent SBRT treatment between November 2014 and October 2024 at Overland Park Veterans Affairs Radiation Oncology Clinic. For the final analysis, 46 patients with < 2 years of follow-up and 22 patients who died from causes other than prostate cancer were excluded, resulting in a cohort of 174 patients with ≥ 24-month follow-up.

Treatment

Patients eligible for staging underwent imaging according to NCCN guidelines, including computed tomography (CT) of the abdomen and pelvis, bone scintigraphy, or, in recent years, prostate-specific membrane antigen positron emission tomography, primarily used for unfavorable intermediate-risk (UIR) and high-risk (HR) cancers. Patients with a negative staging work-up for nodal or skeletal disease were included. Prior to planning the CT simulation, patients were given bowel preparation instructions, including a low-fiber and low-gas-producing diet, simethicone, and enemas, the night before and morning of the simulation. Patients were instructed to arrive with a comfortably full bladder, having not voided for 2 to 3 hours prior to the procedure. At Kansas City Veterans Affairs Medical Center (KCVAMC), SBRT treatment was generally restricted to patients with a baseline American Urological Association symptom score of 15 to 20 out of 35 and a prostate gland size < 80 mL to minimize the risk of acute urinary toxicity. We did not use intraprostatic fiducials, hydrogel rectal spacers, or intravenous contrast agents for planning CT simulation.

Patients were placed in a supine position, and a vacuum bag was used for immobilization. Following the CT simulation, the images were transferred to the Eclipse treatment planning system. The clinical target volume (CTV) encompassed the prostate and the proximal 1.0 cm of the seminal vesicles for Gleason score (GS) 1 to 2, and the entire seminal vesicle was included for GS 3 to 5, which is consistent with KCVAMC practice and established safety protocols. The planning target volume (PTV) was created by uniformly expanding the CTV by 5 to 7 mm, except for the posterior margin, which was limited to 3 to 5 mm. When elective nodal radiotherapy was planned for HR prostate cancer, the pelvic field for CT simulation started at the L-2 upper border, with the lower border extending to the lesser trochanter. The pelvic nodes were delineated per Radiation Therapy Oncology Group (RTOG) guidelines.¹⁷ The CTV nodes (CTVn), including common iliac, external and internal iliac nodes, obturator, and presacral nodes, were created by uniformly expanding the CTVn by 2 to 3 mm. Slice-by-slice corrections were made to avoid bowel overlap in these patients.

The use of androgen deprivation therapy (ADT) for a duration of 6 to 24 months was prescribed for patients with UIR or HR prostate cancer per NCCN guidelines.¹⁶ The prescribed dose to the PTV was 36.25 to 40 Gy (40 Gy was mostly used as a boost to the dominant lesion) in 5 fractions, with each fraction ranging from 7.25 to 8 Gy. For elective nodal radiotherapy in patients at HR, the prescribed dose was 25 Gy in 5 fractions. All patients were planned for VMAT, which aims to deliver ≥ 95% of the prescription dose to 95% of the PTV. Once the physician approved the treatment plan and physics quality assessment was completed, treatments commenced on an every-other-day schedule. Patients received the same bowel preparation instructions for each treatment as for the planning CT simulation. Daily treatment accuracy was confirmed via daily 3-dimensional cone-beam CT (CBCT) for IGRT. No fiducials or hydrogel rectal spacers were used.

Follow-up Schedule and Toxicity Assessment

Follow-up assessments were conducted 4 to 6 weeks after radiation therapy and then repeated every 6 months for 2 to 5 years, and annually thereafter. At each follow-up visit, patients were evaluated for genitourinary (GU) and gastrointestinal (GI) toxicity, according to RTOG toxicity criteria. Prostate-specific antigen (PSA) levels were monitored; in patients receiving ADT, testosterone levels were also checked.

Statistical Analysis

Biochemical failure was defined using the Phoenix definition (nadir PSA + 2 ng/mL). Differences between dose cohorts were assessed using the log-rank test for survival outcomes and X² testing for categorical variables. GU and GI toxicities were summarized as cumulative incidences of RTOG grade ≥ II events. Statistical significance was set at P < .05.

RESULTS

One hundred seventy-four patients were included in the retrospective review. Patients had a median follow-up of 45 months (range, 24-111) (Figure). The median age at treatment was 74 years (range, 51-88), and the median pretreatment PSA level was 11.9 ng/mL (range, 0.6-69.5). Twenty-six patients (14.9%) had a GS 1, 77 (44.3%) had GS 2, 41 (23.6%) had GS 3, 18 (10.3%) had GS 4, and 12 (6.9%) had GS 5. Fifty-one patients (29.3%) received elective pelvic nodal radiotherapy, and 93 patients (53.4%) received ADT (Table 1).

At 24 months follow-up, 6 patients (3.4%) had biochemical failures. One patient died from metastatic prostate cancer, and 5 patients are living with biochemical failure (Table 2). The actuarial 5-year overall survival (OS) rate was 99.4%, and the 5-year disease-free survival (DFS) rate was 96.6%. We performed a subanalysis comparing outcomes of the 36.25 Gy vs 40 Gy SBRT cohorts. There was no statistically significant difference in DFS, OS, or the cumulative incidence of grade II/III toxicity between patients treated with 40 Gy vs 36.25 Gy. Outcomes stratified by NCCN risk groups (low, intermediate, high/very high) are detailed in Table 3. As expected, DFS was slightly lower in the high-risk group, but overall disease control remained high across all stratifications.

The cumulative incidence of RTOG grade II and higher GU toxicity was 28.2% (Table 4). This included 46 patients (26.4%) with grade II GU toxicity and 2 patients (1.2%) who developed grade III GU complications (1 requiring self-catheterization and another a suprapubic catheter for urinary retention). One patient (0.6%) treated with a 40 Gy dose regimen experienced a grade IV GU complication in the form of a rectovesical fistula necessitating surgical intervention.

The cumulative incidence of RTOG grade II or higher GI toxicity was 3.4%, and no grade III or IV gastrointestinal toxicities were observed during the follow-up period. Importantly, intraprostatic fiducials, hydrogel rectal spacers, or intravenous contrast were not routinely used in this cohort of patients.

The high rates of actuarial 5-year DFS and OS observed suggest a favorable initial response to the SBRT regimen employed at KCVAMC. However, given the potential for late recurrence in patients with prostate cancer, longer follow-up is essential to determine the durability of these outcomes. The observed GU toxicity rate of 28.2% for grade II and higher events warrants careful consideration and compares with other published data on SBRT for prostate cancer.¹⁵ The occurrence of a grade IV rectovesical fistula, although rare, is a notable adverse event that warrants discussion in the context of the treatment approach. The low incidence of grade II or higher GI toxicity is an encouraging finding, particularly given that hydrogel rectal spacers are not routinely used to minimize rectal exposure.

DISCUSSION

The primary objective of this retrospective study was to evaluate the outcomes of SBRT for patients with localized prostate cancer treated at KCVAMC and to compare these results with those reported in the literature. Our findings demonstrate promising intermediate-term efficacy, with an estimated 5-year DFS of 96.6% and OS of 99.4% at a median follow-up of 45 months. Furthermore, the observed toxicity profile appears acceptable, with a cumulative grade II and higher GU toxicity rate of 28.2% and a grade II or higher GI toxicity rate of 3.4%. Notably, these outcomes were achieved without the routine use of intraprostatic fiducials or hydrogel rectal spacers.

Two pivotal randomized phase 3 trials have established the noninferiority of ultrahypofractionated radiotherapy (UHRT) with SBRT over conventional fractionation. The HYPO-RT-PC trial compared SBRT (42.7 Gy in 7 fractions) with conventional fractionation (78 Gy in 39 fractions) in intermediate- and high-risk patients with prostate cancer and reported a 5-year biochemical relapse-free survival of 84% in both arms.⁹ The PACE-B trial, which included patients at low- and intermediate-risk, compared SBRT (36.25 Gy in 5 fractions) with conventional or moderate HFRT and reported a 5-year biochemical control rate of 95.8% in the SBRT arm and 94.6% in the control arm.¹⁵

A comprehensive review and meta-analysis of 7 phase 3 studies involving 6795 patients compared different radiotherapy regimens, namely, UHRT, HFRT, and CFRT, and reported that after 5 years, the DFS rates were 85.1% for CFRT, 86% for HFRT, and 85% for UHRT, with no significant difference in toxicity among the 3 different treatment approaches.¹⁸ This suggests that shorter, more intense radiotherapy schedules (UHRT and HFRT) may be as effective and safe as traditional, longer courses of radiation.

There are multiple published nonrandomized prospective trials in which thousands of patients with extreme hypofractionation have been treated with different doses, fractions, and techniques. While heterogeneity and limited long-term follow-up in the existing evidence are acknowledged, these data suggest that prostate SBRT provides appropriate biochemical control with few high-grade toxicities, supporting its ongoing global use and justifying further prospective investigations. Comparative data are shown in Table 5. Several ongoing studies are evaluating noninferiority, superiority, and cost-effectiveness using different methodologies (Table 6).^9,15,19-24

This study’s efficacy outcomes, particularly the high DFS rate, are consistent with the findings from these landmark trials, suggesting that the SBRT regimen used at KCVAMC is effective in achieving early disease control despite 17.2% of patients having high-risk disease. The GU toxicity observed in this study, with a 28.2% rate of grade II or higher events, is also comparable with the 26.9% reported in the 5-fraction SBRT arm of the PACE-B trial, which had a longer median follow-up of 74 months.¹⁵ It is important to note that a portion of these grade II events occurred in patients who were already on a blockers for pre-existing lower urinary tract symptoms before starting radiotherapy, which may inflate the observed cumulative acute toxicity score.

A critical comparison is how SBRT toxicity aligns with moderate hypofractionation (eg, 60 Gy in 20 fractions or 70 Gy in 28 fractions as reported by others).^4,6 Our observed grade III and higher GU toxicity rate (1.7%) and grade III and higher GI toxicity rate (0%) are highly favorable when compared with historical moderate hypofractionation data, which typically report grade III GU toxicity in the range of 2% to 3% and grade III GI toxicity around 1% to 2%. This suggests that despite the higher dose per fraction, SBRT does not necessarily lead to increased severe acute toxicity, potentially offering a superior therapeutic ratio for GI and GU sparing.

However, the occurrence of a grade IV rectovesical fistula in 1 patient (0.6%)—who received the 40 Gy dose—was a serious complication that warrants careful consideration. This rare, but severe, complication in the higher dose cohort underscores the potential for increased organ-at-risk toxicity, particularly in the absence of a hydrogel rectal spacer, which is designed to mitigate high-dose rectal exposure. While the overall rate of significant GU toxicity remains low, this event highlights the potential risks associated with SBRT. Hydrogel rectal spacers are designed to increase the distance between the prostate and the rectum, which can reduce the rectal radiation dose and potentially mitigate the risk of such fistulas. The low rate of grade II or worse GI toxicity (3.4%) in our study is noteworthy, especially considering that hydrogel spacers were not routinely used. This finding aligns with the 2.5% GI toxicity rate reported in the SBRT arm of the PACE-B trial, suggesting that careful treatment planning and delivery techniques, such as VMAT-IMRT and daily CBCT for IGRT, may contribute to minimizing GI toxicity even without the use of rectal spacers.¹⁵ The exclusive use of 3-dimensional CBCT for IGRT in our study, without the use of fiducial markers, suggests that accurate target localization can be achieved with this approach, contributing to the observed efficacy and reduced toxicity.

Strengths and Limitations

This study’s retrospective, single-center design may have introduced selection bias. The median follow-up of 45 months, while substantial, is still relatively short for assessing very late toxicities and long-term oncologic outcomes in prostate cancer, which is known for late recurrences. Additionally, the lack of a direct comparison group within KCVAMC limits the ability to definitively attribute the observed outcomes solely to SBRT treatment. However, the strengths of this study include the inclusion of a consecutive series of veteran patients with localized prostate cancer across all risk categories, providing a real-world perspective on SBRT outcomes in a diverse patient population. Furthermore, the detailed assessment of efficacy and toxicity via standardized RTOG criteria enhances the comparability of our findings with those of other published prospective studies, despite the retrospective nature of the data.

CONCLUSIONS

This single-institution retrospective analysis revealed that short-term SBRT (36.25 to 40 Gy in 5 fractions), with a minimum follow-up of 24 months and a median follow-up of 45 months, for localized prostate cancer, including patients at HR, is associated with promising early efficacy and acceptable toxicity, even in the absence of routine fiducial or hydrogel spacer use. The favorable actuarial 5-year DFS and OS rates, coupled with a manageable toxicity profile, suggest that SBRT is a safe and convenient treatment option for many patients with localized prostate cancer. However, a longer follow-up is necessary to confirm these findings and fully characterize the long-term efficacy and toxicity of this SBRT regimen. Nevertheless, the results contribute to the growing body of evidence suggesting that SBRT is a safe and convenient treatment option for many patients with localized prostate cancer.