Original Research

Children with neurologic impairment (NI; eg, hypoxic-­ischemic encephalopathy, muscular dystrophy) are characterized by functional and/or intellectual impairments resulting from a variety of neurologic diseases.¹ These children commonly have respiratory comorbidities, including central hypoventilation, impaired cough, and oromotor dysfunction, that may lead to chronic respiratory insufficiency and a need for chronic respiratory support at baseline.^2,3 Baseline respiratory support modalities can include supplemental oxygen, noninvasive positive pressure ventilation, or invasive mechanical ventilation.

Acute respiratory infections (ARI; eg, pneumonia, bronchiolitis) are the most common cause of hospitalization, intensive care unit (ICU) admission, and death for children with NI.^1,3 Discharge criteria for otherwise healthy children admitted to the hospital with ARI often include return to respiratory baseline.⁴ Children with complex chronic conditions have longer hospitalizations when hospitalized with respiratory infections,^5-7 because, in part, comorbidities and complications prolong the time to return to baseline. This prolonged return to respiratory baseline in combination with family knowledge, comfort, and skill in managing respiratory support and other complexities at home may alter discharge practices in the population of children with NI. In our clinical experience, discharge before return to baseline respiratory support occurs more frequently in children with NI than in otherwise healthy children when hospitalized with ARI. However, the consequences of discharging children with NI prior to return to respiratory baseline are unknown.

In this study, we sought to determine if discharge prior to return to baseline respiratory support is associated with reutilization among children with NI hospitalized with ARI. We hypothesized that patients discharged prior to return to respiratory baseline would have higher rates of 30-day hospital reutilization.

METHODS

Study Design and Data Source

This single-center, retrospective cohort study of children hospitalized at Cincinnati Children’s Hospital Medical Center (CCHMC) used data from the Pediatric Health Information System (PHIS) and the electronic medical record (EMR). PHIS, an administrative database of 45 not-for-profit, tertiary care, US pediatric hospitals managed by Children’s Hospital Association (Lenexa, Kansas), was used to identify eligible children, examine demographic and clinical variables, and define outcomes. PHIS contains data regarding patient demographics, inpatient resource utilization, and diagnoses. Encrypted medical record numbers in PHIS allowed for local identification of patients’ medical records to complete EMR review to confirm eligibility and obtain detailed patient-level clinical information (eg, respiratory support needs) not available in PHIS.

Pilot medical record reviews allowed for standardized study definitions and procedures. All study staff underwent training with the primary investigator, including detailed review of 10 initial abstractions. Two investigators (K.M. and S.C.) performed repeat abstractions from 40 randomly selected records to enable assessment of interrater reliability. Average reliability, indicated by the κ statistic, indicated substantial to near-perfect reliability⁸ (κ = 0.97, 95% CI 0.90-1.00) for the primary exposure. EMR data were managed using Research Electronic Data Capture (REDCap, Nashville, Tennessee)⁹ and subsequently merged with PHIS data.

Study Population

Hospitalizations of children with NI aged 1 to 18 years at CCHMC between January 2010 and September 2015 were eligible for inclusion if they had a principal discharge diagnosis indicative of ARI and required increased respiratory support from baseline during hospitalization. NI was defined as a high-intensity, chronic neurological diagnosis with substantial functional impairments according to previously defined diagnosis codes.^1,10 ARI was identified using codes in the Clinical Classification Software (Agency for Healthcare Research and Quality, Rockville, MD) respiratory group indicative of ARI (eg, pneumonia, bronchiolitis, influenza; Appendix Table).

Children transferred to CCHMC were excluded because records from their initial illness presentation and management were not available. Because of expected differences in management and outcomes, children with a known diagnosis of tuberculosis or human immunodeficiency virus were excluded. Because exposure criteria were dependent on hospital discharge status, hospitalizations for children who died during admission (4 of 632 hospitalizations, 0.63%) were excluded from the final cohort (Appendix Figure).

Study Definitions

Baseline respiratory support (ie, “respiratory baseline”) was defined as the child’s highest level of respiratory support needed prior to admission when well (ie, no support, supplemental oxygen, continuous positive airway pressure [CPAP] or bilevel positive airway pressure [BiPAP], or ventilator support), and further characterized by night or day/night requirement. Respiratory baseline was identified using EMR documentation of home respiratory support use at the time of index admission. Return to respiratory baseline was defined as the date on which the child achieved documented home respiratory support settings, regardless of clinical symptoms.

Children may have required increased respiratory support from baseline at any time during hospitalization. Maximum respiratory support required was categorized as one of the following: (1) initiation of supplemental oxygen or increase in oxygen flow or duration; (2) initiation of CPAP or BiPAP; (3) increase in pressure settings or duration of pressure support for those with baseline CPAP, BiPAP, or ventilator use; and (4) initiation of full mechanical ventilation. Respiratory support categories were mutually exclusive: children requiring multiple types of increased respiratory support were classified for analysis by the most invasive form of respiratory support used (eg, a child requiring increase in both oxygen flow and pressure settings was categorized as an increase in pressure settings). Children who received heated high-flow nasal cannula therapy as maximum support were categorized as initiation or increase in oxygen support.

Time to return to respiratory baseline was defined as the difference in days between date of return to respiratory baseline and date of admission. Time to return to respiratory baseline was determined only for children who were discharged at respiratory baseline.

Primary Exposure and Outcome Measures

The primary exposure was hospital discharge before return to respiratory baseline (ie, discharge on higher respiratory support than at baseline settings). At our institution, standardized discharge criteria for children with NI do not exist. The primary outcome was all-cause, 30-day hospital reutilization, including hospital readmissions and emergency department (ED) revisits. Secondary outcomes included 30-day reutilization for ARI and hospital length of stay (LOS) in days.

Patient Demographics and Clinical Characteristics

Demographic and patient characteristics that might influence hospital discharge before return to respiratory baseline or readmission were obtained from PHIS (eg, demographic information, age, insurance type, measures of clinical complexity, illness severity) and by EMR review (eg, baseline respiratory support needs, maximum respiratory support during hospitalization). Measures of clinical complexity included comorbid complex chronic conditions (CCCs)^11-14 and technology dependence^14-16 using previously defined diagnostic codes. Measures of illness severity included sepsis¹⁷ and ICU-level care. At our institution, children with baseline ventilator use do not require admission to the ICU unless they are clinically unstable.

Statistical Analysis

Continuous variables were described using medians and interquartile ranges (IQR). Categorical variables were described using counts and percentages. Patient characteristics and outcomes were stratified by primary exposure and compared using chi-square test or Fisher exact test for categorical variables and Wilcoxon rank sum test for continuous variables.

To examine the independent association between discharge before return to respiratory baseline and hospital reutilization, a generalized estimating equation was used that included potential confounders while accounting for within-patient clustering. Patient demographics included age, race, ethnicity, and insurance type; measures of clinical complexity included number of CCCs, technology dependence, and baseline respiratory support; and measures of acute illness severity included ARI diagnosis, degree of increase in respiratory support during hospitalization, and ICU-level care. LOS was also included in the model as a covariate because of its expected association with both exposure and outcome.

Secondary analyses were conducted using the outcome of 30-day reutilization for ARI. Subgroup analysis excluding hospitalizations of children lost to follow-up (ie, no encounters in the 6 months after hospital discharge) was also conducted. All analyses were performed with SAS v9.3 (SAS Institute, Cary, North Carolina). P values < .05 were considered statistically significant. This study was approved by the Institutional Review Board.

RESULTS

Study Cohort

A total of 632 hospitalizations experienced by 366 children with NI who were admitted with ARI were included (Appendix Figure). Most children (66.4%) in the cohort experienced only one hospitalization, 17.5% had two hospitalizations, 7.9% had three hospitalizations, and 8.2% had four or more hospitalizations. The median age at hospitalization was 5.0 years (IQR 2.8-10.5) and most hospitalizations were for children who were male (56.6%), white (78.3%), non-Hispanic (96.0%), and publicly insured (51.7%; Table 1). More than one-quarter (28.6%) of hospitalizations were for children with four or more CCCs, and in 73.4% of hospitalizations, children were technology dependent (Table 1). Baseline respiratory support was common (46.8%), including home mechanical ventilation in 11.1% of hospitalizations (Table 1). Bacterial pneumonia, including aspiration pneumonia, was the most common discharge diagnosis (50.5%, Table 1).

Demographic and Clinical Characteristics

Children were discharged before return to respiratory baseline in 30.4% of hospitalizations (Appendix Figure). Children discharged before return to respiratory baseline were older (median age 5.7 years, IQR 3.1-11.0, vs 4.9 years, IQR 2.6-9.7; P = .04) and more likely to be privately insured (54.2% vs 43.4%; P = .04), compared with children discharged at respiratory baseline (Table 1). Children discharged before return to respiratory baseline were also more likely to have a respiratory CCC (59.9% vs 30.9%; P < .001), have a respiratory technology dependence diagnosis code (44.8% vs 24.1%; P < .001), and have baseline respiratory support needs on EMR review (67.7% vs 37.7%; P < .001), compared with children discharged at baseline (Table 1).

Children discharged before return to respiratory baseline required significantly greater escalation in respiratory support during hospitalization, compared with children discharged at respiratory baseline, including higher rates of initiation of CPAP or BiPAP, increased pressure settings from baseline (for home CPAP, BiPAP, or ventilator users), and initiation of full mechanical ventilation (Table 1). Hospitalizations in which children were discharged before return to respiratory baseline were also more likely to include ICU care than were those for children discharged at baseline (52.1% vs 35.2%; P < .001; Table 1).

Clinical Outcomes and Utilization

Reutilization within 30 days occurred after 32.1% of hospitalizations, with 26.1% requiring hospital readmission and 6.0% requiring ED revisit (Table 2). There was no statistical association in either unadjusted (Table 2) or adjusted (Table 3) analysis between children discharged before return to respiratory baseline and 30-day all-cause hospital reutilizations, hospital readmissions, or ED revisits.

In analysis of secondary outcomes, 30-day reutilization because of ARI occurred after 21.5% of hospitalizations, with 19.0% requiring hospital readmission and 2.5% requiring ED revisit. Median hospital LOS for the cohort was 4 days (IQR 2-8; Table 2). Hospitalizations in which children were discharged before return to respiratory baseline were longer than in those discharged at baseline (median 6 days, IQR 3-11, vs 4 days, IQR 2-7; P < .001; Table 2).

For hospitalizations of children discharged at respiratory baseline, the median time to return to respiratory baseline was 3 days (IQR 1-5, complete range 0-80). In these encounters, discharge occurred soon after return to respiratory baseline (median 1 day, IQR 0-1.5, complete range 0-54).

In subgroup analysis excluding the 18 hospitalizations in which children were lost to follow-up (2.8% of the total cohort), discharge before return to respiratory baseline was not associated with 30-day all-cause hospital reutilization (Table 4).

DISCUSSION

In this retrospective cohort study, children with NI hospitalized with ARI were frequently discharged using increased respiratory support from baseline. However, those discharged before return to respiratory baseline, despite their greater clinical complexity and acute illness severity, did not have increased hospital reutilization, compared with children discharged at respiratory baseline. Our findings suggest that discharge before return to baseline respiratory support after ARI may be clinically appropriate in some children with NI.

With the growing emphasis on decreasing hospital costs, concern exists that patients are being discharged from hospitals “quicker and sicker,”^18,19 with shortening lengths of stay and higher patient instability at discharge. Clinical instability at discharge has been associated with adverse postdischarge outcomes in adults with pneumonia^20-23; however, studies evaluating discharge readiness have not examined the population of children with NI. Our findings of no difference in reutilization for children with NI discharged before return to respiratory baseline, which would be expected to approximate one or more clinical instabilities, contrast these concerns.

Clinicians caring for children with NI hospitalized with ARI may find it difficult to determine a child’s discharge readiness, in part because many children with NI have longer times to return to respiratory baseline and some never return to their pre-­illness baseline.²⁴ In otherwise healthy children hospitalized with respiratory infections such as pneumonia, discharge criteria typically include complete wean from respiratory support prior to discharge.^4,25 In our study’s more complex children, whose parents already manage respiratory support at home, we hypothesize that discharging providers may be comfortable with discharge when the child has certain types of increased respiratory support compatible with home equipment, a parent skilled with monitoring the child’s respiratory status, and the support of an experienced outpatient provider and home nursing providers. At our institution, outpatient respiratory support weans are primarily performed by pediatric pulmonologists and, for isolated weaning of supplemental oxygen or time using support, by parents and outpatient pediatricians.

Another important factor in determining a child’s discharge readiness is the perspective of the child’s parent. Berry et al found that children whose parents believe they are not healthy enough for discharge are more likely to experience unplanned hospital readmissions,²⁴ signaling the role of child- and family-­specific factors in safe discharge decisions. Therefore, parents of children with NI should be proactively involved throughout the multidisciplinary discharge process,^26,27 including the decision to discharge before return to respiratory baseline. Parents have identified ongoing provider support, opportunities to practice home care skills, and written instructions with contingency plans as important components of discharge readiness.²⁸ Further work to create partnerships with these highly skilled caregivers in discharge decision making and transition planning are needed to promote safe discharge practices in this complex population.

In our study, children discharged before return to respiratory baseline were more likely to be older and privately insured compared with children discharged at respiratory baseline. Prior studies have found that social factors including low socioeconomic status influence ED provider admissions decisions for children with pneumonia.^29,30 However, the role of socioeconomic factors in provider discharge decisions for children with NI has not been assessed. These traits may also be proxy markers of other sociodemographic factors, such as parent education level, financial hardship influencing ability to participate in a child’s care at the bedside, access to comprehensive outpatient primary care, and availability of private home nursing. We hypothesize that these related characteristics directly and indirectly influence provider discharge decisions.

Discharging providers are likely more comfortable with discharge prior to return to respiratory baseline when the family has private duty nursing in the home. Home nurses can assist families in providing increased respiratory support and recognizing respiratory problems that may arise following discharge. However, home nursing shortages are common nationwide.^31,32 Low-income children, children with respiratory technology use, and children without Medicaid have been found to have larger gaps in home nursing availability.^31,32 Further studies are needed to understand the role of home nursing availability in provider discharge decisions in this population.

This study has several limitations. The retrospective design of this study creates the potential for residual confounding; there may be other clinical or demographic factors influencing hospital discharge decisions that we are unable to capture using EMR review, including parental knowledge and comfort managing illness, quality of discharge instructions, frequency of follow-up visits, and presence of skilled home nursing services. Categorization of children based on respiratory support status at discharge lends potential for misclassification of exposure; however, our substantial interrater reliability suggests that misclassification bias is small. This study’s primary finding indicated no difference between exposure groups; although we may be unable to detect small differences, we had sufficient power with our sample size to detect meaningful differences in reutilization outcomes.

This study was not designed to capture outpatient time to return to respiratory baseline; prospective studies are needed to identify rates of return to respiratory baseline following ARI in children with NI. We did not measure the level of respiratory support used by children at the time of discharge and, therefore, are unable to estimate the amount of respiratory support weaning needed following discharge or the compatibility of support with home equipment using our data. In addition, this study focused on respiratory support modalities and, thus, did not measure inpatient utilization of mucociliary clearance technologies that might be hypothesized to decrease the time to return to baseline respiratory support. Next steps in evaluating treatment of ARI include investigating the effect of mucociliary clearance on both exposure and outcome in this population.

There may be other clinically meaningful outcomes for this population apart from reutilization that we have not assessed in this study, including increased respiratory support required following discharge, primary care reutilization, healthcare costs, or parent satisfaction with timing of and outcomes after discharge. Finally, although our hospital has reutilization rates for children with NI that are similar to other institutions in the United States,³³ our results may not be generalizable to children with NI hospitalized at other institutions because local discharge processes and systems of care may be different. Prospective, multicenter investigation is needed to evaluate the clinical consequences of discharge before return to respiratory baseline more broadly.

CONCLUSION

At our institution, approximately one-quarter of children with NI hospitalized with ARI were discharged before return to respiratory baseline, but these children were not at increased risk of reutilization, compared with children discharged at respiratory baseline. Our findings suggest that return to baseline respiratory support might not be a necessary component of hospital discharge criteria. In otherwise clinically stable children with NI, discharge before return to respiratory baseline may be reasonable if their parents are comfortable managing respiratory support at home.

Acknowledgments

The authors thank Jonathan Rodean of the Children’s Hospital Association for his assistance with abstraction of PHIS data.

Children with neurologic impairment (NI; eg, hypoxic-­ischemic encephalopathy, muscular dystrophy) are characterized by functional and/or intellectual impairments resulting from a variety of neurologic diseases.¹ These children commonly have respiratory comorbidities, including central hypoventilation, impaired cough, and oromotor dysfunction, that may lead to chronic respiratory insufficiency and a need for chronic respiratory support at baseline.^2,3 Baseline respiratory support modalities can include supplemental oxygen, noninvasive positive pressure ventilation, or invasive mechanical ventilation.

Acute respiratory infections (ARI; eg, pneumonia, bronchiolitis) are the most common cause of hospitalization, intensive care unit (ICU) admission, and death for children with NI.^1,3 Discharge criteria for otherwise healthy children admitted to the hospital with ARI often include return to respiratory baseline.⁴ Children with complex chronic conditions have longer hospitalizations when hospitalized with respiratory infections,^5-7 because, in part, comorbidities and complications prolong the time to return to baseline. This prolonged return to respiratory baseline in combination with family knowledge, comfort, and skill in managing respiratory support and other complexities at home may alter discharge practices in the population of children with NI. In our clinical experience, discharge before return to baseline respiratory support occurs more frequently in children with NI than in otherwise healthy children when hospitalized with ARI. However, the consequences of discharging children with NI prior to return to respiratory baseline are unknown.

In this study, we sought to determine if discharge prior to return to baseline respiratory support is associated with reutilization among children with NI hospitalized with ARI. We hypothesized that patients discharged prior to return to respiratory baseline would have higher rates of 30-day hospital reutilization.

METHODS

Study Design and Data Source

This single-center, retrospective cohort study of children hospitalized at Cincinnati Children’s Hospital Medical Center (CCHMC) used data from the Pediatric Health Information System (PHIS) and the electronic medical record (EMR). PHIS, an administrative database of 45 not-for-profit, tertiary care, US pediatric hospitals managed by Children’s Hospital Association (Lenexa, Kansas), was used to identify eligible children, examine demographic and clinical variables, and define outcomes. PHIS contains data regarding patient demographics, inpatient resource utilization, and diagnoses. Encrypted medical record numbers in PHIS allowed for local identification of patients’ medical records to complete EMR review to confirm eligibility and obtain detailed patient-level clinical information (eg, respiratory support needs) not available in PHIS.

Pilot medical record reviews allowed for standardized study definitions and procedures. All study staff underwent training with the primary investigator, including detailed review of 10 initial abstractions. Two investigators (K.M. and S.C.) performed repeat abstractions from 40 randomly selected records to enable assessment of interrater reliability. Average reliability, indicated by the κ statistic, indicated substantial to near-perfect reliability⁸ (κ = 0.97, 95% CI 0.90-1.00) for the primary exposure. EMR data were managed using Research Electronic Data Capture (REDCap, Nashville, Tennessee)⁹ and subsequently merged with PHIS data.

Study Population

Hospitalizations of children with NI aged 1 to 18 years at CCHMC between January 2010 and September 2015 were eligible for inclusion if they had a principal discharge diagnosis indicative of ARI and required increased respiratory support from baseline during hospitalization. NI was defined as a high-intensity, chronic neurological diagnosis with substantial functional impairments according to previously defined diagnosis codes.^1,10 ARI was identified using codes in the Clinical Classification Software (Agency for Healthcare Research and Quality, Rockville, MD) respiratory group indicative of ARI (eg, pneumonia, bronchiolitis, influenza; Appendix Table).

Children transferred to CCHMC were excluded because records from their initial illness presentation and management were not available. Because of expected differences in management and outcomes, children with a known diagnosis of tuberculosis or human immunodeficiency virus were excluded. Because exposure criteria were dependent on hospital discharge status, hospitalizations for children who died during admission (4 of 632 hospitalizations, 0.63%) were excluded from the final cohort (Appendix Figure).

Study Definitions

Baseline respiratory support (ie, “respiratory baseline”) was defined as the child’s highest level of respiratory support needed prior to admission when well (ie, no support, supplemental oxygen, continuous positive airway pressure [CPAP] or bilevel positive airway pressure [BiPAP], or ventilator support), and further characterized by night or day/night requirement. Respiratory baseline was identified using EMR documentation of home respiratory support use at the time of index admission. Return to respiratory baseline was defined as the date on which the child achieved documented home respiratory support settings, regardless of clinical symptoms.

Children may have required increased respiratory support from baseline at any time during hospitalization. Maximum respiratory support required was categorized as one of the following: (1) initiation of supplemental oxygen or increase in oxygen flow or duration; (2) initiation of CPAP or BiPAP; (3) increase in pressure settings or duration of pressure support for those with baseline CPAP, BiPAP, or ventilator use; and (4) initiation of full mechanical ventilation. Respiratory support categories were mutually exclusive: children requiring multiple types of increased respiratory support were classified for analysis by the most invasive form of respiratory support used (eg, a child requiring increase in both oxygen flow and pressure settings was categorized as an increase in pressure settings). Children who received heated high-flow nasal cannula therapy as maximum support were categorized as initiation or increase in oxygen support.

Time to return to respiratory baseline was defined as the difference in days between date of return to respiratory baseline and date of admission. Time to return to respiratory baseline was determined only for children who were discharged at respiratory baseline.

Primary Exposure and Outcome Measures

The primary exposure was hospital discharge before return to respiratory baseline (ie, discharge on higher respiratory support than at baseline settings). At our institution, standardized discharge criteria for children with NI do not exist. The primary outcome was all-cause, 30-day hospital reutilization, including hospital readmissions and emergency department (ED) revisits. Secondary outcomes included 30-day reutilization for ARI and hospital length of stay (LOS) in days.

Patient Demographics and Clinical Characteristics

Demographic and patient characteristics that might influence hospital discharge before return to respiratory baseline or readmission were obtained from PHIS (eg, demographic information, age, insurance type, measures of clinical complexity, illness severity) and by EMR review (eg, baseline respiratory support needs, maximum respiratory support during hospitalization). Measures of clinical complexity included comorbid complex chronic conditions (CCCs)^11-14 and technology dependence^14-16 using previously defined diagnostic codes. Measures of illness severity included sepsis¹⁷ and ICU-level care. At our institution, children with baseline ventilator use do not require admission to the ICU unless they are clinically unstable.

Statistical Analysis

Continuous variables were described using medians and interquartile ranges (IQR). Categorical variables were described using counts and percentages. Patient characteristics and outcomes were stratified by primary exposure and compared using chi-square test or Fisher exact test for categorical variables and Wilcoxon rank sum test for continuous variables.

To examine the independent association between discharge before return to respiratory baseline and hospital reutilization, a generalized estimating equation was used that included potential confounders while accounting for within-patient clustering. Patient demographics included age, race, ethnicity, and insurance type; measures of clinical complexity included number of CCCs, technology dependence, and baseline respiratory support; and measures of acute illness severity included ARI diagnosis, degree of increase in respiratory support during hospitalization, and ICU-level care. LOS was also included in the model as a covariate because of its expected association with both exposure and outcome.

Secondary analyses were conducted using the outcome of 30-day reutilization for ARI. Subgroup analysis excluding hospitalizations of children lost to follow-up (ie, no encounters in the 6 months after hospital discharge) was also conducted. All analyses were performed with SAS v9.3 (SAS Institute, Cary, North Carolina). P values < .05 were considered statistically significant. This study was approved by the Institutional Review Board.

RESULTS

Study Cohort

A total of 632 hospitalizations experienced by 366 children with NI who were admitted with ARI were included (Appendix Figure). Most children (66.4%) in the cohort experienced only one hospitalization, 17.5% had two hospitalizations, 7.9% had three hospitalizations, and 8.2% had four or more hospitalizations. The median age at hospitalization was 5.0 years (IQR 2.8-10.5) and most hospitalizations were for children who were male (56.6%), white (78.3%), non-Hispanic (96.0%), and publicly insured (51.7%; Table 1). More than one-quarter (28.6%) of hospitalizations were for children with four or more CCCs, and in 73.4% of hospitalizations, children were technology dependent (Table 1). Baseline respiratory support was common (46.8%), including home mechanical ventilation in 11.1% of hospitalizations (Table 1). Bacterial pneumonia, including aspiration pneumonia, was the most common discharge diagnosis (50.5%, Table 1).

Demographic and Clinical Characteristics

Children were discharged before return to respiratory baseline in 30.4% of hospitalizations (Appendix Figure). Children discharged before return to respiratory baseline were older (median age 5.7 years, IQR 3.1-11.0, vs 4.9 years, IQR 2.6-9.7; P = .04) and more likely to be privately insured (54.2% vs 43.4%; P = .04), compared with children discharged at respiratory baseline (Table 1). Children discharged before return to respiratory baseline were also more likely to have a respiratory CCC (59.9% vs 30.9%; P < .001), have a respiratory technology dependence diagnosis code (44.8% vs 24.1%; P < .001), and have baseline respiratory support needs on EMR review (67.7% vs 37.7%; P < .001), compared with children discharged at baseline (Table 1).

Children discharged before return to respiratory baseline required significantly greater escalation in respiratory support during hospitalization, compared with children discharged at respiratory baseline, including higher rates of initiation of CPAP or BiPAP, increased pressure settings from baseline (for home CPAP, BiPAP, or ventilator users), and initiation of full mechanical ventilation (Table 1). Hospitalizations in which children were discharged before return to respiratory baseline were also more likely to include ICU care than were those for children discharged at baseline (52.1% vs 35.2%; P < .001; Table 1).

Clinical Outcomes and Utilization

Reutilization within 30 days occurred after 32.1% of hospitalizations, with 26.1% requiring hospital readmission and 6.0% requiring ED revisit (Table 2). There was no statistical association in either unadjusted (Table 2) or adjusted (Table 3) analysis between children discharged before return to respiratory baseline and 30-day all-cause hospital reutilizations, hospital readmissions, or ED revisits.

In analysis of secondary outcomes, 30-day reutilization because of ARI occurred after 21.5% of hospitalizations, with 19.0% requiring hospital readmission and 2.5% requiring ED revisit. Median hospital LOS for the cohort was 4 days (IQR 2-8; Table 2). Hospitalizations in which children were discharged before return to respiratory baseline were longer than in those discharged at baseline (median 6 days, IQR 3-11, vs 4 days, IQR 2-7; P < .001; Table 2).

For hospitalizations of children discharged at respiratory baseline, the median time to return to respiratory baseline was 3 days (IQR 1-5, complete range 0-80). In these encounters, discharge occurred soon after return to respiratory baseline (median 1 day, IQR 0-1.5, complete range 0-54).

In subgroup analysis excluding the 18 hospitalizations in which children were lost to follow-up (2.8% of the total cohort), discharge before return to respiratory baseline was not associated with 30-day all-cause hospital reutilization (Table 4).

DISCUSSION

In this retrospective cohort study, children with NI hospitalized with ARI were frequently discharged using increased respiratory support from baseline. However, those discharged before return to respiratory baseline, despite their greater clinical complexity and acute illness severity, did not have increased hospital reutilization, compared with children discharged at respiratory baseline. Our findings suggest that discharge before return to baseline respiratory support after ARI may be clinically appropriate in some children with NI.

With the growing emphasis on decreasing hospital costs, concern exists that patients are being discharged from hospitals “quicker and sicker,”^18,19 with shortening lengths of stay and higher patient instability at discharge. Clinical instability at discharge has been associated with adverse postdischarge outcomes in adults with pneumonia^20-23; however, studies evaluating discharge readiness have not examined the population of children with NI. Our findings of no difference in reutilization for children with NI discharged before return to respiratory baseline, which would be expected to approximate one or more clinical instabilities, contrast these concerns.

Clinicians caring for children with NI hospitalized with ARI may find it difficult to determine a child’s discharge readiness, in part because many children with NI have longer times to return to respiratory baseline and some never return to their pre-­illness baseline.²⁴ In otherwise healthy children hospitalized with respiratory infections such as pneumonia, discharge criteria typically include complete wean from respiratory support prior to discharge.^4,25 In our study’s more complex children, whose parents already manage respiratory support at home, we hypothesize that discharging providers may be comfortable with discharge when the child has certain types of increased respiratory support compatible with home equipment, a parent skilled with monitoring the child’s respiratory status, and the support of an experienced outpatient provider and home nursing providers. At our institution, outpatient respiratory support weans are primarily performed by pediatric pulmonologists and, for isolated weaning of supplemental oxygen or time using support, by parents and outpatient pediatricians.

Another important factor in determining a child’s discharge readiness is the perspective of the child’s parent. Berry et al found that children whose parents believe they are not healthy enough for discharge are more likely to experience unplanned hospital readmissions,²⁴ signaling the role of child- and family-­specific factors in safe discharge decisions. Therefore, parents of children with NI should be proactively involved throughout the multidisciplinary discharge process,^26,27 including the decision to discharge before return to respiratory baseline. Parents have identified ongoing provider support, opportunities to practice home care skills, and written instructions with contingency plans as important components of discharge readiness.²⁸ Further work to create partnerships with these highly skilled caregivers in discharge decision making and transition planning are needed to promote safe discharge practices in this complex population.

In our study, children discharged before return to respiratory baseline were more likely to be older and privately insured compared with children discharged at respiratory baseline. Prior studies have found that social factors including low socioeconomic status influence ED provider admissions decisions for children with pneumonia.^29,30 However, the role of socioeconomic factors in provider discharge decisions for children with NI has not been assessed. These traits may also be proxy markers of other sociodemographic factors, such as parent education level, financial hardship influencing ability to participate in a child’s care at the bedside, access to comprehensive outpatient primary care, and availability of private home nursing. We hypothesize that these related characteristics directly and indirectly influence provider discharge decisions.

Discharging providers are likely more comfortable with discharge prior to return to respiratory baseline when the family has private duty nursing in the home. Home nurses can assist families in providing increased respiratory support and recognizing respiratory problems that may arise following discharge. However, home nursing shortages are common nationwide.^31,32 Low-income children, children with respiratory technology use, and children without Medicaid have been found to have larger gaps in home nursing availability.^31,32 Further studies are needed to understand the role of home nursing availability in provider discharge decisions in this population.

This study has several limitations. The retrospective design of this study creates the potential for residual confounding; there may be other clinical or demographic factors influencing hospital discharge decisions that we are unable to capture using EMR review, including parental knowledge and comfort managing illness, quality of discharge instructions, frequency of follow-up visits, and presence of skilled home nursing services. Categorization of children based on respiratory support status at discharge lends potential for misclassification of exposure; however, our substantial interrater reliability suggests that misclassification bias is small. This study’s primary finding indicated no difference between exposure groups; although we may be unable to detect small differences, we had sufficient power with our sample size to detect meaningful differences in reutilization outcomes.

This study was not designed to capture outpatient time to return to respiratory baseline; prospective studies are needed to identify rates of return to respiratory baseline following ARI in children with NI. We did not measure the level of respiratory support used by children at the time of discharge and, therefore, are unable to estimate the amount of respiratory support weaning needed following discharge or the compatibility of support with home equipment using our data. In addition, this study focused on respiratory support modalities and, thus, did not measure inpatient utilization of mucociliary clearance technologies that might be hypothesized to decrease the time to return to baseline respiratory support. Next steps in evaluating treatment of ARI include investigating the effect of mucociliary clearance on both exposure and outcome in this population.

There may be other clinically meaningful outcomes for this population apart from reutilization that we have not assessed in this study, including increased respiratory support required following discharge, primary care reutilization, healthcare costs, or parent satisfaction with timing of and outcomes after discharge. Finally, although our hospital has reutilization rates for children with NI that are similar to other institutions in the United States,³³ our results may not be generalizable to children with NI hospitalized at other institutions because local discharge processes and systems of care may be different. Prospective, multicenter investigation is needed to evaluate the clinical consequences of discharge before return to respiratory baseline more broadly.

CONCLUSION

At our institution, approximately one-quarter of children with NI hospitalized with ARI were discharged before return to respiratory baseline, but these children were not at increased risk of reutilization, compared with children discharged at respiratory baseline. Our findings suggest that return to baseline respiratory support might not be a necessary component of hospital discharge criteria. In otherwise clinically stable children with NI, discharge before return to respiratory baseline may be reasonable if their parents are comfortable managing respiratory support at home.

Acknowledgments

The authors thank Jonathan Rodean of the Children’s Hospital Association for his assistance with abstraction of PHIS data.

Children with neurologic impairment (NI; eg, hypoxic-­ischemic encephalopathy, muscular dystrophy) are characterized by functional and/or intellectual impairments resulting from a variety of neurologic diseases.¹ These children commonly have respiratory comorbidities, including central hypoventilation, impaired cough, and oromotor dysfunction, that may lead to chronic respiratory insufficiency and a need for chronic respiratory support at baseline.^2,3 Baseline respiratory support modalities can include supplemental oxygen, noninvasive positive pressure ventilation, or invasive mechanical ventilation.

Acute respiratory infections (ARI; eg, pneumonia, bronchiolitis) are the most common cause of hospitalization, intensive care unit (ICU) admission, and death for children with NI.^1,3 Discharge criteria for otherwise healthy children admitted to the hospital with ARI often include return to respiratory baseline.⁴ Children with complex chronic conditions have longer hospitalizations when hospitalized with respiratory infections,^5-7 because, in part, comorbidities and complications prolong the time to return to baseline. This prolonged return to respiratory baseline in combination with family knowledge, comfort, and skill in managing respiratory support and other complexities at home may alter discharge practices in the population of children with NI. In our clinical experience, discharge before return to baseline respiratory support occurs more frequently in children with NI than in otherwise healthy children when hospitalized with ARI. However, the consequences of discharging children with NI prior to return to respiratory baseline are unknown.

In this study, we sought to determine if discharge prior to return to baseline respiratory support is associated with reutilization among children with NI hospitalized with ARI. We hypothesized that patients discharged prior to return to respiratory baseline would have higher rates of 30-day hospital reutilization.

METHODS

Study Design and Data Source

This single-center, retrospective cohort study of children hospitalized at Cincinnati Children’s Hospital Medical Center (CCHMC) used data from the Pediatric Health Information System (PHIS) and the electronic medical record (EMR). PHIS, an administrative database of 45 not-for-profit, tertiary care, US pediatric hospitals managed by Children’s Hospital Association (Lenexa, Kansas), was used to identify eligible children, examine demographic and clinical variables, and define outcomes. PHIS contains data regarding patient demographics, inpatient resource utilization, and diagnoses. Encrypted medical record numbers in PHIS allowed for local identification of patients’ medical records to complete EMR review to confirm eligibility and obtain detailed patient-level clinical information (eg, respiratory support needs) not available in PHIS.

Pilot medical record reviews allowed for standardized study definitions and procedures. All study staff underwent training with the primary investigator, including detailed review of 10 initial abstractions. Two investigators (K.M. and S.C.) performed repeat abstractions from 40 randomly selected records to enable assessment of interrater reliability. Average reliability, indicated by the κ statistic, indicated substantial to near-perfect reliability⁸ (κ = 0.97, 95% CI 0.90-1.00) for the primary exposure. EMR data were managed using Research Electronic Data Capture (REDCap, Nashville, Tennessee)⁹ and subsequently merged with PHIS data.

Study Population

Hospitalizations of children with NI aged 1 to 18 years at CCHMC between January 2010 and September 2015 were eligible for inclusion if they had a principal discharge diagnosis indicative of ARI and required increased respiratory support from baseline during hospitalization. NI was defined as a high-intensity, chronic neurological diagnosis with substantial functional impairments according to previously defined diagnosis codes.^1,10 ARI was identified using codes in the Clinical Classification Software (Agency for Healthcare Research and Quality, Rockville, MD) respiratory group indicative of ARI (eg, pneumonia, bronchiolitis, influenza; Appendix Table).

Children transferred to CCHMC were excluded because records from their initial illness presentation and management were not available. Because of expected differences in management and outcomes, children with a known diagnosis of tuberculosis or human immunodeficiency virus were excluded. Because exposure criteria were dependent on hospital discharge status, hospitalizations for children who died during admission (4 of 632 hospitalizations, 0.63%) were excluded from the final cohort (Appendix Figure).

Study Definitions

Baseline respiratory support (ie, “respiratory baseline”) was defined as the child’s highest level of respiratory support needed prior to admission when well (ie, no support, supplemental oxygen, continuous positive airway pressure [CPAP] or bilevel positive airway pressure [BiPAP], or ventilator support), and further characterized by night or day/night requirement. Respiratory baseline was identified using EMR documentation of home respiratory support use at the time of index admission. Return to respiratory baseline was defined as the date on which the child achieved documented home respiratory support settings, regardless of clinical symptoms.

Children may have required increased respiratory support from baseline at any time during hospitalization. Maximum respiratory support required was categorized as one of the following: (1) initiation of supplemental oxygen or increase in oxygen flow or duration; (2) initiation of CPAP or BiPAP; (3) increase in pressure settings or duration of pressure support for those with baseline CPAP, BiPAP, or ventilator use; and (4) initiation of full mechanical ventilation. Respiratory support categories were mutually exclusive: children requiring multiple types of increased respiratory support were classified for analysis by the most invasive form of respiratory support used (eg, a child requiring increase in both oxygen flow and pressure settings was categorized as an increase in pressure settings). Children who received heated high-flow nasal cannula therapy as maximum support were categorized as initiation or increase in oxygen support.

Time to return to respiratory baseline was defined as the difference in days between date of return to respiratory baseline and date of admission. Time to return to respiratory baseline was determined only for children who were discharged at respiratory baseline.

Primary Exposure and Outcome Measures

The primary exposure was hospital discharge before return to respiratory baseline (ie, discharge on higher respiratory support than at baseline settings). At our institution, standardized discharge criteria for children with NI do not exist. The primary outcome was all-cause, 30-day hospital reutilization, including hospital readmissions and emergency department (ED) revisits. Secondary outcomes included 30-day reutilization for ARI and hospital length of stay (LOS) in days.

Patient Demographics and Clinical Characteristics

Demographic and patient characteristics that might influence hospital discharge before return to respiratory baseline or readmission were obtained from PHIS (eg, demographic information, age, insurance type, measures of clinical complexity, illness severity) and by EMR review (eg, baseline respiratory support needs, maximum respiratory support during hospitalization). Measures of clinical complexity included comorbid complex chronic conditions (CCCs)^11-14 and technology dependence^14-16 using previously defined diagnostic codes. Measures of illness severity included sepsis¹⁷ and ICU-level care. At our institution, children with baseline ventilator use do not require admission to the ICU unless they are clinically unstable.

Statistical Analysis

Continuous variables were described using medians and interquartile ranges (IQR). Categorical variables were described using counts and percentages. Patient characteristics and outcomes were stratified by primary exposure and compared using chi-square test or Fisher exact test for categorical variables and Wilcoxon rank sum test for continuous variables.

To examine the independent association between discharge before return to respiratory baseline and hospital reutilization, a generalized estimating equation was used that included potential confounders while accounting for within-patient clustering. Patient demographics included age, race, ethnicity, and insurance type; measures of clinical complexity included number of CCCs, technology dependence, and baseline respiratory support; and measures of acute illness severity included ARI diagnosis, degree of increase in respiratory support during hospitalization, and ICU-level care. LOS was also included in the model as a covariate because of its expected association with both exposure and outcome.

Secondary analyses were conducted using the outcome of 30-day reutilization for ARI. Subgroup analysis excluding hospitalizations of children lost to follow-up (ie, no encounters in the 6 months after hospital discharge) was also conducted. All analyses were performed with SAS v9.3 (SAS Institute, Cary, North Carolina). P values < .05 were considered statistically significant. This study was approved by the Institutional Review Board.

RESULTS

Study Cohort

A total of 632 hospitalizations experienced by 366 children with NI who were admitted with ARI were included (Appendix Figure). Most children (66.4%) in the cohort experienced only one hospitalization, 17.5% had two hospitalizations, 7.9% had three hospitalizations, and 8.2% had four or more hospitalizations. The median age at hospitalization was 5.0 years (IQR 2.8-10.5) and most hospitalizations were for children who were male (56.6%), white (78.3%), non-Hispanic (96.0%), and publicly insured (51.7%; Table 1). More than one-quarter (28.6%) of hospitalizations were for children with four or more CCCs, and in 73.4% of hospitalizations, children were technology dependent (Table 1). Baseline respiratory support was common (46.8%), including home mechanical ventilation in 11.1% of hospitalizations (Table 1). Bacterial pneumonia, including aspiration pneumonia, was the most common discharge diagnosis (50.5%, Table 1).

Demographic and Clinical Characteristics

Children were discharged before return to respiratory baseline in 30.4% of hospitalizations (Appendix Figure). Children discharged before return to respiratory baseline were older (median age 5.7 years, IQR 3.1-11.0, vs 4.9 years, IQR 2.6-9.7; P = .04) and more likely to be privately insured (54.2% vs 43.4%; P = .04), compared with children discharged at respiratory baseline (Table 1). Children discharged before return to respiratory baseline were also more likely to have a respiratory CCC (59.9% vs 30.9%; P < .001), have a respiratory technology dependence diagnosis code (44.8% vs 24.1%; P < .001), and have baseline respiratory support needs on EMR review (67.7% vs 37.7%; P < .001), compared with children discharged at baseline (Table 1).

Children discharged before return to respiratory baseline required significantly greater escalation in respiratory support during hospitalization, compared with children discharged at respiratory baseline, including higher rates of initiation of CPAP or BiPAP, increased pressure settings from baseline (for home CPAP, BiPAP, or ventilator users), and initiation of full mechanical ventilation (Table 1). Hospitalizations in which children were discharged before return to respiratory baseline were also more likely to include ICU care than were those for children discharged at baseline (52.1% vs 35.2%; P < .001; Table 1).

Clinical Outcomes and Utilization

Reutilization within 30 days occurred after 32.1% of hospitalizations, with 26.1% requiring hospital readmission and 6.0% requiring ED revisit (Table 2). There was no statistical association in either unadjusted (Table 2) or adjusted (Table 3) analysis between children discharged before return to respiratory baseline and 30-day all-cause hospital reutilizations, hospital readmissions, or ED revisits.

In analysis of secondary outcomes, 30-day reutilization because of ARI occurred after 21.5% of hospitalizations, with 19.0% requiring hospital readmission and 2.5% requiring ED revisit. Median hospital LOS for the cohort was 4 days (IQR 2-8; Table 2). Hospitalizations in which children were discharged before return to respiratory baseline were longer than in those discharged at baseline (median 6 days, IQR 3-11, vs 4 days, IQR 2-7; P < .001; Table 2).

For hospitalizations of children discharged at respiratory baseline, the median time to return to respiratory baseline was 3 days (IQR 1-5, complete range 0-80). In these encounters, discharge occurred soon after return to respiratory baseline (median 1 day, IQR 0-1.5, complete range 0-54).

In subgroup analysis excluding the 18 hospitalizations in which children were lost to follow-up (2.8% of the total cohort), discharge before return to respiratory baseline was not associated with 30-day all-cause hospital reutilization (Table 4).

DISCUSSION

In this retrospective cohort study, children with NI hospitalized with ARI were frequently discharged using increased respiratory support from baseline. However, those discharged before return to respiratory baseline, despite their greater clinical complexity and acute illness severity, did not have increased hospital reutilization, compared with children discharged at respiratory baseline. Our findings suggest that discharge before return to baseline respiratory support after ARI may be clinically appropriate in some children with NI.

With the growing emphasis on decreasing hospital costs, concern exists that patients are being discharged from hospitals “quicker and sicker,”^18,19 with shortening lengths of stay and higher patient instability at discharge. Clinical instability at discharge has been associated with adverse postdischarge outcomes in adults with pneumonia^20-23; however, studies evaluating discharge readiness have not examined the population of children with NI. Our findings of no difference in reutilization for children with NI discharged before return to respiratory baseline, which would be expected to approximate one or more clinical instabilities, contrast these concerns.

Clinicians caring for children with NI hospitalized with ARI may find it difficult to determine a child’s discharge readiness, in part because many children with NI have longer times to return to respiratory baseline and some never return to their pre-­illness baseline.²⁴ In otherwise healthy children hospitalized with respiratory infections such as pneumonia, discharge criteria typically include complete wean from respiratory support prior to discharge.^4,25 In our study’s more complex children, whose parents already manage respiratory support at home, we hypothesize that discharging providers may be comfortable with discharge when the child has certain types of increased respiratory support compatible with home equipment, a parent skilled with monitoring the child’s respiratory status, and the support of an experienced outpatient provider and home nursing providers. At our institution, outpatient respiratory support weans are primarily performed by pediatric pulmonologists and, for isolated weaning of supplemental oxygen or time using support, by parents and outpatient pediatricians.

Another important factor in determining a child’s discharge readiness is the perspective of the child’s parent. Berry et al found that children whose parents believe they are not healthy enough for discharge are more likely to experience unplanned hospital readmissions,²⁴ signaling the role of child- and family-­specific factors in safe discharge decisions. Therefore, parents of children with NI should be proactively involved throughout the multidisciplinary discharge process,^26,27 including the decision to discharge before return to respiratory baseline. Parents have identified ongoing provider support, opportunities to practice home care skills, and written instructions with contingency plans as important components of discharge readiness.²⁸ Further work to create partnerships with these highly skilled caregivers in discharge decision making and transition planning are needed to promote safe discharge practices in this complex population.

In our study, children discharged before return to respiratory baseline were more likely to be older and privately insured compared with children discharged at respiratory baseline. Prior studies have found that social factors including low socioeconomic status influence ED provider admissions decisions for children with pneumonia.^29,30 However, the role of socioeconomic factors in provider discharge decisions for children with NI has not been assessed. These traits may also be proxy markers of other sociodemographic factors, such as parent education level, financial hardship influencing ability to participate in a child’s care at the bedside, access to comprehensive outpatient primary care, and availability of private home nursing. We hypothesize that these related characteristics directly and indirectly influence provider discharge decisions.

Discharging providers are likely more comfortable with discharge prior to return to respiratory baseline when the family has private duty nursing in the home. Home nurses can assist families in providing increased respiratory support and recognizing respiratory problems that may arise following discharge. However, home nursing shortages are common nationwide.^31,32 Low-income children, children with respiratory technology use, and children without Medicaid have been found to have larger gaps in home nursing availability.^31,32 Further studies are needed to understand the role of home nursing availability in provider discharge decisions in this population.

This study has several limitations. The retrospective design of this study creates the potential for residual confounding; there may be other clinical or demographic factors influencing hospital discharge decisions that we are unable to capture using EMR review, including parental knowledge and comfort managing illness, quality of discharge instructions, frequency of follow-up visits, and presence of skilled home nursing services. Categorization of children based on respiratory support status at discharge lends potential for misclassification of exposure; however, our substantial interrater reliability suggests that misclassification bias is small. This study’s primary finding indicated no difference between exposure groups; although we may be unable to detect small differences, we had sufficient power with our sample size to detect meaningful differences in reutilization outcomes.

This study was not designed to capture outpatient time to return to respiratory baseline; prospective studies are needed to identify rates of return to respiratory baseline following ARI in children with NI. We did not measure the level of respiratory support used by children at the time of discharge and, therefore, are unable to estimate the amount of respiratory support weaning needed following discharge or the compatibility of support with home equipment using our data. In addition, this study focused on respiratory support modalities and, thus, did not measure inpatient utilization of mucociliary clearance technologies that might be hypothesized to decrease the time to return to baseline respiratory support. Next steps in evaluating treatment of ARI include investigating the effect of mucociliary clearance on both exposure and outcome in this population.

There may be other clinically meaningful outcomes for this population apart from reutilization that we have not assessed in this study, including increased respiratory support required following discharge, primary care reutilization, healthcare costs, or parent satisfaction with timing of and outcomes after discharge. Finally, although our hospital has reutilization rates for children with NI that are similar to other institutions in the United States,³³ our results may not be generalizable to children with NI hospitalized at other institutions because local discharge processes and systems of care may be different. Prospective, multicenter investigation is needed to evaluate the clinical consequences of discharge before return to respiratory baseline more broadly.

CONCLUSION

At our institution, approximately one-quarter of children with NI hospitalized with ARI were discharged before return to respiratory baseline, but these children were not at increased risk of reutilization, compared with children discharged at respiratory baseline. Our findings suggest that return to baseline respiratory support might not be a necessary component of hospital discharge criteria. In otherwise clinically stable children with NI, discharge before return to respiratory baseline may be reasonable if their parents are comfortable managing respiratory support at home.

Acknowledgments

The authors thank Jonathan Rodean of the Children’s Hospital Association for his assistance with abstraction of PHIS data.

Adverse Childhood Experiences, or ACEs, include exposure to abuse, neglect, or household dysfunction (eg, having a parent who is mentally ill) as a child.¹ Exposure to ACEs affects health into adulthood, with a dose-response relationship between ACEs and a range of comorbidities.¹ Adults with 6 or more ACEs have a 20-year shorter life expectancy than do those with no ACEs.¹ Still, ACEs are static; once experienced, that experience cannot be undone. However, resilience, or positive adaptation in the context of adversity, can be protective, buffering the negative effects of ACEs.^2,3 Protective factors that promote resilience include social capital, such as positive relationships with caregivers and peers.³

With their clear link to health outcomes across the life-course, there is a movement for pediatricians to screen children for ACEs⁴ and to develop strategies that promote resilience in children, parents, and families. However, screening a child for adversity has challenges because younger children may not have experienced an adverse exposure, or they may be unable to voice their experiences. Studies have demonstrated that parental adversity, or ACEs, may be a marker for childhood adversity.^5,6 Biological models also support this potential intergenerational effect of ACEs. Chronic exposure to stress, including ACEs, results in elevated cortisol via a dysregulated hypothalamic-pituitary-adrenal axis, which results in chronic inflammation.⁷ This “toxic stress” is prolonged, severe in intensity, and can lead to epigenetic changes that may be passed on to the next generation.^8,9

Hospitalization of an ill child, and the transition to home after that hospitalization, is a stressful event for children and families.¹⁰ This stress may be relevant to parents that have a history of a high rate of ACEs or a current low degree of resilience. Our previous work demonstrated that, in the inpatient setting, parents with high ACEs (≥4) or low resilience have increased coping difficulty 14 days after their child’s hospital discharge.¹¹ Our objective here was to evaluate whether a parent’s ACEs and/or resilience would also be associated with that child’s likelihood of reutilization. We hypothesized that more parental ACEs and/or lower parental resilience would be associated with revisits the emergency room, urgent care, or hospital readmissions.

METHODS

Participants and Study Design

We conducted a prospective cohort study of parents of hospitalized children recruited from the “Hospital-to-Home Outcomes” Studies (H2O I and H2O II).^12,13 H2O I and II were prospective, single-center, randomized controlled trials designed to determine the effectiveness of either a nurse-led transitional home visit (H2O I) or telephone call (H2O II) on 30-day unplanned healthcare reutilization. The trials and this study were approved by the Cincinnati Children’s Institutional Review Board. All parents provided written informed consent.

Details of H2O I and II recruitment and design have been described previously.^12,13 Briefly, children were eligible for inclusion in either study if they were admitted to our institution’s general Hospital Medicine or the Hospital Medicine Complex Care Services; for H2O I, children hospitalized on the Neurology and Neurosurgery services were also eligible.^12,13 Patients were excluded if they were discharged to a residential facility, if they lived outside the home healthcare nurse service area, if they were eligible for skilled home healthcare services (eg, intravenous antibiotics), or if the participating caregiver was non-English speaking.^12,13 In H2O I, families were randomized either to receive a single nurse home visit within 96 hours of discharge or standard of care. In H2O II, families enrolled were randomized to receive a telephone call by a nurse within 96 hours of discharge or standard of care. As we have previously published, randomization in both trials successfully balanced the intervention and control arms with respect to key demographic characteristics.^12,13 For the analyses presented here, we focused on a subset of caregivers 18 years and older whose children were enrolled in either H2O I or II between August 2015 and October 2016. In both H2O trials, face-to-face and paper-based questionnaires were completed by parents during the index hospitalization.

Outcome and Predictors

Our primary outcome was unanticipated healthcare reutilization defined as return to the emergency room, urgent care, or unplanned readmission within 30 days of hospital discharge, consistent with the H2O trials. This was measured using the primary institution’s administrative data supplemented by a utilization database shared across regional hospitals.¹⁴ Readmissions were identified as “unplanned” using a previously validated algorithm,¹⁵ and treated as a dichotomous yes/no variable.

Our primary predictors were parental ACEs and resilience (see Appendix Tables). The ACE questionnaire addresses abuse, neglect, and household dysfunction in the first 18 years of life.¹ It is composed of 10 questions, each with a yes/no response.¹ We defined parents as low (ACE 0), moderate (ACE 1-3), or high (ACE ≥4) risk a priori because previous literature has described poor outcomes in adults with 4 or more ACEs.¹⁶

Given the sensitive nature of the questions, respondents independently completed the ACE questionnaire on paper instead of via the face-to-face survey. Respondents returned the completed questionnaire to the research assistant in a sealed envelope. All families received educational information on relevant hospital and community-based resources (eg, social work).

Parental resilience was measured using the Brief Resilience Scale (BRS). The BRS is 6 items, each on a 5-point Likert scale. Responses were averaged, providing a total score of 1-5; higher scores are representative of higher resilience.¹⁷ We treated the BRS score as a continuous variable. BRS has been used in clinical settings; it has demonstrated positive correlation with social support and negative correlation with fatigue.¹⁷ Parents answered BRS questions during the index pediatric hospitalization in a face-to-face interview.

Parent and Child Characteristics

Parent and child sociodemographic variables were also obtained during the face-to-face interview. Parental variables included age, gender, educational attainment, household income, employment status, and financial and social strain.¹¹ Educational attainment was analyzed in 2 categories—high school or less vs more than high school—because most discharge instructions are written at a high school reading level.¹⁸ Parents reported their annual household income in the following categories: <$15,000; $15,000-$29,999; $30,000-$44,999; $45,000-$59,999; $60,000-$89,999; $90,000-$119,999; ≥$120,000. Employment was dichotomized as not employed/student vs any employment. Financial and social strain were assessed using a series of 9 previously described questions.¹⁹ These questions assessed, via self-report, a family’s ability to make ends meet, ability to pay rent/mortgage or utilities, need to move in with others because of financial reasons, and ability to borrow money if needed, as well as home ownership and parental marital status.^15,19 Strain questions were all dichotomous (yes/no, single/not single). A composite variable was then constructed that categorized those reporting no strain items, 1 to 2 items, 3 to 4 items, and 5 or more items.²⁰

Child variables included race, ethnicity, age, primary care access,²¹ payer, and H2O treatment arm. Race categories were white/Caucasian, black/African American, American Indian or Alaskan Native, Asian or Pacific Islander, and other; ethnicity categories were Hispanic/Latino, non-Hispanic/Latino, and unknown. Given relatively low numbers of children reported to be Hispanic/Latino, we combined race and ethnicity into a single variable, categorized as non-Hispanic/white, non-Hispanic/black, and multiracial/Hispanic/other. Primary care access was assessed using the access subscale to the Parent’s Perception of Primary Care questionnaire. This includes assessment of a family’s ability to travel to their doctor, to see their doctor for routine or sick care, and to get help or advice on evenings or weekends. Scores were categorized as always adequate, almost always adequate, or sometimes/never adequate.²¹ Payer was dichotomized to private or public/self-pay.

Statistical Analyses

We examined the distribution of outcomes, predictors, and covariates. We compared sociodemographic characteristics of those respondents and nonrespondents to the ACE screen using the chi-square test for categorical variables or the t test for continuous variables. We used logistic regression to assess for associations between the independent variables of interest and reutilization, adjusting for potential confounders. To build our adjusted, multivariable model, we decided a priori to include child race/ethnicity, primary care access, financial and social strain, and trial treatment arm. We treated the H2O I control group as the referent group. Other covariates considered for inclusion were caregiver education, household income, employment, and payer. These were included in multivariable models if bivariate associations were significant at the P < .1 level. We assessed an ACE-by-resilience interaction term because we hypothesized that those with more ACEs and lower resilience may have more reutilization outcomes than parents with fewer ACEs and higher resilience. We also evaluated interaction terms between trial arm assignment and predictors to assess effects that may be introduced by the randomization. Predictors in the final logistic regression model were significant at the P < .05 level. Logistic regression assumption of little or no multicollinearity among the independent variables was verified in the final models. All analyses were performed with Stata v16 (Stata Corp, College Station, Texas).

RESULTS

There were a total of 1,787 parent-child dyads enrolled in the H2O I and II during the study period; 1,320 parents (74%) completed the ACE questionnaire and were included the analysis. Included parents were primarily female and employed, as well as educated beyond high school (Table 1). Overall, 64% reported one or more ACEs (range 0 to 9); 45% reported 1to 3, and 19% reported 4 or more ACEs. The most commonly reported ACEs were divorce (n = 573, 43%), exposure to alcoholism (n = 306, 23%), and exposure to mental illness (n = 281, 21%; Figure 1). Parents had a mean BRS score of 3.97 (range 1.17-5.00), with the distribution shown in Figure 2.

Of the 1,320 included patients, the average length of stay was 2.5 days, and 82% of hospitalizations were caused by acute medical issues (eg, bronchiolitis). A total of 211 children experienced a reutilization event within 30 days of discharge. In bivariate analysis, children with parents with 4 or more ACEs had a 2.02-times (95% CI 1.35-3.02) higher odds of experiencing a reutilization event than did those with parents reporting no ACEs. Parents with higher resilience scores had children with a lower odds of reutilization (odds ratio [OR] 0.77 95% CI 0.63-0.95).

In addition to our a priori variables, parental education, employment, and insurance met our significance threshold for inclusion in the multivariable model. The ACE-by-resilience interaction term was not significant and not included in the model. Similarly, there was no significant interaction between ACE and resilience and H2O treatment arm; the interaction terms were not included in the final adjusted model, but treatment arm assignment was kept as a covariate. A total of 1,292 children, out of the 1,320 respondents, remained in the final multivariable model; the excluded 28 had incomplete covariate data but were not otherwise different. In this final adjusted model, children with parents reporting 4 or more ACEs had a 1.69-times (95% CI 1.11-2.60) greater odds of reutilization than did those with parents reporting no ACEs (Table 2). Resilience failed to reach statistical significance in the adjusted model (OR 0.86, 95% CI 0.70-1.07).

DISCUSSION

We found that high-risk parents (4 or more ACEs) had children with an increased odds of healthcare reutilization, suggesting intergenerational effects of ACEs. We did not find a similar effect relating to parental resilience. We also did not find an interaction between parental ACEs and resilience, suggesting that a parent’s reported degree of resilience does not modify the effect of ACEs on reutilization risk.

Parental adversity may be a risk factor for a child’s unanticipated reutilization. We previously demonstrated that parents with 4 or more ACEs have more coping difficulty than a parent with no ACEs after a child’s hospitalization.¹¹ It is possible that parents with high adversity may have poorer coping mechanisms when dealing with a stressful situation, such as a child’s hospitalization. This may have resulted in inequitable outcomes (eg, increased reutilization) for their children. Other studies have confirmed such an intergenerational effect of adversity, linking a parent’s ACEs with poor developmental, behavioral, and health outcomes in their children.^6,22,23 O’Malley et al showed an association of parental ACEs to current adversities,²⁴ such as insurance or housing concerns, that affect the entirety of the household, including children. In short, it appears that parental ACEs may be a compelling predictor of current childhood adversity.

Resilience buffers the negative effects of ACEs; however, we did not find significant associations between resilience and reutilization or an interaction between ACEs and resilience. The factors that may contribute to reutilization are complex. In our previous work, parental resilience was associated with coping difficulty after discharge; but again, did not interact with parental ACEs.¹¹ Here, we suggest that while resilience may buffer the negative effects of ACEs, that buffering may not affect the likelihood of reutilization. It is also possible that the BRS tool is of less relevance on how one handles the stress of a child’s hospitalization. While the BRS is one measure of resilience, there are many other relevant constructs to resilience, such as connection to social supports, that also may also contribute to risk of reutilization.²⁵

Reducing the stress of a hospitalization itself and promoting a safe transition from hospital to home is critical to improving child health outcomes. Our data here, and in our previous work, demonstrate that a history of adversity and one’s current coping ability may drive a parent’s response to a child’s hospitalization and affect their capacity to care for that child after hospital discharge.¹¹ Additional in-hospital supports like child life, behavioral health, or pastoral care could reduce the stress of the hospitalization while also building positive coping mechanisms.^26-29 A meta-analysis demonstrated that such coping interventions can help alleviate the stress of a hospitalization.³⁰ Hill et al demonstrated successful stress reduction in parents of hospitalized children using a “Coping Kit for Parents.”³¹ Further studies are warranted to understand which interventions are most effective for children and families and whether they could be more effectively deployed if the inpatient team knew more about parental ACEs.

Screening for parental ACEs could help to identify patients at highest risk for a poor transition to home. Therefore, screening for parental adversity in clinical settings, including inpatient settings, may be relevant and valuable.³² Additionally, by recognizing the high prevalence of ACEs in an inpatient setting, hospitals and healthcare organizations could be motivated to develop and enact trauma-informed approaches. A trauma-informed care approach recognizes the intersection of trauma with health and social problems. With this recognition, care teams can more sensitively address the trauma as they provide relevant services.³³ Trauma-informed care is a secondary public health prevention approach that would help team members identify the prevalence and effects of trauma via screening, recognize the signs of a maladaptive response to stress, and respond by integrating awareness of trauma into practice management.^28,34 Both the National Academy of Medicine and the Agency for Healthcare Research and Quality have called for such a trauma-informed approach in primary care.³⁵ In response, many healthcare organizations have developed trauma-informed practices to better address the needs of the populations they serve. For example, provider training on this approach has led to improved rapport in patient-provider relationships.³⁶

Although ACE awareness is a component of trauma-informed care, there are still limitations of the original ACE questionnaire developed by Felitti et al. The existing tool is not inclusive of all adversities a parent or child may face. Moreover, its focus is on past exposures and experiences and not current health-related social needs (eg, food insecurity) which have known linkages with a range of health outcomes and health disparities.³⁷ Additionally, the original ACE questionnaire was created as a population level tool and not as a screening tool. If used as a screening tool, providers may view the questions as too sensitive to ask, and parents may have difficulty responding to and understanding the relevance to their child’s care. Therefore, we suggest that more evidence is required to understand how to best adapt ACE questions into a screening processes that may be implemented in a medical setting.

More evidence is also needed to determine when and where such screening may be most useful. A primary care provider would be best equipped to screen caregivers for ACEs given their established relationship with parents and patients. Given the potential relevance of such information for inpatient care provision, information could then flow from primary care to the inpatient team. However, because not all patients have established primary care providers and only 4% of pediatricians screen for ACEs,³⁸ it is important for inpatient medical teams to understand their role in identifying and addressing ACEs during hospital stays. Development of a screening tool, with input from all stakeholders—including parents—that is valid and feasible for use in a pediatric inpatient setting would be an important step forward. This tool should be paired with training in how to discuss these topics in a trauma-informed, nonjudgmental, empathic manner. We see this as a way in which providers can more effectively elicit an accurate response while simultaneously educating parents on the relevance of such sensitive topics during an acute hospital stay. We also recommend that screening should always be paired with response capabilities that connect those who screen positive with resources that could help them to navigate the stress experienced during and after a child’s hospitalization. Furthermore, communication with primary care providers about parents that screen positive should be integrated into the transition process.

This work has several limitations. First, our study was a part of randomized controlled trials conducted in one academic setting, which thereby limits generalizability. For example, we limited our cohort to those who were English-speaking patients only. This may bias our results because respondents with limited English proficiency may have different risk profiles than their English-speaking peers. In addition, the administration of the both the ACE and resilience questionnaires occurred during an acutely stressful period, which may influence how a parent responds to these questions. Also, both of the surveys are self-reported by parents, which may be susceptible to memory and response biases. Relatedly, we had a high number of nonrespondents, particularly to the ACE questionnaire. Our results are therefore only relevant to those who chose to respond and cannot be applied to nonrespondents. Further work assessing why one does or does not respond to such sensitive questions is an important area for future inquiry. Lastly, our cohort had limited medical complexity; future studies may consider links between parental ACEs (and resilience) and morbidity experienced by children with medical complexity.

CONCLUSION

Parents history of adversity is linked to their children’s unanticipated healthcare reutilization after a hospital discharge. Screening for parental stressors during a hospitalization may be an important first step to connecting parents and children to evidence-based interventions capable of mitigating the stress of hospitalization and promoting better, more seamless transitions from hospital to home.

Acknowledgments

Group Members: The following H2O members are nonauthor contributors: JoAnne Bachus, BSN, RN; Monica Borell, BSN, RN; Lenisa V Chang, MA, PhD; Patricia Crawford, RN; Sarah Ferris, BA; Jennifer Gold, BSN, RN; Judy A Heilman, BSN, RN; Jane C Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Margo Moore, MS, BSN, RN; Lynne O’Donnell, BSN, RN; Sarah Riddle, MD; Susan N Sherman, DPA; Angela M Statile, MD, MEd; Karen P Sullivan, BSN, RN; Heather Tubbs-Cooley, PhD, RN; Susan Wade-Murphy, MSN, RN; and Christine M White, MD, MAT.

The authors also thank David Keller, MD, for his guidance on the study.

Disclosures

The authors have no financial relationships or conflicts of interest relevant to this article to disclose.

Funding Source

Supported by funds from the Academic Pediatric Young Investigator Award (Dr A Shah) and the Patient-Centered Outcomes Research Institute Award (IHS-1306-0081, to Dr K Auger, Dr S Shah, Dr H Sucharew, Dr J Simmons), the National Institutes of Health (1K23AI112916, to Dr AF Beck), and the Agency for Healthcare Research and Quality (1K12HS026393-01, to Dr A Shah, K08-HS024735- 01A1, to Dr K Auger). Dr J Haney received Summer Undergraduate Research Fellowship funding through the Summer Undergraduate Research Fellowship at Cincinnati Children’s Hospital Medical Center.

Disclaimer

All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

Adverse Childhood Experiences, or ACEs, include exposure to abuse, neglect, or household dysfunction (eg, having a parent who is mentally ill) as a child.¹ Exposure to ACEs affects health into adulthood, with a dose-response relationship between ACEs and a range of comorbidities.¹ Adults with 6 or more ACEs have a 20-year shorter life expectancy than do those with no ACEs.¹ Still, ACEs are static; once experienced, that experience cannot be undone. However, resilience, or positive adaptation in the context of adversity, can be protective, buffering the negative effects of ACEs.^2,3 Protective factors that promote resilience include social capital, such as positive relationships with caregivers and peers.³

With their clear link to health outcomes across the life-course, there is a movement for pediatricians to screen children for ACEs⁴ and to develop strategies that promote resilience in children, parents, and families. However, screening a child for adversity has challenges because younger children may not have experienced an adverse exposure, or they may be unable to voice their experiences. Studies have demonstrated that parental adversity, or ACEs, may be a marker for childhood adversity.^5,6 Biological models also support this potential intergenerational effect of ACEs. Chronic exposure to stress, including ACEs, results in elevated cortisol via a dysregulated hypothalamic-pituitary-adrenal axis, which results in chronic inflammation.⁷ This “toxic stress” is prolonged, severe in intensity, and can lead to epigenetic changes that may be passed on to the next generation.^8,9

Hospitalization of an ill child, and the transition to home after that hospitalization, is a stressful event for children and families.¹⁰ This stress may be relevant to parents that have a history of a high rate of ACEs or a current low degree of resilience. Our previous work demonstrated that, in the inpatient setting, parents with high ACEs (≥4) or low resilience have increased coping difficulty 14 days after their child’s hospital discharge.¹¹ Our objective here was to evaluate whether a parent’s ACEs and/or resilience would also be associated with that child’s likelihood of reutilization. We hypothesized that more parental ACEs and/or lower parental resilience would be associated with revisits the emergency room, urgent care, or hospital readmissions.

METHODS

Participants and Study Design

We conducted a prospective cohort study of parents of hospitalized children recruited from the “Hospital-to-Home Outcomes” Studies (H2O I and H2O II).^12,13 H2O I and II were prospective, single-center, randomized controlled trials designed to determine the effectiveness of either a nurse-led transitional home visit (H2O I) or telephone call (H2O II) on 30-day unplanned healthcare reutilization. The trials and this study were approved by the Cincinnati Children’s Institutional Review Board. All parents provided written informed consent.

Details of H2O I and II recruitment and design have been described previously.^12,13 Briefly, children were eligible for inclusion in either study if they were admitted to our institution’s general Hospital Medicine or the Hospital Medicine Complex Care Services; for H2O I, children hospitalized on the Neurology and Neurosurgery services were also eligible.^12,13 Patients were excluded if they were discharged to a residential facility, if they lived outside the home healthcare nurse service area, if they were eligible for skilled home healthcare services (eg, intravenous antibiotics), or if the participating caregiver was non-English speaking.^12,13 In H2O I, families were randomized either to receive a single nurse home visit within 96 hours of discharge or standard of care. In H2O II, families enrolled were randomized to receive a telephone call by a nurse within 96 hours of discharge or standard of care. As we have previously published, randomization in both trials successfully balanced the intervention and control arms with respect to key demographic characteristics.^12,13 For the analyses presented here, we focused on a subset of caregivers 18 years and older whose children were enrolled in either H2O I or II between August 2015 and October 2016. In both H2O trials, face-to-face and paper-based questionnaires were completed by parents during the index hospitalization.

Outcome and Predictors

Our primary outcome was unanticipated healthcare reutilization defined as return to the emergency room, urgent care, or unplanned readmission within 30 days of hospital discharge, consistent with the H2O trials. This was measured using the primary institution’s administrative data supplemented by a utilization database shared across regional hospitals.¹⁴ Readmissions were identified as “unplanned” using a previously validated algorithm,¹⁵ and treated as a dichotomous yes/no variable.

Our primary predictors were parental ACEs and resilience (see Appendix Tables). The ACE questionnaire addresses abuse, neglect, and household dysfunction in the first 18 years of life.¹ It is composed of 10 questions, each with a yes/no response.¹ We defined parents as low (ACE 0), moderate (ACE 1-3), or high (ACE ≥4) risk a priori because previous literature has described poor outcomes in adults with 4 or more ACEs.¹⁶

Given the sensitive nature of the questions, respondents independently completed the ACE questionnaire on paper instead of via the face-to-face survey. Respondents returned the completed questionnaire to the research assistant in a sealed envelope. All families received educational information on relevant hospital and community-based resources (eg, social work).

Parental resilience was measured using the Brief Resilience Scale (BRS). The BRS is 6 items, each on a 5-point Likert scale. Responses were averaged, providing a total score of 1-5; higher scores are representative of higher resilience.¹⁷ We treated the BRS score as a continuous variable. BRS has been used in clinical settings; it has demonstrated positive correlation with social support and negative correlation with fatigue.¹⁷ Parents answered BRS questions during the index pediatric hospitalization in a face-to-face interview.

Parent and Child Characteristics

Parent and child sociodemographic variables were also obtained during the face-to-face interview. Parental variables included age, gender, educational attainment, household income, employment status, and financial and social strain.¹¹ Educational attainment was analyzed in 2 categories—high school or less vs more than high school—because most discharge instructions are written at a high school reading level.¹⁸ Parents reported their annual household income in the following categories: <$15,000; $15,000-$29,999; $30,000-$44,999; $45,000-$59,999; $60,000-$89,999; $90,000-$119,999; ≥$120,000. Employment was dichotomized as not employed/student vs any employment. Financial and social strain were assessed using a series of 9 previously described questions.¹⁹ These questions assessed, via self-report, a family’s ability to make ends meet, ability to pay rent/mortgage or utilities, need to move in with others because of financial reasons, and ability to borrow money if needed, as well as home ownership and parental marital status.^15,19 Strain questions were all dichotomous (yes/no, single/not single). A composite variable was then constructed that categorized those reporting no strain items, 1 to 2 items, 3 to 4 items, and 5 or more items.²⁰

Child variables included race, ethnicity, age, primary care access,²¹ payer, and H2O treatment arm. Race categories were white/Caucasian, black/African American, American Indian or Alaskan Native, Asian or Pacific Islander, and other; ethnicity categories were Hispanic/Latino, non-Hispanic/Latino, and unknown. Given relatively low numbers of children reported to be Hispanic/Latino, we combined race and ethnicity into a single variable, categorized as non-Hispanic/white, non-Hispanic/black, and multiracial/Hispanic/other. Primary care access was assessed using the access subscale to the Parent’s Perception of Primary Care questionnaire. This includes assessment of a family’s ability to travel to their doctor, to see their doctor for routine or sick care, and to get help or advice on evenings or weekends. Scores were categorized as always adequate, almost always adequate, or sometimes/never adequate.²¹ Payer was dichotomized to private or public/self-pay.

Statistical Analyses

We examined the distribution of outcomes, predictors, and covariates. We compared sociodemographic characteristics of those respondents and nonrespondents to the ACE screen using the chi-square test for categorical variables or the t test for continuous variables. We used logistic regression to assess for associations between the independent variables of interest and reutilization, adjusting for potential confounders. To build our adjusted, multivariable model, we decided a priori to include child race/ethnicity, primary care access, financial and social strain, and trial treatment arm. We treated the H2O I control group as the referent group. Other covariates considered for inclusion were caregiver education, household income, employment, and payer. These were included in multivariable models if bivariate associations were significant at the P < .1 level. We assessed an ACE-by-resilience interaction term because we hypothesized that those with more ACEs and lower resilience may have more reutilization outcomes than parents with fewer ACEs and higher resilience. We also evaluated interaction terms between trial arm assignment and predictors to assess effects that may be introduced by the randomization. Predictors in the final logistic regression model were significant at the P < .05 level. Logistic regression assumption of little or no multicollinearity among the independent variables was verified in the final models. All analyses were performed with Stata v16 (Stata Corp, College Station, Texas).

RESULTS

There were a total of 1,787 parent-child dyads enrolled in the H2O I and II during the study period; 1,320 parents (74%) completed the ACE questionnaire and were included the analysis. Included parents were primarily female and employed, as well as educated beyond high school (Table 1). Overall, 64% reported one or more ACEs (range 0 to 9); 45% reported 1to 3, and 19% reported 4 or more ACEs. The most commonly reported ACEs were divorce (n = 573, 43%), exposure to alcoholism (n = 306, 23%), and exposure to mental illness (n = 281, 21%; Figure 1). Parents had a mean BRS score of 3.97 (range 1.17-5.00), with the distribution shown in Figure 2.

Of the 1,320 included patients, the average length of stay was 2.5 days, and 82% of hospitalizations were caused by acute medical issues (eg, bronchiolitis). A total of 211 children experienced a reutilization event within 30 days of discharge. In bivariate analysis, children with parents with 4 or more ACEs had a 2.02-times (95% CI 1.35-3.02) higher odds of experiencing a reutilization event than did those with parents reporting no ACEs. Parents with higher resilience scores had children with a lower odds of reutilization (odds ratio [OR] 0.77 95% CI 0.63-0.95).

In addition to our a priori variables, parental education, employment, and insurance met our significance threshold for inclusion in the multivariable model. The ACE-by-resilience interaction term was not significant and not included in the model. Similarly, there was no significant interaction between ACE and resilience and H2O treatment arm; the interaction terms were not included in the final adjusted model, but treatment arm assignment was kept as a covariate. A total of 1,292 children, out of the 1,320 respondents, remained in the final multivariable model; the excluded 28 had incomplete covariate data but were not otherwise different. In this final adjusted model, children with parents reporting 4 or more ACEs had a 1.69-times (95% CI 1.11-2.60) greater odds of reutilization than did those with parents reporting no ACEs (Table 2). Resilience failed to reach statistical significance in the adjusted model (OR 0.86, 95% CI 0.70-1.07).

DISCUSSION

We found that high-risk parents (4 or more ACEs) had children with an increased odds of healthcare reutilization, suggesting intergenerational effects of ACEs. We did not find a similar effect relating to parental resilience. We also did not find an interaction between parental ACEs and resilience, suggesting that a parent’s reported degree of resilience does not modify the effect of ACEs on reutilization risk.

Parental adversity may be a risk factor for a child’s unanticipated reutilization. We previously demonstrated that parents with 4 or more ACEs have more coping difficulty than a parent with no ACEs after a child’s hospitalization.¹¹ It is possible that parents with high adversity may have poorer coping mechanisms when dealing with a stressful situation, such as a child’s hospitalization. This may have resulted in inequitable outcomes (eg, increased reutilization) for their children. Other studies have confirmed such an intergenerational effect of adversity, linking a parent’s ACEs with poor developmental, behavioral, and health outcomes in their children.^6,22,23 O’Malley et al showed an association of parental ACEs to current adversities,²⁴ such as insurance or housing concerns, that affect the entirety of the household, including children. In short, it appears that parental ACEs may be a compelling predictor of current childhood adversity.

Resilience buffers the negative effects of ACEs; however, we did not find significant associations between resilience and reutilization or an interaction between ACEs and resilience. The factors that may contribute to reutilization are complex. In our previous work, parental resilience was associated with coping difficulty after discharge; but again, did not interact with parental ACEs.¹¹ Here, we suggest that while resilience may buffer the negative effects of ACEs, that buffering may not affect the likelihood of reutilization. It is also possible that the BRS tool is of less relevance on how one handles the stress of a child’s hospitalization. While the BRS is one measure of resilience, there are many other relevant constructs to resilience, such as connection to social supports, that also may also contribute to risk of reutilization.²⁵

Reducing the stress of a hospitalization itself and promoting a safe transition from hospital to home is critical to improving child health outcomes. Our data here, and in our previous work, demonstrate that a history of adversity and one’s current coping ability may drive a parent’s response to a child’s hospitalization and affect their capacity to care for that child after hospital discharge.¹¹ Additional in-hospital supports like child life, behavioral health, or pastoral care could reduce the stress of the hospitalization while also building positive coping mechanisms.^26-29 A meta-analysis demonstrated that such coping interventions can help alleviate the stress of a hospitalization.³⁰ Hill et al demonstrated successful stress reduction in parents of hospitalized children using a “Coping Kit for Parents.”³¹ Further studies are warranted to understand which interventions are most effective for children and families and whether they could be more effectively deployed if the inpatient team knew more about parental ACEs.

Screening for parental ACEs could help to identify patients at highest risk for a poor transition to home. Therefore, screening for parental adversity in clinical settings, including inpatient settings, may be relevant and valuable.³² Additionally, by recognizing the high prevalence of ACEs in an inpatient setting, hospitals and healthcare organizations could be motivated to develop and enact trauma-informed approaches. A trauma-informed care approach recognizes the intersection of trauma with health and social problems. With this recognition, care teams can more sensitively address the trauma as they provide relevant services.³³ Trauma-informed care is a secondary public health prevention approach that would help team members identify the prevalence and effects of trauma via screening, recognize the signs of a maladaptive response to stress, and respond by integrating awareness of trauma into practice management.^28,34 Both the National Academy of Medicine and the Agency for Healthcare Research and Quality have called for such a trauma-informed approach in primary care.³⁵ In response, many healthcare organizations have developed trauma-informed practices to better address the needs of the populations they serve. For example, provider training on this approach has led to improved rapport in patient-provider relationships.³⁶

Although ACE awareness is a component of trauma-informed care, there are still limitations of the original ACE questionnaire developed by Felitti et al. The existing tool is not inclusive of all adversities a parent or child may face. Moreover, its focus is on past exposures and experiences and not current health-related social needs (eg, food insecurity) which have known linkages with a range of health outcomes and health disparities.³⁷ Additionally, the original ACE questionnaire was created as a population level tool and not as a screening tool. If used as a screening tool, providers may view the questions as too sensitive to ask, and parents may have difficulty responding to and understanding the relevance to their child’s care. Therefore, we suggest that more evidence is required to understand how to best adapt ACE questions into a screening processes that may be implemented in a medical setting.

More evidence is also needed to determine when and where such screening may be most useful. A primary care provider would be best equipped to screen caregivers for ACEs given their established relationship with parents and patients. Given the potential relevance of such information for inpatient care provision, information could then flow from primary care to the inpatient team. However, because not all patients have established primary care providers and only 4% of pediatricians screen for ACEs,³⁸ it is important for inpatient medical teams to understand their role in identifying and addressing ACEs during hospital stays. Development of a screening tool, with input from all stakeholders—including parents—that is valid and feasible for use in a pediatric inpatient setting would be an important step forward. This tool should be paired with training in how to discuss these topics in a trauma-informed, nonjudgmental, empathic manner. We see this as a way in which providers can more effectively elicit an accurate response while simultaneously educating parents on the relevance of such sensitive topics during an acute hospital stay. We also recommend that screening should always be paired with response capabilities that connect those who screen positive with resources that could help them to navigate the stress experienced during and after a child’s hospitalization. Furthermore, communication with primary care providers about parents that screen positive should be integrated into the transition process.

This work has several limitations. First, our study was a part of randomized controlled trials conducted in one academic setting, which thereby limits generalizability. For example, we limited our cohort to those who were English-speaking patients only. This may bias our results because respondents with limited English proficiency may have different risk profiles than their English-speaking peers. In addition, the administration of the both the ACE and resilience questionnaires occurred during an acutely stressful period, which may influence how a parent responds to these questions. Also, both of the surveys are self-reported by parents, which may be susceptible to memory and response biases. Relatedly, we had a high number of nonrespondents, particularly to the ACE questionnaire. Our results are therefore only relevant to those who chose to respond and cannot be applied to nonrespondents. Further work assessing why one does or does not respond to such sensitive questions is an important area for future inquiry. Lastly, our cohort had limited medical complexity; future studies may consider links between parental ACEs (and resilience) and morbidity experienced by children with medical complexity.

CONCLUSION

Parents history of adversity is linked to their children’s unanticipated healthcare reutilization after a hospital discharge. Screening for parental stressors during a hospitalization may be an important first step to connecting parents and children to evidence-based interventions capable of mitigating the stress of hospitalization and promoting better, more seamless transitions from hospital to home.

Acknowledgments

Group Members: The following H2O members are nonauthor contributors: JoAnne Bachus, BSN, RN; Monica Borell, BSN, RN; Lenisa V Chang, MA, PhD; Patricia Crawford, RN; Sarah Ferris, BA; Jennifer Gold, BSN, RN; Judy A Heilman, BSN, RN; Jane C Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Margo Moore, MS, BSN, RN; Lynne O’Donnell, BSN, RN; Sarah Riddle, MD; Susan N Sherman, DPA; Angela M Statile, MD, MEd; Karen P Sullivan, BSN, RN; Heather Tubbs-Cooley, PhD, RN; Susan Wade-Murphy, MSN, RN; and Christine M White, MD, MAT.

The authors also thank David Keller, MD, for his guidance on the study.

Disclosures

The authors have no financial relationships or conflicts of interest relevant to this article to disclose.

Funding Source

Supported by funds from the Academic Pediatric Young Investigator Award (Dr A Shah) and the Patient-Centered Outcomes Research Institute Award (IHS-1306-0081, to Dr K Auger, Dr S Shah, Dr H Sucharew, Dr J Simmons), the National Institutes of Health (1K23AI112916, to Dr AF Beck), and the Agency for Healthcare Research and Quality (1K12HS026393-01, to Dr A Shah, K08-HS024735- 01A1, to Dr K Auger). Dr J Haney received Summer Undergraduate Research Fellowship funding through the Summer Undergraduate Research Fellowship at Cincinnati Children’s Hospital Medical Center.

Disclaimer

All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

Adverse Childhood Experiences, or ACEs, include exposure to abuse, neglect, or household dysfunction (eg, having a parent who is mentally ill) as a child.¹ Exposure to ACEs affects health into adulthood, with a dose-response relationship between ACEs and a range of comorbidities.¹ Adults with 6 or more ACEs have a 20-year shorter life expectancy than do those with no ACEs.¹ Still, ACEs are static; once experienced, that experience cannot be undone. However, resilience, or positive adaptation in the context of adversity, can be protective, buffering the negative effects of ACEs.^2,3 Protective factors that promote resilience include social capital, such as positive relationships with caregivers and peers.³

With their clear link to health outcomes across the life-course, there is a movement for pediatricians to screen children for ACEs⁴ and to develop strategies that promote resilience in children, parents, and families. However, screening a child for adversity has challenges because younger children may not have experienced an adverse exposure, or they may be unable to voice their experiences. Studies have demonstrated that parental adversity, or ACEs, may be a marker for childhood adversity.^5,6 Biological models also support this potential intergenerational effect of ACEs. Chronic exposure to stress, including ACEs, results in elevated cortisol via a dysregulated hypothalamic-pituitary-adrenal axis, which results in chronic inflammation.⁷ This “toxic stress” is prolonged, severe in intensity, and can lead to epigenetic changes that may be passed on to the next generation.^8,9

Hospitalization of an ill child, and the transition to home after that hospitalization, is a stressful event for children and families.¹⁰ This stress may be relevant to parents that have a history of a high rate of ACEs or a current low degree of resilience. Our previous work demonstrated that, in the inpatient setting, parents with high ACEs (≥4) or low resilience have increased coping difficulty 14 days after their child’s hospital discharge.¹¹ Our objective here was to evaluate whether a parent’s ACEs and/or resilience would also be associated with that child’s likelihood of reutilization. We hypothesized that more parental ACEs and/or lower parental resilience would be associated with revisits the emergency room, urgent care, or hospital readmissions.

METHODS

Participants and Study Design

We conducted a prospective cohort study of parents of hospitalized children recruited from the “Hospital-to-Home Outcomes” Studies (H2O I and H2O II).^12,13 H2O I and II were prospective, single-center, randomized controlled trials designed to determine the effectiveness of either a nurse-led transitional home visit (H2O I) or telephone call (H2O II) on 30-day unplanned healthcare reutilization. The trials and this study were approved by the Cincinnati Children’s Institutional Review Board. All parents provided written informed consent.

Details of H2O I and II recruitment and design have been described previously.^12,13 Briefly, children were eligible for inclusion in either study if they were admitted to our institution’s general Hospital Medicine or the Hospital Medicine Complex Care Services; for H2O I, children hospitalized on the Neurology and Neurosurgery services were also eligible.^12,13 Patients were excluded if they were discharged to a residential facility, if they lived outside the home healthcare nurse service area, if they were eligible for skilled home healthcare services (eg, intravenous antibiotics), or if the participating caregiver was non-English speaking.^12,13 In H2O I, families were randomized either to receive a single nurse home visit within 96 hours of discharge or standard of care. In H2O II, families enrolled were randomized to receive a telephone call by a nurse within 96 hours of discharge or standard of care. As we have previously published, randomization in both trials successfully balanced the intervention and control arms with respect to key demographic characteristics.^12,13 For the analyses presented here, we focused on a subset of caregivers 18 years and older whose children were enrolled in either H2O I or II between August 2015 and October 2016. In both H2O trials, face-to-face and paper-based questionnaires were completed by parents during the index hospitalization.

Outcome and Predictors

Our primary outcome was unanticipated healthcare reutilization defined as return to the emergency room, urgent care, or unplanned readmission within 30 days of hospital discharge, consistent with the H2O trials. This was measured using the primary institution’s administrative data supplemented by a utilization database shared across regional hospitals.¹⁴ Readmissions were identified as “unplanned” using a previously validated algorithm,¹⁵ and treated as a dichotomous yes/no variable.

Our primary predictors were parental ACEs and resilience (see Appendix Tables). The ACE questionnaire addresses abuse, neglect, and household dysfunction in the first 18 years of life.¹ It is composed of 10 questions, each with a yes/no response.¹ We defined parents as low (ACE 0), moderate (ACE 1-3), or high (ACE ≥4) risk a priori because previous literature has described poor outcomes in adults with 4 or more ACEs.¹⁶

Given the sensitive nature of the questions, respondents independently completed the ACE questionnaire on paper instead of via the face-to-face survey. Respondents returned the completed questionnaire to the research assistant in a sealed envelope. All families received educational information on relevant hospital and community-based resources (eg, social work).

Parental resilience was measured using the Brief Resilience Scale (BRS). The BRS is 6 items, each on a 5-point Likert scale. Responses were averaged, providing a total score of 1-5; higher scores are representative of higher resilience.¹⁷ We treated the BRS score as a continuous variable. BRS has been used in clinical settings; it has demonstrated positive correlation with social support and negative correlation with fatigue.¹⁷ Parents answered BRS questions during the index pediatric hospitalization in a face-to-face interview.

Parent and Child Characteristics

Parent and child sociodemographic variables were also obtained during the face-to-face interview. Parental variables included age, gender, educational attainment, household income, employment status, and financial and social strain.¹¹ Educational attainment was analyzed in 2 categories—high school or less vs more than high school—because most discharge instructions are written at a high school reading level.¹⁸ Parents reported their annual household income in the following categories: <$15,000; $15,000-$29,999; $30,000-$44,999; $45,000-$59,999; $60,000-$89,999; $90,000-$119,999; ≥$120,000. Employment was dichotomized as not employed/student vs any employment. Financial and social strain were assessed using a series of 9 previously described questions.¹⁹ These questions assessed, via self-report, a family’s ability to make ends meet, ability to pay rent/mortgage or utilities, need to move in with others because of financial reasons, and ability to borrow money if needed, as well as home ownership and parental marital status.^15,19 Strain questions were all dichotomous (yes/no, single/not single). A composite variable was then constructed that categorized those reporting no strain items, 1 to 2 items, 3 to 4 items, and 5 or more items.²⁰

Child variables included race, ethnicity, age, primary care access,²¹ payer, and H2O treatment arm. Race categories were white/Caucasian, black/African American, American Indian or Alaskan Native, Asian or Pacific Islander, and other; ethnicity categories were Hispanic/Latino, non-Hispanic/Latino, and unknown. Given relatively low numbers of children reported to be Hispanic/Latino, we combined race and ethnicity into a single variable, categorized as non-Hispanic/white, non-Hispanic/black, and multiracial/Hispanic/other. Primary care access was assessed using the access subscale to the Parent’s Perception of Primary Care questionnaire. This includes assessment of a family’s ability to travel to their doctor, to see their doctor for routine or sick care, and to get help or advice on evenings or weekends. Scores were categorized as always adequate, almost always adequate, or sometimes/never adequate.²¹ Payer was dichotomized to private or public/self-pay.

Statistical Analyses

We examined the distribution of outcomes, predictors, and covariates. We compared sociodemographic characteristics of those respondents and nonrespondents to the ACE screen using the chi-square test for categorical variables or the t test for continuous variables. We used logistic regression to assess for associations between the independent variables of interest and reutilization, adjusting for potential confounders. To build our adjusted, multivariable model, we decided a priori to include child race/ethnicity, primary care access, financial and social strain, and trial treatment arm. We treated the H2O I control group as the referent group. Other covariates considered for inclusion were caregiver education, household income, employment, and payer. These were included in multivariable models if bivariate associations were significant at the P < .1 level. We assessed an ACE-by-resilience interaction term because we hypothesized that those with more ACEs and lower resilience may have more reutilization outcomes than parents with fewer ACEs and higher resilience. We also evaluated interaction terms between trial arm assignment and predictors to assess effects that may be introduced by the randomization. Predictors in the final logistic regression model were significant at the P < .05 level. Logistic regression assumption of little or no multicollinearity among the independent variables was verified in the final models. All analyses were performed with Stata v16 (Stata Corp, College Station, Texas).

RESULTS

There were a total of 1,787 parent-child dyads enrolled in the H2O I and II during the study period; 1,320 parents (74%) completed the ACE questionnaire and were included the analysis. Included parents were primarily female and employed, as well as educated beyond high school (Table 1). Overall, 64% reported one or more ACEs (range 0 to 9); 45% reported 1to 3, and 19% reported 4 or more ACEs. The most commonly reported ACEs were divorce (n = 573, 43%), exposure to alcoholism (n = 306, 23%), and exposure to mental illness (n = 281, 21%; Figure 1). Parents had a mean BRS score of 3.97 (range 1.17-5.00), with the distribution shown in Figure 2.

Of the 1,320 included patients, the average length of stay was 2.5 days, and 82% of hospitalizations were caused by acute medical issues (eg, bronchiolitis). A total of 211 children experienced a reutilization event within 30 days of discharge. In bivariate analysis, children with parents with 4 or more ACEs had a 2.02-times (95% CI 1.35-3.02) higher odds of experiencing a reutilization event than did those with parents reporting no ACEs. Parents with higher resilience scores had children with a lower odds of reutilization (odds ratio [OR] 0.77 95% CI 0.63-0.95).

In addition to our a priori variables, parental education, employment, and insurance met our significance threshold for inclusion in the multivariable model. The ACE-by-resilience interaction term was not significant and not included in the model. Similarly, there was no significant interaction between ACE and resilience and H2O treatment arm; the interaction terms were not included in the final adjusted model, but treatment arm assignment was kept as a covariate. A total of 1,292 children, out of the 1,320 respondents, remained in the final multivariable model; the excluded 28 had incomplete covariate data but were not otherwise different. In this final adjusted model, children with parents reporting 4 or more ACEs had a 1.69-times (95% CI 1.11-2.60) greater odds of reutilization than did those with parents reporting no ACEs (Table 2). Resilience failed to reach statistical significance in the adjusted model (OR 0.86, 95% CI 0.70-1.07).

DISCUSSION

We found that high-risk parents (4 or more ACEs) had children with an increased odds of healthcare reutilization, suggesting intergenerational effects of ACEs. We did not find a similar effect relating to parental resilience. We also did not find an interaction between parental ACEs and resilience, suggesting that a parent’s reported degree of resilience does not modify the effect of ACEs on reutilization risk.

Parental adversity may be a risk factor for a child’s unanticipated reutilization. We previously demonstrated that parents with 4 or more ACEs have more coping difficulty than a parent with no ACEs after a child’s hospitalization.¹¹ It is possible that parents with high adversity may have poorer coping mechanisms when dealing with a stressful situation, such as a child’s hospitalization. This may have resulted in inequitable outcomes (eg, increased reutilization) for their children. Other studies have confirmed such an intergenerational effect of adversity, linking a parent’s ACEs with poor developmental, behavioral, and health outcomes in their children.^6,22,23 O’Malley et al showed an association of parental ACEs to current adversities,²⁴ such as insurance or housing concerns, that affect the entirety of the household, including children. In short, it appears that parental ACEs may be a compelling predictor of current childhood adversity.

Resilience buffers the negative effects of ACEs; however, we did not find significant associations between resilience and reutilization or an interaction between ACEs and resilience. The factors that may contribute to reutilization are complex. In our previous work, parental resilience was associated with coping difficulty after discharge; but again, did not interact with parental ACEs.¹¹ Here, we suggest that while resilience may buffer the negative effects of ACEs, that buffering may not affect the likelihood of reutilization. It is also possible that the BRS tool is of less relevance on how one handles the stress of a child’s hospitalization. While the BRS is one measure of resilience, there are many other relevant constructs to resilience, such as connection to social supports, that also may also contribute to risk of reutilization.²⁵

Reducing the stress of a hospitalization itself and promoting a safe transition from hospital to home is critical to improving child health outcomes. Our data here, and in our previous work, demonstrate that a history of adversity and one’s current coping ability may drive a parent’s response to a child’s hospitalization and affect their capacity to care for that child after hospital discharge.¹¹ Additional in-hospital supports like child life, behavioral health, or pastoral care could reduce the stress of the hospitalization while also building positive coping mechanisms.^26-29 A meta-analysis demonstrated that such coping interventions can help alleviate the stress of a hospitalization.³⁰ Hill et al demonstrated successful stress reduction in parents of hospitalized children using a “Coping Kit for Parents.”³¹ Further studies are warranted to understand which interventions are most effective for children and families and whether they could be more effectively deployed if the inpatient team knew more about parental ACEs.

Screening for parental ACEs could help to identify patients at highest risk for a poor transition to home. Therefore, screening for parental adversity in clinical settings, including inpatient settings, may be relevant and valuable.³² Additionally, by recognizing the high prevalence of ACEs in an inpatient setting, hospitals and healthcare organizations could be motivated to develop and enact trauma-informed approaches. A trauma-informed care approach recognizes the intersection of trauma with health and social problems. With this recognition, care teams can more sensitively address the trauma as they provide relevant services.³³ Trauma-informed care is a secondary public health prevention approach that would help team members identify the prevalence and effects of trauma via screening, recognize the signs of a maladaptive response to stress, and respond by integrating awareness of trauma into practice management.^28,34 Both the National Academy of Medicine and the Agency for Healthcare Research and Quality have called for such a trauma-informed approach in primary care.³⁵ In response, many healthcare organizations have developed trauma-informed practices to better address the needs of the populations they serve. For example, provider training on this approach has led to improved rapport in patient-provider relationships.³⁶

Although ACE awareness is a component of trauma-informed care, there are still limitations of the original ACE questionnaire developed by Felitti et al. The existing tool is not inclusive of all adversities a parent or child may face. Moreover, its focus is on past exposures and experiences and not current health-related social needs (eg, food insecurity) which have known linkages with a range of health outcomes and health disparities.³⁷ Additionally, the original ACE questionnaire was created as a population level tool and not as a screening tool. If used as a screening tool, providers may view the questions as too sensitive to ask, and parents may have difficulty responding to and understanding the relevance to their child’s care. Therefore, we suggest that more evidence is required to understand how to best adapt ACE questions into a screening processes that may be implemented in a medical setting.

More evidence is also needed to determine when and where such screening may be most useful. A primary care provider would be best equipped to screen caregivers for ACEs given their established relationship with parents and patients. Given the potential relevance of such information for inpatient care provision, information could then flow from primary care to the inpatient team. However, because not all patients have established primary care providers and only 4% of pediatricians screen for ACEs,³⁸ it is important for inpatient medical teams to understand their role in identifying and addressing ACEs during hospital stays. Development of a screening tool, with input from all stakeholders—including parents—that is valid and feasible for use in a pediatric inpatient setting would be an important step forward. This tool should be paired with training in how to discuss these topics in a trauma-informed, nonjudgmental, empathic manner. We see this as a way in which providers can more effectively elicit an accurate response while simultaneously educating parents on the relevance of such sensitive topics during an acute hospital stay. We also recommend that screening should always be paired with response capabilities that connect those who screen positive with resources that could help them to navigate the stress experienced during and after a child’s hospitalization. Furthermore, communication with primary care providers about parents that screen positive should be integrated into the transition process.

This work has several limitations. First, our study was a part of randomized controlled trials conducted in one academic setting, which thereby limits generalizability. For example, we limited our cohort to those who were English-speaking patients only. This may bias our results because respondents with limited English proficiency may have different risk profiles than their English-speaking peers. In addition, the administration of the both the ACE and resilience questionnaires occurred during an acutely stressful period, which may influence how a parent responds to these questions. Also, both of the surveys are self-reported by parents, which may be susceptible to memory and response biases. Relatedly, we had a high number of nonrespondents, particularly to the ACE questionnaire. Our results are therefore only relevant to those who chose to respond and cannot be applied to nonrespondents. Further work assessing why one does or does not respond to such sensitive questions is an important area for future inquiry. Lastly, our cohort had limited medical complexity; future studies may consider links between parental ACEs (and resilience) and morbidity experienced by children with medical complexity.

CONCLUSION

Parents history of adversity is linked to their children’s unanticipated healthcare reutilization after a hospital discharge. Screening for parental stressors during a hospitalization may be an important first step to connecting parents and children to evidence-based interventions capable of mitigating the stress of hospitalization and promoting better, more seamless transitions from hospital to home.

Acknowledgments

Group Members: The following H2O members are nonauthor contributors: JoAnne Bachus, BSN, RN; Monica Borell, BSN, RN; Lenisa V Chang, MA, PhD; Patricia Crawford, RN; Sarah Ferris, BA; Jennifer Gold, BSN, RN; Judy A Heilman, BSN, RN; Jane C Khoury, PhD; Pierce Kuhnell, MS; Karen Lawley, BSN, RN; Margo Moore, MS, BSN, RN; Lynne O’Donnell, BSN, RN; Sarah Riddle, MD; Susan N Sherman, DPA; Angela M Statile, MD, MEd; Karen P Sullivan, BSN, RN; Heather Tubbs-Cooley, PhD, RN; Susan Wade-Murphy, MSN, RN; and Christine M White, MD, MAT.

The authors also thank David Keller, MD, for his guidance on the study.

Disclosures

The authors have no financial relationships or conflicts of interest relevant to this article to disclose.

Funding Source

Supported by funds from the Academic Pediatric Young Investigator Award (Dr A Shah) and the Patient-Centered Outcomes Research Institute Award (IHS-1306-0081, to Dr K Auger, Dr S Shah, Dr H Sucharew, Dr J Simmons), the National Institutes of Health (1K23AI112916, to Dr AF Beck), and the Agency for Healthcare Research and Quality (1K12HS026393-01, to Dr A Shah, K08-HS024735- 01A1, to Dr K Auger). Dr J Haney received Summer Undergraduate Research Fellowship funding through the Summer Undergraduate Research Fellowship at Cincinnati Children’s Hospital Medical Center.

Disclaimer

All statements in this report, including findings and conclusions, are solely those of the authors and do not necessarily represent the views of the Patient-Centered Outcomes Research Institute, its Board of Governors, or the Methodology Committee.

Xerostomia, the subjective sensation of dry mouth, is a common problem developed by geriatric patients. In practice, xerostomia can impair swallowing, speech, and oral hygiene, and if left unchecked, symptoms such as dysphagia and dysarthria can diminish patients’ quality of life (QOL). Salivary gland hypofunction (SGH) is the objective measure of decreased saliva production, determined by sialometry. Although xerostomia and SGH can coexist, the 2 conditions are not necessarily related.^1-4 For this discussion, the term xerostomia will denote dry mouth with or without a concomitant diagnosis of SGH.

Xerostomia is seen in a wide variety of patients with varied comorbidities. It is commonly associated with Sjögren syndrome and head and neck irradiation. The diagnosis and treatment of xerostomia often involves rheumatologists, dentists, otolaryngologists, and oncologists. Additionally, most of the scientific literature about this topic exists in dental journals, such as the Journal of the American Dental Association and the British Dental Journal. Rarer still are studies in the veteran population.⁵

Faced with increasing time pressure to treat the many chronic diseases affecting aging veterans, health care providers (HCPs) tend to deprioritize diagnosing dry mouth. To that point, saliva is often not considered in the same category as other bodily fluids. According to Mandel, “It lacks the drama of blood, the sincerity of sweat…[and] the emotional appeal of tears.”⁶ In reality, saliva plays a critical role in the oral-digestive tract and in swallowing. It contains the first digestive enzymes in the gastrointestinal tract and is key for maintaining homeostasis in the oral cavity.⁷ Decreased saliva production results in difficulties with speech and mastication as well as problems of dysphagia, esophageal dysfunction, dysgeusia, nutritional compromises, new and recurrent dental caries, candidiasis, glossitis, impaired use of dentures, halitosis, and susceptibility to mucosal injury.^7,8 Problems with the production of saliva may lead to loss of QOL, such as enjoying a meal or conversing with others.⁴

Although xerostomia is often associated with advanced age, it is more often explained by the diseases that afflict geriatric patients and the arsenal of medications used to treat them.^2,9-16 Polypharmacy, the simultaneous use of multiple drugs by a single patient for ≥ 1 conditions, is an independent risk factor for xerostomia regardless of the types of medication taken.¹⁶ From 2005 to 2011, older adults in the US significantly increased their prescription medication use and dietary supplements. More than one-third of older adults used ≥ 5 prescription medications concurrently, and two-thirds of older adults used combinations of prescribed medications, over-the-counter medications, and dietary supplements.¹⁷ Several drug classes have the capacity to induce xerostomia, such as antihypertensives, antiulcer agents, anticholinergics, and antidepressants.^2,5,12 Prevalence of dry mouth also can range from 10% to 46%, and women typically are more medicated and symptomatic.^{2,3,9,13,14,16} Xerostomia can also lead to depression and even reduce patients’ will to live.¹⁸ Despite xerostomia’s prevalence and impact on QOL, few patients report it as their chief symptom, and few physicians attempt to treat it.¹⁹

In order to target polypharmacy as a cause of dry mouth, the objectives for this study were to evaluate (1) the prevalence of xerostomia; (2) the relationship between xerostomia and other oral conditions; and (3) the impact of polypharmacy on dry mouth in a veteran population.

 

Methods

This is a retrospective cross-sectional study of all outpatient visits in fiscal year (FY) 2015 (October 1, 2014 to September 30, 2015) at the VA Palo Alto Health Care System (VAPAHCS), a tertiary care US Department of Veterans Affairs (VA) hospital. This study was approved by the Stanford University Institutional Review Board. All patients diagnosed with xerostomia in the 1-year study period were identified using ICD-9 diagnosis codes for dry mouth or disturbance of salivary gland secretion (527.7, 527.8, R68.2) and Systemized Nomenclature of Medicine Clinical Terms (SNOMED CT) codes covering dry mouth, xerostomia, aptyalism, absent salivary secretion, and disturbance of salivary secretion (87715008, 78948009). Data analysts in the VA Office of Business Analytics assisted in gathering data from the Veterans Information Systems and Technology Architecture (VistA) electronic health record.

The statistical analysis of that data was performed using Microsoft Excel. Age and gender distributions were determined for the patients. The relationship between xerostomia and the number and types of medications taken by patients also was examined. A previous Swedish study examining the link between dry mouth and quantities of medications used a scale ranging from 0 to ≥ 7 medications.¹⁶ The scale for this study was made wider to include the following groups: 0-2, 3-5, 6-8, 9-11, and ≥ 12 medications. Items that do not have xerogenic risks, such as medical supplies (eg, gloves, syringes, etc) and topical medications, were excluded from the analysis. Finally, the number of subjects with comorbid problems with speech, dentition, or swallowing (SDS) was recorded. Non-VA medications were included to capture any self- or externally prescribed xerogenic medications.

Results

Of the patients seen at VAPAHCS during FY 2015, 138 had a diagnostic code for xerostomia, including 129 men (93.5%) and 9 women (6.5%). The average (SD) age of this xerostomia cohort was 69.3 (12.6) years, and the 3 most common age groups were 60 to 69 years (37.7%), 70 to 79 years (28.3%), and 80 to 89 years (13.0%) (Table 1). Of those 138 patients with a xerostomia diagnosis, a majority (84; 60.9%) had at least 1 documented SDS problem (Table 2). Conversely, during FY 2015, although 4,971 patients seen at VAPAHCS had documented SDS problems, only 77 (1.5%) had a recorded diagnosis of xerostomia.

Of the 138 patients with xerostomia, 55 (39.9%) were taking ≥ 12 medications, more than twice as many patients as in any of the other groups studied (0-2, 3-5, 6-8, and 9-11 medications taken) (Table 3). On average, each patient with xerostomia filled prescriptions for 10.4 (SD, 7.2) different drugs. In this cohort of 138 patients diagnosed with xerostomia, antihypertensive medications or analgesics were taken by > 50% of patients, while statins, psychiatric medications, antibiotics, proton pump inhibitors, or drugs known to have anticholinergic activity were taken by > 40%. Antihistamines, anticonvulsants, diuretics, or inhaled respiratory agents were used by > 20% of the patients in this cohort (Table 4).

Data on each individual medication were split into 2 categories: the percentage of patients that filled ≥ 1 prescription for that drug, and the total number of prescriptions filled and/or refilled for that drug (ie, including all fills and refills made by individual patients). The 5 most widely used medications in this cohort were omeprazole (39.1%), docusate sodium (29.7%), gabapentin (29.7%), aspirin (27.5%), and hydrocodone/acetaminophen (26.1%) (Table 5). The 5 prescriptions that were cumulatively most filled and/or refilled were omeprazole (128), sildenafil citrate (108), gabapentin (101), hydrocodone/acetaminophen (100), and oxycodone (92) (Table 6). Though sildenafil citrate and oxycodone were among the most-filled prescriptions, these were not included in Table 5 as neither was taken by > 15% of the patients studied. These prescriptions were filled multiple times by a small subset of patients.

Regarding treatment for dry mouth, artificial saliva spray was one of the most widely used (23.2%) and the seventh most-filled prescription within this cohort (86). The only other medication taken by > 15% of patients in a formulation other than a tablet or capsule was chlorhexidine, a germicidal mouthwash used to improve oral care.

Also, 30 (21.7%) patients with a documented xerostomia diagnosis had a history of substance misuse involving use of ≥ 1 of tobacco, alcohol, marijuana, or other illicit drugs.

 

Discussion

Saliva is an essential component for the maintenance of normal oral health.^20,21 Decreased saliva production causes problems, including difficulties with speech, mastication, dysphagia, changes in taste, dental caries, impaired use of prostheses, recurrent infections, halitosis, deterioration of soft tissues, and compromised QOL.^22,23 Among patients with a diagnosed SDS abnormality who were seen at this facility during FY 2015, the prevalence of xerostomia was only 1.5%. However, the true prevalence and incidence of xerostomia among veterans is not known. Given the role of xerostomia as a common risk factor for SDS problems and the polypharmacy exhibited by those presented here with SDS problems, it is probable that xerostomia was underreported in this veteran cohort.

Additionally, although salivary acinar cells are known to atrophy with age, as is consistent with this xerostomia cohort’s average age (SD) of 69.3 (12.6) years, the development of dry mouth is a multifactorial process. The current scientific literature asserts that most salivary loss is due to local and systemic diseases, immunologic disorders, external radiation, and as was highlighted by this study, multiple prescription and nonprescription medications.^24-26

It has also been demonstrated previously that dry mouth complaints and low salivary flow rates are directly proportional to the number of medications taken by patients.^2,27-30 Polypharmacy is therefore an area of great interest, and ≥ 40 categories of xerogenic medications have been identified by investigators such as Sreebny and Schwartz.³¹ Among those, some of the most xerogenic medication classes include antihypertensives, antiulcer agents, anticholinergics, and antidepressants, are all very commonly consumed in this cohort of patients with dry mouth (58.7%, 42.0%, 47.1%, and 38.4%, respectively). The medication regimens within this cohort of veterans with xerostomia were prime examples of polypharmacy as each patient took an average (SD) of 10.4 (7.2) medications, 39.9% took ≥ 12, and 72.5% of patients with xerostomia were taking ≥ 6 prescription drugs during a 1-year period.

Given the dangers of polypharmacy, a more conservative approach to prescribing medications could feasibly help with preventing xerostomia and SGH. In practice, while clinicians try to avoid prescribing anticholinergics, antimuscarinics, and antihistaminergic drugs for geriatric patients, they are tasked with the complex management of medication adverse effects (AEs) when dealing with multiple health conditions. The clinicians’ primary responsibilities are to follow the standard of care and not to introduce unnecessary harm when managing patients, but they also must push for, stay abreast of, and conduct more basic research and clinical trials to inform, adjust, and improve our current standard.

Research into polypharmacy and its role in inducing dry mouth is ongoing. Twenty years ago, Thomson and colleagues identified reduced salivary flow in patients who used antianginals, thyroxine, diuretics, antidepressants, and medications for asthma, while only 5 years earlier Loesche and colleagues reported the role of antiulcer medications, such as proton pump inhibitors, in the development of xerostomia.^2,32 Within the past 5 years, Viljakainen and colleagues and Ohara and colleagues have echoed some of those findings by identifying associations between xerostomia and agents that impact digestive organs.^33,34 A strong association recently was identified between the use of antipsychotic drugs and xerostomia.³⁵ Additionally, when attributing xerostomia to polypharmacy, the interaction between medications is often overlooked in favor of considering the raw number of prescriptions taken. Whereas 1 medication alone may not have drying properties, combinations of medications might be more likely to induce xerostomia. Thomson and colleagues suggested such a situation regarding the interaction between thyroxine and diuretics.³⁶ Future studies should focus on identifying viable substitutes for existing medications that reduce risk for xerostomia without compromising the management of other serious conditions.

 

Treatment

Another pressing question for clinicians concerns artificial saliva. Although 23.2% of patients with dry mouth in this xerostomia cohort used artificial saliva, the efficacy of this treatment is still unproven. Saliva substitutes are often used by patients who cannot produce sufficient amounts of natural saliva. In practice, artificial saliva produces, at best, modest temporary improvement in dry mouth symptoms in up to 40% of patients. At worst, as put forth by the Cochrane Review, artificial saliva may be no better than placebo in treating dry mouth.^37,38 The volumes needed for symptom relief are large, ranging between 40 mL and 150 mL per day depending on the substitute’s composition. Saliva substitutes also must be reapplied throughout each day. This is particularly bothersome when patients must wake up repeatedly to reapply the treatment at night.³⁷ In short, these substances do not seem wholly effective in managing dry mouth, and other options must be made available to patients with refractory xerostomia when artificial saliva and lifestyle modifications fail.

For now, few alternatives exist. Chewing gums and lozenges help to stimulate salivary flow, as do muscarinic agonists like pilocarpine. Unfortunately, muscarinic agonists are seldom used due to cholinergic AEs. Humidifiers are effective in increasing nighttime moisture but are contraindicated in patients with dust mite allergies.³⁹ Reservoir-based devices with automated pumps funnel water and/or salivary substitutes from a fanny pack into patients’ mouths for lubrication.³⁷ Other more esoteric pharmacologic treatments include D-limonene, yohimbine, and amifostine, which purportedly protect salivary progenitor cells, increase peripheral cholinergic activity, and protect salivary glands from free-radical damage during radiation treatment, respectively. Although these agents have shown some promise, D-limonene is difficult to administer given the high dosage required for treatment, yohimbine hasn’t been seriously investigated for improving salivary secretion since 1997, and amifostine isn’t used widely due to its AE profile despite its US Food and Drug Administration approval for prevention of xerostomia.³⁹

Substance Abuse

The impact of smoking on xerostomia remains controversial. Some studies report an association between active smoking and xerostomia; others suggest that the local irritant effect of tobacco smoke may increase salivary gland output.^40,41 The same may be true for chronic alcohol use as there are no epidemiologic studies showing a causal relationship between alcohol use and xerostomia. Studies with rats that are chronically exposed to ethanol have found increased salivary flow rates.⁴² In the xerostomia cohort presented here, 30 patients (21.7%) had a documented history of substance misuse. That percentage is likely underestimated, as substance misuse is often underreported, and frequent use may not always constitute misuse. Therefore, nicotine exposure, alcohol exposure, illicit drug use, and vaping all should be considered during the workup of a patient with xerostomia.

Limitations

It is common for medications to remain in a patient’s health record long after that patient stops taking them. Developing methods to track when patients discontinue their prescriptions will be essential for clearing up uncertainty in our data and in other similar studies. This study also did not account for patients’ medication adherence and the duration of exposure to medications and illicit substances. Furthermore, the results of this veteran study are not easily generalizable as this cohort is disproportionately male, of advanced age, and especially prone to exhibiting both substance use and psychiatric diagnoses relative to the general population. As described by Viljakainen and colleagues, risk factors for xerostomia include advanced age, female gender, low body mass index (BMI), malnutrition, and depressive symptoms, but because the demographic scope of this veteran population was narrow, it was not possible to discern the impact of, for example, gender.³³ Data on variables like BMI, malnutrition, and depressive symptoms were not available. For this study, xerostomia could only be considered as an all-or-nothing phenomenon because the dataset did not describe different levels of dry mouth severity (eg, mild, moderate, severe).

 

Additionally, past polypharmacy studies have acknowledged an inability to tell whether xerostomia is mainly due to medications or to underlying medical conditions. For example, for emphysema, ß-adrenergic stimulation from bronchodilators could cause dry mouth by thickening saliva and decreasing salivary volume, but the pathophysiology and/or cardinal symptoms of emphysema, including chronic obstructive pulmonary disease-associated tachypnea, might contribute independently to dryness.

Though we can make inferences based on the medications taken by this cohort (eg, those taking antihypertensives have high blood pressure), this dataset did not explicitly detail comorbid conditions and ICD codes for chronic diseases that commonly arise with xerostomia. Those conditions, however, are of great clinical importance. Diabetes mellitus, HIV/AIDS, and, classically, Sjögren syndrome, all are known to cause dry mouth.⁴³ Identifying new conditions that co-occur with xerostomia would allow clinicians to describe the root causes of and risk factors for dry mouth and SDS conditions in greater detail. Patients with dry mouth without SDS problems in this dataset are of particular interest as closer examination of their medications and comorbid conditions could help us understand why some individuals and not others develop SDS problems. The subjects of how comorbidities contribute to dry mouth and how their influences can be judged independently from the effects of medications are of great interest to us and will be investigated rigorously in our future studies.

Conclusions

In this cohort, few patients with SDS problems had documentation of a concomitant xerostomia diagnosis. This could represent a true infrequency of dry mouth or more likely, an underrecognition by clinicians. Heightened physician awareness regarding the signs and symptoms of and risk factors for xerostomia is needed to improve providers’ ability to diagnose this condition. 

In particular, polypharmacy should be a major consideration when evaluating patients for xerostomia. This continues to be an important area of research, and some of the latest data on polypharmacy among older patients were compiled in a recent meta-analysis by Tan and colleagues. The authors of that systematic review reiterated the significant association between salivary gland hypofunction and the number of medications taken by patients. They also advocated for the creation of a risk score for medication-induced dry mouth to aid in medication management.⁴⁴ Per their recommendations, it is now as crucial as ever to consider the numbers and types of medications taken by patients, to discontinue unnecessary prescriptions when possible, and to continue developing new strategies for preventing and treating xerostomia.

Xerostomia, the subjective sensation of dry mouth, is a common problem developed by geriatric patients. In practice, xerostomia can impair swallowing, speech, and oral hygiene, and if left unchecked, symptoms such as dysphagia and dysarthria can diminish patients’ quality of life (QOL). Salivary gland hypofunction (SGH) is the objective measure of decreased saliva production, determined by sialometry. Although xerostomia and SGH can coexist, the 2 conditions are not necessarily related.^1-4 For this discussion, the term xerostomia will denote dry mouth with or without a concomitant diagnosis of SGH.

Xerostomia is seen in a wide variety of patients with varied comorbidities. It is commonly associated with Sjögren syndrome and head and neck irradiation. The diagnosis and treatment of xerostomia often involves rheumatologists, dentists, otolaryngologists, and oncologists. Additionally, most of the scientific literature about this topic exists in dental journals, such as the Journal of the American Dental Association and the British Dental Journal. Rarer still are studies in the veteran population.⁵

Faced with increasing time pressure to treat the many chronic diseases affecting aging veterans, health care providers (HCPs) tend to deprioritize diagnosing dry mouth. To that point, saliva is often not considered in the same category as other bodily fluids. According to Mandel, “It lacks the drama of blood, the sincerity of sweat…[and] the emotional appeal of tears.”⁶ In reality, saliva plays a critical role in the oral-digestive tract and in swallowing. It contains the first digestive enzymes in the gastrointestinal tract and is key for maintaining homeostasis in the oral cavity.⁷ Decreased saliva production results in difficulties with speech and mastication as well as problems of dysphagia, esophageal dysfunction, dysgeusia, nutritional compromises, new and recurrent dental caries, candidiasis, glossitis, impaired use of dentures, halitosis, and susceptibility to mucosal injury.^7,8 Problems with the production of saliva may lead to loss of QOL, such as enjoying a meal or conversing with others.⁴

Although xerostomia is often associated with advanced age, it is more often explained by the diseases that afflict geriatric patients and the arsenal of medications used to treat them.^2,9-16 Polypharmacy, the simultaneous use of multiple drugs by a single patient for ≥ 1 conditions, is an independent risk factor for xerostomia regardless of the types of medication taken.¹⁶ From 2005 to 2011, older adults in the US significantly increased their prescription medication use and dietary supplements. More than one-third of older adults used ≥ 5 prescription medications concurrently, and two-thirds of older adults used combinations of prescribed medications, over-the-counter medications, and dietary supplements.¹⁷ Several drug classes have the capacity to induce xerostomia, such as antihypertensives, antiulcer agents, anticholinergics, and antidepressants.^2,5,12 Prevalence of dry mouth also can range from 10% to 46%, and women typically are more medicated and symptomatic.^{2,3,9,13,14,16} Xerostomia can also lead to depression and even reduce patients’ will to live.¹⁸ Despite xerostomia’s prevalence and impact on QOL, few patients report it as their chief symptom, and few physicians attempt to treat it.¹⁹

In order to target polypharmacy as a cause of dry mouth, the objectives for this study were to evaluate (1) the prevalence of xerostomia; (2) the relationship between xerostomia and other oral conditions; and (3) the impact of polypharmacy on dry mouth in a veteran population.

 

Methods

This is a retrospective cross-sectional study of all outpatient visits in fiscal year (FY) 2015 (October 1, 2014 to September 30, 2015) at the VA Palo Alto Health Care System (VAPAHCS), a tertiary care US Department of Veterans Affairs (VA) hospital. This study was approved by the Stanford University Institutional Review Board. All patients diagnosed with xerostomia in the 1-year study period were identified using ICD-9 diagnosis codes for dry mouth or disturbance of salivary gland secretion (527.7, 527.8, R68.2) and Systemized Nomenclature of Medicine Clinical Terms (SNOMED CT) codes covering dry mouth, xerostomia, aptyalism, absent salivary secretion, and disturbance of salivary secretion (87715008, 78948009). Data analysts in the VA Office of Business Analytics assisted in gathering data from the Veterans Information Systems and Technology Architecture (VistA) electronic health record.

The statistical analysis of that data was performed using Microsoft Excel. Age and gender distributions were determined for the patients. The relationship between xerostomia and the number and types of medications taken by patients also was examined. A previous Swedish study examining the link between dry mouth and quantities of medications used a scale ranging from 0 to ≥ 7 medications.¹⁶ The scale for this study was made wider to include the following groups: 0-2, 3-5, 6-8, 9-11, and ≥ 12 medications. Items that do not have xerogenic risks, such as medical supplies (eg, gloves, syringes, etc) and topical medications, were excluded from the analysis. Finally, the number of subjects with comorbid problems with speech, dentition, or swallowing (SDS) was recorded. Non-VA medications were included to capture any self- or externally prescribed xerogenic medications.

Results

Of the patients seen at VAPAHCS during FY 2015, 138 had a diagnostic code for xerostomia, including 129 men (93.5%) and 9 women (6.5%). The average (SD) age of this xerostomia cohort was 69.3 (12.6) years, and the 3 most common age groups were 60 to 69 years (37.7%), 70 to 79 years (28.3%), and 80 to 89 years (13.0%) (Table 1). Of those 138 patients with a xerostomia diagnosis, a majority (84; 60.9%) had at least 1 documented SDS problem (Table 2). Conversely, during FY 2015, although 4,971 patients seen at VAPAHCS had documented SDS problems, only 77 (1.5%) had a recorded diagnosis of xerostomia.

Of the 138 patients with xerostomia, 55 (39.9%) were taking ≥ 12 medications, more than twice as many patients as in any of the other groups studied (0-2, 3-5, 6-8, and 9-11 medications taken) (Table 3). On average, each patient with xerostomia filled prescriptions for 10.4 (SD, 7.2) different drugs. In this cohort of 138 patients diagnosed with xerostomia, antihypertensive medications or analgesics were taken by > 50% of patients, while statins, psychiatric medications, antibiotics, proton pump inhibitors, or drugs known to have anticholinergic activity were taken by > 40%. Antihistamines, anticonvulsants, diuretics, or inhaled respiratory agents were used by > 20% of the patients in this cohort (Table 4).

Data on each individual medication were split into 2 categories: the percentage of patients that filled ≥ 1 prescription for that drug, and the total number of prescriptions filled and/or refilled for that drug (ie, including all fills and refills made by individual patients). The 5 most widely used medications in this cohort were omeprazole (39.1%), docusate sodium (29.7%), gabapentin (29.7%), aspirin (27.5%), and hydrocodone/acetaminophen (26.1%) (Table 5). The 5 prescriptions that were cumulatively most filled and/or refilled were omeprazole (128), sildenafil citrate (108), gabapentin (101), hydrocodone/acetaminophen (100), and oxycodone (92) (Table 6). Though sildenafil citrate and oxycodone were among the most-filled prescriptions, these were not included in Table 5 as neither was taken by > 15% of the patients studied. These prescriptions were filled multiple times by a small subset of patients.

Regarding treatment for dry mouth, artificial saliva spray was one of the most widely used (23.2%) and the seventh most-filled prescription within this cohort (86). The only other medication taken by > 15% of patients in a formulation other than a tablet or capsule was chlorhexidine, a germicidal mouthwash used to improve oral care.

Also, 30 (21.7%) patients with a documented xerostomia diagnosis had a history of substance misuse involving use of ≥ 1 of tobacco, alcohol, marijuana, or other illicit drugs.

 

Discussion

Saliva is an essential component for the maintenance of normal oral health.^20,21 Decreased saliva production causes problems, including difficulties with speech, mastication, dysphagia, changes in taste, dental caries, impaired use of prostheses, recurrent infections, halitosis, deterioration of soft tissues, and compromised QOL.^22,23 Among patients with a diagnosed SDS abnormality who were seen at this facility during FY 2015, the prevalence of xerostomia was only 1.5%. However, the true prevalence and incidence of xerostomia among veterans is not known. Given the role of xerostomia as a common risk factor for SDS problems and the polypharmacy exhibited by those presented here with SDS problems, it is probable that xerostomia was underreported in this veteran cohort.

Additionally, although salivary acinar cells are known to atrophy with age, as is consistent with this xerostomia cohort’s average age (SD) of 69.3 (12.6) years, the development of dry mouth is a multifactorial process. The current scientific literature asserts that most salivary loss is due to local and systemic diseases, immunologic disorders, external radiation, and as was highlighted by this study, multiple prescription and nonprescription medications.^24-26

It has also been demonstrated previously that dry mouth complaints and low salivary flow rates are directly proportional to the number of medications taken by patients.^2,27-30 Polypharmacy is therefore an area of great interest, and ≥ 40 categories of xerogenic medications have been identified by investigators such as Sreebny and Schwartz.³¹ Among those, some of the most xerogenic medication classes include antihypertensives, antiulcer agents, anticholinergics, and antidepressants, are all very commonly consumed in this cohort of patients with dry mouth (58.7%, 42.0%, 47.1%, and 38.4%, respectively). The medication regimens within this cohort of veterans with xerostomia were prime examples of polypharmacy as each patient took an average (SD) of 10.4 (7.2) medications, 39.9% took ≥ 12, and 72.5% of patients with xerostomia were taking ≥ 6 prescription drugs during a 1-year period.

Given the dangers of polypharmacy, a more conservative approach to prescribing medications could feasibly help with preventing xerostomia and SGH. In practice, while clinicians try to avoid prescribing anticholinergics, antimuscarinics, and antihistaminergic drugs for geriatric patients, they are tasked with the complex management of medication adverse effects (AEs) when dealing with multiple health conditions. The clinicians’ primary responsibilities are to follow the standard of care and not to introduce unnecessary harm when managing patients, but they also must push for, stay abreast of, and conduct more basic research and clinical trials to inform, adjust, and improve our current standard.

Research into polypharmacy and its role in inducing dry mouth is ongoing. Twenty years ago, Thomson and colleagues identified reduced salivary flow in patients who used antianginals, thyroxine, diuretics, antidepressants, and medications for asthma, while only 5 years earlier Loesche and colleagues reported the role of antiulcer medications, such as proton pump inhibitors, in the development of xerostomia.^2,32 Within the past 5 years, Viljakainen and colleagues and Ohara and colleagues have echoed some of those findings by identifying associations between xerostomia and agents that impact digestive organs.^33,34 A strong association recently was identified between the use of antipsychotic drugs and xerostomia.³⁵ Additionally, when attributing xerostomia to polypharmacy, the interaction between medications is often overlooked in favor of considering the raw number of prescriptions taken. Whereas 1 medication alone may not have drying properties, combinations of medications might be more likely to induce xerostomia. Thomson and colleagues suggested such a situation regarding the interaction between thyroxine and diuretics.³⁶ Future studies should focus on identifying viable substitutes for existing medications that reduce risk for xerostomia without compromising the management of other serious conditions.

 

Treatment

Another pressing question for clinicians concerns artificial saliva. Although 23.2% of patients with dry mouth in this xerostomia cohort used artificial saliva, the efficacy of this treatment is still unproven. Saliva substitutes are often used by patients who cannot produce sufficient amounts of natural saliva. In practice, artificial saliva produces, at best, modest temporary improvement in dry mouth symptoms in up to 40% of patients. At worst, as put forth by the Cochrane Review, artificial saliva may be no better than placebo in treating dry mouth.^37,38 The volumes needed for symptom relief are large, ranging between 40 mL and 150 mL per day depending on the substitute’s composition. Saliva substitutes also must be reapplied throughout each day. This is particularly bothersome when patients must wake up repeatedly to reapply the treatment at night.³⁷ In short, these substances do not seem wholly effective in managing dry mouth, and other options must be made available to patients with refractory xerostomia when artificial saliva and lifestyle modifications fail.

For now, few alternatives exist. Chewing gums and lozenges help to stimulate salivary flow, as do muscarinic agonists like pilocarpine. Unfortunately, muscarinic agonists are seldom used due to cholinergic AEs. Humidifiers are effective in increasing nighttime moisture but are contraindicated in patients with dust mite allergies.³⁹ Reservoir-based devices with automated pumps funnel water and/or salivary substitutes from a fanny pack into patients’ mouths for lubrication.³⁷ Other more esoteric pharmacologic treatments include D-limonene, yohimbine, and amifostine, which purportedly protect salivary progenitor cells, increase peripheral cholinergic activity, and protect salivary glands from free-radical damage during radiation treatment, respectively. Although these agents have shown some promise, D-limonene is difficult to administer given the high dosage required for treatment, yohimbine hasn’t been seriously investigated for improving salivary secretion since 1997, and amifostine isn’t used widely due to its AE profile despite its US Food and Drug Administration approval for prevention of xerostomia.³⁹

Substance Abuse

The impact of smoking on xerostomia remains controversial. Some studies report an association between active smoking and xerostomia; others suggest that the local irritant effect of tobacco smoke may increase salivary gland output.^40,41 The same may be true for chronic alcohol use as there are no epidemiologic studies showing a causal relationship between alcohol use and xerostomia. Studies with rats that are chronically exposed to ethanol have found increased salivary flow rates.⁴² In the xerostomia cohort presented here, 30 patients (21.7%) had a documented history of substance misuse. That percentage is likely underestimated, as substance misuse is often underreported, and frequent use may not always constitute misuse. Therefore, nicotine exposure, alcohol exposure, illicit drug use, and vaping all should be considered during the workup of a patient with xerostomia.

Limitations

It is common for medications to remain in a patient’s health record long after that patient stops taking them. Developing methods to track when patients discontinue their prescriptions will be essential for clearing up uncertainty in our data and in other similar studies. This study also did not account for patients’ medication adherence and the duration of exposure to medications and illicit substances. Furthermore, the results of this veteran study are not easily generalizable as this cohort is disproportionately male, of advanced age, and especially prone to exhibiting both substance use and psychiatric diagnoses relative to the general population. As described by Viljakainen and colleagues, risk factors for xerostomia include advanced age, female gender, low body mass index (BMI), malnutrition, and depressive symptoms, but because the demographic scope of this veteran population was narrow, it was not possible to discern the impact of, for example, gender.³³ Data on variables like BMI, malnutrition, and depressive symptoms were not available. For this study, xerostomia could only be considered as an all-or-nothing phenomenon because the dataset did not describe different levels of dry mouth severity (eg, mild, moderate, severe).

 

Additionally, past polypharmacy studies have acknowledged an inability to tell whether xerostomia is mainly due to medications or to underlying medical conditions. For example, for emphysema, ß-adrenergic stimulation from bronchodilators could cause dry mouth by thickening saliva and decreasing salivary volume, but the pathophysiology and/or cardinal symptoms of emphysema, including chronic obstructive pulmonary disease-associated tachypnea, might contribute independently to dryness.

Though we can make inferences based on the medications taken by this cohort (eg, those taking antihypertensives have high blood pressure), this dataset did not explicitly detail comorbid conditions and ICD codes for chronic diseases that commonly arise with xerostomia. Those conditions, however, are of great clinical importance. Diabetes mellitus, HIV/AIDS, and, classically, Sjögren syndrome, all are known to cause dry mouth.⁴³ Identifying new conditions that co-occur with xerostomia would allow clinicians to describe the root causes of and risk factors for dry mouth and SDS conditions in greater detail. Patients with dry mouth without SDS problems in this dataset are of particular interest as closer examination of their medications and comorbid conditions could help us understand why some individuals and not others develop SDS problems. The subjects of how comorbidities contribute to dry mouth and how their influences can be judged independently from the effects of medications are of great interest to us and will be investigated rigorously in our future studies.

Conclusions

In this cohort, few patients with SDS problems had documentation of a concomitant xerostomia diagnosis. This could represent a true infrequency of dry mouth or more likely, an underrecognition by clinicians. Heightened physician awareness regarding the signs and symptoms of and risk factors for xerostomia is needed to improve providers’ ability to diagnose this condition. 

In particular, polypharmacy should be a major consideration when evaluating patients for xerostomia. This continues to be an important area of research, and some of the latest data on polypharmacy among older patients were compiled in a recent meta-analysis by Tan and colleagues. The authors of that systematic review reiterated the significant association between salivary gland hypofunction and the number of medications taken by patients. They also advocated for the creation of a risk score for medication-induced dry mouth to aid in medication management.⁴⁴ Per their recommendations, it is now as crucial as ever to consider the numbers and types of medications taken by patients, to discontinue unnecessary prescriptions when possible, and to continue developing new strategies for preventing and treating xerostomia.

Xerostomia, the subjective sensation of dry mouth, is a common problem developed by geriatric patients. In practice, xerostomia can impair swallowing, speech, and oral hygiene, and if left unchecked, symptoms such as dysphagia and dysarthria can diminish patients’ quality of life (QOL). Salivary gland hypofunction (SGH) is the objective measure of decreased saliva production, determined by sialometry. Although xerostomia and SGH can coexist, the 2 conditions are not necessarily related.^1-4 For this discussion, the term xerostomia will denote dry mouth with or without a concomitant diagnosis of SGH.

Xerostomia is seen in a wide variety of patients with varied comorbidities. It is commonly associated with Sjögren syndrome and head and neck irradiation. The diagnosis and treatment of xerostomia often involves rheumatologists, dentists, otolaryngologists, and oncologists. Additionally, most of the scientific literature about this topic exists in dental journals, such as the Journal of the American Dental Association and the British Dental Journal. Rarer still are studies in the veteran population.⁵

Faced with increasing time pressure to treat the many chronic diseases affecting aging veterans, health care providers (HCPs) tend to deprioritize diagnosing dry mouth. To that point, saliva is often not considered in the same category as other bodily fluids. According to Mandel, “It lacks the drama of blood, the sincerity of sweat…[and] the emotional appeal of tears.”⁶ In reality, saliva plays a critical role in the oral-digestive tract and in swallowing. It contains the first digestive enzymes in the gastrointestinal tract and is key for maintaining homeostasis in the oral cavity.⁷ Decreased saliva production results in difficulties with speech and mastication as well as problems of dysphagia, esophageal dysfunction, dysgeusia, nutritional compromises, new and recurrent dental caries, candidiasis, glossitis, impaired use of dentures, halitosis, and susceptibility to mucosal injury.^7,8 Problems with the production of saliva may lead to loss of QOL, such as enjoying a meal or conversing with others.⁴

Although xerostomia is often associated with advanced age, it is more often explained by the diseases that afflict geriatric patients and the arsenal of medications used to treat them.^2,9-16 Polypharmacy, the simultaneous use of multiple drugs by a single patient for ≥ 1 conditions, is an independent risk factor for xerostomia regardless of the types of medication taken.¹⁶ From 2005 to 2011, older adults in the US significantly increased their prescription medication use and dietary supplements. More than one-third of older adults used ≥ 5 prescription medications concurrently, and two-thirds of older adults used combinations of prescribed medications, over-the-counter medications, and dietary supplements.¹⁷ Several drug classes have the capacity to induce xerostomia, such as antihypertensives, antiulcer agents, anticholinergics, and antidepressants.^2,5,12 Prevalence of dry mouth also can range from 10% to 46%, and women typically are more medicated and symptomatic.^{2,3,9,13,14,16} Xerostomia can also lead to depression and even reduce patients’ will to live.¹⁸ Despite xerostomia’s prevalence and impact on QOL, few patients report it as their chief symptom, and few physicians attempt to treat it.¹⁹

In order to target polypharmacy as a cause of dry mouth, the objectives for this study were to evaluate (1) the prevalence of xerostomia; (2) the relationship between xerostomia and other oral conditions; and (3) the impact of polypharmacy on dry mouth in a veteran population.

 

Methods

This is a retrospective cross-sectional study of all outpatient visits in fiscal year (FY) 2015 (October 1, 2014 to September 30, 2015) at the VA Palo Alto Health Care System (VAPAHCS), a tertiary care US Department of Veterans Affairs (VA) hospital. This study was approved by the Stanford University Institutional Review Board. All patients diagnosed with xerostomia in the 1-year study period were identified using ICD-9 diagnosis codes for dry mouth or disturbance of salivary gland secretion (527.7, 527.8, R68.2) and Systemized Nomenclature of Medicine Clinical Terms (SNOMED CT) codes covering dry mouth, xerostomia, aptyalism, absent salivary secretion, and disturbance of salivary secretion (87715008, 78948009). Data analysts in the VA Office of Business Analytics assisted in gathering data from the Veterans Information Systems and Technology Architecture (VistA) electronic health record.

The statistical analysis of that data was performed using Microsoft Excel. Age and gender distributions were determined for the patients. The relationship between xerostomia and the number and types of medications taken by patients also was examined. A previous Swedish study examining the link between dry mouth and quantities of medications used a scale ranging from 0 to ≥ 7 medications.¹⁶ The scale for this study was made wider to include the following groups: 0-2, 3-5, 6-8, 9-11, and ≥ 12 medications. Items that do not have xerogenic risks, such as medical supplies (eg, gloves, syringes, etc) and topical medications, were excluded from the analysis. Finally, the number of subjects with comorbid problems with speech, dentition, or swallowing (SDS) was recorded. Non-VA medications were included to capture any self- or externally prescribed xerogenic medications.

Results

Of the patients seen at VAPAHCS during FY 2015, 138 had a diagnostic code for xerostomia, including 129 men (93.5%) and 9 women (6.5%). The average (SD) age of this xerostomia cohort was 69.3 (12.6) years, and the 3 most common age groups were 60 to 69 years (37.7%), 70 to 79 years (28.3%), and 80 to 89 years (13.0%) (Table 1). Of those 138 patients with a xerostomia diagnosis, a majority (84; 60.9%) had at least 1 documented SDS problem (Table 2). Conversely, during FY 2015, although 4,971 patients seen at VAPAHCS had documented SDS problems, only 77 (1.5%) had a recorded diagnosis of xerostomia.

Of the 138 patients with xerostomia, 55 (39.9%) were taking ≥ 12 medications, more than twice as many patients as in any of the other groups studied (0-2, 3-5, 6-8, and 9-11 medications taken) (Table 3). On average, each patient with xerostomia filled prescriptions for 10.4 (SD, 7.2) different drugs. In this cohort of 138 patients diagnosed with xerostomia, antihypertensive medications or analgesics were taken by > 50% of patients, while statins, psychiatric medications, antibiotics, proton pump inhibitors, or drugs known to have anticholinergic activity were taken by > 40%. Antihistamines, anticonvulsants, diuretics, or inhaled respiratory agents were used by > 20% of the patients in this cohort (Table 4).

Data on each individual medication were split into 2 categories: the percentage of patients that filled ≥ 1 prescription for that drug, and the total number of prescriptions filled and/or refilled for that drug (ie, including all fills and refills made by individual patients). The 5 most widely used medications in this cohort were omeprazole (39.1%), docusate sodium (29.7%), gabapentin (29.7%), aspirin (27.5%), and hydrocodone/acetaminophen (26.1%) (Table 5). The 5 prescriptions that were cumulatively most filled and/or refilled were omeprazole (128), sildenafil citrate (108), gabapentin (101), hydrocodone/acetaminophen (100), and oxycodone (92) (Table 6). Though sildenafil citrate and oxycodone were among the most-filled prescriptions, these were not included in Table 5 as neither was taken by > 15% of the patients studied. These prescriptions were filled multiple times by a small subset of patients.

Regarding treatment for dry mouth, artificial saliva spray was one of the most widely used (23.2%) and the seventh most-filled prescription within this cohort (86). The only other medication taken by > 15% of patients in a formulation other than a tablet or capsule was chlorhexidine, a germicidal mouthwash used to improve oral care.

Also, 30 (21.7%) patients with a documented xerostomia diagnosis had a history of substance misuse involving use of ≥ 1 of tobacco, alcohol, marijuana, or other illicit drugs.

 

Discussion

Saliva is an essential component for the maintenance of normal oral health.^20,21 Decreased saliva production causes problems, including difficulties with speech, mastication, dysphagia, changes in taste, dental caries, impaired use of prostheses, recurrent infections, halitosis, deterioration of soft tissues, and compromised QOL.^22,23 Among patients with a diagnosed SDS abnormality who were seen at this facility during FY 2015, the prevalence of xerostomia was only 1.5%. However, the true prevalence and incidence of xerostomia among veterans is not known. Given the role of xerostomia as a common risk factor for SDS problems and the polypharmacy exhibited by those presented here with SDS problems, it is probable that xerostomia was underreported in this veteran cohort.

Additionally, although salivary acinar cells are known to atrophy with age, as is consistent with this xerostomia cohort’s average age (SD) of 69.3 (12.6) years, the development of dry mouth is a multifactorial process. The current scientific literature asserts that most salivary loss is due to local and systemic diseases, immunologic disorders, external radiation, and as was highlighted by this study, multiple prescription and nonprescription medications.^24-26

It has also been demonstrated previously that dry mouth complaints and low salivary flow rates are directly proportional to the number of medications taken by patients.^2,27-30 Polypharmacy is therefore an area of great interest, and ≥ 40 categories of xerogenic medications have been identified by investigators such as Sreebny and Schwartz.³¹ Among those, some of the most xerogenic medication classes include antihypertensives, antiulcer agents, anticholinergics, and antidepressants, are all very commonly consumed in this cohort of patients with dry mouth (58.7%, 42.0%, 47.1%, and 38.4%, respectively). The medication regimens within this cohort of veterans with xerostomia were prime examples of polypharmacy as each patient took an average (SD) of 10.4 (7.2) medications, 39.9% took ≥ 12, and 72.5% of patients with xerostomia were taking ≥ 6 prescription drugs during a 1-year period.

Given the dangers of polypharmacy, a more conservative approach to prescribing medications could feasibly help with preventing xerostomia and SGH. In practice, while clinicians try to avoid prescribing anticholinergics, antimuscarinics, and antihistaminergic drugs for geriatric patients, they are tasked with the complex management of medication adverse effects (AEs) when dealing with multiple health conditions. The clinicians’ primary responsibilities are to follow the standard of care and not to introduce unnecessary harm when managing patients, but they also must push for, stay abreast of, and conduct more basic research and clinical trials to inform, adjust, and improve our current standard.

Research into polypharmacy and its role in inducing dry mouth is ongoing. Twenty years ago, Thomson and colleagues identified reduced salivary flow in patients who used antianginals, thyroxine, diuretics, antidepressants, and medications for asthma, while only 5 years earlier Loesche and colleagues reported the role of antiulcer medications, such as proton pump inhibitors, in the development of xerostomia.^2,32 Within the past 5 years, Viljakainen and colleagues and Ohara and colleagues have echoed some of those findings by identifying associations between xerostomia and agents that impact digestive organs.^33,34 A strong association recently was identified between the use of antipsychotic drugs and xerostomia.³⁵ Additionally, when attributing xerostomia to polypharmacy, the interaction between medications is often overlooked in favor of considering the raw number of prescriptions taken. Whereas 1 medication alone may not have drying properties, combinations of medications might be more likely to induce xerostomia. Thomson and colleagues suggested such a situation regarding the interaction between thyroxine and diuretics.³⁶ Future studies should focus on identifying viable substitutes for existing medications that reduce risk for xerostomia without compromising the management of other serious conditions.

 

Treatment

Another pressing question for clinicians concerns artificial saliva. Although 23.2% of patients with dry mouth in this xerostomia cohort used artificial saliva, the efficacy of this treatment is still unproven. Saliva substitutes are often used by patients who cannot produce sufficient amounts of natural saliva. In practice, artificial saliva produces, at best, modest temporary improvement in dry mouth symptoms in up to 40% of patients. At worst, as put forth by the Cochrane Review, artificial saliva may be no better than placebo in treating dry mouth.^37,38 The volumes needed for symptom relief are large, ranging between 40 mL and 150 mL per day depending on the substitute’s composition. Saliva substitutes also must be reapplied throughout each day. This is particularly bothersome when patients must wake up repeatedly to reapply the treatment at night.³⁷ In short, these substances do not seem wholly effective in managing dry mouth, and other options must be made available to patients with refractory xerostomia when artificial saliva and lifestyle modifications fail.

For now, few alternatives exist. Chewing gums and lozenges help to stimulate salivary flow, as do muscarinic agonists like pilocarpine. Unfortunately, muscarinic agonists are seldom used due to cholinergic AEs. Humidifiers are effective in increasing nighttime moisture but are contraindicated in patients with dust mite allergies.³⁹ Reservoir-based devices with automated pumps funnel water and/or salivary substitutes from a fanny pack into patients’ mouths for lubrication.³⁷ Other more esoteric pharmacologic treatments include D-limonene, yohimbine, and amifostine, which purportedly protect salivary progenitor cells, increase peripheral cholinergic activity, and protect salivary glands from free-radical damage during radiation treatment, respectively. Although these agents have shown some promise, D-limonene is difficult to administer given the high dosage required for treatment, yohimbine hasn’t been seriously investigated for improving salivary secretion since 1997, and amifostine isn’t used widely due to its AE profile despite its US Food and Drug Administration approval for prevention of xerostomia.³⁹

Substance Abuse

The impact of smoking on xerostomia remains controversial. Some studies report an association between active smoking and xerostomia; others suggest that the local irritant effect of tobacco smoke may increase salivary gland output.^40,41 The same may be true for chronic alcohol use as there are no epidemiologic studies showing a causal relationship between alcohol use and xerostomia. Studies with rats that are chronically exposed to ethanol have found increased salivary flow rates.⁴² In the xerostomia cohort presented here, 30 patients (21.7%) had a documented history of substance misuse. That percentage is likely underestimated, as substance misuse is often underreported, and frequent use may not always constitute misuse. Therefore, nicotine exposure, alcohol exposure, illicit drug use, and vaping all should be considered during the workup of a patient with xerostomia.

Limitations

It is common for medications to remain in a patient’s health record long after that patient stops taking them. Developing methods to track when patients discontinue their prescriptions will be essential for clearing up uncertainty in our data and in other similar studies. This study also did not account for patients’ medication adherence and the duration of exposure to medications and illicit substances. Furthermore, the results of this veteran study are not easily generalizable as this cohort is disproportionately male, of advanced age, and especially prone to exhibiting both substance use and psychiatric diagnoses relative to the general population. As described by Viljakainen and colleagues, risk factors for xerostomia include advanced age, female gender, low body mass index (BMI), malnutrition, and depressive symptoms, but because the demographic scope of this veteran population was narrow, it was not possible to discern the impact of, for example, gender.³³ Data on variables like BMI, malnutrition, and depressive symptoms were not available. For this study, xerostomia could only be considered as an all-or-nothing phenomenon because the dataset did not describe different levels of dry mouth severity (eg, mild, moderate, severe).

 

Additionally, past polypharmacy studies have acknowledged an inability to tell whether xerostomia is mainly due to medications or to underlying medical conditions. For example, for emphysema, ß-adrenergic stimulation from bronchodilators could cause dry mouth by thickening saliva and decreasing salivary volume, but the pathophysiology and/or cardinal symptoms of emphysema, including chronic obstructive pulmonary disease-associated tachypnea, might contribute independently to dryness.

Though we can make inferences based on the medications taken by this cohort (eg, those taking antihypertensives have high blood pressure), this dataset did not explicitly detail comorbid conditions and ICD codes for chronic diseases that commonly arise with xerostomia. Those conditions, however, are of great clinical importance. Diabetes mellitus, HIV/AIDS, and, classically, Sjögren syndrome, all are known to cause dry mouth.⁴³ Identifying new conditions that co-occur with xerostomia would allow clinicians to describe the root causes of and risk factors for dry mouth and SDS conditions in greater detail. Patients with dry mouth without SDS problems in this dataset are of particular interest as closer examination of their medications and comorbid conditions could help us understand why some individuals and not others develop SDS problems. The subjects of how comorbidities contribute to dry mouth and how their influences can be judged independently from the effects of medications are of great interest to us and will be investigated rigorously in our future studies.

Conclusions

In this cohort, few patients with SDS problems had documentation of a concomitant xerostomia diagnosis. This could represent a true infrequency of dry mouth or more likely, an underrecognition by clinicians. Heightened physician awareness regarding the signs and symptoms of and risk factors for xerostomia is needed to improve providers’ ability to diagnose this condition. 

In particular, polypharmacy should be a major consideration when evaluating patients for xerostomia. This continues to be an important area of research, and some of the latest data on polypharmacy among older patients were compiled in a recent meta-analysis by Tan and colleagues. The authors of that systematic review reiterated the significant association between salivary gland hypofunction and the number of medications taken by patients. They also advocated for the creation of a risk score for medication-induced dry mouth to aid in medication management.⁴⁴ Per their recommendations, it is now as crucial as ever to consider the numbers and types of medications taken by patients, to discontinue unnecessary prescriptions when possible, and to continue developing new strategies for preventing and treating xerostomia.

Metrics from 2017 at the Fayetteville Veterans Affairs Heath Care Center (FVAHCC) Anticoagulation Clinic indicate that 43% of patients with atrial fibrillation (AF) who are prescribed warfarin have difficulty maintaining a therapeutic international normalized ratio (INR). These patients require frequent clinic appointments to adjust their regimens to ensure anticoagulation efficacy. FVAHCC policy requires a patient to return to the clinic for repeat INR evaluation within 5 to 14 days of the visit where INR was outside of the established therapeutic range.¹ These frequent INR monitoring appointments increase patient and health care provider burden.

Direct oral anticoagulants (DOACs) are an alternative to warfarin for patients with AF who require anticoagulation. DOACs, which do not require regular efficacy monitoring, can be beneficial to patients who struggle to maintain a therapeutic INR when taking warfarin. FVAHCC policy regarding warfarin therapy monitoring allows for a maximum of 6 weeks between appointments. This period is often extended to 3 to 6 months for patients on DOACs.¹

At FVAHCC, patients prescribed warfarin are managed in a centralized Anticoagulation Clinic led by a clinical pharmacy specialist (CPS). When a patient reports for an appointment, a clinical pharmacy technician performs point-of-care INR testing and asks standardized questions regarding therapy, including an assessment of adherence. The CPS then evaluates the patient’s INR test results, adjusts the dosage of warfarin as indicated, and determines appropriate follow-up.

A patient who is prescribed a DOAC is monitored by a CPS who works within a patient aligned care team (PACT). The PACT, a multidisciplinary team providing health care to veterans, includes physicians, nurses, pharmacists, dieticians, and mental health providers. Each CPS covers 3 or 4 PACTs. These pharmacists monitor all aspects of DOAC therapy at regular intervals, including renal and hepatic function, complete blood counts, medication adherence, and adverse effects.

Clinic and patient INR data are tracked using a time in therapeutic range (TTR) report generated by the US Department of Veterans Affairs (VA). The TTR report provides clinical information to enhance patient anticoagulation care.² The TTR report identifies patients with an active order for warfarin and a diagnosis of AF or venous thromboembolism (VTE) whose INR is within therapeutic range (between 2 and 3) < 60% of the time over the previous 160 days.² The patient must have had at least 3 INR levels drawn within that time frame for a TTR report calculation.² The report excludes patients who were first prescribed warfarin within the previous 42 days and those with mechanical heart valves. The TTR report is used by the VA to see concrete facility-level results for quality improvement efforts.²

A quality improvement screening tool was developed to identify patients with AF being treated with warfarin who may appropriately transition to DOAC therapy. Anticoagulation Clinic patients were eligible for further evaluation if they had a TTR report level of < 60% and were prescribed indefinite warfarin therapy for AF.

The national VA goal is to have patient TTR report levels read > 60%. Therefore, the primary objective of this project was to improve Anticoagulation Clinic TTR metrics by targeting patients with TTR levels below the national goal.²

Patients who were successfully converted from warfarin to a DOAC were no longer included in Anticoagulation Clinic metrics and instead were followed by a PACT CPS. Thus, it was hypothesized that the average number of monthly Anticoagulation Clinic encounters would decrease on successful implementation of the screening tool. A secondary endpoint of the study evaluated the change in the total number of encounters of those who converted from warfarin to a DOAC.

Fewer clinic encounters could increase time available for the CPS to incorporate other initiatives into workflow and could increase clinic availability for newly referred veterans.

 

Methods

As this undertaking was considered to be a quality improvement project, institutional review board approval was not required. During an 8-week screening period (August to September 2018), the DOAC screening tool was implemented into the Anticoagulation Clinic workflow. This screening tool (Figure 1) was established based on VA Pharmacy Benefit Management (PBM) Service’s Criteria for Use for Stroke Prevention in Nonvalvular Atrial Fibrillation, a national set of standards used to determine appropriate candidates for DOAC therapy.³

Exclusion criteria included patients with INR goals < 2 or > 3, patients with a diagnosis of VTE, and patients with weight > 120 kg. Patients with a diagnosis of VTE were excluded due to the variability in therapy duration. Weight cutoffs were based on recommendations by the International Society on Thrombosis and Haemostasis. Due to a lack of available data, it was suggested that clinical judgment be used in patients whose weight was > 120 kg.⁴

During the screening period, weekly TTR reports identified patients in the clinic who had TTR < 60%. When a patient with a TTR report results of < 60% also had a scheduled appointment within a week, a CPS then further reviewed patient eligibility using the DOAC screening tool. On arrival for an appointment, the eligible patient was counseled on DOAC medications and the differences between warfarin and DOACs, including monitoring requirements. Patients had the option to switch to DOAC therapy or remain on warfarin.

The change in the average number of monthly Anticoagulation Clinic encounters for 3 months prior to the screening period (May to July 2018) and 2 months following screening (October to November 2018) was evaluated to measure the impact of the DOAC screening tool. The total number of encounters in the clinic was assessed using the monthly VA reports and were averaged for each period. Then data from the 2 periods were compared.

The monthly encounter reports, a data tool that monitors the number of unique visits per veteran each calendar month, also were used to generate a secondary endpoint showing the number of encounters in the Anticoagulation Clinic associated with patients who switched to a DOAC, including visits prior to changing therapy, and before and after the screening period.

Student’s t test was used to compare the change in encounter frequency before and after screening tool implementation for both primary and secondary endpoints. α was defined as .05 a priori. Continuous data were presented as means and standard deviations. Data were calculated with Microsoft Excel 2016.

Results

For the 3 months before the 8-week screening period, an average of 476 Anticoagulation Clinic encounters per month were documented. Two months of data following the screening period averaged 546 encounters per month. There were an average of 70 additional encounters per month after screening tool implementation (P = .15), reflecting the study’s primary objective.

A total of 219 patients in the Anticoagulation Clinic were identified as having a TTR report results of < 60% during the 8-week screening period (Figure 2). Eighty-two of those patients (37.4%) were considered eligible to switch from warfarin to DOAC therapy. Thirty of those eligible patients (13.7%) switched to a DOAC. A total of 107 clinic encounters (22.5%) was associated with these 30 patients prior to screening and 32 associated encounters (5.9%) following screening (P = .01). Of the remaining 137 patients (62.6%) who were ineligible for DOAC therapy, the most common reason for disqualification was a diagnosis of VTE (Table).

 

Discussion

The general results of this quality improvement project showed that implementation of a screening tool designed to identify patients eligible for DOAC therapy did not decrease the average number of Anticoagulation Clinic encounters. Thirty of 82 eligible patients (36.6%) decided to switch to DOAC therapy during the study period. For those 30 patients, there was a statistically significant decrease in the number of individual clinic encounters. This suggests that the screening tool may positively impact Anticoagulation Clinic metrics when evaluating individual patients, potentially increasing clinic appointment availability.

Confounding Factors

Multiple confounding factors may have affected this project’s results. First, Class I recall for point-of-care test strips used by the clinic was mandated by the US Food and Drug Administration on November 1, 2018.⁵ Before the recall, investigators found that many nontherapeutic INRs using point-of-care testing later showed results that were within the therapeutic INR range using same-day venous blood collection. This may have led to increases in falsely recorded nontherapeutic INRs and lowered TTR report results. Initially, the project was designed to collect monthly clinic encounter data for 3 months following the 8-week screening period; however, data collection was stopped after 2 months because of the test strip recall.

In addition, in early December 2018, all patients were moved from the Anticoagulation Clinic to the Anticoagulation Telephone Clinic that uses venous blood draws and telephone appointments. Data from venous blood draw results had previously been excluded from this project because results were not available on the same day. Patients in this program are contacted by telephone rather than being offered a face-to-face appointment, thus reducing in-clinic encounters.

Another confounding factor was a FVAHCC policy change in August 2018 requiring that any patient initiated on a DOAC make a onetime visit to the Anticoagulation Clinic prior to establishing care with a PACT CPS. Investigators were unable to exclude these patients from monthly encounter data. Some patients transitioning from warfarin to DOAC therapy were required to continue receiving anticoagulation monitoring from the clinic because of limited PACT CPS clinic availability, thus further increasing postscreening encounters.

Health care providers outside of the Anticoagulation Clinic and uninvolved with the quality improvement project also were switching patients from warfarin to DOAC therapies. Although this may have affected encounter data positively, investigators cannot guarantee these patients would have met criteria outlined by the screening tool.

In September 2018 Hurricane Florence disrupted health care delivery during the 8-week screening period. This event disrupted numerous clinic appointments. Although screening of patients was completed during the 8-week screening period, some patients did not switch to DOAC therapies until November 2018.

Secondary Endpoint Results

Promising results can be seen by specifically looking at the secondary endpoint: the number of encounters associated with patients who chose DOAC therapy. There were 107 encounters associated with the 30 patients who switched to a DOAC prior to screening and only 32 associated encounters after screening, a reduction of 70.1%. This suggests that multiple appointment slots were freed when the screening tool led to successful conversion from warfarin to a DOAC. Further assessment is warranted.

 

Future Project Development

Future areas for quality improvement project development include expanding project criteria to include patients taking warfarin for VTE. Eighty-nine of 137 patients (65%) who were deemed ineligible to switch to DOAC therapy were excluded due to a diagnosis of VTE. There are existing VA/Department of Defense Criteria for Use for DOAC use in VTE recommendations. Straightforward modification of the screening tool could include this patient group and may be especially useful for patients on indefinite warfarin therapy for recurrent VTE who have poor TTR report results.⁶

Given the number of confounding factors caused by unforeseen changes to the Anticoagulation Clinic workflow, use of the DOAC screening tool was placed on hold at the conclusion of data collection. This limited the ability to analyze encounter data in the months following project conclusion. Future plans include reimplementation of the screening tool with minor adjustments to include patients on warfarin for VTE and patients with a TTR report results above 60%.

Conclusion

This quality improvement project sought to determine the impact of a screening tool on effecting Anticoagulation Clinic encounter metrics. Results of this project show that the screening tool was unsuccessful in reducing the number of overall clinic encounters. Some promise was shown when evaluating clinic encounters for patients who switched anticoagulation therapies. Numerous confounding factors may have contributed to these results.

Metrics from 2017 at the Fayetteville Veterans Affairs Heath Care Center (FVAHCC) Anticoagulation Clinic indicate that 43% of patients with atrial fibrillation (AF) who are prescribed warfarin have difficulty maintaining a therapeutic international normalized ratio (INR). These patients require frequent clinic appointments to adjust their regimens to ensure anticoagulation efficacy. FVAHCC policy requires a patient to return to the clinic for repeat INR evaluation within 5 to 14 days of the visit where INR was outside of the established therapeutic range.¹ These frequent INR monitoring appointments increase patient and health care provider burden.

Direct oral anticoagulants (DOACs) are an alternative to warfarin for patients with AF who require anticoagulation. DOACs, which do not require regular efficacy monitoring, can be beneficial to patients who struggle to maintain a therapeutic INR when taking warfarin. FVAHCC policy regarding warfarin therapy monitoring allows for a maximum of 6 weeks between appointments. This period is often extended to 3 to 6 months for patients on DOACs.¹

At FVAHCC, patients prescribed warfarin are managed in a centralized Anticoagulation Clinic led by a clinical pharmacy specialist (CPS). When a patient reports for an appointment, a clinical pharmacy technician performs point-of-care INR testing and asks standardized questions regarding therapy, including an assessment of adherence. The CPS then evaluates the patient’s INR test results, adjusts the dosage of warfarin as indicated, and determines appropriate follow-up.

A patient who is prescribed a DOAC is monitored by a CPS who works within a patient aligned care team (PACT). The PACT, a multidisciplinary team providing health care to veterans, includes physicians, nurses, pharmacists, dieticians, and mental health providers. Each CPS covers 3 or 4 PACTs. These pharmacists monitor all aspects of DOAC therapy at regular intervals, including renal and hepatic function, complete blood counts, medication adherence, and adverse effects.

Clinic and patient INR data are tracked using a time in therapeutic range (TTR) report generated by the US Department of Veterans Affairs (VA). The TTR report provides clinical information to enhance patient anticoagulation care.² The TTR report identifies patients with an active order for warfarin and a diagnosis of AF or venous thromboembolism (VTE) whose INR is within therapeutic range (between 2 and 3) < 60% of the time over the previous 160 days.² The patient must have had at least 3 INR levels drawn within that time frame for a TTR report calculation.² The report excludes patients who were first prescribed warfarin within the previous 42 days and those with mechanical heart valves. The TTR report is used by the VA to see concrete facility-level results for quality improvement efforts.²

A quality improvement screening tool was developed to identify patients with AF being treated with warfarin who may appropriately transition to DOAC therapy. Anticoagulation Clinic patients were eligible for further evaluation if they had a TTR report level of < 60% and were prescribed indefinite warfarin therapy for AF.

The national VA goal is to have patient TTR report levels read > 60%. Therefore, the primary objective of this project was to improve Anticoagulation Clinic TTR metrics by targeting patients with TTR levels below the national goal.²

Patients who were successfully converted from warfarin to a DOAC were no longer included in Anticoagulation Clinic metrics and instead were followed by a PACT CPS. Thus, it was hypothesized that the average number of monthly Anticoagulation Clinic encounters would decrease on successful implementation of the screening tool. A secondary endpoint of the study evaluated the change in the total number of encounters of those who converted from warfarin to a DOAC.

Fewer clinic encounters could increase time available for the CPS to incorporate other initiatives into workflow and could increase clinic availability for newly referred veterans.

 

Methods

As this undertaking was considered to be a quality improvement project, institutional review board approval was not required. During an 8-week screening period (August to September 2018), the DOAC screening tool was implemented into the Anticoagulation Clinic workflow. This screening tool (Figure 1) was established based on VA Pharmacy Benefit Management (PBM) Service’s Criteria for Use for Stroke Prevention in Nonvalvular Atrial Fibrillation, a national set of standards used to determine appropriate candidates for DOAC therapy.³

Exclusion criteria included patients with INR goals < 2 or > 3, patients with a diagnosis of VTE, and patients with weight > 120 kg. Patients with a diagnosis of VTE were excluded due to the variability in therapy duration. Weight cutoffs were based on recommendations by the International Society on Thrombosis and Haemostasis. Due to a lack of available data, it was suggested that clinical judgment be used in patients whose weight was > 120 kg.⁴

During the screening period, weekly TTR reports identified patients in the clinic who had TTR < 60%. When a patient with a TTR report results of < 60% also had a scheduled appointment within a week, a CPS then further reviewed patient eligibility using the DOAC screening tool. On arrival for an appointment, the eligible patient was counseled on DOAC medications and the differences between warfarin and DOACs, including monitoring requirements. Patients had the option to switch to DOAC therapy or remain on warfarin.

The change in the average number of monthly Anticoagulation Clinic encounters for 3 months prior to the screening period (May to July 2018) and 2 months following screening (October to November 2018) was evaluated to measure the impact of the DOAC screening tool. The total number of encounters in the clinic was assessed using the monthly VA reports and were averaged for each period. Then data from the 2 periods were compared.

The monthly encounter reports, a data tool that monitors the number of unique visits per veteran each calendar month, also were used to generate a secondary endpoint showing the number of encounters in the Anticoagulation Clinic associated with patients who switched to a DOAC, including visits prior to changing therapy, and before and after the screening period.

Student’s t test was used to compare the change in encounter frequency before and after screening tool implementation for both primary and secondary endpoints. α was defined as .05 a priori. Continuous data were presented as means and standard deviations. Data were calculated with Microsoft Excel 2016.

Results

For the 3 months before the 8-week screening period, an average of 476 Anticoagulation Clinic encounters per month were documented. Two months of data following the screening period averaged 546 encounters per month. There were an average of 70 additional encounters per month after screening tool implementation (P = .15), reflecting the study’s primary objective.

A total of 219 patients in the Anticoagulation Clinic were identified as having a TTR report results of < 60% during the 8-week screening period (Figure 2). Eighty-two of those patients (37.4%) were considered eligible to switch from warfarin to DOAC therapy. Thirty of those eligible patients (13.7%) switched to a DOAC. A total of 107 clinic encounters (22.5%) was associated with these 30 patients prior to screening and 32 associated encounters (5.9%) following screening (P = .01). Of the remaining 137 patients (62.6%) who were ineligible for DOAC therapy, the most common reason for disqualification was a diagnosis of VTE (Table).

 

Discussion

The general results of this quality improvement project showed that implementation of a screening tool designed to identify patients eligible for DOAC therapy did not decrease the average number of Anticoagulation Clinic encounters. Thirty of 82 eligible patients (36.6%) decided to switch to DOAC therapy during the study period. For those 30 patients, there was a statistically significant decrease in the number of individual clinic encounters. This suggests that the screening tool may positively impact Anticoagulation Clinic metrics when evaluating individual patients, potentially increasing clinic appointment availability.

Confounding Factors

Multiple confounding factors may have affected this project’s results. First, Class I recall for point-of-care test strips used by the clinic was mandated by the US Food and Drug Administration on November 1, 2018.⁵ Before the recall, investigators found that many nontherapeutic INRs using point-of-care testing later showed results that were within the therapeutic INR range using same-day venous blood collection. This may have led to increases in falsely recorded nontherapeutic INRs and lowered TTR report results. Initially, the project was designed to collect monthly clinic encounter data for 3 months following the 8-week screening period; however, data collection was stopped after 2 months because of the test strip recall.

In addition, in early December 2018, all patients were moved from the Anticoagulation Clinic to the Anticoagulation Telephone Clinic that uses venous blood draws and telephone appointments. Data from venous blood draw results had previously been excluded from this project because results were not available on the same day. Patients in this program are contacted by telephone rather than being offered a face-to-face appointment, thus reducing in-clinic encounters.

Another confounding factor was a FVAHCC policy change in August 2018 requiring that any patient initiated on a DOAC make a onetime visit to the Anticoagulation Clinic prior to establishing care with a PACT CPS. Investigators were unable to exclude these patients from monthly encounter data. Some patients transitioning from warfarin to DOAC therapy were required to continue receiving anticoagulation monitoring from the clinic because of limited PACT CPS clinic availability, thus further increasing postscreening encounters.

Health care providers outside of the Anticoagulation Clinic and uninvolved with the quality improvement project also were switching patients from warfarin to DOAC therapies. Although this may have affected encounter data positively, investigators cannot guarantee these patients would have met criteria outlined by the screening tool.

In September 2018 Hurricane Florence disrupted health care delivery during the 8-week screening period. This event disrupted numerous clinic appointments. Although screening of patients was completed during the 8-week screening period, some patients did not switch to DOAC therapies until November 2018.

Secondary Endpoint Results

Promising results can be seen by specifically looking at the secondary endpoint: the number of encounters associated with patients who chose DOAC therapy. There were 107 encounters associated with the 30 patients who switched to a DOAC prior to screening and only 32 associated encounters after screening, a reduction of 70.1%. This suggests that multiple appointment slots were freed when the screening tool led to successful conversion from warfarin to a DOAC. Further assessment is warranted.

 

Future Project Development

Future areas for quality improvement project development include expanding project criteria to include patients taking warfarin for VTE. Eighty-nine of 137 patients (65%) who were deemed ineligible to switch to DOAC therapy were excluded due to a diagnosis of VTE. There are existing VA/Department of Defense Criteria for Use for DOAC use in VTE recommendations. Straightforward modification of the screening tool could include this patient group and may be especially useful for patients on indefinite warfarin therapy for recurrent VTE who have poor TTR report results.⁶

Given the number of confounding factors caused by unforeseen changes to the Anticoagulation Clinic workflow, use of the DOAC screening tool was placed on hold at the conclusion of data collection. This limited the ability to analyze encounter data in the months following project conclusion. Future plans include reimplementation of the screening tool with minor adjustments to include patients on warfarin for VTE and patients with a TTR report results above 60%.

Conclusion

This quality improvement project sought to determine the impact of a screening tool on effecting Anticoagulation Clinic encounter metrics. Results of this project show that the screening tool was unsuccessful in reducing the number of overall clinic encounters. Some promise was shown when evaluating clinic encounters for patients who switched anticoagulation therapies. Numerous confounding factors may have contributed to these results.

Metrics from 2017 at the Fayetteville Veterans Affairs Heath Care Center (FVAHCC) Anticoagulation Clinic indicate that 43% of patients with atrial fibrillation (AF) who are prescribed warfarin have difficulty maintaining a therapeutic international normalized ratio (INR). These patients require frequent clinic appointments to adjust their regimens to ensure anticoagulation efficacy. FVAHCC policy requires a patient to return to the clinic for repeat INR evaluation within 5 to 14 days of the visit where INR was outside of the established therapeutic range.¹ These frequent INR monitoring appointments increase patient and health care provider burden.

Direct oral anticoagulants (DOACs) are an alternative to warfarin for patients with AF who require anticoagulation. DOACs, which do not require regular efficacy monitoring, can be beneficial to patients who struggle to maintain a therapeutic INR when taking warfarin. FVAHCC policy regarding warfarin therapy monitoring allows for a maximum of 6 weeks between appointments. This period is often extended to 3 to 6 months for patients on DOACs.¹

At FVAHCC, patients prescribed warfarin are managed in a centralized Anticoagulation Clinic led by a clinical pharmacy specialist (CPS). When a patient reports for an appointment, a clinical pharmacy technician performs point-of-care INR testing and asks standardized questions regarding therapy, including an assessment of adherence. The CPS then evaluates the patient’s INR test results, adjusts the dosage of warfarin as indicated, and determines appropriate follow-up.

A patient who is prescribed a DOAC is monitored by a CPS who works within a patient aligned care team (PACT). The PACT, a multidisciplinary team providing health care to veterans, includes physicians, nurses, pharmacists, dieticians, and mental health providers. Each CPS covers 3 or 4 PACTs. These pharmacists monitor all aspects of DOAC therapy at regular intervals, including renal and hepatic function, complete blood counts, medication adherence, and adverse effects.

Clinic and patient INR data are tracked using a time in therapeutic range (TTR) report generated by the US Department of Veterans Affairs (VA). The TTR report provides clinical information to enhance patient anticoagulation care.² The TTR report identifies patients with an active order for warfarin and a diagnosis of AF or venous thromboembolism (VTE) whose INR is within therapeutic range (between 2 and 3) < 60% of the time over the previous 160 days.² The patient must have had at least 3 INR levels drawn within that time frame for a TTR report calculation.² The report excludes patients who were first prescribed warfarin within the previous 42 days and those with mechanical heart valves. The TTR report is used by the VA to see concrete facility-level results for quality improvement efforts.²

A quality improvement screening tool was developed to identify patients with AF being treated with warfarin who may appropriately transition to DOAC therapy. Anticoagulation Clinic patients were eligible for further evaluation if they had a TTR report level of < 60% and were prescribed indefinite warfarin therapy for AF.

The national VA goal is to have patient TTR report levels read > 60%. Therefore, the primary objective of this project was to improve Anticoagulation Clinic TTR metrics by targeting patients with TTR levels below the national goal.²

Patients who were successfully converted from warfarin to a DOAC were no longer included in Anticoagulation Clinic metrics and instead were followed by a PACT CPS. Thus, it was hypothesized that the average number of monthly Anticoagulation Clinic encounters would decrease on successful implementation of the screening tool. A secondary endpoint of the study evaluated the change in the total number of encounters of those who converted from warfarin to a DOAC.

Fewer clinic encounters could increase time available for the CPS to incorporate other initiatives into workflow and could increase clinic availability for newly referred veterans.

 

Methods

As this undertaking was considered to be a quality improvement project, institutional review board approval was not required. During an 8-week screening period (August to September 2018), the DOAC screening tool was implemented into the Anticoagulation Clinic workflow. This screening tool (Figure 1) was established based on VA Pharmacy Benefit Management (PBM) Service’s Criteria for Use for Stroke Prevention in Nonvalvular Atrial Fibrillation, a national set of standards used to determine appropriate candidates for DOAC therapy.³

Exclusion criteria included patients with INR goals < 2 or > 3, patients with a diagnosis of VTE, and patients with weight > 120 kg. Patients with a diagnosis of VTE were excluded due to the variability in therapy duration. Weight cutoffs were based on recommendations by the International Society on Thrombosis and Haemostasis. Due to a lack of available data, it was suggested that clinical judgment be used in patients whose weight was > 120 kg.⁴

During the screening period, weekly TTR reports identified patients in the clinic who had TTR < 60%. When a patient with a TTR report results of < 60% also had a scheduled appointment within a week, a CPS then further reviewed patient eligibility using the DOAC screening tool. On arrival for an appointment, the eligible patient was counseled on DOAC medications and the differences between warfarin and DOACs, including monitoring requirements. Patients had the option to switch to DOAC therapy or remain on warfarin.

The change in the average number of monthly Anticoagulation Clinic encounters for 3 months prior to the screening period (May to July 2018) and 2 months following screening (October to November 2018) was evaluated to measure the impact of the DOAC screening tool. The total number of encounters in the clinic was assessed using the monthly VA reports and were averaged for each period. Then data from the 2 periods were compared.

The monthly encounter reports, a data tool that monitors the number of unique visits per veteran each calendar month, also were used to generate a secondary endpoint showing the number of encounters in the Anticoagulation Clinic associated with patients who switched to a DOAC, including visits prior to changing therapy, and before and after the screening period.

Student’s t test was used to compare the change in encounter frequency before and after screening tool implementation for both primary and secondary endpoints. α was defined as .05 a priori. Continuous data were presented as means and standard deviations. Data were calculated with Microsoft Excel 2016.

Results

For the 3 months before the 8-week screening period, an average of 476 Anticoagulation Clinic encounters per month were documented. Two months of data following the screening period averaged 546 encounters per month. There were an average of 70 additional encounters per month after screening tool implementation (P = .15), reflecting the study’s primary objective.

A total of 219 patients in the Anticoagulation Clinic were identified as having a TTR report results of < 60% during the 8-week screening period (Figure 2). Eighty-two of those patients (37.4%) were considered eligible to switch from warfarin to DOAC therapy. Thirty of those eligible patients (13.7%) switched to a DOAC. A total of 107 clinic encounters (22.5%) was associated with these 30 patients prior to screening and 32 associated encounters (5.9%) following screening (P = .01). Of the remaining 137 patients (62.6%) who were ineligible for DOAC therapy, the most common reason for disqualification was a diagnosis of VTE (Table).

 

Discussion

The general results of this quality improvement project showed that implementation of a screening tool designed to identify patients eligible for DOAC therapy did not decrease the average number of Anticoagulation Clinic encounters. Thirty of 82 eligible patients (36.6%) decided to switch to DOAC therapy during the study period. For those 30 patients, there was a statistically significant decrease in the number of individual clinic encounters. This suggests that the screening tool may positively impact Anticoagulation Clinic metrics when evaluating individual patients, potentially increasing clinic appointment availability.

Confounding Factors

Multiple confounding factors may have affected this project’s results. First, Class I recall for point-of-care test strips used by the clinic was mandated by the US Food and Drug Administration on November 1, 2018.⁵ Before the recall, investigators found that many nontherapeutic INRs using point-of-care testing later showed results that were within the therapeutic INR range using same-day venous blood collection. This may have led to increases in falsely recorded nontherapeutic INRs and lowered TTR report results. Initially, the project was designed to collect monthly clinic encounter data for 3 months following the 8-week screening period; however, data collection was stopped after 2 months because of the test strip recall.

In addition, in early December 2018, all patients were moved from the Anticoagulation Clinic to the Anticoagulation Telephone Clinic that uses venous blood draws and telephone appointments. Data from venous blood draw results had previously been excluded from this project because results were not available on the same day. Patients in this program are contacted by telephone rather than being offered a face-to-face appointment, thus reducing in-clinic encounters.

Another confounding factor was a FVAHCC policy change in August 2018 requiring that any patient initiated on a DOAC make a onetime visit to the Anticoagulation Clinic prior to establishing care with a PACT CPS. Investigators were unable to exclude these patients from monthly encounter data. Some patients transitioning from warfarin to DOAC therapy were required to continue receiving anticoagulation monitoring from the clinic because of limited PACT CPS clinic availability, thus further increasing postscreening encounters.

Health care providers outside of the Anticoagulation Clinic and uninvolved with the quality improvement project also were switching patients from warfarin to DOAC therapies. Although this may have affected encounter data positively, investigators cannot guarantee these patients would have met criteria outlined by the screening tool.

In September 2018 Hurricane Florence disrupted health care delivery during the 8-week screening period. This event disrupted numerous clinic appointments. Although screening of patients was completed during the 8-week screening period, some patients did not switch to DOAC therapies until November 2018.

Secondary Endpoint Results

Promising results can be seen by specifically looking at the secondary endpoint: the number of encounters associated with patients who chose DOAC therapy. There were 107 encounters associated with the 30 patients who switched to a DOAC prior to screening and only 32 associated encounters after screening, a reduction of 70.1%. This suggests that multiple appointment slots were freed when the screening tool led to successful conversion from warfarin to a DOAC. Further assessment is warranted.

 

Future Project Development

Future areas for quality improvement project development include expanding project criteria to include patients taking warfarin for VTE. Eighty-nine of 137 patients (65%) who were deemed ineligible to switch to DOAC therapy were excluded due to a diagnosis of VTE. There are existing VA/Department of Defense Criteria for Use for DOAC use in VTE recommendations. Straightforward modification of the screening tool could include this patient group and may be especially useful for patients on indefinite warfarin therapy for recurrent VTE who have poor TTR report results.⁶

Given the number of confounding factors caused by unforeseen changes to the Anticoagulation Clinic workflow, use of the DOAC screening tool was placed on hold at the conclusion of data collection. This limited the ability to analyze encounter data in the months following project conclusion. Future plans include reimplementation of the screening tool with minor adjustments to include patients on warfarin for VTE and patients with a TTR report results above 60%.

Conclusion

This quality improvement project sought to determine the impact of a screening tool on effecting Anticoagulation Clinic encounter metrics. Results of this project show that the screening tool was unsuccessful in reducing the number of overall clinic encounters. Some promise was shown when evaluating clinic encounters for patients who switched anticoagulation therapies. Numerous confounding factors may have contributed to these results.

In April 2014, amid concerns for long wait times for care within the US Department of Veterans Affairs (VA) Veterans Health Administration (VHA), the Veterans Access, Choice, and Accountability Act was signed into law. This included the Veterans Choice Program (VCP), which included a provision for veterans to be referred outside of the VA to the community for care if their nearest VHA facility could not provide the requested care within 30 days of the clinically indicated date.¹ Since implementation of the VCP, both media outlets and policy researchers have raised concerns about both the timeliness and quality of care provided through this program.^2-4

Specifically for colonoscopy, referral outside of the VA in the pre-VCP era resulted in lower adenoma detection rate (ADR) and decreased adherence to surveillance guidelines when compared with matched VA control colonoscopies, raising concerns about quality assurance.⁵ Colorectal cancer (CRC) screening and timely colonoscopy is a VA priority; however, the performance of the VCP for colonoscopy timelines and quality has not been examined in detail.

Methods

We identified 3,855 veterans at the VA Pittsburgh Healthcare System (VAPHS) who were referred for colonoscopy in the community by using VCP from June 2015 through May 2017, using a query for colonoscopy procedure orders within the VA Corporate Data Warehouse. A total of 190 patients had a colonoscopy completed in the community by utilizing the VCP during this time frame.

At VAPHS, veterans who are referred for colonoscopy are contacted by a scheduler. The scheduler contacts the patient and offers the first available colonoscopy date at VAPHS and schedules the procedure for this date. However, if this date is > 30 days from the procedure order date, the scheduler gives the veteran the option of being contacted by VCP to schedule a colonoscopy within the community (Figure 1). We measured the time interval from the date of the initially scheduled first available colonoscopy at VAPHS to the date the colonoscopy was actually performed through VCP.

Quality assurance also was assessed by checking for the availability of records of colonoscopies performed through the VCP in the VA electronic health record (EHR) system. Colonoscopy procedure reports also were reviewed to assess for documentation of established colonoscopy quality metrics for examinations performed through the VCP. Additionally, we reviewed records scanned into the VA EHR pertaining to the VCP colonoscopy, including pathology information and pre- or postvisit records if available.

Data extraction was initiated in November 2017 to allow for at least 6 months of lead time for outside health records from the community to be received and scanned into the EHR for the veteran at VAPHS. For colonoscopy quality metrics, we chose 3 metrics that are universally documented for all colonoscopy procedures performed at VAPHS: quality of bowel preparation, cecal withdrawal time, and performance of retroflexion in the rectum. Documentation of these quality metrics is recommended in gastroenterology practice guidelinesand/or required by VA national policy.^6,7

We separately reviewed a sample of 350 of the 3,855 patients referred for colonoscopy through VCP at VAPHS during the same time period to investigate overall VCP utilization. This sample was representative at a 95% CI with 5% margin of error of the total and sampled from 2 high-volume referral months (October and November 2015) and 3 low-volume months (January, February, and March 2017). Detailed data were collected regarding the colonoscopy scheduling and VCP referral process, including dates of colonoscopy procedure request; scheduling within the VAPHS; scheduling through the VCP; and ultimately if, when, and where (VAPHS vs community) a veteran had a colonoscopy performed. Wait times for colonoscopy procedures performed at the VAPHS and those performed through the VCP were compared.

The institutional review board at VAPHS reviewed and approved this quality improvement study.

 

Statistical Analysis

For the 190 veterans who had a colonoscopy performed through VCP, a 1-sample Wilcoxon signed rank test was used with a null hypothesis that the median difference in days between first available VAPHS colonoscopy and community colonoscopy dates was 0. For the utilization sample of 350 veterans, an independent samples median test was used to compare the median wait times for colonoscopy procedures performed at the VA and those performed through VCP. IBM SPSS Version 25 was used for all statistical analysis.

Results

Of the 190 identified colonoscopies completed in the community utilizing VCP, scanned records could not be found for 29 procedures (15.3%) (Table). VCP procedures were performed a median 2 days earlier than the first available VAPHS procedure, but this difference was not statistically significant (P = .62) (Figure 2). Although 52% of colonoscopies occurred sooner through VCP than the initially scheduled VAPHS date, 44% were performed later, and there was wide variability in the difference between these dates, ranging from 49 days sooner to 165 days later.

Pathology results from VCP procedures for which tissue samples were obtained were absent in 11.9% (14 of 118) of procedures. There were no clear follow-up recommendations to referring VA health care providers in the 18% (29 of 161) of available procedure reports. In VCP procedures, documentation of selected quality metrics: bowel preparation, cecal withdrawal time, and rectal retroflexion, were deficient in 27.3%, 70.2%, and 32.9%, respectively (Figure 3).

The utilization dataset sample included 350 veterans who were offered a VCP colonoscopy because the first available VAPHS procedure could not be scheduled for > 30 days. Of these patients, 231 (66%) ultimately had their colonoscopy performed at VAPHS. An additional 26.6% of the patients in the utilization sample were lost in the scheduling process (ie, could not be contacted, cancelled and could not be rescheduled, or were a “no show” their scheduled VAPHS procedure). An unknown number of these patients may have had a procedure outside of the VA, but there are no records to confirm or exclude this possibility. Ultimately, there were only 26 (7.4%) confirmed VCP colonoscopy procedures within the utilization sample (Figure 4). The median actual wait time for colonoscopy was 61 days for VA procedures and 66 days for procedures referred through the VCP, which was not statistically significant (P = .15).

Discussion

This is the first study to evaluate the performance of the VCP for colonoscopy referrals. Consistent with recently reported data in other specialties, colonoscopy referrals through VCP did not lead to more timely procedures overall, although there was wide variation.⁸ The use of VCP for veteran referral to the community for colonoscopy led to fragmentation of care—with 15% of records for VCP colonoscopies unavailable in the VA EHR 6 months after the procedure. In addition, there were 45 pre- or postprocedure visits in the community, which is not standard practice at VAPHS, and therefore may add to the cost of care for veterans.

Documentation of selected colonoscopy quality metrics were deficient in 27.3% to 70.2% of available VCP procedure reports. Although many veterans were eligible for VCP referral for colonoscopy, only 7.4% had a documented procedure through VCP, and two-thirds of veterans eligible for VCP participation had their colonoscopy performed at the VAPHS, reflecting overall low utilization of the program.

The national average wait time for VCP referrals for multiple specialties was estimated to be 51 days in a 2018 Government Accountability Office (GAO) report, which is similar to our findings.⁹ The GAO report also concluded that the VCP does not have timeliness standards and notes missed opportunities to develop a mechanism for record transfer between the community and the VA. Our finding of missing colonoscopy procedure and pathology reports within the VA EHR is consistent with this claim. Our analysis revealed that widely accepted quality standards for colonoscopy, those that are required at the VA and monitored for quality assurance at the VAPHS, are not being consistently reported for veterans who undergo procedures in the community. Last, the overall low utilization rate, combined with overall similar wait times for colonoscopies referred through the VCP vs those done at the VA, should lead to reconsideration of offering community care referral to all veterans based solely on static wait time cutoffs.

 

Limitations

There are several limitations to our analysis. First, all data were extracted via chart review by one author; therefore, some scanned procedure or pathology reports or pre- and postprocedure records may have been missed. Second, these data are representative of a single VA medical center and may not reflect trends nationwide. Third, there are many factors that can influence veteran decision making regarding when and where colonoscopy procedures are performed, which could be related to the VCP community care referral process or independent of this. Finally, colonoscopies performed through the VCP are grouped and may not reflect variability in the performance of community practices that veterans were referred to though the VCP.

Adenoma detection rates (ADR) were not included in the assessment for 2 reasons. First, there was an insufficient number of screening colonoscopies to use for the ADR calculation. Second, a composite non-VA ADR of multiple community endoscopists in different practices would likely be inaccurate and not clinically meaningful. Of note, the VAPHS does calculate and maintain ADR information as a practice for its endoscopists.

Conclusions

Our findings are particularly important as the VA expands access to care in the community through the VA Mission Act, which replaces the VCP but continues to include a static wait time threshold of 28 days for referral to community-based care.¹⁰ Especially for colonoscopies with the indication of screening or surveillance, wait times > 28 days are likely not clinically significant. Additionally, this study demonstrates that there also are delays in access to colonoscopy by community-based care providers, and potentially reflects widespread colonoscopy access issues that are not unique to the VA.

Our findings are similar to other published results and reports and raise similar concerns about the pitfalls of veteran referral into the community, including (1) similar wait times for the community and the VA; (2) the risk of fragmented care; (3) unevenquality of care; and (4) low overall utilization of VCP for colonoscopy.¹¹ We agree with the GAO’s recommendations, which include establishing clinically meaningful wait time thresholds, systemic monitoring of the timeliness of care, and additional mechanisms for seamless transfer of complete records of care into the VA system. If a referral is placed for community-based care, this should come with an expectation that the care will be offered and can be delivered sooner than would be possible at the VA. We additionally recommend that standards for reporting quality metrics, including ADR, also be required of community colonoscopy providers contracted to provide care for veterans through the VA Mission Act. Importantly, we recommend that data for comparative wait times and quality metrics for VA and the community should be publicly available for veterans so that they may make more informed choices about where they receive health care.

Acknowledgments

The authors thank Kaneen Allen, PhD, for her administrative assistance and guidance.

In April 2014, amid concerns for long wait times for care within the US Department of Veterans Affairs (VA) Veterans Health Administration (VHA), the Veterans Access, Choice, and Accountability Act was signed into law. This included the Veterans Choice Program (VCP), which included a provision for veterans to be referred outside of the VA to the community for care if their nearest VHA facility could not provide the requested care within 30 days of the clinically indicated date.¹ Since implementation of the VCP, both media outlets and policy researchers have raised concerns about both the timeliness and quality of care provided through this program.^2-4

Specifically for colonoscopy, referral outside of the VA in the pre-VCP era resulted in lower adenoma detection rate (ADR) and decreased adherence to surveillance guidelines when compared with matched VA control colonoscopies, raising concerns about quality assurance.⁵ Colorectal cancer (CRC) screening and timely colonoscopy is a VA priority; however, the performance of the VCP for colonoscopy timelines and quality has not been examined in detail.

Methods

We identified 3,855 veterans at the VA Pittsburgh Healthcare System (VAPHS) who were referred for colonoscopy in the community by using VCP from June 2015 through May 2017, using a query for colonoscopy procedure orders within the VA Corporate Data Warehouse. A total of 190 patients had a colonoscopy completed in the community by utilizing the VCP during this time frame.

At VAPHS, veterans who are referred for colonoscopy are contacted by a scheduler. The scheduler contacts the patient and offers the first available colonoscopy date at VAPHS and schedules the procedure for this date. However, if this date is > 30 days from the procedure order date, the scheduler gives the veteran the option of being contacted by VCP to schedule a colonoscopy within the community (Figure 1). We measured the time interval from the date of the initially scheduled first available colonoscopy at VAPHS to the date the colonoscopy was actually performed through VCP.

Quality assurance also was assessed by checking for the availability of records of colonoscopies performed through the VCP in the VA electronic health record (EHR) system. Colonoscopy procedure reports also were reviewed to assess for documentation of established colonoscopy quality metrics for examinations performed through the VCP. Additionally, we reviewed records scanned into the VA EHR pertaining to the VCP colonoscopy, including pathology information and pre- or postvisit records if available.

Data extraction was initiated in November 2017 to allow for at least 6 months of lead time for outside health records from the community to be received and scanned into the EHR for the veteran at VAPHS. For colonoscopy quality metrics, we chose 3 metrics that are universally documented for all colonoscopy procedures performed at VAPHS: quality of bowel preparation, cecal withdrawal time, and performance of retroflexion in the rectum. Documentation of these quality metrics is recommended in gastroenterology practice guidelinesand/or required by VA national policy.^6,7

We separately reviewed a sample of 350 of the 3,855 patients referred for colonoscopy through VCP at VAPHS during the same time period to investigate overall VCP utilization. This sample was representative at a 95% CI with 5% margin of error of the total and sampled from 2 high-volume referral months (October and November 2015) and 3 low-volume months (January, February, and March 2017). Detailed data were collected regarding the colonoscopy scheduling and VCP referral process, including dates of colonoscopy procedure request; scheduling within the VAPHS; scheduling through the VCP; and ultimately if, when, and where (VAPHS vs community) a veteran had a colonoscopy performed. Wait times for colonoscopy procedures performed at the VAPHS and those performed through the VCP were compared.

The institutional review board at VAPHS reviewed and approved this quality improvement study.

 

Statistical Analysis

For the 190 veterans who had a colonoscopy performed through VCP, a 1-sample Wilcoxon signed rank test was used with a null hypothesis that the median difference in days between first available VAPHS colonoscopy and community colonoscopy dates was 0. For the utilization sample of 350 veterans, an independent samples median test was used to compare the median wait times for colonoscopy procedures performed at the VA and those performed through VCP. IBM SPSS Version 25 was used for all statistical analysis.

Results

Of the 190 identified colonoscopies completed in the community utilizing VCP, scanned records could not be found for 29 procedures (15.3%) (Table). VCP procedures were performed a median 2 days earlier than the first available VAPHS procedure, but this difference was not statistically significant (P = .62) (Figure 2). Although 52% of colonoscopies occurred sooner through VCP than the initially scheduled VAPHS date, 44% were performed later, and there was wide variability in the difference between these dates, ranging from 49 days sooner to 165 days later.

Pathology results from VCP procedures for which tissue samples were obtained were absent in 11.9% (14 of 118) of procedures. There were no clear follow-up recommendations to referring VA health care providers in the 18% (29 of 161) of available procedure reports. In VCP procedures, documentation of selected quality metrics: bowel preparation, cecal withdrawal time, and rectal retroflexion, were deficient in 27.3%, 70.2%, and 32.9%, respectively (Figure 3).

The utilization dataset sample included 350 veterans who were offered a VCP colonoscopy because the first available VAPHS procedure could not be scheduled for > 30 days. Of these patients, 231 (66%) ultimately had their colonoscopy performed at VAPHS. An additional 26.6% of the patients in the utilization sample were lost in the scheduling process (ie, could not be contacted, cancelled and could not be rescheduled, or were a “no show” their scheduled VAPHS procedure). An unknown number of these patients may have had a procedure outside of the VA, but there are no records to confirm or exclude this possibility. Ultimately, there were only 26 (7.4%) confirmed VCP colonoscopy procedures within the utilization sample (Figure 4). The median actual wait time for colonoscopy was 61 days for VA procedures and 66 days for procedures referred through the VCP, which was not statistically significant (P = .15).

Discussion

This is the first study to evaluate the performance of the VCP for colonoscopy referrals. Consistent with recently reported data in other specialties, colonoscopy referrals through VCP did not lead to more timely procedures overall, although there was wide variation.⁸ The use of VCP for veteran referral to the community for colonoscopy led to fragmentation of care—with 15% of records for VCP colonoscopies unavailable in the VA EHR 6 months after the procedure. In addition, there were 45 pre- or postprocedure visits in the community, which is not standard practice at VAPHS, and therefore may add to the cost of care for veterans.

Documentation of selected colonoscopy quality metrics were deficient in 27.3% to 70.2% of available VCP procedure reports. Although many veterans were eligible for VCP referral for colonoscopy, only 7.4% had a documented procedure through VCP, and two-thirds of veterans eligible for VCP participation had their colonoscopy performed at the VAPHS, reflecting overall low utilization of the program.

The national average wait time for VCP referrals for multiple specialties was estimated to be 51 days in a 2018 Government Accountability Office (GAO) report, which is similar to our findings.⁹ The GAO report also concluded that the VCP does not have timeliness standards and notes missed opportunities to develop a mechanism for record transfer between the community and the VA. Our finding of missing colonoscopy procedure and pathology reports within the VA EHR is consistent with this claim. Our analysis revealed that widely accepted quality standards for colonoscopy, those that are required at the VA and monitored for quality assurance at the VAPHS, are not being consistently reported for veterans who undergo procedures in the community. Last, the overall low utilization rate, combined with overall similar wait times for colonoscopies referred through the VCP vs those done at the VA, should lead to reconsideration of offering community care referral to all veterans based solely on static wait time cutoffs.

 

Limitations

There are several limitations to our analysis. First, all data were extracted via chart review by one author; therefore, some scanned procedure or pathology reports or pre- and postprocedure records may have been missed. Second, these data are representative of a single VA medical center and may not reflect trends nationwide. Third, there are many factors that can influence veteran decision making regarding when and where colonoscopy procedures are performed, which could be related to the VCP community care referral process or independent of this. Finally, colonoscopies performed through the VCP are grouped and may not reflect variability in the performance of community practices that veterans were referred to though the VCP.

Adenoma detection rates (ADR) were not included in the assessment for 2 reasons. First, there was an insufficient number of screening colonoscopies to use for the ADR calculation. Second, a composite non-VA ADR of multiple community endoscopists in different practices would likely be inaccurate and not clinically meaningful. Of note, the VAPHS does calculate and maintain ADR information as a practice for its endoscopists.

Conclusions

Our findings are particularly important as the VA expands access to care in the community through the VA Mission Act, which replaces the VCP but continues to include a static wait time threshold of 28 days for referral to community-based care.¹⁰ Especially for colonoscopies with the indication of screening or surveillance, wait times > 28 days are likely not clinically significant. Additionally, this study demonstrates that there also are delays in access to colonoscopy by community-based care providers, and potentially reflects widespread colonoscopy access issues that are not unique to the VA.

Our findings are similar to other published results and reports and raise similar concerns about the pitfalls of veteran referral into the community, including (1) similar wait times for the community and the VA; (2) the risk of fragmented care; (3) unevenquality of care; and (4) low overall utilization of VCP for colonoscopy.¹¹ We agree with the GAO’s recommendations, which include establishing clinically meaningful wait time thresholds, systemic monitoring of the timeliness of care, and additional mechanisms for seamless transfer of complete records of care into the VA system. If a referral is placed for community-based care, this should come with an expectation that the care will be offered and can be delivered sooner than would be possible at the VA. We additionally recommend that standards for reporting quality metrics, including ADR, also be required of community colonoscopy providers contracted to provide care for veterans through the VA Mission Act. Importantly, we recommend that data for comparative wait times and quality metrics for VA and the community should be publicly available for veterans so that they may make more informed choices about where they receive health care.

Acknowledgments

The authors thank Kaneen Allen, PhD, for her administrative assistance and guidance.

In April 2014, amid concerns for long wait times for care within the US Department of Veterans Affairs (VA) Veterans Health Administration (VHA), the Veterans Access, Choice, and Accountability Act was signed into law. This included the Veterans Choice Program (VCP), which included a provision for veterans to be referred outside of the VA to the community for care if their nearest VHA facility could not provide the requested care within 30 days of the clinically indicated date.¹ Since implementation of the VCP, both media outlets and policy researchers have raised concerns about both the timeliness and quality of care provided through this program.^2-4

Specifically for colonoscopy, referral outside of the VA in the pre-VCP era resulted in lower adenoma detection rate (ADR) and decreased adherence to surveillance guidelines when compared with matched VA control colonoscopies, raising concerns about quality assurance.⁵ Colorectal cancer (CRC) screening and timely colonoscopy is a VA priority; however, the performance of the VCP for colonoscopy timelines and quality has not been examined in detail.

Methods

We identified 3,855 veterans at the VA Pittsburgh Healthcare System (VAPHS) who were referred for colonoscopy in the community by using VCP from June 2015 through May 2017, using a query for colonoscopy procedure orders within the VA Corporate Data Warehouse. A total of 190 patients had a colonoscopy completed in the community by utilizing the VCP during this time frame.

At VAPHS, veterans who are referred for colonoscopy are contacted by a scheduler. The scheduler contacts the patient and offers the first available colonoscopy date at VAPHS and schedules the procedure for this date. However, if this date is > 30 days from the procedure order date, the scheduler gives the veteran the option of being contacted by VCP to schedule a colonoscopy within the community (Figure 1). We measured the time interval from the date of the initially scheduled first available colonoscopy at VAPHS to the date the colonoscopy was actually performed through VCP.

Quality assurance also was assessed by checking for the availability of records of colonoscopies performed through the VCP in the VA electronic health record (EHR) system. Colonoscopy procedure reports also were reviewed to assess for documentation of established colonoscopy quality metrics for examinations performed through the VCP. Additionally, we reviewed records scanned into the VA EHR pertaining to the VCP colonoscopy, including pathology information and pre- or postvisit records if available.

Data extraction was initiated in November 2017 to allow for at least 6 months of lead time for outside health records from the community to be received and scanned into the EHR for the veteran at VAPHS. For colonoscopy quality metrics, we chose 3 metrics that are universally documented for all colonoscopy procedures performed at VAPHS: quality of bowel preparation, cecal withdrawal time, and performance of retroflexion in the rectum. Documentation of these quality metrics is recommended in gastroenterology practice guidelinesand/or required by VA national policy.^6,7

We separately reviewed a sample of 350 of the 3,855 patients referred for colonoscopy through VCP at VAPHS during the same time period to investigate overall VCP utilization. This sample was representative at a 95% CI with 5% margin of error of the total and sampled from 2 high-volume referral months (October and November 2015) and 3 low-volume months (January, February, and March 2017). Detailed data were collected regarding the colonoscopy scheduling and VCP referral process, including dates of colonoscopy procedure request; scheduling within the VAPHS; scheduling through the VCP; and ultimately if, when, and where (VAPHS vs community) a veteran had a colonoscopy performed. Wait times for colonoscopy procedures performed at the VAPHS and those performed through the VCP were compared.

The institutional review board at VAPHS reviewed and approved this quality improvement study.

 

Statistical Analysis

For the 190 veterans who had a colonoscopy performed through VCP, a 1-sample Wilcoxon signed rank test was used with a null hypothesis that the median difference in days between first available VAPHS colonoscopy and community colonoscopy dates was 0. For the utilization sample of 350 veterans, an independent samples median test was used to compare the median wait times for colonoscopy procedures performed at the VA and those performed through VCP. IBM SPSS Version 25 was used for all statistical analysis.

Results

Of the 190 identified colonoscopies completed in the community utilizing VCP, scanned records could not be found for 29 procedures (15.3%) (Table). VCP procedures were performed a median 2 days earlier than the first available VAPHS procedure, but this difference was not statistically significant (P = .62) (Figure 2). Although 52% of colonoscopies occurred sooner through VCP than the initially scheduled VAPHS date, 44% were performed later, and there was wide variability in the difference between these dates, ranging from 49 days sooner to 165 days later.

Pathology results from VCP procedures for which tissue samples were obtained were absent in 11.9% (14 of 118) of procedures. There were no clear follow-up recommendations to referring VA health care providers in the 18% (29 of 161) of available procedure reports. In VCP procedures, documentation of selected quality metrics: bowel preparation, cecal withdrawal time, and rectal retroflexion, were deficient in 27.3%, 70.2%, and 32.9%, respectively (Figure 3).

The utilization dataset sample included 350 veterans who were offered a VCP colonoscopy because the first available VAPHS procedure could not be scheduled for > 30 days. Of these patients, 231 (66%) ultimately had their colonoscopy performed at VAPHS. An additional 26.6% of the patients in the utilization sample were lost in the scheduling process (ie, could not be contacted, cancelled and could not be rescheduled, or were a “no show” their scheduled VAPHS procedure). An unknown number of these patients may have had a procedure outside of the VA, but there are no records to confirm or exclude this possibility. Ultimately, there were only 26 (7.4%) confirmed VCP colonoscopy procedures within the utilization sample (Figure 4). The median actual wait time for colonoscopy was 61 days for VA procedures and 66 days for procedures referred through the VCP, which was not statistically significant (P = .15).

Discussion

This is the first study to evaluate the performance of the VCP for colonoscopy referrals. Consistent with recently reported data in other specialties, colonoscopy referrals through VCP did not lead to more timely procedures overall, although there was wide variation.⁸ The use of VCP for veteran referral to the community for colonoscopy led to fragmentation of care—with 15% of records for VCP colonoscopies unavailable in the VA EHR 6 months after the procedure. In addition, there were 45 pre- or postprocedure visits in the community, which is not standard practice at VAPHS, and therefore may add to the cost of care for veterans.

Documentation of selected colonoscopy quality metrics were deficient in 27.3% to 70.2% of available VCP procedure reports. Although many veterans were eligible for VCP referral for colonoscopy, only 7.4% had a documented procedure through VCP, and two-thirds of veterans eligible for VCP participation had their colonoscopy performed at the VAPHS, reflecting overall low utilization of the program.

The national average wait time for VCP referrals for multiple specialties was estimated to be 51 days in a 2018 Government Accountability Office (GAO) report, which is similar to our findings.⁹ The GAO report also concluded that the VCP does not have timeliness standards and notes missed opportunities to develop a mechanism for record transfer between the community and the VA. Our finding of missing colonoscopy procedure and pathology reports within the VA EHR is consistent with this claim. Our analysis revealed that widely accepted quality standards for colonoscopy, those that are required at the VA and monitored for quality assurance at the VAPHS, are not being consistently reported for veterans who undergo procedures in the community. Last, the overall low utilization rate, combined with overall similar wait times for colonoscopies referred through the VCP vs those done at the VA, should lead to reconsideration of offering community care referral to all veterans based solely on static wait time cutoffs.

 

Limitations

There are several limitations to our analysis. First, all data were extracted via chart review by one author; therefore, some scanned procedure or pathology reports or pre- and postprocedure records may have been missed. Second, these data are representative of a single VA medical center and may not reflect trends nationwide. Third, there are many factors that can influence veteran decision making regarding when and where colonoscopy procedures are performed, which could be related to the VCP community care referral process or independent of this. Finally, colonoscopies performed through the VCP are grouped and may not reflect variability in the performance of community practices that veterans were referred to though the VCP.

Adenoma detection rates (ADR) were not included in the assessment for 2 reasons. First, there was an insufficient number of screening colonoscopies to use for the ADR calculation. Second, a composite non-VA ADR of multiple community endoscopists in different practices would likely be inaccurate and not clinically meaningful. Of note, the VAPHS does calculate and maintain ADR information as a practice for its endoscopists.

Conclusions

Our findings are particularly important as the VA expands access to care in the community through the VA Mission Act, which replaces the VCP but continues to include a static wait time threshold of 28 days for referral to community-based care.¹⁰ Especially for colonoscopies with the indication of screening or surveillance, wait times > 28 days are likely not clinically significant. Additionally, this study demonstrates that there also are delays in access to colonoscopy by community-based care providers, and potentially reflects widespread colonoscopy access issues that are not unique to the VA.

Our findings are similar to other published results and reports and raise similar concerns about the pitfalls of veteran referral into the community, including (1) similar wait times for the community and the VA; (2) the risk of fragmented care; (3) unevenquality of care; and (4) low overall utilization of VCP for colonoscopy.¹¹ We agree with the GAO’s recommendations, which include establishing clinically meaningful wait time thresholds, systemic monitoring of the timeliness of care, and additional mechanisms for seamless transfer of complete records of care into the VA system. If a referral is placed for community-based care, this should come with an expectation that the care will be offered and can be delivered sooner than would be possible at the VA. We additionally recommend that standards for reporting quality metrics, including ADR, also be required of community colonoscopy providers contracted to provide care for veterans through the VA Mission Act. Importantly, we recommend that data for comparative wait times and quality metrics for VA and the community should be publicly available for veterans so that they may make more informed choices about where they receive health care.

Acknowledgments

The authors thank Kaneen Allen, PhD, for her administrative assistance and guidance.

Amid the COVID-19 pandemic, eye care professionals should be aware of important guidelines and consider using telehealth to keep both the health care provider and patient as safe as possible. This article is intended to give an update on the ever-changing landscape of eye care due to COVID-19. The Centers for Disease Control and Prevention (CDC) recommends that health care facilities and clinicians delay all elective ambulatory provider visits.¹ In addition, the American Academy of Ophthalmology (AAO) recommends that all ophthalmologists cease providing any treatment other than urgent or emergent care.² Our goal is to equip the eye care provider with the best practice guidelines for seeing urgent and emergent eye conditions.

COVID-19 is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and its symptoms range from mild to severe respiratory illness, fever, cough, fatigue, and shortness of breath.¹ Diarrhea is common early on with infection and loss of taste and smell have also been reported.¹ Follicular conjunctivitis has also been reported, either as an early sign of infection or during hospitalization for severe COVID-19 disease.^2-4 The incubation period of COVID-19 falls within 2 to 14 days according to the CDC.⁵

It has been confirmed that COVID-19 is transmitted through both respiratory droplets and direct contact. Another possible route of viral transmission is entry through aerosolized droplets into the tears, which then pass through the nasolacrimal ducts and into the respiratory tract.⁶

 

Preparations Prior to Office Visit

It is essential for the eye care provider to prioritize patient care in order of absolute necessity, such as sudden vision loss, sudden onset flashes and floaters, and eye trauma. In cases of potentially sight threatening pathology, it is in the best interest of the patient to conduct a face-to-face appointment. Therefore, it is important to implement new guidelines and protocols as we continue to see these patients (Figure 1).

Prior to the patient entering the medical facility, measures should be implemented to minimize exposure risk. This can be done over the telephone or at vehicle entrance screening stations. The triage technician answering the telephone should have a script of questions to ask. The patient should be instructed to come into the office alone unless, for physical or mental reasons, a caregiver is required.

SARS-CoV-2 Screening Questions

Preparedness through risk mitigation strategies are recommended with a targeted questionnaire and noncontact temperature check at the clinic or hospital entrance. Below are some general questions to further triage patients exposed to SARS-CoV-2.

• Do you have fever or any respiratory symptoms?
• Do you have new or worsening cough or shortness of breath?
• Do you have flulike symptoms?
• Have you been in close contact with someone, including health care workers, confirmed to have the COVID-19?

If the patient answers yes to any of the above questions, the CDC urges health care providers to immediately notify both infection control personnel at your health care facility and your local or state health department.^1,2 In regions currently managing significant outbreaks of COVID-19, the AAO recommends that eye care providers assume that any patient could be infected with SARS-CoV-2 and to proceed accordingly.² If urgent eye care is needed, a referral call should be made to a hospital or center equipped to deal with COVID-19 and urgent eye conditions. When calling the referral center, ensure adequate staffing and space and relay all pertinent information along with receiving approval from the treating physician.

Face-to-Face Office Visits

Once it has been determined that it is in the best interest of the patient to be seen in a face-to-face visit, the patient should be instructed to call the office when they arrive in the parking lot. The CDC recommends limiting points of entry upon arrival and during the visit.¹ As soon as an examination lane is ready, the patient can then be messaged to come into the office and escorted into the examination room.

An urgent or emergent ophthalmic examination for a patient with no respiratory symptoms, no fever, and no COVID-19 risk factors should include proper hand hygiene, use of personal protective equipment (PPE), and proper disinfection. Several studies have documented SARS-CoV-2 infection in asymptomatic and presymptomatic patients, making PPE of the up most importance.^2,7,8 PPE should include mask, face shield, and gloves. Currently, there are national and international shortages on PPE and a heightened topic of discussion concerning mask use, effectiveness with extended wear, and reuse. Please refer to the CDC and AAO websites for up-to-date guidelines (Table).^1,2 According to the CDC, N95 respirators are restricted to those performing or present for an aerosol-generating procedure.⁹

It is recommended that the eye care provider should only perform necessary tests and procedures. Noncontact tonometry should be avoided, as this might cause aerosolization of virus particles. The close proximity between eye care providers and their patients during slit-lamp examination may require further precautions to lower the risk of transmission via droplets or through hand to eye contact. The patient should be advised not to speak during the examination portion and the AAO also recommends a surgical mask or cloth face covering for the patient.² An additional protective device that may be used during the slit-lamp exam is a breath shield or a barrier shield (Figures 2 and 3).² Some manufacturers are offering clinicians free slit-lamp breath shields online.

 

Infection Prevention and Control Measures

Last, once the patient leaves the examination room, it should be properly disinfected. A disinfection checklist may be made to ensure uniform systematic cleaning. Alcohol and bleach-based disinfectants commonly used in health care settings are likely very effective against virus particles that cause COVID-19.¹⁰ During the disinfection process, gloves should be worn and careful attention paid to the contact time. Contact time is the amount of time the surface should appear visibly wet for proper disinfection. For example, Metrex CaviWipes have a recommended contact time of 3 minutes; however, this varies depending on type of virus and formulation, check labels or manufacturers’ websites for further directions.¹⁰ Also, the US Environmental Protection Agency has a database search available for disinfectants that meet their criteria for use against SARS-CoV-2.¹¹

In an ever-changing environment, we offer this article to help equip providers to deliver the best possible patient care when face-to-face encounters are necessary. Currently nonurgent eye care follow-up visits are being conducted by telephone or video clinics. It is our goal to inform fellow practitioners on options and strategies to elevate the safety of staff and patients while minimizing the risk of exposure.

Amid the COVID-19 pandemic, eye care professionals should be aware of important guidelines and consider using telehealth to keep both the health care provider and patient as safe as possible. This article is intended to give an update on the ever-changing landscape of eye care due to COVID-19. The Centers for Disease Control and Prevention (CDC) recommends that health care facilities and clinicians delay all elective ambulatory provider visits.¹ In addition, the American Academy of Ophthalmology (AAO) recommends that all ophthalmologists cease providing any treatment other than urgent or emergent care.² Our goal is to equip the eye care provider with the best practice guidelines for seeing urgent and emergent eye conditions.

COVID-19 is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and its symptoms range from mild to severe respiratory illness, fever, cough, fatigue, and shortness of breath.¹ Diarrhea is common early on with infection and loss of taste and smell have also been reported.¹ Follicular conjunctivitis has also been reported, either as an early sign of infection or during hospitalization for severe COVID-19 disease.^2-4 The incubation period of COVID-19 falls within 2 to 14 days according to the CDC.⁵

It has been confirmed that COVID-19 is transmitted through both respiratory droplets and direct contact. Another possible route of viral transmission is entry through aerosolized droplets into the tears, which then pass through the nasolacrimal ducts and into the respiratory tract.⁶

 

Preparations Prior to Office Visit

It is essential for the eye care provider to prioritize patient care in order of absolute necessity, such as sudden vision loss, sudden onset flashes and floaters, and eye trauma. In cases of potentially sight threatening pathology, it is in the best interest of the patient to conduct a face-to-face appointment. Therefore, it is important to implement new guidelines and protocols as we continue to see these patients (Figure 1).

Prior to the patient entering the medical facility, measures should be implemented to minimize exposure risk. This can be done over the telephone or at vehicle entrance screening stations. The triage technician answering the telephone should have a script of questions to ask. The patient should be instructed to come into the office alone unless, for physical or mental reasons, a caregiver is required.

SARS-CoV-2 Screening Questions

Preparedness through risk mitigation strategies are recommended with a targeted questionnaire and noncontact temperature check at the clinic or hospital entrance. Below are some general questions to further triage patients exposed to SARS-CoV-2.

• Do you have fever or any respiratory symptoms?
• Do you have new or worsening cough or shortness of breath?
• Do you have flulike symptoms?
• Have you been in close contact with someone, including health care workers, confirmed to have the COVID-19?

If the patient answers yes to any of the above questions, the CDC urges health care providers to immediately notify both infection control personnel at your health care facility and your local or state health department.^1,2 In regions currently managing significant outbreaks of COVID-19, the AAO recommends that eye care providers assume that any patient could be infected with SARS-CoV-2 and to proceed accordingly.² If urgent eye care is needed, a referral call should be made to a hospital or center equipped to deal with COVID-19 and urgent eye conditions. When calling the referral center, ensure adequate staffing and space and relay all pertinent information along with receiving approval from the treating physician.

Face-to-Face Office Visits

Once it has been determined that it is in the best interest of the patient to be seen in a face-to-face visit, the patient should be instructed to call the office when they arrive in the parking lot. The CDC recommends limiting points of entry upon arrival and during the visit.¹ As soon as an examination lane is ready, the patient can then be messaged to come into the office and escorted into the examination room.

An urgent or emergent ophthalmic examination for a patient with no respiratory symptoms, no fever, and no COVID-19 risk factors should include proper hand hygiene, use of personal protective equipment (PPE), and proper disinfection. Several studies have documented SARS-CoV-2 infection in asymptomatic and presymptomatic patients, making PPE of the up most importance.^2,7,8 PPE should include mask, face shield, and gloves. Currently, there are national and international shortages on PPE and a heightened topic of discussion concerning mask use, effectiveness with extended wear, and reuse. Please refer to the CDC and AAO websites for up-to-date guidelines (Table).^1,2 According to the CDC, N95 respirators are restricted to those performing or present for an aerosol-generating procedure.⁹

It is recommended that the eye care provider should only perform necessary tests and procedures. Noncontact tonometry should be avoided, as this might cause aerosolization of virus particles. The close proximity between eye care providers and their patients during slit-lamp examination may require further precautions to lower the risk of transmission via droplets or through hand to eye contact. The patient should be advised not to speak during the examination portion and the AAO also recommends a surgical mask or cloth face covering for the patient.² An additional protective device that may be used during the slit-lamp exam is a breath shield or a barrier shield (Figures 2 and 3).² Some manufacturers are offering clinicians free slit-lamp breath shields online.

 

Infection Prevention and Control Measures

Last, once the patient leaves the examination room, it should be properly disinfected. A disinfection checklist may be made to ensure uniform systematic cleaning. Alcohol and bleach-based disinfectants commonly used in health care settings are likely very effective against virus particles that cause COVID-19.¹⁰ During the disinfection process, gloves should be worn and careful attention paid to the contact time. Contact time is the amount of time the surface should appear visibly wet for proper disinfection. For example, Metrex CaviWipes have a recommended contact time of 3 minutes; however, this varies depending on type of virus and formulation, check labels or manufacturers’ websites for further directions.¹⁰ Also, the US Environmental Protection Agency has a database search available for disinfectants that meet their criteria for use against SARS-CoV-2.¹¹

In an ever-changing environment, we offer this article to help equip providers to deliver the best possible patient care when face-to-face encounters are necessary. Currently nonurgent eye care follow-up visits are being conducted by telephone or video clinics. It is our goal to inform fellow practitioners on options and strategies to elevate the safety of staff and patients while minimizing the risk of exposure.

Amid the COVID-19 pandemic, eye care professionals should be aware of important guidelines and consider using telehealth to keep both the health care provider and patient as safe as possible. This article is intended to give an update on the ever-changing landscape of eye care due to COVID-19. The Centers for Disease Control and Prevention (CDC) recommends that health care facilities and clinicians delay all elective ambulatory provider visits.¹ In addition, the American Academy of Ophthalmology (AAO) recommends that all ophthalmologists cease providing any treatment other than urgent or emergent care.² Our goal is to equip the eye care provider with the best practice guidelines for seeing urgent and emergent eye conditions.

COVID-19 is caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and its symptoms range from mild to severe respiratory illness, fever, cough, fatigue, and shortness of breath.¹ Diarrhea is common early on with infection and loss of taste and smell have also been reported.¹ Follicular conjunctivitis has also been reported, either as an early sign of infection or during hospitalization for severe COVID-19 disease.^2-4 The incubation period of COVID-19 falls within 2 to 14 days according to the CDC.⁵

It has been confirmed that COVID-19 is transmitted through both respiratory droplets and direct contact. Another possible route of viral transmission is entry through aerosolized droplets into the tears, which then pass through the nasolacrimal ducts and into the respiratory tract.⁶

 

Preparations Prior to Office Visit

It is essential for the eye care provider to prioritize patient care in order of absolute necessity, such as sudden vision loss, sudden onset flashes and floaters, and eye trauma. In cases of potentially sight threatening pathology, it is in the best interest of the patient to conduct a face-to-face appointment. Therefore, it is important to implement new guidelines and protocols as we continue to see these patients (Figure 1).

Prior to the patient entering the medical facility, measures should be implemented to minimize exposure risk. This can be done over the telephone or at vehicle entrance screening stations. The triage technician answering the telephone should have a script of questions to ask. The patient should be instructed to come into the office alone unless, for physical or mental reasons, a caregiver is required.

SARS-CoV-2 Screening Questions

Preparedness through risk mitigation strategies are recommended with a targeted questionnaire and noncontact temperature check at the clinic or hospital entrance. Below are some general questions to further triage patients exposed to SARS-CoV-2.

• Do you have fever or any respiratory symptoms?
• Do you have new or worsening cough or shortness of breath?
• Do you have flulike symptoms?
• Have you been in close contact with someone, including health care workers, confirmed to have the COVID-19?

If the patient answers yes to any of the above questions, the CDC urges health care providers to immediately notify both infection control personnel at your health care facility and your local or state health department.^1,2 In regions currently managing significant outbreaks of COVID-19, the AAO recommends that eye care providers assume that any patient could be infected with SARS-CoV-2 and to proceed accordingly.² If urgent eye care is needed, a referral call should be made to a hospital or center equipped to deal with COVID-19 and urgent eye conditions. When calling the referral center, ensure adequate staffing and space and relay all pertinent information along with receiving approval from the treating physician.

Face-to-Face Office Visits

Once it has been determined that it is in the best interest of the patient to be seen in a face-to-face visit, the patient should be instructed to call the office when they arrive in the parking lot. The CDC recommends limiting points of entry upon arrival and during the visit.¹ As soon as an examination lane is ready, the patient can then be messaged to come into the office and escorted into the examination room.

An urgent or emergent ophthalmic examination for a patient with no respiratory symptoms, no fever, and no COVID-19 risk factors should include proper hand hygiene, use of personal protective equipment (PPE), and proper disinfection. Several studies have documented SARS-CoV-2 infection in asymptomatic and presymptomatic patients, making PPE of the up most importance.^2,7,8 PPE should include mask, face shield, and gloves. Currently, there are national and international shortages on PPE and a heightened topic of discussion concerning mask use, effectiveness with extended wear, and reuse. Please refer to the CDC and AAO websites for up-to-date guidelines (Table).^1,2 According to the CDC, N95 respirators are restricted to those performing or present for an aerosol-generating procedure.⁹

It is recommended that the eye care provider should only perform necessary tests and procedures. Noncontact tonometry should be avoided, as this might cause aerosolization of virus particles. The close proximity between eye care providers and their patients during slit-lamp examination may require further precautions to lower the risk of transmission via droplets or through hand to eye contact. The patient should be advised not to speak during the examination portion and the AAO also recommends a surgical mask or cloth face covering for the patient.² An additional protective device that may be used during the slit-lamp exam is a breath shield or a barrier shield (Figures 2 and 3).² Some manufacturers are offering clinicians free slit-lamp breath shields online.

 

Infection Prevention and Control Measures

Last, once the patient leaves the examination room, it should be properly disinfected. A disinfection checklist may be made to ensure uniform systematic cleaning. Alcohol and bleach-based disinfectants commonly used in health care settings are likely very effective against virus particles that cause COVID-19.¹⁰ During the disinfection process, gloves should be worn and careful attention paid to the contact time. Contact time is the amount of time the surface should appear visibly wet for proper disinfection. For example, Metrex CaviWipes have a recommended contact time of 3 minutes; however, this varies depending on type of virus and formulation, check labels or manufacturers’ websites for further directions.¹⁰ Also, the US Environmental Protection Agency has a database search available for disinfectants that meet their criteria for use against SARS-CoV-2.¹¹

In an ever-changing environment, we offer this article to help equip providers to deliver the best possible patient care when face-to-face encounters are necessary. Currently nonurgent eye care follow-up visits are being conducted by telephone or video clinics. It is our goal to inform fellow practitioners on options and strategies to elevate the safety of staff and patients while minimizing the risk of exposure.

 

To the Editor:

The diagnosis and treatment of nail conditions are necessary competencies for board-certified dermatologists, but appropriate education often is lacking.¹ The American Academy of Dermatology (AAD) annual meeting is one of the largest and most highly attended dermatology educational conferences worldwide. We sought to determine the number of hours dedicated to nail-related topics at the AAD annual meetings from 2013 to 2019.

We accessed programs from the AAD annual meetings archive online (https://www.aad.org/meetings/previous-meetings-archive), and we used hair and psoriasis content for comparison. Event titles and descriptions were searched for nail-related content (using search terms nail, onychia, and onycho), hair-related content (hair, alopecia, trichosis, hirsutism), and psoriasis content (psoriasis). Data acquired for each event included the date, hours, title, and event type (eg, forum, course, focus session, symposium, discussion group, workshop, plenary session).

The number of hours dedicated to nail education consistently lagged behind those related to hair and psoriasis content during the study period (Figure 1). According to the AAD, the conference runs Friday to Tuesday with higher attendance Friday to Sunday (Tim Moses, personal communication, July 9, 2019). Lectures during the weekend are likely to have a broader reach than lectures on Monday and Tuesday. The proportion of nail content during weekend prime time slots was similar to that of hair and psoriasis (Figure 2). Plenary sessions often are presented by renowned experts on hot topics in dermatology. Notably, hair (2014-2015) and psoriasis (2015-2017) content were represented in the plenary sessions during the study period, while nail content was not featured.

Our study shows that nail-related education was underrepresented at the AAD annual meetings from 2013 to 2019 compared to hair- and psoriasis-related content. Educational gaps in the diagnosis of fignail conditions previously have been delineated, and prioritization of instruction on nail disease pathology and diagnostic procedures has been recommended to improve patient care.¹ The majority of nail unit melanomas are diagnosed at late stages, which has been attributed to deficiencies in clinical knowledge and failure to perform or inadequate biopsy techniques.² Notably, a survey of third-year dermatology residents (N=240) assessing experience in procedural dermatology showed that 58% performed 10 or fewer nail procedures and 30% did not feel competent in performing nail surgery.³ Furthermore, a survey examining the management of longitudinal melanonychia among attending and resident dermatologists (N=402) found that 62% of residents and 28% of total respondents were not confident in managing melanonychia.⁴

A limitation of this study was the lack of online data available for AAD annual meetings before 2013, so we were unable to characterize any long-term trends. Furthermore, we were unable to assess the educational reach of these sessions, as data on attendance are lacking.

This study demonstrates a paucity of nail-related content at the AAD annual meetings. The introduction of the “Hands-on: Nail Surgery” in 2015 is an important step forward to diminish the knowledge gap in the diagnosis of various nail diseases and malignancies. We recommend increasing the number of hours and overall content of didactic nail sessions at the AAD annual meeting to further the knowledge and procedural skills of dermatologists in caring for patients with nail disorders.

 

To the Editor:

The diagnosis and treatment of nail conditions are necessary competencies for board-certified dermatologists, but appropriate education often is lacking.¹ The American Academy of Dermatology (AAD) annual meeting is one of the largest and most highly attended dermatology educational conferences worldwide. We sought to determine the number of hours dedicated to nail-related topics at the AAD annual meetings from 2013 to 2019.

We accessed programs from the AAD annual meetings archive online (https://www.aad.org/meetings/previous-meetings-archive), and we used hair and psoriasis content for comparison. Event titles and descriptions were searched for nail-related content (using search terms nail, onychia, and onycho), hair-related content (hair, alopecia, trichosis, hirsutism), and psoriasis content (psoriasis). Data acquired for each event included the date, hours, title, and event type (eg, forum, course, focus session, symposium, discussion group, workshop, plenary session).

The number of hours dedicated to nail education consistently lagged behind those related to hair and psoriasis content during the study period (Figure 1). According to the AAD, the conference runs Friday to Tuesday with higher attendance Friday to Sunday (Tim Moses, personal communication, July 9, 2019). Lectures during the weekend are likely to have a broader reach than lectures on Monday and Tuesday. The proportion of nail content during weekend prime time slots was similar to that of hair and psoriasis (Figure 2). Plenary sessions often are presented by renowned experts on hot topics in dermatology. Notably, hair (2014-2015) and psoriasis (2015-2017) content were represented in the plenary sessions during the study period, while nail content was not featured.

Our study shows that nail-related education was underrepresented at the AAD annual meetings from 2013 to 2019 compared to hair- and psoriasis-related content. Educational gaps in the diagnosis of fignail conditions previously have been delineated, and prioritization of instruction on nail disease pathology and diagnostic procedures has been recommended to improve patient care.¹ The majority of nail unit melanomas are diagnosed at late stages, which has been attributed to deficiencies in clinical knowledge and failure to perform or inadequate biopsy techniques.² Notably, a survey of third-year dermatology residents (N=240) assessing experience in procedural dermatology showed that 58% performed 10 or fewer nail procedures and 30% did not feel competent in performing nail surgery.³ Furthermore, a survey examining the management of longitudinal melanonychia among attending and resident dermatologists (N=402) found that 62% of residents and 28% of total respondents were not confident in managing melanonychia.⁴

A limitation of this study was the lack of online data available for AAD annual meetings before 2013, so we were unable to characterize any long-term trends. Furthermore, we were unable to assess the educational reach of these sessions, as data on attendance are lacking.

This study demonstrates a paucity of nail-related content at the AAD annual meetings. The introduction of the “Hands-on: Nail Surgery” in 2015 is an important step forward to diminish the knowledge gap in the diagnosis of various nail diseases and malignancies. We recommend increasing the number of hours and overall content of didactic nail sessions at the AAD annual meeting to further the knowledge and procedural skills of dermatologists in caring for patients with nail disorders.

 

To the Editor:

The diagnosis and treatment of nail conditions are necessary competencies for board-certified dermatologists, but appropriate education often is lacking.¹ The American Academy of Dermatology (AAD) annual meeting is one of the largest and most highly attended dermatology educational conferences worldwide. We sought to determine the number of hours dedicated to nail-related topics at the AAD annual meetings from 2013 to 2019.

We accessed programs from the AAD annual meetings archive online (https://www.aad.org/meetings/previous-meetings-archive), and we used hair and psoriasis content for comparison. Event titles and descriptions were searched for nail-related content (using search terms nail, onychia, and onycho), hair-related content (hair, alopecia, trichosis, hirsutism), and psoriasis content (psoriasis). Data acquired for each event included the date, hours, title, and event type (eg, forum, course, focus session, symposium, discussion group, workshop, plenary session).

The number of hours dedicated to nail education consistently lagged behind those related to hair and psoriasis content during the study period (Figure 1). According to the AAD, the conference runs Friday to Tuesday with higher attendance Friday to Sunday (Tim Moses, personal communication, July 9, 2019). Lectures during the weekend are likely to have a broader reach than lectures on Monday and Tuesday. The proportion of nail content during weekend prime time slots was similar to that of hair and psoriasis (Figure 2). Plenary sessions often are presented by renowned experts on hot topics in dermatology. Notably, hair (2014-2015) and psoriasis (2015-2017) content were represented in the plenary sessions during the study period, while nail content was not featured.

Our study shows that nail-related education was underrepresented at the AAD annual meetings from 2013 to 2019 compared to hair- and psoriasis-related content. Educational gaps in the diagnosis of fignail conditions previously have been delineated, and prioritization of instruction on nail disease pathology and diagnostic procedures has been recommended to improve patient care.¹ The majority of nail unit melanomas are diagnosed at late stages, which has been attributed to deficiencies in clinical knowledge and failure to perform or inadequate biopsy techniques.² Notably, a survey of third-year dermatology residents (N=240) assessing experience in procedural dermatology showed that 58% performed 10 or fewer nail procedures and 30% did not feel competent in performing nail surgery.³ Furthermore, a survey examining the management of longitudinal melanonychia among attending and resident dermatologists (N=402) found that 62% of residents and 28% of total respondents were not confident in managing melanonychia.⁴

A limitation of this study was the lack of online data available for AAD annual meetings before 2013, so we were unable to characterize any long-term trends. Furthermore, we were unable to assess the educational reach of these sessions, as data on attendance are lacking.

This study demonstrates a paucity of nail-related content at the AAD annual meetings. The introduction of the “Hands-on: Nail Surgery” in 2015 is an important step forward to diminish the knowledge gap in the diagnosis of various nail diseases and malignancies. We recommend increasing the number of hours and overall content of didactic nail sessions at the AAD annual meeting to further the knowledge and procedural skills of dermatologists in caring for patients with nail disorders.

Male facial hair trends are continuously changing and are influenced by culture, geography, religion, and ethnicity.¹ Although the natural pattern of these hairs is largely androgen dependent, the phenotypic presentation often is a result of contemporary grooming practices that reflect prevailing trends.² Beards are common throughout adulthood, and thus, preserving this facial hair pattern is considered with reconstructive techniques.^3,4 Male facial skin physiology and beard hair biology are a dynamic interplay between both internal (eg, hormonal) and external (eg, shaving) variables. The density of beard hair follicles varies within different subunits, ranging between 20 and 80 follicles/cm². Macroscopically, hairs vary in length, diameter, color, and growth rate across individuals and ethnicities.^1,5

There is a paucity of literature assessing if male facial hair offers a protective role for external insults. One study utilized dosimetry to examine the effectiveness of facial hair on mannequins with varying lengths of hair in protecting against erythemal UV radiation (UVR). The authors concluded that, although facial hair provides protection from UVR, it is not significant.⁶ In a study of 200 male patients with actinic keratosis on the head and face, Liu et al⁷ demonstrated that sheltering mustaches, defined as greater than 9 mm in length, reduced the risk for developing an actinic keratosis on the lower lip by a factor of 16 (P=.0003).

We sought to determine if facial hair growth is implicated in the diagnosis and treatment of cutaneous malignancies. Specifically, we hypothesized that the presence of facial hair leads to a delay in diagnosis with increased subclinical growth given that tumors may be camouflaged and go undetected. Although there is a lack of literature, our anecdotal evidence suggests that male patients with facial hair have larger tumors compared to patients who do not regularly maintain any facial hair.

Methods

We performed a retrospective chart review following approval from the institutional review board at The University of North Carolina at Chapel Hill. We identified all male patients with a cutaneous malignancy located on the face who were treated from January 2015 to December 2018. Photographs were reviewed and patients with tumors located within the following facial hair-bearing anatomic subunits were included: lip, melolabial fold, chin, mandible, preauricular cheek, buccal cheek, and parotid-masseteric cheek. Tumors located within the medial cheek were excluded.

Facial hair growth was determined via image review. Because biopsy photographs were not uploaded into the health record for patients who were referred externally, we reviewed all historical photographs for patients who had undergone prior Mohs micrographic surgery at The University of North Carolina at Chapel Hill, preoperative photographs, and follow-up photographs as a proxy to determine facial hair status. Postoperative photographs taken within 2 weeks following surgery were not reviewed, as any facial hair growth was likely due to disinclination on behalf of the patient to shave near or over the incision. Age, number of days from biopsy to surgery, pathology, preoperative tumor size, number of Mohs layers, and defect size also were extrapolated from our chart review.

Statistical Analysis
Summary statistics were applied to describe demographic and clinical characteristics. An unpaired 2-tailed t test was utilized to test the null hypothesis that the mean difference was zero. The χ² test was used for categorical variables. Results achieving P<.05 were considered statistically significant.

 

Results

We reviewed medical records for 171 patients with facial hair and 336 patients without facial hair. The primary outcomes for this study assessed tumor and defect size in patients with facial hair compared to patients with no facial hair (Table 1). On average, patients who had facial hair were younger (67.5 years vs 74.0 years, P<.001). The median number of days from biopsy to surgery (43.0 vs 44.0 days) was comparable across both groups. The majority of patients (47%) exhibited a beard, while 30% had a mustache and 23% had a goatee. The most common tumor location was the preauricular cheek for both groups (29% and 28%, respectively). The mean preoperative tumor size in the facial hair cohort was 1.40 cm compared to 1.22 cm in the group with no facial hair (P=.03). The mean number of Mohs layers in the facial hair cohort was 1.53 compared to 1.33 in the group with no facial hair (P=.03). The facial hair cohort also had a larger mean postoperative defect size (2.18 cm) compared to the group with no facial hair (1.98 cm); however, this finding was not significant (P=.05).

We then stratified our data to analyze only lip tumors in patients with and without a mustache (Table 2). The mean preoperative tumor size in the mustache cohort was 1.10 cm compared to 0.82 cm in the group with no mustaches (P=.046). The mean number of Mohs layers in the mustache cohort was 1.57 compared to 1.42 in the group with no mustaches (P=.43). The mustache cohort also had a larger mean postoperative defect size (1.63 cm) compared to the group with no facial hair (1.33 cm), though this finding also did not reach significance (P=.13).

Comment

Our findings support anecdotal observations that tumors in men with facial hair are larger, require more Mohs layers, and result in larger defects compared with patients who are clean shaven. Similarly, in lip tumors, men with a mustache had a larger preoperative tumor size. Although these patients also required more Mohs layers to clear and a larger defect size, these parameters did not reach significance. These outcomes may, in part, be explained by a delay in diagnosis, as patients with facial hair may not notice any new suspicious lesions within the underlying skin as easily as patients with glabrous skin.

Although facial hair may shield skin from UVR, we agree with Parisi et al⁶ that this protection is marginal at best and that early persistent exposure to UVR plays a much more notable role in cutaneous carcinogenesis. As more men continue to grow facial hairstyles that emulate historical or contemporary trends, dermatologists should emphasize the risk for cutaneous malignancies within these sun-exposed areas of the face. Although some facial hair practices may reflect cultural or ethnic settings, the majority reflect a desired appearance that is achieved with grooming or otherwise.

Skin cancer screening in men with facial hair, particularly those with a strong history of UVR exposure and/or family history, should be discussed and encouraged to diagnose cutaneous tumors earlier. We encourage men with facial hair to be cognizant that cutaneous malignancies can arise within hair-bearing skin and to incorporate self–skin checks into grooming routines, which is particularly important in men with dense facial hair who forego regular self-care grooming or trim intermittently. Furthermore, we urge dermatologists to continue to thoroughly examine the underlying skin, especially in patients with full beards, during skin examinations. Diagnosing and treating cutaneous malignancies early is imperative to maximize ideal functional and cosmetic outcomes, particularly within perioral and lip subunits, where marginal millimeters can impact reconstructive complexity.

Conclusion

Men with facial hair who had cutaneous tumors in our study exhibited larger tumors, required more Mohs layers, and had a larger defect size compared to men without any facial hair growth. Similar findings also were noted when we stratified and compared lip tumors in patients with and without mustaches. Given these observations, patients and dermatologists should continue to have a high index of suspicion for any concerning lesion located within skin underlying facial hair. Regular screening in men with facial hair should be discussed and encouraged to diagnose and treat potential cutaneous tumors earlier.

Male facial hair trends are continuously changing and are influenced by culture, geography, religion, and ethnicity.¹ Although the natural pattern of these hairs is largely androgen dependent, the phenotypic presentation often is a result of contemporary grooming practices that reflect prevailing trends.² Beards are common throughout adulthood, and thus, preserving this facial hair pattern is considered with reconstructive techniques.^3,4 Male facial skin physiology and beard hair biology are a dynamic interplay between both internal (eg, hormonal) and external (eg, shaving) variables. The density of beard hair follicles varies within different subunits, ranging between 20 and 80 follicles/cm². Macroscopically, hairs vary in length, diameter, color, and growth rate across individuals and ethnicities.^1,5

There is a paucity of literature assessing if male facial hair offers a protective role for external insults. One study utilized dosimetry to examine the effectiveness of facial hair on mannequins with varying lengths of hair in protecting against erythemal UV radiation (UVR). The authors concluded that, although facial hair provides protection from UVR, it is not significant.⁶ In a study of 200 male patients with actinic keratosis on the head and face, Liu et al⁷ demonstrated that sheltering mustaches, defined as greater than 9 mm in length, reduced the risk for developing an actinic keratosis on the lower lip by a factor of 16 (P=.0003).

We sought to determine if facial hair growth is implicated in the diagnosis and treatment of cutaneous malignancies. Specifically, we hypothesized that the presence of facial hair leads to a delay in diagnosis with increased subclinical growth given that tumors may be camouflaged and go undetected. Although there is a lack of literature, our anecdotal evidence suggests that male patients with facial hair have larger tumors compared to patients who do not regularly maintain any facial hair.

Methods

We performed a retrospective chart review following approval from the institutional review board at The University of North Carolina at Chapel Hill. We identified all male patients with a cutaneous malignancy located on the face who were treated from January 2015 to December 2018. Photographs were reviewed and patients with tumors located within the following facial hair-bearing anatomic subunits were included: lip, melolabial fold, chin, mandible, preauricular cheek, buccal cheek, and parotid-masseteric cheek. Tumors located within the medial cheek were excluded.

Facial hair growth was determined via image review. Because biopsy photographs were not uploaded into the health record for patients who were referred externally, we reviewed all historical photographs for patients who had undergone prior Mohs micrographic surgery at The University of North Carolina at Chapel Hill, preoperative photographs, and follow-up photographs as a proxy to determine facial hair status. Postoperative photographs taken within 2 weeks following surgery were not reviewed, as any facial hair growth was likely due to disinclination on behalf of the patient to shave near or over the incision. Age, number of days from biopsy to surgery, pathology, preoperative tumor size, number of Mohs layers, and defect size also were extrapolated from our chart review.

Statistical Analysis
Summary statistics were applied to describe demographic and clinical characteristics. An unpaired 2-tailed t test was utilized to test the null hypothesis that the mean difference was zero. The χ² test was used for categorical variables. Results achieving P<.05 were considered statistically significant.

 

Results

We reviewed medical records for 171 patients with facial hair and 336 patients without facial hair. The primary outcomes for this study assessed tumor and defect size in patients with facial hair compared to patients with no facial hair (Table 1). On average, patients who had facial hair were younger (67.5 years vs 74.0 years, P<.001). The median number of days from biopsy to surgery (43.0 vs 44.0 days) was comparable across both groups. The majority of patients (47%) exhibited a beard, while 30% had a mustache and 23% had a goatee. The most common tumor location was the preauricular cheek for both groups (29% and 28%, respectively). The mean preoperative tumor size in the facial hair cohort was 1.40 cm compared to 1.22 cm in the group with no facial hair (P=.03). The mean number of Mohs layers in the facial hair cohort was 1.53 compared to 1.33 in the group with no facial hair (P=.03). The facial hair cohort also had a larger mean postoperative defect size (2.18 cm) compared to the group with no facial hair (1.98 cm); however, this finding was not significant (P=.05).

We then stratified our data to analyze only lip tumors in patients with and without a mustache (Table 2). The mean preoperative tumor size in the mustache cohort was 1.10 cm compared to 0.82 cm in the group with no mustaches (P=.046). The mean number of Mohs layers in the mustache cohort was 1.57 compared to 1.42 in the group with no mustaches (P=.43). The mustache cohort also had a larger mean postoperative defect size (1.63 cm) compared to the group with no facial hair (1.33 cm), though this finding also did not reach significance (P=.13).

Comment

Our findings support anecdotal observations that tumors in men with facial hair are larger, require more Mohs layers, and result in larger defects compared with patients who are clean shaven. Similarly, in lip tumors, men with a mustache had a larger preoperative tumor size. Although these patients also required more Mohs layers to clear and a larger defect size, these parameters did not reach significance. These outcomes may, in part, be explained by a delay in diagnosis, as patients with facial hair may not notice any new suspicious lesions within the underlying skin as easily as patients with glabrous skin.

Although facial hair may shield skin from UVR, we agree with Parisi et al⁶ that this protection is marginal at best and that early persistent exposure to UVR plays a much more notable role in cutaneous carcinogenesis. As more men continue to grow facial hairstyles that emulate historical or contemporary trends, dermatologists should emphasize the risk for cutaneous malignancies within these sun-exposed areas of the face. Although some facial hair practices may reflect cultural or ethnic settings, the majority reflect a desired appearance that is achieved with grooming or otherwise.

Skin cancer screening in men with facial hair, particularly those with a strong history of UVR exposure and/or family history, should be discussed and encouraged to diagnose cutaneous tumors earlier. We encourage men with facial hair to be cognizant that cutaneous malignancies can arise within hair-bearing skin and to incorporate self–skin checks into grooming routines, which is particularly important in men with dense facial hair who forego regular self-care grooming or trim intermittently. Furthermore, we urge dermatologists to continue to thoroughly examine the underlying skin, especially in patients with full beards, during skin examinations. Diagnosing and treating cutaneous malignancies early is imperative to maximize ideal functional and cosmetic outcomes, particularly within perioral and lip subunits, where marginal millimeters can impact reconstructive complexity.

Conclusion

Men with facial hair who had cutaneous tumors in our study exhibited larger tumors, required more Mohs layers, and had a larger defect size compared to men without any facial hair growth. Similar findings also were noted when we stratified and compared lip tumors in patients with and without mustaches. Given these observations, patients and dermatologists should continue to have a high index of suspicion for any concerning lesion located within skin underlying facial hair. Regular screening in men with facial hair should be discussed and encouraged to diagnose and treat potential cutaneous tumors earlier.

Male facial hair trends are continuously changing and are influenced by culture, geography, religion, and ethnicity.¹ Although the natural pattern of these hairs is largely androgen dependent, the phenotypic presentation often is a result of contemporary grooming practices that reflect prevailing trends.² Beards are common throughout adulthood, and thus, preserving this facial hair pattern is considered with reconstructive techniques.^3,4 Male facial skin physiology and beard hair biology are a dynamic interplay between both internal (eg, hormonal) and external (eg, shaving) variables. The density of beard hair follicles varies within different subunits, ranging between 20 and 80 follicles/cm². Macroscopically, hairs vary in length, diameter, color, and growth rate across individuals and ethnicities.^1,5

There is a paucity of literature assessing if male facial hair offers a protective role for external insults. One study utilized dosimetry to examine the effectiveness of facial hair on mannequins with varying lengths of hair in protecting against erythemal UV radiation (UVR). The authors concluded that, although facial hair provides protection from UVR, it is not significant.⁶ In a study of 200 male patients with actinic keratosis on the head and face, Liu et al⁷ demonstrated that sheltering mustaches, defined as greater than 9 mm in length, reduced the risk for developing an actinic keratosis on the lower lip by a factor of 16 (P=.0003).

We sought to determine if facial hair growth is implicated in the diagnosis and treatment of cutaneous malignancies. Specifically, we hypothesized that the presence of facial hair leads to a delay in diagnosis with increased subclinical growth given that tumors may be camouflaged and go undetected. Although there is a lack of literature, our anecdotal evidence suggests that male patients with facial hair have larger tumors compared to patients who do not regularly maintain any facial hair.

Methods

We performed a retrospective chart review following approval from the institutional review board at The University of North Carolina at Chapel Hill. We identified all male patients with a cutaneous malignancy located on the face who were treated from January 2015 to December 2018. Photographs were reviewed and patients with tumors located within the following facial hair-bearing anatomic subunits were included: lip, melolabial fold, chin, mandible, preauricular cheek, buccal cheek, and parotid-masseteric cheek. Tumors located within the medial cheek were excluded.

Facial hair growth was determined via image review. Because biopsy photographs were not uploaded into the health record for patients who were referred externally, we reviewed all historical photographs for patients who had undergone prior Mohs micrographic surgery at The University of North Carolina at Chapel Hill, preoperative photographs, and follow-up photographs as a proxy to determine facial hair status. Postoperative photographs taken within 2 weeks following surgery were not reviewed, as any facial hair growth was likely due to disinclination on behalf of the patient to shave near or over the incision. Age, number of days from biopsy to surgery, pathology, preoperative tumor size, number of Mohs layers, and defect size also were extrapolated from our chart review.

Statistical Analysis
Summary statistics were applied to describe demographic and clinical characteristics. An unpaired 2-tailed t test was utilized to test the null hypothesis that the mean difference was zero. The χ² test was used for categorical variables. Results achieving P<.05 were considered statistically significant.

 

Results

We reviewed medical records for 171 patients with facial hair and 336 patients without facial hair. The primary outcomes for this study assessed tumor and defect size in patients with facial hair compared to patients with no facial hair (Table 1). On average, patients who had facial hair were younger (67.5 years vs 74.0 years, P<.001). The median number of days from biopsy to surgery (43.0 vs 44.0 days) was comparable across both groups. The majority of patients (47%) exhibited a beard, while 30% had a mustache and 23% had a goatee. The most common tumor location was the preauricular cheek for both groups (29% and 28%, respectively). The mean preoperative tumor size in the facial hair cohort was 1.40 cm compared to 1.22 cm in the group with no facial hair (P=.03). The mean number of Mohs layers in the facial hair cohort was 1.53 compared to 1.33 in the group with no facial hair (P=.03). The facial hair cohort also had a larger mean postoperative defect size (2.18 cm) compared to the group with no facial hair (1.98 cm); however, this finding was not significant (P=.05).

We then stratified our data to analyze only lip tumors in patients with and without a mustache (Table 2). The mean preoperative tumor size in the mustache cohort was 1.10 cm compared to 0.82 cm in the group with no mustaches (P=.046). The mean number of Mohs layers in the mustache cohort was 1.57 compared to 1.42 in the group with no mustaches (P=.43). The mustache cohort also had a larger mean postoperative defect size (1.63 cm) compared to the group with no facial hair (1.33 cm), though this finding also did not reach significance (P=.13).

Comment

Our findings support anecdotal observations that tumors in men with facial hair are larger, require more Mohs layers, and result in larger defects compared with patients who are clean shaven. Similarly, in lip tumors, men with a mustache had a larger preoperative tumor size. Although these patients also required more Mohs layers to clear and a larger defect size, these parameters did not reach significance. These outcomes may, in part, be explained by a delay in diagnosis, as patients with facial hair may not notice any new suspicious lesions within the underlying skin as easily as patients with glabrous skin.

Although facial hair may shield skin from UVR, we agree with Parisi et al⁶ that this protection is marginal at best and that early persistent exposure to UVR plays a much more notable role in cutaneous carcinogenesis. As more men continue to grow facial hairstyles that emulate historical or contemporary trends, dermatologists should emphasize the risk for cutaneous malignancies within these sun-exposed areas of the face. Although some facial hair practices may reflect cultural or ethnic settings, the majority reflect a desired appearance that is achieved with grooming or otherwise.

Skin cancer screening in men with facial hair, particularly those with a strong history of UVR exposure and/or family history, should be discussed and encouraged to diagnose cutaneous tumors earlier. We encourage men with facial hair to be cognizant that cutaneous malignancies can arise within hair-bearing skin and to incorporate self–skin checks into grooming routines, which is particularly important in men with dense facial hair who forego regular self-care grooming or trim intermittently. Furthermore, we urge dermatologists to continue to thoroughly examine the underlying skin, especially in patients with full beards, during skin examinations. Diagnosing and treating cutaneous malignancies early is imperative to maximize ideal functional and cosmetic outcomes, particularly within perioral and lip subunits, where marginal millimeters can impact reconstructive complexity.

Conclusion

Men with facial hair who had cutaneous tumors in our study exhibited larger tumors, required more Mohs layers, and had a larger defect size compared to men without any facial hair growth. Similar findings also were noted when we stratified and compared lip tumors in patients with and without mustaches. Given these observations, patients and dermatologists should continue to have a high index of suspicion for any concerning lesion located within skin underlying facial hair. Regular screening in men with facial hair should be discussed and encouraged to diagnose and treat potential cutaneous tumors earlier.

 

To the Editor:

A thorough full-body skin examination (FBSE) is an integral component of a dermatologic encounter and helps identify potentially malignant and high-risk lesions, particularly in areas that are difficult for the patient to visualize.¹ Despite these benefits, many patients experience discomfort and anxiety about this examination because it involves sensitive anatomical areas. The true psychological impact of an FBSE is not clearly understood; however, research into improving patient comfort in these circumstances can have a broad positive impact.² The purpose of this pilot study was to establish patients’ willingness to complete a pre-encounter questionnaire that defines their FBSE preferences as well as to identify the anatomical areas that are of most concern.

This study was approved by the University of Kansas institutional review board as nonhuman subjects research. A pre-encounter questionnaire that included information about the benefits of FBSEs was administered to 34 patients, allowing them to identify anatomic locations that they wanted to exclude from the FBSE.

Following the patient visit (in which the identified anatomical locations were excluded), patients were given a brief exit survey that asked about (1) their preference for a pre-encounter FBSE questionnaire and (2) the impact of the questionnaire on their anxiety level throughout the encounter. Preference for asking was surveyed using a 10-point scale (10=strong preference for the pre-encounter survey; 1=strong preference against the pre-encounter survey). Change in anxiety was surveyed using a 10-point scale (10=strong reduction in anxiety after the pre-encounter survey; 1=strong increase in anxiety after the pre-encounter survey). Statistical analysis was performed using 2-tailed unpaired t tests, with P<.05 considered statistically significant.

Twenty female and 14 male patients were enrolled (mean age, 53 years)(Table). The most commonly excluded anatomical location on the pre-encounter survey was the genitals, followed by the buttocks, breasts/chest, legs, feet, and abdomen (Table); 10 (71%) male and 13 (65%) female respondents did not exclude any component of the FBSE.

After the provider visit, females had a higher preference for the pre-encounter survey (mean score, 9.0) compared to males (mean score, 7.2; P=.021). Similarly, females had reduced anxiety about the office visit after survey administration compared to males (mean score, 8.3 vs 6.0; P=.001)(Table).

The results of our pilot study showed that a brief pre-encounter questionnaire may reduce the distress associated with an FBSE. Our survey took less than 1 minute to complete and served as a useful guide to direct the provider during the FBSE. Moreover, recognizing that patients do not want certain anatomic locations examined can serve as an opportunity for the dermatologist to provide helpful home skin check instructions and recommendations.

The small sample size was a limitation of this study. Future studies can assess with greater precision the clear benefits of a pre-encounter survey as well as the benefits or drawbacks of a survey compared to other modalities that are aimed at reducing patient anxiety about the FBSE, such as having the physician directly ask the patient about areas to avoid during the examination.

A pre-encounter survey about the FBSE can serve as an efficient means of determining patient preference and reducing self-reported anxiety about the visit.

 

To the Editor:

A thorough full-body skin examination (FBSE) is an integral component of a dermatologic encounter and helps identify potentially malignant and high-risk lesions, particularly in areas that are difficult for the patient to visualize.¹ Despite these benefits, many patients experience discomfort and anxiety about this examination because it involves sensitive anatomical areas. The true psychological impact of an FBSE is not clearly understood; however, research into improving patient comfort in these circumstances can have a broad positive impact.² The purpose of this pilot study was to establish patients’ willingness to complete a pre-encounter questionnaire that defines their FBSE preferences as well as to identify the anatomical areas that are of most concern.

This study was approved by the University of Kansas institutional review board as nonhuman subjects research. A pre-encounter questionnaire that included information about the benefits of FBSEs was administered to 34 patients, allowing them to identify anatomic locations that they wanted to exclude from the FBSE.

Following the patient visit (in which the identified anatomical locations were excluded), patients were given a brief exit survey that asked about (1) their preference for a pre-encounter FBSE questionnaire and (2) the impact of the questionnaire on their anxiety level throughout the encounter. Preference for asking was surveyed using a 10-point scale (10=strong preference for the pre-encounter survey; 1=strong preference against the pre-encounter survey). Change in anxiety was surveyed using a 10-point scale (10=strong reduction in anxiety after the pre-encounter survey; 1=strong increase in anxiety after the pre-encounter survey). Statistical analysis was performed using 2-tailed unpaired t tests, with P<.05 considered statistically significant.

Twenty female and 14 male patients were enrolled (mean age, 53 years)(Table). The most commonly excluded anatomical location on the pre-encounter survey was the genitals, followed by the buttocks, breasts/chest, legs, feet, and abdomen (Table); 10 (71%) male and 13 (65%) female respondents did not exclude any component of the FBSE.

After the provider visit, females had a higher preference for the pre-encounter survey (mean score, 9.0) compared to males (mean score, 7.2; P=.021). Similarly, females had reduced anxiety about the office visit after survey administration compared to males (mean score, 8.3 vs 6.0; P=.001)(Table).

The results of our pilot study showed that a brief pre-encounter questionnaire may reduce the distress associated with an FBSE. Our survey took less than 1 minute to complete and served as a useful guide to direct the provider during the FBSE. Moreover, recognizing that patients do not want certain anatomic locations examined can serve as an opportunity for the dermatologist to provide helpful home skin check instructions and recommendations.

The small sample size was a limitation of this study. Future studies can assess with greater precision the clear benefits of a pre-encounter survey as well as the benefits or drawbacks of a survey compared to other modalities that are aimed at reducing patient anxiety about the FBSE, such as having the physician directly ask the patient about areas to avoid during the examination.

A pre-encounter survey about the FBSE can serve as an efficient means of determining patient preference and reducing self-reported anxiety about the visit.

 

To the Editor:

A thorough full-body skin examination (FBSE) is an integral component of a dermatologic encounter and helps identify potentially malignant and high-risk lesions, particularly in areas that are difficult for the patient to visualize.¹ Despite these benefits, many patients experience discomfort and anxiety about this examination because it involves sensitive anatomical areas. The true psychological impact of an FBSE is not clearly understood; however, research into improving patient comfort in these circumstances can have a broad positive impact.² The purpose of this pilot study was to establish patients’ willingness to complete a pre-encounter questionnaire that defines their FBSE preferences as well as to identify the anatomical areas that are of most concern.

This study was approved by the University of Kansas institutional review board as nonhuman subjects research. A pre-encounter questionnaire that included information about the benefits of FBSEs was administered to 34 patients, allowing them to identify anatomic locations that they wanted to exclude from the FBSE.

Following the patient visit (in which the identified anatomical locations were excluded), patients were given a brief exit survey that asked about (1) their preference for a pre-encounter FBSE questionnaire and (2) the impact of the questionnaire on their anxiety level throughout the encounter. Preference for asking was surveyed using a 10-point scale (10=strong preference for the pre-encounter survey; 1=strong preference against the pre-encounter survey). Change in anxiety was surveyed using a 10-point scale (10=strong reduction in anxiety after the pre-encounter survey; 1=strong increase in anxiety after the pre-encounter survey). Statistical analysis was performed using 2-tailed unpaired t tests, with P<.05 considered statistically significant.

Twenty female and 14 male patients were enrolled (mean age, 53 years)(Table). The most commonly excluded anatomical location on the pre-encounter survey was the genitals, followed by the buttocks, breasts/chest, legs, feet, and abdomen (Table); 10 (71%) male and 13 (65%) female respondents did not exclude any component of the FBSE.

After the provider visit, females had a higher preference for the pre-encounter survey (mean score, 9.0) compared to males (mean score, 7.2; P=.021). Similarly, females had reduced anxiety about the office visit after survey administration compared to males (mean score, 8.3 vs 6.0; P=.001)(Table).

The results of our pilot study showed that a brief pre-encounter questionnaire may reduce the distress associated with an FBSE. Our survey took less than 1 minute to complete and served as a useful guide to direct the provider during the FBSE. Moreover, recognizing that patients do not want certain anatomic locations examined can serve as an opportunity for the dermatologist to provide helpful home skin check instructions and recommendations.

The small sample size was a limitation of this study. Future studies can assess with greater precision the clear benefits of a pre-encounter survey as well as the benefits or drawbacks of a survey compared to other modalities that are aimed at reducing patient anxiety about the FBSE, such as having the physician directly ask the patient about areas to avoid during the examination.

A pre-encounter survey about the FBSE can serve as an efficient means of determining patient preference and reducing self-reported anxiety about the visit.