Diabetes

Two studies published in March by researchers at the Veterans Affairs Saint Louis Healthcare System highlight the clinical significance of glucagon-like peptide 1 receptor agonists (GLP-1s) and their impact on reducing substance use disorder (SUD) risks. The studies also explore the impact of GLP-1 discontinuation or interruption on their effectiveness in protection against the cardiovascular events.

In one study, Al-Aly et al assigned 606,434 veterans with type 2 diabetes to 1 of 2 protocols, comparing GLP-1s with sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and followed the patients for up to 3 years. Al-Aly et al found that GLP-1s were “consistently associated” with a lower risk of developing SUDs, including those involving alcohol, cannabis, cocaine, nicotine, and opioids. The findings suggested “potential preventive effects across a broad range of addictive substances.”

In participants with pre-existing SUDs, GLP-1s were also associated with reduced risks of SUD-related emergency department visits, hospital admissions, and mortality, in addition to drug overdoses and suicidal behaviors. A study published in 2025 from the same research group reported that GLP-1s could have a variety of health benefits, including reducing the risk of incident alcohol and cannabis disorders, neurocognitive disorders (such as Alzheimer's disease and dementia), coagulation disorders, cardiometabolic disorders, infectious illnesses and several respiratory conditions, but less was known about the potential for preventing development of opioid use disorder and other SUDs. 

GLP-1s target the brain’s reward pathways and have recently made attention-grabbing headlines regarding celebrity weight loss, with social media boosting public interest. One study, for example, found 100 videos on TikTok with the #Ozempic viewed nearly 70 million times.

Al-Aly et al used SGLT-2 inhibitors as active comparators because “they have no established direct actions on mesolimbic reward circuits in the brain, whereas GLP-1 receptors are present in areas of the brain involved in impulse control and reward signaling.”

The second study found that quitting or pausing GLP-1 treatment for 6 months could have a rebound effect and possibly reverse any progress. Discontinuing GLP-1 treatment is common, with rates ranging from 36% to 81% in the first year. Stopping or interrupting the treatment is often followed by weight regain and a rebound in inflammation, both major drivers in cardiovascular disease risk. 

The study followed 132,551 VA patients using GLP-1s and 201,136 using sulfonylureas from 2017 through 2023. About two-thirds of participants took semaglutide, prescribed as Ozempic to treat diabetes and Wegovy to reduce obesity. A total of 26% of the participants stopped GLP-1 treatment during the follow-up period, with 64% occurring during the first year. Most (67%) treatment interruptions also came in the first year.

Compared with incident use of sulfonylureas, incident use of GLP-1s was associated with a reduced risk of heart attack, stroke, or death. Patients who took the GLP-1s without interruption > 3 years experienced an 18% lower risk for heart attack or stroke.  

Cardiovascular benefits accumulated with continuous use over 3 years, but even brief periods of discontinuations or interruptions could progressively erode and ultimately reverse this protection, the researchers found. Discontinuing treatment for half a year was associated with an increased risk of major adverse cardiovascular events (incidence risk ratio [IRR], 1.04), while longer gaps were progressively associated with a higher risk of disease (IRR, 1.12 for 1 year; IRR, 1.16 for 2 years of interrupted use, respectively).

Dr. Ziyad Al-Aly, a study author and Chief of the Research and Education Service at the Veterans Affairs Saint Louis Healthcare System, called it “metabolic whiplash.” In an interview, he said it was important to caution patients that these medications “need to be taken for the long haul. This is not something (patients) can take for a month or 2 or 3 and get off of it. It's not going to work like that.”

Two studies published in March by researchers at the Veterans Affairs Saint Louis Healthcare System highlight the clinical significance of glucagon-like peptide 1 receptor agonists (GLP-1s) and their impact on reducing substance use disorder (SUD) risks. The studies also explore the impact of GLP-1 discontinuation or interruption on their effectiveness in protection against the cardiovascular events.

In one study, Al-Aly et al assigned 606,434 veterans with type 2 diabetes to 1 of 2 protocols, comparing GLP-1s with sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and followed the patients for up to 3 years. Al-Aly et al found that GLP-1s were “consistently associated” with a lower risk of developing SUDs, including those involving alcohol, cannabis, cocaine, nicotine, and opioids. The findings suggested “potential preventive effects across a broad range of addictive substances.”

In participants with pre-existing SUDs, GLP-1s were also associated with reduced risks of SUD-related emergency department visits, hospital admissions, and mortality, in addition to drug overdoses and suicidal behaviors. A study published in 2025 from the same research group reported that GLP-1s could have a variety of health benefits, including reducing the risk of incident alcohol and cannabis disorders, neurocognitive disorders (such as Alzheimer's disease and dementia), coagulation disorders, cardiometabolic disorders, infectious illnesses and several respiratory conditions, but less was known about the potential for preventing development of opioid use disorder and other SUDs. 

GLP-1s target the brain’s reward pathways and have recently made attention-grabbing headlines regarding celebrity weight loss, with social media boosting public interest. One study, for example, found 100 videos on TikTok with the #Ozempic viewed nearly 70 million times.

Al-Aly et al used SGLT-2 inhibitors as active comparators because “they have no established direct actions on mesolimbic reward circuits in the brain, whereas GLP-1 receptors are present in areas of the brain involved in impulse control and reward signaling.”

The second study found that quitting or pausing GLP-1 treatment for 6 months could have a rebound effect and possibly reverse any progress. Discontinuing GLP-1 treatment is common, with rates ranging from 36% to 81% in the first year. Stopping or interrupting the treatment is often followed by weight regain and a rebound in inflammation, both major drivers in cardiovascular disease risk. 

The study followed 132,551 VA patients using GLP-1s and 201,136 using sulfonylureas from 2017 through 2023. About two-thirds of participants took semaglutide, prescribed as Ozempic to treat diabetes and Wegovy to reduce obesity. A total of 26% of the participants stopped GLP-1 treatment during the follow-up period, with 64% occurring during the first year. Most (67%) treatment interruptions also came in the first year.

Compared with incident use of sulfonylureas, incident use of GLP-1s was associated with a reduced risk of heart attack, stroke, or death. Patients who took the GLP-1s without interruption > 3 years experienced an 18% lower risk for heart attack or stroke.  

Cardiovascular benefits accumulated with continuous use over 3 years, but even brief periods of discontinuations or interruptions could progressively erode and ultimately reverse this protection, the researchers found. Discontinuing treatment for half a year was associated with an increased risk of major adverse cardiovascular events (incidence risk ratio [IRR], 1.04), while longer gaps were progressively associated with a higher risk of disease (IRR, 1.12 for 1 year; IRR, 1.16 for 2 years of interrupted use, respectively).

Dr. Ziyad Al-Aly, a study author and Chief of the Research and Education Service at the Veterans Affairs Saint Louis Healthcare System, called it “metabolic whiplash.” In an interview, he said it was important to caution patients that these medications “need to be taken for the long haul. This is not something (patients) can take for a month or 2 or 3 and get off of it. It's not going to work like that.”

Two studies published in March by researchers at the Veterans Affairs Saint Louis Healthcare System highlight the clinical significance of glucagon-like peptide 1 receptor agonists (GLP-1s) and their impact on reducing substance use disorder (SUD) risks. The studies also explore the impact of GLP-1 discontinuation or interruption on their effectiveness in protection against the cardiovascular events.

In one study, Al-Aly et al assigned 606,434 veterans with type 2 diabetes to 1 of 2 protocols, comparing GLP-1s with sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and followed the patients for up to 3 years. Al-Aly et al found that GLP-1s were “consistently associated” with a lower risk of developing SUDs, including those involving alcohol, cannabis, cocaine, nicotine, and opioids. The findings suggested “potential preventive effects across a broad range of addictive substances.”

In participants with pre-existing SUDs, GLP-1s were also associated with reduced risks of SUD-related emergency department visits, hospital admissions, and mortality, in addition to drug overdoses and suicidal behaviors. A study published in 2025 from the same research group reported that GLP-1s could have a variety of health benefits, including reducing the risk of incident alcohol and cannabis disorders, neurocognitive disorders (such as Alzheimer's disease and dementia), coagulation disorders, cardiometabolic disorders, infectious illnesses and several respiratory conditions, but less was known about the potential for preventing development of opioid use disorder and other SUDs. 

GLP-1s target the brain’s reward pathways and have recently made attention-grabbing headlines regarding celebrity weight loss, with social media boosting public interest. One study, for example, found 100 videos on TikTok with the #Ozempic viewed nearly 70 million times.

Al-Aly et al used SGLT-2 inhibitors as active comparators because “they have no established direct actions on mesolimbic reward circuits in the brain, whereas GLP-1 receptors are present in areas of the brain involved in impulse control and reward signaling.”

The second study found that quitting or pausing GLP-1 treatment for 6 months could have a rebound effect and possibly reverse any progress. Discontinuing GLP-1 treatment is common, with rates ranging from 36% to 81% in the first year. Stopping or interrupting the treatment is often followed by weight regain and a rebound in inflammation, both major drivers in cardiovascular disease risk. 

The study followed 132,551 VA patients using GLP-1s and 201,136 using sulfonylureas from 2017 through 2023. About two-thirds of participants took semaglutide, prescribed as Ozempic to treat diabetes and Wegovy to reduce obesity. A total of 26% of the participants stopped GLP-1 treatment during the follow-up period, with 64% occurring during the first year. Most (67%) treatment interruptions also came in the first year.

Compared with incident use of sulfonylureas, incident use of GLP-1s was associated with a reduced risk of heart attack, stroke, or death. Patients who took the GLP-1s without interruption > 3 years experienced an 18% lower risk for heart attack or stroke.  

Cardiovascular benefits accumulated with continuous use over 3 years, but even brief periods of discontinuations or interruptions could progressively erode and ultimately reverse this protection, the researchers found. Discontinuing treatment for half a year was associated with an increased risk of major adverse cardiovascular events (incidence risk ratio [IRR], 1.04), while longer gaps were progressively associated with a higher risk of disease (IRR, 1.12 for 1 year; IRR, 1.16 for 2 years of interrupted use, respectively).

Dr. Ziyad Al-Aly, a study author and Chief of the Research and Education Service at the Veterans Affairs Saint Louis Healthcare System, called it “metabolic whiplash.” In an interview, he said it was important to caution patients that these medications “need to be taken for the long haul. This is not something (patients) can take for a month or 2 or 3 and get off of it. It's not going to work like that.”

Semaglutide is linked to a small but significantly elevated risk for nonarteritic anterior ischemic optic neuropathy (NAION), according to the results of a new study.

The risk for NAION was approximately twofold greater among people taking semaglutide than among those using an SGLT2 inhibitor, but the absolute risk was still only about 29 per 10,000 people, study author Jennifer S. Lee, MD, PhD, professor of medicine at Stanford School of Medicine in Palo Alto, California, told Medscape Medical News.

In NAION, the optic nerve suddenly loses its blood supply, leading to vision loss, sometimes described as a “stroke of the eye.”

“Clinicians should balance this rare but serious vision-loss risk against semaglutide’s meaningful cardiometabolic benefits,” said Lee, who is also chief of Research and Clinical Innovations and director of the VA National Center for Collaborative Healthcare Innovation at the VA Palo Alto Healthcare System, US Department of Veterans Affairs.

“We recommend counseling patients to seek prompt evaluation if they experience visual symptoms, particularly sudden vision changes,” she added.

The new findings, recently published in JAMA Ophthalmology, are the latest from several studies examining a possible association between NAION and semaglutide, some of which have been negative.

In the current study, the investigators analyzed US Veterans Health Administration data for a total of 11,478 veterans with type 2 diabetes (T2D) who initiated semaglutide and 90,883 who initiated an SGLT2 inhibitor (mostly empagliflozin) between March 1, 2018, and March 1, 2025. Overlap weighting was used to balance the two groups by age, BMI, A1c, sex, and race.

Over a median follow-up of 2.1 years, NAION developed in 123 initiating semaglutide vs 143 initiating SGLT2 inhibitors, with rates of 123 vs 67 per 100,000 person-years. After overlap weighting, the risk for NAION was 2.3-fold higher with semaglutide (hazard ratio [HR], 2.33, P < .001).

Over a maximum of 7.5 years of follow-up, the overlap-weighted NAION incidence was 0.29% vs 0.13% for semaglutide initiators vs SGLT2 initiators.

While the mechanism(s) for the association remains unclear, hypotheses include hypotension, volume depletion from gastrointestinal side effects, rapid glucose improvement leading to stress on the eye microvasculature, and impaired vascular autoregulation at the head of the optic nerve, Lee said.

Different Studies, Different Answers

“Growing observational evidence links GLP-1 receptor agonists, particularly semaglutide, to NAION, with similar twofold to threefold risks reported in several large analyses,” Lee told Medscape Medical News.

“However, other studies have observed attenuated or null associations,” she said. “The mechanism remains unknown and requires further investigation. More research is needed to determine whether this is a class effect or specific to semaglutide.”

A possible reason that some other studies have not found an association may be due to methodological differences, particularly in how NAION cases are identified, noted Joseph F. Rizzo III, MD, professor of ophthalmology and director of the Neuro-ophthalmology Service at Harvard Medical School in Boston.

Rizzo and colleagues were the first to conduct a study looking at this association in their own neuro-ophthalmology population, thereby guaranteeing that the diagnosis was correct but also introducing selection bias. Database studies rely on International Classification of Diseases, 10th Revision codes, and there isn’t one specifically for NAION — only for the broader category of ischemic optic neuropathy (H47.01), Rizzo told Medscape Medical News.

“There are ways to mitigate that effect, but they’re imperfect…so you may get different answers.” Rizzo previously conducted a similar large database study in people with T2D that also produced a twofold difference, although in a shorter follow-up time (6.7 months). Of the new study, he said, “It’s an active comparator study, and it’s beautiful. This SGLT2 approach is so good because GLP-1 and SGLT2 drugs tend to be used for patients with the same level of diabetes, so you’re minimizing residual confounding. I think it’s a great approach.”

Lee added that her team is looking at whether some patients “are at greater risk of NAION than others. To do this, we are refining our definition of NAION and evaluating the risks of other eye conditions that cause impaired vision. We are evaluating the risk of NAION related to other GLP-1s as well.”

Consider Ocular Risk Factors

“These [GLP-1] medicines have generally been really helpful for a lot of people. I never discourage patients from taking them,” said Rizzo.

“But if they’ve had visual loss for whatever reason…I really feel a moral responsibility to say, ‘Look, just so you’re aware, here’s what we found. The absolute risk is low, but I just want to inform you.’ Then they can make their own judgement about the level of risk.”

Lee noted that “clinicians should consider ocular risk factors such as preexisting optic nerve disease or prior NAION when prescribing.”

Rizzo extended that consideration to any patient who has experienced any type of vision loss in the past, such as due to glaucoma, diabetic retinopathy, or injury.

The study was conducted by the US Veterans Affairs with support from the VA Cooperative Studies Program. Lee reported having no further disclosures, and Rizzo reported having no relevant disclosures.

Miriam E. Tucker is a freelance journalist based in the Washington, DC, area. She is a regular contributor to Medscape, with other work appearing in the Washington Post, NPR’s Shots blog, and Diatribe. She is on X @MiriamETucker and BlueSky @miriametucker.bsky.social.

A version of this article first appeared on Medscape.com.

Semaglutide is linked to a small but significantly elevated risk for nonarteritic anterior ischemic optic neuropathy (NAION), according to the results of a new study.

The risk for NAION was approximately twofold greater among people taking semaglutide than among those using an SGLT2 inhibitor, but the absolute risk was still only about 29 per 10,000 people, study author Jennifer S. Lee, MD, PhD, professor of medicine at Stanford School of Medicine in Palo Alto, California, told Medscape Medical News.

In NAION, the optic nerve suddenly loses its blood supply, leading to vision loss, sometimes described as a “stroke of the eye.”

“Clinicians should balance this rare but serious vision-loss risk against semaglutide’s meaningful cardiometabolic benefits,” said Lee, who is also chief of Research and Clinical Innovations and director of the VA National Center for Collaborative Healthcare Innovation at the VA Palo Alto Healthcare System, US Department of Veterans Affairs.

“We recommend counseling patients to seek prompt evaluation if they experience visual symptoms, particularly sudden vision changes,” she added.

The new findings, recently published in JAMA Ophthalmology, are the latest from several studies examining a possible association between NAION and semaglutide, some of which have been negative.

In the current study, the investigators analyzed US Veterans Health Administration data for a total of 11,478 veterans with type 2 diabetes (T2D) who initiated semaglutide and 90,883 who initiated an SGLT2 inhibitor (mostly empagliflozin) between March 1, 2018, and March 1, 2025. Overlap weighting was used to balance the two groups by age, BMI, A1c, sex, and race.

Over a median follow-up of 2.1 years, NAION developed in 123 initiating semaglutide vs 143 initiating SGLT2 inhibitors, with rates of 123 vs 67 per 100,000 person-years. After overlap weighting, the risk for NAION was 2.3-fold higher with semaglutide (hazard ratio [HR], 2.33, P < .001).

Over a maximum of 7.5 years of follow-up, the overlap-weighted NAION incidence was 0.29% vs 0.13% for semaglutide initiators vs SGLT2 initiators.

While the mechanism(s) for the association remains unclear, hypotheses include hypotension, volume depletion from gastrointestinal side effects, rapid glucose improvement leading to stress on the eye microvasculature, and impaired vascular autoregulation at the head of the optic nerve, Lee said.

Different Studies, Different Answers

“Growing observational evidence links GLP-1 receptor agonists, particularly semaglutide, to NAION, with similar twofold to threefold risks reported in several large analyses,” Lee told Medscape Medical News.

“However, other studies have observed attenuated or null associations,” she said. “The mechanism remains unknown and requires further investigation. More research is needed to determine whether this is a class effect or specific to semaglutide.”

A possible reason that some other studies have not found an association may be due to methodological differences, particularly in how NAION cases are identified, noted Joseph F. Rizzo III, MD, professor of ophthalmology and director of the Neuro-ophthalmology Service at Harvard Medical School in Boston.

Rizzo and colleagues were the first to conduct a study looking at this association in their own neuro-ophthalmology population, thereby guaranteeing that the diagnosis was correct but also introducing selection bias. Database studies rely on International Classification of Diseases, 10th Revision codes, and there isn’t one specifically for NAION — only for the broader category of ischemic optic neuropathy (H47.01), Rizzo told Medscape Medical News.

“There are ways to mitigate that effect, but they’re imperfect…so you may get different answers.” Rizzo previously conducted a similar large database study in people with T2D that also produced a twofold difference, although in a shorter follow-up time (6.7 months). Of the new study, he said, “It’s an active comparator study, and it’s beautiful. This SGLT2 approach is so good because GLP-1 and SGLT2 drugs tend to be used for patients with the same level of diabetes, so you’re minimizing residual confounding. I think it’s a great approach.”

Lee added that her team is looking at whether some patients “are at greater risk of NAION than others. To do this, we are refining our definition of NAION and evaluating the risks of other eye conditions that cause impaired vision. We are evaluating the risk of NAION related to other GLP-1s as well.”

Consider Ocular Risk Factors

“These [GLP-1] medicines have generally been really helpful for a lot of people. I never discourage patients from taking them,” said Rizzo.

“But if they’ve had visual loss for whatever reason…I really feel a moral responsibility to say, ‘Look, just so you’re aware, here’s what we found. The absolute risk is low, but I just want to inform you.’ Then they can make their own judgement about the level of risk.”

Lee noted that “clinicians should consider ocular risk factors such as preexisting optic nerve disease or prior NAION when prescribing.”

Rizzo extended that consideration to any patient who has experienced any type of vision loss in the past, such as due to glaucoma, diabetic retinopathy, or injury.

The study was conducted by the US Veterans Affairs with support from the VA Cooperative Studies Program. Lee reported having no further disclosures, and Rizzo reported having no relevant disclosures.

Miriam E. Tucker is a freelance journalist based in the Washington, DC, area. She is a regular contributor to Medscape, with other work appearing in the Washington Post, NPR’s Shots blog, and Diatribe. She is on X @MiriamETucker and BlueSky @miriametucker.bsky.social.

A version of this article first appeared on Medscape.com.

Semaglutide is linked to a small but significantly elevated risk for nonarteritic anterior ischemic optic neuropathy (NAION), according to the results of a new study.

The risk for NAION was approximately twofold greater among people taking semaglutide than among those using an SGLT2 inhibitor, but the absolute risk was still only about 29 per 10,000 people, study author Jennifer S. Lee, MD, PhD, professor of medicine at Stanford School of Medicine in Palo Alto, California, told Medscape Medical News.

In NAION, the optic nerve suddenly loses its blood supply, leading to vision loss, sometimes described as a “stroke of the eye.”

“Clinicians should balance this rare but serious vision-loss risk against semaglutide’s meaningful cardiometabolic benefits,” said Lee, who is also chief of Research and Clinical Innovations and director of the VA National Center for Collaborative Healthcare Innovation at the VA Palo Alto Healthcare System, US Department of Veterans Affairs.

“We recommend counseling patients to seek prompt evaluation if they experience visual symptoms, particularly sudden vision changes,” she added.

The new findings, recently published in JAMA Ophthalmology, are the latest from several studies examining a possible association between NAION and semaglutide, some of which have been negative.

In the current study, the investigators analyzed US Veterans Health Administration data for a total of 11,478 veterans with type 2 diabetes (T2D) who initiated semaglutide and 90,883 who initiated an SGLT2 inhibitor (mostly empagliflozin) between March 1, 2018, and March 1, 2025. Overlap weighting was used to balance the two groups by age, BMI, A1c, sex, and race.

Over a median follow-up of 2.1 years, NAION developed in 123 initiating semaglutide vs 143 initiating SGLT2 inhibitors, with rates of 123 vs 67 per 100,000 person-years. After overlap weighting, the risk for NAION was 2.3-fold higher with semaglutide (hazard ratio [HR], 2.33, P < .001).

Over a maximum of 7.5 years of follow-up, the overlap-weighted NAION incidence was 0.29% vs 0.13% for semaglutide initiators vs SGLT2 initiators.

While the mechanism(s) for the association remains unclear, hypotheses include hypotension, volume depletion from gastrointestinal side effects, rapid glucose improvement leading to stress on the eye microvasculature, and impaired vascular autoregulation at the head of the optic nerve, Lee said.

Different Studies, Different Answers

“Growing observational evidence links GLP-1 receptor agonists, particularly semaglutide, to NAION, with similar twofold to threefold risks reported in several large analyses,” Lee told Medscape Medical News.

“However, other studies have observed attenuated or null associations,” she said. “The mechanism remains unknown and requires further investigation. More research is needed to determine whether this is a class effect or specific to semaglutide.”

A possible reason that some other studies have not found an association may be due to methodological differences, particularly in how NAION cases are identified, noted Joseph F. Rizzo III, MD, professor of ophthalmology and director of the Neuro-ophthalmology Service at Harvard Medical School in Boston.

Rizzo and colleagues were the first to conduct a study looking at this association in their own neuro-ophthalmology population, thereby guaranteeing that the diagnosis was correct but also introducing selection bias. Database studies rely on International Classification of Diseases, 10th Revision codes, and there isn’t one specifically for NAION — only for the broader category of ischemic optic neuropathy (H47.01), Rizzo told Medscape Medical News.

“There are ways to mitigate that effect, but they’re imperfect…so you may get different answers.” Rizzo previously conducted a similar large database study in people with T2D that also produced a twofold difference, although in a shorter follow-up time (6.7 months). Of the new study, he said, “It’s an active comparator study, and it’s beautiful. This SGLT2 approach is so good because GLP-1 and SGLT2 drugs tend to be used for patients with the same level of diabetes, so you’re minimizing residual confounding. I think it’s a great approach.”

Lee added that her team is looking at whether some patients “are at greater risk of NAION than others. To do this, we are refining our definition of NAION and evaluating the risks of other eye conditions that cause impaired vision. We are evaluating the risk of NAION related to other GLP-1s as well.”

Consider Ocular Risk Factors

“These [GLP-1] medicines have generally been really helpful for a lot of people. I never discourage patients from taking them,” said Rizzo.

“But if they’ve had visual loss for whatever reason…I really feel a moral responsibility to say, ‘Look, just so you’re aware, here’s what we found. The absolute risk is low, but I just want to inform you.’ Then they can make their own judgement about the level of risk.”

Lee noted that “clinicians should consider ocular risk factors such as preexisting optic nerve disease or prior NAION when prescribing.”

Rizzo extended that consideration to any patient who has experienced any type of vision loss in the past, such as due to glaucoma, diabetic retinopathy, or injury.

The study was conducted by the US Veterans Affairs with support from the VA Cooperative Studies Program. Lee reported having no further disclosures, and Rizzo reported having no relevant disclosures.

Miriam E. Tucker is a freelance journalist based in the Washington, DC, area. She is a regular contributor to Medscape, with other work appearing in the Washington Post, NPR’s Shots blog, and Diatribe. She is on X @MiriamETucker and BlueSky @miriametucker.bsky.social.

A version of this article first appeared on Medscape.com.

More than 38 million people in the United States (12%) have diabetes mellitus (DM), though 1 in 5 are unaware they have DM.¹ The prevalence among veterans is even more substantial, impacting nearly 25% of those who received care from the US Department of Veterans Affairs (VA).² DM can lead to increased health care costs in addition to various complications (eg, cardiovascular, renal), especially if left uncontrolled.^1,3 similar impact is found in the perioperative period (defined as at or around the time of an operation), as multiple studies have found that uncontrolled preoperative DM can result in worsened surgical outcomes, including longer hospital stays, more infectious complications, and higher perioperative mortality.^4-6

In contrast, adequate glycemic control assessed with blood glucose levels has been shown to decrease the incidence of postoperative infections.⁷ Optimizing glycemic control during hospital stays, especially postsurgery, has become the standard of care, with most health systems establishing specific protocols. In current literature, most studies examining DM management in the perioperative period are focused on postoperative care, with little attention to the preoperative period.^4,6,7

One study found that patients with poor presurgery glycemic control assessed by hemoglobin A_1c (HbA_1c) levels were more likely to remain hyperglycemic during and after surgery. 8 Blood glucose levels < 200 mg/dL can lead to an increased risk of infection and impaired wound healing, meaning a well-controlled HbA_1c before a procedure serves as a potential factor for success.⁹ The 2025 American Diabetes Association (ADA) Standards of Care (SOC) recommendation is to target HbA_1c < 8% whenever possible, and some health systems require lower levels (eg, < 7% or 7.5%).¹⁰ With that goal in mind and knowing that preoperative hyperglycemia has been shown to be a contributing factor in the delay or cancellation of surgical cases, an argument can be made that attention to preoperative DM management also should be a focus for health care systems performing surgeries.^8,9,11

Attention to glucose control during preoperative care offers an opportunity to screen for DM in patients who may not have been screened otherwise and to standardize perioperative DM management. Since DM disproportionately impacts veterans, this is a pertinent issue to the VA. Veterans can be more susceptible to complications if DM is left uncontrolled prior to surgery. To determine readiness for surgery and control of comorbid conditions such as DM before a planned surgery, facilities often perform a preoperative clinic assessment, often in a multidisciplinary clinic.

At Veteran Health Indiana (VHI), a presurgery clinic visit involving the primary surgery service (physician, nurse practitioner, and/or a physician assistant) is conducted 1 to 2 months prior to the planned procedure to determine whether a patient is ready for surgery. During this visit, patients receive a packet with instructions for various tasks and medications, such as applying topical antibiotic prophylaxis on the anticipated surgical site. This is documented in the form of a note in the VHI Computerized Patient Record System (CPRS). The medication instructions are provided according to the preferences of the surgical team. These may be templated notes that contain general directions on the timing and dosing of specific medications, in addition to instructions for holding or reducing doses when appropriate. The instructions can be tailored by the team conducting the preoperative visit (eg, “Take 20 units of insulin glargine the day before surgery” vs “Take half of your long-acting insulin the night before surgery”). Specific to DM, VHI has a nurse-driven day of surgery glucose assessment where point-of-care blood glucose is collected during preoperative holding for most patients.

There is limited research assessing the level of preoperative glycemic control and the incidence of complications in a veteran population. The objective of this study was to gain a baseline understanding of what, if any, standardization exists for preoperative instructions for DM medications and to assess the level of preoperative glycemic control and postoperative complications in patients with DM undergoing major elective surgical procedures.

Methods

This retrospective, single-center chart review was conducted at VHI. The Indiana University and VHI institutional review boards determined that this quality improvement project was exempt from review.

The primary outcome was the number of patients with surgical procedures delayed or canceled due to hyperglycemia or hypoglycemia. Hyperglycemia was defined as blood glucose > 180 mg/dL and hypoglycemia was defined as < 70 mg/dL, slight variations from the current ADA SOC preoperative specific recommendation of a blood glucose reading of 100 to 180 mg/dL within 4 hours of surgery.¹⁰ The standard outpatient hypoglycemia definition of blood glucose < 70 mg/dL was chosen because the current goal (< 100 mg/dL) was not the standard in previous ADA SOCs that were in place during the study period. Specifically, the 2018 ADA SOC did not provide preoperative recommendations and the 2019-2021 ADA SOC recommended 80 to 180 mg/dL.^10,12-18 For patients who had multiple preoperative blood glucose measurements, the first recorded glucose on the day of the procedure was used.

The secondary outcomes of this study were focused on the preoperative process/care at VHI and postoperative glycemic control. The preoperative process included examining whether medication instructions were given and their quality. Additionally, the number of interventions for hyperglycemia and hypoglycemia were required immediately prior to surgery and the average preoperative HbA_1c (measured within 3 months prior to surgery) were collected and analyzed. For postoperative glycemic control, average blood glucose measurements and number of hypoglycemic (< 70 mg/dL) and hyperglycemic (> 180 mg/dL) events were measured in addition to the frequency of changes made at discharge to patients’ DM medication regimens.

The safety outcome of this study assessed commonly observed postoperative complications and was examined up to 30 days postsurgery. These included acute kidney injury (defined using Kidney Disease: Improving Global Outcomes 2012, the standard during the study period), nonfatal myocardial infarction, nonfatal stroke, and surgical site infections, which were identified from the discharge summary written by the primary surgery service.¹⁹ All-cause mortality also was collected.

Patients were included if they were admitted for major elective surgeries and had a diagnosis of either type 1 or type 2 DM on their problem list, determined by International Classification of Diseases, Tenth Revision codes. Major elective surgery was defined as a procedure that would likely result in a hospital admission of > 24 hours. Of note, patients may have been included in this study more than once if they had > 1 procedure at least 30 days apart and met inclusion criteria within the time frame. Patients were excluded if they were taking no DM medications or chronic steroids (at any dose), residing in a long-term care facility, being managed by a non-VA clinician prior to surgery, or missing a preoperative blood glucose measurement.

All data were collected from the CPRS. A list of surgical cases involving patients with DM who were scheduled to undergo major elective surgeries from January 1, 2018, to December 31, 2021, at VHI was generated. The list was randomized to a smaller number (N = 394) for data collection due to the time and resource constraints for a pharmacy residency project. All data were deidentified and stored in a secured VA server to protect patient confidentiality. Descriptive statistics were used for all results.

Results

Initially, 2362 surgeries were identified. A randomized sample of 394 charts were reviewed and 131 cases met inclusion criteria. Each case involved a unique patient (Figure). The most common reasons for exclusion were 143 patients with diet-controlled DM and 78 nonelective surgeries. The mean (SD) age of patients was 68 (8) years, and the most were male (98.5%) and White (76.3%) (Table 1). 

At baseline, 45 of 131 patients (34.4%) had coronary artery disease and 29 (22.1%) each had autonomic neuropathy and chronic kidney disease. Most surgeries were conducted by orthopedic (32.1%) and peripheral vascular (21.4%) specialties. The mean (SD) length of surgery was 4.6 (2.6) hours and of hospital length of stay was 4 (4) days. No patients stayed longer than the 30-day safety outcome follow-up period. All patients had type 2 DM and took a mean 2 DM medications. The 63 patients taking insulin had a mean (SD) total daily dose of 99 (77) U (Table 2). A preoperative HbA1c was collected in 116 patients within 3 months of surgery, with a mean HbA_1c of 7.0% (range, 5.3-10.7).

No patients had surgeries delayed or canceled because of uncontrolled DM on the day of surgery. The mean preoperative blood glucose level was 146 mg/dL (range, 73-365) (Table 3). No patients had a preoperative blood glucose level of < 70 mg/dL and 19 (14.5%) had a blood glucose level > 180 mg/dL. Among patients with hyperglycemia immediately prior to surgery, 6 (31.6%) had documentation of insulin being provided.

For this sample of patients, the preoperative clinic visit was conducted a mean 22 days prior to the planned surgery date. Among the 131 included patients, 122 (93.1%) had documentation of receiving instructions for DM medications. Among patients who had documented receipt of instructions, only 30 (24.6%) had instructions specifically tailored to their regimen rather than a generic templated form. The mean (SD) preoperative blood glucose was similar for those who received specific perioperative DM instructions at 146 (50) mg/dL when compared with those who did not at 147 (45) mg/dL. The mean (SD) preoperative blood glucose reading for those who had no documentation of receipt of perioperative instructions was 126 (54) mg/dL compared with 147 (46) mg/dL for those who did.

The mean number of postoperative blood glucose events per day was negligible for hypoglycemia and more frequent for hyperglycemia with a mean of 2 events per day. The mean postoperative blood glucose range was 121 to 247 mg/dL with most readings < 180 mg/dL. Upon discharge, most patients continued their home DM regimen with 5 patients (3.8%) having changes made to their regimen upon discharge.

Very few postoperative complications were identified from chart review. The most frequently observed postoperative complications were acute kidney injury, surgical site infections, and nonfatal stroke. There were no documented nonfatal myocardial infarctions. Two patients (1.5%) died within 30 days of the surgery; neither death was deemed to have been related to poor perioperative glycemic control.

Discussion

To our knowledge, this retrospective chart review was the first study to assess preoperative DM management and postoperative complications in a veteran population. VHI is a large, tertiary, level 1a, academic medical center that serves approximately 62,000 veterans annually and performs about 5000 to 6000 surgeries annually, a total that is increasing following the COVID-19 pandemic.²⁰ This study found that the current process of a presurgery clinic visit and day of surgery glucose assessment has prevented surgical delays or cancellations.

Most patients included in this study were well controlled at baseline in accordance with the 2025 ADA SOC HbA_1c recommendation of a preoperative HbA1c of < 8%, which may have contributed to no surgical delays or cancellations.¹⁰ However, not all patients had HbA1c collected within 3 months of surgery or even had one collected at all. Despite the ADA SOC providing no explicit recommendation for universal HbA_1c screening prior to elective procedures, its importance cannot be understated given the body of evidence demonstrating poor outcomes with uncontrolled preoperative DM.^8,10 The glycemic control at baseline may have contributed to the very few postsurgical complications observed in this study.

Although the current process at VHI prevented surgical delays and cancellations in this sample, there are still identified areas for improvement. One area is the instructions the patients received. Patients with DM are often prescribed ≥ 1 medication or a combination of insulins, noninsulin injectables, and oral DM medications, and this study population was no different. Because these medications may influence the anesthesia and perioperative periods, the ADA has specific guidance for altering administration schedules in the days leading up to surgery.¹⁰

Inappropriate administration of DM medications could lead to perioperative hypoglycemia or hyperglycemia, possibly causing surgical delays, case cancellations, and/or postoperative complications.21 Although these data reveal the specificity and documented receipt that the preoperative DM instructions did not impact the first recorded preoperative blood glucose, future studies should examine patient confidence in how to properly administer their DM medications prior to surgery. It is vital that patients receive clear instructions in accordance with the ADA SOC on whether to continue, hold, or adjust the dose of their medications to prevent fluctuations in blood glucose levels in the perioperative period, ensure safety with anesthesia, and prevent postoperative complications such as acute kidney injury. Of note, compliance with guideline recommendations for medication instructions was not examined because the data collection time frame expanded over multiple years and the recommendations have evolved each year as new data emerge.

Preoperative DM Management

The first key takeaway from this study is to ensure patients are ready for surgery with a formal assessment (typically in the form of a clinic visit) prior to the surgery. One private sector health system published their approach to this by administering an automatic preoperative HbA_1c screening for those with a DM diagnosis and all patients with a random plasma glucose ≥ 200 mg/dL.²² Additionally, if the patient's HbA_1c level was not at goal prior to surgery (≥ 8% for those with known DM and ≥ 6.5% with no known DM), patients were referred to endocrinology for further management. Increasing attention to the preoperative visit and extending HbA_1c testing to all patients regardless of DM status also provides an opportunity to identify individuals living with undiagnosed DM.¹

Even though there was no difference in the mean preoperative blood glucose level based on receipt or specificity of preoperative DM instructions, a second takeaway from this study is the importance of ensuring patients receive clear instructions on their DM medication schedule in the perioperative period. A practical first step may be updating the templates used by the primary surgery teams and providing education to the clinicians in the clinic on how to personalize the visits. Because the current preoperative DM process at VHI is managed by the primary surgical team in a clinic visit, there is an opportunity to shift this responsibility to other health care professionals, such as pharmacists—a change shown to reduce unintended omission of home medications following surgery during hospitalization and reduce costs.^23,24

Limitations

This study relied on data included in the patient chart. These data include medication interventions made immediately prior to surgery, which can sometimes be inaccurately charted or difficult to find as they are not documented in the typical medication administration record. Also, the safety outcomes were collected from a discharge summary written by different clinicians, which may lead to information bias. Special attention was taken to ensure these data points were collected as accurately as possible, but it is possible some data may be inaccurate from unintentional human error. Additionally, the safety outcome was limited to a 30-day follow-up, but encompassed the entire length of postoperative stay for all included patients. Finally, given this study was retrospective with no comparison group and the intent was to improve processes at VHI, only hypotheses and potential interventions can be generated from this study. Future prospective studies with larger sample sizes and comparator groups are needed to draw further conclusions.

Conclusions

This study found that the current presurgery process at VHI appears to be successful in preventing surgical delays or cancellations due to hyperglycemia or hypoglycemia. Optimizing DM management can improve surgical outcomes by decreasing rates of postoperative complications, and this study added additional evidence in support of that in a unique population: veterans. Insight on the awareness of preoperative blood glucose management should be gleaned from this study, and based on this sample and site, the preadmission screening process and instructions provided to patients can serve as 2 starting points for optimizing elective surgery.

More than 38 million people in the United States (12%) have diabetes mellitus (DM), though 1 in 5 are unaware they have DM.¹ The prevalence among veterans is even more substantial, impacting nearly 25% of those who received care from the US Department of Veterans Affairs (VA).² DM can lead to increased health care costs in addition to various complications (eg, cardiovascular, renal), especially if left uncontrolled.^1,3 similar impact is found in the perioperative period (defined as at or around the time of an operation), as multiple studies have found that uncontrolled preoperative DM can result in worsened surgical outcomes, including longer hospital stays, more infectious complications, and higher perioperative mortality.^4-6

In contrast, adequate glycemic control assessed with blood glucose levels has been shown to decrease the incidence of postoperative infections.⁷ Optimizing glycemic control during hospital stays, especially postsurgery, has become the standard of care, with most health systems establishing specific protocols. In current literature, most studies examining DM management in the perioperative period are focused on postoperative care, with little attention to the preoperative period.^4,6,7

One study found that patients with poor presurgery glycemic control assessed by hemoglobin A_1c (HbA_1c) levels were more likely to remain hyperglycemic during and after surgery. 8 Blood glucose levels < 200 mg/dL can lead to an increased risk of infection and impaired wound healing, meaning a well-controlled HbA_1c before a procedure serves as a potential factor for success.⁹ The 2025 American Diabetes Association (ADA) Standards of Care (SOC) recommendation is to target HbA_1c < 8% whenever possible, and some health systems require lower levels (eg, < 7% or 7.5%).¹⁰ With that goal in mind and knowing that preoperative hyperglycemia has been shown to be a contributing factor in the delay or cancellation of surgical cases, an argument can be made that attention to preoperative DM management also should be a focus for health care systems performing surgeries.^8,9,11

Attention to glucose control during preoperative care offers an opportunity to screen for DM in patients who may not have been screened otherwise and to standardize perioperative DM management. Since DM disproportionately impacts veterans, this is a pertinent issue to the VA. Veterans can be more susceptible to complications if DM is left uncontrolled prior to surgery. To determine readiness for surgery and control of comorbid conditions such as DM before a planned surgery, facilities often perform a preoperative clinic assessment, often in a multidisciplinary clinic.

At Veteran Health Indiana (VHI), a presurgery clinic visit involving the primary surgery service (physician, nurse practitioner, and/or a physician assistant) is conducted 1 to 2 months prior to the planned procedure to determine whether a patient is ready for surgery. During this visit, patients receive a packet with instructions for various tasks and medications, such as applying topical antibiotic prophylaxis on the anticipated surgical site. This is documented in the form of a note in the VHI Computerized Patient Record System (CPRS). The medication instructions are provided according to the preferences of the surgical team. These may be templated notes that contain general directions on the timing and dosing of specific medications, in addition to instructions for holding or reducing doses when appropriate. The instructions can be tailored by the team conducting the preoperative visit (eg, “Take 20 units of insulin glargine the day before surgery” vs “Take half of your long-acting insulin the night before surgery”). Specific to DM, VHI has a nurse-driven day of surgery glucose assessment where point-of-care blood glucose is collected during preoperative holding for most patients.

There is limited research assessing the level of preoperative glycemic control and the incidence of complications in a veteran population. The objective of this study was to gain a baseline understanding of what, if any, standardization exists for preoperative instructions for DM medications and to assess the level of preoperative glycemic control and postoperative complications in patients with DM undergoing major elective surgical procedures.

Methods

This retrospective, single-center chart review was conducted at VHI. The Indiana University and VHI institutional review boards determined that this quality improvement project was exempt from review.

The primary outcome was the number of patients with surgical procedures delayed or canceled due to hyperglycemia or hypoglycemia. Hyperglycemia was defined as blood glucose > 180 mg/dL and hypoglycemia was defined as < 70 mg/dL, slight variations from the current ADA SOC preoperative specific recommendation of a blood glucose reading of 100 to 180 mg/dL within 4 hours of surgery.¹⁰ The standard outpatient hypoglycemia definition of blood glucose < 70 mg/dL was chosen because the current goal (< 100 mg/dL) was not the standard in previous ADA SOCs that were in place during the study period. Specifically, the 2018 ADA SOC did not provide preoperative recommendations and the 2019-2021 ADA SOC recommended 80 to 180 mg/dL.^10,12-18 For patients who had multiple preoperative blood glucose measurements, the first recorded glucose on the day of the procedure was used.

The secondary outcomes of this study were focused on the preoperative process/care at VHI and postoperative glycemic control. The preoperative process included examining whether medication instructions were given and their quality. Additionally, the number of interventions for hyperglycemia and hypoglycemia were required immediately prior to surgery and the average preoperative HbA_1c (measured within 3 months prior to surgery) were collected and analyzed. For postoperative glycemic control, average blood glucose measurements and number of hypoglycemic (< 70 mg/dL) and hyperglycemic (> 180 mg/dL) events were measured in addition to the frequency of changes made at discharge to patients’ DM medication regimens.

The safety outcome of this study assessed commonly observed postoperative complications and was examined up to 30 days postsurgery. These included acute kidney injury (defined using Kidney Disease: Improving Global Outcomes 2012, the standard during the study period), nonfatal myocardial infarction, nonfatal stroke, and surgical site infections, which were identified from the discharge summary written by the primary surgery service.¹⁹ All-cause mortality also was collected.

Patients were included if they were admitted for major elective surgeries and had a diagnosis of either type 1 or type 2 DM on their problem list, determined by International Classification of Diseases, Tenth Revision codes. Major elective surgery was defined as a procedure that would likely result in a hospital admission of > 24 hours. Of note, patients may have been included in this study more than once if they had > 1 procedure at least 30 days apart and met inclusion criteria within the time frame. Patients were excluded if they were taking no DM medications or chronic steroids (at any dose), residing in a long-term care facility, being managed by a non-VA clinician prior to surgery, or missing a preoperative blood glucose measurement.

All data were collected from the CPRS. A list of surgical cases involving patients with DM who were scheduled to undergo major elective surgeries from January 1, 2018, to December 31, 2021, at VHI was generated. The list was randomized to a smaller number (N = 394) for data collection due to the time and resource constraints for a pharmacy residency project. All data were deidentified and stored in a secured VA server to protect patient confidentiality. Descriptive statistics were used for all results.

Results

Initially, 2362 surgeries were identified. A randomized sample of 394 charts were reviewed and 131 cases met inclusion criteria. Each case involved a unique patient (Figure). The most common reasons for exclusion were 143 patients with diet-controlled DM and 78 nonelective surgeries. The mean (SD) age of patients was 68 (8) years, and the most were male (98.5%) and White (76.3%) (Table 1). 

At baseline, 45 of 131 patients (34.4%) had coronary artery disease and 29 (22.1%) each had autonomic neuropathy and chronic kidney disease. Most surgeries were conducted by orthopedic (32.1%) and peripheral vascular (21.4%) specialties. The mean (SD) length of surgery was 4.6 (2.6) hours and of hospital length of stay was 4 (4) days. No patients stayed longer than the 30-day safety outcome follow-up period. All patients had type 2 DM and took a mean 2 DM medications. The 63 patients taking insulin had a mean (SD) total daily dose of 99 (77) U (Table 2). A preoperative HbA1c was collected in 116 patients within 3 months of surgery, with a mean HbA_1c of 7.0% (range, 5.3-10.7).

No patients had surgeries delayed or canceled because of uncontrolled DM on the day of surgery. The mean preoperative blood glucose level was 146 mg/dL (range, 73-365) (Table 3). No patients had a preoperative blood glucose level of < 70 mg/dL and 19 (14.5%) had a blood glucose level > 180 mg/dL. Among patients with hyperglycemia immediately prior to surgery, 6 (31.6%) had documentation of insulin being provided.

For this sample of patients, the preoperative clinic visit was conducted a mean 22 days prior to the planned surgery date. Among the 131 included patients, 122 (93.1%) had documentation of receiving instructions for DM medications. Among patients who had documented receipt of instructions, only 30 (24.6%) had instructions specifically tailored to their regimen rather than a generic templated form. The mean (SD) preoperative blood glucose was similar for those who received specific perioperative DM instructions at 146 (50) mg/dL when compared with those who did not at 147 (45) mg/dL. The mean (SD) preoperative blood glucose reading for those who had no documentation of receipt of perioperative instructions was 126 (54) mg/dL compared with 147 (46) mg/dL for those who did.

The mean number of postoperative blood glucose events per day was negligible for hypoglycemia and more frequent for hyperglycemia with a mean of 2 events per day. The mean postoperative blood glucose range was 121 to 247 mg/dL with most readings < 180 mg/dL. Upon discharge, most patients continued their home DM regimen with 5 patients (3.8%) having changes made to their regimen upon discharge.

Very few postoperative complications were identified from chart review. The most frequently observed postoperative complications were acute kidney injury, surgical site infections, and nonfatal stroke. There were no documented nonfatal myocardial infarctions. Two patients (1.5%) died within 30 days of the surgery; neither death was deemed to have been related to poor perioperative glycemic control.

Discussion

To our knowledge, this retrospective chart review was the first study to assess preoperative DM management and postoperative complications in a veteran population. VHI is a large, tertiary, level 1a, academic medical center that serves approximately 62,000 veterans annually and performs about 5000 to 6000 surgeries annually, a total that is increasing following the COVID-19 pandemic.²⁰ This study found that the current process of a presurgery clinic visit and day of surgery glucose assessment has prevented surgical delays or cancellations.

Most patients included in this study were well controlled at baseline in accordance with the 2025 ADA SOC HbA_1c recommendation of a preoperative HbA1c of < 8%, which may have contributed to no surgical delays or cancellations.¹⁰ However, not all patients had HbA1c collected within 3 months of surgery or even had one collected at all. Despite the ADA SOC providing no explicit recommendation for universal HbA_1c screening prior to elective procedures, its importance cannot be understated given the body of evidence demonstrating poor outcomes with uncontrolled preoperative DM.^8,10 The glycemic control at baseline may have contributed to the very few postsurgical complications observed in this study.

Although the current process at VHI prevented surgical delays and cancellations in this sample, there are still identified areas for improvement. One area is the instructions the patients received. Patients with DM are often prescribed ≥ 1 medication or a combination of insulins, noninsulin injectables, and oral DM medications, and this study population was no different. Because these medications may influence the anesthesia and perioperative periods, the ADA has specific guidance for altering administration schedules in the days leading up to surgery.¹⁰

Inappropriate administration of DM medications could lead to perioperative hypoglycemia or hyperglycemia, possibly causing surgical delays, case cancellations, and/or postoperative complications.21 Although these data reveal the specificity and documented receipt that the preoperative DM instructions did not impact the first recorded preoperative blood glucose, future studies should examine patient confidence in how to properly administer their DM medications prior to surgery. It is vital that patients receive clear instructions in accordance with the ADA SOC on whether to continue, hold, or adjust the dose of their medications to prevent fluctuations in blood glucose levels in the perioperative period, ensure safety with anesthesia, and prevent postoperative complications such as acute kidney injury. Of note, compliance with guideline recommendations for medication instructions was not examined because the data collection time frame expanded over multiple years and the recommendations have evolved each year as new data emerge.

Preoperative DM Management

The first key takeaway from this study is to ensure patients are ready for surgery with a formal assessment (typically in the form of a clinic visit) prior to the surgery. One private sector health system published their approach to this by administering an automatic preoperative HbA_1c screening for those with a DM diagnosis and all patients with a random plasma glucose ≥ 200 mg/dL.²² Additionally, if the patient's HbA_1c level was not at goal prior to surgery (≥ 8% for those with known DM and ≥ 6.5% with no known DM), patients were referred to endocrinology for further management. Increasing attention to the preoperative visit and extending HbA_1c testing to all patients regardless of DM status also provides an opportunity to identify individuals living with undiagnosed DM.¹

Even though there was no difference in the mean preoperative blood glucose level based on receipt or specificity of preoperative DM instructions, a second takeaway from this study is the importance of ensuring patients receive clear instructions on their DM medication schedule in the perioperative period. A practical first step may be updating the templates used by the primary surgery teams and providing education to the clinicians in the clinic on how to personalize the visits. Because the current preoperative DM process at VHI is managed by the primary surgical team in a clinic visit, there is an opportunity to shift this responsibility to other health care professionals, such as pharmacists—a change shown to reduce unintended omission of home medications following surgery during hospitalization and reduce costs.^23,24

Limitations

This study relied on data included in the patient chart. These data include medication interventions made immediately prior to surgery, which can sometimes be inaccurately charted or difficult to find as they are not documented in the typical medication administration record. Also, the safety outcomes were collected from a discharge summary written by different clinicians, which may lead to information bias. Special attention was taken to ensure these data points were collected as accurately as possible, but it is possible some data may be inaccurate from unintentional human error. Additionally, the safety outcome was limited to a 30-day follow-up, but encompassed the entire length of postoperative stay for all included patients. Finally, given this study was retrospective with no comparison group and the intent was to improve processes at VHI, only hypotheses and potential interventions can be generated from this study. Future prospective studies with larger sample sizes and comparator groups are needed to draw further conclusions.

Conclusions

This study found that the current presurgery process at VHI appears to be successful in preventing surgical delays or cancellations due to hyperglycemia or hypoglycemia. Optimizing DM management can improve surgical outcomes by decreasing rates of postoperative complications, and this study added additional evidence in support of that in a unique population: veterans. Insight on the awareness of preoperative blood glucose management should be gleaned from this study, and based on this sample and site, the preadmission screening process and instructions provided to patients can serve as 2 starting points for optimizing elective surgery.

More than 38 million people in the United States (12%) have diabetes mellitus (DM), though 1 in 5 are unaware they have DM.¹ The prevalence among veterans is even more substantial, impacting nearly 25% of those who received care from the US Department of Veterans Affairs (VA).² DM can lead to increased health care costs in addition to various complications (eg, cardiovascular, renal), especially if left uncontrolled.^1,3 similar impact is found in the perioperative period (defined as at or around the time of an operation), as multiple studies have found that uncontrolled preoperative DM can result in worsened surgical outcomes, including longer hospital stays, more infectious complications, and higher perioperative mortality.^4-6

In contrast, adequate glycemic control assessed with blood glucose levels has been shown to decrease the incidence of postoperative infections.⁷ Optimizing glycemic control during hospital stays, especially postsurgery, has become the standard of care, with most health systems establishing specific protocols. In current literature, most studies examining DM management in the perioperative period are focused on postoperative care, with little attention to the preoperative period.^4,6,7

One study found that patients with poor presurgery glycemic control assessed by hemoglobin A_1c (HbA_1c) levels were more likely to remain hyperglycemic during and after surgery. 8 Blood glucose levels < 200 mg/dL can lead to an increased risk of infection and impaired wound healing, meaning a well-controlled HbA_1c before a procedure serves as a potential factor for success.⁹ The 2025 American Diabetes Association (ADA) Standards of Care (SOC) recommendation is to target HbA_1c < 8% whenever possible, and some health systems require lower levels (eg, < 7% or 7.5%).¹⁰ With that goal in mind and knowing that preoperative hyperglycemia has been shown to be a contributing factor in the delay or cancellation of surgical cases, an argument can be made that attention to preoperative DM management also should be a focus for health care systems performing surgeries.^8,9,11

Attention to glucose control during preoperative care offers an opportunity to screen for DM in patients who may not have been screened otherwise and to standardize perioperative DM management. Since DM disproportionately impacts veterans, this is a pertinent issue to the VA. Veterans can be more susceptible to complications if DM is left uncontrolled prior to surgery. To determine readiness for surgery and control of comorbid conditions such as DM before a planned surgery, facilities often perform a preoperative clinic assessment, often in a multidisciplinary clinic.

At Veteran Health Indiana (VHI), a presurgery clinic visit involving the primary surgery service (physician, nurse practitioner, and/or a physician assistant) is conducted 1 to 2 months prior to the planned procedure to determine whether a patient is ready for surgery. During this visit, patients receive a packet with instructions for various tasks and medications, such as applying topical antibiotic prophylaxis on the anticipated surgical site. This is documented in the form of a note in the VHI Computerized Patient Record System (CPRS). The medication instructions are provided according to the preferences of the surgical team. These may be templated notes that contain general directions on the timing and dosing of specific medications, in addition to instructions for holding or reducing doses when appropriate. The instructions can be tailored by the team conducting the preoperative visit (eg, “Take 20 units of insulin glargine the day before surgery” vs “Take half of your long-acting insulin the night before surgery”). Specific to DM, VHI has a nurse-driven day of surgery glucose assessment where point-of-care blood glucose is collected during preoperative holding for most patients.

There is limited research assessing the level of preoperative glycemic control and the incidence of complications in a veteran population. The objective of this study was to gain a baseline understanding of what, if any, standardization exists for preoperative instructions for DM medications and to assess the level of preoperative glycemic control and postoperative complications in patients with DM undergoing major elective surgical procedures.

Methods

This retrospective, single-center chart review was conducted at VHI. The Indiana University and VHI institutional review boards determined that this quality improvement project was exempt from review.

The primary outcome was the number of patients with surgical procedures delayed or canceled due to hyperglycemia or hypoglycemia. Hyperglycemia was defined as blood glucose > 180 mg/dL and hypoglycemia was defined as < 70 mg/dL, slight variations from the current ADA SOC preoperative specific recommendation of a blood glucose reading of 100 to 180 mg/dL within 4 hours of surgery.¹⁰ The standard outpatient hypoglycemia definition of blood glucose < 70 mg/dL was chosen because the current goal (< 100 mg/dL) was not the standard in previous ADA SOCs that were in place during the study period. Specifically, the 2018 ADA SOC did not provide preoperative recommendations and the 2019-2021 ADA SOC recommended 80 to 180 mg/dL.^10,12-18 For patients who had multiple preoperative blood glucose measurements, the first recorded glucose on the day of the procedure was used.

The secondary outcomes of this study were focused on the preoperative process/care at VHI and postoperative glycemic control. The preoperative process included examining whether medication instructions were given and their quality. Additionally, the number of interventions for hyperglycemia and hypoglycemia were required immediately prior to surgery and the average preoperative HbA_1c (measured within 3 months prior to surgery) were collected and analyzed. For postoperative glycemic control, average blood glucose measurements and number of hypoglycemic (< 70 mg/dL) and hyperglycemic (> 180 mg/dL) events were measured in addition to the frequency of changes made at discharge to patients’ DM medication regimens.

The safety outcome of this study assessed commonly observed postoperative complications and was examined up to 30 days postsurgery. These included acute kidney injury (defined using Kidney Disease: Improving Global Outcomes 2012, the standard during the study period), nonfatal myocardial infarction, nonfatal stroke, and surgical site infections, which were identified from the discharge summary written by the primary surgery service.¹⁹ All-cause mortality also was collected.

Patients were included if they were admitted for major elective surgeries and had a diagnosis of either type 1 or type 2 DM on their problem list, determined by International Classification of Diseases, Tenth Revision codes. Major elective surgery was defined as a procedure that would likely result in a hospital admission of > 24 hours. Of note, patients may have been included in this study more than once if they had > 1 procedure at least 30 days apart and met inclusion criteria within the time frame. Patients were excluded if they were taking no DM medications or chronic steroids (at any dose), residing in a long-term care facility, being managed by a non-VA clinician prior to surgery, or missing a preoperative blood glucose measurement.

All data were collected from the CPRS. A list of surgical cases involving patients with DM who were scheduled to undergo major elective surgeries from January 1, 2018, to December 31, 2021, at VHI was generated. The list was randomized to a smaller number (N = 394) for data collection due to the time and resource constraints for a pharmacy residency project. All data were deidentified and stored in a secured VA server to protect patient confidentiality. Descriptive statistics were used for all results.

Results

Initially, 2362 surgeries were identified. A randomized sample of 394 charts were reviewed and 131 cases met inclusion criteria. Each case involved a unique patient (Figure). The most common reasons for exclusion were 143 patients with diet-controlled DM and 78 nonelective surgeries. The mean (SD) age of patients was 68 (8) years, and the most were male (98.5%) and White (76.3%) (Table 1). 

At baseline, 45 of 131 patients (34.4%) had coronary artery disease and 29 (22.1%) each had autonomic neuropathy and chronic kidney disease. Most surgeries were conducted by orthopedic (32.1%) and peripheral vascular (21.4%) specialties. The mean (SD) length of surgery was 4.6 (2.6) hours and of hospital length of stay was 4 (4) days. No patients stayed longer than the 30-day safety outcome follow-up period. All patients had type 2 DM and took a mean 2 DM medications. The 63 patients taking insulin had a mean (SD) total daily dose of 99 (77) U (Table 2). A preoperative HbA1c was collected in 116 patients within 3 months of surgery, with a mean HbA_1c of 7.0% (range, 5.3-10.7).

No patients had surgeries delayed or canceled because of uncontrolled DM on the day of surgery. The mean preoperative blood glucose level was 146 mg/dL (range, 73-365) (Table 3). No patients had a preoperative blood glucose level of < 70 mg/dL and 19 (14.5%) had a blood glucose level > 180 mg/dL. Among patients with hyperglycemia immediately prior to surgery, 6 (31.6%) had documentation of insulin being provided.

For this sample of patients, the preoperative clinic visit was conducted a mean 22 days prior to the planned surgery date. Among the 131 included patients, 122 (93.1%) had documentation of receiving instructions for DM medications. Among patients who had documented receipt of instructions, only 30 (24.6%) had instructions specifically tailored to their regimen rather than a generic templated form. The mean (SD) preoperative blood glucose was similar for those who received specific perioperative DM instructions at 146 (50) mg/dL when compared with those who did not at 147 (45) mg/dL. The mean (SD) preoperative blood glucose reading for those who had no documentation of receipt of perioperative instructions was 126 (54) mg/dL compared with 147 (46) mg/dL for those who did.

The mean number of postoperative blood glucose events per day was negligible for hypoglycemia and more frequent for hyperglycemia with a mean of 2 events per day. The mean postoperative blood glucose range was 121 to 247 mg/dL with most readings < 180 mg/dL. Upon discharge, most patients continued their home DM regimen with 5 patients (3.8%) having changes made to their regimen upon discharge.

Very few postoperative complications were identified from chart review. The most frequently observed postoperative complications were acute kidney injury, surgical site infections, and nonfatal stroke. There were no documented nonfatal myocardial infarctions. Two patients (1.5%) died within 30 days of the surgery; neither death was deemed to have been related to poor perioperative glycemic control.

Discussion

To our knowledge, this retrospective chart review was the first study to assess preoperative DM management and postoperative complications in a veteran population. VHI is a large, tertiary, level 1a, academic medical center that serves approximately 62,000 veterans annually and performs about 5000 to 6000 surgeries annually, a total that is increasing following the COVID-19 pandemic.²⁰ This study found that the current process of a presurgery clinic visit and day of surgery glucose assessment has prevented surgical delays or cancellations.

Most patients included in this study were well controlled at baseline in accordance with the 2025 ADA SOC HbA_1c recommendation of a preoperative HbA1c of < 8%, which may have contributed to no surgical delays or cancellations.¹⁰ However, not all patients had HbA1c collected within 3 months of surgery or even had one collected at all. Despite the ADA SOC providing no explicit recommendation for universal HbA_1c screening prior to elective procedures, its importance cannot be understated given the body of evidence demonstrating poor outcomes with uncontrolled preoperative DM.^8,10 The glycemic control at baseline may have contributed to the very few postsurgical complications observed in this study.

Although the current process at VHI prevented surgical delays and cancellations in this sample, there are still identified areas for improvement. One area is the instructions the patients received. Patients with DM are often prescribed ≥ 1 medication or a combination of insulins, noninsulin injectables, and oral DM medications, and this study population was no different. Because these medications may influence the anesthesia and perioperative periods, the ADA has specific guidance for altering administration schedules in the days leading up to surgery.¹⁰

Inappropriate administration of DM medications could lead to perioperative hypoglycemia or hyperglycemia, possibly causing surgical delays, case cancellations, and/or postoperative complications.21 Although these data reveal the specificity and documented receipt that the preoperative DM instructions did not impact the first recorded preoperative blood glucose, future studies should examine patient confidence in how to properly administer their DM medications prior to surgery. It is vital that patients receive clear instructions in accordance with the ADA SOC on whether to continue, hold, or adjust the dose of their medications to prevent fluctuations in blood glucose levels in the perioperative period, ensure safety with anesthesia, and prevent postoperative complications such as acute kidney injury. Of note, compliance with guideline recommendations for medication instructions was not examined because the data collection time frame expanded over multiple years and the recommendations have evolved each year as new data emerge.

Preoperative DM Management

The first key takeaway from this study is to ensure patients are ready for surgery with a formal assessment (typically in the form of a clinic visit) prior to the surgery. One private sector health system published their approach to this by administering an automatic preoperative HbA_1c screening for those with a DM diagnosis and all patients with a random plasma glucose ≥ 200 mg/dL.²² Additionally, if the patient's HbA_1c level was not at goal prior to surgery (≥ 8% for those with known DM and ≥ 6.5% with no known DM), patients were referred to endocrinology for further management. Increasing attention to the preoperative visit and extending HbA_1c testing to all patients regardless of DM status also provides an opportunity to identify individuals living with undiagnosed DM.¹

Even though there was no difference in the mean preoperative blood glucose level based on receipt or specificity of preoperative DM instructions, a second takeaway from this study is the importance of ensuring patients receive clear instructions on their DM medication schedule in the perioperative period. A practical first step may be updating the templates used by the primary surgery teams and providing education to the clinicians in the clinic on how to personalize the visits. Because the current preoperative DM process at VHI is managed by the primary surgical team in a clinic visit, there is an opportunity to shift this responsibility to other health care professionals, such as pharmacists—a change shown to reduce unintended omission of home medications following surgery during hospitalization and reduce costs.^23,24

Limitations

This study relied on data included in the patient chart. These data include medication interventions made immediately prior to surgery, which can sometimes be inaccurately charted or difficult to find as they are not documented in the typical medication administration record. Also, the safety outcomes were collected from a discharge summary written by different clinicians, which may lead to information bias. Special attention was taken to ensure these data points were collected as accurately as possible, but it is possible some data may be inaccurate from unintentional human error. Additionally, the safety outcome was limited to a 30-day follow-up, but encompassed the entire length of postoperative stay for all included patients. Finally, given this study was retrospective with no comparison group and the intent was to improve processes at VHI, only hypotheses and potential interventions can be generated from this study. Future prospective studies with larger sample sizes and comparator groups are needed to draw further conclusions.

Conclusions

This study found that the current presurgery process at VHI appears to be successful in preventing surgical delays or cancellations due to hyperglycemia or hypoglycemia. Optimizing DM management can improve surgical outcomes by decreasing rates of postoperative complications, and this study added additional evidence in support of that in a unique population: veterans. Insight on the awareness of preoperative blood glucose management should be gleaned from this study, and based on this sample and site, the preadmission screening process and instructions provided to patients can serve as 2 starting points for optimizing elective surgery.

Type 2 diabetes mellitus (T2DM) is a chronic disease becoming more prevalent each year and is the seventh-leading cause of death in the United States.¹ The most common reason for hospitalization for patients with T2DM is uncontrolled glycemic levels.² Nearly 25% of the US Department of Veterans Affairs (VA) patient population has T2DM.³ T2DM is the leading cause of blindness, end-stage renal disease, and amputation for VA patients.⁴

According to the 2023 American Diabetes Association (ADA) guidelines, treatment goals of T2DM include eliminating symptoms, preventing or delaying complications, and attaining glycemic goals. A typical hemoglobin A_1c (HbA_1c) goal range is < 7%, but individual goals can vary up to < 9% due to a multitude of factors, including patient comorbidities and clinical status.⁵

Initial treatment recommendations are nonpharmacologic and include comprehensive lifestyle interventions such as optimizing nutrition, physical activity, and behavioral therapy. When pharmacologic therapy is required, metformin is the preferred first-line treatment for the majority of newly diagnosed patients with T2DM and should be added to continued lifestyle management.⁵ If HbA_1c levels remains above goal, the 2023 ADA guidelines recommend adding a second medication, including but not limited to insulin, a glucagonlike peptide-1 receptor agonist (GLP-1RA), or a sodium-glucose cotransporter 2 inhibitor. Medication choice is largely based on the patient’s concomitant conditions (eg, atherosclerotic cardiovascular disease, heart failure, or chronic kidney disease). The 2023 ADA guidelines suggest initiating insulin therapy when a patient's blood glucose ≥ 300 mg/dL, HbA_1c > 10%, or if the patient has symptoms of hyperglycemia, even at initial diagnosis. Initiating medications to minimize or avoid hypoglycemia is a priority, especially in high-risk individuals.⁵

Clinical evidence shows that GLP-1RAs may provide similar glycemic control to insulin with lower risk of hypoglycemia.⁶ Other reported benefits of GLP-1RAs include weight loss, blood pressure reduction, and improved lipid levels. The most common adverse events (AEs) with GLP-1RAs are gastrointestinal. Including GLP-1RAs in T2DM pharmacotherapy may lower the risk of hypoglycemia, especially in patients at high risk of hypoglycemia.

The 2023 ADA guidelines indicate that it is appropriate to initiate GLP-]1RAs in patients on insulin.⁵ However, while GLP-1RAs do not increase the risk of hypoglycemia independently, combination treatment with GLP-1RAs and insulin can still result in hypoglycemia.⁶ Insulin is the key suspect of this hypoglycemic risk.⁷ Thus, if insulin dosage can be reduced or discontinued, this might reduce the risk of hypoglycemia.

The literature is limited on how the addition of a GLP-1RA to insulin treatment will affect the patient's daily insulin doses, particularly for the veteran population. The goal of this study is to examine this gap in current research by examining semaglutide, which is the current formulary preferred GLP-1RA at the VA.

Semaglutide is subcutaneously initiated at a dose of 0.25 mg once weekly for 4 weeks to reduce gastrointestinal symptoms, then increased to 0.5 mg weekly. Additional increases to a maintenance dose of 1 mg or 2 mg weekly can occur to achieve glycemic goals. The SUSTAIN-FORTE randomized controlled trial sought to determine whether there was a difference in HbA_1c level reduction and significant weight loss with the 2-mg vs 1-mg dose.⁸ Patients in the trial were taking metformin but needed additional medication to control their HbA_1c. They were not using insulin and may or may not have been taking sulfonylureas prior to semaglutide initiation. Semaglutide 2 mg was found to significantly improve HbA_1c control and promote weight loss compared with semaglutide 1 mg, while maintaining a similar safety profile.

Because this study involved patients who required additional HbA_1c control, although semaglutide reduced HbA_1c, not all patients were able to reduce their other diabetes medications, which depended on the baseline HbA_1c level and the level upon completion of semaglutide titration. Dose reductions for the patients’ other T2DM medications were not reported at trial end. SUSTAIN-FORTE established titration up to semaglutide 2 mg as effective for HbA_1c reduction, although it did not study patients also on insulin.⁸

Insulin is associated with hypoglycemic risk, weight gain, and other AEs.^7,8 This study analyzed whether increasing semaglutide could reduce insulin doses and therefore reduce risk of AEs in patients with T2DM.

Methods

A retrospective, single-center, chart review was conducted at VA Sioux Falls Health Care System (VASFHCS). Data were collected through manual review of VASFHCS electronic medical records. Patients aged ≥ 18 years with active prescriptions for at least once-daily insulin who were initiated on 2-mg weekly dose of semaglutide at the VASFHCS clinical pharmacy practitioner medication management clinic between January 1, 2021, and September 1, 2023, were included. VASFHCS clinical pharmacy practitioners have a scope of practice that allows them to initiate, modify, or discontinue medication therapy within medication management clinics.

The most frequently used prandial insulin at VASFHCS is insulin aspart, and the most frequently used basal insulin is insulin glargine. Patients were retrospectively monitored as they progressed from baseline (the point in time where semaglutide 0.5 mg was initiated) to ≥ 3 months on semaglutide 2-mg therapy. Patients were excluded if they previously used a GLP-1RA or if they were on sliding scale insulin without an exact daily dosage.

The primary endpoint was the percent change in total daily insulin dose from baseline to each dose increase after receiving semaglutide 2 mg for ≥ 3 months. Secondary endpoints included changes in daily prandial insulin dose, daily basal insulin dose, HbA_1c, and number of hypoglycemic events reported. Data collected included age, race, weight, body mass index, total daily prandial insulin dose, total daily basal insulin dose, HbA_1c, and hypoglycemic events reported at the visit when semaglutide was initiated.

Statistical Analysis

The sample size was calculated prior to data collection, and it was determined that for α = .05, 47 patients were needed to achieve 95% power. The primary endpoint was assessed using a paired t test, as were each secondary endpoint. Results with P < .05 were considered statistically significant.

Results

Sixty-two patients were included. The mean HbA_1c level at baseline was 7.7%, the baseline mean prandial and insulin daily doses were 41.5 units and 85.1 units, respectively (Table 1) From baseline to initiation of a semaglutide 1-mg dose, the daily insulin dose changed –5.6% (95% CI, 2.2-14.0; P = .008). From baseline to 2-mg dose initiation daily insulin changed -22.2% (95% CI, 22.0-35.1; P < .001) and for patients receiving semaglutide 2 mg for ≥ 3 months it changed -36.9% (95% CI, 37.4-56.5; P < .001) (Figure).

After receiving the 2-mg dose for ≥ 3 months, the mean daily dose of prandial insulin decreased from 41.5 units to 24.6 units (95% CI, 12.6-21.2; P < .001); mean daily dose of basal insulin decreased from 85.1 units to 52.1 units (95% CI, 23.9-42.0; P < .001); and mean HbA_1c level decreased from 7.7% to 7.1% (95% CI, 0.3-0.8; P < .001). Mean number of hypoglycemic events reported was not statistically significant, changing from 3.6 to 3.2 (95% CI, –0.6 to 0.1; P = .21) (Table 2).

Discussion

This study investigated the effect of subcutaneous semaglutide dose escalation on total daily insulin dose for patients with T2DM. There was a statistically significant decrease in total daily insulin dose from baseline to 1 mg initiation; this decrease continued with further insulin dose reduction seen at the 2-mg dose initiation and additional insulin dose reduction at ≥ 3 months at this dose. It was hypothesized there would be a significant total daily insulin dose reduction at some point, especially when transitioning from the semaglutide 1-mg to the 2-mg dose, based on previous research. ^9,10 The additional reduction in daily insulin dose when continuing on semaglutide 2 mg for ≥ 3 months was an unanticipated but added benefit, showing that if tolerated, maintaining the 2-mg dose will help patients reduce their insulin doses.

In terms of secondary endpoints, there was a statistically significant decrease in mean total daily dose individually for prandial and basal insulin from baseline to ≥ 3 months after semaglutide 2 mg initiation. The change in HbA_1c level was also statistically significant and decreased from baseline, even as insulin doses were reduced. This change in HbA_1c level was expected; previous literature has shown a significant link between improving HbA_1c control when semaglutide doses are increased to 2 mg weekly.¹⁰ Due to having been shown in previous trials, it was expected that HbA_1c levels would decrease even when the insulin doses were being reduced.¹⁰ Insulin dose reduction can potentially be added to the growing evidence of semaglutide benefits. The change in the number of hypoglycemic events was not statistically significant, which was unexpected since previous research show a trend in patients taking GLP-1RAs having fewer hypoglycemic events than those taking insulin.⁶ Further investigation with a larger sample size and prospective trial could determine whether this result is an outlier. In this study, there was no increase in HbA_1c or hypoglycemic events reported with increasing semaglutide doses, which provides further evidence of the safety of semaglutide even at higher doses.

These data suggest that for a patient with T2DM who is already taking insulin, the recommended titration of semaglutide is to start with 0.5 mg and titrate up to a 2-mg subcutaneous weekly dose and to then continue at that dose. As long as the 2-mg dose is tolerated, it will provide patients with the most HbA_1c control and lead to a reduction of their total daily insulin doses according to these results.

Strengths and Limitations

This study compared patient data at different points. This method did not require a second distinct control group, which would potentially introduce confounding factors, such as different baseline characteristics. Another strength is that documentation was available for all patients throughout the study so no one was lost to follow-up. This allowed comprehensive data collection and provided a stronger conclusion given the completeness of the data from baseline to follow-up.

Limitations include the retrospective design and small sample size. In addition, the study design did not allow for randomization. There is no documentation of adherence to medication regimen, which was difficult to determine due to the retrospective nature. Other changes to the patients’ medication regimen were not collected in aggregate and thus, it is possible the total daily insulin dose was impacted by other medication changes. There is also potential for inconsistent documentation of the patients’ true total daily insulin dose in the medical record, thus leading to inaccuracy of recorded data.

Conclusions

A small sample of veterans with T2DM had statistically significant reductions in total daily insulin dose when subcutaneous semaglutide was initiated, as well as after each dose increase. There was also a statistically significant reduction in HbA_1c levels from baseline even as patient insulin doses were reduced. These results support the current practice of using semaglutide to treat T2DM, suggesting it may be safe and effective at reducing HbA_1c levels as the dose is titrated up to 2 mg. There was no statistically significant change in the number of hypoglycemic events reported as semaglutide was titrated up. Thus, when semaglutide is increased to the maximum recommended dose of 2 mg for T2DM, patients may experience a reduction of their total daily dose of insulin and HbA_1c levels. These benefits may reduce the risk of insulin-related AEs while maintaining appropriate glycemic control.

Type 2 diabetes mellitus (T2DM) is a chronic disease becoming more prevalent each year and is the seventh-leading cause of death in the United States.¹ The most common reason for hospitalization for patients with T2DM is uncontrolled glycemic levels.² Nearly 25% of the US Department of Veterans Affairs (VA) patient population has T2DM.³ T2DM is the leading cause of blindness, end-stage renal disease, and amputation for VA patients.⁴

According to the 2023 American Diabetes Association (ADA) guidelines, treatment goals of T2DM include eliminating symptoms, preventing or delaying complications, and attaining glycemic goals. A typical hemoglobin A_1c (HbA_1c) goal range is < 7%, but individual goals can vary up to < 9% due to a multitude of factors, including patient comorbidities and clinical status.⁵

Initial treatment recommendations are nonpharmacologic and include comprehensive lifestyle interventions such as optimizing nutrition, physical activity, and behavioral therapy. When pharmacologic therapy is required, metformin is the preferred first-line treatment for the majority of newly diagnosed patients with T2DM and should be added to continued lifestyle management.⁵ If HbA_1c levels remains above goal, the 2023 ADA guidelines recommend adding a second medication, including but not limited to insulin, a glucagonlike peptide-1 receptor agonist (GLP-1RA), or a sodium-glucose cotransporter 2 inhibitor. Medication choice is largely based on the patient’s concomitant conditions (eg, atherosclerotic cardiovascular disease, heart failure, or chronic kidney disease). The 2023 ADA guidelines suggest initiating insulin therapy when a patient's blood glucose ≥ 300 mg/dL, HbA_1c > 10%, or if the patient has symptoms of hyperglycemia, even at initial diagnosis. Initiating medications to minimize or avoid hypoglycemia is a priority, especially in high-risk individuals.⁵

Clinical evidence shows that GLP-1RAs may provide similar glycemic control to insulin with lower risk of hypoglycemia.⁶ Other reported benefits of GLP-1RAs include weight loss, blood pressure reduction, and improved lipid levels. The most common adverse events (AEs) with GLP-1RAs are gastrointestinal. Including GLP-1RAs in T2DM pharmacotherapy may lower the risk of hypoglycemia, especially in patients at high risk of hypoglycemia.

The 2023 ADA guidelines indicate that it is appropriate to initiate GLP-]1RAs in patients on insulin.⁵ However, while GLP-1RAs do not increase the risk of hypoglycemia independently, combination treatment with GLP-1RAs and insulin can still result in hypoglycemia.⁶ Insulin is the key suspect of this hypoglycemic risk.⁷ Thus, if insulin dosage can be reduced or discontinued, this might reduce the risk of hypoglycemia.

The literature is limited on how the addition of a GLP-1RA to insulin treatment will affect the patient's daily insulin doses, particularly for the veteran population. The goal of this study is to examine this gap in current research by examining semaglutide, which is the current formulary preferred GLP-1RA at the VA.

Semaglutide is subcutaneously initiated at a dose of 0.25 mg once weekly for 4 weeks to reduce gastrointestinal symptoms, then increased to 0.5 mg weekly. Additional increases to a maintenance dose of 1 mg or 2 mg weekly can occur to achieve glycemic goals. The SUSTAIN-FORTE randomized controlled trial sought to determine whether there was a difference in HbA_1c level reduction and significant weight loss with the 2-mg vs 1-mg dose.⁸ Patients in the trial were taking metformin but needed additional medication to control their HbA_1c. They were not using insulin and may or may not have been taking sulfonylureas prior to semaglutide initiation. Semaglutide 2 mg was found to significantly improve HbA_1c control and promote weight loss compared with semaglutide 1 mg, while maintaining a similar safety profile.

Because this study involved patients who required additional HbA_1c control, although semaglutide reduced HbA_1c, not all patients were able to reduce their other diabetes medications, which depended on the baseline HbA_1c level and the level upon completion of semaglutide titration. Dose reductions for the patients’ other T2DM medications were not reported at trial end. SUSTAIN-FORTE established titration up to semaglutide 2 mg as effective for HbA_1c reduction, although it did not study patients also on insulin.⁸

Insulin is associated with hypoglycemic risk, weight gain, and other AEs.^7,8 This study analyzed whether increasing semaglutide could reduce insulin doses and therefore reduce risk of AEs in patients with T2DM.

Methods

A retrospective, single-center, chart review was conducted at VA Sioux Falls Health Care System (VASFHCS). Data were collected through manual review of VASFHCS electronic medical records. Patients aged ≥ 18 years with active prescriptions for at least once-daily insulin who were initiated on 2-mg weekly dose of semaglutide at the VASFHCS clinical pharmacy practitioner medication management clinic between January 1, 2021, and September 1, 2023, were included. VASFHCS clinical pharmacy practitioners have a scope of practice that allows them to initiate, modify, or discontinue medication therapy within medication management clinics.

The most frequently used prandial insulin at VASFHCS is insulin aspart, and the most frequently used basal insulin is insulin glargine. Patients were retrospectively monitored as they progressed from baseline (the point in time where semaglutide 0.5 mg was initiated) to ≥ 3 months on semaglutide 2-mg therapy. Patients were excluded if they previously used a GLP-1RA or if they were on sliding scale insulin without an exact daily dosage.

The primary endpoint was the percent change in total daily insulin dose from baseline to each dose increase after receiving semaglutide 2 mg for ≥ 3 months. Secondary endpoints included changes in daily prandial insulin dose, daily basal insulin dose, HbA_1c, and number of hypoglycemic events reported. Data collected included age, race, weight, body mass index, total daily prandial insulin dose, total daily basal insulin dose, HbA_1c, and hypoglycemic events reported at the visit when semaglutide was initiated.

Statistical Analysis

The sample size was calculated prior to data collection, and it was determined that for α = .05, 47 patients were needed to achieve 95% power. The primary endpoint was assessed using a paired t test, as were each secondary endpoint. Results with P < .05 were considered statistically significant.

Results

Sixty-two patients were included. The mean HbA_1c level at baseline was 7.7%, the baseline mean prandial and insulin daily doses were 41.5 units and 85.1 units, respectively (Table 1) From baseline to initiation of a semaglutide 1-mg dose, the daily insulin dose changed –5.6% (95% CI, 2.2-14.0; P = .008). From baseline to 2-mg dose initiation daily insulin changed -22.2% (95% CI, 22.0-35.1; P < .001) and for patients receiving semaglutide 2 mg for ≥ 3 months it changed -36.9% (95% CI, 37.4-56.5; P < .001) (Figure).

After receiving the 2-mg dose for ≥ 3 months, the mean daily dose of prandial insulin decreased from 41.5 units to 24.6 units (95% CI, 12.6-21.2; P < .001); mean daily dose of basal insulin decreased from 85.1 units to 52.1 units (95% CI, 23.9-42.0; P < .001); and mean HbA_1c level decreased from 7.7% to 7.1% (95% CI, 0.3-0.8; P < .001). Mean number of hypoglycemic events reported was not statistically significant, changing from 3.6 to 3.2 (95% CI, –0.6 to 0.1; P = .21) (Table 2).

Discussion

This study investigated the effect of subcutaneous semaglutide dose escalation on total daily insulin dose for patients with T2DM. There was a statistically significant decrease in total daily insulin dose from baseline to 1 mg initiation; this decrease continued with further insulin dose reduction seen at the 2-mg dose initiation and additional insulin dose reduction at ≥ 3 months at this dose. It was hypothesized there would be a significant total daily insulin dose reduction at some point, especially when transitioning from the semaglutide 1-mg to the 2-mg dose, based on previous research. ^9,10 The additional reduction in daily insulin dose when continuing on semaglutide 2 mg for ≥ 3 months was an unanticipated but added benefit, showing that if tolerated, maintaining the 2-mg dose will help patients reduce their insulin doses.

In terms of secondary endpoints, there was a statistically significant decrease in mean total daily dose individually for prandial and basal insulin from baseline to ≥ 3 months after semaglutide 2 mg initiation. The change in HbA_1c level was also statistically significant and decreased from baseline, even as insulin doses were reduced. This change in HbA_1c level was expected; previous literature has shown a significant link between improving HbA_1c control when semaglutide doses are increased to 2 mg weekly.¹⁰ Due to having been shown in previous trials, it was expected that HbA_1c levels would decrease even when the insulin doses were being reduced.¹⁰ Insulin dose reduction can potentially be added to the growing evidence of semaglutide benefits. The change in the number of hypoglycemic events was not statistically significant, which was unexpected since previous research show a trend in patients taking GLP-1RAs having fewer hypoglycemic events than those taking insulin.⁶ Further investigation with a larger sample size and prospective trial could determine whether this result is an outlier. In this study, there was no increase in HbA_1c or hypoglycemic events reported with increasing semaglutide doses, which provides further evidence of the safety of semaglutide even at higher doses.

These data suggest that for a patient with T2DM who is already taking insulin, the recommended titration of semaglutide is to start with 0.5 mg and titrate up to a 2-mg subcutaneous weekly dose and to then continue at that dose. As long as the 2-mg dose is tolerated, it will provide patients with the most HbA_1c control and lead to a reduction of their total daily insulin doses according to these results.

Strengths and Limitations

This study compared patient data at different points. This method did not require a second distinct control group, which would potentially introduce confounding factors, such as different baseline characteristics. Another strength is that documentation was available for all patients throughout the study so no one was lost to follow-up. This allowed comprehensive data collection and provided a stronger conclusion given the completeness of the data from baseline to follow-up.

Limitations include the retrospective design and small sample size. In addition, the study design did not allow for randomization. There is no documentation of adherence to medication regimen, which was difficult to determine due to the retrospective nature. Other changes to the patients’ medication regimen were not collected in aggregate and thus, it is possible the total daily insulin dose was impacted by other medication changes. There is also potential for inconsistent documentation of the patients’ true total daily insulin dose in the medical record, thus leading to inaccuracy of recorded data.

Conclusions

A small sample of veterans with T2DM had statistically significant reductions in total daily insulin dose when subcutaneous semaglutide was initiated, as well as after each dose increase. There was also a statistically significant reduction in HbA_1c levels from baseline even as patient insulin doses were reduced. These results support the current practice of using semaglutide to treat T2DM, suggesting it may be safe and effective at reducing HbA_1c levels as the dose is titrated up to 2 mg. There was no statistically significant change in the number of hypoglycemic events reported as semaglutide was titrated up. Thus, when semaglutide is increased to the maximum recommended dose of 2 mg for T2DM, patients may experience a reduction of their total daily dose of insulin and HbA_1c levels. These benefits may reduce the risk of insulin-related AEs while maintaining appropriate glycemic control.

Type 2 diabetes mellitus (T2DM) is a chronic disease becoming more prevalent each year and is the seventh-leading cause of death in the United States.¹ The most common reason for hospitalization for patients with T2DM is uncontrolled glycemic levels.² Nearly 25% of the US Department of Veterans Affairs (VA) patient population has T2DM.³ T2DM is the leading cause of blindness, end-stage renal disease, and amputation for VA patients.⁴

According to the 2023 American Diabetes Association (ADA) guidelines, treatment goals of T2DM include eliminating symptoms, preventing or delaying complications, and attaining glycemic goals. A typical hemoglobin A_1c (HbA_1c) goal range is < 7%, but individual goals can vary up to < 9% due to a multitude of factors, including patient comorbidities and clinical status.⁵

Initial treatment recommendations are nonpharmacologic and include comprehensive lifestyle interventions such as optimizing nutrition, physical activity, and behavioral therapy. When pharmacologic therapy is required, metformin is the preferred first-line treatment for the majority of newly diagnosed patients with T2DM and should be added to continued lifestyle management.⁵ If HbA_1c levels remains above goal, the 2023 ADA guidelines recommend adding a second medication, including but not limited to insulin, a glucagonlike peptide-1 receptor agonist (GLP-1RA), or a sodium-glucose cotransporter 2 inhibitor. Medication choice is largely based on the patient’s concomitant conditions (eg, atherosclerotic cardiovascular disease, heart failure, or chronic kidney disease). The 2023 ADA guidelines suggest initiating insulin therapy when a patient's blood glucose ≥ 300 mg/dL, HbA_1c > 10%, or if the patient has symptoms of hyperglycemia, even at initial diagnosis. Initiating medications to minimize or avoid hypoglycemia is a priority, especially in high-risk individuals.⁵

Clinical evidence shows that GLP-1RAs may provide similar glycemic control to insulin with lower risk of hypoglycemia.⁶ Other reported benefits of GLP-1RAs include weight loss, blood pressure reduction, and improved lipid levels. The most common adverse events (AEs) with GLP-1RAs are gastrointestinal. Including GLP-1RAs in T2DM pharmacotherapy may lower the risk of hypoglycemia, especially in patients at high risk of hypoglycemia.

The 2023 ADA guidelines indicate that it is appropriate to initiate GLP-]1RAs in patients on insulin.⁵ However, while GLP-1RAs do not increase the risk of hypoglycemia independently, combination treatment with GLP-1RAs and insulin can still result in hypoglycemia.⁶ Insulin is the key suspect of this hypoglycemic risk.⁷ Thus, if insulin dosage can be reduced or discontinued, this might reduce the risk of hypoglycemia.

The literature is limited on how the addition of a GLP-1RA to insulin treatment will affect the patient's daily insulin doses, particularly for the veteran population. The goal of this study is to examine this gap in current research by examining semaglutide, which is the current formulary preferred GLP-1RA at the VA.

Semaglutide is subcutaneously initiated at a dose of 0.25 mg once weekly for 4 weeks to reduce gastrointestinal symptoms, then increased to 0.5 mg weekly. Additional increases to a maintenance dose of 1 mg or 2 mg weekly can occur to achieve glycemic goals. The SUSTAIN-FORTE randomized controlled trial sought to determine whether there was a difference in HbA_1c level reduction and significant weight loss with the 2-mg vs 1-mg dose.⁸ Patients in the trial were taking metformin but needed additional medication to control their HbA_1c. They were not using insulin and may or may not have been taking sulfonylureas prior to semaglutide initiation. Semaglutide 2 mg was found to significantly improve HbA_1c control and promote weight loss compared with semaglutide 1 mg, while maintaining a similar safety profile.

Because this study involved patients who required additional HbA_1c control, although semaglutide reduced HbA_1c, not all patients were able to reduce their other diabetes medications, which depended on the baseline HbA_1c level and the level upon completion of semaglutide titration. Dose reductions for the patients’ other T2DM medications were not reported at trial end. SUSTAIN-FORTE established titration up to semaglutide 2 mg as effective for HbA_1c reduction, although it did not study patients also on insulin.⁸

Insulin is associated with hypoglycemic risk, weight gain, and other AEs.^7,8 This study analyzed whether increasing semaglutide could reduce insulin doses and therefore reduce risk of AEs in patients with T2DM.

Methods

A retrospective, single-center, chart review was conducted at VA Sioux Falls Health Care System (VASFHCS). Data were collected through manual review of VASFHCS electronic medical records. Patients aged ≥ 18 years with active prescriptions for at least once-daily insulin who were initiated on 2-mg weekly dose of semaglutide at the VASFHCS clinical pharmacy practitioner medication management clinic between January 1, 2021, and September 1, 2023, were included. VASFHCS clinical pharmacy practitioners have a scope of practice that allows them to initiate, modify, or discontinue medication therapy within medication management clinics.

The most frequently used prandial insulin at VASFHCS is insulin aspart, and the most frequently used basal insulin is insulin glargine. Patients were retrospectively monitored as they progressed from baseline (the point in time where semaglutide 0.5 mg was initiated) to ≥ 3 months on semaglutide 2-mg therapy. Patients were excluded if they previously used a GLP-1RA or if they were on sliding scale insulin without an exact daily dosage.

The primary endpoint was the percent change in total daily insulin dose from baseline to each dose increase after receiving semaglutide 2 mg for ≥ 3 months. Secondary endpoints included changes in daily prandial insulin dose, daily basal insulin dose, HbA_1c, and number of hypoglycemic events reported. Data collected included age, race, weight, body mass index, total daily prandial insulin dose, total daily basal insulin dose, HbA_1c, and hypoglycemic events reported at the visit when semaglutide was initiated.

Statistical Analysis

The sample size was calculated prior to data collection, and it was determined that for α = .05, 47 patients were needed to achieve 95% power. The primary endpoint was assessed using a paired t test, as were each secondary endpoint. Results with P < .05 were considered statistically significant.

Results

Sixty-two patients were included. The mean HbA_1c level at baseline was 7.7%, the baseline mean prandial and insulin daily doses were 41.5 units and 85.1 units, respectively (Table 1) From baseline to initiation of a semaglutide 1-mg dose, the daily insulin dose changed –5.6% (95% CI, 2.2-14.0; P = .008). From baseline to 2-mg dose initiation daily insulin changed -22.2% (95% CI, 22.0-35.1; P < .001) and for patients receiving semaglutide 2 mg for ≥ 3 months it changed -36.9% (95% CI, 37.4-56.5; P < .001) (Figure).

After receiving the 2-mg dose for ≥ 3 months, the mean daily dose of prandial insulin decreased from 41.5 units to 24.6 units (95% CI, 12.6-21.2; P < .001); mean daily dose of basal insulin decreased from 85.1 units to 52.1 units (95% CI, 23.9-42.0; P < .001); and mean HbA_1c level decreased from 7.7% to 7.1% (95% CI, 0.3-0.8; P < .001). Mean number of hypoglycemic events reported was not statistically significant, changing from 3.6 to 3.2 (95% CI, –0.6 to 0.1; P = .21) (Table 2).

Discussion

This study investigated the effect of subcutaneous semaglutide dose escalation on total daily insulin dose for patients with T2DM. There was a statistically significant decrease in total daily insulin dose from baseline to 1 mg initiation; this decrease continued with further insulin dose reduction seen at the 2-mg dose initiation and additional insulin dose reduction at ≥ 3 months at this dose. It was hypothesized there would be a significant total daily insulin dose reduction at some point, especially when transitioning from the semaglutide 1-mg to the 2-mg dose, based on previous research. ^9,10 The additional reduction in daily insulin dose when continuing on semaglutide 2 mg for ≥ 3 months was an unanticipated but added benefit, showing that if tolerated, maintaining the 2-mg dose will help patients reduce their insulin doses.

In terms of secondary endpoints, there was a statistically significant decrease in mean total daily dose individually for prandial and basal insulin from baseline to ≥ 3 months after semaglutide 2 mg initiation. The change in HbA_1c level was also statistically significant and decreased from baseline, even as insulin doses were reduced. This change in HbA_1c level was expected; previous literature has shown a significant link between improving HbA_1c control when semaglutide doses are increased to 2 mg weekly.¹⁰ Due to having been shown in previous trials, it was expected that HbA_1c levels would decrease even when the insulin doses were being reduced.¹⁰ Insulin dose reduction can potentially be added to the growing evidence of semaglutide benefits. The change in the number of hypoglycemic events was not statistically significant, which was unexpected since previous research show a trend in patients taking GLP-1RAs having fewer hypoglycemic events than those taking insulin.⁶ Further investigation with a larger sample size and prospective trial could determine whether this result is an outlier. In this study, there was no increase in HbA_1c or hypoglycemic events reported with increasing semaglutide doses, which provides further evidence of the safety of semaglutide even at higher doses.

These data suggest that for a patient with T2DM who is already taking insulin, the recommended titration of semaglutide is to start with 0.5 mg and titrate up to a 2-mg subcutaneous weekly dose and to then continue at that dose. As long as the 2-mg dose is tolerated, it will provide patients with the most HbA_1c control and lead to a reduction of their total daily insulin doses according to these results.

Strengths and Limitations

This study compared patient data at different points. This method did not require a second distinct control group, which would potentially introduce confounding factors, such as different baseline characteristics. Another strength is that documentation was available for all patients throughout the study so no one was lost to follow-up. This allowed comprehensive data collection and provided a stronger conclusion given the completeness of the data from baseline to follow-up.

Limitations include the retrospective design and small sample size. In addition, the study design did not allow for randomization. There is no documentation of adherence to medication regimen, which was difficult to determine due to the retrospective nature. Other changes to the patients’ medication regimen were not collected in aggregate and thus, it is possible the total daily insulin dose was impacted by other medication changes. There is also potential for inconsistent documentation of the patients’ true total daily insulin dose in the medical record, thus leading to inaccuracy of recorded data.

Conclusions

A small sample of veterans with T2DM had statistically significant reductions in total daily insulin dose when subcutaneous semaglutide was initiated, as well as after each dose increase. There was also a statistically significant reduction in HbA_1c levels from baseline even as patient insulin doses were reduced. These results support the current practice of using semaglutide to treat T2DM, suggesting it may be safe and effective at reducing HbA_1c levels as the dose is titrated up to 2 mg. There was no statistically significant change in the number of hypoglycemic events reported as semaglutide was titrated up. Thus, when semaglutide is increased to the maximum recommended dose of 2 mg for T2DM, patients may experience a reduction of their total daily dose of insulin and HbA_1c levels. These benefits may reduce the risk of insulin-related AEs while maintaining appropriate glycemic control.

Diabetes mellitus (DM) is a national health crisis affecting > 38 million people (11.6%) in the United States.¹ American Indian and Alaska Native (AI/AN) adults are disproportionately affected, with a prevalence of 14.5%—the highest among all racial and ethnic groups.¹ Type 2 DM (T2DM) accounts for 90% to 95% of all DM cases and is a leading cause of morbidity and mortality due to its association with cardiovascular disease, kidney failure, and other complications.²

Maintaining glycemic control is important for managing T2DM and preventing microvascular and macrovascular complications.³ The cornerstone of diabetes self-management has been patient self-monitored blood glucose (SMBG) using finger-stick glucometers.⁴ However, SMBG provides measurements from a single point in time and requires frequent, painful, and inconvenient finger pricks, leading to decreased adherence.^5,6 These limitations negatively affect patient engagement and overall glycemic control.⁷

Continuous glucose monitors (CGMs) offer real-time, continuous glucose readings and trends.⁸ CGMs improve glycemic control and reduce hypoglycemic episodes in patients who are insulin-dependent.^9,10 Flash glucose monitors, a type of CGM that requires scanning to obtain glucose readings, provide similar benefits.¹¹ Despite these demonstrated advantages, research has primarily focused on insulin-dependent populations, leaving a significant gap in understanding the effect of CGMs on patients with T2DM who are not insulin-dependent.¹²

Given the high prevalence of T2DM among AI/AN populations and the potential benefits of CGMs, this study sought to evaluate the effect of CGM use on glycemic control and other health metrics in patients with non–insulin-dependent T2DM in an AI/AN population. This focus addresses a critical knowledge gap and may inform clinical practices and policies to improve diabetes management in this high-risk group.

Methods

A retrospective observational study was conducted using deidentified electronic health records (EHRs) from 2019 to 2024 at a federally operated outpatient Indian Health Service (IHS) clinic serving an AI/AN population in the IHS Portland Area (Oregon, Washington, Idaho). The study protocol was reviewed and deemed exempt by institutional review boards at Washington State University and the Portland Area IHS.

Study Population

This study included patients diagnosed with non–insulin-dependent T2DM, had used a CGM for ≥ 1 year, and had hemoglobin A_1c (HbA_1c) measurements within 4 months prior to CGM initiation (baseline) and within ± 4 months after 1 year of CGM use. For other health metrics, including blood pressure (BP), weight, low-density lipoprotein cholesterol (LDL-C), and estimated glomerular filtration rate (eGFR), this study required measurements within 6 months before CGM initiation and within 6 months after 1 year of CGM use. The baseline HbA_1c in the dataset ranged from 5.3% to > 14%.

Patients were excluded if they used insulin during the study period, had incomplete laboratory or clinical data for the required time frame, or had < 1 year of CGM use. The dataset did not include detailed information on oral DM medications; thus, we could not report or account for the type or number of oral hypoglycemic agents used by the patients. The IHS clinical applications coordinator compiled the dataset from the EHR, identifying patients who were prescribed and received a CGM at the clinic. All patients used the Abbott Freestyle Libre CGM, the only formulary CGM available at the clinic during the study period.

A 1-year follow-up endpoint was selected for several reasons: (1) to capture potential seasonal variations in diet and activity; (2) to align with the clinic’s standard practice of annual comprehensive diabetes evaluations; and (3) to allow sufficient time for patients to adapt to CGM use and reflect any meaningful changes in glycemic control.

All patients received standard DM care according to clinic protocols, which included DM self-management education and training. Patients met with the diabetes educator at least once, during which the educator emphasized making informed decisions using CGM data, such as adjusting dietary choices and physical activity levels to manage blood glucose concentrations effectively.

A total of 302 patients were initially identified. After applying exclusion criteria, 132 were excluded due to insulin use, and 77 were excluded due to incomplete HbA_1c data within the specified time frames (Figure 1). The final sample included 93 patients.

Measures

The primary outcome was the change in HbA_1c levels from baseline to 1 year after CGM initiation. Secondary outcomes included changes in weight, systolic and diastolic BP, LDL-C concentrations, and eGFR. For the primary outcome, HbA_1c values were collected within a grace period of ± 4 months from the baseline and 1-year time points. The laboratory’s upper reporting limit for HbA_1c was 14%; values reported as “> 14%” were recorded as 14.1% for data analysis, although the actual values could have been higher.

For secondary outcomes, data were included if measurements were obtained within ± 6 months of the baseline and 1-year time points. Patients who did not have measurements within these time frames for specific metrics were excluded from secondary outcome analysis but remained in the overall study if they met the criteria for HbA_1c and CGM use.

Statistical Analysis

Statistical analysis was performed using R statistical software version 4.4.2. Paired t tests were conducted to compare baseline and 1-year follow- up measurements for variables with parametric distributions. Wilcoxon signed-rank test was used for nonparametric data. A linear regression analysis was conducted to examine the relationship between baseline HbA_1c levels and the change in HbA_1c after 1 year of CGM use. Differences were considered significant at P < .05 set a priori. To guide future research, a posthoc power analysis was performed using Cohen’s d to estimate the required sample sizes for detecting significant effects, assuming a similar population.

Results

The study included 93 patients, with a mean (SD) age of 55 (13) years (range, 29-83 years). Of the participants, 56 were female (60%) and 37 were male (40%). All participants were identified as AI/AN and had non–insulin-dependent T2DM.

Primary Outcomes

A significant reduction in HbA_1c levels was observed after 1 year of CGM use. The mean (SD) baseline HbA1c was 9.5% (2.4%), which decreased to 7.6% (2.2%) at 1-year follow-up (Table 1). This difference represents a mean change of -1.86% (2.4%) (95% CI, -2.35 to -1.37; P < .001 [paired t test, -7.53]).

A linear regression model evaluated the relationship between baseline HbA_1c (predictor) and the change in HbA_1c after 1 year (outcome). The change in HbA_1c was calculated as the difference between 1-year follow-up and baseline values. The regression model revealed a significant negative association between baseline HbA_1c and the change in HbA_1c (Β = -0.576; P < .001), indicating that higher baseline HbA_1c values were associated with greater reductions in HbA_1c over the year. The regression equation was: Change in HbA_1c = 3.587 – 0.576 × Baseline HbA_1c

The regression coefficient for baseline HbA_1c was -0.576 (standard error, 0.083; t = -6.931; P < .001), indicating that for each 1% increase in baseline HbA_1c, the reduction of HbA_1c after 1 year increased by approximately 0.576% (Figure 2). The model explained 34.6% of the variance in HbA_1c change (R² = .345; adjusted R² = .338).

Secondary Outcomes

Systolic BP decreased by a mean (SD) -4.9 (17) mm Hg; 95% CI, -8.6 to -1.11; P = .01, paired t test). However, no significant change was observed for diastolic BP (P = .77, paired t test). Similarly, no significant changes were observed in weight, LDL-C concentrations, or eGFR after 1 year of CGM use. A posthoc power analysis indicated that the study was underpowered to detect smaller effect sizes in secondary outcomes. For example, sample size estimates indicated that detecting significant changes in weight and LDL-C concentrations would require sample sizes of 152 and 220 patients, respectively (Table 2).

Discussion

This study found a clinically significant reduction in HbA_1c levels after 1 year among AI/AN patients with non–insulin-dependent T2DM who used CGMs. The mean HbA_1c decreased 1.9%, from 9.5% at baseline to 7.6% after 1 year. This reduction is not only statistically significant (P < .001), it is clinically meaningful—even a 1% decrease in HbA_1c is associated with substantial reductions in the risk of microvascular complications.³ The magnitude of the HbA_1c reduction observed suggests CGM use may be associated with improved glycemic control in this high-risk population. By achieving lower HbA_1c levels, patients may experience improved long-term health outcomes and a reduced burden of DM-related complications.

Changes in oral DM medications during the study period may have contributed to the observed improvements in HbA_1c levels. While the dataset lacked detailed information on types or dosages of oral hypoglycemic agents used, adjustments in medication regimens are common in DM management and could significantly affect glycemic control. The inability to account for these changes results in an inability to attribute the improvements in HbA_1c solely to CGM use. Future studies should collect comprehensive medication data to better isolate the effects of CGM use from other treatment modifications.

Another factor that may have contributed to the improved glycemic control is the DM self-management education and training patients received as part of standard care. Patients met with diabetes educators at least once and learned how to use the CGM device and interpret the data for self-management decisions. This education may have enhanced patient engagement and empowerment, enabling them to make informed choices about diet, physical activity, and medication adherence. Studies have shown that DM self-management education can significantly improve glycemic control and patient outcomes.¹³ By combining the CGM technology with targeted education, patients may have been better equipped to manage their condition, contributing to the observed reduction in HbA_1c levels. Future studies should consider synergistic effects of CGM use and DM education when evaluating interventions for glycemic control.

The significant reduction in HbA_1c indicates CGM use is associated with improved glycemic control in non–insulin-dependent T2DM. The linear regression analysis suggests patients with poorer glycemic control at baseline experienced greater reductions in HbA_1c over the course of 1 year. This finding aligns with previous studies that have shown greater HbA_1c reductions in patients with higher initial levels when using CGMs. Yaron et al reported similar findings: higher baseline HbA_1c levels predicted more substantial improvements with CGM use in patients with T2DM on insulin therapy.¹⁴

This study contributes to existing research by examining the association between CGM use and glycemic control in patients with non– insulin-dependent T2DM within an AI/AN population, a group that has been underreported in previous studies. Most prior research has focused on insulin-dependent patients or populations with different ethnic backgrounds.12 By focusing on patients with non–insulin-dependent T2DM, this study highlights the broader applicability of CGMs beyond traditional use, showcasing their potential association with benefits in earlier stages of DM management. Targeting the AI/AN population addresses a critical knowledge gap, given the disproportionately high prevalence of T2DM and associated complications in this group. The findings of this study suggest integrating CGM technology into the standard care of AI/AN patients with non–insulin-dependent T2DM may be associated with improved glycemic control and may help reduce health disparities.

The modest decrease in systolic BP observed in this study may indicate potential cardiovascular benefits associated with CGM use, possibly due to improved glycemic control and increased patient engagement in self-management. However, given the limited sample size and exclusion criteria, the study lacked sufficient power to detect significant associations between CGM use and other secondary outcomes such as BP, weight, LDL-C, and eGFR. Therefore, the significant finding with systolic BP should be interpreted with caution.

The lack of significant changes in secondary outcomes may be attributed to the study’s limited sample size and the relatively short duration for observing changes in these parameters. Larger studies are needed to assess the full impact of CGM on these variables. The required sample sizes for achieving adequate power in future studies were calculated, highlighting the utility of our study as a pilot, providing critical data for the design of larger, adequately powered studies.

Limitations

The retrospective design of this study limits causal inferences. Moreover, potential confounding variables were not controlled, such as changes in medication regimens (other than insulin use), dietary counseling, or physical activity. Additionally, we could not account for the type or number of oral DM medications prescribed to patients. The dataset included only information on insulin use, without detailed records of other antidiabetic medications. This limitation may have influenced the observed change in glycemic control, as variations in medication regimens could affect HbA_1c levels.

Because this study lacked a comparator group, the effect of CGM use cannot be definitively isolated from other factors (eg, medication changes, dietary modifications, or physical activity). Moreover, CGM devices can be costly and are not universally covered by all insurance or IHS programs, potentially limiting widespread implementation. Policy-level restrictions and patient-specific barriers may also hinder feasibility in other settings.

The small sample size may limit the generalizability of the findings. Of the initial 302 patients, about 69% were excluded due to insulin use or incomplete laboratory data. A ± 4-month window was selected to balance data quality with real-world practices. Extending this window further (eg, ± 6 months) might have included more participants but risked diluting the 1-year endpoint consistency. The lack of statistical significance in secondary metrics may be due to insufficient power rather than the absence of an effect.

Exclusion of patients due to incomplete data may have introduced selection bias. However, patients were included in the overall analysis if they met the criteria for HbA_1c and CGM use, even if they lacked data for secondary outcomes. Additionally, the laboratory’s upper reporting limit for HbA_1c was 14%, with values above this reported as “> 14%.” For analysis, these were recorded as 14.1%, which may underestimate the true baseline HbA_1c levels and impact of the assessment of change. This occurred for 4 of the 93 patients included.

All patients used the Freestyle Libre CGM, which may limit the generalizability of the findings to other CGM brands or models. Differences in device features, accuracy, scanning frequency, and user experience may influence outcomes, and results might differ with other CGM technologies. The dataset did not include patients’ scanning frequency because this metric was not consistently included in the EHRs.

Conclusions

This study found that CGM use was significantly associated with improved glycemic control in patients with non–insulin-dependent T2DM within an AI/AN population, particularly among patients with higher baseline HbA_1c levels. The findings suggest that CGMs may be a valuable tool for managing T2DM beyond insulin-dependent populations.

Additional research with larger sample sizes, control groups, and extended follow-up periods is recommended to explore long-term benefits and impacts on other health metrics. The sample size estimates derived from this study serve as a valuable resource for researchers designing future studies aimed at addressing these gaps. Future research that expands on our findings by including larger, more diverse cohorts, accounting for medication use, and exploring different CGM technologies will enhance understanding and contribute to more effective diabetes management strategies for varied populations.

Diabetes mellitus (DM) is a national health crisis affecting > 38 million people (11.6%) in the United States.¹ American Indian and Alaska Native (AI/AN) adults are disproportionately affected, with a prevalence of 14.5%—the highest among all racial and ethnic groups.¹ Type 2 DM (T2DM) accounts for 90% to 95% of all DM cases and is a leading cause of morbidity and mortality due to its association with cardiovascular disease, kidney failure, and other complications.²

Maintaining glycemic control is important for managing T2DM and preventing microvascular and macrovascular complications.³ The cornerstone of diabetes self-management has been patient self-monitored blood glucose (SMBG) using finger-stick glucometers.⁴ However, SMBG provides measurements from a single point in time and requires frequent, painful, and inconvenient finger pricks, leading to decreased adherence.^5,6 These limitations negatively affect patient engagement and overall glycemic control.⁷

Continuous glucose monitors (CGMs) offer real-time, continuous glucose readings and trends.⁸ CGMs improve glycemic control and reduce hypoglycemic episodes in patients who are insulin-dependent.^9,10 Flash glucose monitors, a type of CGM that requires scanning to obtain glucose readings, provide similar benefits.¹¹ Despite these demonstrated advantages, research has primarily focused on insulin-dependent populations, leaving a significant gap in understanding the effect of CGMs on patients with T2DM who are not insulin-dependent.¹²

Given the high prevalence of T2DM among AI/AN populations and the potential benefits of CGMs, this study sought to evaluate the effect of CGM use on glycemic control and other health metrics in patients with non–insulin-dependent T2DM in an AI/AN population. This focus addresses a critical knowledge gap and may inform clinical practices and policies to improve diabetes management in this high-risk group.

Methods

A retrospective observational study was conducted using deidentified electronic health records (EHRs) from 2019 to 2024 at a federally operated outpatient Indian Health Service (IHS) clinic serving an AI/AN population in the IHS Portland Area (Oregon, Washington, Idaho). The study protocol was reviewed and deemed exempt by institutional review boards at Washington State University and the Portland Area IHS.

Study Population

This study included patients diagnosed with non–insulin-dependent T2DM, had used a CGM for ≥ 1 year, and had hemoglobin A_1c (HbA_1c) measurements within 4 months prior to CGM initiation (baseline) and within ± 4 months after 1 year of CGM use. For other health metrics, including blood pressure (BP), weight, low-density lipoprotein cholesterol (LDL-C), and estimated glomerular filtration rate (eGFR), this study required measurements within 6 months before CGM initiation and within 6 months after 1 year of CGM use. The baseline HbA_1c in the dataset ranged from 5.3% to > 14%.

Patients were excluded if they used insulin during the study period, had incomplete laboratory or clinical data for the required time frame, or had < 1 year of CGM use. The dataset did not include detailed information on oral DM medications; thus, we could not report or account for the type or number of oral hypoglycemic agents used by the patients. The IHS clinical applications coordinator compiled the dataset from the EHR, identifying patients who were prescribed and received a CGM at the clinic. All patients used the Abbott Freestyle Libre CGM, the only formulary CGM available at the clinic during the study period.

A 1-year follow-up endpoint was selected for several reasons: (1) to capture potential seasonal variations in diet and activity; (2) to align with the clinic’s standard practice of annual comprehensive diabetes evaluations; and (3) to allow sufficient time for patients to adapt to CGM use and reflect any meaningful changes in glycemic control.

All patients received standard DM care according to clinic protocols, which included DM self-management education and training. Patients met with the diabetes educator at least once, during which the educator emphasized making informed decisions using CGM data, such as adjusting dietary choices and physical activity levels to manage blood glucose concentrations effectively.

A total of 302 patients were initially identified. After applying exclusion criteria, 132 were excluded due to insulin use, and 77 were excluded due to incomplete HbA_1c data within the specified time frames (Figure 1). The final sample included 93 patients.

Measures

The primary outcome was the change in HbA_1c levels from baseline to 1 year after CGM initiation. Secondary outcomes included changes in weight, systolic and diastolic BP, LDL-C concentrations, and eGFR. For the primary outcome, HbA_1c values were collected within a grace period of ± 4 months from the baseline and 1-year time points. The laboratory’s upper reporting limit for HbA_1c was 14%; values reported as “> 14%” were recorded as 14.1% for data analysis, although the actual values could have been higher.

For secondary outcomes, data were included if measurements were obtained within ± 6 months of the baseline and 1-year time points. Patients who did not have measurements within these time frames for specific metrics were excluded from secondary outcome analysis but remained in the overall study if they met the criteria for HbA_1c and CGM use.

Statistical Analysis

Statistical analysis was performed using R statistical software version 4.4.2. Paired t tests were conducted to compare baseline and 1-year follow- up measurements for variables with parametric distributions. Wilcoxon signed-rank test was used for nonparametric data. A linear regression analysis was conducted to examine the relationship between baseline HbA_1c levels and the change in HbA_1c after 1 year of CGM use. Differences were considered significant at P < .05 set a priori. To guide future research, a posthoc power analysis was performed using Cohen’s d to estimate the required sample sizes for detecting significant effects, assuming a similar population.

Results

The study included 93 patients, with a mean (SD) age of 55 (13) years (range, 29-83 years). Of the participants, 56 were female (60%) and 37 were male (40%). All participants were identified as AI/AN and had non–insulin-dependent T2DM.

Primary Outcomes

A significant reduction in HbA_1c levels was observed after 1 year of CGM use. The mean (SD) baseline HbA1c was 9.5% (2.4%), which decreased to 7.6% (2.2%) at 1-year follow-up (Table 1). This difference represents a mean change of -1.86% (2.4%) (95% CI, -2.35 to -1.37; P < .001 [paired t test, -7.53]).

A linear regression model evaluated the relationship between baseline HbA_1c (predictor) and the change in HbA_1c after 1 year (outcome). The change in HbA_1c was calculated as the difference between 1-year follow-up and baseline values. The regression model revealed a significant negative association between baseline HbA_1c and the change in HbA_1c (Β = -0.576; P < .001), indicating that higher baseline HbA_1c values were associated with greater reductions in HbA_1c over the year. The regression equation was: Change in HbA_1c = 3.587 – 0.576 × Baseline HbA_1c

The regression coefficient for baseline HbA_1c was -0.576 (standard error, 0.083; t = -6.931; P < .001), indicating that for each 1% increase in baseline HbA_1c, the reduction of HbA_1c after 1 year increased by approximately 0.576% (Figure 2). The model explained 34.6% of the variance in HbA_1c change (R² = .345; adjusted R² = .338).

Secondary Outcomes

Systolic BP decreased by a mean (SD) -4.9 (17) mm Hg; 95% CI, -8.6 to -1.11; P = .01, paired t test). However, no significant change was observed for diastolic BP (P = .77, paired t test). Similarly, no significant changes were observed in weight, LDL-C concentrations, or eGFR after 1 year of CGM use. A posthoc power analysis indicated that the study was underpowered to detect smaller effect sizes in secondary outcomes. For example, sample size estimates indicated that detecting significant changes in weight and LDL-C concentrations would require sample sizes of 152 and 220 patients, respectively (Table 2).

Discussion

This study found a clinically significant reduction in HbA_1c levels after 1 year among AI/AN patients with non–insulin-dependent T2DM who used CGMs. The mean HbA_1c decreased 1.9%, from 9.5% at baseline to 7.6% after 1 year. This reduction is not only statistically significant (P < .001), it is clinically meaningful—even a 1% decrease in HbA_1c is associated with substantial reductions in the risk of microvascular complications.³ The magnitude of the HbA_1c reduction observed suggests CGM use may be associated with improved glycemic control in this high-risk population. By achieving lower HbA_1c levels, patients may experience improved long-term health outcomes and a reduced burden of DM-related complications.

Changes in oral DM medications during the study period may have contributed to the observed improvements in HbA_1c levels. While the dataset lacked detailed information on types or dosages of oral hypoglycemic agents used, adjustments in medication regimens are common in DM management and could significantly affect glycemic control. The inability to account for these changes results in an inability to attribute the improvements in HbA_1c solely to CGM use. Future studies should collect comprehensive medication data to better isolate the effects of CGM use from other treatment modifications.

Another factor that may have contributed to the improved glycemic control is the DM self-management education and training patients received as part of standard care. Patients met with diabetes educators at least once and learned how to use the CGM device and interpret the data for self-management decisions. This education may have enhanced patient engagement and empowerment, enabling them to make informed choices about diet, physical activity, and medication adherence. Studies have shown that DM self-management education can significantly improve glycemic control and patient outcomes.¹³ By combining the CGM technology with targeted education, patients may have been better equipped to manage their condition, contributing to the observed reduction in HbA_1c levels. Future studies should consider synergistic effects of CGM use and DM education when evaluating interventions for glycemic control.

The significant reduction in HbA_1c indicates CGM use is associated with improved glycemic control in non–insulin-dependent T2DM. The linear regression analysis suggests patients with poorer glycemic control at baseline experienced greater reductions in HbA_1c over the course of 1 year. This finding aligns with previous studies that have shown greater HbA_1c reductions in patients with higher initial levels when using CGMs. Yaron et al reported similar findings: higher baseline HbA_1c levels predicted more substantial improvements with CGM use in patients with T2DM on insulin therapy.¹⁴

This study contributes to existing research by examining the association between CGM use and glycemic control in patients with non– insulin-dependent T2DM within an AI/AN population, a group that has been underreported in previous studies. Most prior research has focused on insulin-dependent patients or populations with different ethnic backgrounds.12 By focusing on patients with non–insulin-dependent T2DM, this study highlights the broader applicability of CGMs beyond traditional use, showcasing their potential association with benefits in earlier stages of DM management. Targeting the AI/AN population addresses a critical knowledge gap, given the disproportionately high prevalence of T2DM and associated complications in this group. The findings of this study suggest integrating CGM technology into the standard care of AI/AN patients with non–insulin-dependent T2DM may be associated with improved glycemic control and may help reduce health disparities.

The modest decrease in systolic BP observed in this study may indicate potential cardiovascular benefits associated with CGM use, possibly due to improved glycemic control and increased patient engagement in self-management. However, given the limited sample size and exclusion criteria, the study lacked sufficient power to detect significant associations between CGM use and other secondary outcomes such as BP, weight, LDL-C, and eGFR. Therefore, the significant finding with systolic BP should be interpreted with caution.

The lack of significant changes in secondary outcomes may be attributed to the study’s limited sample size and the relatively short duration for observing changes in these parameters. Larger studies are needed to assess the full impact of CGM on these variables. The required sample sizes for achieving adequate power in future studies were calculated, highlighting the utility of our study as a pilot, providing critical data for the design of larger, adequately powered studies.

Limitations

The retrospective design of this study limits causal inferences. Moreover, potential confounding variables were not controlled, such as changes in medication regimens (other than insulin use), dietary counseling, or physical activity. Additionally, we could not account for the type or number of oral DM medications prescribed to patients. The dataset included only information on insulin use, without detailed records of other antidiabetic medications. This limitation may have influenced the observed change in glycemic control, as variations in medication regimens could affect HbA_1c levels.

Because this study lacked a comparator group, the effect of CGM use cannot be definitively isolated from other factors (eg, medication changes, dietary modifications, or physical activity). Moreover, CGM devices can be costly and are not universally covered by all insurance or IHS programs, potentially limiting widespread implementation. Policy-level restrictions and patient-specific barriers may also hinder feasibility in other settings.

The small sample size may limit the generalizability of the findings. Of the initial 302 patients, about 69% were excluded due to insulin use or incomplete laboratory data. A ± 4-month window was selected to balance data quality with real-world practices. Extending this window further (eg, ± 6 months) might have included more participants but risked diluting the 1-year endpoint consistency. The lack of statistical significance in secondary metrics may be due to insufficient power rather than the absence of an effect.

Exclusion of patients due to incomplete data may have introduced selection bias. However, patients were included in the overall analysis if they met the criteria for HbA_1c and CGM use, even if they lacked data for secondary outcomes. Additionally, the laboratory’s upper reporting limit for HbA_1c was 14%, with values above this reported as “> 14%.” For analysis, these were recorded as 14.1%, which may underestimate the true baseline HbA_1c levels and impact of the assessment of change. This occurred for 4 of the 93 patients included.

All patients used the Freestyle Libre CGM, which may limit the generalizability of the findings to other CGM brands or models. Differences in device features, accuracy, scanning frequency, and user experience may influence outcomes, and results might differ with other CGM technologies. The dataset did not include patients’ scanning frequency because this metric was not consistently included in the EHRs.

Conclusions

This study found that CGM use was significantly associated with improved glycemic control in patients with non–insulin-dependent T2DM within an AI/AN population, particularly among patients with higher baseline HbA_1c levels. The findings suggest that CGMs may be a valuable tool for managing T2DM beyond insulin-dependent populations.

Additional research with larger sample sizes, control groups, and extended follow-up periods is recommended to explore long-term benefits and impacts on other health metrics. The sample size estimates derived from this study serve as a valuable resource for researchers designing future studies aimed at addressing these gaps. Future research that expands on our findings by including larger, more diverse cohorts, accounting for medication use, and exploring different CGM technologies will enhance understanding and contribute to more effective diabetes management strategies for varied populations.

Diabetes mellitus (DM) is a national health crisis affecting > 38 million people (11.6%) in the United States.¹ American Indian and Alaska Native (AI/AN) adults are disproportionately affected, with a prevalence of 14.5%—the highest among all racial and ethnic groups.¹ Type 2 DM (T2DM) accounts for 90% to 95% of all DM cases and is a leading cause of morbidity and mortality due to its association with cardiovascular disease, kidney failure, and other complications.²

Maintaining glycemic control is important for managing T2DM and preventing microvascular and macrovascular complications.³ The cornerstone of diabetes self-management has been patient self-monitored blood glucose (SMBG) using finger-stick glucometers.⁴ However, SMBG provides measurements from a single point in time and requires frequent, painful, and inconvenient finger pricks, leading to decreased adherence.^5,6 These limitations negatively affect patient engagement and overall glycemic control.⁷

Continuous glucose monitors (CGMs) offer real-time, continuous glucose readings and trends.⁸ CGMs improve glycemic control and reduce hypoglycemic episodes in patients who are insulin-dependent.^9,10 Flash glucose monitors, a type of CGM that requires scanning to obtain glucose readings, provide similar benefits.¹¹ Despite these demonstrated advantages, research has primarily focused on insulin-dependent populations, leaving a significant gap in understanding the effect of CGMs on patients with T2DM who are not insulin-dependent.¹²

Given the high prevalence of T2DM among AI/AN populations and the potential benefits of CGMs, this study sought to evaluate the effect of CGM use on glycemic control and other health metrics in patients with non–insulin-dependent T2DM in an AI/AN population. This focus addresses a critical knowledge gap and may inform clinical practices and policies to improve diabetes management in this high-risk group.

Methods

A retrospective observational study was conducted using deidentified electronic health records (EHRs) from 2019 to 2024 at a federally operated outpatient Indian Health Service (IHS) clinic serving an AI/AN population in the IHS Portland Area (Oregon, Washington, Idaho). The study protocol was reviewed and deemed exempt by institutional review boards at Washington State University and the Portland Area IHS.

Study Population

This study included patients diagnosed with non–insulin-dependent T2DM, had used a CGM for ≥ 1 year, and had hemoglobin A_1c (HbA_1c) measurements within 4 months prior to CGM initiation (baseline) and within ± 4 months after 1 year of CGM use. For other health metrics, including blood pressure (BP), weight, low-density lipoprotein cholesterol (LDL-C), and estimated glomerular filtration rate (eGFR), this study required measurements within 6 months before CGM initiation and within 6 months after 1 year of CGM use. The baseline HbA_1c in the dataset ranged from 5.3% to > 14%.

Patients were excluded if they used insulin during the study period, had incomplete laboratory or clinical data for the required time frame, or had < 1 year of CGM use. The dataset did not include detailed information on oral DM medications; thus, we could not report or account for the type or number of oral hypoglycemic agents used by the patients. The IHS clinical applications coordinator compiled the dataset from the EHR, identifying patients who were prescribed and received a CGM at the clinic. All patients used the Abbott Freestyle Libre CGM, the only formulary CGM available at the clinic during the study period.

A 1-year follow-up endpoint was selected for several reasons: (1) to capture potential seasonal variations in diet and activity; (2) to align with the clinic’s standard practice of annual comprehensive diabetes evaluations; and (3) to allow sufficient time for patients to adapt to CGM use and reflect any meaningful changes in glycemic control.

All patients received standard DM care according to clinic protocols, which included DM self-management education and training. Patients met with the diabetes educator at least once, during which the educator emphasized making informed decisions using CGM data, such as adjusting dietary choices and physical activity levels to manage blood glucose concentrations effectively.

A total of 302 patients were initially identified. After applying exclusion criteria, 132 were excluded due to insulin use, and 77 were excluded due to incomplete HbA_1c data within the specified time frames (Figure 1). The final sample included 93 patients.

Measures

The primary outcome was the change in HbA_1c levels from baseline to 1 year after CGM initiation. Secondary outcomes included changes in weight, systolic and diastolic BP, LDL-C concentrations, and eGFR. For the primary outcome, HbA_1c values were collected within a grace period of ± 4 months from the baseline and 1-year time points. The laboratory’s upper reporting limit for HbA_1c was 14%; values reported as “> 14%” were recorded as 14.1% for data analysis, although the actual values could have been higher.

For secondary outcomes, data were included if measurements were obtained within ± 6 months of the baseline and 1-year time points. Patients who did not have measurements within these time frames for specific metrics were excluded from secondary outcome analysis but remained in the overall study if they met the criteria for HbA_1c and CGM use.

Statistical Analysis

Statistical analysis was performed using R statistical software version 4.4.2. Paired t tests were conducted to compare baseline and 1-year follow- up measurements for variables with parametric distributions. Wilcoxon signed-rank test was used for nonparametric data. A linear regression analysis was conducted to examine the relationship between baseline HbA_1c levels and the change in HbA_1c after 1 year of CGM use. Differences were considered significant at P < .05 set a priori. To guide future research, a posthoc power analysis was performed using Cohen’s d to estimate the required sample sizes for detecting significant effects, assuming a similar population.

Results

The study included 93 patients, with a mean (SD) age of 55 (13) years (range, 29-83 years). Of the participants, 56 were female (60%) and 37 were male (40%). All participants were identified as AI/AN and had non–insulin-dependent T2DM.

Primary Outcomes

A significant reduction in HbA_1c levels was observed after 1 year of CGM use. The mean (SD) baseline HbA1c was 9.5% (2.4%), which decreased to 7.6% (2.2%) at 1-year follow-up (Table 1). This difference represents a mean change of -1.86% (2.4%) (95% CI, -2.35 to -1.37; P < .001 [paired t test, -7.53]).

A linear regression model evaluated the relationship between baseline HbA_1c (predictor) and the change in HbA_1c after 1 year (outcome). The change in HbA_1c was calculated as the difference between 1-year follow-up and baseline values. The regression model revealed a significant negative association between baseline HbA_1c and the change in HbA_1c (Β = -0.576; P < .001), indicating that higher baseline HbA_1c values were associated with greater reductions in HbA_1c over the year. The regression equation was: Change in HbA_1c = 3.587 – 0.576 × Baseline HbA_1c

The regression coefficient for baseline HbA_1c was -0.576 (standard error, 0.083; t = -6.931; P < .001), indicating that for each 1% increase in baseline HbA_1c, the reduction of HbA_1c after 1 year increased by approximately 0.576% (Figure 2). The model explained 34.6% of the variance in HbA_1c change (R² = .345; adjusted R² = .338).

Secondary Outcomes

Systolic BP decreased by a mean (SD) -4.9 (17) mm Hg; 95% CI, -8.6 to -1.11; P = .01, paired t test). However, no significant change was observed for diastolic BP (P = .77, paired t test). Similarly, no significant changes were observed in weight, LDL-C concentrations, or eGFR after 1 year of CGM use. A posthoc power analysis indicated that the study was underpowered to detect smaller effect sizes in secondary outcomes. For example, sample size estimates indicated that detecting significant changes in weight and LDL-C concentrations would require sample sizes of 152 and 220 patients, respectively (Table 2).

Discussion

This study found a clinically significant reduction in HbA_1c levels after 1 year among AI/AN patients with non–insulin-dependent T2DM who used CGMs. The mean HbA_1c decreased 1.9%, from 9.5% at baseline to 7.6% after 1 year. This reduction is not only statistically significant (P < .001), it is clinically meaningful—even a 1% decrease in HbA_1c is associated with substantial reductions in the risk of microvascular complications.³ The magnitude of the HbA_1c reduction observed suggests CGM use may be associated with improved glycemic control in this high-risk population. By achieving lower HbA_1c levels, patients may experience improved long-term health outcomes and a reduced burden of DM-related complications.

Changes in oral DM medications during the study period may have contributed to the observed improvements in HbA_1c levels. While the dataset lacked detailed information on types or dosages of oral hypoglycemic agents used, adjustments in medication regimens are common in DM management and could significantly affect glycemic control. The inability to account for these changes results in an inability to attribute the improvements in HbA_1c solely to CGM use. Future studies should collect comprehensive medication data to better isolate the effects of CGM use from other treatment modifications.

Another factor that may have contributed to the improved glycemic control is the DM self-management education and training patients received as part of standard care. Patients met with diabetes educators at least once and learned how to use the CGM device and interpret the data for self-management decisions. This education may have enhanced patient engagement and empowerment, enabling them to make informed choices about diet, physical activity, and medication adherence. Studies have shown that DM self-management education can significantly improve glycemic control and patient outcomes.¹³ By combining the CGM technology with targeted education, patients may have been better equipped to manage their condition, contributing to the observed reduction in HbA_1c levels. Future studies should consider synergistic effects of CGM use and DM education when evaluating interventions for glycemic control.

The significant reduction in HbA_1c indicates CGM use is associated with improved glycemic control in non–insulin-dependent T2DM. The linear regression analysis suggests patients with poorer glycemic control at baseline experienced greater reductions in HbA_1c over the course of 1 year. This finding aligns with previous studies that have shown greater HbA_1c reductions in patients with higher initial levels when using CGMs. Yaron et al reported similar findings: higher baseline HbA_1c levels predicted more substantial improvements with CGM use in patients with T2DM on insulin therapy.¹⁴

This study contributes to existing research by examining the association between CGM use and glycemic control in patients with non– insulin-dependent T2DM within an AI/AN population, a group that has been underreported in previous studies. Most prior research has focused on insulin-dependent patients or populations with different ethnic backgrounds.12 By focusing on patients with non–insulin-dependent T2DM, this study highlights the broader applicability of CGMs beyond traditional use, showcasing their potential association with benefits in earlier stages of DM management. Targeting the AI/AN population addresses a critical knowledge gap, given the disproportionately high prevalence of T2DM and associated complications in this group. The findings of this study suggest integrating CGM technology into the standard care of AI/AN patients with non–insulin-dependent T2DM may be associated with improved glycemic control and may help reduce health disparities.

The modest decrease in systolic BP observed in this study may indicate potential cardiovascular benefits associated with CGM use, possibly due to improved glycemic control and increased patient engagement in self-management. However, given the limited sample size and exclusion criteria, the study lacked sufficient power to detect significant associations between CGM use and other secondary outcomes such as BP, weight, LDL-C, and eGFR. Therefore, the significant finding with systolic BP should be interpreted with caution.

The lack of significant changes in secondary outcomes may be attributed to the study’s limited sample size and the relatively short duration for observing changes in these parameters. Larger studies are needed to assess the full impact of CGM on these variables. The required sample sizes for achieving adequate power in future studies were calculated, highlighting the utility of our study as a pilot, providing critical data for the design of larger, adequately powered studies.

Limitations

The retrospective design of this study limits causal inferences. Moreover, potential confounding variables were not controlled, such as changes in medication regimens (other than insulin use), dietary counseling, or physical activity. Additionally, we could not account for the type or number of oral DM medications prescribed to patients. The dataset included only information on insulin use, without detailed records of other antidiabetic medications. This limitation may have influenced the observed change in glycemic control, as variations in medication regimens could affect HbA_1c levels.

Because this study lacked a comparator group, the effect of CGM use cannot be definitively isolated from other factors (eg, medication changes, dietary modifications, or physical activity). Moreover, CGM devices can be costly and are not universally covered by all insurance or IHS programs, potentially limiting widespread implementation. Policy-level restrictions and patient-specific barriers may also hinder feasibility in other settings.

The small sample size may limit the generalizability of the findings. Of the initial 302 patients, about 69% were excluded due to insulin use or incomplete laboratory data. A ± 4-month window was selected to balance data quality with real-world practices. Extending this window further (eg, ± 6 months) might have included more participants but risked diluting the 1-year endpoint consistency. The lack of statistical significance in secondary metrics may be due to insufficient power rather than the absence of an effect.

Exclusion of patients due to incomplete data may have introduced selection bias. However, patients were included in the overall analysis if they met the criteria for HbA_1c and CGM use, even if they lacked data for secondary outcomes. Additionally, the laboratory’s upper reporting limit for HbA_1c was 14%, with values above this reported as “> 14%.” For analysis, these were recorded as 14.1%, which may underestimate the true baseline HbA_1c levels and impact of the assessment of change. This occurred for 4 of the 93 patients included.

All patients used the Freestyle Libre CGM, which may limit the generalizability of the findings to other CGM brands or models. Differences in device features, accuracy, scanning frequency, and user experience may influence outcomes, and results might differ with other CGM technologies. The dataset did not include patients’ scanning frequency because this metric was not consistently included in the EHRs.

Conclusions

This study found that CGM use was significantly associated with improved glycemic control in patients with non–insulin-dependent T2DM within an AI/AN population, particularly among patients with higher baseline HbA_1c levels. The findings suggest that CGMs may be a valuable tool for managing T2DM beyond insulin-dependent populations.

Additional research with larger sample sizes, control groups, and extended follow-up periods is recommended to explore long-term benefits and impacts on other health metrics. The sample size estimates derived from this study serve as a valuable resource for researchers designing future studies aimed at addressing these gaps. Future research that expands on our findings by including larger, more diverse cohorts, accounting for medication use, and exploring different CGM technologies will enhance understanding and contribute to more effective diabetes management strategies for varied populations.

Cardiovascular disease (CVD) is the leading cause of death among women in the United States.¹ Most CVD is due to the buildup of plaque (ie, cholesterol, proteins, calcium, and inflammatory cells) in artery walls.² The plaque may lead to atherosclerotic cardiovascular disease (ASCVD), which includes coronary heart disease, cerebrovascular disease, peripheral artery disease, and aortic atherosclerotic disease.^2,3 Control and reduction of ASCVD risk factors, including high cholesterol levels, elevated blood pressure, insulin resistance, smoking, and a sedentary lifestyle, can contribute to a reduction in ASCVD morbidity and mortality.² People with type 2 diabetes mellitus (T2DM) have an increased prevalence of lipid abnormalities, contributing to their high risk of ASCVD.^4,5

The prescribing of statins (3-hydroxy-3-methyl-glutaryl-coenzmye A reductase inhibitors) is the cornerstone of lipid-lowering therapy and cardiovascular risk reduction for primary and secondary prevention of ASCVD.⁶ The American Diabetes Association (ADA) and American College of Cardiology/American Heart Association (ACC/AHA) recommend moderate- to high-intensity statins for primary prevention in patients with T2DM and high-intensity statins for secondary prevention in those with or without diabetes when not contraindicated.^4,5,7 Despite eligibility according to guideline recommendations, research predominantly shows that women are less likely to receive statin therapy; however, this trend is improving. ^[6,8-11] To explain the sex differences in statin use, Nanna et al found that there is a combination of women being offered statin therapy less frequently, declining therapy more frequently, and discontinuing treatment more frequently.¹¹ One possibility for discontinuing treatment could be statin-associated muscle symptoms (SAMS), which occur in about 10% of patients.¹² The incidence of adverse effects (AEs) may be related to the way statins are metabolized.

Pharmacogenomic testing is free for veterans through the US Department of Veterans Affairs (VA) PHASER program, which offers information and recommendations for a panel of 11 gene variants. The panel includes genes related to common medication classes such as anticoagulants, antiplatelets, proton pump inhibitors, nonsteroidal anti-inflammatory drugs, opioids, antidepressants, and statins. The VA PHASER panel includes the solute carrier organic anion transporter family member 1B1 (SLCO1B1) gene, which is predominantly expressed in the liver and facilitates the hepatic uptake of most statins.^13,14 A reduced function of SLCO1B1 can lead to higher statin levels, resulting in increased concentrations that may potentially cause SAMS.^13,14 Some alleles associated with reduced function include SLCO1B1*5, *15, *23, *31, and *46 to *49, whereas others are associated with increased function, such as SLCO1B1 *14 and *20 (Appendix).¹⁵ Supporting evidence shows the SLCO1B1*5 nucleotide polymorphism increases plasma levels of simvastatin and atorvastatin, affecting effectiveness or toxicity. ¹³ Females tend to have a lower body weight and higher percentage of body fat compared with males, which might lead to higher concentrations of lipophilic drugs, including atorvastatin and simvastatin, which may be exacerbated by decreased function of SLCO1B1*5.¹⁵ With pharmacogenomic testing, therapeutic recommendations can be made to improve the overall safety and efficacy of statins, thus improving adherence using a patient-specific approach.^14,15

Methods

Carl Vinson VA Medical Center (CVVAMC) serves about 42,000 veterans in Central and South Georgia, of which about 15% are female. Of the female veterans enrolled in care, 63% identify as Black, 27% White, and 1.5% as Asian, American Indian/Alaska Native, or Native Hawaiian/Other Pacific Islander. The 2020 Veterans Chartbook report showed that female veterans and minority racial and ethnic groups had worse access to health care and higher mortality rates than their male and non-Hispanic White counterparts.¹⁶

The Primary Care Equity Dashboard (PCED) was developed to engage the VA health care workforce in the process of identifying and addressing inequities in local patient populations.¹⁷ Using electronic quality measure data, the PCED provides Veterans Integrated Service Network-level and facility-level performance on several metrics.¹⁸ The PCED had not been previously used at the CVVAMC, and few publications or quality improvement projects regarding its use have been reported by the VA Office of Health Equity. PCED helped identify disparities when comparing female to male patients in the prescribing of statin therapy for patients with CVD and statin therapy for patients with T2DM.

VA PHASER pharmacogenomic analyses provided an opportunity to expand this quality improvement project. Sanford Health and the VA collaborated on the PHASER program to offer free genetic testing for veterans. The program launched in 2019 and expanded to various VA sites, including CVVAMC in March 2023. This program has been extended to December 31, 2025.

The primary objective of this quality improvement project was to increase statin prescribing among female veterans with T2DM and/or CVD to reduce cardiovascular risk. Secondary outcomes included increased pharmacogenomic testing and the assessment of pharmacogenomic results related to statin therapy. This project was approved by the CVVAMC Pharmacy and Therapeutics Committee. The PCED was used to identify female veterans with T2DM and/or CVD without an active prescription for a statin between July and October 2023. A review of Computerized Patient Record System patient charts was completed to screen for prespecified inclusion and exclusion criteria. Veterans were included if they were assigned female at birth, were enrolled in care at CVVAMC, and had a diagnosis of T2DM or CVD (history of myocardial infarction, coronary bypass graft, percutaneous coronary intervention, or other revascularization in any setting).

Veterans were excluded if they were currently pregnant, trying to conceive, breastfeeding, had a T1DM diagnosis, had previously documented hypersensitivity to a statin, active liver failure or decompensated cirrhosis, previously documented statin-associated rhabdomyolysis or autoimmune myopathy, an active prescription for a proprotein convertase subtilisin/kexin type 9 inhibitor, or previously documented statin intolerance (defined as the inability to tolerate ≥ 3 statins, with ≥ 1 prescribed at low intensity or alternate-day dosing). The female veterans were compared to 2 comparators: the facility's male veterans and the VA national average, identified via the PCED.

Once a veteran was screened, they were telephoned between October 2023 and February 2024 and provided education on statin use and pharmacogenomic testing using a standardized note template. An order was placed for participants who provided verbal consent for pharmacogenomic testing. Those who agreed to statin initiation were referred to a clinical pharmacist practitioner (CPP) who contacted them at a later date to prescribe a statin following the recommendations of the 2019 ACC/AHA and 2023 ADA guidelines and pharmacogenomic testing, if applicable.^4,5,7 Appropriate monitoring and follow-up occurred at the discretion of each CPP. Data collection included: age, race, diagnoses (T2DM, CVD, or both), baseline lipid panel (total cholesterol, triglycerides, high-density lipoprotein, low-density lipoprotein), hepatic function, name and dose of statin, reasons for declining statin therapy, and pharmacogenomic testing results related to SLCO1B1.

Results

At baseline in July 2023, 77.8% of female veterans with T2DM were prescribed a statin, which exceeded the national VA average (77.0%), but was below the rate for male veterans (78.7%) in the facility comparator group.17 Additionally, 82.2% of females with CVD were prescribed a statin, which was below the national VA average of 86.0% and the 84.9% of male veterans in the facility comparator group.17 The PCED identified 189 female veterans from July 2023 to October 2023 who may benefit from statin therapy. Thirty-three females met the exclusion criteria. Of the 156 included veterans, 129 (82.7%) were successfully contacted and 27 (17.3%) could not be reached by telephone after 3 attempts (Figure 1). The 129 female veterans contacted had a mean age of 59 years and the majority were Black (82.9%) (Table 1).

Primary Outcomes

Of the 129 contacted veterans, 31 (24.0%) had a non-VA statin prescription, 13 (10.1%) had an active VA statin prescription, and 85 (65.9%) did not have a statin prescription, despite being eligible. Statin adherence was confirmed with participants, and the medication list was updated accordingly.

Of the 85 veterans with no active statin therapy, 37 (43.5%) accepted a new statin prescription and 48 (56.5%) declined. There were various reasons provided for declining statin therapy: 17 participants (35.4%) declined due to concern for AEs (Table 2).

From July 2023 to March 2024, the percentage of female veterans with active statin therapy with T2DM increased from 77.8% to 79.0%. For those with active statin therapy with CVD, usage increased from 82.2% to 90.2%, which exceeded the national VA average and facility male comparator group (Figures 2 and 3).¹⁷

Secondary Outcomes

Seventy-one of 129 veterans (55.0%) gave verbal consent, and 47 (66.2%) completed the pharmacogenomic testing; 58 (45.0%) declined. Five veterans (10.6%) had a known SLCO1B1 allele variant present. One veteran required a change in statin therapy based on the results (eAppendix).

Discussion

This project aimed to increase statin prescribing among female veterans with T2DM and/or CVD to reduce cardiovascular risk and increase pharmacogenomic testing using the PCED and care managed by CPPs. The results of this quality improvement project illustrated that both metrics have improved at CVVAMC as a result of the intervention. The results in both metrics now exceed the PCED national VA average, and the CVD metric also exceeds that of the facility male comparator group. While there was only a 1.2% increase from July 2023 to March 2024 for patients with T2DM, there was an 8.0% increase for patients with CVD. Despite standardized education on statin use, more veterans declined therapy than accepted it, mostly due to concern for AEs. Recording the reasons for declining statin therapy offered valuable insight that can be used in additional discussions with veterans and clinicians.

Pharmacogenomics gives clinicians the unique opportunity to take a proactive approach to better predict drug responses, potentially allowing for less trial and error with medications, fewer AEs, greater trust in the clinician, and improved medication adherence. The CPPs incorporated pharmacogenomic testing into their practice, which led to identifying 5 SLCO1B1 gene abnormalities. The PCED served as a powerful tool for advancing equity-focused quality improvement initiatives on a local level and was crucial in prioritizing the detection of veterans potentially receiving suboptimal care.

Limitations

The nature of “cold calls” made it challenging to establish contact for inclusion in this study. An alternative to increase engagement could have been scheduled phone or face-to-face visits. While the use of the PCED was crucial, data did not account for statins listed in the non-VA medication list. All 31 patients with statins prescribed outside the VA had a start date added to provide the most accurate representation of the data moving forward.

Another limitation in this project was its small sample size and population. CVVAMC serves about 6200 female veterans, with roughly 63% identifying as Black. The preponderance of Black individuals (83%) in this project is typical for the female patient population at CVVAMC but may not reflect the demographics of other populations. Other limitations to this project consisted of scheduling conflicts. Appointments for laboratory draws at community-based outpatient clinics were subject to availability, which resulted in some delay in completion of pharmacogenomic testing.

Conclusions

CPPs can help reduce inequity in health care delivery. Increased incorporation of the PCED into regular practice within the VA is recommended to continue addressing sex disparities in statin use, diabetes control, blood pressure management, cancer screenings, and vaccination needs. CVVAMC plans to expand its use through another quality improvement project focused on reducing sex disparities in blood pressure management. Improving educational resources made available to veterans on the importance of statin therapy and potential to mitigate AEs through use of the VA PHASER program also would be helpful. This project successfully improved CVVAMC metrics for female veterans appropriately prescribed statin therapy and increased access to pharmacogenomic testing. Most importantly, it helped close the sex-based gap in CVD risk reduction care.

Cardiovascular disease (CVD) is the leading cause of death among women in the United States.¹ Most CVD is due to the buildup of plaque (ie, cholesterol, proteins, calcium, and inflammatory cells) in artery walls.² The plaque may lead to atherosclerotic cardiovascular disease (ASCVD), which includes coronary heart disease, cerebrovascular disease, peripheral artery disease, and aortic atherosclerotic disease.^2,3 Control and reduction of ASCVD risk factors, including high cholesterol levels, elevated blood pressure, insulin resistance, smoking, and a sedentary lifestyle, can contribute to a reduction in ASCVD morbidity and mortality.² People with type 2 diabetes mellitus (T2DM) have an increased prevalence of lipid abnormalities, contributing to their high risk of ASCVD.^4,5

The prescribing of statins (3-hydroxy-3-methyl-glutaryl-coenzmye A reductase inhibitors) is the cornerstone of lipid-lowering therapy and cardiovascular risk reduction for primary and secondary prevention of ASCVD.⁶ The American Diabetes Association (ADA) and American College of Cardiology/American Heart Association (ACC/AHA) recommend moderate- to high-intensity statins for primary prevention in patients with T2DM and high-intensity statins for secondary prevention in those with or without diabetes when not contraindicated.^4,5,7 Despite eligibility according to guideline recommendations, research predominantly shows that women are less likely to receive statin therapy; however, this trend is improving. ^[6,8-11] To explain the sex differences in statin use, Nanna et al found that there is a combination of women being offered statin therapy less frequently, declining therapy more frequently, and discontinuing treatment more frequently.¹¹ One possibility for discontinuing treatment could be statin-associated muscle symptoms (SAMS), which occur in about 10% of patients.¹² The incidence of adverse effects (AEs) may be related to the way statins are metabolized.

Pharmacogenomic testing is free for veterans through the US Department of Veterans Affairs (VA) PHASER program, which offers information and recommendations for a panel of 11 gene variants. The panel includes genes related to common medication classes such as anticoagulants, antiplatelets, proton pump inhibitors, nonsteroidal anti-inflammatory drugs, opioids, antidepressants, and statins. The VA PHASER panel includes the solute carrier organic anion transporter family member 1B1 (SLCO1B1) gene, which is predominantly expressed in the liver and facilitates the hepatic uptake of most statins.^13,14 A reduced function of SLCO1B1 can lead to higher statin levels, resulting in increased concentrations that may potentially cause SAMS.^13,14 Some alleles associated with reduced function include SLCO1B1*5, *15, *23, *31, and *46 to *49, whereas others are associated with increased function, such as SLCO1B1 *14 and *20 (Appendix).¹⁵ Supporting evidence shows the SLCO1B1*5 nucleotide polymorphism increases plasma levels of simvastatin and atorvastatin, affecting effectiveness or toxicity. ¹³ Females tend to have a lower body weight and higher percentage of body fat compared with males, which might lead to higher concentrations of lipophilic drugs, including atorvastatin and simvastatin, which may be exacerbated by decreased function of SLCO1B1*5.¹⁵ With pharmacogenomic testing, therapeutic recommendations can be made to improve the overall safety and efficacy of statins, thus improving adherence using a patient-specific approach.^14,15

Methods

Carl Vinson VA Medical Center (CVVAMC) serves about 42,000 veterans in Central and South Georgia, of which about 15% are female. Of the female veterans enrolled in care, 63% identify as Black, 27% White, and 1.5% as Asian, American Indian/Alaska Native, or Native Hawaiian/Other Pacific Islander. The 2020 Veterans Chartbook report showed that female veterans and minority racial and ethnic groups had worse access to health care and higher mortality rates than their male and non-Hispanic White counterparts.¹⁶

The Primary Care Equity Dashboard (PCED) was developed to engage the VA health care workforce in the process of identifying and addressing inequities in local patient populations.¹⁷ Using electronic quality measure data, the PCED provides Veterans Integrated Service Network-level and facility-level performance on several metrics.¹⁸ The PCED had not been previously used at the CVVAMC, and few publications or quality improvement projects regarding its use have been reported by the VA Office of Health Equity. PCED helped identify disparities when comparing female to male patients in the prescribing of statin therapy for patients with CVD and statin therapy for patients with T2DM.

VA PHASER pharmacogenomic analyses provided an opportunity to expand this quality improvement project. Sanford Health and the VA collaborated on the PHASER program to offer free genetic testing for veterans. The program launched in 2019 and expanded to various VA sites, including CVVAMC in March 2023. This program has been extended to December 31, 2025.

The primary objective of this quality improvement project was to increase statin prescribing among female veterans with T2DM and/or CVD to reduce cardiovascular risk. Secondary outcomes included increased pharmacogenomic testing and the assessment of pharmacogenomic results related to statin therapy. This project was approved by the CVVAMC Pharmacy and Therapeutics Committee. The PCED was used to identify female veterans with T2DM and/or CVD without an active prescription for a statin between July and October 2023. A review of Computerized Patient Record System patient charts was completed to screen for prespecified inclusion and exclusion criteria. Veterans were included if they were assigned female at birth, were enrolled in care at CVVAMC, and had a diagnosis of T2DM or CVD (history of myocardial infarction, coronary bypass graft, percutaneous coronary intervention, or other revascularization in any setting).

Veterans were excluded if they were currently pregnant, trying to conceive, breastfeeding, had a T1DM diagnosis, had previously documented hypersensitivity to a statin, active liver failure or decompensated cirrhosis, previously documented statin-associated rhabdomyolysis or autoimmune myopathy, an active prescription for a proprotein convertase subtilisin/kexin type 9 inhibitor, or previously documented statin intolerance (defined as the inability to tolerate ≥ 3 statins, with ≥ 1 prescribed at low intensity or alternate-day dosing). The female veterans were compared to 2 comparators: the facility's male veterans and the VA national average, identified via the PCED.

Once a veteran was screened, they were telephoned between October 2023 and February 2024 and provided education on statin use and pharmacogenomic testing using a standardized note template. An order was placed for participants who provided verbal consent for pharmacogenomic testing. Those who agreed to statin initiation were referred to a clinical pharmacist practitioner (CPP) who contacted them at a later date to prescribe a statin following the recommendations of the 2019 ACC/AHA and 2023 ADA guidelines and pharmacogenomic testing, if applicable.^4,5,7 Appropriate monitoring and follow-up occurred at the discretion of each CPP. Data collection included: age, race, diagnoses (T2DM, CVD, or both), baseline lipid panel (total cholesterol, triglycerides, high-density lipoprotein, low-density lipoprotein), hepatic function, name and dose of statin, reasons for declining statin therapy, and pharmacogenomic testing results related to SLCO1B1.

Results

At baseline in July 2023, 77.8% of female veterans with T2DM were prescribed a statin, which exceeded the national VA average (77.0%), but was below the rate for male veterans (78.7%) in the facility comparator group.17 Additionally, 82.2% of females with CVD were prescribed a statin, which was below the national VA average of 86.0% and the 84.9% of male veterans in the facility comparator group.17 The PCED identified 189 female veterans from July 2023 to October 2023 who may benefit from statin therapy. Thirty-three females met the exclusion criteria. Of the 156 included veterans, 129 (82.7%) were successfully contacted and 27 (17.3%) could not be reached by telephone after 3 attempts (Figure 1). The 129 female veterans contacted had a mean age of 59 years and the majority were Black (82.9%) (Table 1).

Primary Outcomes

Of the 129 contacted veterans, 31 (24.0%) had a non-VA statin prescription, 13 (10.1%) had an active VA statin prescription, and 85 (65.9%) did not have a statin prescription, despite being eligible. Statin adherence was confirmed with participants, and the medication list was updated accordingly.

Of the 85 veterans with no active statin therapy, 37 (43.5%) accepted a new statin prescription and 48 (56.5%) declined. There were various reasons provided for declining statin therapy: 17 participants (35.4%) declined due to concern for AEs (Table 2).

From July 2023 to March 2024, the percentage of female veterans with active statin therapy with T2DM increased from 77.8% to 79.0%. For those with active statin therapy with CVD, usage increased from 82.2% to 90.2%, which exceeded the national VA average and facility male comparator group (Figures 2 and 3).¹⁷

Secondary Outcomes

Seventy-one of 129 veterans (55.0%) gave verbal consent, and 47 (66.2%) completed the pharmacogenomic testing; 58 (45.0%) declined. Five veterans (10.6%) had a known SLCO1B1 allele variant present. One veteran required a change in statin therapy based on the results (eAppendix).

Discussion

This project aimed to increase statin prescribing among female veterans with T2DM and/or CVD to reduce cardiovascular risk and increase pharmacogenomic testing using the PCED and care managed by CPPs. The results of this quality improvement project illustrated that both metrics have improved at CVVAMC as a result of the intervention. The results in both metrics now exceed the PCED national VA average, and the CVD metric also exceeds that of the facility male comparator group. While there was only a 1.2% increase from July 2023 to March 2024 for patients with T2DM, there was an 8.0% increase for patients with CVD. Despite standardized education on statin use, more veterans declined therapy than accepted it, mostly due to concern for AEs. Recording the reasons for declining statin therapy offered valuable insight that can be used in additional discussions with veterans and clinicians.

Pharmacogenomics gives clinicians the unique opportunity to take a proactive approach to better predict drug responses, potentially allowing for less trial and error with medications, fewer AEs, greater trust in the clinician, and improved medication adherence. The CPPs incorporated pharmacogenomic testing into their practice, which led to identifying 5 SLCO1B1 gene abnormalities. The PCED served as a powerful tool for advancing equity-focused quality improvement initiatives on a local level and was crucial in prioritizing the detection of veterans potentially receiving suboptimal care.

Limitations

The nature of “cold calls” made it challenging to establish contact for inclusion in this study. An alternative to increase engagement could have been scheduled phone or face-to-face visits. While the use of the PCED was crucial, data did not account for statins listed in the non-VA medication list. All 31 patients with statins prescribed outside the VA had a start date added to provide the most accurate representation of the data moving forward.

Another limitation in this project was its small sample size and population. CVVAMC serves about 6200 female veterans, with roughly 63% identifying as Black. The preponderance of Black individuals (83%) in this project is typical for the female patient population at CVVAMC but may not reflect the demographics of other populations. Other limitations to this project consisted of scheduling conflicts. Appointments for laboratory draws at community-based outpatient clinics were subject to availability, which resulted in some delay in completion of pharmacogenomic testing.

Conclusions

CPPs can help reduce inequity in health care delivery. Increased incorporation of the PCED into regular practice within the VA is recommended to continue addressing sex disparities in statin use, diabetes control, blood pressure management, cancer screenings, and vaccination needs. CVVAMC plans to expand its use through another quality improvement project focused on reducing sex disparities in blood pressure management. Improving educational resources made available to veterans on the importance of statin therapy and potential to mitigate AEs through use of the VA PHASER program also would be helpful. This project successfully improved CVVAMC metrics for female veterans appropriately prescribed statin therapy and increased access to pharmacogenomic testing. Most importantly, it helped close the sex-based gap in CVD risk reduction care.

Cardiovascular disease (CVD) is the leading cause of death among women in the United States.¹ Most CVD is due to the buildup of plaque (ie, cholesterol, proteins, calcium, and inflammatory cells) in artery walls.² The plaque may lead to atherosclerotic cardiovascular disease (ASCVD), which includes coronary heart disease, cerebrovascular disease, peripheral artery disease, and aortic atherosclerotic disease.^2,3 Control and reduction of ASCVD risk factors, including high cholesterol levels, elevated blood pressure, insulin resistance, smoking, and a sedentary lifestyle, can contribute to a reduction in ASCVD morbidity and mortality.² People with type 2 diabetes mellitus (T2DM) have an increased prevalence of lipid abnormalities, contributing to their high risk of ASCVD.^4,5

The prescribing of statins (3-hydroxy-3-methyl-glutaryl-coenzmye A reductase inhibitors) is the cornerstone of lipid-lowering therapy and cardiovascular risk reduction for primary and secondary prevention of ASCVD.⁶ The American Diabetes Association (ADA) and American College of Cardiology/American Heart Association (ACC/AHA) recommend moderate- to high-intensity statins for primary prevention in patients with T2DM and high-intensity statins for secondary prevention in those with or without diabetes when not contraindicated.^4,5,7 Despite eligibility according to guideline recommendations, research predominantly shows that women are less likely to receive statin therapy; however, this trend is improving. ^[6,8-11] To explain the sex differences in statin use, Nanna et al found that there is a combination of women being offered statin therapy less frequently, declining therapy more frequently, and discontinuing treatment more frequently.¹¹ One possibility for discontinuing treatment could be statin-associated muscle symptoms (SAMS), which occur in about 10% of patients.¹² The incidence of adverse effects (AEs) may be related to the way statins are metabolized.

Pharmacogenomic testing is free for veterans through the US Department of Veterans Affairs (VA) PHASER program, which offers information and recommendations for a panel of 11 gene variants. The panel includes genes related to common medication classes such as anticoagulants, antiplatelets, proton pump inhibitors, nonsteroidal anti-inflammatory drugs, opioids, antidepressants, and statins. The VA PHASER panel includes the solute carrier organic anion transporter family member 1B1 (SLCO1B1) gene, which is predominantly expressed in the liver and facilitates the hepatic uptake of most statins.^13,14 A reduced function of SLCO1B1 can lead to higher statin levels, resulting in increased concentrations that may potentially cause SAMS.^13,14 Some alleles associated with reduced function include SLCO1B1*5, *15, *23, *31, and *46 to *49, whereas others are associated with increased function, such as SLCO1B1 *14 and *20 (Appendix).¹⁵ Supporting evidence shows the SLCO1B1*5 nucleotide polymorphism increases plasma levels of simvastatin and atorvastatin, affecting effectiveness or toxicity. ¹³ Females tend to have a lower body weight and higher percentage of body fat compared with males, which might lead to higher concentrations of lipophilic drugs, including atorvastatin and simvastatin, which may be exacerbated by decreased function of SLCO1B1*5.¹⁵ With pharmacogenomic testing, therapeutic recommendations can be made to improve the overall safety and efficacy of statins, thus improving adherence using a patient-specific approach.^14,15

Methods

Carl Vinson VA Medical Center (CVVAMC) serves about 42,000 veterans in Central and South Georgia, of which about 15% are female. Of the female veterans enrolled in care, 63% identify as Black, 27% White, and 1.5% as Asian, American Indian/Alaska Native, or Native Hawaiian/Other Pacific Islander. The 2020 Veterans Chartbook report showed that female veterans and minority racial and ethnic groups had worse access to health care and higher mortality rates than their male and non-Hispanic White counterparts.¹⁶

The Primary Care Equity Dashboard (PCED) was developed to engage the VA health care workforce in the process of identifying and addressing inequities in local patient populations.¹⁷ Using electronic quality measure data, the PCED provides Veterans Integrated Service Network-level and facility-level performance on several metrics.¹⁸ The PCED had not been previously used at the CVVAMC, and few publications or quality improvement projects regarding its use have been reported by the VA Office of Health Equity. PCED helped identify disparities when comparing female to male patients in the prescribing of statin therapy for patients with CVD and statin therapy for patients with T2DM.

VA PHASER pharmacogenomic analyses provided an opportunity to expand this quality improvement project. Sanford Health and the VA collaborated on the PHASER program to offer free genetic testing for veterans. The program launched in 2019 and expanded to various VA sites, including CVVAMC in March 2023. This program has been extended to December 31, 2025.

The primary objective of this quality improvement project was to increase statin prescribing among female veterans with T2DM and/or CVD to reduce cardiovascular risk. Secondary outcomes included increased pharmacogenomic testing and the assessment of pharmacogenomic results related to statin therapy. This project was approved by the CVVAMC Pharmacy and Therapeutics Committee. The PCED was used to identify female veterans with T2DM and/or CVD without an active prescription for a statin between July and October 2023. A review of Computerized Patient Record System patient charts was completed to screen for prespecified inclusion and exclusion criteria. Veterans were included if they were assigned female at birth, were enrolled in care at CVVAMC, and had a diagnosis of T2DM or CVD (history of myocardial infarction, coronary bypass graft, percutaneous coronary intervention, or other revascularization in any setting).

Veterans were excluded if they were currently pregnant, trying to conceive, breastfeeding, had a T1DM diagnosis, had previously documented hypersensitivity to a statin, active liver failure or decompensated cirrhosis, previously documented statin-associated rhabdomyolysis or autoimmune myopathy, an active prescription for a proprotein convertase subtilisin/kexin type 9 inhibitor, or previously documented statin intolerance (defined as the inability to tolerate ≥ 3 statins, with ≥ 1 prescribed at low intensity or alternate-day dosing). The female veterans were compared to 2 comparators: the facility's male veterans and the VA national average, identified via the PCED.

Once a veteran was screened, they were telephoned between October 2023 and February 2024 and provided education on statin use and pharmacogenomic testing using a standardized note template. An order was placed for participants who provided verbal consent for pharmacogenomic testing. Those who agreed to statin initiation were referred to a clinical pharmacist practitioner (CPP) who contacted them at a later date to prescribe a statin following the recommendations of the 2019 ACC/AHA and 2023 ADA guidelines and pharmacogenomic testing, if applicable.^4,5,7 Appropriate monitoring and follow-up occurred at the discretion of each CPP. Data collection included: age, race, diagnoses (T2DM, CVD, or both), baseline lipid panel (total cholesterol, triglycerides, high-density lipoprotein, low-density lipoprotein), hepatic function, name and dose of statin, reasons for declining statin therapy, and pharmacogenomic testing results related to SLCO1B1.

Results

At baseline in July 2023, 77.8% of female veterans with T2DM were prescribed a statin, which exceeded the national VA average (77.0%), but was below the rate for male veterans (78.7%) in the facility comparator group.17 Additionally, 82.2% of females with CVD were prescribed a statin, which was below the national VA average of 86.0% and the 84.9% of male veterans in the facility comparator group.17 The PCED identified 189 female veterans from July 2023 to October 2023 who may benefit from statin therapy. Thirty-three females met the exclusion criteria. Of the 156 included veterans, 129 (82.7%) were successfully contacted and 27 (17.3%) could not be reached by telephone after 3 attempts (Figure 1). The 129 female veterans contacted had a mean age of 59 years and the majority were Black (82.9%) (Table 1).

Primary Outcomes

Of the 129 contacted veterans, 31 (24.0%) had a non-VA statin prescription, 13 (10.1%) had an active VA statin prescription, and 85 (65.9%) did not have a statin prescription, despite being eligible. Statin adherence was confirmed with participants, and the medication list was updated accordingly.

Of the 85 veterans with no active statin therapy, 37 (43.5%) accepted a new statin prescription and 48 (56.5%) declined. There were various reasons provided for declining statin therapy: 17 participants (35.4%) declined due to concern for AEs (Table 2).

From July 2023 to March 2024, the percentage of female veterans with active statin therapy with T2DM increased from 77.8% to 79.0%. For those with active statin therapy with CVD, usage increased from 82.2% to 90.2%, which exceeded the national VA average and facility male comparator group (Figures 2 and 3).¹⁷

Secondary Outcomes

Seventy-one of 129 veterans (55.0%) gave verbal consent, and 47 (66.2%) completed the pharmacogenomic testing; 58 (45.0%) declined. Five veterans (10.6%) had a known SLCO1B1 allele variant present. One veteran required a change in statin therapy based on the results (eAppendix).

Discussion

This project aimed to increase statin prescribing among female veterans with T2DM and/or CVD to reduce cardiovascular risk and increase pharmacogenomic testing using the PCED and care managed by CPPs. The results of this quality improvement project illustrated that both metrics have improved at CVVAMC as a result of the intervention. The results in both metrics now exceed the PCED national VA average, and the CVD metric also exceeds that of the facility male comparator group. While there was only a 1.2% increase from July 2023 to March 2024 for patients with T2DM, there was an 8.0% increase for patients with CVD. Despite standardized education on statin use, more veterans declined therapy than accepted it, mostly due to concern for AEs. Recording the reasons for declining statin therapy offered valuable insight that can be used in additional discussions with veterans and clinicians.

Pharmacogenomics gives clinicians the unique opportunity to take a proactive approach to better predict drug responses, potentially allowing for less trial and error with medications, fewer AEs, greater trust in the clinician, and improved medication adherence. The CPPs incorporated pharmacogenomic testing into their practice, which led to identifying 5 SLCO1B1 gene abnormalities. The PCED served as a powerful tool for advancing equity-focused quality improvement initiatives on a local level and was crucial in prioritizing the detection of veterans potentially receiving suboptimal care.

Limitations

The nature of “cold calls” made it challenging to establish contact for inclusion in this study. An alternative to increase engagement could have been scheduled phone or face-to-face visits. While the use of the PCED was crucial, data did not account for statins listed in the non-VA medication list. All 31 patients with statins prescribed outside the VA had a start date added to provide the most accurate representation of the data moving forward.

Another limitation in this project was its small sample size and population. CVVAMC serves about 6200 female veterans, with roughly 63% identifying as Black. The preponderance of Black individuals (83%) in this project is typical for the female patient population at CVVAMC but may not reflect the demographics of other populations. Other limitations to this project consisted of scheduling conflicts. Appointments for laboratory draws at community-based outpatient clinics were subject to availability, which resulted in some delay in completion of pharmacogenomic testing.

Conclusions

CPPs can help reduce inequity in health care delivery. Increased incorporation of the PCED into regular practice within the VA is recommended to continue addressing sex disparities in statin use, diabetes control, blood pressure management, cancer screenings, and vaccination needs. CVVAMC plans to expand its use through another quality improvement project focused on reducing sex disparities in blood pressure management. Improving educational resources made available to veterans on the importance of statin therapy and potential to mitigate AEs through use of the VA PHASER program also would be helpful. This project successfully improved CVVAMC metrics for female veterans appropriately prescribed statin therapy and increased access to pharmacogenomic testing. Most importantly, it helped close the sex-based gap in CVD risk reduction care.

A recent worldwide survey found the United States to have the highest reported prevalence of type 2 diabetes mellitus (T2DM) among young people aged 10 to 19 years. Research on the prevalence of the disease among Indigenous populations is scarce, however, leaving these individuals at a potentially greater risk.

The estimated prevalence of T2DM has nearly doubled over the past 2 decades, with cases per 1000 youths aged 10 to 19 years increasing from 0.34 in 2001 to 0.46 in 2009 to 0.67 in 2017, a relative increase of 95.3% over 16 years. In 2012, the SEARCH study of youth-onset T2DM found American Indians and non-Hispanic Black individuals had the highest incidence (46.5/100,000/year in American Indians and 32.6/100,000/year in non-Hispanic Black individuals), compared with non-Hispanic White individuals (3.9/100,000/year).

About 28,000 US youth aged < 20 years had T2DM in 2017, a figure expected to reach 48,000 in 2060 based on increasing prevalence and incidence rates. Assuming the trends observed between 2002 and 2017 continue, an estimated 220,000 young people will have T2DM. 

However, the lack of recent research of T2DM in young indigenous populations may have masked a serious problem among Native Americans. A 2025 literature review of 49 studies call it a “type 2 diabetes crisis” among Indigenous communities; not because of the disease, but due to high rates of complications. Though Indigenous peoples are estimated to inhabit > 90 countries and collectively represent > 370 million people, the studies included in the review involved individuals from 6 countries and 2 self-governing states (US, Canada, Australia, Aotearoa New Zealand, Nauru, Argentina, the Cook Islands, and Niue) and at least 45 Indigenous populations after search criteria were satisfied. Data were derived from population-based screening and health databases, including from 432 IHS facilities and 6 IHS regions.

Of the study populations, 27 (75%) reported diabetes prevalence above 1 per 1000. Age-specific data, available in 44 studies, showed increased prevalence with age: 0 to 4 per 1000 at age < 10 years; 0 to 44 per 1000 at age 10 to 19 years; and 0 to 64 per 1000 at age 15 to 25 years. 

In young adults aged 15 to 25 years, prevalence was highest in Akimel O’odham and Tohono O’odham Peoples from the Gila River Indian Community in Arizona. Among children aged < 10 years, the highest prevalence was reported in Cherokee Nation children. Some groups reported no diabetes, such as the Northern Plains Indians from Montana and Wyoming.

Statistics showing the speed of expanding prevalence were particularly notable. For Akimel O’odham and Tohono O’odham Indian youth, diabetes prevalence increased more than eightfold over 2 decades (particularly in those aged < 15).

A 2021 study of 500 participants who were diagnosed with T2DM in youth were followed for a mean of 13 years. By the time they were 26, 67.5% had hypertension, 51.6% had dyslipidemia, 54.8% had diabetic kidney disease, and 32.4% had nerve disease. 

Indigenous North American children may also have an even greater risk for later complications. A Canadian study found that among Canadian First Nations Peoples the incidence of end-stage kidney disease was 2.8 times higher and the mortality rate was double that of non-Indigenous people with youth-onset T2DM despite similar age at diagnosis and duration of disease. 

To combat the steady increase of T2DM prevalence among Indigenous youth, researchers advise “urgent action” to improve data equity through the inclusion of Indigenous populations in health surveillance, routine disaggregation by Indigenous status, and culturally safe research partnerships led by Indigenous communities. Standardized age group classifications, age- and gender-specific reporting, and assessment of comorbid obesity are essential, they add, to define health care needs and identify regions that would benefit from enhanced early detection and management.

A recent worldwide survey found the United States to have the highest reported prevalence of type 2 diabetes mellitus (T2DM) among young people aged 10 to 19 years. Research on the prevalence of the disease among Indigenous populations is scarce, however, leaving these individuals at a potentially greater risk.

The estimated prevalence of T2DM has nearly doubled over the past 2 decades, with cases per 1000 youths aged 10 to 19 years increasing from 0.34 in 2001 to 0.46 in 2009 to 0.67 in 2017, a relative increase of 95.3% over 16 years. In 2012, the SEARCH study of youth-onset T2DM found American Indians and non-Hispanic Black individuals had the highest incidence (46.5/100,000/year in American Indians and 32.6/100,000/year in non-Hispanic Black individuals), compared with non-Hispanic White individuals (3.9/100,000/year).

About 28,000 US youth aged < 20 years had T2DM in 2017, a figure expected to reach 48,000 in 2060 based on increasing prevalence and incidence rates. Assuming the trends observed between 2002 and 2017 continue, an estimated 220,000 young people will have T2DM. 

However, the lack of recent research of T2DM in young indigenous populations may have masked a serious problem among Native Americans. A 2025 literature review of 49 studies call it a “type 2 diabetes crisis” among Indigenous communities; not because of the disease, but due to high rates of complications. Though Indigenous peoples are estimated to inhabit > 90 countries and collectively represent > 370 million people, the studies included in the review involved individuals from 6 countries and 2 self-governing states (US, Canada, Australia, Aotearoa New Zealand, Nauru, Argentina, the Cook Islands, and Niue) and at least 45 Indigenous populations after search criteria were satisfied. Data were derived from population-based screening and health databases, including from 432 IHS facilities and 6 IHS regions.

Of the study populations, 27 (75%) reported diabetes prevalence above 1 per 1000. Age-specific data, available in 44 studies, showed increased prevalence with age: 0 to 4 per 1000 at age < 10 years; 0 to 44 per 1000 at age 10 to 19 years; and 0 to 64 per 1000 at age 15 to 25 years. 

In young adults aged 15 to 25 years, prevalence was highest in Akimel O’odham and Tohono O’odham Peoples from the Gila River Indian Community in Arizona. Among children aged < 10 years, the highest prevalence was reported in Cherokee Nation children. Some groups reported no diabetes, such as the Northern Plains Indians from Montana and Wyoming.

Statistics showing the speed of expanding prevalence were particularly notable. For Akimel O’odham and Tohono O’odham Indian youth, diabetes prevalence increased more than eightfold over 2 decades (particularly in those aged < 15).

A 2021 study of 500 participants who were diagnosed with T2DM in youth were followed for a mean of 13 years. By the time they were 26, 67.5% had hypertension, 51.6% had dyslipidemia, 54.8% had diabetic kidney disease, and 32.4% had nerve disease. 

Indigenous North American children may also have an even greater risk for later complications. A Canadian study found that among Canadian First Nations Peoples the incidence of end-stage kidney disease was 2.8 times higher and the mortality rate was double that of non-Indigenous people with youth-onset T2DM despite similar age at diagnosis and duration of disease. 

To combat the steady increase of T2DM prevalence among Indigenous youth, researchers advise “urgent action” to improve data equity through the inclusion of Indigenous populations in health surveillance, routine disaggregation by Indigenous status, and culturally safe research partnerships led by Indigenous communities. Standardized age group classifications, age- and gender-specific reporting, and assessment of comorbid obesity are essential, they add, to define health care needs and identify regions that would benefit from enhanced early detection and management.

A recent worldwide survey found the United States to have the highest reported prevalence of type 2 diabetes mellitus (T2DM) among young people aged 10 to 19 years. Research on the prevalence of the disease among Indigenous populations is scarce, however, leaving these individuals at a potentially greater risk.

The estimated prevalence of T2DM has nearly doubled over the past 2 decades, with cases per 1000 youths aged 10 to 19 years increasing from 0.34 in 2001 to 0.46 in 2009 to 0.67 in 2017, a relative increase of 95.3% over 16 years. In 2012, the SEARCH study of youth-onset T2DM found American Indians and non-Hispanic Black individuals had the highest incidence (46.5/100,000/year in American Indians and 32.6/100,000/year in non-Hispanic Black individuals), compared with non-Hispanic White individuals (3.9/100,000/year).

About 28,000 US youth aged < 20 years had T2DM in 2017, a figure expected to reach 48,000 in 2060 based on increasing prevalence and incidence rates. Assuming the trends observed between 2002 and 2017 continue, an estimated 220,000 young people will have T2DM. 

However, the lack of recent research of T2DM in young indigenous populations may have masked a serious problem among Native Americans. A 2025 literature review of 49 studies call it a “type 2 diabetes crisis” among Indigenous communities; not because of the disease, but due to high rates of complications. Though Indigenous peoples are estimated to inhabit > 90 countries and collectively represent > 370 million people, the studies included in the review involved individuals from 6 countries and 2 self-governing states (US, Canada, Australia, Aotearoa New Zealand, Nauru, Argentina, the Cook Islands, and Niue) and at least 45 Indigenous populations after search criteria were satisfied. Data were derived from population-based screening and health databases, including from 432 IHS facilities and 6 IHS regions.

Of the study populations, 27 (75%) reported diabetes prevalence above 1 per 1000. Age-specific data, available in 44 studies, showed increased prevalence with age: 0 to 4 per 1000 at age < 10 years; 0 to 44 per 1000 at age 10 to 19 years; and 0 to 64 per 1000 at age 15 to 25 years. 

In young adults aged 15 to 25 years, prevalence was highest in Akimel O’odham and Tohono O’odham Peoples from the Gila River Indian Community in Arizona. Among children aged < 10 years, the highest prevalence was reported in Cherokee Nation children. Some groups reported no diabetes, such as the Northern Plains Indians from Montana and Wyoming.

Statistics showing the speed of expanding prevalence were particularly notable. For Akimel O’odham and Tohono O’odham Indian youth, diabetes prevalence increased more than eightfold over 2 decades (particularly in those aged < 15).

A 2021 study of 500 participants who were diagnosed with T2DM in youth were followed for a mean of 13 years. By the time they were 26, 67.5% had hypertension, 51.6% had dyslipidemia, 54.8% had diabetic kidney disease, and 32.4% had nerve disease. 

Indigenous North American children may also have an even greater risk for later complications. A Canadian study found that among Canadian First Nations Peoples the incidence of end-stage kidney disease was 2.8 times higher and the mortality rate was double that of non-Indigenous people with youth-onset T2DM despite similar age at diagnosis and duration of disease. 

To combat the steady increase of T2DM prevalence among Indigenous youth, researchers advise “urgent action” to improve data equity through the inclusion of Indigenous populations in health surveillance, routine disaggregation by Indigenous status, and culturally safe research partnerships led by Indigenous communities. Standardized age group classifications, age- and gender-specific reporting, and assessment of comorbid obesity are essential, they add, to define health care needs and identify regions that would benefit from enhanced early detection and management.

Click here to view more from Federal Health Care Data Trends 2025.

Click here to view more from Federal Health Care Data Trends 2025.

Click here to view more from Federal Health Care Data Trends 2025.

New data suggest that glucagon-like peptide 1 (GLP-1) receptor agonists, used to treat diabetes and obesity, may also help guard against obesity-related cancers.

In a large observational study, new GLP-1 agonist users with obesity and diabetes had a significantly lower risk for 14 obesity-related cancers than similar individuals who received dipeptidyl peptidase-4 (DPP-4) inhibitors, which are weight-neutral.

This study provides a “reassuring safety signal” showing that GLP-1 drugs are linked to a modest drop in obesity-related cancer risk, and not a higher risk for these cancers, said lead investigator Lucas Mavromatis, medical student at NYU Grossman School of Medicine in New York City, during a press conference at American Society of Clinical Oncology (ASCO) 2025 annual meeting.

However, there were some nuances to the findings. The protective effect of GLP-1 agonists was only significant for colon and rectal cancers and for women, Mavromatis reported. And although GLP-1 users had an 8% lower risk of dying from any cause, the survival benefit was also only significant for women.

Still, the overall “message to patients is GLP-1 receptor treatments remain a strong option for patients with diabetes and obesity and may have an additional, small favorable benefit in cancer,” Mavromatis explained at the press briefing.

'Intriguing Hypothesis'

Obesity is linked to an increased risk of developing more than a dozen cancer types, including esophageal, colon, rectal, stomach, liver, gallbladder, pancreatic, kidney, postmenopausal breast, ovarian, endometrial and thyroid, as well as multiple myeloma and meningiomas.

About 12% of Americans have been prescribed a GLP-1 medication to treat diabetes and/or obesity. However, little is known about how these drugs affect cancer risk.

To investigate, Mavromatis and colleagues used the Optum healthcare database to identify 170,030 adults with obesity and type 2 diabetes from 43 health systems in the United States.

Between 2013 and 2023, half started a GLP-1 agonist and half started a DPP-4 inhibitor, with propensity score matching used to balance characteristics of the two cohorts.

Participants were a mean age of 56.8 years, with an average body mass index of 38.5; more than 70% were White individuals and more than 14% were Black individuals.

During a mean follow-up of 3.9 years, 2501 new obesity-related cancers were identified in the GLP-1 group and 2671 in the DPP-4 group — representing a 7% overall reduced risk for any obesity-related cancer in the GLP-1 group (hazard ratio [HR], 0.93).

When analyzing each of the 14 obesity-related cancers separately, the protective link between GLP-1 use and cancer was primarily driven by colon and rectal cancers. GLP-1 users had a 16% lower risk for colon cancer (HR, 0.84) and a 28% lower risk for rectal cancer (HR, 0.72).

“No other cancers had statistically significant associations with GLP-1 use,” Mavromatis told briefing attendees. But “importantly, no cancers had statistically significant adverse associations with GLP-1 use,” he added.

Experts have expressed some concern about a possible link between GLP-1 use and pancreatic cancer given that pancreatitis is a known side effect of GLP-1 use. However, “this is not borne out by epidemiological data,” Mavromatis said.

“Additionally, we were not able to specifically assess medullary thyroid cancer, which is on the warning label for several GLP-1 medications, but we did see a reassuring lack of association between GLP-1 use and thyroid cancer as a whole,” he added.

During follow-up, there were 2783 deaths in the GLP-1 group and 2961 deaths in the DPP-4 group — translating to an 8% lower risk for death due to any cause among GLP-1 users (HR, 0.92; P = .001).

Mavromatis and colleagues observed sex differences as well. Women taking a GLP-1 had an 8% lower risk for obesity-related cancers (HR, 0.92; P = .01) and a 20% lower risk for death from any cause (HR, 0.80; P < .001) compared with women taking a DPP-4 inhibitor.

Among men, researchers found no statistically significant difference between GLP-1 and DPP-4 use for obesity-related cancer risk (HR, 0.95; P = .29) or all-cause mortality (HR, 1.04; P = .34).

Overall, Mavromatis said, it’s important to note that the absolute risk reduction seen in the study is “small and the number of patients that would need to be given one of these medications to prevent an obesity-related cancer, based on our data, would be very large.”

Mavromatis also noted that the length of follow-up was short, and the study assessed primarily older and weaker GLP-1 agonists compared with newer agents on the market. Therefore, longer-term studies with newer GLP-1s are needed to confirm the effects seen as well as safety.

In a statement, ASCO President Robin Zon, MD, said this trial raises the “intriguing hypothesis” that the increasingly popular GLP-1 medications might offer some benefit in reducing the risk of developing cancer.

Zon said she sees many patients with obesity, and given the clear link between cancer and obesity, defining the clinical role of GLP-1 medications in cancer prevention is “important.”

This study “leads us in the direction” of a potential protective effect of GLP-1s on cancer, but “there are a lot of questions that are generated by this particular study, especially as we move forward and we think about prevention of cancers,” Zon told the briefing.

This study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health. Mavromatis reported no relevant disclosures. Zon reported stock or ownership interests in Oncolytics Biotech, TG Therapeutics, Select Sector SPDR Health Care, AstraZeneca, CRISPR, McKesson, and Berkshire Hathaway.

A version of this article first appeared on Medscape.com.

New data suggest that glucagon-like peptide 1 (GLP-1) receptor agonists, used to treat diabetes and obesity, may also help guard against obesity-related cancers.

In a large observational study, new GLP-1 agonist users with obesity and diabetes had a significantly lower risk for 14 obesity-related cancers than similar individuals who received dipeptidyl peptidase-4 (DPP-4) inhibitors, which are weight-neutral.

This study provides a “reassuring safety signal” showing that GLP-1 drugs are linked to a modest drop in obesity-related cancer risk, and not a higher risk for these cancers, said lead investigator Lucas Mavromatis, medical student at NYU Grossman School of Medicine in New York City, during a press conference at American Society of Clinical Oncology (ASCO) 2025 annual meeting.

However, there were some nuances to the findings. The protective effect of GLP-1 agonists was only significant for colon and rectal cancers and for women, Mavromatis reported. And although GLP-1 users had an 8% lower risk of dying from any cause, the survival benefit was also only significant for women.

Still, the overall “message to patients is GLP-1 receptor treatments remain a strong option for patients with diabetes and obesity and may have an additional, small favorable benefit in cancer,” Mavromatis explained at the press briefing.

'Intriguing Hypothesis'

Obesity is linked to an increased risk of developing more than a dozen cancer types, including esophageal, colon, rectal, stomach, liver, gallbladder, pancreatic, kidney, postmenopausal breast, ovarian, endometrial and thyroid, as well as multiple myeloma and meningiomas.

About 12% of Americans have been prescribed a GLP-1 medication to treat diabetes and/or obesity. However, little is known about how these drugs affect cancer risk.

To investigate, Mavromatis and colleagues used the Optum healthcare database to identify 170,030 adults with obesity and type 2 diabetes from 43 health systems in the United States.

Between 2013 and 2023, half started a GLP-1 agonist and half started a DPP-4 inhibitor, with propensity score matching used to balance characteristics of the two cohorts.

Participants were a mean age of 56.8 years, with an average body mass index of 38.5; more than 70% were White individuals and more than 14% were Black individuals.

During a mean follow-up of 3.9 years, 2501 new obesity-related cancers were identified in the GLP-1 group and 2671 in the DPP-4 group — representing a 7% overall reduced risk for any obesity-related cancer in the GLP-1 group (hazard ratio [HR], 0.93).

When analyzing each of the 14 obesity-related cancers separately, the protective link between GLP-1 use and cancer was primarily driven by colon and rectal cancers. GLP-1 users had a 16% lower risk for colon cancer (HR, 0.84) and a 28% lower risk for rectal cancer (HR, 0.72).

“No other cancers had statistically significant associations with GLP-1 use,” Mavromatis told briefing attendees. But “importantly, no cancers had statistically significant adverse associations with GLP-1 use,” he added.

Experts have expressed some concern about a possible link between GLP-1 use and pancreatic cancer given that pancreatitis is a known side effect of GLP-1 use. However, “this is not borne out by epidemiological data,” Mavromatis said.

“Additionally, we were not able to specifically assess medullary thyroid cancer, which is on the warning label for several GLP-1 medications, but we did see a reassuring lack of association between GLP-1 use and thyroid cancer as a whole,” he added.

During follow-up, there were 2783 deaths in the GLP-1 group and 2961 deaths in the DPP-4 group — translating to an 8% lower risk for death due to any cause among GLP-1 users (HR, 0.92; P = .001).

Mavromatis and colleagues observed sex differences as well. Women taking a GLP-1 had an 8% lower risk for obesity-related cancers (HR, 0.92; P = .01) and a 20% lower risk for death from any cause (HR, 0.80; P < .001) compared with women taking a DPP-4 inhibitor.

Among men, researchers found no statistically significant difference between GLP-1 and DPP-4 use for obesity-related cancer risk (HR, 0.95; P = .29) or all-cause mortality (HR, 1.04; P = .34).

Overall, Mavromatis said, it’s important to note that the absolute risk reduction seen in the study is “small and the number of patients that would need to be given one of these medications to prevent an obesity-related cancer, based on our data, would be very large.”

Mavromatis also noted that the length of follow-up was short, and the study assessed primarily older and weaker GLP-1 agonists compared with newer agents on the market. Therefore, longer-term studies with newer GLP-1s are needed to confirm the effects seen as well as safety.

In a statement, ASCO President Robin Zon, MD, said this trial raises the “intriguing hypothesis” that the increasingly popular GLP-1 medications might offer some benefit in reducing the risk of developing cancer.

Zon said she sees many patients with obesity, and given the clear link between cancer and obesity, defining the clinical role of GLP-1 medications in cancer prevention is “important.”

This study “leads us in the direction” of a potential protective effect of GLP-1s on cancer, but “there are a lot of questions that are generated by this particular study, especially as we move forward and we think about prevention of cancers,” Zon told the briefing.

This study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health. Mavromatis reported no relevant disclosures. Zon reported stock or ownership interests in Oncolytics Biotech, TG Therapeutics, Select Sector SPDR Health Care, AstraZeneca, CRISPR, McKesson, and Berkshire Hathaway.

A version of this article first appeared on Medscape.com.

New data suggest that glucagon-like peptide 1 (GLP-1) receptor agonists, used to treat diabetes and obesity, may also help guard against obesity-related cancers.

In a large observational study, new GLP-1 agonist users with obesity and diabetes had a significantly lower risk for 14 obesity-related cancers than similar individuals who received dipeptidyl peptidase-4 (DPP-4) inhibitors, which are weight-neutral.

This study provides a “reassuring safety signal” showing that GLP-1 drugs are linked to a modest drop in obesity-related cancer risk, and not a higher risk for these cancers, said lead investigator Lucas Mavromatis, medical student at NYU Grossman School of Medicine in New York City, during a press conference at American Society of Clinical Oncology (ASCO) 2025 annual meeting.

However, there were some nuances to the findings. The protective effect of GLP-1 agonists was only significant for colon and rectal cancers and for women, Mavromatis reported. And although GLP-1 users had an 8% lower risk of dying from any cause, the survival benefit was also only significant for women.

Still, the overall “message to patients is GLP-1 receptor treatments remain a strong option for patients with diabetes and obesity and may have an additional, small favorable benefit in cancer,” Mavromatis explained at the press briefing.

'Intriguing Hypothesis'

Obesity is linked to an increased risk of developing more than a dozen cancer types, including esophageal, colon, rectal, stomach, liver, gallbladder, pancreatic, kidney, postmenopausal breast, ovarian, endometrial and thyroid, as well as multiple myeloma and meningiomas.

About 12% of Americans have been prescribed a GLP-1 medication to treat diabetes and/or obesity. However, little is known about how these drugs affect cancer risk.

To investigate, Mavromatis and colleagues used the Optum healthcare database to identify 170,030 adults with obesity and type 2 diabetes from 43 health systems in the United States.

Between 2013 and 2023, half started a GLP-1 agonist and half started a DPP-4 inhibitor, with propensity score matching used to balance characteristics of the two cohorts.

Participants were a mean age of 56.8 years, with an average body mass index of 38.5; more than 70% were White individuals and more than 14% were Black individuals.

During a mean follow-up of 3.9 years, 2501 new obesity-related cancers were identified in the GLP-1 group and 2671 in the DPP-4 group — representing a 7% overall reduced risk for any obesity-related cancer in the GLP-1 group (hazard ratio [HR], 0.93).

When analyzing each of the 14 obesity-related cancers separately, the protective link between GLP-1 use and cancer was primarily driven by colon and rectal cancers. GLP-1 users had a 16% lower risk for colon cancer (HR, 0.84) and a 28% lower risk for rectal cancer (HR, 0.72).

“No other cancers had statistically significant associations with GLP-1 use,” Mavromatis told briefing attendees. But “importantly, no cancers had statistically significant adverse associations with GLP-1 use,” he added.

Experts have expressed some concern about a possible link between GLP-1 use and pancreatic cancer given that pancreatitis is a known side effect of GLP-1 use. However, “this is not borne out by epidemiological data,” Mavromatis said.

“Additionally, we were not able to specifically assess medullary thyroid cancer, which is on the warning label for several GLP-1 medications, but we did see a reassuring lack of association between GLP-1 use and thyroid cancer as a whole,” he added.

During follow-up, there were 2783 deaths in the GLP-1 group and 2961 deaths in the DPP-4 group — translating to an 8% lower risk for death due to any cause among GLP-1 users (HR, 0.92; P = .001).

Mavromatis and colleagues observed sex differences as well. Women taking a GLP-1 had an 8% lower risk for obesity-related cancers (HR, 0.92; P = .01) and a 20% lower risk for death from any cause (HR, 0.80; P < .001) compared with women taking a DPP-4 inhibitor.

Among men, researchers found no statistically significant difference between GLP-1 and DPP-4 use for obesity-related cancer risk (HR, 0.95; P = .29) or all-cause mortality (HR, 1.04; P = .34).

Overall, Mavromatis said, it’s important to note that the absolute risk reduction seen in the study is “small and the number of patients that would need to be given one of these medications to prevent an obesity-related cancer, based on our data, would be very large.”

Mavromatis also noted that the length of follow-up was short, and the study assessed primarily older and weaker GLP-1 agonists compared with newer agents on the market. Therefore, longer-term studies with newer GLP-1s are needed to confirm the effects seen as well as safety.

In a statement, ASCO President Robin Zon, MD, said this trial raises the “intriguing hypothesis” that the increasingly popular GLP-1 medications might offer some benefit in reducing the risk of developing cancer.

Zon said she sees many patients with obesity, and given the clear link between cancer and obesity, defining the clinical role of GLP-1 medications in cancer prevention is “important.”

This study “leads us in the direction” of a potential protective effect of GLP-1s on cancer, but “there are a lot of questions that are generated by this particular study, especially as we move forward and we think about prevention of cancers,” Zon told the briefing.

This study was funded by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health. Mavromatis reported no relevant disclosures. Zon reported stock or ownership interests in Oncolytics Biotech, TG Therapeutics, Select Sector SPDR Health Care, AstraZeneca, CRISPR, McKesson, and Berkshire Hathaway.

A version of this article first appeared on Medscape.com.