Receipt of Glucagon-Like Peptide-1 Receptor Agonists Among Rural Patients with Obstructive Sleep Apnea and Excess Weight

Abstract

Rationale:

Successful weight loss is difficult for those with obstructive sleep apnea (OSA), and gaps in weight management care are more pronounced in rural areas. Glucagon-Like Peptide-1 Receptor Agonists (GLP-1RAs) are promising agents to improve the effectiveness of weight management care in patients with OSA and excess weight living in rural areas.

Objective:

We sought to examine the association of rural residence (vs urban) with receipt of GLP-1RAs among patients with OSA and excess weight.

Methods:

Using medical record and administrative data from the Veterans Health Administration, we constructed a national cohort of patients with OSA (defined by the presence of an ICD-10 code for OSA and procedure code for a sleep study) and excess weight (defined as BMI ≥ 27 kg/m2) who participated in a lifestyle program for weight loss. The exposure was rural home address as compared to urban home address as defined by Rural-Urban Commuting Area codes. The outcome was the receipt of GLP-1RAs in the year after index date. We performed a mixed effects logistic regression analysis of the association between rurality and receipt of GLP-1RAs while adjusting for confounding variables. We then performed a model-based causal mediation analysis evaluating two potential mediators of this association: 1) drive time to care and 2) drive distance to care.

Results:

Of the 68,680 patients meeting inclusion criteria from 1/1/2018 to 8/1/2023, 26.0% lived in a rural area, and 74.0% lived in an urban area. Overall, 8.0% received a GLP-1RA in the year after index date. Rural residents had ~10% lesser odds of receipt of a GLP-1RA approved for chronic weight loss in comparison to urban peers, after adjusting for confounders. Drive distance and drive time to primary and specialty care did not mediate this association.

Conclusions:

Rural patients with OSA and excess weight were less likely to receive a GLP-1RA as compared to urban peers. This association was not mediated through drive time or drive distance to care. Further work is needed to understand the factors that contribute to rural-urban differences in healthcare delivery beyond geographic domains of access.

Introduction:

Obstructive sleep apnea (OSA) is a highly prevalent condition, with over 425 million adults worldwide and 26 million adults in the United States estimated to have moderate to severe OSA.1 OSA impairs quality of life and carries a substantial comorbidity burden.2,3 Excess weight is the largest modifiable risk factor for OSA, accounting for nearly 60% of the attributable risk.4 Randomized trials of weight management interventions demonstrate improvements in OSA severity, symptoms, and metabolic parameters of obesity related conditions.5,6 Despite these benefits, successful weight loss is challenging, particularly among patients with OSA. Patients with OSA lose less weight than their peers, even though studies suggest that patients with OSA may demonstrate greater engagement with lifestyle-based weight management programs.7,8

In addition to these challenges in the treatment of excess weight among those with OSA, gaps in weight management care are particularly pronounced in rural areas. Over 46 million Americans live in rural areas and carry a disproportionate burden of obesity-related comorbidities but are less likely to receive healthcare to address these burdens.9,10 While primary care services are present in these areas, there is a lack of integrated OSA specialty care. Sleep medicine and weight management specialists are predominantly located in urban areas that are geographically and culturally distant from rural areas.11 We previously have shown that rural patients with OSA receive fewer weight management services (lifestyle programs, nutrition consults, pharmacotherapy, and bariatric surgery) in comparison to urban peers, but gaps remain in understanding the delivery of specific highly efficacious medication classes.12

Glucagon-Like Peptide-1 Receptor Agonists (GLP-1RAs) hold promise to improve outcomes in OSA for rural patients. These medications allow most patients to achieve clinically meaningful weight loss, improve severity of OSA, and improve cardiometabolic comorbidities.13,14 In addition, GLP-1RAs have relatively few requirements for laboratory and clinical monitoring, as well as few contraindications, providing a potentially effective and scalable option to improve the effectiveness of weight management care for patients with OSA living in rural areas.15,16 Early evidence in the management of type II diabetes mellitus suggests lower uptake for GLP-1RAs among patients in rural areas compared with urban peers.17 However, use of GLP-1RAs among patients with OSA living in rural areas is unknown. In addition, little is known about potential geographic barriers to receipt of these medications. To address this, we sought to determine: 1) if receipt of GLP1-RAs was different among rural patients with OSA and excess weight in comparison to urban peers; and 2) to what extent the effect of rurality on GLP-1RA prescription was mediated through the geographic domains of access to care, including drive time and drive distance to healthcare sites.

Methods:

Design/Data Source:

We conducted a cohort study to test the association of rurality with the receipt of GLP-1RAs among patients with OSA and excess weight. We used nationwide electronic health record (EHR) and administrative data obtained through the VA Corporate Data Warehouse (CDW). The CDW compiles comprehensive data around patient level factors and health system factors that may inform healthcare delivery, including patient demographics, vital signs, comorbidities, outpatient encounters, pharmacotherapy, procedures, and locations of treatments.18,19 The Institutional Review Boards of the VA Puget Sound (Approval #1615202) and the University of Washington (Approval #00018209) approved this study.

Population:

We identified adults with a diagnosis of OSA and excess weight within the medical record with index dates from 1/1/2018–8/1/2023. OSA was defined as having a sleep study (identified by CPT codes for either an in lab polysomnogram or home sleep apnea test) and an ICD-10 code for OSA.20–22 Excess weight was defined as BMI ≥ 27 kg/m2. We calculated the BMI using the latest pre index clinical weight in the 12 months prior to sleep study and last known height. We restricted the sample to patients with BMI ≥ 27 kg/m2 in order to be consistent with Veterans Health Administration (VHA) and FDA indications for GLP-1RAs.23–26 Within the VHA, participation in a lifestyle-based weight management program is a further requirement for use of weight management medications, including GLP-1RAs. We therefore restricted our sample to patients with active participation in these programs, defined as an encounter for nutrition counseling or comprehensive lifestyle intervention in the 12 months prior to and 3 months following qualifying sleep study. We excluded patients that died within 3 months of the qualifying sleep study. We defined date of index for our study as the date that all cohort entry criteria were met.

Variables:

The exposure was defined at the patient level as rural vs urban location, based on the patient’s home address. We categorized rurality based on the VHA Urban/Rural/Highly Rural (URH) classification system, which is the standard classification system used by the VA Office of Rural Health.27 This system assigns a designation of “urban,” “rural,” or “highly rural” based on the veterans geocoded home address and its corresponding Rural-Urban Commuting Area (RUCA) codes.28 We then collapsed this into a dichotomous categorical variable of rural vs urban due to small numbers in the highly rural category, though we acknowledge that this loses some of the granularity afforded by multiple levels of rurality. In addition, the URH classification system includes a separate designation for patients living on U.S. territorial islands including the U.S. Virgin Islands, Guam, Palau, and the Northern Marianas Islands. We categorized the one patient in our cohort living on a U.S. territorial island as rural. We defined the outcome as receipt of a GLP-1RA that is FDA approved for chronic weight loss (liraglutide, semaglutide, tirzepatide) in the 12 months following the index date. We then constructed a directed acyclic graph (DAG) to clearly define our a priori assumptions regarding covariate relationships with the exposure and outcome variables (Figure 1).29,30 We included patient level factors of age, sex, race, Hispanic ethnicity, and eligibility for VA benefits through a service-connected condition.31 Patient level medical factors included BMI at index, obesity related comorbidities (hypertension, coronary artery disease, type 2 diabetes), and mental health related comorbidities (anxiety, depression, and post-traumatic stress disorder). We also included geographic determinants of access to care, defined as drive time and drive distance to primary care and specialty care. These variables were obtained through geo-mapping software based on data sourced from the Planning Systems Support Group that calculates drive distance and average drive time to the nearest primary and specialty care center from the patient’s home address for the fiscal-year quarter of the qualifying sleep study. Finally, to understand treating site capabilities, we captured the Clinical Complexity Index designation for the VA Medical Center of each patient’s sleep study. Clinical complexity is based on the VHA Facility Complexity Model.32 This designation classifies facilities into five levels of complexity based on the patient population served at the site, clinical services complexity, and education/research activities.

Figure 1:

Directed acyclic graph of the relationship between rurality and receipt of Glucagon-Like Peptide-1 Receptor Agonists (GLP-1RAs)

Primary Analysis:

We used generalized linear mixed effects models with a logit link function fit with restricted maximum likelihood estimation to evaluate the association between rural patient home address (vs urban home address) and the receipt of GLP-1RAs in the 12 months following index. In all analyses, patient’s VA Medical Center (defined as the medical center of the qualifying sleep study) was adjusted by a random intercept to account for clustering of practice patterns. We performed a series of staged analyses to assess the impacts of confounding as outlined in our DAG. In model 1, we included the demographic factors of age, race, Hispanic ethnicity, and sex. In model 2, we included medical history related confounders as additional adjustment variables to model 1, including BMI, hypertension, diabetes mellitus, coronary artery disease, post-traumatic stress disorder, anxiety, and depression. In model 3, we included health system-related covariates of patient service-connected status and site complexity to model 2. Finally, in model 4 we included patient level geographic factors of drive distance and drive time to both primary and specialty care. STATA version 18.0 (College Station, TX) was used for data management. Analyses were performed using R statistical software version 4.5.

Mediation Analysis:

We performed a causal mediation analysis hypothesizing that patient’s drive distance and drive time to nearest VA primary care center are mediating variables. We also examined the mediating effects attributable to drive distance and drive time to nearest specialty care center. We used a model-based causal mediation analysis involving the use of a mediator regression model and outcome regression model.33,34 We constructed and fit a mixed effects model for the mediator variable assuming the mean is a linear function of rurality and the covariates listed above. A generalized linear mixed effects model of the outcome was also fit along with the mediator model in which the log odds of the receipt of GLP-1RAs was modeled as a linear function of the mediator, rurality, and confounding variables. In both the mediator and outcome model, we continued to include a random intercept shared between site. From these regressions, we estimated the total effect, average direct effect, and average causal mediation (indirect) effect.35 Uncertainty estimates and confidence intervals were calculated through a bootstrap where resampling was implemented with a quasi-Bayesian Monte Carlo method across 1,000 iterations. This process was repeated individually for each mediator variable (drive distance to primary care, drive time to primary care, drive distance to specialty care, drive time to specialty care). These analyses were performed using the mediate package and R statistical software.36

Sensitivity Analyses:

Since the FDA approved indications, prescribing practices, and access to GLP-1RAs changed over time, we performed a sensitivity analysis including index year as an additional adjustment variable to model 4 of the primary analysis. We also included an interaction term between index year and rurality to assess for changes in magnitude of difference in receipt of GLP-1RAs among rural vs urban patients over time.

Due to the use of GLP-1RAs for treatment of diabetes mellitus, we performed a sensitivity analysis of the association between rural home address and receipt of a GLP-1RA stratified by presence of a diagnosis of diabetes. We repeated the analytic plan of the primary analysis separately for those with and without a diagnosis of diabetes.

Results:

There were 68,680 patients identified within the VA CDW from January 1, 2018, to August 1, 2023 who met inclusion criteria. Of these, 17,880 (26.0%) lived in a rural area, and 50,800 (74.0%) lived in an urban area. The baseline characteristics of the rural and urban populations are presented in Table 1. Rural populations had a higher proportion of patients that were identified as white (78.2% in rural areas vs. 56.9% in urban areas) and lower proportion of patients identified as Black (13.0% in rural areas vs. 31.6% in urban areas). Rural areas also had a lower proportion of patients of Hispanic ethnicity (5.2% in rural areas vs. 13.2% in urban areas). As expected, the drive time and drive distance to primary and specialty care were greater in rural areas as compared to urban areas (Table 1). Age, female proportion, and comorbidities were largely balanced between rural and urban areas with standardized mean differences <0.2.

Table 1:

Cohort Characteristics (n=68,680)

Rural (n=17,880)
Mean (SD) / N(%)
Urban (n=50,800)
Mean (SD) / N(%)
Std diff

Patient Demographics

 Age, yr
56.5 (13.8)
54.4 (14.1)
0.15

 Female sex
2,703 (15.1%)
9,341 (18.4%)
0.09

 Race

0.50

  White
13,988 (78.2%)
28,921 (56.9%)

  Black
2,331 (13.0%)
16,064 (31.6%)

  NHPI
136 (0.8%)
685 (1.3%)

  AIAN
255 (1.4%)
563 (1.1%)

  Asian
88 (0.5%)
683 (1.3%)

  None provided/Unknown
1,082 (6.1%)
3,884 (7.6%)

 Hispanic ethnicity
935 (5.2%)
6,692 (13.2%)
0.28

 Service-connected
13,562 (75.9%)
37,992 (74.8%)
0.02

Drive Distance and Drive Time

 Drive distance to primary care, mi
25.1 (19.1)
9.9 (7.6)
1.05

 Drive distance to specialty care, mi
61.6 (41.3)
28.3 (29.2)
0.93

 Drive time to primary care, min
30.6 (19.6)
14.9 (8.3)
1.04

 Drive time to specialty care, min
66.4 (39.7)
33.1 (27.2)
0.98

Medical Characteristics

 Body Mass index, kg/m2

36.3 (6.2)
36.0 (6.2)
0.04

 Charlson Score, unit
2.2 (2.8)
2.1 (2.9)
0.03

 Comorbidities

  Diabetes mellitus
7,518 (42.0%)
19,666 (38.7%)
0.07

  Coronary artery disease
3,650 (20.4%)
8,488 (16.7%)
0.10

  Hypertension
11,900 (66.6%)
31,983 (63.0%)
0.08

  Anxiety
5,363 (30.0%)
16,396 (32.3%)
0.05

  Depression
8,790 (49.2%)
26,583 (52.3%)
0.06

  Post-traumatic stress disorder
5,927 (33.1%)
17,026 (33.5%)
0.01

Healthcare system structure

 Site complexity

0.45

  1a- high complexity
6,868 (38.4%)
28,892 (56.9%)

  1b- high complexity
3,933 (22.0%)
10,595 (20.9%)

  1c- high complexity
2,717 (15.2%)
5,812 (11.4%)

  2- moderate complexity
2,157 (12.1%)
3,176 (6.3%)

  3- low complexity
2,205 (12.3%)
2,325 (4.6%)

Among the 17,880 rural patients, 1,502 (8.4%) received a GLP-1RA approved for chronic weight loss in the year after index date. Of the 50,800 urban patients, 4,006 (7.9%) received a GLP-1RA in the year after OSA diagnosis. From 2018 to 2023, this proportion increased for both rural and urban patients, with ~4% of patients from 2018 receiving a GLP-1RA to ~15% of patients in 2023 receiving a GLP-1RA approved for chronic weight loss (Figure 2). In comparison, 11,614 (65.0%) of rural patients and 32,652 (64.3%) of urban patients received PAP over this time period.

Figure 2:

Yearly proportions of GLP-1RA receipt

GLP1-RAs- Glucagon-Like Peptide-1 receptor agonists

*Index dates included for 2023 are 1/1/2023–8/1/2023.

After accounting for demographic, medical history, and other health system related confounders in our fully adjusted model (Model 4), we found that patients with a rural home address had almost 10% lesser odds of receiving GLP-1RAs in comparison to their urban peers (OR 0.91; 95% CI 0.83–0.99; Table 2). Adjusted models with fewer covariates demonstrated similar trends (Models 1–3). The addition of comorbidities in model 2 and health system factors in model 3 yielded almost identical results to the final model. The intraclass coefficient among all models ranged from 0.059–0.070, indicating that the random intercept (patient’s site) accounted for only 5.9–7.0% of the variability in the association between exposure and outcome. Overall, the clustering by patient’s site contributed only a small amount to the overall models (Table 2).

Table 2:

Mixed effects logistic regression models for the association between rural residence (vs urban) with receipt of GLP-1RAs within 12 months after index date among patients with OSA and excess weight

Parameter
Model 1
Model 2
Model 3
Model 4

Sample size
68,680
68,680
68,680
68,680

Patients that received GLP-1RA
5,508
5,508
5,508
5,508

OR (95% CI) for association of rurality with receipt of a GLP-1RA
0.96 (0.89–1.02)
0.92* (0.86–0.99)
0.92* (0.85–0.99)
0.91* (0.83–0.99)

ICC of random intercept
0.059
0.070
0.069
0.070

In the mediation analysis of geographic variables, we did not identify any significant effect mediated through drive distance to primary care, drive time to primary care, drive distance to specialty care, or drive time to specialty care (Table 3).

Table 3:

Causal mediation analysis of the association of rural residence and GLP-1RA receipt as mediated through geographic domains of access to care (n=68,680)

Mediation variable
Total Effect
Average direct effect
Average causal mediation (indirect) effect
Mediation proportion (95% CI)
P-value for mediation effect of variable

Drive distance to primary care (for every 10 miles)
−0.59%* (−1.05% to −0.11%)
−0.73%* (−1.24% to −0.23%)
0.14% (−0.08% to 0.37%)
−24.6% (−116.4% to 18.9%)
0.21

Drive time to primary care (for every 10 minutes)
−0.58%* (−1.07% to −0.12%)
−0.63%* (−1.15% to −0.08%)
0.04% (−0.20% to 0.29%)
−7.9% (−102.9% to 54.3%)
0.71

Drive distance to specialty care (for every 10 miles)
−0.57%* (−1.05% to −0.08%)
−0.54%* (−1.07% to −0.03%)
0.02% (−0.21% to 0.18%
−4.4% (−48.8% to 72.8%)
0.78

Drive time to specialty care (for every 10 minutes)
−0.58%* (−1.02% to −0.11%)
−0.54% (−1.04% to 0.02%)
−0.04% (−0.24% to 0.15%)
6.6% (−37.3% to 71.9%
0.69

In the sensitivity analysis that accounted for index year as an adjustment variable, rural patients continued to demonstrate lower receipt of GLP-1RAs as compared to urban patients (OR 0.85; 95% CI 0.73–0.99). Inclusion of an interaction term between rural home address and index year did not demonstrate significant change in the association between rurality and GLP-1RA receipt over time (OR for interaction by year; 1.02; 95% CI 0.98–1.06).

In sensitivity analyses stratified by diabetes diagnosis, we found that 15.6% of patients with diabetes (n=27,184) were prescribed a GLP-1RA in comparison to 3.0% of those without diabetes (n=41,496). These proportions increased over the course of the time period of the cohort for both those with and without diabetes (Appendix 1 and 2). The odds of receiving a GLP-1RA among rural vs urban patients in both strata were similar to each other and to the overall primary analysis (Appendix 3 and 4). The odds ratio for the receipt of a GLP-1RA among rural patients with diabetes as compared to urban was 0.91 (95% CI 0.83–1.00; Appendix 3, Model 4). Among those without diabetes, the odds ratio for the receipt of a GLP-1RA among rural patients as compared to urban was 0.89 (95% CI 0.75–1.05, Appendix 4, Model 4).

Discussion/Conclusions:

Among patients with OSA and excess weight participating in a lifestyle program for weight loss, 8.0% received a GLP-1RA approved for chronic weight loss in the year after index date. Although the use of GLP-1RA agents increased across the time period of the cohort, usage overall remained low. Patients living in rural areas had 10% lesser odds of receipt of a GLP-1RA in comparison to urban peers after adjusting for confounders, and this association was not mediated through drive time or drive distance to primary or specialty care.

For decades, patients living in rural areas have demonstrated higher burdens of obesity, chronic comorbidities, and mortality.37–39 The difference in age adjusted mortality rates between rural and urban areas has widened throughout the past 20 years, with cardiovascular and metabolic comorbidities representing a substantial proportion of these deaths.9 In the context of OSA, rural patients are less likely to undergo sleep testing for symptoms of OSA and those that live farther from medical care are more likely to have a greater severity of OSA when they undergo sleep testing.40,41 These differences between rural and urban patients represent a complex interaction of barriers to access and delivery of care, societal factors affecting health care, and health system delivery of care.42,43

Given the disproportionate burden of disease in rural areas, the effective treatment of OSA, excess weight, and the associated cardiometabolic comorbidities represents an opportunity to substantially improve health outcomes among patients with OSA living in rural areas. In real world settings, only 20% of patients with OSA who enroll in a lifestyle program achieve clinically meaningful weight loss.44 While options such as pharmacotherapy and bariatric surgery can augment weight loss, they are rarely received by patients with OSA.12 GLP-1RAs offer a scalable and effective option for rural patients with OSA. Many patients treated with GLP-1RAs achieve clinically meaningful weight loss while simultaneously demonstrating improvements in cardiometabolic health.45,46 In addition, these medications have relatively few requirements for laboratory and in person monitoring, as well as few contraindications, which should reduce barriers to access of these medications for patients in rural areas.47,48 Despite the potential for a reduced barrier to entry, we found the opposite association for actual receipt of GLP-1RAs. Among patients with OSA and excess weight participating in a lifestyle program for weight loss, patients with a rural home address had about 10% lesser odds of receipt of a GLP-1RA approved for chronic weight loss in comparison to their urban peers, after accounting for confounding factors. While this difference is modest in magnitude, the receipt of GLP-1RAs increased over the years of this cohort and prescribing is likely to continue to increase with expanding availability of the medication and subsequent FDA indication of Tirzepatide for treatment of OSA in 2024.49 Given the high prevalence of OSA in the population, it will be important to ensure that barriers to access and receipt of effective medications do not further drive persistent and/or widening disparities in the use GLP-1RAs.

As described by Fortney et. al., access is a multidimensional concept including geographic, temporal, financial, digital, and cultural domains, and barriers to care can be considered within these areas.42 Among patients in rural areas, geographic barriers come first to mind. These geographic determinants are a particular concern for obesity treatment in OSA as sleep medicine and advanced weight management services (integrated programs, pharmacotherapy, bariatric surgery) specialists tend to be siloed in urban areas.11 However, in our analyses, drive distance to primary care, drive time to primary care, drive distance to specialty care, and drive time to specialty care do not appear to mediate the relationship between lower receipt of GLP-1RAs among rural patients with OSA and excess weight in comparison to urban peers cared for by the VHA. Our causal mediation analysis did not demonstrate any significant effect through the mediation pathway for any of these variables. This would suggest that, at least within the VHA, the geographic domain of access as measured through drive time and drive distance to care sites is not responsible for the lower receipt of these medications among rural patients. This may reflect the impact of VHA specific health care policies and initiatives, such as large-scale investments in widely distributed community based outpatient clinics, and increased accessibility for telehealth care, including legislation authorizing telehealth visits for patients and providers across state lines/jurisdictions.50–52

If geographic barriers are not mediating disparities in GLP-1RA delivery in rural areas, we must consider other drivers. Is it that greater wait times for OSA care in rural areas also extend to wait times for the delivery of GLP-1RAs to rural patients, yielding a temporal barrier to care?53 Or do greater patient borne costs of OSA for rural patients impact the use of GLP-1RAs, even in a healthcare system that has universal coverage of enrollees?53 Alternatively, is stigma associated with excess weight compounded with the marginalization that rural patients can perceive when interacting with urban based health care systems, leading to lesser receipt of GLP-1RAs to rural patients with OSA and excess weight?54 Future work will need to understand temporal, financial, digital, and cultural domains which may differentially impact patients in rural areas.

A notable strength of this study is the health system in which it was conducted. The VHA is the largest integrated health care system in the United States with a national geographic reach including representative rural areas.55,56 In addition, the VHA collects nationwide electronic health record and administrative data, allowing for excellent capture of variables included in the analysis. The VHA also has universal coverage of enrollees regardless of location, limiting confounding by insurance status and coverage. Finally, the centralized medication purchasing and distribution within the VHA may help to limit differential rural vs urban effects of drug shortages of GLP-1RAs.57

Despite these strengths, there are some limitations to the inferences drawn in this study. First, while the VHA has national reach of services, there is a disproportionate sample of patients with male sex, which is not representative of the U.S. population at large. Despite this, 12,069 patients of the 68,862 patients included in the sample were of female sex. Second, data regarding race and ethnicity categories as recorded in the electronic medical record are largely reflections of social constructs. Nonetheless, we included these variables within our analysis as they may have important implications in how healthcare is delivered.58 Third, while universal coverage for medical services and prescriptions within the VHA is a major strength in our analysis, it also limits generalizability of the findings to other health systems structures and insurance coverage types. While universal coverage may improve access, it is important to note that other important social determinants of health (income, education) were not available in this data set. Measurable variables such as drive time, drive distance, and rurality alone do not encompass all aspects of access. Important patient level determinants of healthcare and provider and health system characteristics (e.g., local prescribing patterns, medications availability) are not as easily measurable but may provide more granular detail regarding healthcare delivery. While we attempted to account for system level determinants through the use of the Clinical Complexity Index and including clinical site as a random intercept, we were not able to include important factors that may affect receipt of GLP-1RAs, such as availability of endocrinology and local prescribing restrictions. Finally, it should be noted that we limited the entry criteria for the cohort to participation in lifestyle programs within the VHA. While this may have excluded patients participating in a lifestyle program outside of the VHA, there is a strong cost incentive for patients to participate within the VHA, as these programs are free to access for Veterans.59 This cost incentive likely limited any inadvertent exclusion of patients from our cohort.

Among patients with OSA and excess weight participating in a lifestyle program for weight loss identified from 1/2018 to 8/2023, 8.0% received a GLP-1RA approved for chronic weight loss in the year after index date. Though this proportion increased over the years of the study population, the receipt of these effective medications remained low at ~15% overall in patients that entered the cohort in 2023. While patients living in rural areas had 10% lesser odds of receipt of a GLP-1RA in comparison to urban peers after adjusting for confounders, this association was not mediated through drive time or drive distance to primary or specialty care. Given the overall low receipt of these medications, there is a need to improve and expand access to GLP-1RAs among all patients with OSA and excess weight. In order to address barriers to the receipt of these effective medications for OSA, obesity, and associated cardiometabolic comorbidities, innovative strategies are necessary improve the access to these medications, while also mitigating disparities for patients that live in rural areas. In addition, further work is necessary to understand the factors that continue to drive rural-urban differences in the delivery in healthcare beyond geographic domains of access.

Supplementary Material

supplementary tables

Funding information:

AGL received support during the period of this work from the NIH/NHLBI (5T32 HL007287-45 and 1F32HL179793-01). AGL also received funding from the American Thoracic Society through the ATS Academic Sleep Pulmonary Integrated REsearch/Clinical (ASPIRE) Fellowship.

LMD received support from the VA Health Systems Research (IIR 20-240) which contributed to the creation of the cohort in this manuscript.

Footnotes

Disclaimers:

The work represents the views of the authors and do not necessarily represent the views of the Department of Veterans Affairs or the United States government.

This work was presented as an oral presentation at the Annual Meeting of the Associated Professional Sleep Societies in Seattle, WA on June 11, 2025.

A previous version of this manuscript was submitted to the University of Washington as a thesis for completion of a Master of Science degree.

References:

• 1.Benjafield AV, Ayas NT, Eastwood PR, et al.
  Estimation of the global prevalence and burden of obstructive sleep apnoea: a literature-based analysis. Lancet Respir Med. 2019;7(8):687–698. doi: 10.1016/S2213-2600(19)30198-5
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 2.Vinnikov D, Blanc PD, Alilin A, Zutler M, Holty JEC. Fatigue and sleepiness determine respiratory quality of life among veterans evaluated for sleep apnea. Health Qual Life Outcomes. 2017;15:48. doi: 10.1186/s12955-017-0624-x
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 3.Lee W, Nagubadi S, Kryger MH, Mokhlesi B. Epidemiology of Obstructive Sleep Apnea: a Population-based Perspective. Expert Rev Respir Med. 2008;2(3):349–364. doi: 10.1586/17476348.2.3.349
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 4.Young T, Peppard PE, Taheri S. Excess weight and sleep-disordered breathing. J Appl Physiol Bethesda Md 1985. 2005;99(4):1592–1599. doi: 10.1152/japplphysiol.00587.2005 [DOI] [PubMed] [Google Scholar]
• 5.Carneiro-Barrera A, Díaz-Román A, Guillén-Riquelme A, Buela-Casal G. Weight loss and lifestyle interventions for obstructive sleep apnoea in adults: Systematic review and meta-analysis. Obes Rev Off J Int Assoc Study Obes. 2019;20(5):750–762. doi: 10.1111/obr.12824 [DOI] [PubMed] [Google Scholar]
• 6.Haase CL, Lopes S, Olsen AH, Satylganova A, Schnecke V, McEwan P. Weight loss and risk reduction of obesity-related outcomes in 0.5 million people: evidence from a UK primary care database. Int J Obes 2005. 2021;45(6):1249–1258. doi: 10.1038/s41366-021-00788-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
• 7.Kline CE, Burke LE, Sereika SM, et al.
  Bidirectional Relationships Between Weight Change and Sleep Apnea in a Behavioral Weight Loss Intervention. Mayo Clin Proc. 2018;93(9):1290–1298. doi: 10.1016/j.mayocp.2018.04.026
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 8.Littman AJ, Boyko EJ, McDonell MB, Fihn SD. Evaluation of a Weight Management Program for Veterans. Prev Chronic Dis. 2012;9:E99. doi: 10.5888/pcd9.110267
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 9.Thomas KL, Dobis EA, McGranahan DA. The Nature of the Rural-Urban Mortality Gap. U.S. Department of Agriculture, Economic Research Service; 2024. [Google Scholar]
• 10.Krishna S, Gillespie KN, McBride TM. Diabetes burden and access to preventive care in the rural United States. J Rural Health Off J Am Rural Health Assoc Natl Rural Health Care Assoc. 2010;26(1):3–11. doi: 10.1111/j.1748-0361.2009.00259.x [DOI] [PubMed] [Google Scholar]
• 11.Donovan LM, Shah A, Chai-Coetzer CL, Barbé F, Ayas NT, Kapur VK. Redesigning Care for OSA. Chest. 2020;157(4):966–976. doi: 10.1016/j.chest.2019.10.002
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 12.Leonhard AG, McDowell J, Hoerster KD, et al.
  Receipt of Weight Management Services Among Patients with OSA and Obesity. CHEST Pulm. Published online March 4, 2025:100156. doi: 10.1016/j.chpulm.2025.100156
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 13.Blackman A, Foster GD, Zammit G, et al.
  Effect of liraglutide 3.0 mg in individuals with obesity and moderate or severe obstructive sleep apnea: the SCALE Sleep Apnea randomized clinical trial. Int J Obes 2005. 2016;40(8):1310–1319. doi: 10.1038/ijo.2016.52 [DOI] [PMC free article] [PubMed] [Google Scholar]
• 14.Malhotra A, Grunstein RR, Fietze I, et al.
  Tirzepatide for the Treatment of Obstructive Sleep Apnea and Obesity. N Engl J Med. 2024;391(13):1193–1205. doi: 10.1056/NEJMoa2404881
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 15.Ghusn W, Hurtado MD. Glucagon-like Receptor-1 agonists for obesity: Weight loss outcomes, tolerability, side effects, and risks. Obes Pillars. 2024;12:100127. doi: 10.1016/j.obpill.2024.100127
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 16.Latif W, Lambrinos KJ, Patel P, Rodriguez R. Compare and Contrast the Glucagon-Like Peptide-1 Receptor Agonists (GLP1RAs). In: StatPearls. StatPearls Publishing; 2025. Accessed March 17, 2025. http://www.ncbi.nlm.nih.gov/books/NBK572151/ [PubMed] [Google Scholar]
• 17.Chen WH, Li Y, Yang L, et al.
  Geographic variation and racial disparities in adoption of newer glucose-lowering drugs with cardiovascular benefits among US Medicare beneficiaries with type 2 diabetes. PLOS ONE. 2024;19(1):e0297208. doi: 10.1371/journal.pone.0297208
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 18.Corporate Data Warehouse (CDW). January
  11, 2023. Accessed August 7, 2025. https://www.hsrd.research.va.gov/for_researchers/cdw.cfm
• 19.Culbreath C, Gonsoulin M. VIReC Corporate Data Warehouse (CDW) Domain Descriptions. Hines IL: U.S. Department of Veterans Affairs, Health Services Research & Development Service, VA Information Resource Center; 2019. [Google Scholar]
• 20.McIsaac DI, Gershon A, Wijeysundera D, Bryson GL, Badner N, van Walraven C. Identifying Obstructive Sleep Apnea in Administrative Data: A Study of Diagnostic Accuracy. Anesthesiology. 2015;123(2):253–263. doi: 10.1097/ALN.0000000000000692
  [DOI] [PubMed] [Google Scholar]
• 21.Jolley RJ, Liang Z, Peng M, et al.
  Identifying Cases of Sleep Disorders through International Classification of Diseases (ICD) Codes in Administrative Data. Int J Popul Data Sci. 3(1):448. doi: 10.23889/ijpds.v3i1.448 [DOI] [PMC free article] [PubMed] [Google Scholar]
• 22.Donovan LM, Wai T, Spece LJ, et al.
  Sleep Testing and Mortality in a Propensity-matched Cohort of Patients with Chronic Obstructive Pulmonary Disease. Ann Am Thorac Soc. Published online August 14, 2023. doi: 10.1513/AnnalsATS.202303-275OC [DOI] [PubMed] [Google Scholar]
• 23.ZEPBOUND (tirzepatide) injection. United States prescribing information. Revised November 2023. US Food and Drug Administration. Accessed May 23, 2024. https://www.accessdata.fda.gov/drugsatfda_docs/label/2023/217806s000lbl.pdf [Google Scholar]
• 24.WEGOVY (semaglutide) injection, for subcutaneous use. US Food and Drug Administration (FDA) approved product information. Revised July 2023. Accessed May 23, 2024. https://www.accessdata.fda.gov/drugsatfda_docs/label/2023/215256s007lbl.pdf
• 25.VA Pharmacy Benefits Management Services and National Formulary Committee. Semaglutide (WEGOVY) Criteria for Use April 2024. Published online April 2024.
• 26.VA Pharmacy Benefits Management Services and National Formulary Committee. Tirzepatide (ZEPBOUND) Subcutaneous Injection Criteria for Use December 2023. Published online December 2023.
• 27.Veterans Health Administration Office of Rural Health. Fact Sheet: How to Define Rurality. U.S. Department of Veterans Affairs; 2023. ruralhealth.va.gov [Google Scholar]
• 28.U.S. Depatment of Agriculture Economic Research Service. 2010 Rural-Urban Commuting Codes. Rural-Urban Commuting Area Codes. September
  25, 2023. Accessed April 13, 2024. https://www.ers.usda.gov/data-products/rural-urban-commuting-area-codes/documentation/ [Google Scholar]
• 29.Lederer DJ, Bell SC, Branson RD, et al.
  Control of Confounding and Reporting of Results in Causal Inference Studies. Guidance for Authors from Editors of Respiratory, Sleep, and Critical Care Journals. Ann Am Thorac Soc. 2019;16(1):22–28. doi: 10.1513/AnnalsATS.201808-564PS
  [DOI] [PubMed] [Google Scholar]
• 30.Textor J, Van Der Zander B, Gilthorpe MS, Liśkiewicz M, Ellison GTH. Robust causal inference using directed acyclic graphs: the R package ‘dagitty.’
  Int J Epidemiol. Published online January 15, 2017:dyw341. doi: 10.1093/ije/dyw341 [DOI] [PubMed] [Google Scholar]
• 31.Eligibility for VA disability benefits. Veterans Affairs. April
  23, 2025. Accessed May 18, 2025. https://www.va.gov/disability/eligibility/ [Google Scholar]
• 32.National Academies of Sciences E, Education D of B and SS and, Integration B on HS, Sciences D on E and P, Environment B on I and the C, Administration C on FSR for VH. Nature of Veterans Health Administration Facilities Management (Engineering) Tasks and Staffing. In: Facilities Staffing Requirements for the Veterans Health Administration-Resource Planning and Methodology for the Future. National Academies Press (US); 2019. Accessed December 5, 2023. https://www.ncbi.nlm.nih.gov/books/NBK555777/ [PubMed] [Google Scholar]
• 33.Imai K, Yamamoto T. Identification and Sensitivity Analysis for Multiple Causal Mechanisms: Revisiting Evidence from Framing Experiments. Polit Anal. 2013;21(2):141–171. doi: 10.1093/pan/mps040 [DOI] [Google Scholar]
• 34.Imai K, Keele L, Yamamoto T. Identification, Inference and Sensitivity Analysis for Causal Mediation Effects. Stat Sci. 2010;25(1). doi: 10.1214/10-STS321 [DOI] [Google Scholar]
• 35.Lee H, Cashin AG, Lamb SE, et al.
  A Guideline for Reporting Mediation Analyses of Randomized Trials and Observational Studies. JAMA. 2021;326(11):1045–1056. doi: 10.1001/jama.2021.14075
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 36.Tingley D, Yamamoto T, Hirose K, Keele L, Imai K. mediation : R Package for Causal Mediation Analysis. J Stat Softw. 2014;59(5). doi: 10.18637/jss.v059.i05 [DOI] [Google Scholar]
• 37.Befort CA, Nazir N, Perri MG. Prevalence of obesity among adults from rural and urban areas of the United States: findings from NHANES (2005–2008). J Rural Health Off J Am Rural Health Assoc Natl Rural Health Care Assoc. 2012;28(4):392–397. doi: 10.1111/j.1748-0361.2012.00411.x [DOI] [PMC free article] [PubMed] [Google Scholar]
• 38.Table 39. Number of Respondent-Reported Chronic Conditions from 10 Selected Conditions among Adults Aged 18 and over, by Selected Characteristics: United States, Selected Years 2002–2016. National CEnter for Health Statistics. Health, United States; 2017. [Google Scholar]
• 39.Moy E
  Leading Causes of Death in Nonmetropolitan and Metropolitan Areas — United States, 1999–2014. MMWR Surveill Summ. 2017;66. doi: 10.15585/mmwr.ss6601a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
• 40.Spagnuolo CM, McIsaac M, Dosman J, Karunanayake C, Pahwa P, Pickett W. Distance to Specialist Medical Care and Diagnosis of Obstructive Sleep Apnea in Rural Saskatchewan. Can Respir J. 2019;2019:1683124. doi: 10.1155/2019/1683124
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 41.Allen AJMH, Amram O, Tavakoli H, Almeida FR, Hamoda M, Ayas NT. Relationship between Travel Time from Home to a Regional Sleep Apnea Clinic in British Columbia, Canada, and the Severity of Obstructive Sleep. Ann Am Thorac Soc. 2016;13(5):719–723. doi: 10.1513/AnnalsATS.201509-613BC
  [DOI] [PubMed] [Google Scholar]
• 42.Fortney JC, Burgess JF, Bosworth HB, Booth BM, Kaboli PJ. A Re-conceptualization of Access for 21st Century Healthcare. J Gen Intern Med. 2011;26(Suppl 2):639–647. doi: 10.1007/s11606-011-1806-6
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 43.Gong G, Phillips SG, Hudson C, Curti D, Philips BU. Higher US Rural Mortality Rates Linked To Socioeconomic Status, Physician Shortages, And Lack Of Health Insurance. Health Aff Proj Hope. 2019;38(12):2003–2010. doi: 10.1377/hlthaff.2019.00722 [DOI] [PubMed] [Google Scholar]
• 44.Janney CA, Kilbourne AM, Germain A, et al.
  The Influence of Sleep Disordered Breathing on Weight Loss in a National Weight Management Program. Sleep. 2016;39(1):59–65. doi: 10.5665/sleep.5318
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 45.Wilding JPH, Batterham RL, Calanna S, et al.
  Once-Weekly Semaglutide in Adults with Overweight or Obesity. N Engl J Med. 2021;384(11):989–1002. doi: 10.1056/NEJMoa2032183
  [DOI] [PubMed] [Google Scholar]
• 46.Jastreboff AM, Aronne LJ, Ahmad NN, et al.
  Tirzepatide Once Weekly for the Treatment of Obesity. N Engl J Med. 2022;387(3):205–216. doi: 10.1056/NEJMoa2206038
  [DOI] [PubMed] [Google Scholar]
• 47.Idrees Z, Cancarevic I, Huang L. FDA-Approved Pharmacotherapy for Weight Loss Over the Last Decade. Cureus. 14(9):e29262. doi: 10.7759/cureus.29262 [DOI] [PMC free article] [PubMed] [Google Scholar]
• 48.Khera R, Murad MH, Chandar AK, et al.
  Association of Pharmacological Treatments for Obesity With Weight Loss and Adverse Events: A Systematic Review and Meta-analysis. JAMA. 2016;315(22):2424–2434. doi: 10.1001/jama.2016.7602
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 49.Commissioner O of the. FDA Approves First Medication for Obstructive Sleep Apnea. FDA. December
  27, 2024. Accessed January 22, 2026. https://www.fda.gov/news-events/press-announcements/fda-approves-first-medication-obstructive-sleep-apnea [Google Scholar]
• 50.Bridging the Digital Divide | Telehealth VA. Accessed June 7, 2025. https://telehealth.va.gov/digital-divide
• 51.Heyworth L, Kirsh S, Zulman D, Ferguson JM, Kizer KW. Expanding Access through Virtual Care: The VA’s Early Experience with Covid-19. Catal Non-Issue Content. 2020;1(4). doi: 10.1056/CAT.20.0327 [DOI] [Google Scholar]
• 52.Kirsh S, Su GL, Sales A, Jain R. Access to outpatient specialty care: solutions from an integrated health care system. Am J Med Qual Off J Am Coll Med Qual. 2015;30(1):88–90. doi: 10.1177/1062860614542844 [DOI] [PubMed] [Google Scholar]
• 53.Fatima D, Tsai WH, Corrigan J, et al.
  Exploring patient-borne costs and wait times for obstructive sleep apnea (OSA) care among rural and urban adults. Can J Respir Crit Care Sleep Med. Published online January 2, 2023. Accessed July 2, 2025. https://www.tandfonline.com/doi/full/10.1080/24745332.2022.2156936 [Google Scholar]
• 54.Brundisini F, Giacomini M, DeJean D, Vanstone M, Winsor S, Smith A. Chronic disease patients’ experiences with accessing health care in rural and remote areas: a systematic review and qualitative meta-synthesis. Ont Health Technol Assess Ser. 2013;13(15):1–33. [PMC free article] [PubMed] [Google Scholar]
• 55.Administration UD of VA Veterans Health. VA.gov | Veterans Affairs. MOVE! Weight Management Program. Accessed January 24, 2024. https://www.move.va.gov/MOVE/QandA.asp [Google Scholar]
• 56.Bureau UC. Veterans in Rural America: 2011–2015. Census.gov. Accessed May 23, 2024. https://www.census.gov/library/publications/2017/acs/acs-36.html
• 57.Putka S
  Diabetes Patients Also Struggle to Access GLP-1 Agonists | MedPage Today. January
  24, 2023. Accessed June 5, 2025. https://www.medpagetoday.com/special-reports/features/102773
• 58.Aggarwal R, Chiu N, Loccoh EC, Kazi DS, Yeh RW, Wadhera RK. Rural-Urban Disparities: Diabetes, Hypertension, Heart Disease, and Stroke Mortality Among Black and White Adults, 1999–2018. J Am Coll Cardiol. 2021;77(11):1480–1481. doi: 10.1016/j.jacc.2021.01.032
  [DOI] [PMC free article] [PubMed] [Google Scholar]
• 59.Rogers TB. MOVE! Weight Management Program | Veterans Affairs. U.S. Department of Veterans Affairs. Accessed June 5, 2025. https://www.move.va.gov/MOVE/index.asp [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplementary tables