Insomnia, Sleep Apnea, and Incidence of Hypertension and Cardiovascular Disease Among Men and Women US Veterans

Authors: Allison E. Gaffey, Matthew M. Burg, Henry K. Yaggi, Kaicheng Wang, Cynthia A. Brandt, Sally G. Haskell, Lori A. Bastian, Tiffany E. Chang, Allison Levine, Melissa Skanderson, Andrey Zinchuk

Abstract

Background

Comorbid insomnia and obstructive sleep apnea (COMISA) is associated with cardiovascular disease (CVD) in older adults. The associations of COMISA with cardiovascular risk among military veterans, who show a greater risk for hypertension and CVD than non‐Veterans, and whether associations differ by sex are unknown. Thus, we examined associations of COMISA with incident hypertension and CVD risk in post‐9/11 Veterans.

Methods

This retrospective cohort included patients who enrolled in Veterans Health Administration care from 2001 to 2021. Insomnia and obstructive sleep apnea were defined by ≥2 outpatient diagnoses. Hypertension was defined by ≥2 outpatient‐coded diagnoses or ≥1 antihypertensive medication fill; CVD by ≥1 inpatient or ≥2 outpatient diagnoses. Time‐varying Cox proportional hazard models adjusted for demographics, behavioral, and clinical factors were conducted overall and by sex. Sensitivity analyses accounted for health care use, a 180‐day washout, sleep study–confirmed diagnoses, outpatient blood pressure data, and inclusion of patients deceased during follow‐up.

Results

Among 937 598 veterans (12% women; median age, 41 years), COMISA was associated with increased hypertension risk overall (adjusted hazard ratio [aHR], 2.43 [95% CI, 2.36–2.50]), in men (aHR, 2.09 [95% CI, 2.02–2.16]) and in women (aHR, 2.20 [95% CI, 2.00–2.42]). Insomnia alone (aHR, 1.27–1.44) and obstructive sleep apnea only (aHR, 2.00–2.26) were also associated with elevated risk. COMISA was similarly associated with CVD risk overall (aHR, 3.81 [95% CI, 3.64–3.99]), in men (aHR, 3.81 [95% CI, 3.63–4.00]) and women (aHR, 3.44 [95% CI, 2.98–3.98]), as were insomnia (aHR, 1.36–1.37) and obstructive sleep apnea (aHR, 3.32–2.62). Sensitivity analyses were consistent.

Conclusions

COMISA conferred the greatest risk of hypertension and CVD among post‐9/11 Veterans. Identifying disordered sleep among men and women may be an important CVD prevention priority.

Clinical PerspectiveWhat Is New? In a nationwide cohort of nearly 1 million post‐9/11 Veterans, comorbid insomnia and obstructive sleep apnea was independently associated with a >2‐fold increase in incident hypertension and a 3‐fold increase in cardiovascular disease.Comorbid insomnia and obstructive sleep apnea conferred substantially greater cardiovascular risk than either insomnia or sleep apnea alone, emphasizing the clinical relevance of co‐occurring sleep disorders as underrecognized determinants of cardiovascular disease in younger and middle‐aged Veteran populations. What Are the Clinical Implications? Incorporating systematic screening and integrated management of insomnia and sleep apnea into cardiovascular risk assessment frameworks may enhance early identification, optimize prevention strategies, and mitigate long‐term cardiovascular disease burden among Veterans and the general population.

Insomnia and obstructive sleep apnea (OSA) are the most common sleep disorders and often co‐occur, a condition known as COMISA. ^1^ Both contribute to the risk for hypertension, cardiovascular disease (CVD), and related death. ^2^ , ^3^ , ^4^ , ^5^ , ^6^ , ^7^ , ^8^ , ^9^ Although the cardiovascular consequences of each disorder are well recognized and the American Heart Association recently added sleep to its Essential 8 metrics for cardiovascular health, ^10^ research concerning COMISA has largely focused on middle‐aged and older adults. ^7^ , ^8^ , ^11^ As a result, there is limited understanding about how COMISA contributes to cardiovascular risk, particularly in younger adults and in military veterans, who face unique occupational, behavioral, and psychosocial exposures and bear a higher burden of sleep and cardiovascular conditions than non‐Veterans. ^12^ , ^13^ , ^14^ , ^15^ , ^16^ Resolving these questions is valuable for optimizing hypertension management and CVD prevention, especially because sleep disturbances represent a modifiable prevention target.

Sex differences in sleep disorder prevalence and presentation may further influence cardiovascular risk. Insomnia is more prevalent among women, ^17^ , ^18^ while OSA is more commonly diagnosed in men, although this gap narrows after menopause. ^19^ Among military personnel, COMISA differences have also been reported. Women are more likely to report psychiatric comorbidities, whereas men more often exhibit severe sleep‐disordered breathing. ^20^ Among Veterans treated in the Veterans Health Administration (VA), women are assessed up to 1 year later than men for sleep disorders and are less likely to be diagnosed with any sleep disorder than men. ^16^ Insomnia and OSA symptom profiles also differ by sex, with women more likely to present with fatigue, difficulty falling or staying asleep, or mood disturbance, and men more likely to exhibit classic apneic symptoms such as loud snoring, witnessed apneas, and excessive daytime sleepiness. ^21^ , ^22^ , ^23^ Further, women are more often symptomatic at a lower apnea–hypopnea index. ^24^ These differences in symptom presentation may contribute to underrecognition and delayed diagnosis of OSA in women, which can influence treatment initiation (eg, referral and uptake of positive airway pressure [PAP] therapy) and potentially alter disease progression trajectories. ^23^ , ^25^ In turn, sex differences in presentation, detection, and management plausibly modify the observed associations between sleep disorders and cardiovascular outcomes. ^21^ , ^24^ Hypertension and CVD risk trajectories also differ by age and sex, ^26^ underscoring the need for analyses by sex.

The primary objective of this investigation was to determine whether insomnia, OSA, and COMISA were associated with risk of incident hypertension and CVD in a large, nationwide cohort of primarily younger and middle‐aged Veterans. We also conducted sex‐specific analyses to assess whether these associations differed by sex. We hypothesized that (1) each sleep disorder would independently predict greater cardiovascular risk; (2) COMISA would confer greater risk than either disorder alone; and (3) effects would vary by sex, with stronger associations expected among men due to their greater likelihood of receiving a sleep disorder diagnosis, particularly OSA, within the VA system.

Methods

Data used for these analyses are unavailable for distribution by the authors as they are covered by a Department of Veterans Affairs proprietary data use agreement. The VA Connecticut Healthcare System Institutional Review Board provided exemption for the research described.

Data Sources

The VA’s Defense Manpower Data Center–Contingency Tracking System was used to identify eligible patients on the basis of military discharge and VA care initiation dates. This list was merged with national VA Corporate Data Warehouse electronic health record data, which consist of all vital signs, procedural administrative, pharmacy, and diagnostic records (based on the International Classification of Diseases, Ninth Revision, Clinical Modification, and Tenth Revision, Clinical Modification [ICD‐9‐CM and ICD‐10‐CM] codes and dates), from Veterans’ VA health care visits (see Table S1 for codes).

Study Population

The WVCS (Women Veterans Cohort Study) is a nationwide, prospective examination of men and women Veterans who served during conflicts in Iraq and Afghanistan (ie, post‐9/11), were discharged from service October 1, 2001, to September 30, 2021, and had their first VA outpatient visit before September 30, 2021. Patients were excluded if they were diagnosed with insomnia, OSA, hypertension, or CVD before VA enrollment; had a history of severe psychiatric disorder (ie, bipolar or schizophrenia); were missing demographics or covariates; lacked a date of follow‐up care; or were deceased during follow‐up and we could not determine cause of death. Those with a prior history of CVD were also excluded from analyses of hypertension. Study characteristics were reported on the basis of the Strengthening the Reporting of Observational Studies in Epidemiology guidelines (Table S2). ^27^

Exposures: Insomnia, Sleep Apnea, COMISA

Patients were categorized as having insomnia or OSA if they had ICD‐9/10 diagnostic codes on ≥2 outpatient visits. Using these diagnoses, 4 groups were insomnia only, OSA only, COMISA, and neither diagnosis. The timing of diagnoses was used to calculate time to each outcome event.

Outcomes: Incident Hypertension and CVD

The primary outcome of hypertension was defined by ICD‐9/10 codes on ≥2 outpatient visits or a filled prescription for antihypertensive medication (ie, angiotensin‐converting enzyme inhibitors, angiotensin II receptor blockers, beta blockers, calcium channel blockers, or diuretics). The secondary outcome, incident CVD, was defined as ≥1 inpatient or 2 outpatient diagnoses of heart failure, peripheral vascular disease, atrial fibrillation, myocardial infarction, or ischemic stroke, consistent with prior VA studies. ^28^ A third, exploratory outcome was ≥2 outpatient blood pressure (BP) readings exceeding 140/90 mm Hg American Heart Association/American College of Cardiology guidelines (stage II hypertension). ^29^ Because BP is routinely collected at outpatient visits, this approach helps reduce risk of delayed or missed hypertension diagnoses.

Other Descriptive Variables and Covariates

Patients were characterized on the basis of marital status, the availability of additional non‐VA health insurance at baseline, baseline BP, and history of sleep assessment and treatment with medication for insomnia (ie, doxepin, eszopiclone, ramelteon, suvorexant, temazepam, trazodone, zaleplon, zolpidem, or zolpidem tartrate) or PAP therapy (eg, Current Procedural Terminology codes E0601, E0470, E0561, and A7027). A priori covariates were recorded at baseline (ie, defined as the date of the first primary care or other outpatient visit) or the first occurrence during follow‐up.

Covariates included age, biological sex (documented as woman and man [reference]), race and ethnicity (non‐Hispanic Black, Hispanic, Other [American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, or more than 1 race], non‐Hispanic White [reference]), rurality (urban [reference], rural, very rural); behavioral smoking history (never [reference], current, past), body mass index (categories of normal/underweight [reference], overweight, obese), and history of an alcohol or substance use disorder (ever diagnosed versus never [reference]); and other clinical risk factors (ever diagnosed or screened positive versus never [reference]): type 2 diabetes or dyslipidemia. Due to the high prevalence of psychiatric diagnoses among Veterans, major depressive disorder, generalized anxiety disorder, and posttraumatic stress disorder were also separately included as clinical covariates. For sensitivity analyses, we derived a variable for health care use based on the number of patients’ primary care visits during the first 2 years of VA care, as used in past analyses. ^28^

Statistical Analysis

Univariate and bivariate statistics (χ^2^, ANOVA, and Mann–Whitney) were used to present descriptive data and to compare factors by sex. To minimize reverse causality, we excluded individuals with any documented history of hypertension or CVD before baseline. Patients were followed from baseline until their first hypertension or CVD diagnosis, last VA encounter, or administrative censoring. Those without a sleep diagnosis were the reference group. If a participant was diagnosed with hypertension or CVD before receiving a diagnosis of insomnia or OSA, they were classified in the reference group, and follow‐up time was censored after outcome diagnosis. Follow‐up was censored at the date of hypertension or CVD diagnosis or the date of a patient’s last VA encounter on the basis of the event that occurred first. Patients contributed person‐time to exposed groups only from the date of their respective sleep disorder diagnosis onward. This approach strengthens the temporal ordering of exposure and outcome, reducing bias due to misclassification or reverse causation.

Cumulative incidence of hypertension and CVD were calculated overall and for men and women separately as cases per 1000 person‐years. Time‐varying Cox proportional hazards models were used to examine the associations of insomnia, OSA, and COMISA (versus no sleep disorder), in which insomnia and OSA diagnoses were treated as time‐updated exposures, with incident hypertension (primary) and CVD (secondary). Patients contributed exposed person‐time only from the date of diagnosis onward. Models were conducted overall and stratified by sex, and adjusted sequentially for demographic, behavioral, and clinical covariates (see Tables S3 and S5, models 1–3). To evaluate whether COMISA conferred higher risk than either condition alone, OSA was used as the reference group in additional models. Interactions by sex were also tested.

Sensitivity analyses included (1) health care use, (2) a 180‐day gap between exposure and outcome diagnoses to confirm temporality, (3) restricting analyses to patients with a sleep study to verify sleep diagnoses with objective data for models of incident hypertension, (4) outpatient BP readings in the hypertension definition, and (5) including patients who died during follow‐up (see Table S4 for models 4–8). Covariate data were assumed to be missing at random. Multiple comparisons were performed using the Bonferroni correction with P<0.016 defined as statistically significant for each of the 3 pairwise comparisons. Analyses were conducted with SAS version 9.4 (SAS Institute, Cary, NC).

Results

The final analytic sample included 937 598 patients (12% were women [n=113 381]), of which 13% had a diagnosis of insomnia, 20% had a diagnosis of OSA, and 14% were diagnosed with COMISA (Figure 1). The median age at baseline was 41 years; 61% of patients were non‐Hispanic White individuals; 50% were married; 72% lived in an urban location; and 35% had additional, nongovernment health insurance (Table 1). One‐third were obese at baseline, and the most prevalent clinical disorders were posttraumatic stress disorder, dyslipidemia, and major depressive disorder (42%–31%). Across the cohort, most patients completed 1 primary care visit in the initial 2 years of VA care. During 7 088 554 person‐years of follow‐up (median, 7.2 [interquartile range, 4.0–10.8; range, 1–22]), 10% had completed a sleep study. Of those with COMISA, 72% were first diagnosed with insomnia, and 28% were first diagnosed with OSA.

When comparing patient characteristics by sex, a greater percentage of men than women were non‐Hispanic White individuals (62% versus 48%) and were married (51% versus 27%). More men were diagnosed with a lipid disorder than women (39% versus 27%), but fewer men were diagnosed with major depressive disorder (30% versus 42%). A greater percentage of men had a diagnosis of OSA than women (32% versus 21%), although there was a similar proportion of insomnia diagnoses by sex (23% versus 24%). Among the group with COMISA, about 73% of men were diagnosed with insomnia before OSA compared with 64% of women. Among the COMISA group, while more men had a history of a sleep study or were prescribed a PAP device than women (10% versus 8% and 17% versus 9%, respectively), slightly fewer men were ever prescribed insomnia medication (44% versus 49%).

Risk for Incident Hypertension

Across the 20 years of VA follow‐up, 310 048 patients developed hypertension (33%), and the median follow‐up time was 7 (interquartile range, 4–11) years. The incidence of hypertension was 31.6 per 1000 person‐years overall, 32.2 for men and 27.5 for women. In unadjusted multivariable models, COMISA, insomnia alone, and OSA alone were each associated with significantly greater risk for hypertension (COMISA: hazard ratio [HR], 2.98 [95% CI, 2.90–3.07]; HR, 1.51 [95% CI, 1.49–1.53], and OSA: HR, 2.41 [95% CI, 2.36–2.45]; Table 2) compared with patients without any sleep diagnosis. After adjustment for demographics, behavioral, and clinical factors across the cohort, there was a greater risk for hypertension in those with COMISA (adjusted hazard ratio [aHR], 2.43 [95% CI, 2.36–2.50]), insomnia (aHR, 1.44 [95% CI, 1.42–1.46]), and OSA (aHR, 2.26 [95% CI, 2.22–2.30]) compared with those without either diagnosis.

Among men, there were significant unadjusted effects of COMISA (HR, 2.94 [95% CI, 2.86–3.04]), insomnia (HR, 1.49 [95% CI, 1.47–1.51]), and OSA (HR, 2.39 [95% CI, 2.35–2.44]) on risk for incident hypertension. For women, COMISA (HR, 3.18 [95% CI, 2.94–3.45]), insomnia (HR, 1.63 [95% CI, 1.57–1.69]), and OSA (HR, 2.34, 95% CI, 2.21–2.48) were each associated with greater risk for hypertension. After adjusting the models for men, COMISA remained associated with hypertension risk (aHR, 2.09 [95% CI, 2.02–2.16]), as did insomnia (aHR, 1.27 [95% CI, 1.25–1.29]) and OSA (aHR, 2.00 [95% CI, 1.96–2.05]). Among women only, the fully adjusted models were again significant for COMISA (aHR, 2.21 [95% CI, 2.00–2.42]), insomnia (aHR, 1.40 [95% CI, 1.32–1.47]), and OSA (aHR, 2.04 [95% CI, 1.91–2.19]).

Next, the reference group was changed from no sleep disorder to OSA to understand how much the relative risk of COMISA exceeded that of OSA. In the unadjusted analyses of the entire cohort and by sex, those with COMISA had a >28% to 34% greater risk of hypertension than patients with OSA only. In fully adjusted models, overall, for men, and for women those with COMISA had a 17% to 26% greater risk of hypertension. Next, to examine the role of sex, we included a Sex×Insomnia×OSA interaction term in the fully adjusted model of the whole cohort. While the overall interaction was not significant (P=0.817), interaction of Sex×Insomnia was associated with risk for hypertension (P<0.001), indicating a greater effect of insomnia for women (Table S3) for results based on stepwise adjustment. Sensitivity analyses including health care use, with a washout period, restricting the sample to those with a sleep study, and including BP readings of 140/90 mm Hg in the hypertension definition also aligned with the primary findings (Table S4).

Risk of Incident CVD

Overall, 3.2% of patients developed CVD. The cumulative incidence of CVD per 1000 person‐years was 14.3 overall, 14.3 for men, and 14.9 for women. In unadjusted analyses across the cohort, COMISA (HR, 5.43 [95% CI, 5.19–5.68]), insomnia (HR, 1.59 [95% CI, 1.54–1.65]), and OSA (HR, 4.10 [95% CI, 3.97–4.24]) were all associated with greater risk for all‐cause CVD (Table 3). After adjusting for demographics, health behaviors, and clinical factors in the entire cohort, COMISA (aHR, 3.81 [95% CI, 3.64–3.99]), insomnia (aHR, 1.37 [95% CI, 1.32–1.41]), and OSA (aHR, 3.32 [95% CI, 3.21–3.43]) were each associated with greater risk for all‐cause CVD than that observed for patients without a disorder.

Men with COMISA (HR, 5.54 [95% CI, 5.28–5.80]), insomnia (HR, 1.58 [95% CI, 1.53–1.64]), and OSA (HR, 4.21 [95% CI, 4.07–4.36]) each had a greater, unadjusted risk of CVD, compared with men without a sleep disorder. These effects were similar among women, where the unadjusted effect was significant for COMISA and each sleep disorder (COMISA: HR, 4.45 [95% CI, 4.05–5.13]; HR, 1.60 [95% CI, 1.54–1.65]; OSA: HR, 3.07 [95% CI, 2.72–3.46]). For men, after adjusting for all covariates, COMISA was again associated with a greater risk of CVD (aHR, 3.81 [95% CI, 3.63–4.00]), along with insomnia (aHR, 1.36 [95% CI, 1.31–1.41]) and OSA (aHR, 3.36 [95% CI, 3.25–3.48]). When modeling the adjusted effects of sleep disorders among women, the risk for CVD was greatest with COMISA (aHR, 3.44 [95% CI, 2.98–3.98]) and was also significant for insomnia (aHR, 1.36 [95% CI, 1.22–1.51]) and OSA (aHR, 2.62 [95% CI, 2.32–2.97]; Table S5 for results based on stepwise adjustment).

In unadjusted analyses with OSA as the comparison group, in the entire cohort, and for men and women separately, patients with COMISA had a 57% to 90% greater risk of all‐cause CVD than those without. After adjustment, patients with COMISA still had a 30% to 53% greater CVD risk than those with OSA only. Next, the primary models were rerun with an interaction between sex and the sleep disorders. While there was a significant interaction of Sex×OSA in relation to risk for all‐cause CVD (P<0.001), demonstrating the greater OSA‐associated risk of CVD among men, the Sex×Insomnia×OSA interaction was not significant (P=0.281). Finally, in all sensitivity analyses, the effects remained overall, for men, and for women (Table S6).

Discussion

Across this distinctive nationwide cohort of nearly 1 million men and women who were served by the VA over 20 years, insomnia, OSA, and COMISA were each associated with significantly increased risk of incident hypertension and CVD. COMISA conferred the highest risk, with a >2‐fold increase in hypertension and a >3‐fold increase in CVD after adjusting for demographic, behavioral, and clinical covariates. These associations remained robust across sex‐stratified and sensitivity analyses, underscoring the consistent cardiovascular relevance of disordered sleep. Although COMISA has been linked to cardiometabolic outcomes in prior studies, this investigation extends the evidence to a more recent, predominantly underrepresented Veteran population. The effect sizes for COMISA in this study were notably larger than those in previous research, ^7^ , ^8^ , ^9^ , ^11^ , ^30^ likely reflecting the high comorbidity burden and younger age of this cohort.

Diagnostic and clinical challenges remain. About half of patients were diagnosed with a sleep disorder (13% with insomnia, 21% with OSA, and a separate 14% had COMISA). However, these rates were lower than those observed in military personnel referred for sleep assessment (33% with insomnia, 30% with OSA, and 37% with COMISA), ^31^ indicating that sleep disorders may be more prevalent in active military or underrecognized in the VA. ^12^ , ^16^ , ^32^ Although Veterans commonly experience conditions that are comorbid with insomnia and OSA, ^20^ , ^31^ and data from the general population show greater OSA rates in men and insomnia in women, ^17^ , ^33^ insomnia was diagnosed equally across sexes, and OSA was more prevalent among men than women. VA patients with insomnia may be overlooked or misdiagnosed (eg, depression), and OSA may be underdiagnosed in women Veterans, as seen in the general population. ^5^ , ^34^

While sex differences in sleep disorder presentation are well documented, ^17^ , ^33^ we found few clinically meaningful interactions by sex. COMISA was equally predictive of hypertension and CVD in men and women. Slightly stronger associations were noted for insomnia in women (hypertension) and for OSA in men (CVD). Despite the empirical rationale for conducting sex‐specific analyses, there have been few efforts in relation to COMISA and CVD. In a 43‐month prospective cohort study of 868 older patients (37% women), men with COMISA had a 2.8‐fold increased risk of major adverse cardiovascular events, with no association for women. ^35^ Another 12‐year prospective study of 471 patients with type 2 diabetes (25% women) showed a 4‐fold greater risk of CVD in those with COMISA, with no interaction by sex. ^8^ Investigating sex differences in the effects of disordered sleep on CVD risk should be prioritized in future studies. However, given established sex differences in vulnerability to insomnia and OSA, ^17^ , ^33^ even in the absence of strong statistical interactions, sex‐informed screening, diagnosis, and treatment may be beneficial.

Mechanisms of COMISA and CVD

There is likely a bidirectional association between risk for OSA and insomnia. ^1^ , ^11^ , ^36^ , ^37^ On one hand, OSA can predispose patients to insomnia if respiratory events and postapneic waking periods are interpreted as difficulty staying asleep, ^11^ , ^36^ which can increase patients’ alertness and anxiety about their sleep quality. ^37^ Some patients treated with PAP therapies show greater sleep fragmentation, ^37^ which may be linked to increased hypothalamic–pituitary–adrenal axis activity and adverse metabolic changes. ^37^ Although there is far less evidence, another theory suggests that insomnia‐related sleep deprivation could predispose individuals to a lower arousal threshold, thereby increasing patients’ vulnerability to OSA. ^1^ Either causal association is plausible, and it is also unknown if first developing insomnia versus OSA then contributes differently to cardiovascular risk.

The pathophysiological consequences of COMISA on CVD may occur via several shared mechanistic pathways—sleep duration, fragmented sleep, and arousals—which could be stronger for those who are diagnosed with both sleep disorders. ^38^ Independently and additively, these factors promote autonomic dysfunction and systemic inflammation, followed by increased heart rate and BP, myocyte toxicity, atherogenesis, and endothelial dysfunction. In addition to potential shared pathways, insomnia and OSA also have distinct effects on cardiovascular risk. For insomnia, dysregulation of the hypothalamic–pituitary–adrenal axis axis can alter insulin resistance and lipid levels, while OSA is associated with downstream changes in organ dysfunction, increased ventricular afterload, and aortic dilation. ^38^ Scant data are available about sex differences, but it has been theorized that women with insomnia or OSA have a greater predisposition to adverse cardiovascular effects via inflammation. ^39^ , ^40^

Risk Assessment and Management of COMISA

These findings underscore the need for regular sleep screening in patients with cardiovascular risk factors. Our finding that the elevated CVD risk associated with COMISA is not solely attributable to OSA reinforces the need to assess both disorders in clinical care. Routine screening for both insomnia and OSA is especially critical in the VA system, where more recent Veterans face disproportionately high risks for hypertension and CVD than their counterparts in the general population or earlier cohorts of Veterans. ^13^ Detection of 1 disorder could be leveraged to assess for the other, and incorporating sleep‐focused assessments into primary care and mental health visits may improve early identification and management.

While treatment approaches for COMISA are expanding, they largely draw from interventions that were validated for either condition. ^41^ The VA and Department of Defense have issued guidelines for managing insomnia and OSA, ^42^ yet no guidance currently exists for treating COMISA specifically, and guidelines are also not available from the American Academy of Sleep Medicine. Combined treatment using cognitive behavioral therapy for insomnia (CBTi) alongside PAP has shown greater efficacy than sequential or single‐modality interventions. ^11^ , ^37^ , ^38^ , ^43^ These findings may also inform VA policy by underscoring the importance of integrated care models that bridge sleep medicine, primary care, and cardiology. Although clinical practice guidelines exist for insomnia and OSA separately, ^44^ the absence of VA or Department of Defense guidance specific to COMISA highlights an opportunity for policy and guideline development. Given the robust association between COMISA and CVD risk, implications extend to health care resource planning within the VA, including allocation of sleep study capacity, PAP resources, and access to CBTi and other behavioral interventions.

While CBTi is a promising technique for COMISA management, caution is warranted with nonselective use of CBTi components. For instance, a retrospective study of Veterans showed reduced benefit from brief behavioral therapy among non‐White patients and those with a short sleep duration (<4.1 hours). ^45^ When medications are appropriate, newer nonbenzodiazepine sedative hypnotics and antidepressant agents could be helpful among patients with COMISA. ^37^ Adjunctive strategies such as sleep hygiene and education are also supported. ^46^ Importantly, treatment should take into account symptom profiles and preferences among men and women, respectively. Women are less likely to report classic OSA symptoms ^47^ and are more likely to prefer nonpharmacological treatments. ^32^ Additionally, a survey of 1141 post‐9/11 Veterans (53% women) showed that while women were more likely to endorse sleep as a behavior that they believe could prevent CVD, ^48^ men and women Veterans were as unlikely to endorse getting adequate sleep as part of their CVD prevention. Acknowledging such differences and similarities can enhance detection and education and optimize individualized care. ^49^

Strengths and Limitations

Strengths of this investigation included the robust estimation of risks and subgroup analyses, especially differences between men and women. The 20‐year follow‐up also offered a strong temporal relation between sleep disorders and cardiovascular outcomes, thereby making causality more plausible. Also, highlighting elevated cardiovascular risk among younger adults and emphasizing early detection and treatment of sleep disorders are clinically meaningful and potentially practice changing. Using administrative and electronic health record data and rigorous statistical methods including sensitivity analyses strengthens the validity of the findings.

There are also limitations. First, the sample was limited to post‐9/11 Veterans receiving VA care, reducing generalizability to other Veteran cohorts, non‐VA users, or the general population due to the unique psychosocial characteristics of this group. For example, although military service members are initially viewed as healthier than the general population (ie, the healthy soldier effect ^50^ ), after service they have a greater risk for CVD than non‐Veterans. ^51^

Second, the attainment of diagnoses poses challenges. Because insomnia and OSA are underdiagnosed in the VA system than their true prevalence would suggest, ^16^ , ^32^ it is possible that some Veterans with preexisting disorders were misclassified as incident cases if their conditions were not coded until after VA entry. Exclusion of individuals with prebaseline diagnoses may also introduce selection bias, yielding a cohort that is less representative of the full Veteran population, particularly those whose sleep problems developed during active duty. Future linkage of VA and Department of Defense records may help reduce this bias and improve characterization of sleep disorder trajectories. Also, hypertension may be underdiagnosed among younger Veterans and women. ^52^ , ^53^ To conservatively define the hypertension outcome, patients were not considered hypertensive on the basis of qualifying BP measurements alone, although results were similar when outpatient BP readings were added to the hypertension definition.

Third, as this retrospective cohort study involved electronic health record data, risks of misclassification and missing data remain. For example, the electronic health record does not capture symptom onset, severity, or timing before diagnosis, so the diagnostic sequence may reflect surveillance bias. The finding that 72% of Veterans were first diagnosed with insomnia could represent a diagnostic cascade rather than true new‐onset comorbidity, potentially inflating COMISA case identification. In addition, data on OSA or insomnia severity and PAP adherence were unavailable. Future prospective studies could help confirm these findings. ^13^ , ^14^ Relatedly, there was no information available on OSA or insomnia severity or PAP adherence. Future prospective studies could help to confirm these findings as well.

Fourth, we could not assess insomnia phenotypes such as short sleep duration, which may be more strongly associated with hypertension and CVD risk, ^10^ , ^54^ possibly leading to effect size underestimation. Fifth, residual confounding from unmeasured lifestyle factors, including sleep duration and quality, physical activity, diet, and stress, is likely and may vary by sex, race, ethnicity, and socioeconomic status. ^55^ , ^56^ Although we adjusted for a range of behavioral and clinical covariates, future research should incorporate more detailed lifestyle and environmental data. Sixth, indications for antihypertensive medications were not available. Finally, non‐VA care was not captured.

Further research is needed to clarify the biological and behavioral mechanisms linking COMISA to cardiovascular risk in Veterans, including the potential moderating roles of psychiatric comorbidity and sex. Prospective and mechanistic studies are required to examine if COMISA’s cardiovascular impact differs by diagnostic sequence (ie, insomnia versus OSA onset) or by specific pathophysiologic profiles (eg, inflammation, autonomic dysfunction). In addition, implementation‐focused research within the VA is warranted to evaluate integrated interventions, such as combined CBTi and PAP therapy, and to test strategies that address barriers unique to Veterans, including high rates of psychiatric comorbidity, medical multimorbidity, and rural residence. Such work will be critical to inform precision approaches to COMISA detection and management within the VA system.

In summary, COMISA was associated with a >2‐fold increase in hypertension risk, and a >3‐fold increase in CVD risk, consistent across sexes. These findings strengthen the rationale for improved recognition of sleep disorders in cardiovascular health and highlight the importance of early detection and integrated management. While further research is needed to evaluate the effects of guideline‐directed sleep care, early detection and integrated treatment of insomnia and OSA may be a critical, modifiable pathway to improving hypertension and CVD outcomes.

Sources of Funding

A.E. Gaffey’s effort was supported by a grant from the National Heart, Lung, and Blood Institute of the National Institutes of Health (K23HL168233).

Disclosures

None.

Supporting information