C75-17 Insomnia, Sleep Apnea, and Incidence of Pre-diabetes Among U.S. Military Veterans

Background

Prediabetes is a prevalent, clinically actionable stage of dysglycemia that precedes type 2 diabetes mellitus (T2DM), and is strongly associated with future cardiovascular disease, and healthcare utilization within the Veterans Health Administration (VA). Because progression to T2DM is often preventable, identifying modifiable risk factors is a key public health priority. Insomnia and obstructive sleep apnea (OSA), two of the most common sleep disorders among the general population and Veterans, are each independently associated with impaired glucose regulation and insulin resistance. Yet, past studies have largely focused on incident T2DM rather than prediabetes, where prevention efforts may have the greatest impact. Moreover, the risk of incident prediabetes associated with comorbid insomnia and OSA (COMISA), beyond either disorder alone, remains incompletely characterized, particularly in younger and middle-aged Veteran populations. Sex differences in sleep disorders and their metabolic consequences also warrant consideration.

Methods

We conducted a retrospective cohort study of U.S. military Veterans free of pre-diabetes at baseline and followed them for incident pre-diabetes. Exposures were categorized into four mutually exclusive groups: neither disorder (reference), insomnia only, OSA only, and COMISA. Time-to-event analyses were performed using Cox proportional hazards models. Models were evaluated for the entire sample and by sex, first unadjusted and then adjusted for demographics, body mass index category, smoking status, alcohol and drug abuse, cardiometabolic comorbidities, and psychiatric comorbidities. Effect modification was assessed using an insomnia-by-OSA interaction term.

Results

In multivariable Cox models, insomnia only, OSA only, and COMISA were each independently associated with increased risk of incident prediabetes compared with neither condition. Overall, adjusted hazard ratios (HR; 95% CI) were 1.33 (1.16-1.53 [insomnia]), 2.91 (2.58-3.29 [OSA]), and 3.58 (3.04-4.20 [COMISA]. In men, HR were 1.25 (1.07-1.45 [insomnia]), 3.05 (2.69-3.46 [OSA]), and 3.84 (3.25-4.55 [COMISA]). In women, HRs were 1.81 (1.31-2.48 [insomnia]), 1.78 (1.14-2.78 [OSA]), and 1.82 (1.00-3.33 [COMISA]). The insomnia×OSA interaction term was not significant after adjustment, indicating no effect modification between OSA and insomnia.

Conclusion

Among U.S. Veterans, both insomnia and OSA were independently associated with increased risk of incident pre-diabetes, with the highest risk observed in those with COMISA. Although COMISA identifies a high-risk sleep phenotype, the lack of an interaction shows that risk may be distinct to each. Associations were broadly consistent by sex, supporting relevance in both men and women Veterans. These findings highlight the potential value of early identification and treatment of sleep disorders as part of pre-diabetes prevention efforts in high-risk populations.

This abstract is funded by: None