Independent association between hourly apnea-hypopnea duration and prevalence of diabetes in male adults

Abstract

Objective

Although the apnea-hypopnea index (AHI) is currently used as a diagnostic indicator to assess the severity of obstructive sleep apnea (OSA), the AHI alone cannot adequately identify the risk of diabetes mellitus. On the other hand, the hourly apnea-hypopnea duration (HAD) has been shown to be a better predictor of nocturnal hypoxemia in patients with OSA. This study aimed to evaluate whether HAD is associated with the prevalence of diabetes.

Methods

The study sample was 590 male participants with suspected OSA who completed a home sleep apnea test. The main outcome of the study was the prevalence of diabetes, and a multivariable logistic regression model was constructed to estimate the odds ratio (ORs) of HAD to the prevalence of diabetes. The main model was adjusted for AHI, age, body mass index, tobacco smoking, alcohol, insomnia, Epworth Sleepiness Scale score, sleep efficiency, hypertension, and cardiovascular disease.

Results

The prevalence of diabetes was 13.90 % in the selected population (quartile age 48.00–68.00 years). The proportion of diabetes was higher in the second and third tertile groups of HAD, with adjusted ORs of 2.48 (95 % CI 1.14, 5.39) and 4.37 (95 % CI 1.22, 15.70), respectively. In the sensitivity analysis, when the oxygen desaturation index (ODI) and lowest oxygen saturation by pulse oximetry were included simultaneously in the same statistical model, the results remained statistically significant. In the multivariable regression analysis, neither AHI nor ODI showed a significant association with the prevalence of diabetes (P > 0.05).

Conclusion

After adjusting for multiple potential confounders, higher HAD tertiles were associated with a higher prevalence of diabetes. In the clinical assessment of OSA-related adverse health outcomes, the HAD deserves more consideration.