Associations of body roundness index and body mass index with obstructive sleep apnea: evidence from two cohorts study.

BACKGROUND

Obstructive sleep apnea (OSA) is closely associated with obesity. Traditional anthropometric indices body mass index (BMI) could not assess body fat content precisely. The body roundness index (BRI) has emerged as a novel adiposity metric reflecting central fat distribution. This study aimed to compare BRI and BMI in assessing the risk and worsening of OSA severity in community-based cohort study.

METHODS

We analyzed data from two well-characterized cohorts: the Osteoporotic Fractures in Men Study (MrOS) and the Sleep Heart Health Study (SHHS). Statistical and machine learning approaches were used to assess associations between BRI, BMI, and OSA. Sex-stratified analyses and models evaluating of future OSA severity categories using baseline features were also performed.

RESULTS

BRI showed stronger associations with severe baseline OSA than BMI in both cohorts (MrOS: OR = 1.50 vs. 1.22, SHHS: OR = 1.21 vs. 1.12). BRI demonstrated significant dose-response trends across quartiles, while BMI was only significant in the highest quartile. RCS analyses indicated linear relationships between both BRI, BMI, with severe baseline OSA. A significant sex interaction was observed (P for interaction = 0.015), with a stronger association in men (OR = 1.44, 95% CI: 1.16-1.66) than in women (OR = 1.13, 95% CI: 1.02-1.26). Random forest models using baseline features showed higher AUCs for BRI than BMI in assessing worsening of follow-up OSA severity (MrOS: 0.67 vs. 0.57, p < 0.05, SHHS: 0.73 vs. 0.68, p < 0.01). SHAP interpretation confirmed BRI as a superior feature, with greater feature contribution than BMI.

CONCLUSION

BRI showed closer associations with severe baseline OSA than BMI and worsening OSA severity during follow-up in two community-based cohorts. These findings suggest that BRI may serve as a useful anthropometric marker for identifying individuals at higher risk of severe or worsening OSA.