CPAP Adherence Predictors in a Randomized Trial of Moderate-to-Severe OSA Enriched With Women and Minorities.

Background: Suboptimal CPAP adherence in OSA clinical trials involving predominantly white men limits interpretability and generalizability. We examined predictors of CPAP adherence in a clinical trial enriched with minorities.

Methods: The Sleep Apnea Stress Study-a randomized, double-blind, sham-controlled trial of patients with moderate-to-severe OSA-included participants with complete 8-week adherence data (n = 138). Overnight 14-channel polysomnography, anthropometry, socioeconomic status, mood questionnaires, and week 1 CPAP adherence were analyzed via adjusted linear models relative to CPAP adherence (average minutes per night usage).

Results: Overall, age was 51 ± 12 years, 55% of the patients were male, 55% were white, BMI was 36.7 ± 7.7 kg/m, and median apnea-hypopnea index was 20 (interquartile range, 13-37). In univariate analyses adherence increased with randomization to active CPAP (81 min; 95% CI, 30-132), increasing age (35 min/decade; 95% CI, 13-57), white race (78 min, 95% CI, 26-129), and per hour of week 1 adherence (41 min, 95% CI, 32-51). Active CPAP (48 min, 95% CI, 6-91), increasing age (27 min/decade, 95% CI, 10-44), and higher 1-week adherence (36 min/h, 95% CI, 27-46) were significantly associated with improved adherence in multivariable analyses. Subgroup analyses showed stronger associations of adherence with treatment arm in whites and increasing age in minorities. Increasing age and white race were more strongly associated with adherence in women.

Conclusions: In this trial with near-even sex distribution and high ethnic minority representation, we identified CPAP assignment, increasing age, and early adherence to be associated with improved adherence in addition to sex-specific and race-specific adherence differences. These results can inform targeted clinical trial adherence optimization strategies.

Trial Registry: ClinicalTrials.gov; No.: NCT00607893; URL: www.clinicaltrials.gov.