Abstract
BACKGROUND: Obstructive sleep apnea (OSA) causes high morbidity and mortality and is independently associated with an increased likelihood of multiple complications. The diagnosis of OSA is presently time-consuming, labor-intensive and inaccessible. AIM: This study sought to develop a simple and efficient model for identifying OSA in Chinese adult population. METHODS: In this study, the efficiency of Epworth Sleepiness Scale (ESS) and a new established prediction model for screening OSA were evaluated in the test cohort (2,032 participants) and confirmed in an independent validation cohort (784 participants). RESULTS: In the test cohort, a high specificity (82.77%, 95% confidence interval [CI], 77.36-87.35) and a moderate sensitivity (61.65%, 95% CI, 59.35-63.91) were obtained at the threshold of nine for the ESS alone. Notably, sex-stratified analysis revealed different optimum cut-off points: nine for males and six for females. The new generated screening model, including age, waist circumference, ESS score, and minimum oxygen saturation (SaO2) as independent variables, revealed a higher sensitivity (89.13%, 95% CI, 87.60-90.53) and specificity (90.34%, 95% CI, 85.85-93.77) at the best cut-off point. Through receiver operating characteristics curve analysis, the area under the receiver operating characteristics curve of the model was found significantly larger than that of the ESS alone (0.955 vs. 0.774, P<0.0001). All these results were confirmed in the validation cohort. CONCLUSIONS: A practical screening model comprising minimum SaO2 and other parameters could efficiently identify undiagnosed OSA from the high-risk patients. Additionally, a sex-specific difference should be considered if the ESS alone is used.