Simulated intention-to-treat analysis based on clinical parameters of patients at high risk for sleep apnea derivated to respiratory polygraphy

Abstract

Purpose: Obstructive Sleep Apnea-Hypopnea Syndrome (OSAS) is a public health problem. We designed a pilot study to validate empiric indication of CPAP therapy in a population with moderate- to-high pre-test probabilities who underwent self-administered home-based respiratory polygraphy (RP). Methods: A cross-sectional simulation study was performed. CPAP therapy could be indicated by two independent blind observers. Observer 1's decision was based on the results of STOP-BANG (SBQ) and Epworth Sleepiness Scale (ESS) and Observer 2 used all objective data provided by RP + SBQ + ESS. Results: We evaluated 1763 patients; 1060 men and 703 women (39.2%) with a mean age of 53.6±13.8 and a body mass index (BMI) of 32.8±7.5 kg/m2. We found evidence of mild (34.1%), moderate (26.6%), and severe (18.3%) There were Apnea-Hypopnea Index (AHI) relationship between > 5 or < 5 SBQ and RP AHI (p<0.05). BMI > 25 kg/m2 + snoring (S) + observed apnea (O) + 1 of the following: ESS > 11, hypertension (HT) or > 5 SBQ components showed sensitivity of 40% (CI95%: 37.3-43) and specificity of 95.1% (CI93.4-96.4). The performance of 5 SBQ components with regard to gender and empirical CPAP therapy was; (women vs. men): AUC-ROC 0.625 (CI95%: 0.599-0.651) vs. 0.70 (CI95%: 0.68-0.72), p<0.01, respectively. Conclusions: STOP-BANG and ESS made it possible to indicate CPAP reliably (low rate of false-positive results) in 20-40% of patients who needed such therapy according to clinical history and RP results. These clinical criteria performed better in male.