A Predictive Model For The Obstructive Sleep Apnea-hypopnea Syndrome

Objective To establish a relatively reliable predictive model for Obstructive Sleep Apnea-Hypopnea Syndrome(OSAHS) according to clinical findings and pulse oximetry data in OSAHS patientsMethods Case-control study was used .109OSAHS patients and 127 non-OSAHS patients were included in the light of defined criterions. Their data were analyzed by singular factor analysis .We found out significantly statistical variable and then used logistic regression. Two logistic regression equations were established . Mode one :The regression about commonly clinical features. Mode two: The regression about commonly clinical features joined pulse oximetry data in OSAHS patients . Receive Operator Characteristic Curves were made and the cutoff points of probabilities were decided . We calculated the sensitivity and specificity about two models. We analyzed the missed data of patients .At last , We enlarged samples and appreciated two models. The reliability of model two is better than that of model one .Results The predicted probability of OSAHS was calculated using the following equation:P=e~x/ (1+e~x) , x=-1.977+1.27a-1.13b+0.13c ,a=l whennocturnal respiratory effort-related arousal is reported and zero when it is not, b=l when it is male and 2 when it is female c=DI(desaturation index ).A probability cutoff point(=0.15) was decided .They were considered to be patients when their probabilities were equal to or greater than 0.15.The sensitivity was 96.3% and the specificity was about 50%.Conclusion This study shows that analysis of clinical features joined pulse oximetry data of patients presenting with suspected OSAHS can predict and screen patients.