Quality Control of the Ambulatory Polygraphy Using Automatic Analysis

Abstract

It is necessary to ensure the quality of sleep studies conducted at home given that there can be potential variations. Automatic analysis is simple and could help in an audit. The objective is to find a predictive model of visual reading using an automatic analysis of saturation and respiratory signal in order to establish a reading standard with a polygraph used at home on patients who have sleep apnea-hypopnea clinical symptoms. The analysis was carried out using the following two definitions of hypopnea: an event with a duration of >or= 10 s with a decrease of > 30% of the respiratory signal; and an event associated either with a desaturation of >or= 3% or with a desaturation of >or= 4%. A total of 189 studies were selected from a representative sample of 218 patients. Two pneumologists carried out the readings together. The agreement between the visual respiratory disturbance index (RDI) [ie, apneas plus hypopneas] for both definitions and the automatic respiratory signal analysis (ie, automatic RDI [RDIa]) or the automatic desaturation index of 3% (DI3%a) and of 4% (DI4%a) showed limits from a Bland-Altman plot that were too large. However, a multiple linear regression analysis with RDIa and DI3%a or RDIa and DI4%a presented an acceptable level of agreement with RDI for both definitions (p < 0.001; r(2) = 96.2% and 97%, respectively). The 95% confidence interval for the differences between the RDI and the model was +/- 10.1 or +/- 8.8 events per hour, so a study should be revised outside of these limits. A predictive multiple regression model that uses the automatic analysis of the oximetry and respiratory signal could establish a standard for the visual reading of polygraphy at home.