Comparative Study between Sequential Automatic and Manual Home Respiratory Polygraphy Scoring Using a Three-Channel Device: Impact of the Manual Editing of Events to Identify Severe Obstructive Sleep Apnea

Objective. According to current guidelines, autoscoring of respiratory events in respiratory polygraphy requires manual scoring. The aim of this study was to evaluate the agreement between automatic analysis and manual scoring to identify patients with suspected OSA. Methods. This retrospective study analyzed 791 records from respiratory polygraphy (RP) performed at home. The association grade between automatic scoring and manual scoring was evaluated using Kappa coefficient and the agreement using Bland and Altman test and intraclass correlation coefficient (CCI). To determine the accuracy in the identification of AHI ≥ 30 eV/h, the ROC curve analysis was used. Results. The population analyzed consisted of 493 male (62.3%) and 298 female patients, with an average age of 54.7 ± 14.20 years and BMI of 32.7 ± 8.21 kg/m(2). There was no significant difference between automatic and manual apnea/hypopnea indexes (aAHI, mAHI): aAHI 17.25 (SD: 17.42) versus mAHI 21.20 ± 7.96 (p; NS). The agreement between mAHI and aAHI to AHI ≥ 30 was 94%, with a Kappa coefficient of 0.83 (p < 0.001) and a CCI of 0.83. The AUC-ROC, sensitivity, and specificity were 0.99 (CI 95%: 0.98-0.99, p < 0.001), 86% (CI 95%: 78.7-91.4), and 97% (CI 95%: 96-98.3), respectively. Conclusions. We observed good agreement between automatic scoring and sequential manual scoring to identify subjects with AHI ≥ 30 eV/h.