Differences between manual and automatic analysis in determining the severity of obstructive sleep apnea using home sleep apnea testing

Introduction: Home sleep apnea testing (HSAT) is a diagnostic measure for obstructive sleep apnea hypopnea syndrome (OSAHS) in moderate/high risk patients. Some HSAT companies contain automatic analysis (AA). However, guidelines recommend manual analysis (MA) despite the weak evidence for this recommendation. Objective: Evaluate the concordance between AA and MA of HSAT to make either a diagnosis and severity classification. Methods: We evaluated AA and MA of HSAT between 2015 and 2016. The study was a blind analysis reviewed by two physicians using currents recommendations. The differences between AA and MA were compared with single variable T analysis, inter-scorer agreement for diagnosis was evaluated with Cohen Kappa coefficient, correlation was examined using Tau-b Kendall, and Bland-Altman plot was constructed to analyze differences between AA and MA. Results: One hundred and ninety-eight patients were included. In our study, the mean age was 50 +/- 15 years, 83% male, BMI 30 +/- 5 and neck circumference 41 +/- 4 cm. Eighty-two percent of subjects showed an apnea-hypopnea index (AHI) > 5 ev/h. Thirty-five percent of patients with OSAHS were mild (AHI: 5 -15 ev/h), 34% moderate and 31% severe (>30 ev/h). The kappa coefficient between physicians was 1.0 (high), between AA and MA was 0.58 (moderate) for the diagnosis of OSAHS and 0.33 (weak) for severity with 0.70 Tau-b. The AA underestimates the IAH -8 ev/h, (95% CI -9 to -7 ev/h, p < 0.001) and delivers a misclassification of severity by 47%. Conclusions: AA underestimates the rate of respiratory events and alters the classification of the severity of the disease and may modify the therapeutic approach. (C) 2018 Elsevier B.V. All rights reserved.