Comparison between automatic and manual scoring of 301 home respiratory polygraphs with ApneaLink device

Abstract

Introduction: Obstructive sleep apnea syndrome (OSAS) is associated with increased car­diovascular and metabolic morbidity and mortality, and poor quality of life. Its effective diagnosis and treatment improve individual and public health. Aim: To evaluate concor- dance between automatic versus manual analysis of the ApneaLink device to diagnose and classify OSAS in patients with clinical suspicion. Material and Method: Retrospective evaluation of 301 respiratory polygraphs from HOSCAR. Correlation, general agreement and concordance between parameters obtained manually and automatically are measured using Pearson’s coefficient, intraclass correlation coefficient, and Bland and Altman graph. Results: In 11.3% of cases, the automatic analysis misinterpreted the flow signal. There were no significant differences between automatic (AHIa 18.9 ± 17.5) and manual (AHIm 20.8 ± 19.4) apnea-hypopnea indices r = 0.97 (95% CI:0.9571 to 0.9728, p < 0.0001) and nor between automatic (82.1 ± 7.6) and manual (83.1 ± 6.8) minimum oxygen saturation r = 0.85 (95% CI: 0.8108 to 0.8766, p < 0.0001). There was no good correlation between au- tomatic and manual analysis in the classification of central apneas, r = 0.51(95% CI:0.4238 to 0.5942, p < 0.0001). There was an underestimation of the severity of OSAS by automatic analysis in 11% of cases. Conclusion: The diagnosis delivered automatically by ApneaLink could be accepted without additional manual confirmation only in cases classified as severe. For minors AHI, confirmation through manual expert analysis would be required.