Comparison of the automatic analysis versus the manual scoring from ApneaLink™ device for the diagnosis of obstructive sleep apnoea syndrome.

OBJECTIVE

The purpose of this study was to compare the performance of the automated detection versus the manual scoring from the ApneaLink™ device to diagnose obstructive sleep apnoea syndrome (OSAS).

METHODS

All participants (96) performed the ApneaLink™ (AL) and polysomnography (PSG) simultaneously in the sleep laboratory. The two recordings were interpreted blindly. The hypopnoea criterion used for the analysis of both automatic and manual ApneaLink™ was a fall in airflow ≥50% of baseline for ≥10 s. The agreement between AL and PSG and the interobserver concordance was calculated. ROC analysis, sensitivity and specificity were assessed for the different ApneaLink™ and OSAS criteria.

RESULTS

Ninety patients were included (69 men; mean age, 49.6; median RDI, 13.9; median BMI, 29.3 kg/m(2)). The automatic apnoea/hypopnoea index (AHI-a) showed a lower agreement with the respiratory disturbance index (RDI) than the manual apnoea/hypopnoea (AHI-m) [AHI-a/RDI: intraclass correlation coefficient (ICC) 0.88 versus AHI-m/RDI: ICC 0.91]. The manual scoring (MS) showed a similar sensitivity and a higher specificity than the automatic scoring (AA) for the detection of OSAS, defined as an RDI ≥ 5 (sensitivity and specificity AA and MS: 89%/89%, 60%/86.7%, respectively). The accuracy of the automatic and manual scoring of the AL was similar when OSAS was defined as an RDI ≥ 20 or 30. The ApneaLink™ manual scoring had a very good interobserver agreement (k = 0.86).

CONCLUSIONS

The manual scoring of an ApneaLink™ recording was better than the automatic scoring in terms of agreement with RDI and to discriminate patients with OSAS. The hand scoring did not improve the accuracy of automatic scoring in patients with severe OSAS.