Reliability and validity testing of automated scoring in obstructive sleep apnea diagnosis with the Embletta X100.

Abstract

OBJECTIVES/HYPOTHESIS: To verify the reliability and validity of automated scoring and compare it to that of manual scoring for diagnosing obstructive sleep apnea using an Embletta X100 level 2 portable device.

STUDY DESIGN: Retrospective study.

METHODS: A total of 116 patients with suspected obstructive sleep apnea who had successfully received portable polysomnography with the Embletta X100 were examined. All polysomnography data were analyzed by automated and manual methods. Manual scoring was performed according to the revised American Academy of Sleep Medicine 2012 criteria. Automated scoring was analyzed using the automatic algorithm, which was updated with the American Academy of Sleep Medicine 2012 criteria. All parameters were evaluated statistically using correlation analysis and paired t tests.

RESULTS: The apnea-hypopnea index for automated scoring and manual scoring with the Embletta X100 were moderately correlated (r = 0.76, P < .001). However, there was poor agreement (Bland-Altman plot, κ = 0.34, 0.33, and 0.26; cutoff value = 5, 15, and 30), and the apnea-hypopnea index data were generally excessively underestimated based on diagnostic agreement and disagreement criteria. Furthermore, the apnea-hypopnea index severity (Kendall tau-b = 0.62) between automated and manual scoring lacked good concordance.

CONCLUSIONS: Automated scoring using the Embletta X100 was statistically moderately related to the manual scoring results. However, automated scoring tended to excessively underestimate the apnea-hypopnea index data compared to manual scoring. Thus, manual scoring by a sleep expert is essential for obstructive sleep apnea diagnosis with the Embletta X100.

LEVEL OF EVIDENCE: 4.