Comparison of Manual and Automatic Scoring of limited channel sleep studies:Noxturnal Software correlates well with manual scoring in severe OSA

Abstract

Introduction: Current practice at Birmingham Heartlands Hospital is for a physiologist to manually score limited channel sleep studies with apnoea-hypopnea index (AHI) as the outcome. Analysis software available to score these events may reduce the delay between diagnosis and commencing treatment, which is useful in severe patients (AHI≥30/hour). Aim: To determine the reliability of auto scoring using Noxturnal software compared with manual scoring of sleep studies with AHI≥30/hour. Methods: First ten severe OSA cases identified each month during 2015 (n=120) were included. All subjects received NOX T3 during one night. Time frame (lights off and on) was adjusted to the same range for auto and manual scoring. Agreement between measurements of AHI was determined by Bland-Altman plot. Pearson’s correlation coefficient was used for other events. Results: The average recording quality was 91.73%. Pearson coefficient shows a strong correlation in all the events analysed. Data was skewed towards higher AHI with manual scoring. Nine studies changed OSA severity category to moderate by using the autoscoring system. Conclusion: Automatic analysis is reliable and efficient. There is no clinical difference in the outcome. More research including larger sample studies as well as other OSA severities is needed.