Manual vs. automated analysis of polysomnographic recordings in patients with chronic obstructive pulmonary disease

Purpose The sleep quality, as assessed by polysomnography (PSG), of patients with chronic obstructive pulmonary disease (COPD) can be severely disturbed. The manual analysis of PSGs is time-consuming, and computer systems have been developed to automatically analyze PSGs. Studies on the reliability of automated analyses in healthy subjects show varying results, and the purpose of this study was to assess whether automated analysis of PSG by one certain automatic system in patients with COPD provide accurate outcomes when compared to manual analysis. Methods In a retrospective study, the full-night polysomnographic recordings of patients with and without COPD were analyzed automatically by Matrix Sleep Analysis software and manually. The outcomes of manual and automated analyses in both groups were compared using Bland–Altman plots and Students’ paired t tests. Results Fifty PSGs from patients with COPD and 57 PSGs from patients without COPD were included. In both study groups, agreement between manual and automated analysis was poor in nearly all sleep and respiratory parameters, like total sleep time, sleep efficiency, sleep latency, amount of rapid eye movement sleep and other sleep stages, number of arousals, apnea–hypopnea index, and desaturation index. Conclusion Automated analysis of PSGs by the studied automated system in patients with COPD has poor agreement with manual analysis when looking at sleep and respiratory parameters and should, therefore, not replace the manual analysis of PSG recordings in patients with COPD.