Comparison of Somnologica 3 Computerized Polysomnographic Systems Analysis and Manual Assessment of Sleep Apnea

使用Somnologica 3電腦分析睡眠呼吸中止和人工評估的比較

10.29806/TM.200608.0002

張博瑞(Po-Jui Chang)；謝孟亨(Meng-Heng Hsieh)；羅友倫(Yu-Lun Lo)；羅君禹(Chun-Yu Lo)；鍾福財(Fu-Tsai Chung)；郭志熙(Chih-Hsia Kuo)；林鴻銓(Horng-Chyuan Lin)

阻塞性睡眠呼吸中止症候群 ； 電腦化多重睡眠檢查系統 ； obstructive sleep apnea syndrome ； computerized polysomnographic systems

胸腔醫學

21卷4期（2006 / 08 / 01）

313 - 320

英文

Sleep disorders are increasingly being recognized by physicians, raising demand at sleep laboratories, and prompting a search for user-friendly methods of data analysis. Computerized polysomnographic systems have become a commonly used tool in sleep laboratories in Taiwan and throughout the world. This study was designed to compare the accuracy of data analysis using computerized Somnologica 3 systems and manual interpretation. Twenty-three patients with suspected respiratory sleep disorders were referred to our department from outpatient clinics. Each individual underwent a full-night sleep study in our sleep laboratory. Recorded data were simultaneously analyzed using the computerized Somnologica 3 system (Version Ⅱ software) and interpreted by a pulmonologist thoroughly experienced in polysomnography, using the standard Rechtschaffen and Kales criteria. When comparing autoscoring and manual analysis, our results revealed that the computerized system produced more substantial errors with respect to the duration of each sleep stage (S1 sleep time: 34.9±5.7 vs. 74.2±10.4min; S2: 94.5±15.1 vs. 208.9±15.8min; S4: 36.4±5.4 vs. 14.5±4.1min; REM: 26.1±5.4 vs. 55.4±6.2min, respectively; all p＜0.05), thus leading to a reduction not only in estimated total sleep time (226.2±19.9 vs. 376.4±16.4 min; p＜0.05), but also a decrease in the number of hypopnea/apnea events (AHI 13.0±19.5 vs. 18.9±3.6, p＜0.05). Based on the evidence from this study, therefore, we suggest that polysomnographic records autoscored by Somnologica 3 systems are inaccurate and will underestimate the severity of sleep-related disorders. A thoroughly experienced polysomnographer is needed for all data interpretation in clinical practice.

Sleep disorders are increasingly being recognized by physicians, raising demand at sleep laboratories, and prompting a search for user-friendly methods of data analysis. Computerized polysomnographic systems have become a commonly used tool in sleep laboratories in Taiwan and throughout the world. This study was designed to compare the accuracy of data analysis using computerized Somnologica 3 systems and manual interpretation. Twenty-three patients with suspected respiratory sleep disorders were referred to our department from outpatient clinics. Each individual underwent a full-night sleep study in our sleep laboratory. Recorded data were simultaneously analyzed using the computerized Somnologica 3 system (Version Ⅱ software) and interpreted by a pulmonologist thoroughly experienced in polysomnography, using the standard Rechtschaffen and Kales criteria. When comparing autoscoring and manual analysis, our results revealed that the computerized system produced more substantial errors with respect to the duration of each sleep stage (S1 sleep time: 34.9±5.7 vs. 74.2±10.4min; S2: 94.5±15.1 vs. 208.9±15.8min; S4: 36.4±5.4 vs. 14.5±4.1min; REM: 26.1±5.4 vs. 55.4±6.2min, respectively; all p＜0.05), thus leading to a reduction not only in estimated total sleep time (226.2±19.9 vs. 376.4±16.4 min; p＜0.05), but also a decrease in the number of hypopnea/apnea events (AHI 13.0±19.5 vs. 18.9±3.6, p＜0.05). Based on the evidence from this study, therefore, we suggest that polysomnographic records autoscored by Somnologica 3 systems are inaccurate and will underestimate the severity of sleep-related disorders. A thoroughly experienced polysomnographer is needed for all data interpretation in clinical practice.