Kim Y, Kurachi M, Horita M, Matsuura K, Kamikawa Y
Department of Neuropsychiatry, Toyama Medical and Pharmaceutical University, Japan.
Jpn J Psychiatry Neurol. 1993 Mar;47(1):91-7. doi: 10.1111/j.1440-1819.1993.tb02035.x.
The purpose of this study was to elucidate the agreement of visual scoring of all-night polysomnographic recordings among many scores from different laboratories. Ten scorers including the author from different laboratories in Japan scored the same paper recordings of two young male subjects. We calculated the agreement rate for each stage using an epoch by epoch analysis. In both records, the agreement rates for stages 2 and R were high; on the contrary, those for stages 3 and 4 were low. After adding a supplementary definition of high voltage slow wave in deep sleep, we scored the first NREM period of another subject. The mean agreement rate for stage 3 among 10 scorers was significantly higher than those of the two former subjects. However, the agreement for stage 4 did not change so much. This result demonstrates that there is much interrater (laboratory) variability of visual scoring, especially in slow wave sleep. When the result of automatic scoring is compared to that of the visual scoring to evaluate the reliability of automatic scoring, these findings must be considered.
本研究的目的是阐明不同实验室的众多评分者对整夜多导睡眠图记录进行视觉评分的一致性。包括作者在内的来自日本不同实验室的10名评分者对两名年轻男性受试者相同的纸质记录进行了评分。我们通过逐段分析计算了每个阶段的一致率。在两份记录中,第2阶段和快速眼动(R)阶段的一致率较高;相反,第3阶段和第4阶段的一致率较低。在对深度睡眠中的高电压慢波添加补充定义后,我们对另一名受试者的第一个非快速眼动(NREM)期进行了评分。10名评分者中第3阶段的平均一致率显著高于前两名受试者。然而,第4阶段的一致性变化不大。这一结果表明,视觉评分存在很大的评分者间(实验室间)差异,尤其是在慢波睡眠方面。在将自动评分结果与视觉评分结果进行比较以评估自动评分的可靠性时,必须考虑这些发现。