Abstract: Objective　To evaluate the application of Pediatric Sleep Questionnaire （PSQ） by item analysis using item response theory.Methods　Data were obtained from the survey of children at the age of 3-14 years from 7 districts in Beijing from May to July 2015.By using multi-stage stratified cluster random sampling method，a total of 11 420 children from 25 kindergartens，primary and middle schools were selected in 7 districts （Xicheng，Chaoyang，Changping，Shunyi，Fangshan，Huairou and Mentougou）．All the parents of children were recruited to complete the PSQ，which consists of four dimensions related to sleep disordered breathing in children：breathing，sleep，behavior and others.The item response theory with two-parameter Logistic model was used to conduct item analysis to calculate the discrimination and difficulty of every item and draw item characteristics curve.Results　The analysis of two-parameter Logistic model showed that most of the items had a good discrimination，but the items of “Does your child occasionally wet the bed” and “Is your child overweight” had a significant lower discrimination parameter（0.441 and 0.124，respectively） and a higher difficulty parameter （4.378 and 6.172，respectively）．The test information function showed almost the same curves of the questionnaire with or without the two items.Conclusion　A simplified version of PSQ is recommended in Chinese children，although a good evaluation effect in the screening of children's sleep disorder is found.

Key words: Sleep, Questionnaires, Item response theory, Logistic models

摘要： 目的　采用项目反应理论对儿童睡眠问卷进行项目分析，评价儿童睡眠问卷在国内的适用性。方法　2015年5—7月采用多阶段分层整群随机抽样方法，随机抽取了北京市西城、朝阳、昌平、顺义、房山、怀柔和门头沟共7个区县的25所幼儿园和中小学学校，共计11 420例儿童。采用儿童睡眠问卷调查其睡眠情况，该问卷主要包含与儿童睡眠呼吸障碍相关的4个维度，即呼吸、睡眠、行为、其他。统计分析采用基于二参数Logistic模型的项目反应理论，计算各条目的区分度和难度参数，并绘制项目反应特征曲线。结果　大多数条目具有较好的区分度，但“睡觉偶尔尿床”和“是否超重”的区分度明显低于其他条目（区分度参数估计值分别为0.441和0.124），且难度参数均较大（分别为4.378和6.172）。测验信息量结果提示删除这两个条目后，问卷提供的信息量几乎没有变化。结论　儿童睡眠问卷在评价儿童睡眠障碍中可以提供较好的初筛效果，但国内应用可以考虑对其进行简化。

关键词: 睡眠, 问卷调查, 项目反应理论, Logistic模型