Yarlas Aaron, White Michelle K, St Pierre Danielle G, Bjorner Jakob B
QualityMetric, 1301 Atwood Ave, Suite 216E, Johnston, RI, USA.
Department of Public Health, University of Copenhagen, Gothersgade 160, DK-1123, Copenhagen, Denmark.
J Patient Rep Outcomes. 2021 May 19;5(1):40. doi: 10.1186/s41687-021-00311-3.
The 12-item Medical Outcomes Study Sleep Scale (MOS Sleep Scale) has been used to capture patient-reported sleep problems in hundreds of studies. A revised version of the MOS Sleep Scale (MOS Sleep-R) was developed that uses simplified response sets, provides interpretable norm-based scoring, and has two recall versions (one-week or four-week). The objective of this study was to evaluate the psychometric properties (reliability and construct validity) of the MOS Sleep-R using data from a representative sample of U.S. adults.
Standardization of raw scores into norm-based T-scores (mean = 50, standard deviation = 10) was based on data from a 2009 U.S. internet-based general population survey. The internal consistency reliability of multi-item subscales and global sleep problems indices for both one-week and four-week recall forms of the MOS Sleep-R were examined using Cronbach's alphas and inter-item correlations. Construct validity was tested by comparing item-scale correlations between items within subscales with item-scale correlations across subscales. Scale-level convergent validity was tested using correlations with measures including generic health-related quality of life (i.e., SF-36v2) and other relevant outcomes (e.g., job performance, number of days in bed due to illness or injury, happiness/satisfaction with life, frequency of stress/pressure in daily life, the impact of stress/pressure on health, and overall health).
The one-week and four-week recall forms of the MOS Sleep-R were completed by 2045 and 2033 respondents, respectively. The psychometric properties of the one-week and four-week forms were similar. All multi-item subscales and global index scores showed adequate internal consistency reliability (all Cronbach's alpha > 0.75). Patterns of inter-item and item-scale correlations support the scaling assumptions of the MOS Sleep-R. Patterns of correlations between MOS Sleep-R scores with criterion measures of health-related quality of life and other outcomes indicated adequate construct validity.
The MOS Sleep-R introduces a number of revisions to the original survey, including simplified response sets, the introduction of a one-week recall form, and norm-based scoring that enhances interpretability of scores. Both the one-week and four-week recall period forms of the MOS Sleep-R demonstrated good internal consistency reliability and construct validity in a U.S. general population sample.
12项医学结局研究睡眠量表（MOS睡眠量表）已在数百项研究中用于获取患者报告的睡眠问题。开发了MOS睡眠量表的修订版（MOS睡眠-R），其使用简化的反应集，提供基于常模的可解释评分，并有两个回忆版本（一周或四周）。本研究的目的是使用来自美国成年人代表性样本的数据评估MOS睡眠-R的心理测量特性（信度和结构效度）。
基于2009年美国基于互联网的一般人群调查数据，将原始分数标准化为基于常模的T分数（均值 = 50，标准差 = 10）。使用克朗巴赫α系数和项目间相关性，检验了MOS睡眠-R一周和四周回忆形式的多项目子量表和总体睡眠问题指数的内部一致性信度。通过比较子量表内项目之间的项目-量表相关性与跨子量表的项目-量表相关性来测试结构效度。使用与包括一般健康相关生活质量（即SF-36v2）和其他相关结局（如工作绩效、因病或受伤卧床天数、生活幸福/满意度、日常生活中压力/压力频率、压力/压力对健康的影响以及总体健康）的测量指标的相关性来测试量表水平的收敛效度。
分别有2045名和2033名受访者完成了MOS睡眠-R的一周和四周回忆形式。一周和四周形式的心理测量特性相似。所有多项目子量表和总体指数得分均显示出足够的内部一致性信度（所有克朗巴赫α系数>0.75）。项目间和项目-量表相关性模式支持MOS睡眠-R的量表假设。MOS睡眠-R得分与健康相关生活质量和其他结局的标准测量指标之间的相关性模式表明具有足够的结构效度。
MOS睡眠-R对原始调查进行了多项修订，包括简化反应集、引入一周回忆形式以及基于常模的评分，提高了分数的可解释性。MOS睡眠-R的一周和四周回忆期形式在美国一般人群样本中均表现出良好的内部一致性信度和结构效度。
