WHOQOL-BREF之心理計量特徵的回顧研究

A review on the psychometric properties of the WHOQOL-BREF

世界衛生組織生活品質問卷簡明版（WHOQOL-BREF）是一個廣泛使用於醫藥、公衛界的跨文化測量工具，目前已發展50餘國版本，然缺乏在不同國家或地區檢驗其心理計量特徵，本研究採用系統性回顧法，以評估其在不同樣本族群的信、效度。本研究於PsycINFO與MEDLINE資料庫，搜尋1998至2019年6月檢驗WHOQOL-BREF心理計量特徵的英文研究，共50篇文獻。研究發現WHOQOL-BREF之生理、心理、社會關係、與環境範疇具有高內部一致性信度與再測信度。此4範疇可有效區辨健康組與疾病組，並與健康相關問卷有顯著正相關，與疾病相關問卷有顯著負相關，具良好的區辨與效標關聯效度。然其建構效度不佳，部份研究無法驗證世界衛生組織所提出的4因素模型。綜觀此50篇研究發現，採取較不嚴謹的分析方法可能是造成WHOQOL-BREF建構效度不佳的原因。本研究依據心理計量理論及其模擬研究結果，建議未來在檢驗WHOQOL-BREF之建構效度時，於EFA採用迭代主軸法、多重方法決定因素數、與promax轉軸。於CFA採用較嚴謹的模型適配指標切結分數，並謹慎採用模型修正指數。

The short version of the World Health Organization Quality of Life (WHOQOL-BREF) questionnaire, a cross cultural instrument, is widely used in medical and public health fields, with versions developed in over 40 countries. Because research to evaluate the psychometric properties of the WHOQOL-BREF in diverse sample groups is scant, this study systematically reviewed the reliability and validity of such research. We searched studies, written in English, of psychometric properties examined in the WHOQOL-BREF in MEDLINE and PsycINFO (from 1998 to the May 2019). Fifty papers were finally included. The findings showed that the physical, psychological, and social relationships and the environmental domains of the WHOQOL-BREF had appropriate Cronbach's α and test–retest reliabilities. All these four domains showed good discriminant and criterion-related validity, representing a significant differentiation between the healthy and unhealthy groups, and exhibited significantly positive and negative correlations on questionnaires measuring health and syndrome of illness symptoms. However, the WHOQOL-BREF performed poorly in construct validity. The four-factor model proposed by the WHO was not fully validated in studies reviewed. An overview of these 50 papers indicates that inappropriate statistical analysis approaches may be a reason for poor construct validity. Based on the psychometric theory and its results from simulation studies, this study recommends conducting exploratory factor analysis using iterative principle axis factoring, promax rotation, and multiple methods to determine the number of factors as well as using confirmatory factor analysis with strict cut-off scores in goodness-of-fit indices and cautiously adopting a modification index for evaluating the psychometric properties of the WHOQOL-BREF.