[Reliability and validity of depression scales of Chinese version: a systematic review].

Objective: Through systematically reviewing the reliability and validity of depression scales of Chinese version in adults in China to evaluate the psychometric properties of depression scales for different groups. Methods: Eligible studies published before 6 May 2016 were retrieved from the following database: CNKI, Wanfang, PubMed and Embase. The HSROC model of the diagnostic test accuracy (DTA) for Meta-analysis was used to calculate the pooled sensitivity and specificity of the PHQ-9. Results: A total of 44 papers evaluating the performance of depression scales were included. Results showed that the reliability and validity of the common depression scales were eligible, including the Beck depression inventory (BDI), the Hamilton depression scale (HAMD), the center epidemiological studies depression scale (CES-D), the patient health questionnaire (PHQ) and the Geriatric depression scale (GDS). The Cronbach' s coefficient of most tools were larger than 0.8, while the test-retest reliability and split-half reliability were larger than 0.7, indicating good internal consistency and stability. The criterion validity, convergent validity, discrimination validity and screening validity were acceptable though different cut-off points were recommended by different studies. The pooled sensitivity of the 11 studies evaluating PHQ-9 was 0.88 (95%CI: 0.85-0.91) while the pooled specificity was 0.89 (95%CI: 0.82-0.94), which demonstrated the applicability of PHQ-9 in screening depression. Conclusion: The reliability and validity of different depression scales of Chinese version are acceptable. The characteristics of different tools and study population should be taken into consideration when choosing a specific scale.目的： 通过对以中国成年人为研究对象的中文版抑郁量表的信效度研究进行系统综述，评价适用于不同人群的抑郁量表的信效度。 方法： 基于万方、中国知网、PubMed和Embase四个数据库，检索2016年5月6日前发表的对中文版抑郁量表进行信效度评估的文献，评价量表的整体信效度，并采用诊断试验准确性Meta分析的HSROC模型对病人健康问卷抑郁量表（PHQ-9）的灵敏度和特异度进行分析。 结果： 共纳入文献44篇，结果表明常见中文版抑郁量表如贝克抑郁量表、汉密顿抑郁量表、流调中心抑郁量表、病人健康问卷、老年抑郁量表等信效度均较高。多数量表的Cronbach系数>0.8，重测信度和分半信度>0.7，内部一致性和稳定性较好。不同研究者对量表结构的评价不一，但量表的效标效度、聚合效度、区分效度及筛查效度良好。对PHQ-9的筛查指标进行诊断试验准确性Meta分析的结果显示：合并后灵敏度为0.88（95%CI：0.85~ 0.91），特异度为0.89 (95%CI：0.82~ 0.94），表明PHQ-9可作为筛查工具进行抑郁的初筛。 结论： 适用于不同人群的中文版抑郁量表的信效度均较高，可作为筛查工具进行研究对象抑郁状态的评估。同时建议在选择量表时应考虑不同量表的适用人群特点。.