探究纳入文献质量评价不一致性的原因——以针灸治疗膝骨关节炎的系统评价/Meta分析为例＊

目的 评估国内针灸治疗膝骨关节炎的系统评价/Meta分析中纳入文献质量评价的不一致性,并探究产生不一致性的可能原因。方法 检索国内关于针灸治疗膝骨关节炎系统评价/Meta分析,提取其中纳入的随机对照试验（randomized controlled trials,RCTs）基本特征及质量评价记录,筛选出重复的质量评价记录,观察不同研究者对相同RCT所作判断的不一致性。结果 205条重复的质量评价记录中,除随机和盲法外均存在较大不一致性,结果数据的完整性、选择性报道、其他偏倚、Jadad评分占比分别为58%、47.5%、79%和51%。另外,在纳入的研究中提及“由2名成员独立评价”完成的质量评价在大多数领域的不一致性反而更高。结论 国内针灸治疗膝骨关节炎的系统评价/Meta分析中纳入文献质量评价结果在多个领域存在较大不一致性,其原因可能与报告不规范、未实施至少2名成员独立评价,以及研究者对质量评价工具掌握程度参差不齐有关,需要进一步改进质量评价工具的应用。

Exploration on Reasons of Inconsistencies in Quality Evaluation on Included Literatures Based on Systematic Reviews/Meta-analysis on Acupuncture Treatment for Knee Osteoarthritis

Objective To evaluate inconsistencies in quality evaluation on literatures included in systematic reviews/meta-analysis (SRs/MAs) on acupuncture treatment for knee osteoarthritis in China, and to explore possible reasons of the inconsistencies.Methods SRs/MAs on acupuncture treatment for knee osteoarthritis in China were retrieved, among which basic characteristics and quality evaluation records of included randomized controlled trials (RCTs) were extracted, and repeated quality evaluation records were screened out to observe inconsistencies of different researchers" judgments on the same RCT.Results In the 205 repeated quality evaluation records, there were significant inconsistencies in other aspects except randomization and blind methods, mainly reflected in completeness of outcome data, selective reporting, other bias, Jadad scores, etc., whose inconsistencies accounted for 58%, 47.5%, 79% and 51%, respectively. In addition, the inconsistency of quality evaluation referring to "completed by independent evaluation and cross-checking by two members" in the included studies is more inconsistent in most domains.Conclusion There are significant inconsistencies in most areas in quality evaluation results of included literatures in SRs/MAs on acupuncture treatment for knee osteoarthritis in China, which may be due to irregularities in reporting, non-implementation of at least 2 members" independent evaluations and cross-checks, and uneven mastery of quality evaluation tools by researchers. Further work is needed to improve application of the quality evaluation tools.