对射频消融术与肝切除术治疗小肝癌系统评价/Meta分析的方法学质量再评价

目的评价射频消融术(RFA)与肝切除术(HR)治疗早期肝癌的系统评价(SR)/Meta分析的方法学质量及其对结局的影响。方法系统检索PubMed等6个电子数据库和5个卫生技术评估(HTA)官方网站中的HTA、SR或Meta分析,检索时限均从建库至2012年11月11日。用OQAQ量表评价纳入研究的方法学质量;用Cochrane协作网的推荐工具和改良的MINORS量表分别评价随机对照试验(RCT)和非随机对照试验(NRCT)的方法学质量;用Stata 10.0软件整合Meta分析的OR值及其95%CI。结果共纳入1个HTA、3个SR和15个Meta分析,其OQAQ平均得分3.3分[95%CI(2.6,4.1)];仅5个研究(26.3%)可认为只有小缺陷,方法学质量评为Good。7个研究错用统计学模型,3个研究修正模型后改变原结局方向;5个研究(5/19)作者将回顾性对照研究作为RCT纳入。纳入的39个原始研究中,3个RCT质量为B级,35个NRCT质量中等,MINORS(总分18分)平均得分15.0分[95%CI(14.6,15.4)],仅13个(37.1%)得分≥16分。17个(43.6%)原始研究不符合纳入标准,其中9个(23.1%)混杂其他有效处理措施(TACE、PEI等),4个(10.3%)纳入非原发性肝癌患者。结论目前针对RFA和HR治疗小肝癌的HTA、SR和Meta分析的总体质量较差,异质性较高,证据水平较低。临床医生在使用这些证据指导临床决策时尚需谨慎。

A Systematic Assessment of the Quality of Systematic Reviews/Meta-Analyses in Radiofrequency Ablation versus Hepatic Resection for Small Hepatocellular Carcinoma

Objective To evaluate the methodological quality and impacts on outcomes for systematic reviews （SRs） of radiofrequency ablation （RFA） versus hepatic resection （HR） for early hepatocellular carcinoma （HCC）. Methods We comprehensively searched six databases and five official websites for health technology assessment （HTA）, to collect HTAs, SRs, or meta-analyses from inception to Nov. 11th, 2012. The Overview Quality Assessment Questionnaire （OQAQ） was applied for quality assessment of included studies, the tools recommended by the Cochrane Collaboration was applied for quality assessment for randomized controlled trials （RCTs）, and the modified MINORS score was ap- plied to assess non-randomized controlled trials （NRCTs）. The odds ratios （ORs） and 95% confidence intervals （CIs） were integrated using Stata 10.0 software. Results One HTA, 3 SRs and 15 meta-analyses were included in total. The mean OQAQ score was 3.3 with 95%CI 2.6 to 4.1. Only five （26.3%） SRs were assessed as good quality. Seven studies misused statistical models, and 3 of them changed outcome direction after modification. Five studies （5/19） included retrospective controlled studies as RCTs. A total of 39 primary studies referenced by SRs were included, of which, 3 RCTs were levelled grade B, 35 NRCTs were of moderate quality, with an estimated mean MINORS score of 15.0 （totally, scored 18） with 95%CI 14.6 to 15.4, and only 13 studies （37.1%） scored more than 16. Seventeen primary studies （43.6%） did not meet in- clusion criteria of the S Rs, of which, 9 （23.1%） studies were mixed with other effective interventions in both groups （TACE, PEI, etc.）. Four studies included patients with non-primary HCC. Conclusion Currently, the overall quality of HTAs, SRs and meta-analyses about comparing the effects between RFA and HR for early HCC is fairly poor （high heterogeneity exists, and the evidence level is low. Physicians should apply the evidence with caution in clinical practice.