From the Faculty of Medicine and the Division of Plastic and Reconstructive Surgery, Dalhousie University; the Division of Plastic and Reconstructive Surgery, University of Toronto; and the Division of Plastic and Reconstructive Surgery, Department of Surgery, King Abdulaziz University.
Plast Reconstr Surg. 2019 Aug;144(2):519-530. doi: 10.1097/PRS.0000000000005880.
Meta-analyses are common in the plastic surgery literature, but studies concerning their quality are lacking. The authors assessed the overall quality of meta-analyses in plastic surgery, and attempted to identify variables associated with scientific quality.
A systematic review of meta-analyses published in seven plastic surgery journals between 2007 and 2017 was undertaken. Publication descriptors and methodologic details were extracted. Articles were assessed using the following two instruments: A Measurement Tool to Assess Systematic Reviews (AMSTAR) and AMSTAR 2.
Seventy-four studies were included. The number of meta-analyses per year increased. Most meta-analyses assessed a single intervention (59.5 percent), and pooled a mean of 20.9 studies (range, two to 134), including a mean of 2463 patients (range, 44 to 14,884). Most meta-analyses were published in Plastic and Reconstructive Surgery (44.6 percent) and included midlevel evidence (II to IV) primary studies. Only 16.2 percent of meta-analyses included randomized controlled trials. Meta-analyses generally reported positive (81.1 percent) and significant results (77.0 percent). Median AMSTAR score was 7 of 11 (interquartile range, 5 to 8). Higher AMSTAR scores correlated with more recent meta-analyses that provided a rationale for statistical pooling, and appropriately managed methodologic heterogeneity (r = 0.66; p < 0.01).
Despite an increase in number and quality, meta-analyses are at high risk of bias because of the low level of evidence of included primary studies and heterogeneity within and between primary studies. Plastic surgeons should be aware of the pitfalls of conducting and interpreting meta-analyses.
荟萃分析在整形外科学文献中很常见，但缺乏对其质量的研究。作者评估了整形外科学中荟萃分析的总体质量，并试图确定与科学质量相关的变量。
对 2007 年至 2017 年期间在 7 种整形外科学期刊上发表的荟萃分析进行了系统回顾。提取了出版描述符和方法学细节。使用以下两种工具评估文章：评估系统评价的测量工具（AMSTAR）和 AMSTAR 2。
共纳入 74 项研究。每年发表的荟萃分析数量增加。大多数荟萃分析评估了单一干预措施（59.5%），汇总了平均 20.9 项研究（范围 2 至 134 项），包括平均 2463 例患者（范围 44 至 14884 例）。大多数荟萃分析发表在《整形与重建外科学》（44.6%）中，包括中级证据（II 至 IV 级）的原始研究。只有 16.2%的荟萃分析纳入了随机对照试验。荟萃分析通常报告阳性（81.1%）和显著结果（77.0%）。AMSTAR 中位数评分为 11 分中的 7 分（四分位距为 5 至 8 分）。较高的 AMSTAR 评分与最近发表的荟萃分析相关，这些分析提供了统计汇总的理由，并适当处理了方法学异质性（r = 0.66；p < 0.01）。
尽管数量和质量有所增加，但荟萃分析仍存在很高的偏倚风险，原因是纳入的原始研究证据水平较低，以及原始研究内部和之间存在异质性。整形外科医生应该意识到进行和解释荟萃分析的陷阱。
