文献速递美容整形

整形外科随机对照试验注册与发表结局不一致的系统综述

Discrepancies between Registered and Published Primary and Secondary Outcomes in Randomized Controlled Trials within the Plastic Surgery Literature: A Systematic Review.
PubMed 文献 · Plast Reconstr Surg · 2020-01-01

BACKGROUND: Recent studies have identified a high incidence of discrepancy between registered and published outcomes in registered medical and surgical randomized controlled trials. This has not yet been studied in the plastic surgery literature.…

要点

  • 整形外科RCT注册率仅39%,发表结局与注册方案不一致率高达88%。
  • 次要结局不一致更严重(80%),常见为发表未注册的新次要结局(65%)。
  • 44%的结局不一致使发表结果更倾向于统计学显著阳性。

核心结论

  • 整形外科文献中注册与发表试验结局不一致率很高。
  • 次要结局的结局报告不一致问题更为突出,此前研究较少。
  • 高比例的不一致倾向于统计学显著和更患者相关的结局,可能反映高影响期刊发表压力。

研究设计与样本

  • 系统评估2012至2016年七本高影响整形外科期刊发表的随机对照试验。
  • 从注册网站和发表文章使用标准化数据提取表收集数据。

主要发现

  • 43项(88%)试验存在注册与发表结局不一致。
  • 主要结局不一致26项(53%),次要结局不一致39项(80%)。
  • 单个试验主要结局不一致数量1至7个,次要结局1至12个。
  • 美容外科结局不一致试验数最多(n=15)。
  • 未报告的注册主要结局占13%,次要结局占38%。

关键数值与推荐

  • 注册的非显著主要结局在30%试验中被发表为非显著次要结局。
  • 发表新的未注册次要结局(65%)和改变发表主要结局的评估时间(61%)是最常见的不一致类型。
  • 43项结局不一致试验中19项(44%)不一致有利于统计学显著阳性结局。
  • 主要结局不一致中8项(16%)、次要结局不一致中14项(29%)使发表结局患者相关性改善。

临床怎么用

  • 阅读整形外科RCT时应核对注册方案与发表结局的一致性。
  • 关注次要结局是否存在未注册的新增或时间点改变。
  • 对声称阳性结果的试验需警惕结局报告偏倚。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Recent studies have identified a high incidence of discrepancy between registered and published outcomes in registered medical and surgical randomized controlled trials. This has not yet been studied in the plastic surgery literature. METHODS: The authors systematically assessed plastic surgery randomized controlled trials published between 2012 and 2016 in seven high-impact plastic surgery journals. Data were collected from the registration website and published articles using a standardized data extraction form. RESULTS: A total of 145 randomized controlled trials were identified, with a 39 percent trial registration rate (n = 57). Forty-nine trials were included in the final analysis. Forty-three (88 percent) had a discrepancy between registered and published outcomes: 26 (53 percent) for primary outcome(s), and 39 (80 percent) for secondary outcome(s). The number of discrepancies in an individual trial ranged from one to seven for primary outcomes and one to 12 for secondary outcomes. Aesthetic surgery had the largest number of trials with outcome discrepancies (n = 15). The prevalence of unreported registered outcomes was 13 percent for primary outcomes and 38 percent for secondary outcomes. Registered nonsignificant primary outcomes were published as nonsignificant secondary outcomes in 30 percent of trials. Publishing new nonregistered secondary outcomes (65 percent) and changing the assessment timing of published primary outcomes (61 percent) were the most common types of discrepancies. Discrepancies favored a statistically significant positive outcome in 19 (44 percent) of the 43 trials with an outcome discrepancy. Discrepancies that resulted in published outcomes with improved patient relevance were found in eight trials (16 percent) for primary outcome discrepancies and 14 trials (29 percent) for secondary outcome discrepancies. CONCLUSIONS: The plastic surgery literature has high rates of discrepancies between registered and published trial outcomes. Outcome reporting discrepancy is even more problematic for secondary outcomes, an area of analysis that has previously been poorly studied. The high rate of discrepancy change favoring a statistically significant outcome and more patient-relevant outcomes may indicate the pressure to demonstrate significant results to be accepted for publication in high-impact journals.

来源:Plast Reconstr Surg,查看原文。版权归原作者所有。

引用

  1. Hudson AS, Morzycki AD, Samargandi OA, et al. Discrepancies between Registered and Published Primary and Secondary Outcomes in Randomized Controlled Trials within the Plastic Surgery Literature: A Systematic Review. Plast Reconstr Surg. 2020. PMID: 31609284. doi:10.1097/PRS.0000000000006370

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。