文献速递关节外科

医学期刊中的髋膝关节置换骨科文献是否存在阴性偏倚

Hip and Knee Arthroplasty Orthopedic Literature in Medical Journals-Is It Negatively Biased?
PubMed 文献 · J Arthroplasty · 2017-09-19

该研究通过PubMed检索1976年1月至2016年12月《新英格兰医学杂志》《柳叶刀》和《美国医学会杂志》上的文献,由2名评价者将研究分为阳性、中性和阴性偏倚三类。结果共46篇阳性(30.3%)、46篇阴性(30.3%)、60篇中性(39.5%),三组比例无统计学显著差异(P=.905);报告下肢关节置换并发症的文献比例较高(55.9%),但无统计学显著性(z-score=1.38;95%CI 0.48-0.64;P=.17);两个时间段间无差异(P=.548)。结论认为三大医学期刊无下肢关节置换报告发表偏倚的证据,但需更多工作验证。

要点

  • 三大期刊中阳性、阴性、中性报告比例分别为30.3%、30.3%、39.5%,无显著差异(P=.905)。
  • 报告关节置换并发症的文献占55.9%,但无统计学显著性(95%CI 0.48-0.64;P=.17)。
  • 1975-1990年与2000-2016年两个时间段间期刊报告无差异(P=.548)。

核心结论

  • 三大医学期刊(JAMA、NEJM、Lancet)无下肢关节置换报告发表偏倚的证据。
  • 报告下肢关节置换并发症的研究更多,可能源于期刊编辑的系统性偏倚或非并发症报告投稿较少。
  • 未发现时间段是主要医学期刊骨科文献偏倚报告的因素。
  • 需更多工作验证本研究结果。

研究设计与样本

  • 对NEJM、Lancet和JAMA三大医学期刊1976年1月至2016年12月的文献进行综述,使用PubMed电子数据库检索。
  • 由2名评价者将研究分为阳性、中性和阴性偏倚三类。
  • 将基于深静脉血栓、感染、金属相关并发症、脂肪栓塞、再入院或死亡等并发症的研究归类为下肢关节置换相关并发症报告。
  • 比较1975-1990年与2000-2016年两个时间段的期刊偏倚。
  • 采用描述性分析评估频率,分类变量用卡方分析,二分类数据用z检验。

主要发现

  • 三大期刊合计阳性报告46篇(30.3%)、阴性报告46篇(30.3%)、中性报告60篇(39.5%)。
  • 三组分类比例间无统计学显著差异(P=.905)。
  • 报告关节置换并发症的医学文献比例较高(55.9%),但无统计学显著性(z-score=1.38;95%CI 0.48-0.64;P=.17)。
  • 1975-1990年与2000-2016年两个时间段间总体期刊报告无差异(P=.548)。

临床怎么用

  • 原文未据此给出推荐。

局限性

  • 需更多工作验证本研究结果。

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: Healthcare policy is often determined by well-designed studies most often published in high-impact medical journals. However, concern about the presence of publication bias against lower-extremity arthroplasty-related studies has called into question some of the validity of certain reports. There are only a few studies investigating the presence of the bias in high-impact medical journals against lower-extremity arthroplasty intervention, particularly in the Journal of American Medical Association (JAMA), New England Journal of Medicine (NEJM), and the Lancet. Thus, the purpose of this study was to assess (1) the distribution of positive, neutral, and negative results; (2) the number of reports focused on lower-extremity arthroplasty complications among these 3 journals; and (3) difference in bias between 2 time periods (1975 to 1990 and 2000 to 2016). METHODS: A review of the literature from 3 major medical journals (NEJM, Lancet, and JAMA) was performed using PubMed electronic databases, which retrieved articles between January 1976 and December 2016. Studies were categorized as being positive, neutral, and negatively biased by 2 reviewers. Studies were categorized as reporting on lower-extremity arthroplasty-related complications if they were based on complications including deep vein thrombosis, infection, metal-related complication, fat embolism, readmission, or mortality. In addition, we have compared the journal bias between 2 different time points (1975 to 1990 and 2000 to 2016). Descriptive analyses were performed to assess frequencies. Chi-squared analysis was conducted for categorical variables, whereas a z-test was performed for dichotomous data. RESULTS: When assessing all 3 journals, there were 46 positive (30.3%), 46 negative (30.3%), and 60 neutral reports (39.5%). There was no statistically significant difference in classification proportions between the 3 groups (P = .905). There was a higher percentage of medical literature reporting on the complications of arthroplasty (55.9%); however, this was not statistically significant (z-score = 1.38; 95% confidence interval, 0.48-0.64; P = .17). There was no difference in overall journal reporting between 1975 to 1990 and 2000 to 2016 (P = .548). CONCLUSION: There was no evidence of publication bias of lower-extremity arthroplasty reports in the major medical journals (JAMA, NEJM, and Lancet). However, there were more published studies reporting on complications of lower-extremity arthroplasty. This may be due to systematic bias among journal editors in these journals, or due to low journal submission reporting noncomplications after arthroplasty intervention. We did not find the time period to be a factor in bias reporting of orthopedic literature in major medical journals. More work is needed to verify the results of this study.

来源:J Arthroplasty,查看原文。版权归原作者所有。

引用

  1. Delanois RE, Gwam CU, Piuzzi NS, et al. Hip and Knee Arthroplasty Orthopedic Literature in Medical Journals-Is It Negatively Biased? J Arthroplasty. 2017. PMID: 28993088. doi:10.1016/j.arth.2017.09.017

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。