We screened intervention reviews from the Cochrane Database of Systematic Reviews (CDSR) and sought well-appraised meta-analyses. We used the individual RCT study estimates\' relative deviation from the pooled effect estimate as a proxy for the deviation of the study results from the truth. The effect of the JIF on the relative deviation was estimated with linear regression and with local polynomial regression, both with adjustment for the relative size of studies. Several sensitivity analyses for various sub-group analyses and for alternative impact metrics were conducted.
In 2459 results from 446 meta-analyses, results with a higher JIF were on average closer to \"truth\" than the results with a lower JIF. The relative deviation decreased on average by -0.023 per JIF (95% CI -0.32 to -0.21). A decrease was consistently found in all sensitivity analyses.
Our results indicate that study results published in higher-impact journals are on average closer to truth. However, the JIF is only one weak and impractical indicator among many that determine a studies\' accuracy.
方法:我们从Cochrane系统评价数据库(CDSR)筛选干预评价,并寻求评价良好的荟萃分析。我们使用单个RCT研究估计与合并效应估计的相对偏差作为研究结果与事实的偏差的替代。用线性回归和局部多项式回归估计JIF对相对偏差的影响,两者都对研究的相对规模进行了调整。对各种亚组分析和替代影响指标进行了几项敏感性分析。
结果:在来自446个荟萃分析的2459个结果中,具有较高JIF的结果平均比具有较低JIF的结果更接近“真理”。相对偏差平均每JIF下降-0.023(95%CI-0.32至-0.21)。在所有敏感性分析中始终发现下降。
结论:我们的研究结果表明,发表在影响力较高的期刊上的研究结果平均更接近真理。然而,在许多决定研究准确性的指标中,JIF只是一个薄弱和不切实际的指标。