Validity of data extraction in evidence synthesis practice of adverse events: reproducibility study

Ref ID 837
First Author C. Xu
Journal BMJ
Year Of Publishing 2022
URL https://www.bmj.com/content/377/bmj-2021-069155.full
Keywords • Harms
• Reproducibility
• Error
• Non-Cochrane reviews
Problem(s) • Data extraction errors and double counting
Article Type Empirical
Article Subtype Meta-epidemiological analysis
First Author Country China
Aim To assess the the validity of data extraction in systematic reviews of adverse events published between 1 January 2015 and 1 January 2020 and to assess the effect of data extraction errors on the results, and to develop a classification framework for data extraction errors.
Level of Investigation Descriptive
Summary of Findings From 201 included systematic reviews of RCTs indexed on Pubmed between 1 January 2015 and 1 January 2020. The 201 comprising reviews included 829 pairwise meta-analyses, and data extraction could not be reproduced in 1762 (17.0%) of 10 386 trials. In 554 (66.8%) of 829 meta-analyses, at least one randomised controlled trial had data extraction errors; 171 (85.1%) of 201 systematic reviews had at least one meta-analysis with data extraction errors. Meta-analyses that had two or more different types of errors were more susceptible to these changes than those with only one type of error (for moderate changes, 11 (28.2%) of 39 v 26 (10.4%) 249, P=0.002; for large changes, 5 (12.8%) of 39 v 8 (3.2%) of 249, P=0.01).
Number of systematic reviews included 201
Number of eligible systematic reviews assessed 18636
Treatment impacted Yes
Treatment impacted description
Interpretation impacted Yes
Interpretation impacted description Data extraction errors led to 10 (3.5%) of 288 meta-analyses changing the direction of the effect and 19 (6.6%) of 288 meta-analyses changing the significance of the P value.