Flaws in the application and interpretation of statistical analyses in systematic reviews of therapeutic interventions were common: a cross-sectional analysis

Ref ID 406
First Author M. J. Page
Journal JOURNAL OF CLINICAL EPIDEMIOLOGY
Year Of Publishing 2018
URL https://www.jclinepi.com/article/S0895-4356(17)30735-7/fulltext
Keywords Statistical
Expertise
Inference
Subgroup
General medical
Problem(s) Lack of statistical expertise in handling of quantitative data
Unplanned or unjustified subgroup or sensitivity analyses
Incorrect interpretation or statistical inference error from meta-analysis
Number of systematic reviews included 110
Summary of Findings For the primary meta-analysis of each included systematic review, just over half (56%) used the random-effects model, but few (8%) interpreted the meta-analytic effect correctly or presented an accompanying prediction interval. In the 15% of overall reviews displaying funnel plot asymmetry only 24% included the recommended number of at least 10 studies. In the 38% of overall reviews which presented subgroup analyses, findings were not interpreted with respect to a test for interaction in 69% of cases and the issue of potential confounding in the subgroup analyses was not raised in any systematic review. For the 50% of overall reviews that did sensitivity analyses, no rationale was provided for any of the sensitivity analyses accompanying 73% of index meta-analyses.
Did the article find that the problem(s) led to qualitative changes in interpretation of the results? Not Applicable
Are the methods of the article described in enough detail to replicate the study? Yes