Recommendation Integrity

Affected recommendations are classified as unlikely, possibly, or likely distorted after reassessing the repaired evidence base. A recommendation is assumed likely distorted if removing problematic evidence leaves no supporting evidence. The evidence base will be repaired by removing problematic evidence from each recommendation. Meta-analyses that included retracted trials will have pooled effects recalculated after removing those trials. The study will judge whether each recommendation using problematic evidence is favorable toward an intervention or against it.