Methodological quality was assessed by two independent reviewers using an adapted version of the STROBE statement. Criteria for validity assessment included controlling for confounders (specifically age, gender and comorbidity), selection criteria, appropriateness of study design, potential sources of bias and statistical methods. The authors included a second category of validity in which they assessed quality of data extraction in individual studies. Studies were scored according to four levels: sufficient, moderate, insufficient and inapplicable. Criteria for data extraction included description of inclusion and exclusion criteria, reporting of effect, control for confounding, description of medication exposure and generalisability.
Disagreements were resolved by consensus.