OBJECTIVE: To systematically review and appraise misinterpretation of pharmacovigilance disproportionality analysis results in published studies. STUDY DESIGN AND SETTING: We randomly selected 100 studies that performeddisproportionality analysesand indexed in Medline identified during a systematic literature search. Titles, abstracts and main texts (results, discussion and conclusion) were evaluated for spin independently by 2 reviewers. Spin in pharmacovigilance studies was classified according to 3 main categories: inappropriate interpretation, inappropriate extrapolations and misleading reporting. RESULTS: Of the 100 studies evaluated, we found that 63%, 56% and 51% had at least one type of spin in their abstract, main text or conclusion respectively, and 40%used causal language to interpret their results in the abstract or conclusion.Spin in titles and results were exclusively represented by inappropriate interpretations of findings (12% and 21% respectively), with terms such as "risk of" or "risks associated with" or results erroneously presented as regular Odds Ratios. Spin in discussion sections mostly concernedinappropriate interpretations (38%)and misleading reporting (12%). Misleading reporting, notably failing to acknowledge the limitations of disproportionality analyses, was the most frequent type of spin in abstracts (55%) and conclusion sections (37%). CONCLUSION: We found that spin is frequent in publications of pharmacovigilance disproportionality analyses, notably in abstracts.This consisted notably in an over-interpretation of the results suggesting a proven causative link between a drug use and the risk of an event.< Réduire