Evaluating risk of bias using prevalence-question systematic review as a model

Article type
Authors
Rajic A, Wilhelm B, Young I, Waddell L, Parker S, McEwen S
Abstract
Background: Prevalence-question systematic reviews (SRs) are utilized by program managers/policy makers to evaluate the extent of a ‘problem’ and to undertake potential actions. Objectives: To evaluate risk of bias using one specific prevalence-question systematic review as a model. Methods: Prevalence-question was chosen based on its relevance to stakeholders engaged in ‘humans, animals and environment’ interface. Demand for organic food has increased rapidly over the past two decades. Contradictory opinions on the microbial safety of organic food exist. A SR was utilized to address the question ‘‘What is the prevalence of zoonotic bacteria in organic food chain of animal origin?’’ Electronic searches were conducted in 9 databases; 11 contacted to identify ‘in-press’ studies; reference lists from 10 articles were handsearched. Two reviewers screened all retrieved citations, evaluated methodological soundness of relevant studies using five ‘Risk of bias’ criteria, and extracted data. Results: From 162 relevant studies only 31 were suitable for quality assessment. No difference was observed in prevalence of zoonotic bacteria and antimicrobial resistance between two chain types. None of the studies met all five criteria. In >half of the studies, clustering and/or repeated sampling was not adjusted for. Conclusions: Current evidence on microbial food safety of organic food chain of animal origin is limited to a small number of studies and does not support an association between prevalence of microbial food safety hazards and chain (organic vs. conventional). Future primary research studies in this field should utilize random or systematic sampling, and appropriate statistical analyses.