Completeness of reporting of systematic reviews in the animal health literature: A meta-research study

Abstract

Systematic reviews are a valuable tool for evaluating the efficacy of interventions and for quantifying associations. To be properly assessed, reviews must be comprehensively reported. The primary objective of this study was to evaluate the completeness of reporting of systematic reviews and meta-analyses in animal health. The secondary objective was to further characterize methods for literature searches and risk of bias assessments and to document whether the risk of bias component represented an assessment of risk of bias, study quality, or levels of evidence based on the primary studies included. The dataset comprised 91 systematic reviews or meta-analyses of interventions or exposures with at least one health outcome measured at the animal or animal byproduct level, in any companion or food animal species and published between 2014 and 2018. Two reviewers independently collected information on whether each item in the PRISMA reporting guidelines was reported, with disagreements resolved by consensus. There was considerable variability in the completeness of reporting among reviews; some items, such as eligibility criteria for inclusion, were reported in most reviews (>65 %). Other items were not consistently reported; for instance, in 60 % (54) of the reviews there was no information provided on the sample size of individual studies, populations, interventions and comparators, outcomes, or follow up period. Although 89 % (81) of systematic reviews with meta-analysis included the effect size estimate and confidence intervals, it was not possible to determine which study designs were included for 30 % (14) of reviews. Results from individual PRISMA item questions were combined to determine whether all aspects of each recommended item were reported; 71 % of items were adequately reported in less than half the systematic reviews without a meta-analysis, 35 % of the items were adequately reported in less than half the systematic reviews with a meta-analysis, and 71 % of items were adequately reported in less than half of the meta-analyses without a systematic review component. An assessment of individual study level bias was included in 64 % of the reviews, although this component included an evaluation of risk of bias (35 reviews), study quality (25 reviews), or levels of evidence based on study design (12 reviews). Reporting guidelines or clinical guidelines were inappropriately used to assess risk of bias in 9 reviews. Overall, the results of this study reveal that reporting of systematic reviews in the animal health literature is suboptimal and improvements are needed to enhance utility of these reviews.