Quality of systematic reviews in HIV: The case of clinical outcomes associated with patient medication adherence.

Aim: Use of systematic reviews (SRs) as first-level evidence for guideline recommendations hinges on review quality. In particular, US guidelines for adherence-related recommendations in the treatment of human immunodeficiency virus (HIV) are not based on available SRs of adherence-outcome relationships; it is unclear why. No published studies report on the quality of SRs on HIV adherence and outcomes, which may be driving the lack of use. We describe the quality of this body of literature.

Methods: Literature searches were conducted in Ovid MEDLINE, EMBASE, CINAHL, PubMed Central, the Cochrane Library, Science Citation Index, Web of Science, ScIELO Citation Index, and Ovid Emcare. Screening and quality assessments were performed in duplicate using AMSTAR 2. Funding sources and impact factors of publishing journals were also extracted, and correlations between quality rankings and numbers of critical weaknesses versus impact factors were assessed using Spearman's rank correlation coefficient.

Results: Nine SRs of 1141 records met eligibility criteria. Overall confidence in the results was critically low for most (78%) SRs. Underperformance was found across all AMSTAR 2 domains. Impact factor (a surrogate or journal reputation) did not correlate with quality.

Conclusions: SRs do not necessarily comprise top-level evidence despite the availability of quality appraisal tools and reporting guidance, which could explain the lack of SR evidence in US HIV medication adherence-related guideline recommendations. All parties to evidence synthesis publication should require quality assessment of studies.