Frequent inappropriate use of unweighted summary statistics in systematic reviews of pathogen genotypes or genogroups

Abstract

ObjectivesOur study aimed to systematically assess and report the methodological quality used in epidemiological systematic reviews (SRs) and meta-analysis (MA) of pathogen genotypes/genogroups. Study Design and SettingNine electronic databases and manual search of reference lists were used to identify relevant studies. The method types were divided into three groups: 1) with weighted pooling analysis (which we call MA), (2) unweighted analysis of the study-level measures (which we call summary statistics), and (3) without any data pooling (which we call SR only). Characteristics were evaluated using Assessment of Multiple Systematic Reviews (AMSTAR), Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA), and Risk Of Bias In Systematic reviews (ROBIS) tools. The protocol was registered in PROSPERO with CRD42017078146. ResultsAmong 36 included articles, 5 (14%) studies conducted SR only, 16 (44%) performed MA, and 15 (42%) used summary statistics. The univariable and multivariable linear regression of AMSTAR and PRISMA scores showed that MA had higher quality compared with those with summary statistics. The SR only and summary statistics groups had approximately equal scores among three scales of AMSTAR, PRISMA, and ROBIS. The methodological quality of epidemiological studies has improved from 1999 to 2017. ConclusionDespite the frequent use of unweighted summary statistics, MA remains the most suitable method for reaching rational conclusions in epidemiological studies of pathogen genotypes/genogroups.