RESUMO
BACKGROUND: Reporting of systematic reviews (SRs) using PRISMA increases transparency and reproducibility; adherence in the dermatology literature has not been assessed. OBJECTIVES: To assess selected, primarily methodological items from the PRISMA reporting guideline among SRs published in dermatology journals. METHODS: We reviewed SRs published from 2013 to 2017 in the five highest-impact dermatology journals according to the Science Citation Index. We descriptively assessed reporting of selected PRISMA items, the proportion of PRISMA items fully and partially reported, and whether SRs described using a preregistered protocol. We used univariate and multivariate linear regression to evaluate associations between exposures (year, protocol registration, funding source, type of included study, disease and journal), and outcomes (proportion of PRISMA items fully reported, and fully and partially reported, for each SR). RESULTS: We identified 136 SRs. All had more than one inadequately reported PRISMA item. Protocol registration (73%) and risk of bias (38%) were most often unreported. Reporting improved over time in our primary multivariate analysis [fully reported vs. partially and not reported, ß = 2·48; 95% confidence interval (CI) 0·73-4·27] and secondary analysis (fully and partially reported vs. not reported, ß = 1·28, 95% CI 0·06-2·50). Only 15% (20 of 136) of SRs stated that their protocols were registered; this was associated with PRISMA adherence to the evaluated PRISMA items in our primary multivariate analysis (ß = 10·05, 95% CI 2·89-17·2) and secondary analysis (ß = 8·87, 95% CI 3·84-13·9). CONCLUSIONS: SR reporting in dermatology journals is often inadequate but improving over time; protocol registration is associated with better reporting. What's already known about this topic? No studies to date have examined the adherence of dermatology systematic reviews (SRs) to reporting guidelines, such as PRISMA. In other medical fields, reporting is variable with some improvement in adherence to reporting standards over time. What does this study add? Among SRs published in five dermatology journals from 2013 to 2017, all (n = 136) had at least one inadequately reported PRISMA item, while 93% (127 of 136) had at least one fully nonreported item. Reporting improved over time and SRs that stated use of a preregistered protocol were associated with better reporting. Several items remain commonly underreported in dermatology SRs. Authors, reviewers, journal editors and editorial committees should encourage preregistration of SR protocols and improved SR reporting.
