OBJECTIVES

: To assess the overall reporting quality of recent Meta-Analyses (MAs) in psychology through the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-analyses) statement.

METHODS

: We searched for MAs indexed in the PsycINFO database in 2016 through a pre-specified search strategy (N=2159). In accordance with our study objective and statistical power analysis, we randomly selected 207 MAs, 100 with the explicit statement, in their methods section, of the use of the PRISMA statement (PRISMA group) and 107 without (No PRISMA group). Two authors have extracted, independently, data on reporting quality using the 27 items of PRISMA.

RESULTS

: The overall median adherence to PRISMA is significantly different in the two groups (PRISMA=85%, No PRISMA=70%, p<0.0001). In fact, 13 out of the 27 items of the PRISMA statement are significantly more encountered in the PRISMA group than in the no PRISMA group (i.e. structured summary, protocol and registration, information sources, search strategy, study characteristics, results of individual studies, funding and, in both methods and results sections, study selection, risk of bias in individual study and bias across studies (all p-values are <0.05)). In the PRISMA group, for example, the probability to have information about protocol and registration (OR: 10.95, IC95%:3.18-37.67), the risk of bias in individual studies in either the methods (OR: 3.88; IC95%:2.18-6.92) or the results (OR: 5.75; IC95%:3.15-10.49) sections or the search strategy (OR: 4.18; IC95%:2.34-7.51) is significantly higher than in the no PRISMA group.

CONCLUSIONS

: The clear use of PRISMA has a positive influence on the reporting quality of MAs, even if in MAs using PRISMA, the reporting is not optimal. The use of PRISMA should be more widespread to facilitate the critical assessment of the validity of the published MAs in psychology.