The influence of 'grey' literature on meta-analysis.

Abstract

Introduction. The impact of the inclusion/exclusion of grey literature in meta-analysis (MA) is unclear. Objectives. To investigate, in a sample of published MA; the prevalence of grey literature, the quality of reporting at the MA and trial levels, and the impact of grey literature on the point estimate and precision of the results. Methods. Analysis of Variance and regression models were used to consider the quality of reporting and the impact of grey literature on estimates of efficacy. Results. Grey literature was included in 33% of the MA. Grey inclusive MA tended to be of higher quality than those that excluded it. At the trial level, grey literature was of lower quality than published literature. The exclusion of grey literature led to increases in both the reported effectiveness of the intervention and the precision of the results. Conclusions. MA that exclude grey literature run the risk of producing biased estimates of intervention effectiveness. Grey literature must be sought and included when it meets pre-defined inclusion criteria.