Comparison of statistical analysis using the random-effects and inverse variance heterogeneity models for a meta-analysis

Abstract

In our previous meta-analysis of the neutrophil-to-lymphocyte and platelet-to-lymphocyte ratios in Henoch-Schonlein purpura (W. Fu, et al., Int. Immunopharmacol 94 (2021) 107454.), the meta-analysis was performed using a random-effects model. In this study, the same data were re-analyzed using the Inverse Variance Heterogeneity (IVhet) model. Similar results were obtained from both models except for the different heterogeneities in the subgroup analysis of age. The analysis with the IVhet model also showed that the NLR played a more critical role than the PLR in HSP and its complications, and using multiple models would be helpful for better evaluating the clinical value of the study.