Skip to main content

Table 5 Strategies for modelling survival data obtained from many dissimilar sources

From: Geographical variation in Plasmodium vivax relapse

Statistical method Accounts for individual-level variation Accounts for between-study variation R packages
Fixed-effects meta-analysis No; operates on summary statistics No Many software packages, e.g., meta and metafor
Mixed-effects meta-analysis No; operates on summary statistics Yes Many packages, e.g., meta and metafor; also general-purpose software such as lme4 may be used
Survival analysis for pooled data Yes No A number of packages, e.g., survival, eha and flexsurv
Survival analysis with mixed effects Yes Yes Most notably R/coxme; flexible software seems to be hard to find
  1. Data were analysed using mixed-effects meta-analysis, which is common for this type of study. All of the methods have strengths and weaknesses.