Fig. 2From: Investigating treatment-effect modification by a continuous covariate in IPD meta-analysis: an approach using fractional polynomialsStudywise TEFs (FP1, flex3) (top with restricted TEF scale). Bottom part: the same TEFs with full scale (a) and corresponding variances, weights (random effects and fixed effects) and metaTEFs. For comparison the ‘pooled’ estimate is also givenBack to article page