Fig. 6From: Interval-cohort designs and bias in the estimation of per-protocol effects: a simulation studyIP weighted estimates of per-protocol survival under strong confounding and decreasing values of mBack to article page