Fig. 4

Comparison of our model’s predictions for overestimation of benefit with real-world overestimates of effect. Figure shows our model predictions, displayed as observed relative risk reduction/true relative risk reduction, overlayed on top of the summary figure from Bassler et al. [9]. The Bassler estimates come from comparing the effect estimate in truncated trials with non-truncated trials answering the same question. As in our Figure 2, the purple line represents typical truncated trials, the red line represents typical truncated trials if instead carried to completion, and the orange line represents large truncated trials