Figure 1

Comparison of the effects of trial drop-outs on estimated failure rates in 100 patients followed for 63 days after a treatment with a true 20% failure rate. The "per protocol" method (PP), of estimating cure rates in which the denominator at each time point is the combined number of patients who were followed until that point and continued to be aparasitaemic and those who had a recrudescence observed, is shown as a red solid line and compared with the Kaplan-Meier survival method. The drop-outs all occurred after 28 days of follow-up. Three different scenarios are presented with differing proportions of failures (20 in total) presenting before and after 28 days: blue dash line – 2 failures before and 18 after day 28; orange dash line – 10 failures before and 10 failures after day 28; green dash line – 18 failures before and 2 failures after day 28.