Table 1. Spearman correlation between time to first recurrence and time from first to second recurrence in the simulated data and the Thailand-Myanmar data excluding censored data.

All 1,000,000 simulated individuals who had at least two recurrences during the 1-year-simulation were used to compute the Spearman correlation. Note that the very weak negative correlation for models 1 and 2 is most likely due the censoring in the simulated data after 1 year of follow-up (see Table L in S1 Tables for details). All p-values are below 0.0001 due to the large number of simulations. For the Thailand-Myanmar data, we show here the Spearman correlation for artesunate or chloroquine treated individuals, excluding censored individuals (for the correlations of all individuals including censored data see Table K in S1 Tables).

	Artesunate	Chloroquine
Model 1: constant relapse rate	-0.06	-0.09
Model 2: temporal heterogeneity	-0.01	-0.03
Model 3: population heterogeneity	0.54	0.41
Thailand-Myanmar data	0.59	0.42