Two sets of assumptions about ivermectin efficacy in ONCHOSIM.

Assumption set 1 was quantified such that ONCHOSIM could reproduce trends in skin mf levels as observed in a community trial that encompassed five consecutive annual ivermectin treatments [14], [19]. Assumption set 2 was quantified such that ONCHOSIM could reproduce trends in worm survival during three years of 3-monthly and 6-monthly mass treatment, as estimated from nodulectomy data [23], and trends in skin mf levels up to two years after a single dose of ivermectin as reported in a published meta-analysis [31]. Parameter values were fitted to the data with maximum likelihood, using the mean output of 100 repeated ONCHOSIM simulations as expected values (see S1 Text for details).
aExcess mortality has been reported for both female [23]–[30] and male worms [25], [26]. In the current study, excess mortality due to ivermectin was allowed to differ between male and female worms, reflecting the relative absence of male worms from subcutaneous nodules after repeated ivermectin treatment [22]–[30]. The macrofilaricidal effects of ivermectin were allowed to vary per treatment; however, this variation could not be estimated due to the aggregated nature of the Guatemalan data [23]. Instead, we arbitrarily assumed beta distributions with sample size 50 and mean 6% for males (2.5% and 97.5% percentiles 1.3%–14.0%) and 12% (3.9%–19.0%) for females, with the macrofilaricidal effects on male and female worms being perfectly correlated. Macrofilaricidal effects were assumed to be independent of earlier exposure to ivermectin and worm age, and hence reproductive capacity of the worm. In the sensitivity analysis, we set the average macrofilaricidal effects to either 4% and 8% (for males and females), or 9% and 18% (difference of factor 2/3 or 3/2) while keeping the sample size of the beta distribution at 50.
bThis treatment effect was assumed to vary per worm and treatment; 2.5% and 97.5% percentiles 2–24 months.
cThis assumption represents the notion that ivermectin causes temporary congestion of female worm uteri with dead mf, effectively preventing insemination and release of microfilariae [20], [21]. Time until recovery was assumed to vary per worm and treatment, and to follow an exponential distribution with mean 3.5 years (fitted to data [31]). This implies that 5% of adult female worms can be inseminated and release microfilariae within two months after exposure to ivermectin. Likewise, congestion resolves in 25%, 50%, 75%, and 95% of adult female worms within 1, 2.5, 5, and 10.5 years after exposure to ivermectin, respectively.
dTo account for variation in treatment efficacy between persons and treatments, for every simulated person and treatment, the average reduction was multiplied with a random value drawn from a Weibull distribution with mean 1 and shape 2 (see also S1 Text). In the sensitivity analysis, the average reduction was set to 23% or 52% (difference of factor 2/3 or 3/2).
Two sets of assumptions about ivermectin efficacy in ONCHOSIM.