Table 1.

Methods for calculating onchocerciasis-associated epilepsy (OAE) cases in the African Programme for Onchocerciasis Control (APOC) countries in 1995 (pre-control) and in 2015

Figure 1a shows the functional relationship describing the community-level association between the prevalence of Onchocerca volvulus skin microfilariae and all-cause epilepsy (case definition as in the International League Against Epilepsy guidelines [31]), as published by Pion et al. [4]. The predicted prevalence of epilepsy in areas with zero O. volvulus microfilariae prevalence was removed from the analysis. The prevalence of OAE in onchocerciasis-endemic areas was calculated by subtracting the predicted prevalence of all-cause epilepsy for APOC-areas using the functional relationship from an averaged all-cause background epilepsy prevalence for Sub-Saharan Africa (0.36%, 95% CI: 0.26–0.47% [11]). We linked the functional relationship to a published map of nodule prevalence in adult males in Africa (Fig. 1b) after converting this map to skin microfilariae prevalence in the general population (age 5 and above) at the pixel level (1 × 1 km raster) using a published statistical model (Fig. 1c) [69]. We assumed that the association between all-cause epilepsy and microfilariae prevalence was entirely driven by geographical variation in onchocerciasis prevalence, which we assume to be uncorrelated with other important causes of epilepsy in developing countries, like neurocysticercosis Next, the pre-control number of OAE cases was estimated by multiplying the average OAE prevalence in an area (averaged over pixels) with the size of the population at risk (based on APOC census data), assuming that the population density is homogeneous throughout the area. We stratified the pixels by pre-control nodule prevalence in adult males (>  0%–<  20%, ≥ 20%–<  40%, ≥ 40%) and the population at risk proportional to the number of pixels in each endemicity category. To extrapolate the number of OAE cases to 2015, we assumed that the population at risk and hence the potential number of OAE cases (counterfactual assuming no control) increased annually due to population growth. Population growth between 1995 and 2015 was assumed to be 2.74% based on UN population prospects for SSA [70]. For years that areas remained untreated, we assumed that prevalence of epilepsy remained proportionally stable (i.e. as estimated for 1995). Next, we corrected the number of cases for the presence of MDA, assuming that treatment has no effect on prevalent cases of OAE but prevents incidence of new cases after a scaling-up period of 3 years (i.e. accounting for low treatment coverage in the first few years of MDA programmes). Ivermectin was assumed to reduce OAE incidence to zero (after on three years of non-optimal MDA) on the basis of studies that suggest a reduction in the incidence of epilepsy after ivermectin treatment [26, 27, 71]. We further assumed that once incidence of OAE is zero, the number of prevalent OAE cases declines by 3.5% annually due to mortality, based on a reported 70% cumulative 10-year survival probability among epilepsy cases [16] ($1-\sqrt[10]{0.7}=0.035$). All baseline tables and calculations are shown in Additional file 2. Furthermore, multivariate sensitivity analyses were performed around our assumption of survival probability and number of years of suboptimal ivermectin before OAE incidence drops to zero (Additional file 2).