We incorporated OAE into EPIONCHO-IBM using the dose-response relationship between O. volvulus mf load in childhood and the probability of developing epilepsy later in life reported in ref. 10 (Fig. 1) and used a 3–15-year age range for the onset of OAE10,11,17. The EPIONCHO-IBM-modelled relationship between ABR and epilepsy prevalence was strongly non-linear (Fig. 3) and consistent with the range of epilepsy prevalence reported for SSA18. The higher epilepsy prevalence for males compared to females in Fig. 3 results from the age- and sex-exposure functions in EPIONCHO-IBM (Supplementary Text S1 and Supplementary Fig. S4). For the 3–15-year age group (contributing to OAE onset in our model), boys are more highly exposed than girls, with the exposure of both being greater than 0 at birth. This enables the model to capture the high mf loads reported in ref. 10 for children aged 5–10 years (Supplementary Tables S1, S4) and to generate a modelled epilepsy prevalence in the 30–35-year age group (the one participating in the epilepsy survey conducted in ref. 10, 25 years after collection of parasitological data in 1991–93) of 8.6% in males and 6.7% in females (compared to the observed 9.5% in males and 6.8% in females) after accounting, in the model, for the 19 years of CDTI experienced in the study area. According to ref. 10, the difference in epilepsy prevalence between the sexes was not statistically significant, and in our model, we do not assume a differential susceptibility to developing OAE according to sex; therefore, our results are due to EPIONCHO-IBM exposure functions. These functions had been derived from fitting an age- and sex-structured precursor of our model to data on mf infection profiles from Cameroon, as described in Supplementary Text S1.The modelled epilepsy prevalence as a function of O. volvulus mf prevalence showed remarkable agreement between the results of our full-(stochastic) transmission model and those of the statistical model fitted in ref. 7 (Fig. 4). By calibrating the model with age-specific mf data for children aged 5–10 years at baseline (1991–93)10, and simulating the CDTI history of the study area19, EPIONCHO-IBM was able to reproduce pre-control parasitological and entomological conditions14,15 (Table 1), as well as the epilepsy prevalence and incidence values reported for the Mbam Valley in the retrospective cohort study10 (Supplementary Table S2) and cross-sectional study16 (Fig. 2), after 19 years of annual CDTI. An entomological study in the Mbam Valley (2016–2017), approximately at the same time as the studies in refs. 10,16, reported annual transmission potentials (ATPs) of 0–4488 L3/person/year (mean of 173820), very similar to our modelled baseline ATP (1700, Table 1), suggesting intense ongoing transmission in this highly holoendemic area despite prolonged CDTI. Also, in 2017, IgG4 seropositivity to the O. volvulus Ov16 antigen (an indicator of exposure) in 7–10-year-olds was 42–55%21. These results highlight the importance of improving programmatic performance and deploying alternative and/or complementary interventions in highly endemic areas if onchocerciasis transmission and OAE are to be eliminated or considerably reduced.The scenario-based simulation results indicated that 25 years of CDTI would lead to substantial reductions in OAE prevalence and incidence in hyperendemic settings under biannual treatment with 80% coverage of the total population. Under these programmatic conditions, the number of new cases declined to near 0 and did not increase in the following 75 years after treatment cessation (Supplementary Fig. S2). In holoendemic settings, annual or biannual treatment could reduce incidence by 60–80%, but premature treatment cessation would lead to the resurgence of transmission and OAE, as the modelled strategies would not lead to interruption of transmission (consistent with the observations of refs. 20,21 for annual CDTI). Of note, the ABR of 7300, used to model 80% mf prevalence (Fig. 5), is notably lower than the value of c.42,000 necessary to capture the (highly holoendemic) baseline situation in the Mbam Valley (nearly 90% mf prevalence), owing to the strongly non-linear relationship between ABR and mf prevalence at endemic equilibrium in EPIONCHO-IBM13.Our model predicts that untreated children (under-fives, specifically those aged 3–4 years) contribute 10–12% of the total OAE incidence prior to CDTI, becoming partially or totally protected by transmission reductions resulting from the treatment of eligibles (particularly from enhanced-coverage biannual treatment in the hyperendemic setting). Our results support the notion that OAE is preventable by strengthening onchocerciasis control and elimination efforts17.LimitationsAs the only driver of OAE in our model is the mf load in childhood (3–15 years), we have not accounted for differences in susceptibility due to other factors. However, the risk of developing epilepsy later in life may be more likely for younger children10,11. Furthermore, genetic, immunological and co-infection factors may render some individuals more inherently susceptible to developing epilepsy10. Besides, our model was parameterised with Simulium damnosum sensu lato as the vector, but in Uganda, where OAE was first reported in Africa22, transmission was due to S. neavei23. Therefore, it cannot be ruled out that the transmission characteristics of different vector groups may influence OAE patterns (but see below).Although not all suspected cases of epilepsy (SCE) identified by Chesnais et al.10 would have been OAE cases, a study conducted in the same area and at the same time (in 2017–18) ascertained that 93.2% of persons with epilepsy fulfilled the OAE diagnostic criteria24. We acknowledge that cysticercosis (another infection associated with epilepsy) has been reported in the Mbam Valley25. However, a case-control study conducted in the village of Bilomo (located in the valley) failed to reveal a statistically significant association between seropositivity for cysticercosis and epilepsy, with the authors proposing onchocerciasis as an alternative explanation for the high epilepsy prevalence found in the village26. More generally, a better understanding and mapping of the co-endemicity of onchocerciasis and taeniasis/cysticercosis in relation to epilepsy in SSA constitutes an important research gap that needs to be urgently addressed27.Currently, EPIONCHO-IBM does not include excess human mortality. A dose-response relationship between mf load and relative risk of mortality has been reported5. Bhattacharyya et al., using the ONCHOSIM transmission model, assumed the operation of excess mortality in individuals with OAE, modelled as a reduction of their residual life expectancy28. If excess mortality were compensated by high birth rates, as observed in South Sudan (LuÃs-Jorge Amaral, pers. comm.), incorporation of these processes into the demographic structure of EPIONCHO-IBM would lead to a younger population and likely generate higher OAE incidence rates than those currently predicted. Expanding the age range to 3–18 years for OAE onset6,8,17 could also be explored, but this was not pursued here, given the lack of mf data beyond 15 years of age in refs. 10,11. Interestingly, the relative mortality risk as a function of mf load was statistically significantly higher for those aged < 20 years compared to adults5.In conclusion, by integrating OAE into EPIONCHO-IBM, via the dose-response relationship between O. volvulus mf load in childhood and risk of developing epilepsy later in life reported in ref. 10, the model closely reproduces observed prevalence and incidence after 19 years of annual CDTI in the Mbam Valley area of Cameroon10,16, as well as the relationship between mf prevalence and epilepsy prevalence across several SSA epidemiological settings7 (Fig. 4). This suggests that such dose-response relationship captures a fundamental process by which O. volvulus infection is associated with epilepsy rather than being an observation specific to the study area, conferring a high degree of generalisability to our modelling framework. Scenario-based simulations highlight the urgency of strengthening MDA-based programmes by increasing treatment frequency and coverage and minimising non-adherence to effect maximal reductions in OAE prevalence and incidence, whose benefits may continue after treatment cessation. In holoendemic settings, alternative and/or complementary strategies will likely be required to markedly reduce transmission and protect gains against OAE. Modelling the impact of larviciding (as done for the Sanaga river29, of which the Mbam is its most important tributary), and/or of other vector control methods such as ‘slash-and-clear’30, alone or in combination with a switch to moxidectin MDA would be an important avenue for further work, given the superior efficacy of moxidectin and similar safety profile compared to ivermectin in clinical trials and its potential to accelerate onchocerciasis elimination as shown in modelling studies31. Given that our work highlights the contribution to OAE incidence of currently untreated children, moxidectin could help address this gap. Paediatric dose-finding studies seeking to investigate the pharmacokinetics and safety of moxidectin for the treatment of 4–11-year-olds have been completed32, and modelling studies should be undertaken to investigate the impact of moxidectin MDA on OAE prevalence and incidence compared to ivermectin. Modelling studies should also be undertaken to quantify the impact of supplementing annual CDTI with an extra treatment round targeted at school-age children33. More generally, our model can help inform onchocerciasis control and elimination programmes aimed to interrupt transmission as proposed by the World Health Organization 2021–2030 Roadmap on NTDs34 and consequently eliminate OAE. Our study strengthens the argument to include OAE in future iterations of the GBD Study to fully capture the range of onchocerciasis-associated disease sequelae in addition to the currently considered cutaneous and ocular clinical manifestations3. To this end, it will be crucial to incorporate OAE-associated premature mortality into EPIONCHO-IBM.
