CLTS, community-led total sanitation intervention; DALY, disability-adjusted life year; EOT, elimination of transmission; FOI, Force-of-infection; MDA, mass drug administration; M&E, monitoring and evaluation; NCC, neurocysticercosis; NTD, neglected tropical disease; PCC, porcine cysticercosis; PAHO, Pan American Health Organization; SAC, school-age children; SSA, sub-Saharan Africa; TS, Taenia solium; WHO, World Health Organization; zDALY, zoonotic disability-adjusted life year.

The views and opinions expressed in this article are those of the authors and do not necessarily reflect those of the World Health Organization. Publication in Gates Open Research does not imply endorsement by the Gates Foundation.

Taeniasis and (neuro)cysticercosis are infections caused by the cestode Taenia solium (TS), involving a complex transmission cycle between the intermediate pig host and the definitive (also accidental intermediate) human host. When humans act as the accidental intermediate host, localization of larval-stage cysticerci in the central nervous system can result in neurocysticercosis (NCC), the main condition contributing to TS-associated morbidity and mortality, including epileptic seizures/epilepsy. TS is endemic across Latin America, sub-Saharan Africa (SSA), and Asia, especially South and South East Asia, in settings with low hygiene conditions where open defecation practices prevail and/or sanitation systems are insufficient to prevent exposure of infective material in human faeces to pigs ¹ . Most recent estimates of disease burden indicate that TS resulted in 1.61 (95% uncertainty interval=1.05–2.23) million disability-adjusted life years (DALYs) globally in 2017 for NCC-associated morbidity and mortality ² . This likely represents a substantial underestimation based on the difficulties inherent to assessing NCC prevalence and the way that disability weights have been attributed to NCC, which need to consider not only epilepsy but, at a minimum, also headache and neuropsychiatric co-morbidities ^{3– 5} . Recent reviews have suggested that TS may also be present in other regions, such as Eastern Europe ⁶ . Large economic consequences do not only pertain to the human public health sector, but also to the animal sector, resulting from reduced market value and market distortion associated with pig infection in the food-value chain, which disproportionately impacts the poorest farmers and communities ^{7– 12} .

A variety of intervention options are available to tackle TS transmission in endemic settings. In the pig host, vaccines are available including TSOL18, alongside anthelmintic treatment using oxfendazole ¹³ . The TSOL18 ¹⁴ vaccine has been licenced and made commercially available in India since November 2016, with registration underway in Uganda, Tanzania, Kenya, Nepal, Philippines, Thailand, and Sri Lanka, while oxfendazole was registered in Morocco for treatment against porcine cysticercosis (PCC) in June 2013 (see WHO article on Paranthic ^TM and Cysvax ^TM ). Intervention options under consideration include treatment of human taeniasis carriers based on mass drug administration (MDA) or on targeted treatment with either praziquantel or niclosamide ¹³ . These treatments, e.g. praziquantel, could potentially be integrated with other neglected tropical disease (NTD) programs ¹⁵ , such as those for schistosomiasis and soil-transmitted helminthiases. However, possible adverse neurological outcomes associated with treatment of NCC cases, especially at higher doses, may restrict the utility of praziquantel ¹⁶ . Further structural changes/interventions that generate broader positive externalities such as impacting TS transmission, include improved sanitation and pig husbandry practices; however, their wide-scale implementation will mostly depend on longer-term economic development ¹⁷ . Health education, such as the computer-based educational tool ‘The Vicious Worm’ ¹⁸ , provides a low-cost, locally-adaptable, and implementable intervention for both short- and longer-term impact. Studies demonstrate improved and sustained knowledge uptake in Tanzanian health- and agriculture-sector professionals ^{19, 20} , as well as in rural Zambian primary-school children ²¹ . In a rural Mexican community, health education, developed with community-participation, showed reductions in pig foraging behaviour and access to infective material, accompanied by reductions in pig cysticercosis prevalence ²² . Improvement in knowledge areas associated with reducing risk factors through health education, directed at school-age children ²³ , has highlighted the role of health education campaigns. Other community-based participatory educational interventions in Burkina Faso have also demonstrated a marked reduction in human cysticercosis incidence and prevalence ²⁴ . A community-led total sanitation (CLTS) intervention in Zambia did not reduce porcine cysticercosis prevalence, with sanitation practices and cysticercosis awareness largely unchanging ²⁵ , further indicating the importance of knowledge uptake. A recent systematic review analysed the available evidence on the effectiveness of TS intervention options ²⁶ , concluding that combined human- and pig-focussed interventions are the most promising strategies for achieving rapid declines in infection ²⁷ and enhancing prospects for regional elimination ²⁸ . This further supports the argument for a One Health approach, including humans and non-human animals as well as the environment. In addition, integrated knowledge translation, where knowledge-users work with researchers throughout the research process ²⁹ , as well as contextualized policy- and practice-transfer mechanisms seem important for sustainability of the obtained results. In 2012, the World Health Organization (WHO) NTD roadmap called for the establishment of a validated strategy for TS control and elimination by 2015, and for interventions to be scaled up in selected countries by 2020 ³⁰ . However, a validated strategy has not yet been established. There are currently TS pilot control programs under assessment in Madagascar, with pilots also planned for Vietnam and China ³¹ . Currently, there are no specific TS programmatic goals as identified for other NTD programs, reflecting the relatively early stage of consolidation of an optimal disease control strategy compared to other NTDs.

A recent WHO consultation was held to gather evidence to support a new NTD roadmap for post-2020 targets and milestones. Transmission dynamics models can help address evidence gaps by assessing, in silico, the feasibility and potential impact of different intervention strategies to achieve proposed 2021–2030 targets. Dixon et al. ³² recently conducted a systematic review and analysis of available TS transmission models, identifying four mathematical/computational/statistical models (among them three transmission dynamics models), and a conceptual ‘logical’ framework ³³ . Models included a decision-tree ³⁴ , Reed-Frost ³⁵ , individual-based ³⁶ , and population-based ³⁷ frameworks. (Other models have been published since this review ^{38, 39} .) On-going development of cystiSim ³⁶ and EPICYST ³⁷ within a collaborative umbrella, has led to the establishment of ‘CystiTeam’, a partnership between the groups that developed these two models (based at the University of Copenhagen/Scientific Institute of Public Health, Brussels, and Imperial College London, respectively) and groups of epidemiologists, veterinarians, clinicians, one-health experts and program stakeholders. With the development of the new 2030 NTD goals in mind, Table 1 outlines the current NTD goals for TS (2015, 2020) and the proposed goals for 2030, with a summary of their technical feasibility, requirements, and most prominent knowledge gaps and associated risks.

A quantitative comparison of TS transmission dynamics models has not yet been performed, and therefore it is not possible to formally compare the (cystiSim ³⁶ and EPICYST ³⁷ ) models at this stage (an aim of the CystiTeam partnership). The lack of standardised programmatic targets also restricts the ability to determine the effectiveness of different interventions with the available modelling frameworks. Current models, however, do shed some light on the potential impact of interventions under more generalised, illustrative scenarios. The population-based, deterministic, transmission model EPICYST ³⁷ has highlighted, based on simulating a generic SSA setting, the higher efficacy (measured as the percentage of human cysticercosis cases prevented) of biomedical interventions (human test-and-treat, pig MDA and pig vaccination) compared to structural change-based interventions (improved husbandry, sanitation and meat inspection), although insufficient data and knowledge currently exist to parameterize accurately the latter ( Figure 1). In EPICYST ³⁷ , further developments are on-going to reflect age-specific stratification of interventions such as pig vaccination scheduling and to test the impact of praziquantel MDA in school-age children (SAC). cystiSim ³⁶ , an agent-based, stochastic, model is able to simulate age-structured, field-realistic interventions as well as bespoke treatment efficacy for varying settings. cystiSim has identified, within the context of an endemic district in Tanzania, that two-host interventions (human MDA plus pig vaccination and treatment), are optimal strategies to achieve elimination of transmission (EOT) if high coverage can be reached and sustained for prolonged periods ³⁶ . Both EPICYST ³⁷ and cystiSim ³⁶ agree that pig- and human-directed interventions are sensitive to coverage levels. Biomedical interventions were identified as more robust to changes in both coverage and efficacy compared to structural change-based interventions in EPICYST ³⁷ , although modelling of the latter requires collection of robust data. cystiSim ³⁶ showed that coverage is particularly important when interventions target a single host, but the addition of an intervention targeting the second host can compensate for lower coverages ³⁶ . More recently, Braae et al. ⁴² used cystiSim ³⁶ to explore intervention simulations deemed to be closely aligned to the (assumed) population biology of the parasite; for example, testing a combined pig intervention (TSOL18 vaccine and oxfendazole MDA) for a duration of 3 years to reflect the modelled average lifespan of the adult tapeworm ⁴³ . The EOT probability was >90% in this scenario (coverage: 75%; frequency: 3-monthly; duration: 3 years), suggesting a role for pig-only strategies if these can be implemented at high coverage and frequency for sufficiently long. This modelling study also indicated that the program duration could be reduced to 2 years with a similar EOT probability (>85%) with addition of human MDA after the first year (coverage: 80%; frequency: 6 monthly; no. of treatment rounds: 3; duration: 2 years). Inclusion of pig-focussed interventions (with or without human MDA) was substantially more effective than human MDA-only strategies (coverage: 80%; frequency: annual or 6-monthly; duration: 5 or 10 years) ( Figure 2).

cystiSim has been used to inform potential control activities under the Pan American Health Organization (PAHO) and to support intervention design (intervention selection, coverage, frequency, duration) and assessment of options under consideration in the community-based intervention pilot project ‘CYSTISTOP’ in Zambia. CYSTISTOP commenced in 2015 and is due to end in 2020 ⁴⁴ . Modelling comparisons examine different intervention strategies, including mass or targeted treatment programs assumed to be feasible in ‘lower input/investment’ systems versus more intensive elimination strategies which focus on combined, higher-frequency interventions ⁴⁰ . Yearly pig oxfendazole MDA strategies (drug efficacy: 100%; therapeutic coverage: 90% of pigs aged ≥2 months; frequency: annual; no. rounds: 12; duration: 12 years) provided the most effective ‘control’ approach (EOT probability =75%), compared to human praziquantel MDA (drug efficacy: 95%; therapeutic coverage: 85% of humans aged ≥5 years; frequency: annual; no. rounds: 12; duration: 12 years), with EoT probability=1%. In terms of highly intensive elimination options, combined interventions (human and pig MDA with drug efficacy and coverage as above plus pig vaccination; frequency: 4-monthly; no. rounds: 6; duration: 2 years) was the most effective (EOT probability =96.5%) ( Figure 3). These simulated interventions will be evaluated using the final CYSTISTOP results for model validation.

Addressing existing models’ structural and parametric uncertainty is a critical step towards enabling such models to increase their predictive capacity to assess the 2030 TS targets and provide robust support. There are several frameworks available ^{32– 39} with groups in CystiTeam working on improving the cystiSim and EPICYST models collaboratively. Knowledge gaps have been highlighted to further model development, including age-specific infection trends and local practices to inform setting-specific parameterization. Other biological parameters requiring further investigation include the average and distribution of the adult tapeworm lifespan, processes regulating parasite acquisition in humans and pigs, and exposure heterogeneity, which may manifest as aggregated (overdispersed) infection distributions at the population level (as attested by the distribution of cysticerci in pig populations ^{49, 62, 63} ). Identifying how current demographic assumptions impact transmission, such as the pig-to-people population ratio, need to be systematically tested between current TS transmission models. Figure 4 presents key research gaps and data needs to move the field forward.