Evaluating the variation in the projected benefit of community-wide mass treatment for schistosomiasis: Implications for future economic evaluations

Background The majority of schistosomiasis control programmes focus on targeting school-aged children. Expanding the use of community-wide mass treatment to reach more adults is under consideration. However, it should be noted that this would require a further increase in programmatic resources, international aid, and commitment for the provision of praziquantel. Consequently, it is important to understand (i) where a change of strategy would have the greatest benefit, and (ii) how generalisable the conclusions of field trials and analytical studies based on mathematical models investigating the impact of community-wide mass treatment, are to a broad range of settings. Methods In this paper, we employ a previously described deterministic fully age-structured schistosomiasis transmission model and evaluate the benefit of community-wide mass treatment both in terms of controlling morbidity and eliminating transmission for Schistosoma mansoni, across a wide range of epidemiological settings and programmatic scenarios. This included variation in the baseline relative worm pre-control burden in adults, the overall level of transmission in defined settings, choice of effectiveness metric (basing morbidity calculations on prevalence or intensity), the level of school enrolment and treatment compliance. Results Community-wide mass treatment was found to be more effective for controlling the transmission of schistosome parasites than using a school-based programme only targeting school-aged children. However, in the context of morbidity control, the potential benefit of switching to community-wide mass treatment was highly variable across the different scenarios analysed. In contrast, for areas where the goal is to eliminate transmission, the projected benefit of community-wide mass treatment was more consistent. Conclusion Whether community-wide mass treatment is appropriate will depend on the local epidemiological setting (i.e. the relative pre-control burden in adults and transmission intensity), and whether the goal is morbidity control or eliminating transmission. This has important implications regarding the generalisability of cost-effectiveness analyses of schistosomiasis interventions. Our results indicate that areas with poor school-enrolment/coverage could benefit more from community-wide treatment of praziquantel and should potentially be prioritised for any change in strategy. This work highlights the importance of not over-generalising conclusions and policy in this area, but of basing decisions on high quality epidemiological data and quantitative analyses of the impact of interventions in a range of settings. Electronic supplementary material The online version of this article (doi:10.1186/s13071-017-2141-5) contains supplementary material, which is available to authorized users.


Supplementary information
The Model The fundamental model used to describe the mean worm burden of individuals of a given age and the quantity of infectious eggs in the environment was developed from the founding work of Anderson and May [1]. The current version of the model is described in detail in [2][3][4][5].
The key parameters in Supporting Table 1 and Supporting Table 2.
The model was further adapted to allow for systematic non-compliance (individuals never taking treatment) [6,7]. A more detailed description of the model is provided in Truscott et al. [8].  [9,10] Aggregation parameter, k 0.24 [11] Density dependence fecundity, γ 0.0007/female worm [3] Drug efficacy (proportion of worms killed by praziquantel) 0.86 [12] Egg output per female worm in absence of density dependence (in terms of faecal egg counts) 0.14 [13] Human demography Based on Uganda's demographical profile [14,15] Life expectancy of the infected snails 4 weeks [16] Baseline scenarios

Supporting
Based on the available data [3], we developed three age-intensity profiles representing a range of possible scenarios regarding the relative burden in adults.
When changing the assumed the life span of the adult worms, the age-specific parameters for the exposure and contribution to the infectious reservoir were refittedso that the shape of the age profile does not change.
Evaluating the variation in the projected benefit of community-wide mass treatment for schistosomiasis: Implications for future economic evaluations Supporting

The model was used to simulate two transmission settings; a higher transmission setting with an age-weighted mean worm burden of 155, and a lower transmission setting with a mean worm burden of 60. The data used to define the three scenarios is presented in [3].
The model was used to simulate two transmission settings; a higher transmission setting with an age-weighted mean worm burden of 155 (based on model fits to the data [5]), and a lower transmission setting with a mean worm burden of 60. To ensure the results for the different scenarios are comparable, the R0 was adjusted such that the different scenarios had the same pre-control mean worm burden (i.e. we ensured that we are not comparing the impact of both a different age-infection profile and a different pre-control burden when comparing the different scenarios).
Evaluating the variation in the projected benefit of community-wide mass treatment for schistosomiasis: Implications for future economic evaluations Supporting The range in each cell shows the variation in the relative increase in effectiveness when using annual community-wide versus school-based treatment to the assumed level of treatment coverage in adults (55% vs. 75%). The scenarios for the relative pre-control burden in adults are shown in Figure 2 ( Evaluating the variation in the projected benefit of community-wide mass treatment for schistosomiasis: Implications for future economic evaluations Supporting The

range in each cell shows the variation in the relative increase in effectiveness when using annual community-wide versus school-based treatment to the assumed mean life expectancy of the adult worms (4 years vs. 5.71 years). The scenarios for the relative pre-control burden in adults are shown in Figure 2 (note they have the same age-weighted overall mean worm burden). The results assume a treatment coverage of 75% and 5% systematic non-compliance. The analysis was conducted with a five-year implementation period and a 15-year time horizon (i.e. looking at the impact of five years of treatment over 15 years).
Evaluating the variation in the projected benefit of community-wide mass treatment for schistosomiasis: Implications for future economic evaluations Supporting The range in each cell shows the variation in the relative increase in effectiveness when using annual community-wide versus school-based treatment to the assumed level of systematic non-compliance (0% vs. 20%). The scenarios for the relative pre-control burden in adults are shown in Figure 2 (note they have the same age-weighted overall mean worm burden). The results assume a treatment coverage of 75%. The analysis was conducted with a five-year implementation period and a 15-year time horizon (i.e. looking at the impact of five years of treatment over 15 years). The results assume the systematic non-compliance rate is 20% for the school-based programme, and 5% when using community-wide mass treatment (simulating a scenario where many of the non-enrolled SAC are only reached when using a community-based programme). The scenarios for the relative pre-control burden in adults are shown in Figure  2 (note they have the same age-weighted overall mean worm burden). The treatment coverage was assumed to be 75%.

The analysis was conducted with a five-year implementation period and a 15-year time horizon (i.e. looking at the impact of five years of treatment over 15 years).
Evaluating the variation in the projected benefit of community-wide mass treatment for schistosomiasis: Implications for future economic evaluations Supporting In the first results column the model was fitted to fully age-structured data (and therefore accounts for the true shape of the age-intensity profile) [5]. In the second results column, the model was only fitted to reproduce the estimated mean pre-control worm burdens in SAC and adults from the same dataset. The data used in this example is from the Iietune village (Kenya) [60]. The results assume a treatment coverage of 75% and no systematic non-compliance. The analysis was conducted with a five-year implementation period and a 15-year time horizon (i.e. looking at the effect of five years of treatment for 15 years).