Capturing heterogeneity inOpisthorchis viverriniepidemiology and control

Abstract

AbstractOpisthorchis viverriniis a parasitic liver fluke affecting over 10 million people despite sustained control efforts. High intensity infections are a risk factor for the often fatal bile duct cancer, cholangiocarcinoma. Similar to other helminthiases, the distribution of worm burden in humans is highly uneven within populations. We developed multiple models which allow us to capture heterogeneity in transmission and interventions dynamics and the resulting impact on worm distribution: An agent-based model with the common assumption of gamma-distributed transmission parameters; an agent-based model with an alternative nonparametric distribution of transmission parameters; and a simpler ordinary differential equation model. We calibrated all models to prevalence and intensity of infection data in humans, and prevalence data for reservoir hosts and intermediate hosts from southern Lao People’s Democratic Republic. We simulated the impact of multiple interventions on prevalence, intensity of infection and the distribution of worm burden in humans. Our results showed significant overlap in predictions of prevalence and intensity of infection over time between the agent-based models and the ordinary differential equation model, corroborating both the simple and more complex models; however, the nonparametric model was better able to capture the distribution of the highest intensity burden in individuals. Under assumptions of homogeneous adherence to mass drug administration campaigns, no model was able to capture the changing heterogeneity of worm burden over time seen in the epidemiological data. Allowing for heterogeneous adherence in these campaigns, which was only possible in the agent-based models, allowed us to explain the changes seen in the worm distribution and burden seen in the data. This result highlights the added benefit of agent-based models in capturing the changing heterogeneity in worm burden in areas with repeated mass treatments. Appropriately capturing this heterogeneity is essential in understanding the relationship between worm burden, control interventions and subsequent disease burden.Author summaryOpisthorchis viverriniis a parasitic liver fluke affecting over 10 million people despite sustained control efforts. The distribution of worm burden in humans is highly uneven within populations with high intensity infections being a major risk factor for bile duct cancer. We developed and present multiple models, some of which allow us to capture this uneven distribution in susceptibility to infection as well as in adherence to treatment: Two agent-based models of high complexity and a simpler population-based model. We calibrated all models to replicate worm burden data collected in southern Lao People’s Democratic Republic. We simulated the impact of multiple interventions and showed significant overlap of all models in many aspects, corroborating both the simple and more complex models. However, we show that the agent-based models have the added benefit of being able to better capture the unevenness of worm burden before and especially after interventions.