Measuring elimination of gambiense human African trypanosomiasis: a comparison of deceptively different metrics.

Modelling is an effective and widely used tool for predicting the future trajectory of infectious diseases. One of its strengths is its ability to measure and predict metrics that are unobservable in real life, such as the true prevalence, or when infection events occur. Policy goals, on the other hand, must necessarily be based on observable metrics in order to be verifiable. The WHO has targeted gambiense human African trypanosomiasis (gHAT) for elimination of transmission (EoT) by 2030. In order to verify this, the WHO HAT Elimination Technical Advisory Group (HAT-e-TAG) has recently defined the country-level indicator for EoT as 5 years of no new locally-infected reported cases, along with a sufficient level of surveillance. While this indicator is a useful and concrete metric that can be clearly measured, it does not directly measure the actual transmission or prevalence of the disease. In some cases, the timing of the last transmission event or the point when infection is no longer present in a country may differ quite substantially from when the country reaches the WHO indicator for EoT. In this article, we discuss the difference between these different metrics and show with some examples how much they can disagree. Furthermore, modelling papers have previously been published using various approximations of EoT, which has the potential to cause confusion. In this paper, we highlight the potential misunderstandings which could occur due to seemingly trivial differences in the definition of elimination and its indicators, especially in the context of such a slow-progressing disease as gHAT. We conclude that while there is value in models predicting both observable and unobservable metrics, modellers need to ensure that they are clear about definitions when communicating results, and policymakers need to be sure that they know what definitions are being used when drawing conclusions from model results. While this paper focuses specifically on three indicators of EoT for gHAT, the general points made here could have applications in other infections approaching the endgame, such as onchocerciasis, or even beyond public health.