Unraveling the invisible leptospirosis in mainland Southeast Asia and its fate under climate change

Abstract

Leptospirosis is a neglected waterborne zoonosis of growing concern in tropical and low-income regions. Endemic in Southeast Asia, its distribution and environmental factors such as climate controlling its dynamics remain poorly documented. In this paper, we investigate for the first time the current and future leptospirosis burden at a local scale in mainland Southeast Asia. We adjusted machine-learning models on incidence reports from the Thai surveillance system to identify environmental determinants of leptospirosis. The explanatory variables tested in our models included climate, topographic, land cover and soil variables. The model performing the best in cross-validation was used to estimate the current incidence regionally in Thailand, Myanmar, Cambodia, Vietnam and Laos. It then allowed to predict the spatial distribution of leptospirosis future burden from 2021 to 2100 based on an ensemble of CMIP6 climate model projections and 4 Shared Socio-economics Pathways ranging from the most optimistic to the no-climate policy outcomes (SSP1–2.6, SSP2–4.5, SSP3–7.0 and SSP5–8.5). Leptospirosis incidence was best estimated by 10 environmental variables: four landscape-, four rainfall-, two temperature-related variables. Of all tested scenario, the worst-case scenario of climate change (SSP5–8.5) surprisingly appeared as the best-case scenario for the future of leptospirosis since it would induce a significant global decline in disease incidence in Southeast Asia mainly driven by the increasing temperatures. These global patterns are however contrasted regionally with some regions showing increased incidence in the future. Our work highlights climate and the environment as major drivers of leptospirosis incidence in Southeast Asia. Applying our model to regions where leptospirosis is not routinely monitored suggests an overlooked burden in the region. As our model focuses on leptospirosis responses to environmental drivers only, some other factors, such as poverty, lifestyle or behavioral changes, could further influence these estimated future patterns.