# Sea surface temperature modulates El Niño and La Niña driven leptospirosis patterns: Evidence from causal machine learning in Colombia

## TL;DR

This study shows how El Niño and La Niña affect leptospirosis cases in Colombia, with climate factors like sea surface temperature playing a key role.

## Contribution

The study introduces causal machine learning to analyze the impact of ENSO on leptospirosis at a municipal level in Colombia.

## Key findings

- El Niño increases the probability of excess leptospirosis cases by 7.2 percentage points compared to neutral conditions.
- La Niña reduces the probability of excess leptospirosis cases by 1.2 percentage points compared to neutral conditions.
- Higher sea surface temperatures reduce the impact of ENSO on leptospirosis cases.

## Abstract

Leptospirosis is a zoonotic disease prevalent in tropical regions influenced by climatic factors such as precipitation and soil moisture, which are regulated by the El Niño-Southern Oscillation (ENSO). This study examines the causal relationship between El Niño and La Niña episodes and leptospirosis cases in Colombia at the municipal level from 2007 to 2023. Using an ecological longitudinal design, we analyzed laboratory-confirmed cases from the National Public Health Surveillance System, environmental data from remote sensing, and socioeconomic data, employing a causal machine learning framework with doubly robust estimation and overlap weighting. We estimated the Average Treatment Effect (ATE) and the Conditional Average Treatment Effect (CATE) for three scenarios: Neutral vs. La Niña, Neutral vs. El Niño, and El Niño vs. La Niña. Results showed 10,629 cases, predominantly in males, with the highest incidence in Cali, Barranquilla, San José del Guaviare, and Cartagena. La Niña was associated with a 1.2 percentage point reduction in the probability of excess leptospirosis cases (ATE = -0.012, 95% CI: -0.015 – -0.008), while El Niño corresponded with a 7.2 percentage point increase in the probability of excess leptospirosis cases (ATE = 0.072, 95% CI: 0.041 – 0.103) compared to Neutral episodes. The El Niño vs. La Niña comparison showed no significant effect. As sea surface temperatures rose in the Pacific Ocean off the Colombian coast, the impact of both El Niño and La Niña episodes was observed to diminish, according to the CATE analysis. Regional variations, particularly in the Orinoco and Amazon regions, seem to drive these national trends, probably due to inverse hydro-climatic responses to ENSO. Refutation tests indicated the presence of remaining bias for the scenarios Neutral vs. El Niño and El Niño vs. La Niña. These findings highlight the complex interplay between climate and leptospirosis, underscoring the need for region-specific public health strategies to mitigate climate-driven disease risks in Colombia.