Rainfall and Leptospirosis in the Dominican Republic, 2012–2026: A Distributed-Lag Time-Series and Province-Panel Analysis
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
The biological mechanism linking rainfall with leptospirosis transmission is well established, but rigorous quantitative evidence for the insular Caribbean remains scarce, limited mostly to descriptive reports of post-hurricane outbreaks. We aimed to quantify the association between rainfall and leptospirosis incidence in the Dominican Republic, where leptospirosis is endemic, between 2012 and 2026 using a distributed-lag approach.
Methods
We conducted an ecological time-series study complemented by a province panel with fixed effects. Case data (5,412 valid cases) were obtained from the national surveillance system (SIP-0276FA51); rainfall data were obtained from 15 INDOMET rain-gauge stations across 13 provinces (2000–2026). We calculated the cross-correlation function between lagged monthly rainfall (0–6 months) and case counts, fitted a negative binomial distributed-lag regression model (lags 0–3 months) adjusted for seasonality and trend, triangulated findings with a 13-province fixed-effects panel, compared four rainfall exposure metrics, assessed extreme-rainfall threshold sensitivity, and estimated the population attributable fraction (PAF) with a parametric bootstrap.
Results
The rainfall-case association peaked at a 1-month lag (r = 0.551; 95 % CI 0.437–0.647; p < 0.001) and remained significant through 3 months. In the distributed-lag model, all four lags were independently significant, with the 1-month lag showing the strongest effect (incidence rate ratio [IRR] = 1.156 per additional 50mm; 95 % CI 1.090–1.226; p < 0.001). The province panel yielded an almost identical 1-month lag effect (IRR = 1.147; 95 % CI 1.127–1.167). Rainfall above the historical 90th percentile increased case risk the following month (rate ratio = 1.95), with effect magnitude increasing monotonically with threshold stringency. An estimated 28.2 % (95 % CI 19.0–36.6 %) of cases were attributable to rainfall above the recorded historical minimum.
Conclusions
Rainfall is a robust, consistent predictor of leptospirosis incidence in the Dominican Republic, with the strongest association observed at a 1-month lag and a lag pattern broadly consistent with findings from distributed-lag studies in Thailand and the Philippines. These findings, which to our knowledge constitute the first formal quantification of this association for the insular Caribbean, provide an evidence base for rainfall-linked early-warning systems, with the strongest signal at approximately one month and elevated risk extending through three months after rainfall.
