# Dengue transmission heterogeneity across Indonesia’s archipelago: Climate-driven spatiotemporal patterns and policy implications

## TL;DR

Dengue outbreaks in Indonesia follow a climate-driven west-to-east timing pattern, with El Niño events linked to larger outbreaks, suggesting the need for localized early warning systems.

## Contribution

The study identifies structured dengue-climate heterogeneity across Indonesia and proposes a two-tier early warning system based on regional coherence.

## Key findings

- A west-to-east gradient in dengue wave timing aligns with monsoon progression in western Indonesia.
- El Niño events are strongly associated with increased dengue incidence, nearly doubling during strong events.
- 18 provinces show consistent rainfall-dengue timing, indicating potential for localized early warning systems.

## Abstract

Indonesia has the highest dengue burden in Southeast Asia, with 488 of 514 districts reporting cases annually across its 17,000-island archipelago. Despite this substantial burden, spatiotemporal transmission patterns remain poorly characterised. We analysed province-level dengue surveillance data (2010–2024) from Indonesia’s Ministry of Health alongside local and regional climate variables to characterise heterogeneity in dengue periodicity and identify provinces where climate-based early warning may be feasible. Using wavelet phase analysis, dynamic time warping clustering, and distributed lag non-linear models, we examined relationships between climate and dengue incidence across 34 provinces. A systematic west-to-east gradient in dengue wave timing was identified, with Northern Sumatran provinces peaking earlier than other provinces, aligning with Australian-Asian monsoon progression. This gradient was robust in western Indonesia (Spearman ρ = 0.7 between longitude and phase lag) but weakened in eastern provinces. Multi-annual outbreak peaks (2015–2016, 2023–2024) coincided with strong El Niño events, with mean incidence during strong El Niño years was 96% higher than other years. The Indian Ocean Dipole showed no significant association. Phase coherence analysis identified 18 provinces where precipitation-dengue timing was sufficiently consistent (coherence ≥0.85) for potential early warning applications and DLNM confirmed significant dose-response associations in 11 of these. Indonesia’s dengue-climate relationships exhibit structured heterogeneity that precludes uniform national prediction approaches but may enable province-specific early warning in high-coherence areas. A two-tier system combining ENSO monitoring for strategic preparedness with local climate monitoring for tactical intervention timing could improve outbreak response across Indonesia’s diverse epidemiological landscapes.

Indonesia reports hundreds of thousands of dengue cases annually, ranking among the highest-burden countries in Southeast Asia. Effective disease control requires understanding when and where outbreaks occur, yet Indonesia’s vast archipelago spanning over 5,000 kilometres presents unique challenges for prediction and response. We analysed 15 years of dengue surveillance data across all 34 Indonesian provinces to characterise how outbreak timing varies geographically and relates to climate patterns. We found a consistent west-to-east gradient in outbreak timing across western Indonesia, with provinces in Sumatra peaking up to four months earlier than those in Java and Bali, following the progression of the monsoon system. However, this pattern breaks down in eastern Indonesia, where climate regimes differ markedly. El Niño events were strongly associated with nationwide outbreak years, with incidence nearly doubling during strong El Niño periods. We identified 18 provinces where rainfall-dengue timing relationships remained consistent across years, suggesting where climate-based early warning systems may be most feasible. Our findings support province-specific rather than uniform national approaches to dengue preparedness, with coordination between early- and late-peaking provinces to optimise resource deployment.