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

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.