Using individual surveillance records from 2019–2021, this study examined both malaria case patterns and district factors associated with elimination across Sumatra. Of 13,386 reported cases, 71.8% were Plasmodium vivax; among 11,390 investigated cases, 15.3% were classified as imported. Multilevel models identified associations with mobile populations and clinical characteristics, while district elimination was associated with higher human development, lower historical burden, and fewer endemic neighbours. These are observational surveillance associations, not causal effects of a policy.
Key findings
- Among 13,386 reports, 9,617 (71.8%) were P. vivax, 2,467 (18.4%) P. falciparum, and 369 (2.8%) probable P. knowlesi. The P. knowlesi figure remains dependent on microscopic classification.
- Among 11,390 investigated cases, 8,748 (76.8%) were indigenous, 1,741 (15.3%) imported, and 901 (7.9%) unclassified.
- Imported cases were associated with ages 15–24 years (aOR 1.79) and 25–64 years (1.97) versus under 15, male sex (1.70), and mobile or migrant occupations (1.46). Hospitalisation had an aOR of 15.29, an observational reporting association rather than a causal risk estimate.
- At district level, higher human development was associated with greater odds of elimination (aOR 1.15), while higher historical burden (0.46) and more endemic neighbours (0.78) were associated with lower odds. Local clusters were detected for multiple parasite species.
Why this matters globally
As countries approach elimination, malaria becomes concentrated in specific populations and places. This analysis illustrates the value of linking individual cases with district context to distinguish importation from local transmission and target cross-boundary surveillance. The Sumatran model should not be transferred directly elsewhere without validating reporting systems, case definitions, and mobility patterns.
Thai researcher contribution
The publisher lists Karina D Lestari, Sri B Fajariyani, Pimphen Charoen, Saranath Lawpoolsri, and Wirichada Pan-ngum with Mahidol University’s Faculty of Tropical Medicine; Wirichada is also affiliated with MORU. The work formed part of Mahidol’s Biomedical and Health Informatics master’s programme under SPARK. The Early Version only states that Wirichada Pan-ngum and Iqbal R F Elyazar contributed equally and does not expose individual CRediT roles, so no further responsibilities are inferred.
Limitations to consider
This retrospective 2019–2021 surveillance analysis depends on completeness of reporting, testing, and travel histories. Importation used self-reported travel within two to four weeks, and 7.9% of investigated cases remained unclassified. Probable P. knowlesi relied on microscopy, which can misclassify species. District factors are ecological associations and establish neither causation nor intervention effects. Results cover Sumatra across a pre-pandemic and pandemic-era period, limiting transfer. The publisher page is an accepted, peer-reviewed Early Version still subject to editing before the Version of Record.