Monthly persistence and reappearance of reported malaria occurrence due to Plasmodium falciparum and Plasmodium vivax in the Peruvian Amazon, 2009–2024: a longitudinal geospatial study: Original research article

Vera-Ponce V.J. · Acta Tropica (2026) · DOI: 10.1016/j.actatropica.2026.108351

Resumen

Malaria elimination requires distinguishing continued reporting from return after zero-reporting periods. In this longitudinal ecological study, we compared monthly persistence and reappearance of reported Plasmodium falciparum and Plasmodium vivax malaria in the Peruvian Amazon from 2009 to 2024. Reported surveillance episodes, which were not assumed to represent unique individuals, were aggregated by month, species, and 350 harmonized geographic units while retaining zero counts. A dynamic hurdle model combined a Bernoulli occurrence component, a zero-truncated negative binomial positive-count component, species-specific spatial effects, and geospatial and temporal covariates. For temporal validation, models were fitted on 2009–2022 and evaluated in 2023–2024. Among 585,510 reported surveillance episodes, 81.8% were attributable to P. vivax. In the 76-unit eligible common-support universe, adjusted persistence was 0.790 for P. falciparum and 0.848 for P. vivax; the difference was 0.053 (approximate 95% uncertainty interval [UI], −0.088 to 0.299). Under the corrected reappearance risk set requiring a strictly earlier positive report, adjusted reappearance was 0.071 and 0.146, respectively; the difference was 0.075 (approximate 95% UI, 0.030 to 0.128). Predicted-to-observed ratios were 0.821 for P. falciparum and 0.900 for P. vivax; both met the prespecified validation criteria, although underprediction was greater for P. falciparum. Larger population, lower elevation, and greater forest cover were associated with reported occurrence. The persistence contrast remained uncertain; corrected grouped reappearance was higher for P. vivax, although duration-specific contrasts did not uniformly exclude zero. Separating surveillance-defined continuity from return of reporting may support subnational monitoring after prospective validation; climate associations remained exploratory because of inadequate calibration and instability.

Instituciones: Universidad Nacional Toribio Rodríguez de Mendoza de Amazonas