A diagnostic prediction model for early diagnosis of imported malaria in a European non-endemic setting: Development and temporal validation in a 17-year cohort

Introduction Imported malaria is the most common potentially life-threatening tropical disease among travelers. Early diagnosis remains a clinical challenge and determines its severity and mortality. The aim was to develop and temporally validate a simple clinical model for malaria diagnosis. Methods Retrospective observational cohort study of adults with suspected malaria at a secondary hospital in Madrid over 17 years. Pregnant women, people living with HIV, and submicroscopic malaria cases were included. Inclusion criteria were travel to an endemic area with fever or symptoms prompting parasitological testing. Patients were divided into a derivation cohort (2004–2015) and a temporal validation cohort (2016–2020). Independent predictors were identified using multivariable logistic regression with LASSO selection. Model performance was assessed using ROC curves. Results A total of 649 patients were included; 296 (45.6%) had confirmed malaria due to Plasmodium falciparum (90.9%). Five predictors were identified: thrombocytopenia, absence of leukocytosis, elevated lactate dehydrogenase, fever, and gastrointestinal symptoms. The model achieved an AUC of 0.85 in the derivation cohort and 0.82 (95% CI: 0.73–0.90) in the validation cohort. Using a cut-off ≥14, specificity reached 90%. Conclusions We developed and temporally validated a simple diagnostic prediction model with good discriminatory performance that may support decision-making in patients with suspected imported malaria. The score may help identify travelers at higher probability of malaria.