Validation of a prognostic score for complicated malaria

Abstract

Introduction: Early identification of subjects with malaria who may develop complications is critical to guide treatment. A predictive tool based on the clinical characteristics of the disease is required to predict complications.

Objective: To develop and validate a clinical and paraclinical tool to predict complicated malaria.

Methods: Subjects with a diagnosis of complicated malaria according to the World Health Organization criteria were sequentially recruited. Clinical characteristics associated with complicated malaria were identified and incorporated into a binary logistic prediction model in an initial derivation cohort and, subsequently, validated in a different cohort of subjects with the infection.

Results: 584 subjects were analyzed in the derivation cohort and 898 in the validation cohort. The incidence of complicated malaria was 21.2% and 10.9%, respectively. Variables associated with complicated malaria were dyspnea, fever for more than 72 hours, hemoglobin < 13 mg/dL, total bilirubin > 6 mg/dL, urea nitrogen > 20 mg/dL and Plasmodium vivax. The area under the receiver operating characteristics curve for the score constructed with these variables in the derivation cohort was 0.70 and in the validation cohort, it was 0.69.

Conclusion: A prognostic score that evaluates the presence of dyspnea, persistent fever, hemoglobin level, total bilirubin, urea nitrogen and P. vivax infection in patients with malaria may be useful to classify in a timely and objective manner subjects who may develop complications.