Sensitivity and specificity of fever as a clinical predictor for malaria in Loreto, Peru

Abstract

Introduction: Malaria is a parasitic disease caused by Plasmodium sporozoites. Peruvian National Regulations for Malaria and Severe Malaria state that active surveillance of febrile patients is the main activity to perform, since it allows early case detection.
Objective: To determine the usefulness of fever as a clinical predictor for malaria in endemic areas of Loreto; sensitivity and specificity of fever were also assessed.
Material and Method: We prospectively included 400 febrile patients who presented to the Malaria Program at Loreto Regional Hospital from October to December 2002.
Results: We found a 42.1% sensitivity (95% CI: 31.6-53.3), 62.7% specificity (95% CI: 57.3-67.7), 20.9% positive predictive value (95% CI: 15.2-28.0), 82.2% negative predictive value 82.2 (95% CI: 76.9-86.5), 58,8% accuracy (95% IC: 53,9-63,5). Fever (axillary temperature > 37,5 degrees Celsius) was associated to a final diagnosis of malaria (OR = 1,22; 95% CI = 0,73 2,03). In our study, fever does not have enough sensitivity nor specificity for detecting malaria cases; thus, it is not a good clinical predictor in endemic areas. Health facilities could mistakenly withhold antimalarial treatment to 57,9% of patients with a positive thick blood smear because of not having fever (false negatives), leaving an important human reservoir perpetuating transmission. Also, they could provide antimalarial treatment to 37,3 % of patient with a negative thick blood smear, only because of having fever (false positives).
Conclusion: Malaria diagnosis requires of more sensitive and more specific clinical criteria. Improved case definitions for malaria must be elaborated, so they may be used particularly in endemic areas for malaria where no microscopy is available.