Are platelet indices useful in diagnosis of tropical acute febrile illnesses?

Abstract

Background: Diagnosis and treatment of tropical acute febrile illnesses is challenging with respect to health facilities and personnel and diagnostic kits. Platelet indices are available at no extra cost or blood requirements. They are useful in diagnosing both infective and non-infective diseases. However, these indices have not been used previously to predict the differential diagnosis of common tropical febrile illnesses using a decision tree. Methods: A retrospective analysis of clinical and laboratory data of 402 patients with fever lasting for >24 h and < 3 weeks was performed. Subjects were divided into seven diagnostic groups: bacterial, viral, rickettsial, parasitic, mycobacterial, undifferentiated, and non-infective. Bivariate correlation and distribution of platelet indices among the diagnostic groups were examined. The area under the curve was obtained using these indices for each group. A classification tree was also constructed with these indices and other variables sequentially, to predict the likely cause of the fever. Results: Subjects were predominantly males (255) and young individuals (242). The most common causes of fever were viral (18.2%) and bacterial (16.8%) infections. Platelet indices varied significantly across all the groups, even after controlling for other clinical/laboratory parameters. Receiver-operating characteristic analysis showed that high platelet count had the best sensitivity and specificity for diagnosis of tuberculosis. The classification tree constructed using platelet indices alone had a higher margin of error than that using a combination of clinical symptoms, complete blood count, liver function tests, and ECG heart rate. Conclusions: Platelet indices vary significantly across tropical illnesses. Thrombocytosis has good sensitivity and specificity for diagnosis of tuberculosis. Combined with clinical symptoms and routine blood tests, these indices predicted a likely diagnosis in two-thirds of patients. Prospective validation in different localities is needed prior to the use of these indices in the diagnosis of acute febrile illnesses.