Scrub typhus among hospitalised patients with febrile illness in South India: magnitude and clinical predictors.

ABSTRACT

OBJECTIVES: To derive a clinical algorithm for diagnosis of scrub typhus among patients hospitalized with febrile illness and to determine predictors of bad prognosis. METHODS: Patients hospitalized with febrile illness of 5-30 days duration were evaluated for common aetiologies. Sera were tested for antibodies (IgM and IgG) to Orientia tsutsugamushi using ELISA kit. RESULTS: Among 207 patients, 50 had elevated levels of IgM antibodies. The data of these patients were compared with that of 16 controls having febrile illnesses in whom rickettsial infection was ruled out. Transaminase elevation (>twice normal) was present in 90% and was significantly (P=0.004) more common in those with scrub typhus. If a combination of elevated transaminases, thrombocytopenia and leukocytosis is used, the specificity and positive predictive value are about 80%. Case fatality rate was 14%. Univariate analysis showed that hyperbilirubinemia (>1.5mg%) has a RR of 9 (95% CI=1.48-58.5) and elevated creatinine level (>1.4 mg%) had a RR of 43.99 (95% CI=3.65-530.5) for death. Elevated creatinine level was found to be an independent predictor of mortality (P=0.02). CONCLUSION: In developing countries with limited diagnostic facilities, it is prudent to recommend empiric therapy in patients with undifferentiated febrile illness having evidence of multiple system involvement especially if there is transaminase elevation. Elevated creatinine may predict bad outcome.