PERIPHERAL BLOOD COUNT FOR DENGUE SEVERITY PREDICTION: A PROSPECTIVE STUDY IN THAI CHILDREN

INTRODUCTION: Dengue viral infection has a wide range of severity levels and requires different levels of medical attention. Early severity prediction using clinical features is difficult. Certain lymphocytic subtypes can be used to predict severity; we postulate that peripheral blood counts can also predict severity, which would be more useful in smaller rural hospitals. OBJECTIVE: We aimed to compare the peripheral blood counts between patients with mild dengue infection and those with severe dengue infection and identify simple yet sensitive early severity predictors. METHODS: We enrolled 91 patients with serologically confirmed dengue infection who were admitted to King Chulalongkorn Memorial Hospital. Their leukocytic counts on admission were compared. Potential predictors were identified by using receiver-operating-characteristic analysis. RESULTS: Compared with patients with mild infection, those with severe infection (dengue hemorrhagic fever grade II or worse) had a higher leukocyte count (3580 vs 3050 cells per μL; P = .04), and fewer had leukopenia on admission (70% vs 89%; P = .03). They also had a lower percentage of “typical” lymphocytes (24% vs 40%; P = .02). Two predictors were identified; either one classified ∼19% of all admitted patients as being at low risk. Typical lymphocyte counts of <40% excluded patients with mild disease with 89% sensitivity and 24% specificity (negative predictive value: 77%; positive predictive value: 45%). A combination of parameters [(white blood cells per μL) + 470 × (% typical lymphocytes) + 5 × (atypical lymphocytes per μL) ≥−14 950] improved the sensitivity and specificity to 92% and 26% (negative predictive value: 82%; positive predictive value: 46%). CONCLUSIONS: The absence of leukopenia and a low percentage of typical lymphocytes predict severe dengue illness. Simple hematologic parameters may be used to reduce unnecessary admissions of patients with suspected dengue infection in the absence of more sophisticated predictors.