Developing a severity index on day 4 to predict severe dengue infection in adults

Abstract

INTRODUCTION: Dengue causes high morbidity and mortality among adults in Sri Lanka. Early prediction of severe illness would help to reduce morbidity and mortality. Studies to identify predictors of severe dengue in adults are sparse. AIMS: To identify predictors of severe dengue infection by the fourth day of illness. Methods: Symptoms, signs and investigation results on the 4th day of illness were compared between two groups of patients with serologically confirmed dengue over 6 months from 1st of March 2011; Group A (severe illness: evidence of fluid leakage, compensated shock, profound shock). Group B (non-severe illness). RESULTS: Of 117 adults 9meanage 32 yrs (SD= 13.3) and 95 males) 27 fell into Group A and 90 into group B. On day 4 of illness serum aminotransferases (AST and ALT) were significantly higher in Group A than group B [AST: 260 iu/1 (SD=168.8] vs 145 iu/l(SD 135.11), p=0.005; ALT: 247 iu/1 (SD= 161.5) vs 105 iu/1 (SD= 91.5), p=0.002]. Overall AST (r=0.3, p=0.038) and ALT (r=0.3, p=0.045) had a positive correlation with haematocrit (PCV). An index was developed using stepwi-se multivariate discriminant function analysis to predict severe infection by the 4th day. A severity Index, [(0.082 x PCVD4) + (0.02 x PlateIetD4) + (-0.006 x ASTD4) -3.677] of <-0.258 predicted severe infection with 73.7 % specificity, 73.8% sensitivity, 56% positive predictive value and 86% negative predictive value. CONCLUSIONS: A severity index <-0.258 calculated on the 4th day of illness may predict severe infection among adult dengue patients. This must now be validated prospectively.