Univariate and multivariable logistic regression analysis examining relationships between clinical and laboratory features present at enrolment and diagnosis of dengue in the study cohort of 856 patients.

Logistic regression models were adjusted for age, sex, day of illness at enrolment and study site (Clinic A vs Clinic B). Interaction tests revealed no evidence for a difference in the effect of UACR by study site (p = 0.13 for univariate analysis and 0.27 for multivariable analysis, likelihood ratio test). AUC and sensitivity/specificity were calculated without adjustment.All factors from the univariate analysis were included in the multivariable model apart from microalbuminuria, since this was already represented by the UACR value.There were (a) 6, (b) 15, (c) 16 and (d) 85 cases with missing values for these parameters.(e)A total of 752 patients with complete covariate information were included in the multivariable model.OR: odds ratio.AUC: area under the receiver operating characteristic curve.