High mortality risk among individuals assumed to be TB‐negative can be predicted using a simple test

Abstract

To determine mortality among assumed TB negative (aTBneg) individuals in Guinea-Bissau and to investigate whether plasma levels of soluble urokinase receptor (suPAR) can be used to determine post-consultation mortality risk. This prospective West-African cohort study included 1007 aTBneg individuals who were enrolled from 2004 to 2006; 4983 age-matched controls were followed for comparison. Plasma suPAR levels were measured using the suPARnostic ELISA. Survival was analysed using Cox regression, ROC curves and Kaplan-Meier analysis. After 3 months of follow-up, mortality was 21 per 100 person-year-observation (PYO) among aTBneg individuals and three per 100 PYO among the control population [mortality rate ratio (MRR) = 6.92 (95% CI 4.48-10.7)]. SuPAR values ranged between 0.9 and 45 ng/ml in aTBneg individuals. A log-linear relationship was found between suPAR levels <15 ng/ml and mortality. In the log-linear range, a 1 ng/ml increase was associated with a 46% increase in the mortality rate: MRR = 1.46 (95% CI 1.34-1.59). The area under the ROC curves was 0.88 for HIV-positive individuals and 0.79 for HIV-negative individuals. Our study showed a high mortality rate among aTBneg individuals and demonstrated that suPAR measurements can provide prognostic information on mortality among individuals without disease diagnosis. Measuring suPAR is a technically simple method for determining mortality risk in individuals that are assumed to be TB-negative.