Abstract

Background: Tuberculosis (TB) is a bacterial disease caused by Mycobacterium tuberculosis, resulting in an estimated one million new cases and over 200,000 deaths annually among children. TB is the leading cause of death in HIV-infected children globally (Swaminathan & Rekha, 2010), but few studies have evaluated cofactors of pediatric TB mortality the era ART. We evaluated predictors of TB mortality in a cohort of HIV-infected hospitalized Kenyan children initiating antiretroviral therapy (ART).

Methods: HIV-infected children age(PUSH) trial. Children were ART-naïve and started ART within 2 weeks of enrollment. All children underwent intensified TB case finding at enrollment and were evaluated for TB with symptom screening, physical exam and microbiologic evaluation (two sputum or gastric aspirate samples for AFB, Xpert and culture and one stool Xpert). Children with suspected tuberculosis were treated by hospital clinicians according to Kenyan Ministry of Health guidelines and were followed for six months. We evaluated cofactors of mortality using Kaplan-Meier curves and univariate and multivariate Cox proportional hazard models.

Results: Of 181 ART-naive children enrolled in the study, 14 (8%) had confirmed TB, 81 (45%) had unconfirmed TB, and 86 (47%) had unlikely TB). Overall, mortality was higher among children with confirmed TB compared children with Unlikely TB [HR 3.9, 95% CI 1.50 – 9.97). In multivariate analysis of children with confirmed and unconfirmed TB, higher mortality was observed among participants without anti-TB treatment (aHR 6.5; 95% CI 2.24–18.84; p0.378 (aHR 4·5; 1.50 - 13.81; p=0.008).

Conclusion: We observed high mortality among hospitalized HIV-infected children with confirmed TB. Lack of anti-tuberculosis treatment, high monocyte-to-lymphocyte ratio, and OVC status were significant predictors of TB mortality. Earlier identification and treatment of TB/HIV co-infection is urgently needed.