A data-driven approach to identify putative tuberculosis among adolescents in resource-constrained settings

Abstract

Introduction: There has been recent evidence of low rates of tuberculosis (TB) testing in Southern Africa, and lack of evidence age-disaggregated data on pulmonary TB rates to identify TB outcomes among adolescents.

Aim/objective: To evaluate the potential of self-reported tuberculosis (TB) clinic diagnosis and self-reported TB symptoms for estimating TB rates among adolescents living with HIV in a resource-constrained setting in South Africa.

Methods: This study is based on data from three waves of a longitudinal study cohort of 1 060 ALHIV, ART-initiated adolescents in Eastern Cape province of South Africa. First, we use latent class analysis (LCA) and group-based trajectory modelling (GBTM) to model TB symptom trajectory groups based on past-year self-reported TB symptoms. Second, we then use multinomial logistic regression to assess the association between TB symptom trajectory groups and self-reported TB test results.

Results: We identified three long-term TB symptomology groups namely: high TB symptomatic (16.6%), decreasing symptomatic (23.3%), and asymptomatic (60.1%). Both symptomatic groups were significantly associated with self-reported TB diagnosis. Baseline self-reported positive TB diagnosis (OR2.02 95%CI 1.42-2.89, p<0.001) and HIV-positive status (OR1.94 95%CI 1.31-2.87, p<0.001) were significantly associated with high TB symptomatic group controlling for age, gender, and rural residence. These findings were consistent over time.

Conclusion: Self-reported symptom screening is useful for ruling out TB and identify cases in need of further diagnostic assessment for TB. Given high rates of TB symptomatic ALHIV better diagnostics are urgently needed.