Predictors of mortality in Ugandan children with TB, 2016-2021.

BACKGROUND

The burden of pediatric TB is high in Uganda. Our objective was to evaluate predictors of mortality during TB treatment among children at an urban and a rural referral hospital.

METHODS

We designed a historical cohort study of TB cases at Mulago National Referral Hospital, Kampala; and Fort Portal Regional Referral Hospital, Fort Portal, Uganda, in children aged <15 years from 2016 to 2021. We used Kaplan-Meier models to estimate survival and fit multivariable Cox regression models to determine mortality hazards during TB treatment.

RESULTS

We identified 1,658 children diagnosed with TB from 2016 to 2021. Of 1,623 children with known treatment outcomes, 127/1,623 (7.8%) died during TB treatment, 1,298/1,623 (78.3%) completed treatment, 150/1,623 (9.2%) were lost to follow-up, and two children failed treatment. Using Kaplan-Meier functions, the median time to death was 27 days following treatment initiation. In adjusted Cox models, predictors of mortality included HIV (aHR 1.68, 95% CI 1.01-2.81), moderate malnutrition (aHR 2.22, 95% CI 1.18-4.16), and severe malnutrition (aHR 2.92, 95% CI 1.75-4.87).

CONCLUSION

Mortality was high at an urban and a rural referral hospital among children who initiated TB treatment from 2016 to 2021, with the majority of deaths occurring during the intensive phase of TB treatment. Malnutrition and HIV were significant predictors of death during treatment.