External validation of a treatment decision algorithm for tuberculosis in children living with HIV: a diagnostic...

Tuberculosis is the leading cause of death in children living with HIV, and is challenging to diagnose owing to the limitations of microbiological approaches. The Pediatric Asian African Network for Tuberculosis and HIV Research (PAANTHER) treatment decision algorithm (TDA) was previously developed to improve the diagnosis of tuberculosis in children living with HIV, but has not yet been validated. We aimed to externally validate this TDA in a prospective cohort of children from four countries in Africa.

The TB-Speed HIV study was a prospective diagnostic cohort study conducted in seven tertiary hospitals: two in Côte d'Ivoire, two in Mozambique, one in Uganda, and two in Zambia. Eligible participants were children living with HIV, aged 1 month to 14 years, with presumptive tuberculosis. Children who were currently receiving tuberculosis treatment, or who had received treatment within the past 3 months, were excluded. At enrolment, children were evaluated for suggestive tuberculosis symptoms (fever for >2 weeks, unremitting cough, haemoptysis and/or weight loss in the past 4 weeks, and tachycardia) and underwent Xpert MTB/RIF Ultra testing on respiratory and stool samples, chest radiography, and abdominal ultrasonography. Children were given a PAANTHER TDA score, with a score of less than 100 defined as negative and a score of 100 or higher defined as positive and prompting the initiation of treatment. At the end of the study, children were retrospectively classified by an endpoint review committee as having confirmed, unconfirmed, or unlikely tuberculosis, forming the composite reference standard against which diagnostic accuracy was assessed. The primary outcome was the proportion of children with missed tuberculosis (false negatives) among those who were not initiated on treatment as per the PAANTHER TDA. We assessed the diagnostic accuracy (sensitivity, specificity, and negative and positive predictive values) and feasibility (uptake and time to treatment initiation) of the PAANTHER TDA. The protocol-defined negative predictive value threshold for validation was 75%. This study is registered at ClinicalTrials.gov, NCT04121026, and is complete.

From Oct 2, 2019, to Dec 31, 2021, 1706 children were prescreened for eligibility, of whom 713 children living with HIV were screened for tuberculosis symptoms; 423 were eligible for the study and 277 were enrolled. Of these 277 children, 272 (98%) had a complete TDA evaluation: 215 (79%) had a score of 100 or higher, including 24 (9%) with a positive Xpert MTB/RIF Ultra test. Treatment was initiated in 185 (86%) of 215 children with a positive score and 12 (20%) of 60 children with a negative score, at a median of 1 day [IQR 0-4] after inclusion. After assessment by the endpoint review committee, the proportion of children with tuberculosis in the classifiable study population (n=273) was 56·8% (95% CI 50·9-62·5), with 131 having unconfirmed tuberculosis and 24 having confirmed tuberculosis. Four (7%) of the 60 children with a negative score were subsequently classified as having tuberculosis by the endpoint review committee (missed tuberculosis). The TDA had a negative predictive value of 93·3% (95% CI 84·1-97·4), a positive predictive value of 70·9% (95% CI 64·5-76·6), a sensitivity of 97·4% (95% CI 93·6-99·0), and a specificity of 47·5% (95% CI 38·7-56·4).

The PAANTHER TDA was externally validated in a cohort of children living with HIV as per the study protocol. The high sensitivity of this TDA and its potential to enable rapid treatment initiation could contribute to the reduction of mortality in children living with HIV.

Unitaid.