New algorithms could cut childhood TB costs at primary care

Diagnosing tuberculosis (TB) in children is difficult, and delays or missed cases can have serious consequences. To address this, researchers led by Claudia M. Denkinger evaluated new treatment-decision algorithms (TDAs) designed for children under age 10, some that use chest X-ray (CXR) and some that do not. Their goal was practical: to estimate how well these TDAs would work and how much they would cost if rolled out in Uganda. The team focused on two common care levels where children are first seen: primary health care centres (PHC) and district hospitals (DH). Instead of running a field trial, they built decision models that follow the pathway from clinical evaluation to the decision to treat, allowing them to compare different combinations of tests and care pathways. By modelling a large group of children with suspected TB, the researchers could compare accuracy and costs for realistic program choices and provide information useful for health planners deciding where and how to use these algorithms.

The researchers used decision-tree models following the TDA pathway and compared six scenarios that combined different diagnostic tests at PHCs and DHs. The tests considered included stool and respiratory Xpert, urine lipoarabinomannan, and/or CXR. They evaluated outcomes for a cohort of 10,000 children with presumptive TB, using a Monte Carlo simulation from a health system perspective and reporting costs in 2024 International dollars. Across all scenarios the TDAs showed high sensitivity (80.8–91.9%) but relatively low specificity (51.2–60.9%), meaning they were good at detecting true TB cases but produced many false positives. Total diagnostic and treatment costs for the cohort ranged from I$1,768,958 to I$2,458,790, with much of the expense driven by overtreatment of false-positive cases. Diagnostic spending often offset some overtreatment, and the cost per correct treatment-decision was lowest when a mobile CXR was available at PHC (I$287.40) and highest when children were referred to DHs (I$445.84).

The study suggests practical trade-offs for TB programs: the TDAs are reliable at finding most childhood TB cases and can be implemented at primary health centres, where they may be less costly overall than relying on district hospital referral. However, the relatively low specificity means many children without TB could receive unnecessary treatment, which is a major driver of total costs. The authors point to two clear ways to make wide implementation more affordable: improve specificity of the algorithms so fewer false positives are treated, and reduce the costs associated with treatment of false-positive cases. In short, using TDAs at the primary care level appears promising for increasing case detection and managing costs, but attention to diagnostic accuracy and treatment expenses will determine whether programs can scale these approaches sustainably.