Care-cascade failures as a determinant of infectious burden in tuberculosis: an agent-based modelling study in India

Background. Despite effective treatment, care failures can prolong tuberculosis (TB) infectiousness, sustain transmission and hinder progress towards End TB targets. Counts of patients diagnosed and treated obscure these delays. We examined how care-cascade failures shape infectious burden and intervention impact in India. Methods. We used an existing EMOD-TBHIV agent-based modelling platform for India, calibrated the model to national incidence, prevalence and mortality, and assessed agreement with published care-cascade benchmarks. We classified simulated TB disease episodes by diagnostic failures (false-negative tests or pre-treatment default), treatment failure, both, or neither. We quantified infectiousness-weighted person-time across presymptomatic disease, patient delay and health-system delay. We projected effects of active case finding, improved diagnostic sensitivity and treatment support from 2025 to 2035, individually and combined, and examined sensitivity across 20 calibrated parameter sets. Findings. Among 25,894 simulated TB episodes beginning from 2015 to 2024, 39% had neither recorded failure type, 26% had diagnostic failures only and 35% had treatment failure, with or without diagnostic failures. Median effective infectious time was 3 months among episodes with neither failure type versus 16 months among those with treatment failure. The latter accounted for 63% of infectiousness-weighted burden. Overall, 64% of active TB person-time accrued after the first diagnostic test. Treatment support and active case finding reduced 2035 incidence relative to baseline by 18% (95% CI 15 to 21) and 17% (13 to 20), respectively; Diagnostic improvement reduced incidence by 3% (-1 to 6), and the combined package by 34% (31 to 36). Relative to 2015, the package reduced prevalence by 85% and deaths by 89% but incidence by only 57%, short of the 90% End TB target in every parameter set. Interpretation. Repeated care failures prolonged infectiousness among patients already in care, so that supporting treatment completion reduced transmission as much as finding new cases. Monitoring their frequency and duration alongside conventional cascade indicators could identify opportunities to shorten infectious time and inform programme priorities.