The potential impact, cost and cost-effectiveness of tuberculosis interventions - a modelling exercise

Abstract

Background: While a range of interventions exist for tuberculosis prevention, screening, diagnosis, and treatment, their potential population impact and cost-effectiveness are seldom directly compared, or evaluated between settings with different background TB epidemiology and structural drivers.

Methods: We calibrated a deterministic TB model to epidemiological indicators in Brazil, India, and South Africa. We implemented seven interventions across countries focusing on prevention, screening and diagnosis, and treatment of TB, as well as TB screening in prisons in Brazil and nutritional supplementation in India. We standardised scale-up (2025-2030), coverage (80% of target population), and strength of evidence for epidemiological impact using published efficacy data. We estimated epidemiological impact and incremental cost-effectiveness ratios (ICERs), expressed as costs per disability-adjusted life year (DALY) averted by 2050.

Results: Only three interventions prevented >10% of incident TB episodes by 2050: vaccination (median 15-28% across countries), symptom-agnostic community-wide screening (32-38%) and screening in prisons (23%). The impact of other interventions was more limited, ranging from 0% (shortened drug-susceptible treatment) to 5% (nutritional supplementation). ICERs varied widely by intervention and setting. Shortened drug-resistant treatment was cost-saving across settings, with the next lowest ICERs for prison screening in Brazil (72 USD/DALY) and nutritional supplementation in India (167 USD/DALY). Within each country, both low-cost community-wide screening and TB vaccine campaigns had lower USD/DALY than TB preventive treatment.

Conclusion: Interventions with meaningful epidemiological impact can also be cost-effective, but need to target populations beyond clinic-diagnosed individuals or their households. Achieving such potential requires a priority shift in funding, policy and product development.

FIGURE 1:

Epidemiological model structure and subdivisions Model structure for the natural history of TB, incorporating both general subdivisions and country-specific structural determinants: incarceration (Brazil), undernutrition (India), and HIV (South Africa). Background mortality is assumed for all states but not shown; arrows exiting from the corners of certain states represent specific mortality rates. The model is age-stratified in 5-year bands up to 75+ years. TB states shaded in light grey represent individuals harbouring viable Mycobacterium tuberculosis and are further subdivided by drug resistance; this also applies to those recently treated, assuming relapse would involve the same strain. Dark grey TB states appear only in the previously treated stratum within the treatment subdivisions.

FIGURE 2:

Incidence trends for Business as Usual and interventions BAU = Business-As-Usual, Vacc = Vaccination, TPT = Tuberculosis Preventive Treatment, Comm Scr = Community-wide screening, Impr Diag = Improved diagnosis in clinics, DST for all = Drug Susceptibility Testing for all clinic-diagnosed individuals, Short DS = Shortened treatment for Drug Susceptible TB, Short DR = Shortened treatment for Drug Resistant TB, Pri Scr = Mass screening in incarcerated individuals, Nutr = Nutritional support for households of individuals receiving TB treatment. Lines show BAU incidence trends (black dashed line) and intervention trends in incidence for each country. Shading reflects 95% Uncertainty Intervals (UI). Note UIs overlap for most interventions. Error bars show 2023 incidence calibration target. Pop-outs show last five years for incidence curves (without uncertainty) that lie close to the baseline (black dashes line).

FIGURE 3:

Percentage of incident symptomatic TB episodes averted (2025–2050) Vacc = Vaccination, TPT = Tuberculosis Preventive Treatment, Comm Scr = Community-wide screening, Impr Diag = Improved diagnosis in clinics, DST for all = Drug Susceptibility Testing for all clinic-diagnosed individuals, Short DS = Shortened treatment for Drug Susceptible TB, Short DR = Shortened treatment for Drug Resistant TB, Pri Scr = Mass screening in incarcerated individuals, Nutr = Nutritional support for households of individuals receiving TB treatment. BRA = Brazil, IND = India, ZAF = South Africa. Error bars reflect 95% Uncertainty Intervals.

FIGURE 4:

Incremental Cost-Effectiveness Ratios for interventions plots Error bars show 95% uncertainty intervals for DALYs averted (horizontal lines) and incremental budget (vertical lines). Dashed line shows cost-effectiveness thresholds. Vacc = Vaccination, TPT = Tuberculosis Preventive Treatment, Comm Scr = Community-wide screening (low diagnostic cost), Impr Diag = Improved diagnosis in clinics, DST for all = Drug Susceptibility Testing for all clinic-diagnosed individuals, Short DS = Shortened treatment for Drug Susceptible TB, Short DR = Shortened treatment for Drug Resistant TB, Pri Scr = Mass screening in incarcerated individuals, Nutr = Nutritional support for households of individuals receiving TB treatment. Note, see supp mat S4, section 3 for community-wide screening with high cost.