3.1. Baseline scenarios

Two baseline scenarios were adopted for this analysis. The need for two baseline scenarios reflects uncertainty over the expected short- and long-term impacts of potential disruptions to TB care brought about by the COVID-19 pandemic and the speed of recovery from these disruptions.

The first baseline, or the “standard baseline,” reflects historical experience with TB prior to the COVID-19 pandemic. The disruptions that occurred during the pandemic are assumed to be reversed. Assumed spending in the baseline is relatively constant, ranging between a narrow band of US$ 6.3 to US$ 6.9 billion (constant US$ 2020) between 2023 and 2050. This somewhat reflects historical trends of TB spending over recent years (sources: IHME (2021) and Global Plan, WHO (2022a, b). Figures are reported in constant 2020 US$.).Footnote ⁶ In the short run, TB incidence follows the slow, almost flat trajectory seen since the mid-2010s. Annual absolute changes in incidence are no greater than 0.25% per year, up to 2030. However, with a growing population, funding is unable to keep TB incidence at a steady state, and TB cases and deaths begin to increase steadily until 2050. In 2050, there are an expected 10.8 million cases and 1.6 million deaths (Figures 1 and 2).

In the second baseline, known as the “disruption baseline,” TB service delivery fails to recover to pre-COVID levels. Under this assumption, the model predicts that in 2023, cases are 30% higher than in the standard baseline, around 10.3 million, with deaths at 1.6 million. Due to insufficient service provision, TB cannot be contained, leading to a rapid rise in incidence and TB mortality. By 2030, cases are at 15.5 million with 2.5 million deaths. By 2050, cases are at 20.0 million with 3.6 million deaths. This scenario requires reduced funding being made available to overcome the TB burden. This baseline can be construed as a plausible “worst-case scenario” for TB until the middle of the century.

In this report, we focus more on results using the standard baseline. We also report comparisons to the disruption baseline.

3.2. Intervention scenarios

Two intervention scenarios are considered in this report. The first scenario includes all the elements of the Global Plan, except the new vaccine. This implies the use of existing tools, and some new, to-be-developed tools such as improved diagnostics and medicines. The second scenario includes vaccine research and deployment.

The intervention scenarios were constructed to ensure key targets were met. These include

1. (i) At least 95% of people with TB will receive a TB diagnosis.

2. (ii) All high-risk and key and vulnerable populations will be able to access periodic screening.

3. (iii) 50 million people will access appropriate TB treatment, including 4.7 million children and 3.32 million people with drug-resistant (DR-) TB. 35 million people will access TB preventive treatment (TPT).

4. (iv) At least one new TB vaccine will be introduced for widespread use by 2026 (vaccine scenario only).

The TB Impact and Estimates (TIME) model (Houben et al., Reference Houben, Lalli, Sumner, Hamilton, Pedrazzoli, Bonsu and Hippner2016) and supplementary modeling work were used to estimate the epidemiological impact of meeting the above targets for each country.

Regarding screening and treatment, screening rates were increased in an S-shaped curve starting in 2023 and ending in 2030. Existing tools were explicitly modeled in TIME predominantly using X-ray screening and rapid molecular tests, generating a sensitivity and specificity of 84.8% and 99.7%, respectively, for systematic screening of household contacts and high-risk groups. For people with HIV who are newly enrolled on antiretroviral therapy (ART) and those who are already on ART, sensitivity was set at 72% and 65% and specificity at 98% and 97%, respectively. Treatment success was increased in the model from 2019 levels (carried to 2023) to 90% by 2030.

The Global Plan 2023–2030 continues the focus of the previous Global Plan 2018–2022 on TPT, calling for 100% coverage of contact tracing (for finding TB and offering TPT) in the household of all people diagnosed with bacteriologically positive TB by 2022 and onward. Furthermore, it is assumed that all new people taking ART for HIV/AIDS and those already on ART and eligible for TPT will receive TPT. Estimates for the distribution of active and latent TB in adults and children in households of index cases were based on Fox et al. (Reference Fox, Barry, Britton and Marks2013). Household size estimates and the percentage of the household under 5 years of age were based on demographic health surveys where available, and a global average was used where not (household size of five and 15% of household members under the age of 5). ART cohort sizes were estimated using the Spectrum AIM model used annually to produce estimates for the UNAIDS Annual Global Report on the AIDS pandemic, among other purposes.

The TIME model does not directly model the finding and treatment of subclinical TB or TB prevention through large-scale vaccine programs. Insights into the additional impact of these “new” tools, when added to a program implementing existing tools at full scale, were obtained via supplementary modeling work. This supplementary modeling conducted more detailed analysis of intervention combinations in four focus countries: Indonesia, Ukraine, Uzbekistan, and Kenya, to provide estimates of impacts from new tools. The results suggest that these tools can lead to the achievement of the 2030 impact milestones.Footnote ⁷ Coverage scales up gradually reaching 30% for treating subclinical TB and 60% for a post-exposure TB vaccine by 2030.

The epidemiological impacts in terms of incident cases and deaths from each scenario are depicted in Figures 1 and 2, respectively. The Global Plan sees a rapid decrease in both cases and deaths over the 8 years from 2023 to 2030. In 2030, cases are expected to fall to 2.1 million (with vaccine) or 3.1 million (without vaccine). Deaths are expected to fall substantially to 145,000 (with vaccine) or 197,000 (without vaccine) in the same year. From 2030 to 2050; the incidence and mortality profile of the Global Plan without vaccine essentially stays in a steady state. The Global Plan with vaccine continues a gradual reduction of incident cases and deaths such that in 2050 cases are predicted to be 196,000 with 46,000 deaths.

4.1. Costs of service delivery

In a departure from previous Global Plan exercises, the 2023–2030 Global Plan adopts a “normative approach” to TB treatment, meaning that the projected implementation of tools (e.g., diagnostics, medicines) and services (e.g., patient support) is consistent with current and anticipated international guidelines. This approach has allowed for more detailed and complete projections of resource needs with service delivery costs estimated using a bottom-up, ingredients-based approach. Seven types of costs were estimated: diagnosis, treatment, prevention, health systems, enablers, programs costs, and vaccination costs, with methods differing by type of cost.

Diagnosis, treatment, and prevention costs were estimated by first establishing common TB services including case finding, diagnosis, and treatment. The services are:

1. (i) Passive case finding, pulmonary TB

2. (ii) Passive case finding, extrapulmonary TB

3. (iii) Systematic screening, household and close contacts

4. (iv) Systematic screening, preventative treatment for household and close contacts

5. (v) Systematic screening, people living with HIV

6. (vi) Systematic screening, other key and vulnerable population groups

7. (vii) Detection of drug resistance and comorbidities (for all patients diagnosed through passive and systemic screening)

8. (viii) TB treatment regimens and services, 6-month regimen consisting of isoniazid, rifampin, pyrazinamide and ethambutol for two months and then isoniazid plus rifampin for four months (2HRZE/4HR).

9. (ix) TB treatment regimens and services, 4-month regimen consisting of rifapentine-moxifloxacin regimen for the treatment of drug-susceptible pulmonary TB. (RPT-Mox)

10. (x) TB treatment regimens and services, rifampicin-susceptible, isoniazid-resistant TB

11. (xi) TB treatment regimens and services, all-oral shorter regimen

12. (xii) TB treatment regimens and services, Bedaquiline + Pretomanid + Linezolid (BPaL) regimen

13. (xiii) TB treatment regimens and services, 18–24-month drug-resistant TB regimen

14. (xiv) TB treatment regimens and services, delamanid-based regimen (children only)

Next, the primary cost drivers of each service were established including composite interventions (e.g., clinical assessment, X-ray, sputum transportation, etc.), the staff time required to deliver the service, and health system costs for in-patient and out-patient care. Fifty-four separate interventions were identified and costed as part of this exercise. Moreover, service profiles varied according to age, pulmonary status, HIV status, MDR status, and passive versus active TB of patients. Detailed descriptions of these service profiles are reported in the Global Plan appendix (Stop TB Partnership, 2022).

Unit costs for the 54 interventions were sourced from the Value TB project (Sweeney et al., Reference Sweeney2021). Detailed data were available for five countries: Ethiopia, Georgia, India, Kenya, and Philippines. Unit cost data were then extrapolated to other countries by first determining how much of each profile represented tradeable goods, non-tradeable goods, and labor. The share of tradeable goods was converted using market exchange rates or directly sourced from the latest prices available in relevant procurement catalogues. Non-tradeable goods were transferred to target countries using purchasing-power parity exchange rates. Labor costs were converted using ratios of GDP per capita between the target and reference country (Serje et al., Reference Serje2018).

For the cost transfer approach:

1. (i) Georgia was used as a reference for upper–middle-income TB burden countries.

2. (ii) India was used as a reference for lower–middle-income high TB burden countries in South Asia.

3. (iii) The Philippines was used as a reference for middle-income high TB burden countries in the Western Pacific Region.

4. (iv) Kenya was used as a reference for middle-income high TB burden countries in Africa.

5. (v) Ethiopia was used as a reference for lower-income high TB burden countries.

Once unit costs were established for each country, costs were applied to expected coverage levels under the intervention scenarios. If costs were unavailable from Value TB (primarily treatment costs), a constant parameter value was assumed for all countries sourced from procurement catalogues such as Stop TB Partnership’s Global Drug Facility.Footnote ⁸ Assumptions used in the costing analysis and the methodological approach for each assumption are presented in Table 1.

Table 1. Unit cost values used across 54 interventions in TB modeling in US$.

Note: “Value TB” means that the parameter was converted from the Value TB database using the transfer method described in the text. “Constant” means that a single parameter was used for all countries. “WHO” means that the figures were sourced from the WHO CHOICE model.2HRZE/4HR describes the treatment regimen containing isoniazid, rifampin, pyrazinamide and ethambutol for two months/isoniazid plus rifampin for four months. 4 RPT-Mox describes the four-month rifapentine-moxifloxacin regimen for the treatment of drug-susceptible pulmonary TB. 3 HP describes the regimen for treatment of latent TB infection, consisting of weekly doses of isoniazid and rifapentine for three months.

Health systems costs represent facility-level costs required for in- and out-patient visits across each intervention. Unit costs were sourced from WHO-CHOICE database and are applied to requirements for each service as detailed in the Global Plan appendix (Stop TB Partnership, 2022).

Enablers comprise specific “enabling” activities, including direct patient support (5%), advocacy and communications (1%), community rights and gender activities (6%), and public–private management (12% for countries with a high degree of private healthcare sector presence). Enabling costs were estimated as percentage markups on prevention, diagnosis, treatment, and health system costs and were based on the detailed budgets of the Democratic Republic of the Congo, Georgia, India, Philippines, and Tajikistan, which were judged to be representative in terms of budgeting for enabling activities.

Program costs consist of above patient level costs and were also estimated as a percentage mark up on direct services costs. The markup value was based on average expenditure data reported to the WHO and is equal to 70%.

Lastly, vaccination costs are assumed to equal US$ 6 per dose (US$ 4 for vaccination and US$ 2 for delivery), requiring two doses per person.

4.2. Total costs of the Global Plan

The cost profile of the two different baseline scenarios and two different intervention scenarios is presented in Figure 3.Footnote ⁹

Figure 3. Cost profile of baseline and intervention scenarios, LLMCs. Sources: IHME (2021) and Global Plan, WHO (2022a, b). Figures are reported in constant 2020 US$. OOP = out of pocket expenses

The baseline requires spending of US$ 6.5 billion over the period of analysis. The disruption baseline assumes US$ 3.0–3.3 billion in spending every year.

The two intervention scenario cost profiles require substantial increases in spending US$ 10.4 billion initially, rising to US$ 12.7 billion by 2026. In 2027, when the new vaccine is assumed to be ready and deployed, the costs for the Global Plan with vaccine rise sharply for 4 years to around US$ 18-US$ 19 billion, reflecting the initial rollout to an unvaccinated population. From 2031 onwards, both profiles decline gradually, although they are still several billion dollars more than baseline spending. The vaccine plan is slightly more expensive than the non-vaccine plan after 2031 to account for vaccinating the new cohorts of unvaccinated Figure 3.

4.3. Benefits

In this cost-benefit analysis, only averted deaths are incorporated as benefits, not averted cases. In a previous cos-benefit analysis, we note that averted mortality comprised nearly all of the benefits (Rudman et al., Reference Rudman, Adusi-Poku, Razvi, Wong and Houben2020). Averted mortality is valued using a standardized approach across all Halftime SDG papers that follow the recommendations of Robinson et al. (Reference Robinson, Hammitt, Cecchini, Chalkidou, Claxton, Cropper and Eozenou2019). Moreover, we do not include other important but difficult to quantify benefits related to promoting partnerships, the reduction of stigma associated with TB, and spillover research benefits beyond the TB sector.

To estimate the value of averted mortality, we take a reference value for the U.S. VSL of US$ 9.4 million (2015 dollars), representing approximately 160 times income as measured by income per capita PPP. This is transferred to the entire low- and lower–middle-income population via the ratio of GDP per capita, using an income elasticity of 1.5.

To estimate these values, we take the population-weighted PPP GDP per capita figure in 2020 US$ for the group of LLMCs and the United States of America and estimate the VSL at time t = 0, 2020.

$$ {VSL}_t={\left(\frac{PPP\hskip0.35em GDP\hskip0.35em {pc}_{LLMC,t}}{PPP\hskip0.35em GDP\hskip0.35em {pc}_{USA,t}}\right)}^{e-1}\ast \hskip0.35em 160\hskip0.35em \ast \hskip0.35em GDP\hskip0.35em {pc}_{LLMC,t}\hskip0.35em $$

Following Cropper et al. (Reference Cropper, Guttikunda, Jawahar, Lazri, Malik, Song and Yao2019), we estimate each subsequent VSL in the time series according to the following formula:

$$ {VSL}_{t+1}={VSL}_t\hskip0.35em \ast \hskip0.35em {\left(1+{g}_t\right)}^e $$

where g_t is the real GDP per capita growth rate between period t and t + 1 (SSP Database, IIASA GDP Model, Scenario SSP2_v9_130219) and e = 1.5.

The GDP growth in this group of countries outpaces the population growth, so that the VSL grows rapidly over time. In constant 2020 US$ values, the benefit of an averted death is US$ 98,700 (2020), US$ 149,800 (2025), US$ 212,000 (2030), US$ 276,300 (2035), US$ 338,100 (2040), US$ 396,800 (2045), and US$ 456,000 (2050).

4.4. Cost–benefit analysis results

Results of the cost–benefit analysis are presented in Table 2. We present incremental costs, incremental benefits, and BCRs for both scenarios relative to the two baselines. In all specifications, BCRs are high, and the number of averted deaths is large.

Table 2. Benefits, costs, and BCRs of the Global Plan, 2023–2050.

Note: Incremental costs and benefits are discounted at 8%. BCR = benefit-cost ratio

In the main specification (without vaccines, standard baseline), incremental costs equal US$ 56.4 billion over 2023–2050 while incremental benefits are a substantial US$ 2,595 billion. The BCR is 46.0. Deaths averted are estimated at 27.3 million over 28 years, at a cost per death averted of US$ 2063. With vaccines, estimated costs are 32% higher, while benefits are 7% higher. The BCR is 37.4 with a cost per death averted of US$ 2520. Comparisons to disruption baseline yield higher costs and substantially higher benefits. BCRs are 71.0 and 61.4 without and with vaccine development and deployment, respectively.

4.5. Sensitivity analyses

In this section, we generate several alternative specifications of costs and benefits to test the sensitivity of results against the underlying assumptions.

We consider four sensitivity analyses:

1. (i) Including the costs of R&D into the analysis

2. (ii) Reducing the Global Plan’s costs beyond 2035 for the vaccine scenario

3. (iii) Increasing the discount rate to 12%

4. (iv) Decreasing the discount rate to 5%

Results of these sensitivity analyses are presented in Table 3. Across all analyses, BCRs remain very high, ranging between 28 and 84.

Table 3. Benefit–cost ratios from sensitivity analyses.

4.6. Including the costs of R&D

The Global Plan includes a substantial funding request for the costs of R&D. We have chosen not to include these costs in the main cost-benefit analysis under the assumption that the costs of R&D will be embedded in the price of the new tools. Including R&D costs likely represents double counting of costs, in the same way that including, for example, the cost of pharmaceutical companies’ marketing or employee wages in addition to the price of the new tools would represent double counting. Nevertheless, given that R&D is specifically highlighted in the Global Plan, we demonstrate the impact on BCRs from including R&D costs.

Note that the benefits of R&D include an improvement in the efficiency and/or effectiveness of TB management. These have been incorporated into the impact calculations noted previously.

Costs of R&D for TB were estimated by the Stop TB Partnership New Tools Working Group and reported in the Global Plan. Total R&D costs for vaccines, diagnostics, treatments, and basic research are estimated at almost US$ 40 billion over the period 2023–2030 (Figure 4).

Figure 4. R&D costs for TB over the period 2023–2030, billions of US$. Source: Global Plan to End TB 2023–2030.

The Global Plan does not specify the time profile of these investments. Moreover, these costs require attribution to LLMCs and our two main scenarios (with and without vaccines). While most TB R&D costs will likely be borne by high-income countries, we attribute these to LLMCs for the purpose of sensitivity analysis and on the basis that these will likely come from budgets earmarked for global health. For the purposes of the cost-benefit analysis, we assume that:

1. (i) All R&D costs are incurred equally over 4 years in 2023, 2024, 2025, and 2026. This ensures all necessary technologies are ready for deployment in 2027, the first year in which vaccines are included as part of the activity profile.

2. (ii) Half of the basic research costs and all medicines and diagnosis R&D costs are attributed to the Global Plan scenario without vaccines.

3. (iii) Of these R&D costs, the relevant share for LLMCs is based on their expected case numbers as a share of global case numbers over the period 2027–2050. This equals 83%.

4. (iv) Half of the basic research costs and all vaccine R&D costs are attributed to vaccines scenario.

5. (v) Of the global vaccine R&D costs, costs are apportioned to LLMCs based on their share of global population in 2025–2030 as estimated by the UN. This is done on the basis that vaccine activity and service delivery is highly tied to population. The share attributed to LLMCs is 53%.

Including R&D costs reduces the BCRs by 17–26% depending on the combination of baseline and intervention scenarios. The BCR of our preferred specification (Global Plan without vaccines against standard baseline) is 34.0 with R&D costs included.

4.7. Lower intervention spending beyond 2035 for the vaccine scenario

In the Global Plan’s vaccine scenario, cases and deaths fall to very low levels by 2030. However, funding requirements remain high, more than US$ 9 billion annually. While the funding requirements per case are relatively stable for the baseline and without vaccine intervention scenario across the outer years, 2035–2050,Footnote ¹⁰ the Global Plan with vaccine scenario sees an increase in the cost per case from US$ 11,334 in 2035, rising to US$ 48,625 in 2050.

This potentially overestimates the required funding for that scenario. Therefore, we consider an alternative scenario where the cost per case is fixed at US$ 11,334 across the entire period 2035–2050. The new cost profile for this sensitivity analysis is presented in Figure 5. Costs fall with incidence such that counterfactual costs in the standard baseline are higher after 2041.

Figure 5. Alternative cost profile for Global Plan with vaccine scenario.

The BCR of this scenario is 46.3 compared to the standard baseline (Table 3), essentially the same BCR as the Global Plan without vaccine (46.0). Compared to the disruption baseline, the BCR with the new cost profile is 70.1.

4.8. Changing discount rates

As a final sensitivity analysis, we alter the discount rate to 5% and 12%. As expected, the BCR of our preferred specification rises with the lower discount rate to 55.4 and falls with the higher discount rate to 36.6.