Economic evaluation of integrating nutritional support intervention in India’s National Tuberculosis Elimination Programme: implications for low-income and middle-income countries

Abstract

Objectives

This study aimed to evaluate the cost-effectiveness of integrating nutritional support into India’s National Tuberculosis Elimination Programme (NTEP) using the MUKTI initiative.

Design

Economic evaluation.

Setting

Primary data on the cost of delivering healthcare services, out-of-pocket expenditure and health-related quality of life among patients with tuberculosis (TB) were collected from Dhar district of Madhya Pradesh, India.

Intervention

Integration of nutritional support (MUKTI initiative) into the NTEP of India.

Control

Routine standard of care in the NTEP of India.

Primary outcome measure

Incremental cost per quality-adjusted life year (QALY) gained.

Methods

A mathematical model, combining a Markov model and a compartmental susceptible–infected–recovered model, was used to simulate outcomes for patients with pulmonary TB under NTEP and MUKTI protocols. Primary data collected from 2615 patients with TB, supplemented with estimates from published literature, were used to model progression of disease, treatment outcomes and community transmission dynamics over a 2-year time horizon. Health-related quality of life was assessed using the EuroQol 5-Dimension 5-Level scale. Costs to the health system and out-of-pocket expenditures were included. A multivariable probabilistic sensitivity analysis was undertaken to estimate the effect of joint parameter uncertainty. A scenario analysis explored outcomes without considering community transmission. Results are presented based on health-system and abridged societal perspectives.

Results

Over 2 years, patients in the NTEP plus MUKTI programme had higher life years (1.693 vs 1.622) and QALYs (1.357 vs 1.294) than those in NTEP alone, with increased health system costs (₹11 538 vs ₹6807 (US$139 vs US$82)). Incremental cost per life year gained and QALY gained were ₹67 164 (US$809) and ₹76 306 (US$919), respectively. At the per capita gross domestic product threshold of ₹161 500 (US$1946) for India, the MUKTI programme had a 99.9% probability of being cost-effective but exceeded the threshold when excluding community transmission.

Conclusion

The findings highlight the potential benefits of a cost-effective, holistic approach that addresses socio-economic determinants such as nutrition. Reduction in community transmission is the driver of cost-effectiveness of nutritional interventions in patients with TB.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study employed a hybrid modelling framework that combined a patient-level Markov model with a population-level susceptible–infected–recovered transmission model to simulate both direct and indirect effects of nutritional support on tuberculosis (TB) outcomes in India.

• The study uses primary data on the cost of delivering healthcare services as well as on health-related quality of life and out-of-pocket expenditure from patients with TB, strengthening the contextual validity of model inputs.

• The model uses India-specific EuroQol 5-Dimension 5-Level value set to generate quality adjusted life years (QALYs), in alignment with national health economic evaluation guidelines.

• The use of the susceptible–infected–recovered model rather than the susceptible–exposed–infected–recovered may limit the capture of long-term transmission dynamics and latent disease progression, although sensitivity analyses were conducted to test this assumption.

• The analysis excludes measurement of nutritional supplementation’s effect on extrapulmonary TB cases. However, as extrapulmonary TB cases represent only 15% of the total TB cases, the exclusion might not exert a substantial influence on the observed findings.

Introduction

Despite notable progress in the field of tuberculosis (TB) control, the toll of TB on public health remains substantial. India bears the greatest share of this burden, contributing around 25% of global TB cases and 38% of global TB-related mortality in 2021.1,3 The National Strategic Plan for Tuberculosis Elimination in India aims to achieve an 80% decrease in TB incidence and a 90% reduction in TB mortality by 2025, in comparison with the baseline estimates recorded in 2015.⁴ It necessitates a comprehensive approach towards TB control, with an augmented emphasis on its social determinants.

One such social determinant is undernutrition, as it is a common comorbidity of TB in low- and middle-income countries (LMICs) and plays a pivotal role in the trajectory of TB control and management.⁵ Undernutrition is a critical determinant of the vulnerability to TB infection as well as the success of TB treatment because it is a consistent risk factor for mortality, drug toxicity, delayed sputum conversion and recurrence.6,8 The vicious interplay between nutrition and TB underscores the importance of integrating nutritional support interventions into the National Tuberculosis Elimination Programme (NTEP) to enhance its effectiveness. Such an intervention becomes all the more important in the Indian context, wherein 19.6% of the adult men and 22.4% of the adult women are underweight for their age, and there is a high prevalence of moderate to severe undernutrition in both men and women with active TB across India.^{9 10}

The systematic integration of nutritional support into India’s NTEP seeks to address several key issues (figure 1). First, it aims to improve the nutritional status of patients with TB, enhancing their ability to tolerate anti-TB medications, thus potentially improving the rate of sputum conversion.¹¹ Earlier sputum conversion also diminishes the extent of disease transmission to others, since patients with TB tend to remain in the infectious stage for a shorter duration now. Second, nutritional support diminishes the likelihood of relapse in patients with TB, which not only improves the health outcomes but also reduces the number of secondary infections caused by relapsed cases.¹² Finally, these types of interventions serve to enhance the synergy and interaction between the healthcare system and patients, fostering a more robust and collaborative therapeutic relationship.¹³ This, in turn, augments patient adherence to treatment protocols, resulting in improved compliance and ultimately translating into more favourable health outcomes. Evidence from India has also established a strong association between TB and malnutrition.^{9 14 15} It has been demonstrated that nearly half of the active TB cases in India are attributable to malnutrition.^{10 16} Studies have indicated that reducing malnutrition could potentially prevent up to 70% of TB-related deaths in India.^{17 18} Therefore, efforts to combat TB in India must be integrated with strategies to address hunger and malnutrition.

Figure 1. Causal mediation pathway demonstrating the effect of nutritional support in tuberculosis control. ATD, anti tubercular drugs; NTEP, National Tuberculosis Elimination Programme; QALY, quality-adjusted life year; QoL, quality of life; TB, tuberculosis.

Despite all the evident benefits of improved TB treatment outcomes and reduced transmission, the economic aspects of integrating nutritional support interventions into NTEP are a critical concern given the resource constraints of India’s healthcare system. Recent studies have begun to shed light on the economic implications of nutritional support in the context of TB control in India, and our work seeks to contribute meaningfully to this emerging evidence base. Recent model-based estimates have shown that providing nutritional support to 50% of adults on TB treatment and their households could avert approximately 4·6% of TB deaths and 2·2% of TB episodes in India.¹⁹ Likewise, another recent Markov state transition model simulated TB incidence, treatment and TB-attributable mortality among household contacts receiving nutritional intervention and demonstrated that it is a cost-effective approach from both government and societal perspectives in India.²⁰ Earlier attempts to evaluate the cost-effectiveness of nutritional support in the context of TB control in India have also highlighted the potential of improving the nutritional status of undernourished individuals to reduce subsequent TB cases and deaths, although they had assessed the effect of providing nutritional support to undernourished individuals not infected by TB and did not factor in the effect of community transmission.²¹ These studies have made important contributions to understanding the broader effect of nutrition on TB outcomes and provide a strong foundation for future research and policy dialogue.

This study complements and builds on these efforts by focusing specifically on the integration of nutritional support within India’s NTEP, evaluating its cost-effectiveness from both health system and societal perspectives. This study seeks to leverage the primary data to explore the complex interplay between nutrition and TB in the Indian context by evaluating the cost-effectiveness of integrating nutritional support interventions for patients with TB in the NTEP. This shift in focus aims to ascertain the effect of nutritional support within the targeted population by studying the effect of differing cure rates, relapse rates and community transmission. The study aimed to highlight the economic considerations associated with the integration of nutritional support into the NTEP (MUKTI initiative), thereby augmenting the understanding of effective strategies in the context of TB management. The initiative is named MUKTI (a Hindi language word meaning ‘liberation’ or ‘freedom’), as it symbolises the effort to free patients from the burden of TB. Specifically, this study evaluates the incremental cost per quality-adjusted life year (QALY) gained by implementing the package of interventions as part of the MUKTI initiative alongside the NTEP, as compared with routine standards of care under NTEP alone. Taken together, these complementary studies help advance the policy conversation around holistic and cost-effective strategies to strengthen TB care in India and other similar settings.

Methods

Conceptual framework

We considered a patient recently diagnosed with pulmonary TB who could potentially receive treatment either according to the existing standards of care outlined in the NTEP or be enrolled in the MUKTI programme (details of which are elaborated in the subsequent section). The treatment process is envisioned to induce changes in the patient’s sputum status and/or nutritional status (figure 2). Notably, differences between the two treatment arms are anticipated in terms of the recovery and relapse rates. Furthermore, differences are also expected to occur in the costs borne by the health system, out-of-pocket (OOP) expenditure incurred by the patient, and the effect of altered nutrition on the patient’s health-related quality of life. The differences between the time to become sputum-negative may also affect the duration over which a patient with TB remains infectious. This temporal variance carries implications for community transmission dynamics since a more extended period of infectivity increases the potential for the spread of infection to close contacts. Our evaluation of the cost-effectiveness of the MUKTI intervention involves a comparison of costs and health outcomes associated with the NTEP and comparisons of these with those of the MUKTI programme, thus providing a comprehensive perspective on the potential health effects and economic implications of both treatment arms. Notably, figure 2 represents the modelled conceptual framework, and figure 1 is included for conceptual completeness and policy relevance, as not all pathways depicted in figure 1 were intended to be modelled in the economic evaluation.

Figure 2. Conceptual framework for assessment of cost-effectiveness of integrating nutritional support intervention in NTEP. NTEP, National Tuberculosis Elimination Programme; OOP, out-of-pocket; TB, tuberculosis.

Treatment arms: control and intervention

In the control arm, the patients were assumed to receive treatment under current standards of care as specified in NTEP.²² In addition to the current standards of care, four components were implemented as part of the MUKTI programme for patients with TB in the intervention arm.²³ These included counselling support through home visits, provision of locally procured protein-rich food baskets to patients, group-based community sessions (positive deviance (PD) sessions) for enabling peer-to-peer learning and facilitation of linkages to nutrition support schemes provided by the government, including the direct benefit transfer scheme. Cluster coordinators of the MUKTI programme made monthly home visits to the patients and counselled them on the importance of treatment compliance and nutrition intake. Food baskets were distributed to patients during home visits or during PD sessions. The food baskets contained wheat flour, groundnuts, yellow split pigeon peas, lentils and a flour made from a mixture of roasted and ground pulses and cereals, such as barley and gram. PD sessions were conducted at a central place in clusters of 3–4 villages or 8–10 patients, wherein successful practices of positive deviants were identified and adapted as local solutions. The final component of the intervention included providing necessary support to the patients to open bank accounts and assisting the government in registering patients with TB on the Nikshay portal. This was to facilitate compliance with the Nikshay Poshan Yojana, wherein a monthly financial incentive of ₹500 is provided to each notified patient with TB during their treatment duration as a part of NTEP.²⁴ The details of the sampling approach and study settings of the MUKTI programme have been reported separately.²³

Model structure

A mathematical model was developed to estimate costs and health outcomes in a hypothetical cohort of 1000 patients with newly diagnosed pulmonary TB following treatment in either the control or the intervention arm. A comprehensive modelling approach consisting of the Markov model and susceptible–infected–recovered (SIR) model was employed to estimate health outcomes and associated costs within different treatment arms. The progression of the cohort through various health states was simulated using a Markov model, which included distinct health states determined based on the sputum status as well as the nutritional status of a patient with TB (figure 3). The following health states were considered in the Markov model: (1) undernourished sputum-positive, (2) ≥ normal body mass index (BMI) sputum-positive, (3) undernourished sputum-negative, (4) ≥ normal BMI sputum-negative, (5) all-cause mortality and (6) TB-related mortality. Patients with BMI values <18.5 kg/m² were considered undernourished.²⁵

Figure 3. Markov model for estimation of health outcomes among patients with tuberculosis. BMI, body mass index; TB, tuberculosis.

The cohort was followed up for a time horizon of 2 years. This timeframe aligns with the prescribed pulmonary TB treatment duration in the NTEP, which consists of a 6-month regimen (comprising 2 months of intensive phase followed by 4 months of continuation phase).²² Given that the majority of relapses tend to occur within 1 year post-treatment completion or discontinuation, our focus extended to 18 months.¹² To ensure a comprehensive assessment of all health outcomes and associated costs, a 2-year follow-up period was adopted. A cycle length of 1 month was used to simulate the dynamic transition of the patients across different health states associated with TB progression and treatment. In consideration of the time preference associated with costs incurred and health outcomes gained in the future, a discounting approach was applied, wherein the future costs and health outcomes were discounted at an annual rate of 3% in the base case analysis. This discount rate aligns with the recommendations of conducting health economic evaluations in India.²⁶

In addition to the consideration of individual health states within the cohort, we acknowledged the importance of capturing the broader dynamics of TB transmission within the community. Recognising the potential for disease spread from sputum-positive patients, a crucial aspect in the epidemiology of TB, our modelling framework was augmented by integrating an SIR model. This component of analysis allowed us to simulate and understand the dynamics of community transmission, offering a more comprehensive perspective on the effect of different treatment strategies on both individual and population-level health outcomes. When employed together, Markov and SIR models provided a comprehensive framework for evaluating the health and economic implications of different treatment strategies within the context of TB management. As per the recommendations of conducting health economic evaluations in India, the analysis was based on an abridged societal perspective, which included health system costs and OOP expenditure.²⁶

Valuation of health outcomes

Based on the data of 2615 patients newly diagnosed with pulmonary TB enrolled in the MUKTI programme, 77.3% were assumed to have an undernourished status (BMI<18.5 kg/m²) upon treatment initiation. To model the progression of patients to distinct health states of the Markov model, time-specific sputum conversion and relapse rates were derived using time-to-event analysis. Time-to-event analysis was conducted to estimate monthly transition probabilities for sputum conversion and relapse, informed by published empirical estimates. Individual patient-level data were reconstructed from published India-specific Kaplan–Meier survival curves using the open-access digital extraction tool, Automeris.^{27 28} This reconstructed dataset was then used to derive monthly probabilities of cure and relapse in the control scenario, which models the progression of patients under the standard NTEP pathway. To capture the differential effect of nutritional support intervention (MUKTI), relative effect sizes derived from published literature were incorporated. In the time-to-event analysis, the median time to sputum conversion among patients with TB with normal nutritional status was 35 days.²⁹ Undernourished patients with TB were assumed to experience slower sputum conversion, modelled using a relative risk of 2.01.³⁰ The cumulative 24-month post-treatment relapse rate was considered to be 3.56%, although the time-dependent relapse rates were obtained from the time-to-event analysis.²⁷ Relapse probabilities modified for undernourished patients with TB were considered almost three times higher than those observed in patients with a BMI ≥normal based on long-term follow-up data of 857 patients with TB in the Tuberculosis Trials Consortium study.³¹ TB-specific mortality was modelled as a function of BMI using HRS hazard ratios reported by a large Indian cohort with over 774 000 person-years of follow-up.³² Age-specific all-cause mortality from each health state was obtained from India’s Sample Registration System life tables.³³

To capture both individual-level disease progression and the broader transmission effect of the intervention, a hybrid modelling approach that integrated a cohort-based Markov model with an SIR transmission model was employed. The Markov model simulated the health trajectories, costs and health outcomes of a cohort of patients with TB over a 2-year time horizon using monthly cycles. All individuals were assumed to be actively infected and, thus infectious, at the start of the simulation. Monthly transition probabilities for key events (eg, sputum conversion, cure, relapse and death) were derived using time-to-event data stratified by nutritional status, as detailed in the earlier section. As the Markov model progressed over time, the number of infectious individuals in the cohort decreased according to these transition probabilities.

To assess the potential benefits of nutritional support through transmission reduction, an SIR model that simulated TB spread among household contacts of the index patients was embedded. The SIR model was run over a 6-month period, representing the window during which most household transmission occurs and susceptible individuals are likely to be exposed.³⁴ Euler integration was used to numerically solve the system of ordinary differential equations governing the SIR dynamics. An average household size of five individuals per index case was assumed.³⁵ The transmission rate (β) was set at 1.44 per quarter, and the recovery rate (τ) and reactivation rate (η) were set at 0.8 and 0.1 per quarter, respectively, based on published Indian estimates.36,38 The time-varying number of infectious individuals from the Markov model informed the infection pressure in the SIR model, ensuring internal consistency across the two frameworks.

New secondary cases generated by the SIR model were assumed to progress to active TB disease within the 2-year analytic time frame and were incorporated into the original Markov model structure. Each secondary case was treated as initiating a new disease trajectory and was modelled from the infectious state onward using the same transition probabilities, costs and health utility parameters as the primary cohort. This approach enabled the estimation of costs, life years and QALYs attributable to secondary transmission. Double-counting was avoided by limiting the SIR model to a single generation of transmission and modelling secondary cases independently. Reinfection and onward transmission beyond household contacts were not included to maintain tractability and prevent overestimation of indirect effects.

The health-related quality of life for individuals across various health states was evaluated by analysing the data obtained from patients with pulmonary TB enrolled in the MUKTI programme and interviewed using the EuroQol 5-Dimension, 5-Level instrument.²³ The health profiles derived from the collected data were transformed into utility scores using the Indian EQ-5D-5L value set.³⁹ The patients classified as ‘undernourished sputum positives’ exhibited a utility score of 0.755, whereas those with a ≥normal BMI sputum-positive status demonstrated a higher utility score of 0.808. For individuals categorised as ‘undernourished sputum negative’, the utility score was measured at 0.781, whereas their counterparts with a ‘≥ normal BMI sputum-negative’ status exhibited a slightly higher utility score of 0.834. These utility scores served as a foundation for the computation of QALYs.⁴⁰ The health outcomes were valued in terms of life years, QALYs and deaths.

Assessment of cost

The health system cost and OOP expenditure were assessed across both treatment arms. The assessment of health system cost for the delivery of TB services was undertaken from an economic perspective.^{41 42} The cost of inputs associated with the components delivered under the MUKTI programme (counselling sessions, provision of food baskets, PD sessions and linking patients to various government schemes) was obtained from the financial records of the programme from the implementing agency. The cost of each food basket was assessed to be ₹520 (US$6.27), whereas the cost of transportation of the food basket to the patient was set as ₹26.45 (US$0.32). Likewise, per patient cost of PD sessions, counselling and healthcare workers’ incentive were considered as ₹962 (US$11.59), ₹47 (US$0.57) and ₹250 (US$3.01), respectively. As it was a pay-for-performance model, the normative provider payment rates per person covered under the MUKTI programme were considered to be reflective of the programme costs. A similar approach was used to assess the costs associated with the TB service package under NTEP.

For the assessment of treatment costs, the standard treatment protocol for patients with TB under both arms was followed; therefore, similar stage-specific treatment costs for patients with TB were used.²² To assess the cost of treatment of patients with TB, a mix of top-down and bottom-up micro-costing approaches was followed. A similar approach has been followed in several Indian costing studies and recommended as a part of the National Health System Cost Database.43,46 The collection of cost data took place in the Dhar district of Madhya Pradesh, India. The costing analysis was conducted at five primary health centres and three community health centres within the specified region. All cost centres pivotal to service delivery were identified, and resource use was measured in the reference period. The costs were segregated based on the nature of the resource, ie, human resource, capital, equipment, drugs, consumables, furniture and overhead costs. The data collection process for different resources comprised a review of financial and stock records, physical observation of building space used and interviews of personnel involved in service delivery. Capital resources/costs were annualised to estimate the exact value of resources utilised in the reference period. The total cost of the recurrent resources (drugs and consumables) was estimated by multiplying the unit price with the quantity of respective resource consumed. The health system cost per outpatient visit and in-patient admission was estimated.

The data on the OOP expenditure incurred by the patients during the treatment were collected by trained data collectors, who conducted face-to-face interviews with 2615 patients with TB at their homes using structured questionnaires. The details of OOP expenditures on user fee (consultation/hospital charges), medicines, diagnostics, travel, boarding and lodging and other miscellaneous expenditures were obtained. The resulting estimates of OOP expenditures, amounting to ₹5089 (US$61.31) for the control arm and ₹3630 (US$43.73) for the intervention arm over the 6-month treatment period, were subsequently incorporated into the model to assign OOP expenditure across various health states. All costs are reported in ₹ and US$ using the average conversion rate of 1 US$= ₹83 in 2023–2024.⁴⁷ Details of the model input parameters along with their sensitivity ranges and distributions have been provided in online supplemental material 3.

Sensitivity analysis

A multivariable probabilistic sensitivity analysis was undertaken to estimate the effect of joint parameter uncertainty. Under the probabilistic sensitivity analysis, all cost parameters were assigned a gamma distribution, whereas utility values and probabilities/proportions were assigned a beta distribution; 20% and 50% variations on either side of the base value were used for the clinical and cost parameters, respectively. Based on 999 Monte Carlo simulations, the median value of the incremental cost-effectiveness ratio (ICER) along with the 2.5^th and 97.5^th percentiles was calculated. As the estimation of the willingness-to-pay per QALY threshold for India is still underway, the annual gross domestic product (GDP) per capita of India (₹161 500 (US$1946)) was considered as the threshold of cost-effectiveness.⁴⁸ To assess the effect of discounting the future costs and outcomes at an increased rate, a sensitivity analysis was conducted wherein the future costs and outcomes were discounted at 5%.

A scenario analysis was also conducted, which did not include the subsequent benefits of the intervention on community transmission of TB to the contacts of the patient with TB. The decision to exclude community transmission in the scenario analysis was driven primarily by two considerations. First, we aimed to identify the direct effect of the implemented interventions on costs and treatment outcomes for an individual patient with TB. Second, this approach aligns with a conservative estimation strategy, allowing us to establish a baseline understanding of the intervention’s effectiveness in the absence of broader epidemiological influences.

The Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) statement was used to describe different aspects of the study (online supplemental material 2).⁴⁹

Role of funding source

The funder had no role in the design, conduct, preparation of the manuscript, or decision to publish the present study.

Patient and public involvement

The patients were actively involved at multiple stages of this research study. Primary data collection involved face-to-face interviews with patients with TB to assess OOP expenditures incurred during treatment and evaluate their health-related quality of life. The study’s research question and outcome measures were informed by the patients’ clinical and socioeconomic profiles, as well as their preferences and experiences with the healthcare system. The study design incorporated inputs from patients regarding treatment adherence and barriers to nutrition intake, which were gathered during counselling sessions and home visits as part of the MUKTI programme. The dissemination plan includes presenting study findings to policymakers and stakeholders to inform programme decisions, with potential opportunities to involve patients in shaping dissemination strategies, ensuring results are communicated effectively to relevant communities.

Results

Costs and health outcomes

The health system cost associated with implementing the MUKTI intervention was estimated to be ₹16 267 (US$195.99) per patient throughout the treatment period. In comparison, the total health system cost under the prevailing standards of care within the NTEP was estimated at ₹8634 (US$104.02) over a comparable duration. We found that a patient with TB can infect a maximum of 2.2 healthy household contacts over the course of his/her lifetime (figure 4). In the evaluation of health outcomes over a 2-year study horizon, patients enrolled in the MUKTI programme demonstrated an improvement in life years lived compared with those under the current standards of care within the NTEP. Specifically, patients in the MUKTI programme experienced 1.693 life years lived, surpassing the 1.622 life years observed in the control (discounted estimates, table 1). This also translated into a corresponding increase in QALYs for the MUKTI group, with 1.357 QALYs lived in the duration of 2 years compared with 1.294 QALYs in the NTEP group (discounted estimates).

Figure 4. Transmission dynamics of patients with tuberculosis as per the susceptible–infected–recovered model.

Table 1. Costs, health outcomes and incremental values per patient with tuberculosis in the intervention and control arms across the study horizon (2 years).

Findings		Current standards of care under NTEP	Inclusion of nutritional support intervention
Life years lived	Undiscounted Discounted	1.674 (1.668 to 1.679) 1.622 (1.616 to 1.627)	1.745 (1.739 to 1.751) 1.693 (1.687 to 1.698)
QALYs lived	Undiscounted Discounted	1.336 (1.203 to 1.441) 1.294 (1.166 to 1.395)	1.399 (1.253 to 1.511) 1.357 (1.215 to 1.465)
Health system cost (₹)	Undiscounted Discounted	6931 (5740 to 8285) 6807 (5639 to 8134)	11 713 (9795 to 14 140) 11 538 (9654 to 13 923)
Out of pocket expenditure (₹)	Undiscounted Discounted	4071 (3401 to 4864) 3999 (3342 to 4777)	2614 (2211 to 3145) 2574 (2179 to 3095)
Incremental cost-effectiveness ratio (ICER) of Integrating Nutritional Support Intervention in NTEP
		Health system perspective	Abridged societal perspective
Per life year gained (₹)	Undiscounted Discounted	66 969 (33 655 to 105 422) 67 164 (33 872 to 105 460)	46 140 (12 491 to 89 160) 46 581 (12 690 to 88 966)
Per QALY gained (₹)	Undiscounted Discounted	75 890 (35 239 to 140 325) 76 306 (35 752 to 139 484)	52 762 (13 731 to 115 443) 53 292 (14 452 to 115 026)

Figures in parentheses represent 95% CIs.

NTEP, National Tuberculosis Elimination Programme; QALY, quality-adjusted life year.

Modelling the study cohort over the 2-year time horizon revealed that the health system costs associated with the MUKTI programme would be ₹11 538 (US$139) per patient with TB. In 2 years, the control arm under the current NTEP standards would incur health system costs of ₹6807 (US$82) in the corresponding duration (discounted estimates). In parallel, OOP expenditures, indicative of the financial burden borne by the patients, demonstrated an opposite trend. Patients in the MUKTI programme were estimated to have lower OOP expenditures (₹2574 (US$31)) than those in the NTEP control arm (₹3999 (US$48.18)) over the 2-year study horizon (discounted estimates, table 1).

Cost-effectiveness of the MUKTI programme

In line with the CHEERS guidance, all ICER values are based on discounted costs and outcomes. From a health system perspective, the incremental cost per life year gained for integrating nutritional support intervention in NTEP was estimated at ₹67 164 (US$809), whereas the incremental cost per QALY gained stood at ₹76 306 (US$919) (table 1). In an abridged societal perspective that considered health system costs along with OOP expenditures, the incremental cost per life year gained reduced to ₹46 581 (US$561), and the incremental cost per QALY gained was ₹53 292 (US$642). Even when a higher discount rate of 5% was applied to both costs and health outcomes, the ICER values remained relatively stable. Under this scenario, the incremental costs per QALY gained were ₹77 420 (US$933) and ₹55 040 (US$663) from the health system and societal perspectives, respectively. These findings indicate that the intervention remains cost-effective even under more conservative discounting assumptions, with ICERs well within the commonly accepted threshold.

Notably, the MUKTI programme exhibited a robust probability of cost-effectiveness when evaluated against the threshold aligned with the current annual per-capita GDP of India. Even after accounting for the joint parameter uncertainties of the analysis, the assessment using probabilistic sensitivity analysis demonstrated that the MUKTI programme exhibited 91% probability of being a cost-effective intervention compared with current standards of care under NTEP (figures5 6). Nevertheless, in the scenario that did not account for the effect of community transmission of TB, the incremental cost of producing an additional life year and QALY exceeded the predefined cost-effectiveness threshold (CET) (table 2). As a result, in the absence of considering the community transmission, the intervention was deemed cost-ineffective within the Indian context.

Figure 5. Cost-effectiveness plane comparing incremental costs and benefits across different simulation scenarios of incorporating nutritional support intervention in the National Tuberculosis Elimination Programme. QALYs, quality-adjusted life years.

Figure 6. Probability of intervention being cost-effective at different willingness-to-pay levels. CET, cost-effectiveness threshold; GDP, gross domestic product; ICER, Incremental cost-effectiveness ratio; QALY, quality-adjusted life year; WTP, willingness-to-pay.

Table 2. Costs, health outcomes and incremental values per patient with tuberculosis in intervention and control arms across the study horizon (2 years) if community transmission is not considered.

Findings		Current standards of care under NTEP	Inclusion of nutritional support intervention
Life years lived	Undiscounted Discounted	1.947 (1.94 to 1.953) 1.892 (1.885 to 1.898)	1.951 (1.943 to 1.957) 1.896 (1.889 to 1.902)
QALYs lived	Undiscounted Discounted	1.552 (1.391 to 1.682) 1.508 (1.352 to 1.634)	1.562 (1.401 to 1.693) 1.518 (1.362 to 1.646)
Health system cost (₹)	Undiscounted Discounted	6805 (5662 to 8137) 6731 (5604 to 8042)	11 518 (9715 to 13 690) 11 410 (9635 to 13 561)
Out of pocket expenditure (₹)	Undiscounted Discounted	4008 (3343 to 4755) 3967 (3309 to 4700)	2582 (2129 to 3090) 2557 (2110 to 3061)
Incremental cost-effectiveness ratio (ICER) of Integrating Nutritional Support Intervention in NTEP
		Health system perspective	Abridged societal perspective
Per life year gained (₹)	Undiscounted Discounted	711 069 730 104	447 461 460 859
Per QALY gained (₹)	Undiscounted Discounted	313 318 321 301	209 791 215 735

Figures in the parentheses represent 95% CIs.

NTEP, National Tuberculosis Elimination Programme; QALYs, quality-adjusted life years.

Discussion

TB remains a formidable global health challenge, particularly in LMICs, where it accounts for a significant share of both incidence and mortality.⁵⁰ Undernutrition exacerbates TB vulnerability and complicates treatment, making nutritional interventions a crucial component of comprehensive TB control strategies.⁵ Notably, nutritional support interventions lead to substantial improvements in clinical outcomes of patients with TB, including enhanced functional capacity and reduced mortality.¹³ Nevertheless, the cost-effectiveness of integrating the nutritional support intervention in India’s TB control programme has not been assessed so far. In striving towards addressing this critical knowledge gap, this study highlights the interplay between nutrition, TB control and the consequent economic implications of integrating nutritional support interventions into the NTEP. Overall, we found that incorporating nutritional support interventions within the framework of the NTEP represents a cost-effective strategy in the Indian context. Moreover, when the health benefits of this integration were extrapolated to a national scale, it is projected to result in a gain of 0.14 million QALYs within the Indian population.

This study underscores the economic viability of adopting a comprehensive strategy to address the burden of TB, which encompasses the consideration of social determinants alongside curative treatment. As LMICs are grappling with the triple challenges of TB, undernutrition and resource constraints, such interventions become quintessential.¹⁰ The study findings carry significant implications to facilitate such strategic policy decisions, offering essential insights into the judicious allocation of resources for TB control efforts in such countries. Bangladesh, Indonesia, Nigeria and Ethiopia face similar epidemiological profiles and systemic challenges—including limited healthcare resources, widespread poverty, high OOP health expenditures and nutritional insecurity among TB-affected populations.⁵¹ In these settings, integrating nutritional support into national TB programmes may yield comparable health and economic benefits, especially for underserved and marginalised communities. However, the generalisability of the findings of this study should be approached with caution, as factors such as healthcare infrastructure, baseline nutritional status, TB incidence and cost structures may differ across countries. For instance, countries with stronger social welfare systems or lower prevalence of undernutrition may observe smaller marginal benefits. Additionally, differences in cost-effectiveness thresholds, funding mechanisms and programme priorities may affect the transferability of the ICER estimates presented in this study. Nevertheless, the core insight that addressing undernutrition within TB care pathways can be both clinically advantageous and economically viable offers a compelling rationale for other high-burden LMICs to consider similar integrated approaches in their TB control strategies.⁵²

One notable facet of the integration of nutritional support intervention in the TB control programme is its effect on patient–health system dynamics, fostering increased patient interaction. This heightened engagement not only instils confidence in the patient towards the health system but also serves as a crucial deterrent against treatment default, a pivotal determinant of TB control programme success.⁵³ Moreover, the integration of nutritional support contributes to a shift in care-seeking behaviour from the private sector to the public sector. This shift, in turn, mitigates the economic implications of healthcare seeking for other comorbidities, as OOP expenditure tends to be higher in the private sector.^{54 55} Another salient aspect of this integration is its role in expanding the scope of care beyond TB. By enhancing patient–health system interactions, the integration encourages care-seeking for additional morbidities, consequently leading to improved overall health outcomes. These ancillary benefits highlight the multifaceted advantages of incorporating nutritional support interventions into TB control programmes, underscoring the potential to bolster both programme efficacy and broader healthcare utilisation patterns.

Beyond its cost-effectiveness, the integration of nutritional support into TB care also holds important equity implications, particularly within the Indian context, where undernutrition and poverty remain widespread and deeply interconnected.¹⁶ TB disproportionately affects socioeconomically disadvantaged populations, many of whom experience food insecurity, limited access to healthcare and a high risk of incurring catastrophic health expenditures. Nutritional support interventions can play a critical role in mitigating these inequities by directly targeting a key vulnerability—undernutrition—which not only increases the risk of developing TB but also impairs treatment adherence and outcomes.^{13 19} By improving nutritional status, such interventions may enhance immune recovery, reduce the risk of relapse and promote treatment completion, thereby benefiting those most at risk of poor outcomes. Moreover, providing nutritional support may help offset indirect costs borne by patients and caregivers, such as lost income due to illness or caregiving responsibilities, which are particularly burdensome for economically marginalised households. In this regard, the intervention aligns with the equity-focused objectives of the NTEP and global TB strategies, which emphasise reducing health disparities and protecting vulnerable populations from financial hardship.^{4 24} While this study primarily evaluates economic efficiency, the broader social value of such interventions—particularly their potential to reduce health and financial inequities—should be considered in future evaluations and policy decisions.

Our analysis indicates that, during the treatment phase, a patient with pulmonary TB is estimated to live between 0.647 and 0.678 QALYs in 1 year. These findings align with previous empirical assessments, which reported a range of 0.59–0.71 QALYs per year for individuals with pulmonary TB.^{56 57} This consistency serves as validation for our model estimations. Nevertheless, certain limitations warrant consideration in interpreting the findings of this study. An inherent limitation of the present analysis lies in its exclusive focus on pulmonary TB, which omits extrapulmonary TB cases. This exclusion could overestimate the cost-effectiveness of the nutritional intervention, given that community transmission, a key determinant of cost-effectiveness, is absent in extrapulmonary TB. Nonetheless, given that extrapulmonary TB cases represent 15% of the total TB cases, the potential overestimation might not exert a substantial influence on the observed findings.⁵⁸ Future research should aim to address this limitation by incorporating extrapulmonary TB cases for a more comprehensive evaluation of nutritional interventions across the diverse manifestations of the disease.

Moreover, the model’s assumptions and outcomes are contingent on the accuracy and representativeness of the data used, particularly the survival curves and progression probabilities derived from the specific patient cohorts. Although in the sensitivity analysis we varied the values of clinical parameters by 20% on either side of the base value, the variability in patient characteristics, healthcare infrastructure and socioeconomic factors across different regions might affect the generalisability of the results to diverse settings within LMICs. Additionally, the exclusion of community transmission dynamics in the scenario analysis, while providing valuable insights, simplifies the complex reality of TB epidemiology.

This study employed QALY as the primary outcome measure to assess cost-effectiveness, whereas disability-adjusted life years (DALYs) have historically been used more widely in TB modelling studies, particularly in LMICs.^{19 20} Our choice of QALYs was guided by national methodological guidelines and the availability of country-specific data. Specifically, the Indian Reference Case for the conduct of economic evaluations recommends the use of QALYs as the preferred metric for measuring health outcomes.²⁶ This recommendation is also endorsed by Health Technology Assessment India, the country’s central Health Technology Assessment (HTA) agency, as it ensures comparability between different cost-effectiveness studies conducted in the country. Moreover, India has recently developed its EQ-5D-5L value set, which reflects the health state preferences of the Indian general population.^{39 40} This development has enabled researchers to use QALYs in a manner that is not only methodologically rigorous but also culturally and contextually relevant.^{59 60} In this study, primary data on health-related quality of life were collected from patients with TB using the EQ-5D-5L instrument, and utility values were generated using the Indian value set. This approach enhances the internal validity of the analysis and aligns with India’s movement towards evidence-informed priority setting through the use of locally derived data. Nonetheless, the use of QALYs may limit comparability with other studies that adopt DALYs as their outcome metric, especially in cross-country analyses or global health comparisons. However, given the increasing adoption of QALYs within India’s HTA ecosystem, this choice strengthens the policy relevance of our findings for national-level decision-making. By acknowledging this trade-off, we have discussed the implications of using QALYs in the Indian context to support transparent and consistent interpretation of our results.

Another limitation of this study is the use of a 2-year analytic time horizon, which may not fully capture the long-term health and economic benefits associated with TB cure. Although health economic evaluation guidelines generally recommend adopting a lifetime horizon for chronic infectious diseases, in this study, a 2-year time horizon was selected based on programme relevance and clinical rationale. Specifically, this period corresponds with the standard TB treatment duration under the NTEP, which comprises a 6-month treatment regimen—2 months of an intensive phase followed by 4 months of a continuation phase.²² Furthermore, evidence indicates that the majority of TB relapses occur within the first year following treatment completion or discontinuation.¹² Therefore, extending the follow-up period to 18 months post-treatment allows for the capture of most clinically significant relapse events. An additional 6 months in the model were included to ensure that all relevant health outcomes and costs associated with treatment, relapse and early post-treatment phases were comprehensively assessed. Although this approach provides meaningful insights into the short- to medium-term cost-effectiveness of the intervention, it may underestimate the full benefits of TB cure, particularly the long-term survival advantages and broader societal gains from sustained transmission reduction. Future modelling efforts incorporating a lifetime horizon could help further validate and extend the conclusions drawn from this analysis.

Although TB is known for its prolonged latent phase, which is often more appropriately represented through susceptible–exposed–infected–recovered (SEIR) or susceptible–latent–infected–recovered modelling structures, this study employed a simplified SIR model to simulate short-term transmission within household contacts. The SIR model was used to capture the proximal transmission effects arising from the altered duration of infectiousness among index patients with TB receiving nutritional support. In the Indian context, where household sizes are large, living conditions often crowded and undernutrition highly prevalent, close contacts are frequently exposed to intense and prolonged transmission during the early phase of infectious disease. As our model focused on a 2-year analytic window and specifically assessed first-generation, short-term secondary TB cases, it was reasonable to limit the transmission dynamics to the SIR framework. Nevertheless, by excluding a latent phase, the SIR model may overestimate secondary disease incidence under the assumption of uniform rapid progression. To account for this structural simplification, a sensitivity analysis was conducted wherein the number of secondary TB cases generated through the SIR model was halved, effectively modelling a scenario in which fewer individuals progress to active disease within the study time horizon. Under this conservative scenario, the ICER increased to ₹156 800 from the health system perspective and ₹111 109 from the abridged societal perspective. Despite this increase, the intervention remained cost-effective, with ICERs falling well within the cost-effectiveness threshold. These findings affirm the robustness of our conclusions, even under less favourable assumptions regarding transmission and disease progression. Although a more complex SEIR model might be considered in future work to explore longer-term population-level dynamics, our current modelling framework was adequate for evaluating the near-term, household-level effect of nutritional support, which was the primary aim of this analysis.

Lastly, the cost-effectiveness threshold employed, anchored to the annual per-capita GDP of India, might not universally apply to all LMICs, necessitating caution in generalising the cost-effectiveness findings. The choice of an appropriate cost-effectiveness threshold is a subject of ongoing debate in India and other LMICs. In the present analysis, India’s per capita GDP was used as the willingness-to-pay threshold in the base-case analysis, consistent with the Indian Reference Case for Health Technology Assessment, which recommends the use of the per capita GDP as a benchmark until an empirically estimated demand-side threshold becomes available.²⁶ However, the current threshold may not fully capture the opportunity cost of health spending in the present Indian context. Recent literature suggests that lower, supply-side CETs may also be used, given the health system’s budgetary constraints and marginal productivity of healthcare spending.^{61 62} Despite these considerations, the decision to adopt a per capita GDP-based threshold was guided by several factors. First, India is in a phase of increasing investment in health, with expanding fiscal space and growing political commitment towards achieving universal health care. In such a context, demand-side CETs that reflect societal preferences may provide a more appropriate benchmark for resource allocation. Second, the Indian healthcare system is characterised by multiple payers and diverse financing mechanisms—ranging from additional budget allocations to reallocation within existing programmes—implying that the shadow price of health may vary across settings. Nevertheless, in recognition of the concerns around potential overstatement of the WTP threshold, a probabilistic sensitivity analysis was conducted, and a cost-effectiveness acceptability curve was presented, showing the probability of cost-effectiveness across a range of thresholds (figure 6). Importantly, the intervention remains cost-effective even under stricter assumptions, as from the societal perspective, the ICER is approximately ₹53 000 per QALY gained, which lies well within the more conservative CET ranges. This reinforces the robustness of the findings and the value of integrating nutritional support into India’s TB control efforts. Despite these limitations, the analysis offers a robust foundation for understanding the potential benefits and challenges associated with integrating nutritional support into TB control programmes in resource-constrained settings.

Moreover, the viability of the implementation of the MUKTI programme has been evaluated recently, which revealed that implementing such a nutrition supplementation and education programme for patients with TB in India is feasible.⁶³ However, the evaluation incorporated marginal financial costs of implementing the programme and recommended further research to determine the cost-effectiveness of the MUKTI programme by examining the total costs from an economic perspective. Therefore, the present analysis incorporated costs from an economic perspective. There is a slight discrepancy in the cost estimates between our analysis and that of Howell et al, which is explicable through several key methodological distinctions and operational efficiencies. Our estimated marginal cost of implementing the MUKTI programme is approximately 35% lower than the cost reported by Howell et al. This variance primarily stems from our economic approach for cost analysis, which includes annualising start-up costs, as opposed to the method by Howell et al, which fully accounted for these costs within the first year. Additionally, our analysis benefits from a larger dataset, encompassing 2615 patients across all three phases of the study, compared with 1000 patients from phase 1 in the study by Howell et al. This broader scope allows us to capture economies of scale and the learning curve effect, wherein increased operational efficiency and stabilised implementation intensity over time reduce the marginal cost per patient. Furthermore, a threshold analysis was conducted, considering the higher implementation cost reported by Howell et al (details in Supplementary Material 1). Even under this conservative scenario, the MUKTI intervention remains cost-effective. That is, even with the higher cost reported by Howell et al, the MUKTI intervention remains cost-effective.

In conclusion, our analysis highlights the potential benefits of holistic approaches to TB control that prioritise not only medical but also socioeconomic determinants. This study underscores the intricate relationship between nutrition and TB control in India. The MUKTI programme, integrating nutritional support into the NTEP, presents an economically efficient strategy with potential health benefits. The economic evaluation, while highlighting cost-effectiveness, raises considerations, particularly concerning community transmission dynamics. These findings contribute valuable insights to inform policy decisions on resource allocation for TB control in India, emphasising the need for holistic, evidence-based interventions in the elimination efforts against TB.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.