Cost-effectiveness of BPaL/BPaLM as compared to mixed standard of care bedaquiline containing regimen for MDR/RR-TB

Study setting

In India, 119,000 MDR/RR-TB cases have been estimated for 2021. In 2022 compared to 2021, there were 32 per cent more MDR/RR-TB patients notified by National Tuberculosis Elimination Programme (NTEP)¹⁴. MDR/RR-TB diagnosis and therapy have seen substantial change in the last few decades. However, there are still a number of challenges hindering optimal disease management.

Increasing resistance to WHO-recommended Group-A and Group-B medications, the high rates of catastrophic costs experienced by patients with MDR/RR-TB, the wide variations in private providers’ involvement in TB treatment, and the lack of private sector engagement are the most urgent problems.

The most pressing issues include: (i) increasing resistance levels to WHO recommended Group-A and Group-B drugs, (ii) substantial catastrophic costs facing individuals with MDR/RR-TB, (iii) lack of steady participation of private providers in TB management and (iv) insufficient private sector engagement. These challenges are further compounded due to inadequate national investment in health. Nonetheless, India has the potential to guide the global battle against MDR/RR-TB¹⁵. Given that India carries a significant portion of the global TB burden, achieving success in the country which would greatly influence global TB control efforts. In order to achieve the TB elimination goal, there is a need for the implementation of rapid diagnostic tools and newer drugs that can cure MDR/RR-TB and accelerate the step towards ‘End-TB’ goal. In India, under the National Tuberculosis Elimination Programme, all notified MDR/RR-TB patients were taking treatment with the existing mixed SoC regimen, in which 58 per cent of patients placed on the longer regimen (18-20 months) and 42 per cent were treated with the shorter regimen (9-11 months).

Study design

We used Markov model for this economic evaluation. Our study centred on evaluating the effects of two treatment regimens (i) BPaL and (ii) BPaLM compared with a mixed SoC regimen which is in current practice. We conducted this assessment using MDR/RR-TB patients aged over 14 yr, regardless of their fluoroquinolones resistance status. Specifically, we included individuals who had been exposed to bedaquiline, linezolid, pretomanid or delamanid for less than one month in the past.

Study perspective

Using a health system perspective, this cost-effective analysis solely took into account the expenses incurred by the health system, including the costs of patient incentives, nutritional assistance and regimens (shorter SoC, Longer SoC, BPaL and BPaLM).The costs for pre investigations such as smear examination by Ziehl-Neelsen smear microscopy, CBNAAT, solid sputum culture, ECG, HIV rapid test, full haemogram, electrolyte, creatinine, blood sugar, thyroid stimulating hormone test, chest X-ray and liver function test were also added.

Intervention and comparator

In the current study, a comparison was made between the costs and outcomes of BPaL/BPaLM regimens with the current strategy of treating MDR/RR-TB which is a mixed SoC regimen in the NTEP (Table I).

Intervention

The BPaL regimen (6-9 months) consisted of bedaquiline, pretomanid and linezolid; while the BPaLM regimen (6-9 months) included BPaL along with moxifloxacin.

Comparator

The comparator was the currently used mixed SoC regimen for treating MDR/RR-TB following the National Tuberculosis Elimination Programme guidelines¹⁶. The shorter SoC regimen included levofloxacine, bedaquiline, clofazimine, ethambutol, ethionamide and pyrazinamide. The regimen comprised of a 4-months initial phase extendable to 6-months and a 5-months continuation phase for a total duration of 9 to 11-months. Bedaquiline was administered for 6-months. The longer SoC regimen included levofloxacin, bedaquiline, clofazimine, linezolid and cycloserine and was administered for 18-20 months.

Time horizon

Costs and outcomes of the two comparative regimens were modelled using a life time horizon. Based on the literature, the average age of a TB patients was 32 yr¹⁷, and life expectancy at that age was utilised. Both costs and outcomes were adjusted by three per cent discount rate¹⁸. The population’s health status was defined by the model, which tracked it until it was either cured or death.

Description of model

A cohort of 48,563 MDR/RR-TB patients was monitored by the model¹⁹.The model considered TB treatment outcomes between the two regimens. Regardless of their fluoroquinolones resistance status, we only took into consideration individuals over the age of 14 yr who accessed the government health facilities every two weeks for treatment. Specifically, we included individuals who had been exposed to bedaquiline, linezolid, pretomanid or delamanid for less than one month in the past. Table II describes the demographic characteristics of the TB patients. Disease recurrence was considered in transition health state. Treatment related mortality was also taken into account. Life years and QALYs obtained by patients receiving two distinct regimens were the model outcomes. The analyses were performed using a Microsoft Excel spreadsheet.

Markov model

The Markov model considered five health states of TB treatment outcomes (Figure). Based on specific probabilities, each patient moved to a different health condition after a year.

Close

Figure.

Markov model of the sequence of possible TB treatment outcomes and health states.

Export to PPT

Model input parameters

The important model input parameters include demographic values, treatment outcomes attributable to the regimen BPaL, BPaLM and mixed SoC, transition probabilities and various costs of the health system data. The detailed information on the collected input parameters are given in table II. It also included life expectancy and all-cause mortality which was used from the SRS based life table (2012-2016)²⁰.

Cost data

MDR/RR-TB treatment guidelines, previously published prices and expert opinions were used to estimate treatment costs from a provider perspective. Costs related to treatment regimens such as cost of full course of the treatment regimen, nutritional support to patients, incentives to treatment supporters were provided by Central TB Division (CTD) through personal communication. Medication costs for all regimens were also provided by CTD. Pre-treatment investigation costs were taken from the published literature².

Effectiveness data

The clinical outcomes of the BPaL and BPaLM regimens including recurrence were sourced from randomised control trials^11,13,21-24, the Nix-TB and TB-PRACTECAL trials. The Nix-TB trial was done at three sites in South Africa among 109 patients. The use of oral, bedaquiline, pretomanid and linezolid was examined. The intention-to-treat analysis showed that 90 per cent favourable outcome, which is similar to that obtained with the SoC for DS-TB (isoniazid, rifampicin, pyrazinamide and ethambutol). The TB-PRACTECAL trial was done in Belarus, South Africa and Uzbekistan during 2017 to 2021 with four comparator groups. One was the SoC group and other three are BPaL groups such as BPaL, BPaLC and BPaLM. This multi country randomised controlled trial showed that treatment with BPaLM was more effective and had a better safety profile than SoC. BPaLC and BPaL were also highly efficacious. Treatment outcomes of SoC by shorter and longer regimen were collected from the TB control Programme as well as published literature^13,14,21,25. The collected treatment outcomes of BPaL, BPaLM and the SoC were pooled using meta-analysis. The primary parameters of the model were demographic information and TB treatment outcomes, including cure, failure, lost-to-follow-up and death. Information on the recurrence of SoC regimen was collected from the published literature². Data on the quality of life for cured and other outcomes of TB patients were collected from the published literature^26,27.

Model outcome parameters

The model outcomes were presented in the form of life years, QALYs and the total cost incurred by the population for all the regimens. Life years were estimated by multiplying the number of patients by their respective remaining life expectancy. QALYs were derived by multiplying life years by the utility score of quality of life related with each health state. In this study, utility scores were obtained directly from the published sources used for our model parameters. These utility values were applied to estimate the QALYs for each treatment regimen. Using the cost of regimens and the corresponding QALYs estimated from model, we calculated the incremental cost effectiveness ratio (ICER) calculated to compare alternatives.

Willingness to pay

Cost-effectiveness of the proposed regimens was assessed by comparing the willingness to pay (WTP) criterion, which is the one time GDP per capita (INR 1,15,746) for the year 2022-2023²⁸.

Sensitivity analysis

We generated 1,000 input sets by uniformly sampling from reasonable ranges for the input parameters. By varying the input parameters by 20 per cent above or below their typical values, sensitivity analysis was used to assess the model’s robustness. Model results were examined in relation to change in input parameters using one-way sensitivity analysis (OWSA). A tornado diagram was used to illustrate the uncertainty in the outcome factors and how it affected the incremental cost effectiveness ratio. Further, the model was validated using Microsoft Excel through Probability Sensitivity Analysis (PSA) utilising 1,000 Monte Carlo simulation iterations that included 95 per cent confidence intervals (CI). A cost-effectiveness acceptability curve (CEAC) was developed to illustrate the model’s probabilistic response to various cost-effectiveness threshold and the resulting ICER values were displayed in a scatter plot.

Cost threshold analysis

The price of the BPaL and BPaLM was taken from the CTD. However, the cost-effectiveness was mostly determined by the price at which the drugs were procured. We assessed the threshold prices at which BPaL and BPaLM would become most cost-saving treatment options.

Budget impact analysis (BIA)

This economic evaluation calculated the financial effects of switching from mixed SoC to BPaL/BPaLM. Cost-effectiveness analysis was supplemented with BIA to help make well informed decisions. The budget estimates were based on current unit costs of drugs and service delivery within the National Tuberculosis Elimination Programme.

Base case analysis

The base case analysis for the population of 48,563 showed that the total discounted costs by health system perspective for the BPaL, BPaLM and the current mixed SoC regimen were INR 2515, INR 2644 and INR 2630 million, respectively while the total undiscounted costs acquired were INR 4648, INR 4887 and INR 4860 million for the same, respectively. It was observed that the cost for the drug was higher for BPaLM regimen as against BPaL and mixed SoC. The human resources cost, diagnostic, nutritional support, incentives to treatment supporters were considered same for both intervention regimens. In terms of effectiveness, QALY gained by BPaLM regimen is higher than the BPaL and mixed SoC (2.01 vs. 1.93 vs. 1.63 million) (Table III).

Incremental cost-effectiveness ratio (ICER)

The ICER was calculated by the estimated incremental cost and incremental QALYs. When BPaL was compared with the mixed SoC regimen, the ICER was INR -379 which indicated that (the intervention is less cost and more effective) we have to spend INR 379 less per patient for BPaL than the mixed SoC to gain one additional QALY. When BPaLM was compared with the mixed SoC regimen, the ICER was INR 37 which indicates that (the intervention is more cost and more effective) we have to spend INR 37 additionally per patient for BPaLM than the mixed SoC to gain one QALY (Table III). The cost-effectiveness plane illustrates the ICER values (Supplementary Fig. 1).

Scenario analysis

We carried out multiple scenario analyses by altering the proportion of patients receiving each regimen ranging from the current distribution of 58 per cent on the longer vs. 42 per cent on the shorter regimens to 50 per cent in longer regimen vs. 50 per cent in shorter regimen. These alternative mixes were compared against the BPaL and BPaLM regimens. The corresponding ICER values for each scenario were calculated and summarised in table IV. The ICER for the proportion of mixed SoC 10 per cent vs. 90 per cent calculated for BPaLM and BPaL was INR 1521 and INR 1507, respectively. When the mixed SoC proportion was adjusted to 50:50 ratio, the estimated ICER for BPaLM and BPaL were INR 285 and INR -64, respectively. As the proportion of longer regimen was increased in the mixed SoC, BPaLM regimen turned more cost-effective and BPaL regimen turned more cost-saving.

One way sensitivity analysis (OWSA)

The OWSA for the BPaLM regimen revealed that factors such as health-related quality of life utility score of BPaLM cure, medicine cost of BPaLM, medicine cost for longer SoC and health related quality of life utility score of shorter SoC cure had a substantial impact on the ICER value (Supplementary Fig. 2A). Similarly, the OWSA for the BPaL regimen revealed that factors such as the health related quality of life utility score of BPaL cure, medicine cost for longer SoC medicine cost of BPaL and health related quality of life utility score of shorter SoC cure had a significant influence on the ICER value (Supplementary Fig. 2B).

Probability sensitivity analysis (PSA)

The PSA indicated that when joined incremental cost and effectiveness was considered and measured in QALY, BPaLM was found to be cost-effective in 48.3 per cent and cost-saving in 45.5 per cent of the iterations as shown in supplementary figure 3A. The PSA results indicated that the joint incremental cost and effectiveness, measured in QALY favored BPaL as being cost-saving in 61 per cent and cost-effective in 30.3 per cent of the iterations, as illustrated in supplementary figure 3B. Additionally, the CEAC showed that, across a range of cost-effectiveness thresholds, the 6-month BPaL regimen had a 91 per cent change of being an economically dominant strategy when compared to the mixed SoC regimen, as displayed in supplementary figure 3C.

Cost threshold analysis (CTA)

The cost threshold analysis showed that BPaLM turns to be cost-saving when the cost is decreased from INR 39738 to INR 39438 (Supplementary Fig. 4). This indicates that BPaLM turns to be cost-saving if the drug cost is reduced by approximately one per cent.

Budget impact analysis

Table V shows the additional budget required for implementation of BPaL or BPaLM regimen to treat MDR/RR-TB patients in India. If BPaLM regimen is implemented, health system needs to invest around INR 13 million additionally per year. Whereas for the BPaL regimen, health system needs to invest INR 106 million lesser than the current investment.