Abstract
Background
New and effective tools for detecting drug-resistant tuberculosis (DR-TB) include GeneXpert XDR and targeted Next Generation Sequencing (tNGS). However, data on their implementation in high TB-burden settings is limited. We aimed to determine cost-effectiveness of different strategies using GeneXpert XDR or tNGS for DR-TB detection in high TB-burden, low-resource settings.
Methods
A dynamic simulation model was calibrated to WHO-reported TB data for Philippines and Thailand. Intervention scenarios for expanded diagnostic testing of drug-resistance were simulated for 2025 – 2035. Health benefits were estimated using disability-adjusted life years. Cost-effectiveness was calculated from a health system perspective using country-level TB diagnosis and treatment costs. Analyses include incremental cost-effectiveness ratios (ICERs) and incremental net monetary benefit (INMB).
Results
Implementing GeneXpert XDR or tNGS for DR-TB detection improves TB health outcomes. Scenarios using GeneXpert XDR are more likely to be cost-effective than scenarios using tNGS. Interventions targeting previously treated cases reduce costs but also reduce health benefits. Testing all TB cases with GeneXpert XDR is cost-effective (Philippines ICER = $1,808, INMB = $210M; Thailand ICER = $5,251, INMB = $26M) with a 1 x GDP willingness-to-pay threshold (WTP). Targeting GeneXpert XDR to previously treated cases is also cost-effective (Philippines ICER = $1,288, INMB = $52M; Thailand ICER = $3,667, INMB = $9.2M) but results in lower INMB. tNGS is cost-effective at higher WTP.
Interpretation
In high TB-burden countries, GeneXpert XDR is cost-effective as an additional DR-TB diagnostic test. tNGS is not cost-effective for routine clinical DR-TB testing but has potential for application to high-risk populations, especially with introduction of new TB treatment regimens.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-025-24934-z.
Keywords: Tuberculosis, Drug-resistance, Diagnostic testing, Genomic sequencing, Cost-effectiveness
Introduction
Drug resistant tuberculosis (DR-TB), defined as resistance to one or more drugs in the standard TB treatment regimen, is a major public health concern in South-East Asia, which continues to be the region with the highest number of incident cases of DR-TB globally [1]. WHO guidelines currently recommend GeneXpert (Xpert MTB/RIF or Xpert Ultra) as the initial diagnostic test for adults and children with signs and symptoms of pulmonary Mycobacterium tuberculosis (MTB) [2]. However, this test is limited to detection of rifampicin-resistant TB (RR-TB), potentially leading to the misdiagnosis and undertreatment of other forms of drug resistance, contributing to increased treatment failure and TB-related mortality [3, 4].
Modelling studies have shown that DR-TB treatment failure can lead to amplification of drug resistance, contributing to higher rates of multidrug-resistant TB (MDR-TB) [5, 6]. To address this issue, GeneXpert XDR has been identified as an additional molecular diagnostic tool, demonstrating high sensitivity for detection of TB drug resistance to isoniazid (INH), fluoroquinolones (FLQ), ethionamide (ETH), and second-line injectables (amikacin, kanamycin, capreomycin) [7]. As a result of the recent global scale-up of GeneXpert rapid testing, there is an added benefit of requiring minimal training in low resource settings already familiar with the GeneXpert platform.
Recently updated WHO guidelines for TB detection include new guidance for the use of targeted next-generation sequencing (tNGS) for detection of drug resistance [2]. While genomic sequencing has predominantly been used for TB control in high resource, low burden settings[8], low- and middle-income countries (LMICs) have increased their sequencing capacity following the COVID-19 pandemic and have shown interest in applying this technology towards TB control [9]. Furthermore, recent advances in technology allow tNGS to be conducted directly from uncultured sputum samples, allowing a faster turnaround time and requiring less data storage, while also demonstrating higher sensitivity and specificity than other existing diagnostic tools [10].
In a 2023 rapid communication, WHO has determined that use of NGS is accurate, cost-effective depending on context, acceptable and implementable under routine conditions, despite inherent complexity [11]. Although recent studies have investigated the implementation feasibility and benefits of genome sequencing as a diagnostic tool for DR-TB in high-burden, low-resource settings[12], evidence to support cost-effectiveness of tNGS in these settings is limited [13, 14]. Similarly, while previous costing studies have been conducted on the implementation of GeneXpert MTB/RIF, this study explores the additional use of GeneXpert XDR for DR-TB detection.
The aim of this study is to determine the cost-effectiveness of expanded diagnostic testing for DR-TB, through either GeneXpert XDR or tNGS. The Philippines and Thailand are used as case studies, building on a previous study modelling the progression of DR-TB in high burden settings [15]. Although these diagnostic tests will incur additional costs, they are anticipated to reduce DR-TB treatment failure and mortality, resulting in population level health benefits. In addition, as DR-TB is more expensive to treat than drug susceptible tuberculosis (DS-TB), interventions that reduce incidence and prevalence of DR-TB have a high potential for averting future expenditure and reducing the occurrence of catastrophic TB costs.
Methods
We evaluate the cost-effectiveness of GeneXpert XDR and tNGS applied as a rapid molecular diagnostic for DR-TB over a ten-year period, from 2025 to 2035. The scenarios are run using a previously published dynamic transmission model calibrated to TB epidemiology in the Philippines and Thailand [15].
Dynamic Transmission Model
We projected different intervention outcomes using a compartmental transmission model developed on Vensim DSS v10.1.4. (Fig. 1) The model includes fourteen stocks to represent TB dynamics in each country population as individuals move from being susceptible, to infected with latent TB, progressing to active TB, diagnosed, treated and recovered. Additional details on model parameters are summarized in Supplementary Table 1. The model was calibrated to Philippines and Thailand WHO estimates for TB incidence and mortality from 2010 to 2019, as reported in the WHO Global TB Report 2024.
Fig. 1.
Simplified Model Structure. S= susceptible population, LA = latent population, LB = population in late latency, I = infected population, D = diagnosed population, T = treated population, R = recovered population, DS-TB = drug susceptible TB, DR-TB = drug resistant TB. Subscripts S,RandX denote stocks and transitions related to DS-TB, DR-TB, and DR-TB diagnosed and treated as DS-TB, respectively
In addition, we disaggregated the model using a subscript to track previously treated TB cases through all model compartments. Previously treated cases are defined as individuals who have received any anti-TB treatment for a period of one month or more, for either DS-TB or DR-TB, regardless of treatment outcome. This was done to allow for costs to be calculated for targeted interventions towards previously treated patients.
Intervention Scenarios
The Baseline scenario represents the status quo in both the Philippines and Thailand, where GeneXpert MTB/RIF is currently used as the initial diagnostic test for TB. In the context of this study, bacteriologically confirmed TB cases are defined as individuals testing positive for MTB using GeneXpert MTB/RIF.
Baseline use of GeneXpert MTB/RIF is compared with additional diagnostic testing in four intervention scenarios:
• Scenario 1: GeneXpert XDR for all bacteriologically confirmed TB cases;
• Scenario 2: tNGS for all bacteriologically confirmed TB cases;
• Scenario 3: GeneXpert XDR for bacteriologically confirmed, previously treated TB cases, and;
• Scenario 4: tNGS for bacteriologically confirmed, previously treated TB cases.
Scenarios 1 and 2 explore the use of an additional diagnostic test for improved detection of DR-TB, which include GeneXpert XDR and tNGS, respectively. Scenarios 3 and 4 explore targeted interventions for intensified DR-TB testing among previously treated patients, as this target group is known to have higher rates of DR-TB than new cases. The cost-effectiveness analysis is conducted assuming each of the scenarios is mutually exclusive.
For all scenarios, GeneXpert MTB/RIF is maintained as the initial diagnostic to determine the presence of MTB; additional use of GeneXpert XDR or tNGS is proposed for diagnosis of DR-TB in positively identified cases. Due to issues related to test specificity and applicability, GeneXpert XDR and tNGS are not recommended as stand-alone diagnostic tests [2].
Cost-effectiveness Analysis
Inputs used for health benefits and cost calculations are summarized in Table 1. Population level health estimates are expressed in disability adjusted life years (DALYs) without age weighing. To calculate years of life lost (YLL), TB related deaths are multiplied by the proportion of deaths in each five-year age group as per Global Burden of Disease (GBD) estimates[16], and by the average remaining life expectancy for that age group, assuming total life expectancy of 72 years and 80 years for the Philippines and Thailand, respectively [17]. To calculate years of life lived with disability (YLD), TB incidence is multiplied by the DALY disability weight for TB and the average duration of active TB infection. Average duration of active infection is the sum of the delay from developing active TB to receiving treatment (estimated at 3 months) [18, 19] plus the average duration of treatment depending on TB drug resistance (see Table 1). A mean DALY weight of 0.333 is used for all individuals with active TB, regardless of drug-resistance infection or treatment status, according 2021 GBD valuations [20]. DALYs and costs are calculated over a ten-year period (2025–2035), discounted at 5% annually in line with discounting rates appropriate for the economic context of low- and middle-income countries [21].
Table 1.
Parameters for health benefits and cost calculations
| Unit | Mean Philippines | S.D. | Mean Thailand | S.D. | Source | |
|---|---|---|---|---|---|---|
| Health Parameters | ||||||
| Expectation of life* | Years | 17.18 | 1.68 | 11.49 | 2.22 | [16] |
| Delay from symptom onset to diagnosis | Years | 0.15 | 0.060 | 0.15 | 0.060 | [18, 19] |
| Time from diagnosis to treatment, DS-TB* | Years | 0.014 | 0.004 | 0.014 | 0.004 | [24] |
| Time from diagnosis to treatment, DR-TB | Years | 0.019 | 0.003 | 0.019 | 0.003 | [24] |
| Duration of treatment DS-TB* | Years | 0.42 | 0.043 | 0.42 | 0.043 | [23] |
| Duration of treatment DR-TB | Years | 0.63 | 0.250 | 0.63 | 0.250 | [23] |
| Parameters for Diagnostic Costs | ||||||
| Sputum testing (x3)* | $USD | 36.95 | 24.47 | 6.22 | 0.62 | [25, 26] |
| Sputum culture | $USD | 31.07 | 2.12 | 16.59 | 1.66 | [25, 26] |
| Culture with DST | $USD | 110.96 | 13.87 | 27.65 | 2.77 | [25, 26] |
| Diagnostic visit | $USD | 3.98 | 2.36 | 2.08 | 0.35 | [25, 26] |
| TB cases receiving rapid test | Percent | 76 | 7.60 | 75 | 7.50 | [1] |
| GeneXpert MTB/RIF | $USD | 18.77 | 1.25 | 24.34 | 2.43 | [25, 26] |
| GeneXpert XDR | $USD | 23.98 | 2.40 | 55.31 | 5.53 | [25, 26] |
| Targeted NGS | $USD | 150.00 | 15.00 | 150.00 | 15.00 | Table S2 |
| Parameters for Treatment Costs | ||||||
| Outpatient treatment* | $USD/visit | 3.15 | 2.02 | 2.08 | 0.35 | [25, 26] |
| Follow up visits, DS-TB | Number | 6 | 0.75 | 12 | 1.20 | [1] |
| Follow up visits, DR-TB | Number | 13 | 1.63 | 23 | 2.30 | [1] |
| Inpatient bed-day* | $USD/day | 37.18 | 12.51 | 17.98 | 3.46 | [25, 26] |
| DS-TB cases hospitalized* | Percent | 3 | 0.30 | 30 | 3.00 | [1] |
| DR-TB cases hospitalized | Percent | 1 | 0.10 | 100 | 5.00 | [1] |
| Duration of DS-TB hospitalization* | Days | 5 | 0.63 | 14 | 1.40 | [1] |
| Duration of DR-TB hospitalization | Days | 10 | 1.25 | 30 | 3.00 | [1] |
| Treatment cost, DS-TB* | $USD | 47.00 | 4.70 | 47.00 | 4.70 | [23] |
| Treatment cost, DR-TB | $USD | 382.00 | 38.20 | 382.00 | 38.20 | [23] |
Expectation of life Average remaining life expectancy at age of TB related death, DS-TB Drug sensitive tuberculosis, DR-TB Drug resistant tuberculosis, NGS Next generation sequencing, SD Standard deviation
*These parameters were used in multivariable probabilistic sensitivity analysis
For the cost estimation, we adopted a health systems perspective to estimate total TB diagnostic and treatment costs. Conversion from local currency was conducted using an average exchange rate from Jan 2024 – June 2024 of 36.16THB and 56.92PHP to 1USD [22]. For GeneXpert MTB/RIF and XDR diagnostic tests, total costs include capital, consumables, overhead and staff costs, presented as 2025 $USD using 3.9% annual inflation adjustment (see Supplementary Material). Costs are assumed to remain the same between these tests except for consumables, as the Xpert XDR cartridge is higher in cost. For tNGS, an estimated cost of $150 per sample was used, which includes the cost of equipment (amortized over 10 years), reagents and other consumables, personnel, and sample transportation. Sequencing costs are based on Illumina Deeplex, a tNGS platform for TB commercially available in the Philippines and Thailand at the time of the study (see Supplementary Material). For all scenarios, the baseline cost of diagnosis with GeneXpert MTB/RIF is considered, with additional costs incorporated depending on use of GeneXpert XDR or tNGS.
Total diagnostic costs are the sum of the diagnostic test, the cost of the outpatient diagnostic visit, sputum culture for previously treated DS-TB cases and the cost of culture with drug susceptibility testing (DST) performed on positive DR-TB cases. Other additional screening costs or initial diagnostics such as chest X-ray are assumed to remain equal regardless of diagnostic test performed and proportion of DR-TB cases identified.
Treatment costs are the sum of the costs for outpatient treatment visits, inpatient costs, and anti-TB medication. These costs are considered for each different intervention scenario, as they vary according to the proportion of DR-TB cases identified. Frequency of health center visits and average hospitalization days for DS-TB and DR-TB patients are taken from WHO estimates [1]. Antibiotic treatment costs are differentiated between DS-TB cases and DR-TB cases. Anti-TB medication costs have been estimated from the Stop TB Partnership Global Drug Facility, 2025 list of prices for TB treatment regimens [23].
Incremental Cost Effectiveness Ratios (ICERs)
Strategies are compared using incremental cost effectiveness ratios (ICERs) that represent the additional cost of each strategy per DALY averted, relative to the previous least expensive alternative. A willingness-to-pay (WTP) threshold of 1 x GDP is used; equivalent to $4,150 for the Philippines and $7,530 for Thailand [27]. Strategies with an ICER below the WTP threshold are considered cost-effective. In addition, each strategy was mapped to the cost-effectiveness plane according to total cost and DALYs averted. The cost-effectiveness frontier is the line that connects all potentially cost-effective interventions (Fig. 2).
Fig. 2.
Cost Effectiveness Frontier. The cost effectiveness frontier is illustrated here as a line connecting all potentially cost-effective interventions. Scenarios 1– 4 are each plotted based on total costs and DALYs averted. Scenario 4 is dominated in both the Philippines and Thailand, falling below the cost-effectiveness frontier. XDR = GeneXpert XDR, NGS = next generation sequencing, PT = previously treated cases
Incremental Net Monetary Benefit (INMB)
The incremental net monetary benefit (INMB) is also used to compare intervention strategies. To calculate INMB, the incremental health benefit for each scenario (DALYs averted compared to baseline) is multiplied by the WTP threshold of 1 x GDP. This represents the total monetary benefit of the intervention, from which the incremental cost of the scenario from baseline is subtracted. INMB is calculated for 2025–2035, with 5% discounting for each year to determine net present value of DALYs and costs in 2025. A positive INMB indicates that a proposed strategy is cost-effective compared to the baseline at the specified WTP threshold (Fig. 3).
Fig. 3.
Incremental Net Monetary Benefit. Incremental net monetary benefit (INMB) is calculated using a willingness to pay threshold of 1 x GDP per capita. Probabilistic sensitivity analysis is conducted across 1000 runs to generate probability of highest NMB across scenarios. Mean NMB for each scenario is shown as a dotted line and zero intercept as a solid line
Using probabilistic sensitivity analysis across 1000 runs, the probability of each scenario having the highest net monetary benefit was calculated to identify which intervention is most likely to be cost-effective. Finally, INMB acceptability curves are calculated, showing the relative increase in net monetary benefit with increased WTP thresholds for each intervention (Fig. 4) ranging from $0 to $50,000.
Fig. 4.
Incremental Net Monetary Benefit Acceptability. Distribution of incremental net monetary benefit for each scenario is shown, according to varying willingness to pay (WTP) thresholds. Note: WTP threshold of 1 x GDP (2025) is equivalent to $4,150 for the Philippines and$7,530 for Thailand
Sensitivity analysis
Univariate sensitivity analysis was conducted on 37 study parameters; 13 model parameters (Supplementary Table 1) and 24 cost-effectiveness analysis parameters listed in Table 1. Variable ranges were determined from published literature where available, otherwise a range of +/- 20% was applied. Gamma distribution was used for cost parameters and normal distribution was used for model parameters, except for spontaneous self-cure and delay to treatment, which were fitted to gamma distribution.
From the results of the univariate sensitivity analysis, 15 parameters were identified that resulted in ≥5% variation in total DALYs and/or total costs for either the Philippines or Thailand.
Multivariable probabilistic sensitivity analysis was conducted using Monte Carlo simulation, with simultaneous variation of the 15 key parameters identified through univariate sensitivity analysis. The mean value across 1000 simulated runs was used as a point estimate for the Cost and DALY of the baseline and four implementation scenarios, as shown in Table 2. The incremental cost-effectiveness ratio (ICER) is calculated as the ratio of incremental costs and incremental DALYs (Cost/DALY) of each scenario compared to the previous least-expensive scenario. Results from the multivariate probabilistic sensitivity analysis were also used to generate 95% confidence intervals for TB Costs, DALYs and INMB (Table 2).
Table 2.
Cost-Effectiveness analysis of scenarios 1–4, compared with baseline
| Philippines | |||||
|---|---|---|---|---|---|
| Scenario |
Cost, $USD (±95% CI) |
DALYs (±95% CI) |
ICER, $USD per DALY averted |
INMB, $USD (±95% CI) |
Probability of highest NMB |
| Baseline | 581,488,494 (6,487,164) | 22,905,889 (249,576) | N/A | N/A | N/A |
|
Scenario 1 (XDR) |
727,208,032 (7,753,288) | 22,820,107 (248,742) | 1,808 |
210,273,006 (5.22E + 06) |
100% |
|
Scenario 2 (tNGS) |
1,048,447,976 (10,581,674) |
22,796,348 (248,512) | 13,521 |
−12,366,137 (5.95E + 06) |
0% |
|
Scenario 3 (XDR for PT) |
604,712,321 (6,718,665) | 22,887,857 (249,396) | 1,288 |
51,608,887 (1.20E + 06) |
0% |
|
Scenario 4 (tNGS for PT) |
636,861,476 (7,045,439) | 22,882,864 (249,346) | 6,440 (Dominated) |
40,178,305 (1.37E + 06) |
0% |
| Thailand | |||||
|---|---|---|---|---|---|
| Scenario |
Costs, $USD (±95% UI) |
DALYs (±95% UI) |
ICER, $USD per DALY averted |
INMB, $USD (±95% CI) |
Probability of highest NMB |
| Baseline | 92,795,736 (1,048,810) | 2,371,358 (31,187) | N/A | N/A | N/A |
|
Scenario 1 (XDR) |
141,198,459 (1,540,532) | 2,361,421 (31,028) | 5,251 |
26,418,011 (1.70E + 06) |
80% |
|
Scenario 2 (tNGS) |
183,637,948 (1,953,457) | 2,358,819 (30,987) | 16,311 |
3,570,222 (2.10E + 06) |
0% |
|
Scenario 3 (XDR for PT) |
101,520,138 (1,145,766) | 2,368,978 (31,149) | 3,667 |
9,192,245 (4.23E + 05) |
20% |
|
Scenario 4 (tNGS for PT) |
106,931,062 (1,202,688) | 2,368,353 (31,139) | 8,654 (Dominated) |
8,489,584 (5.20E + 05) |
0% |
ICERs are calculated comparing least expensive Scenario (Scenario 3) to Baseline, after which each subsequent Scenario is compared to the previous least expensive Scenario. For calculation of INMB, all Scenarios are compared to Baseline
DALY Disability adjusted life year, XDR GeneXpert XDR, tNGS Targeted next-generation sequencing, PT Previously treated cases, ICER Incremental cost effectiveness ratio, INMB Incremental net monetary benefit, NMB Net monetary benefit, CI Confidence interval
Results
Supplementing the current rapid diagnostic test (GeneXpert MTB/RIF) with an additional tool for detection of drug resistance, in this case GeneXpert XDR or tNGS, can improve DR-TB outcomes and increase DALYs averted, but at additional cost. The interventions that are applied to all bacteriologically confirmed TB cases (Scenario 1 for use of GeneXpert XDR and Scenario 2 for use of tNGS) have the highest impact on improving DR-TB outcomes. Targeting interventions to previously treated cases (Scenarios 3 and 4) significantly reduces costs but reduces the overall health impact.
Cost-effectiveness of different scenarios can be evaluated using incremental cost-effectiveness ratios (ICERs) as listed in Table 2. Scenario 1 is cost-effective in the Philippines (ICER = $1,808) and Thailand (ICER = $5,251), with an ICER below the 1 x GDP WTP threshold in both countries. Scenario 3 is also cost-effective in the Philippines (ICER =$1,288) and Thailand (ICER = $3,667) with ICERs below those of Scenario 1. Scenario 2 (tNGS) averts the highest number of DALYs but has ICER values above the WTP threshold for both countries and is therefore not considered cost-effective. Targeting tNGS to previously treated patients (Scenario 4) reduces costs but also reduces health impact, resulting in dominance by Scenario 1. This is illustrated by the location Scenario 4 outside the cost-effectiveness frontier (Fig. 2), indicating it is not cost-effective.
As a complement to ICERs, the incremental net monetary benefit (INMB) can give additional context for relative cost-effectiveness of interventions. Scenario 1 results in the highest mean INMB for both the Philippines (INMB = $210M) and Thailand (INMB = $26M), indicating that this intervention produces the highest monetary benefit or greatest value for money. Scenario 3 has the next highest mean INMB ($52M for the Philippines, $9.2M for Thailand). However, Scenario 1 has the highest probability of returning the highest NMB in both countries (Fig. 3), meaning it is the most likely to be cost-effective compared to all other scenarios. In the Philippines, Scenario 2 has a negative INMB ($-12M), indicating that it is not cost-effective compared to Baseline. In Thailand, Scenarios 2, 3 and 4 have overlap of INMB in 95% confidence intervals (CI), which signifies higher uncertainty for these results (see Table 2).
The INMB acceptability curves (Fig. 4) illustrate how higher WTP thresholds can result in changes to which interventions are considered cost-effective. At the WTP threshold of 1 x GDP, Scenario 1 (XDR) has the highest INMB for both the Philippines and Thailand. However, at higher WTP thresholds, Scenario 2 (tNGS) displays a steep increase in INMB. At a WTP threshold of ≥$14,000 in the Philippines and ≥$17,000 in Thailand, Scenario 2 is the intervention with the highest INMB, and therefore the highest value for money.
From the univariate sensitivity analysis, the top five parameters affecting the value for DALYs in both the Philippines and Thailand were the rate of slow activation, expectation of life, time to treatment for DS-TB, case detection rate and rate of progression to late latency. Regarding costs, slow activation rate and time to treatment for DS-TB resulted in the highest variation in TB costs for both countries. However, in the Philippines the sputum test cost and outpatient cost were more influential (>10% variation), whereas in Thailand the inpatient cost and inpatient duration for DS-TB resulted in larger variation from the mean (5 – 10%). For detailed univariate sensitivity analysis results see Supplemental Material.
Discussion
Study results have shown that the use of an additional diagnostic test for DR-TB detection, either GeneXpert XDR or tNGS, can lead to improved health outcomes, as measured in DALYs averted. These health benefits are the result of reductions in DR-TB treatment failure, mortality and incidence, as modelled in the Philippines and Thailand contexts. Scenarios 1 and 2, which implement DR-TB testing among all bacteriologically confirmed TB cases, demonstrate the highest health benefits compared to baseline. Targeting additional DR-TB diagnostic testing to previously treated cases (Scenarios 3 and 4) reduces the cost of the intervention, but also reduces the health benefits. The results of a cost-effectiveness analysis can help policy makers compare scenarios in more detail and weigh tradeoffs for informed decision making.
According to study results, interventions using GeneXpert XDR for DR-TB detection (Scenarios 1 and 3) are more likely to be cost-effective than interventions using tNGS (Scenarios 2 and 4), at least in the context of low-resource, high TB-burden settings. The use of GeneXpert XDR on all bacteriologically confirmed TB cases (Scenario 1) is cost-effective in both the Philippines (ICER = $1,808, INMB = $210 M) and Thailand (ICER = $5,251, NMB = $26 M). Targeted use of GeneXpert XDR on previously treated cases (Scenario 3) is also cost-effective in both countries: Philippines ICER = $1,288, NMB = $52 M; Thailand ICER = $3,667, NMB = $9.2 M. Although Scenario 3 is less expensive to implement, it has a lower health impact than Scenario 1 given its focus on only previously treated cases. Scenario 1 may be more expensive but considering the value of the health outcomes (e.g., DALYs averted), it produces the highest net monetary benefit across all scenarios. From a policy perspective, this suggests that while both Scenarios 1 and 3 are cost-effective, implementation of Scenario 1 will result in greater health benefits at the population level and highest value for money.
A surprising outcome from the cost-effectiveness analysis was the small incremental health benefit (DALYs averted) from the use of tNGS or Xpert XDR, despite the large increases in diagnosed and treated DR-TB cases. Several reasons contribute to this outcome; the main reason being that these diagnostic tools directly benefit only DR-TB cases, which represent a small proportion of overall TB cases in the Philippines and Thailand. In addition, introduction of the BPaL regimen (bedaquiline, pretomanid and linezolid), recommended by WHO in 2022, has reduced the average treatment duration for DR-TB to 6–9 months which is only slightly longer than for DS-TB (4–6 months) [28]. The shorter duration of DR-TB infection reduces overall years of life lost from disability (YLD) which in turn reduces population level health benefits from accurate diagnosis and treatment of DR-TB. Finally, TB mortality is experienced mainly in older age groups, which results in a low value for years of life lost due to premature mortality (YLL) when calculating DALYs. However, the model does not consider long term post-tuberculosis sequelae for previously treated TB patients which can result in higher disability and mortality from post-TB complications [29]. This omission may underestimate the health benefits of improved TB diagnostics on DALYs averted over time.
The study has several limitations, including the reliance on WHO data for health parameters such as number of follow up visits, proportion of cases hospitalized and duration of hospitalization. While the model itself can be modified to reflect the application of various scenarios to different country contexts, the results of this analysis are based on the disease burden and TB costs for the Philippines and Thailand. The model is conducted at a national level and does not differentiate between disease burden at the subnational level, between different geographical areas, or among different population groups. In addition, the costing analysis is conducted from the health system perspective and does not include costs from the patient perspective. The costs considered are primarily direct medical costs associated with health center visits, diagnostic and treatment costs and indirect medical costs associated with staff time. Although diagnosis and treatment in both the Philippines and Thailand are subsidized through public health care, there may be a significant financial burden borne by the patient from non-medical costs (e.g. transportation to health facilities) as well as indirect costs from loss of productivity and income [30–32]. Consideration of these additional costs could further increase the cost-effectiveness of proposed improvements in diagnostic testing. Finally, parametric sensitivity analysis highlights uncertainty of results given the wide confidence intervals for DALYs, costs and INMB per intervention. More accurate country level TB diagnostic and treatment costs are required to better inform policy development and decision making.
Despite its limitations, this study has several policy implications which can help inform future DR-TB testing policy in the Philippines and Thailand, as well as in low-resource, high TB-burden areas globally. The benefits of GeneXpert as an established platform, highest probability for cost-effectiveness and ease of use as a point-of-care diagnostic, contribute to supporting widespread application of GeneXpert XDR as an additional diagnostic tool for DR-TB detection in high-burden settings. However, it is important to consider that GeneXpert XDR does not detect resistance to antibiotics in the BPaL regimen. Although resistance to drugs in the BPaL and BPaL-containing regimens is low in the Asia region given its recent introduction, the lack of an effective and routine test for resistance to these new anti-TB drugs is concerning and poses potential risk for undetected spread.
The study findings indicate that tNGS remains too expensive to be applied as a first-line DR-TB test in high-burden settings. As shown in Fig. 2, Scenario 4 (tNGS for PT) is dominated by Scenario 1 (XDR), while Scenario 2 (tNGS) is located at the flattened area of the cost-effective frontier, indicating diminishing health returns for increased cost. Costs for tNGS in this study are based on the Illumina platform, as opposed to other platforms such as Oxford Nanopore, given the more widespread use of Illumina machines at the time of the study. Interestingly, if a reduced cost of $50 per sample is applied for tNGS, Scenario 4 is no longer dominated by Scenario 1, and can be considered a cost-effective intervention in both the Philippines and Thailand. Considering recent advances in tNGS technology and anticipated reductions in cost, $50 per sample is a realistically achievable target. As a result, application of NGS toward high-risk populations such as previously treated cases becomes a much more viable policy alternative for future implementation.
As LMICs, the 1 x GDP per capita WTP threshold used for the Philippines ($4,150) and Thailand ($7,530) limit the interventions that can be considered cost-effective. However, from the INMB acceptability curves in Fig. 4, Scenario 2 demonstrates the highest increase in monetary benefit with increased WTP thresholds. This can help explain why high-income countries use tNGS for detection of DR-TB; a higher WTP threshold would support the cost-effectiveness of these more expensive interventions.
Tuberculosis has the potential to be at the forefront for application of genomic technologies. Given the changing cost landscape for tNGS, it will be important to periodically re-evaluate cost-effectiveness of diagnostic interventions using this technology. tNGS faces other barriers to implementation such as high infrastructure and equipment costs, training needs for skilled human resources, and complicated workflows that take place in central laboratories. Nevertheless, despite the current cost and implementation barriers for clinical use, tNGS has many other valuable applications for outbreak investigation and pathogen surveillance.
In conclusion, the study has found that high TB burden countries would benefit from implementing a rapid molecular diagnostic test such as GeneXpert XDR, ideally for all bacteriologically confirmed TB-cases, or at a minimum for high-risk populations such as previously treated cases. At the same time, new technologies such as tNGS are becoming more accessible as costs continue to fall, and show potential for targeted use among high-risk populations, or as an effective tool to protect the introduction of new TB treatment regimens such as those containing BPaL. These findings should encourage the reduction of tNGS costs for LMICs to make this technology more accessible in high-burden settings. Investment in new technologies for TB control and prevention of drug-resistance can have widespread effects, contributing to regional and broader global health security.
Supplementary Information
Acknowledgements
The authors would like to thank the National TB Control Program of the Disease Prevention and Control Bureau of the Philippine Department of Health for their support and provision of data for the model.
Patient and public involvement
Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Authors’ contributions
MG contributed to the conceptualization, investigation, methodology, formal analysis, validation, writing the original draft, and review & editing. JPA contributed to conceptualization, formal analysis, methodology, supervision and review. DL, RB, FT and CC contributed to investigation, methodology, validation and review. DM contributed to conceptualization, methodology, supervision and review. All authors reviewed and approved the final version of the manuscript.
Funding
No funding was received for conducting this study.
Data availability
For modelling scenarios, publicly available TB data from WHO Global Tuberculosis Report 2024 was used: (https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/data). Additional costing estimates are available in Supplemental Material.
Declarations
Ethics approval and consent to participate
The study was granted exemption from IRB review from the Duke-NUS Departmental Ethics Review Committee (Reference: DERC-19-231205).
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Global Tuberculosis Report 2024. (World Health Organization, Geneva, 2024). Licence: CC BY-NC-SA 3.0 IGO. https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2024
- 2.WHO consolidated guidelines on tuberculosis. Module 3: diagnosis - rapid diagnostics for tuberculosis detection, third edition. (World Health Organization, Geneva, 2024). https://www.who.int/publications/i/item/9789240089488 [PubMed]
- 3.Gegia M, Winters N, Benedetti A, van Soolingen D, Menzies D. Treatment of isoniazid-resistant tuberculosis with first-line drugs: a systematic review and meta-analysis. Lancet Infect Dis. 2017;17:223–34. 10.1016/S1473-3099(16)30407-8. [DOI] [PubMed] [Google Scholar]
- 4.Zurcher K, et al. Drug susceptibility testing and mortality in patients treated for tuberculosis in high-burden countries: a multicentre cohort study. Lancet Infect Dis. 2019;19:298–307. 10.1016/S1473-3099(18)30673-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Karmakar M, Ragonnet R, Ascher DB, Trauer JM, Denholm JT. Estimating tuberculosis drug resistance amplification rates in high-burden settings. BMC Infect Dis. 2022;22:82. 10.1186/s12879-022-07067-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Romanowski K, et al. The impact of improved detection and treatment of Isoniazid resistant tuberculosis on prevalence of multi-drug resistant tuberculosis: a modelling study. PLoS ONE. 2019;14:e0211355. 10.1371/journal.pone.0211355. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pillay S, et al. Xpert MTB/XDR for detection of pulmonary tuberculosis and resistance to isoniazid, fluoroquinolones, ethionamide, and Amikacin. Cochrane Database Syst Rev. 2022;5(CD014841). 10.1002/14651858.CD014841.pub2. [DOI] [PMC free article] [PubMed]
- 8.Mugwagwa T, Abubakar I, White PJ. Using molecular testing and whole-genome sequencing for tuberculosis diagnosis in a low-burden setting: a cost-effectiveness analysis using transmission-dynamic modelling. Thorax. 2021;76:281–91. 10.1136/thoraxjnl-2019-214004. [DOI] [PubMed] [Google Scholar]
- 9.Getchell M, et al. Pathogen genomic surveillance status among lower resource settings in Asia. Nat Microbiol. 2024;9:2738–47. 10.1038/s41564-024-01809-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Murphy SG, et al. Direct detection of drug-resistant Mycobacterium tuberculosis using targeted next generation sequencing. Front Public Health. 2023;11:1206056. 10.3389/fpubh.2023.1206056. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Use of targeted next- generation sequencing to detect drug-resistant tuberculosis: rapid communication, July 2023. (World Health Organization). https://www.who.int/publications/i/item/9789240076372
- 12.de Araujo L, et al. Implementation of targeted next-generation sequencing for the diagnosis of drug-resistant tuberculosis in low-resource settings: a programmatic model, challenges, and initial outcomes. Front Public Health. 2023;11:1204064. 10.3389/fpubh.2023.1204064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cates L, et al. Budget impact of next-generation sequencing for diagnosis of TB drug resistance in Moldova. Int J Tuberc Lung Dis. 2022;26:963–9. 10.5588/ijtld.22.0104. [DOI] [PubMed] [Google Scholar]
- 14.Shrestha S, Addae A, Miller C, Ismail N, Zwerling A. Cost-effectiveness of targeted next-generation sequencing (tNGS) for detection of tuberculosis drug resistance in India, South Africa and Georgia: a modeling analysis. EClinicalMedicine. 2025;79:103003. 10.1016/j.eclinm.2024.103003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Getchell M, et al. Dynamic modelling of improved diagnostic testing for drug-resistant tuberculosis in high burden settings. BMC Infect Dis. 2024;24:1247. 10.1186/s12879-024-10027-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Institute for Health Metrics and Evaluation (IHME). GBD Results. Available from https://vizhub.healthdata.org/gbd-results/. (Seattle, WA.: IHME, Univeristy of Washington, 2024).
- 17.World Bank. World Development Indicators, <https://wdi.worldbank.org/table/2.18>(2022).
- 18.Cai J, et al. Factors associated with patient and provider delays for tuberculosis diagnosis and treatment in Asia: a systematic review and meta-analysis. PLoS ONE. 2015;10:e0120088. 10.1371/journal.pone.0120088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Teo AKJ, Singh SR, Prem K, Hsu LY, Yi S. Duration and determinants of delayed tuberculosis diagnosis and treatment in high-burden countries: a mixed-methods systematic review and meta-analysis. Respir Res. 2021;22:251. 10.1186/s12931-021-01841-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2021. (GBD 2021) Disability Weights. Institution for Health Metrics and Evaluation (IHME). https://ghdx.healthdata.org/record/ihme-data/gbd-2021-disability-weights. (Seattle, United States of America, 2024).
- 21.Haacker M, Hallett TB, Atun R. On discount rates for economic evaluations in global health. Health Policy Plan. 2020;35:107–14. 10.1093/heapol/czz127. [DOI] [PubMed] [Google Scholar]
- 22.International Monetary Fund. Exchange Rate Archives by Month, https://www.imf.org/external/np/fin/data/param_rms_mth.aspx (2024).
- 23.Stop TB, Partnership, Global Drug Facility. GDF Estimated Prices for Select TB Treatment Regimens. (https://www.stoptb.org/what-we-do/facilitate-access-tb-drugs-diagnostics/global-drug-facility-gdf/buyers/plan-order Accessed 25 January 2025).
- 24.National Tuberculosis Control Program, Manual of Procedures, 6th edition. (Department of Health, Philippines, 2020). https://ntp.doh.gov.ph/download/ntp-mop-6th-edition/
- 25.Capeding TPJ, et al. Cost of TB prevention and treatment in the Philippines in 2017. Int J Tuberc Lung Dis. 2022;26:392–8. 10.5588/ijtld.21.0622. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Laboratory Service Manual [คู่มือบริการตรวจทางห้องปฏิบัติการ 2567]. Office of Disease Prevention and Control, Region 11, Nakhon Si Thammarat. Department of Disease Control, Ministry of Public Health, Thailand. (2024).
- 27.International Monetary Fund. GDP per capita, current prices, https://www.imf.org/external/datamapper/NGDPDPC@WEO/THA/IDN/PHL/VNM/MYS (2024).
- 28.WHO consolidated guidelines on tuberculosis. Module 4: treatment and care. World Health Organization; 2025. https://www.who.int/publications/i/item/9789240063129 [PubMed]
- 29.Menzies NA, et al. Lifetime burden of disease due to incident tuberculosis: a global reappraisal including post-tuberculosis sequelae. Lancet Glob Health. 2021;9:e1679–87. 10.1016/S2214-109X(21)00367-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Youngkong S, et al. Catastrophic costs incurred by tuberculosis affected households from Thailand’s first national tuberculosis patient cost survey. Sci Rep. 2024;14:11205. 10.1038/s41598-024-56594-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Portnoy A, et al. Costs incurred by people receiving tuberculosis treatment in low-income and middle-income countries: a meta-regression analysis. Lancet Glob Health. 2023;11:e1640–7. 10.1016/S2214-109X(23)00369-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Florentino JL, et al. Expansion of social protection is necessary towards zero catastrophic costs due to TB: the first National TB patient cost survey in the Philippines. PLoS ONE. 2022;17:e0264689. 10.1371/journal.pone.0264689. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
For modelling scenarios, publicly available TB data from WHO Global Tuberculosis Report 2024 was used: (https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/data). Additional costing estimates are available in Supplemental Material.