Discussion
To our knowledge, this is the first study to empirically assess costs of health service components for XDR-TB and quantify the budget impact of switching to the recently FDA-approved pretomanid-containing BPaL regimen in three geographically diverse high DR-TB burden countries. On a perpatient basis, the BPaL regimen can be two-to-five fold cheaper to treat patients with XDR-TB compared with the conventional regimens, assuming full adherence to the respective care paths outlined in the respective NTP guidelines. We showed that gradual adoption of BPaL would result in an average reduction of between 15% and 32% in budgets required to manage patients with XDR-TB in the respective countries.
Of all the health systems service components, BPaL drug costs constituted the largest contributor to the overall cost-savings. Across the three countries included in this study, using BPaL would result in at least 57% (and as high as 90%) reduction in per-patient drug costs compared with the current regimens to treat XDR-TB. This was primarily due to a reduction in the number of drug types and shortened duration of treatment for the BPaL regimen compared with conventional regimens. Furthermore, procurement prices of key drugs used to treat XDR-TB largely contributed to the difference in cost-savings across the three countries. For example, the unit cost used for one tablet of bedaquiline was US$5.71 in Indonesia, which is more than two times as high as the price charged through GDF. Similarly, the unit cost of one tablet of linezolid was US$6.39, which is more than five times higher than the price charged through GDF. In Indonesia, it is anticipated that these key drugs for XDR-TB treatment will be not procured through GDF for the foreseeable future. Likewise, Indonesia had highest per-patient cost of XDR-TB treatment using BPaL (US$4559), resulting in lowest absolute cost-savings compared with other countries
Another notable contributing factor to the cost-savings associated with the BPaL regimen was the reduction in health service utilisation required to manage treatment of patients with XDR-TB. If the BPaL regimen would be used in the three countries, we anticipate that the number of visits to clinics for outpatient consultation, number and types of patient safety and treatment monitoring tests would be dramatically reduced due to simplified standardised drug regimen and a 14-month reduction in treatment duration compared with the conventional XDR-TB treatment course. Furthermore, if factoring in programmatic (eg, simplified procurement and supply chain management) and operational (decentralisation of XDR-TB treatment) benefits of the simplified and standardised treatment regimen, we expect that the economic case for adopting BPaL regimen would become more favourable.
In our sensitivity analyses, we showed that the average costs per BPaL treatment completed in Indonesia were most sensitive to halving the dosage of linezolid, which showed to be efficacious and more tolerable in the ZeNIX trial.26 Prescribing BPaL to the other WHO-recommended patient populations, patients who are either unable to tolerate or failed MDR-TB treatment would increase the reduction in the net budget (figure 2B). Increasing the speed of BPaL roll-out would reduce the XDR-TB-related expenditure, particularly when the proportion of patients with XDR-TB being enrolled on BPaL would increase over the years.
Our findings should be interpreted in light of the following limitations. First, because the BPaL regimen is a novel regimen that has not yet been widely adopted or studied in large scale, we primarily relied on the data available from the Nix trial to populate the BPaL treatment parameters in the model. Some parameter values in the model may, therefore, have been optimistic in a simplified model structure that may not fully capture the complexities of the XDR-TB patient care. For example, in our study, we used 1.4% LTFU rates of BPaL regimen as reported in the Nix trial. In reality, LTFU rates may be higher, and this would result in higher cost estimate per patient completing BPaL treatment, reducing the overall cost-savings associated with the introduction of BPaL regimen. Second, while we accounted for the impact of adverse events on treatment outcomes and overall treatment costs, we assumed types of adverse events, and management of adverse events would be similar between the two regimens (eg, 10% of the patients would require hospitalisation due to myelosuppression for both regimens). As such, we did not assess costs specific to managing adverse events, resulting from respective XDR-TB treatment. If frequency and types of adverse events associated with the programmatic use of BPaL regimen are higher compared with the conventional regimen, we expect that the overall cost-savings for BPaL regimen will also subsequently be reduced. While uncertainties around these parameters did not impact our overall cost-saving and budget impact estimates for BPaL, ‘real-world’ cost implications may be more significant on overall costs associated with the introduction and use of BPaL regimen. These factors are being evaluated in on-going operational research projects in various settings by the TB Alliance and KNCV.
Second, the overall cost and budget estimates for BPaL introduction were estimated based on the anticipated number of patients who will be initiated on the BPaL regimen in the respective years between 2020 and 2024 in each country. As our study was done prior to the FDA approval, we took a conservative approach in estimating these numbers with the key stakeholders from the NTP in the respective countries. Likewise, if countries take more rigorous and inclusive approach to introducing the BPaL regimen, we expect that the overall cost-savings and budget impact be greater than what was projected in our analyses. Third, in our budget impact analyses, we did not consider initial diagnostic costs and the costs associated with the implementation when transitioning to the novel regimen. While we expect that initial diagnostic process will not change for the decision to initiate patients BPaL, if the diagnostic process becomes simplified for BPaL, this would further favour adoption of BPaL regimen. Furthermore, while we expect that the costs associated with the implementation of the new regimen are an important factor, if the regimen can be scaled-up and maintained for the longer term, these costs will be marginalised.28 However, these implementation costs will vary considerably depending on the operational conditions, training needs, coverage and speed of implementation to the lower levels of health systems. Therefore, we encourage future studies to thoroughly investigate programmatic and implementation costs for introducing new treatment regimens for TB.28 Finally, as our analyses were restricted to the health service provider perspective, we did not factor potential patient benefits and cost that could result from the simplified treatment regimen for XDR-TB. We encourage future studies to empirically assess patient perspective costs and benefits of simplified standardised regimens for DR-TB.11