Table 1.
Key regimen properties to be modelled, as relevant to the BMGF product development portfolio.
We modelled the impact of such an ‘idealised’ pan-TB regimen in India; we also modelled the impact of a ‘less-than-ideal’ future regimen, that meets all but one of these criteria.
Table 2.
All costs were subject to +/- 20% uncertainty, captured by a lognormal distribution. Sources: Global Health Cost Consortium [19], unless indicated otherwise.
Fig 1.
Projected impact of a pan-TB regimen in India.
(A, B) Projections for all TB incidence (panel A) and RR-TB incidence (panel B), assuming that the regimen is introduced in 2022 and scaled up in a linear way over the subsequent three years. Table 3 shows the percent cases averted under the different deployment scenarios shown here. (C) Projected impact on ‘conventional’ programmatic spending (i.e. those not associated with the new regimen), under a pan-TB deployment scenario of universal indication. Diagnosis costs include diagnostic tests (smear microscopy as well as drug sensitivity testing with Xpert and culture in a certain proportion of cases). First- and second-line treatment include the costs of conventional drugs, as well as the staff costs involved in counselling and patient adherence support. Shaded areas show 95% credible intervals, while solid lines show median estimates. See Table 2 for details of cost inputs.
Table 3.
Summary of epidemiological and budget impact of a pan-TB regimen under the different scenarios modelled her.
Fig 2.
Projected incremental cost to the India national TB programme between 2022 and 2030, from implementation of a pan-TB regimen, for different assumed values for the regimen cost (x-axis).
The right-hand panel shows a close-up of the left-hand, at lower regimen costs. The horizontal, dashed line shows the level at which a pan-TB regimen would be overall cost-neutral to the programme. Shaded areas show 95% credible intervals, while solid lines show median estimates. If a pan-TB regimen is used as a replacement for current second-line therapy, it will be cost-saving as long as it costs less than USD 6386 (95% CrI 5849–6980) (left-hand panel, point ‘A’). If the regimen cost could be reduced to below USD 358.5 (95% CrI 286.7–441.4) per regimen, then it is more cost-saving to deploy with a universal indication, than with a RR-only indication (point ‘B’).
Fig 3.
Influence of different regimen properties.
Numbered 'properties' on the x-axis correspond to those listed in Table 1. As described in the main text, we first calculated the 'impact' X of an idealized regimen (percent cases averted between 2022 and 2030), one satisfying all the conditions listed in Table 1 (see Table 3 for impact estimates). We then simulated the impact Y of a 'less-than-ideal' (LTI) regimen: for any given regimen property of interest, we then quantified its 'attributable impact' as (X-Y)/X: that is, the proportional reduction in incidence impact, that results from a regimen failing to meet this criterion. Thus, higher bars represent properties that are more important for epidemiological impact. Red bars show an LTI regimen that meets only one of the regimen requirements in Table 1, while blue bars show an LTI regimen that meets all but one of the regimen requirements.
Fig 4.
Model projections under an alternative baseline (private sector engagement), of initiatives to engage with 85% of private healthcare providers in India between 2019 and 2022.
We assume that engaged private providers have their quality of TB care (accuracy of diagnosis, levels of DST and treatment success) raised to the same standard as in the public sector (see also S2 Table in S1 Appendix); and further that the pan-TB regimen will be deployed amongst all engaged private providers in addition to the public sector. Upper panels show epidemiological impact, as in (Fig 1A and 1B). Lower panels show incremental programmatic spending for a range of scenarios for regimen cost, as in Fig 2.
Fig 5.
Sensitivity analysis of threshold regimen cost, in order for a pan-TB regimen to be cost-saving to the programme, in relation to different assumptions for regimen characteristics.
We model ‘all-but-one’ regimen scenarios, as in Fig 3, but here focusing on the most prominent attributes in Fig 3, along with regimen duration. While the latter is assumed to be 2 months in the main analysis, here we consider a regimen of 4 months duration. Results suggest that—although threshold regimen costs appear relatively robust to these different characteristics, the one exception is RR-TB cure rate under a universal indication. For this attribute, a decrease to 50% (from 85%) is sufficient to bring the threshold cost down by roughly half (that is, posing a more stringent condition for budget neutrality). These results highlight that the economic value of a pan-TB regimen would derive largely from its ability to benefit patients with RR-TB, who might otherwise receive inappropriate therapy.