Hey Spencer!

From the 2022 South Africa paper, it appears that the bedaquiline-based regimen actually consists of 8 different drugs (see table S1), with a total cost per treatment of $6402 in the base-case.  It's not clear to me how much each drug contributes to the total cost, but you should be able to work this out from the regimen info from table S1 and the drug cost data from the medicines catalog  (from reference 24 of the paper). Presumably if you've done it right you should end up with ~$6,402. Then you can just tweak the cost of bedaquiline (I'm assuming no other drugs have also changed price). 

If you manage to do this, then quite plausibly your updated cost per treatment will fall within the range of the paper's one-way sensitivity analysis ($5,122 to $7,682), i.e. Figure 2. From this you can eyeball what the new ICER will be:

Alternatively, you could try reaching out to the lead author - assuming they still have their model to hand, it should be very easy for them to adjust the cost of BDQ and see what the model outputs :)

Also, just to caution, the South Africa paper is "from the perspective of the South Africa national healthcare provider" and so uses the GDP per capita of South Africa as the willingness-to-pay threshold, and some of the health and cost data is also South Africa-specific. So I don't think this can easily be used to infer the change in cost-effectiveness in other settings (though it might give a rough indication of whether or not there could be a substantial change).

That's great news thanks for the update! Its fantastic  how big pharma has recently been throwing some pretty big bones to the not-for-profit world to  manage TB, with this and the vaccine now in control of non-profits

Overseeing the production and delivery of drugs and vaccines is a big step for non profits, and this is going to be an important tests whether they can do it or not - I hope it works out really well!

In terms of cost-effectiveness, from an EA perspective at least the major benefit of treating MDR/XDR TB probably comes from reducing the chance of a disastrous strain of TB which could readily spread and be basically untreatable, rather than just looking a short distance ahead like this paper did.

"All three prior analyses in South Africa were performed over a long (10 years) time horizon to capture the downstream cost and DALYs of multidrug-resistant TB-related treatment failure"

Looking 10 years ahead is actually pretty impressive, but the cost-effectiveness analysis could look a lot more impressive if they looked 100 years ahead, and also considered low probablility catastrophic outcomes too. Indeed it would be very difficult to calculate with huge errors, but not as hard as things like AI cost-effectiveness calculations! Unfortunately we won't see that kind of analys much in peer-reviewed papers, because it wouldn't get accepted like it would here on the forum ;)

Reducing the cost of the drugs is great, and might have an incremental effect on cost-effectiveness, but perhaps from my (and a more 
"EA perspective" the biggest potential benefits here are that if we can treat more people with MDR/XDR, we have a better chance of preventing catastrophe later.