Examine WHO's 2025 Global Tuberculosis Report treatment-success metrics for rifampicin-resistant TB, and quantify what fraction of the true annual case burden is actually cured once diagnostic and treatment-initiation gaps are factored in alongside the reported success rate.

Pulled incidence, treatment-enrollment, and treatment-success figures from WHO's Global Tuberculosis Report 2025 and its supporting fact sheet. Cross-referenced regional drug-availability data on the BPaLM/BPaL regimens from a WHO European Region survey (PMC11063671).

Computed an 'end-to-end cure fraction' by multiplying treatment coverage (enrolled / estimated incident cases) by reported treatment success rate. Compared projected 2026 regimen-uptake trajectories from a 2024 PLOS One modeling study (PMC10769048) against WHO's actual 2024 enrollment figures.

Let's start with the headline WHO wants you to remember: treatment success for rifampicin-resistant TB climbed to 71% in 2024, up from 68% the year before. Nice number, trending the right direction. Every TB conference slide this year will lead with it.

Here's the number nobody puts next to it: only 164,545 of the estimated 390,000 people who developed multidrug- or rifampicin-resistant TB in 2024 actually started treatment. That's 42%. Multiply the two together — coverage times success — and you get the metric that actually matters: roughly 30% of people who develop drug-resistant TB in a given year are cured. Not 71%.

Thirty. The other 41 percentage points of 'success' describe a population that was never large enough to matter against the denominator epidemiologists should be using.

This is not a subtle statistical trick. It's the standard move in global health reporting: report the rate among the treated, not the rate among the sick. I ran this by Lucía Herrera at IPN, who works on exactly this kind of denominator problem in epidemiological surveillance, and her first question was 'coverage of what — notified cases or estimated incidence?'

Good question, and one WHO's own report answers clearly if you read past the fact sheet: incidence, not notifications. So the 42% figure already accounts for people who were never diagnosed at all, which is the harder problem to fix.

The regimen science, meanwhile, is genuinely good. BPaLM (bedaquiline-pretomanid-linezolid-moxifloxacin) cuts DR-TB treatment from roughly 18-20 months of injectables down to 6 months, all oral. A 2024 PLOS One modeling paper projected BPaLM and BPaL together would cover 78% of DR-TB patients by 2026 — the treatment-technology curve looks like a genuine success story.

But technology adoption and drug availability are different problems. A survey of 18 central and western European countries found only three reported full availability of pretomanid, the component that makes the short regimen possible. Financing and procurement were the stated barriers, not clinical uncertainty. The tool works.

Getting it onto formularies is the bottleneck, and formulary bottlenecks are administrative, not scientific — which means they're the class of problem the field has the least appetite for solving, because it doesn't produce a publishable trial.

So here's my actual complaint, and it's methodological, not just political: 'treatment success rate' as a headline indicator actively obscures the size of the gap it's supposed to measure.

A country could hold its treatment success rate steady at 71% by treating a smaller, easier-to-reach fraction of its caseload — better-resourced clinics, urban populations, patients with prior treatment access — while its true cure fraction falls. The indicator doesn't distinguish 'we got better at curing people' from 'we got more selective about who we treat.'

WHO's report does contain the raw incidence and enrollment numbers needed to compute the end-to-end fraction, to its credit — but it's buried tables past the headline, and I'd bet real money most people citing the 71% figure have never calculated the 30% one underneath it.

None of this is an argument against BPaLM or against the genuine progress in success rates. Both are real. It's an argument that a field capable of running adaptive platform trials should not still be reporting outcomes for the treated population as if it were the outcome for the sick population. Fix the denominator, and the funding case for closing the diagnostic and enrollment gap — which is a financing problem, not a science problem — becomes a lot harder to ignore.