Thread by @tomayates

Some thoughts about the Nix-TB paper - nejm.org/doi/full/10.10….

First, it is great that these 109 patients had such good clinical outcomes and that the substantial reported toxicity seemed to be, mostly, reversible.

[1/n]

There are reasons to doubt such good outcomes would be seen using the regimen in routine practice.

The Nix-TB patients had survived a median of 12m since DR TB diagnosis, with median BMI in the normal range, fairly good Karnofsky scores and 15% culture -ve at baseline.

[2/n]

Compare this with the first XDR cases reported at Tugela Ferry, where 52/53 died with a median time to death of sixteen days.

Remember, we saw 80% treatment success with old WHO MDR regimen in the STREAM 1 trial vs 50% in routine practice.

[3/n]

Patients in clinical trials do better for a number of reasons. These include, selection effects - patients who die at 16 days are less likely to make it into clinical trials. Also, patients in clinical trials often benefit from better clinical care.

[4/n]

The editorial by @ThwaitesGuy & Payam Nahid is spot on

nejm.org/doi/full/10.10…

'we should not be dependent on one small, single-group, single-country study...such approaches for the development of tuberculosis regimens do not correspond with the magnitude of the problem'

[5/n]

I find the weight inadequate observational studies have carried in decision making in recent years concerning.

See, for example, @docwak77's important critique of the Collaborative Group individual patient meta-analysis of MDR-TB treatment

thelancet.com/journals/lance…

[6/n]

Or my critique of Schnippel's paper on mortality with BDQ containing regimens.

researchgate.net/publication/32…

Determinants of treatment success are complex, incompletely understood & usually not well measured. This makes it hard to adequately adjust for them.

[7/n]

The Nix-TB team were right to segue quickly into ZeNix, an RCT which will yield really useful data about linezolid dosing.

We need to ramp up DR TB RCT capacity, so we can quickly work out how best to use novel drugs in regimens. Funders need to step up to the plate.

[8/n]

Also regulations need to be amended. It makes no sense that experimental regimens can be used more readily in cohorts/clinical practice, whereas the use of the same regimen in an RCT results in substantial additional bureaucracy, slowing things down and increasing costs.

[9/n]

There is a role for cohorts/observational research, e.g. looking for rare side effects post RCT. But the correct study design for robustly evaluating the relative safety/efficacy of TB treatment regimens is the RCT. We do TB patients no favours by taking short cuts.

[10/10]