Also a reason we need nationwide state medical licensing reform. The cross-state forgery in this case would have been much more difficult to achieve with uniform, nationwide licensing of physicians and physicians assistants. Plus this would limit abusers crossing state lines to avoid discipline.

Medical licensing can be a frustrating process and there are plenty of bugs in the system, but stories like this really drive home why we need regulation of the medical profession. Turns out the “PA” in this case NEVER attended PA school (she also unsurprisingly didn’t show up to the hearing…)

Such complex modeling strategies and assumptions are directly my area of research expertise haha. The blog post is very nice and makes excellent points, but is not really relevant to what I’m trying to say here. Regardless, appreciate your engagement on this topic!

After all, one can calculate an NNT, but even in a magical world with perfect applicability/generalization, no one knows which is the 1/N patient to benefit. I feel that individual decision making isn’t a question of ARR vs RRR- its a process of updating priors on the basis of patient AND trial data

The question I am more deeply interested in is whether a trial’s risk reduction (for me, I prefer ARR) generalizes to the clinical setting (and population) in question. For individual decision-making, I think inevitably the decision is often driven by other, non-trial data.

You make an interesting argument. For me, as a public health physician, I tend to separate individual risk-benefit from population risk-benefit. Across an entire population, HTE (within a trial population) shouldn’t matter (on balance) but for the individual heterogeneous patient of course it does.

The approach described above almost inevitably overestimates the magnitude of benefit by taking population-level base-rate risk (almost always higher than clinical trials due to survivorship) and comparing it to RCT RRR (higher magnitude of benefit from therapy than seen in gen. pop) (4/4)

RRR is not “more robust” by any defensible statistical definition. In fact, empirical data from as far back as 1990s shows that physicians over-estimate magnitude of benefit when presented with RRR instead of ARR, hence CONSORT requirement to report both pubmed.ncbi.nlm.nih.gov/1443954/ (3/4)

That counterfactual comparison is only valid from its sample population, which usually is directly and explicitly NON-comparable to the pop used for an absolute risk tool (due to inclusion/exclusion criteria) (2/4)

I see, thanks for explaining. Have to hard disagree- absolute risk from population-based risk prediction tools should NOT be directly combined with relative risk reductions from clinical trials. RRR from an RCT derives from a counterfactual comparison (1/4)

Seems to me you are engaging in bad faith argumentation against a strawman that I or the other folks on this thread are interested in withdrawing treatment of hypertension from older adults. But no one in this thread has advocated that (again, I literally said the opposite in my first post).

Still haven’t answered my question. What RCT data do you use in “shared decision making” if not ARR? I don’t need the lecture on the limitations of NNT, but you seem to be dodging my actual question. Also, who said anything about deprescribing? I have advocated treatment this whole thread.

My apologies, I was not trying to say that. Trying to understand: NNT is derived directly from key trial data via 1/ARR, so if you aren’t interested in that specific trial data, what data do you use to guide shared decision making? Are you opposed to the general concept of absolute risk reduction?

Gonna have to disagree with you there. Relying on data about benefits and harms is clearly better medicine than ignoring that data entirely to avoid “dichotomania.” Not sure what alternative you are suggesting other than to reject evidence-based medicine entirely.

I think you may be misunderstanding me: the tool suggests we SHOULD intensively treat blood pressure in the vast majority of older adults given how rapidly the benefits accrue. To not do so is ageism in my opinion.