Very similar to the ARB story

Yes those are the 2 options to create an indication: (1) compare to placebo when standard of care (SoC) contraindicated or (2) head2head vs SoC for OG indication. 1 "easier" & addresses a potentially bigger gap. I meant that 1 is ethically justifiable vs placebo instead of SoC, whereas 2 is not

Seeing this incl criterion ("cannot tolerate Spiro") a lot with ongoing nsMRAs/ASIs RCTs. Makes ethical sense since can't offer placebo in spiro-eligible pts, but usual problems are hyperK and dec eGFR; does any1 think these drugs will really be better tolerated? Spiro will not be usurped easily

It'd look like many cardiology drug CI/KM curves and would still impress many!

These results are unquestionably cool, but that might be the most egregiously misleading y axis I've ever seen. It feels like they are doing everything they can to obfuscate the absolute risk reduction here. When do we ever report cumulative incidence with raw event # on the y axis?

Use of a DOAC after a procedure undertaken because you couldn't take a DOAC, to prevent a complication from that procedure

Also check out @hildabast.bsky.social post on PubMed alternatives

absolutelymaybe.plos.org/2025/02/14/w...

Not with the AI piece, but I find the device and its app very clunky

And absolute risk reduction

Does folate supplementation work here like it does for phenytoin in kids? pubmed.ncbi.nlm.nih.gov/21482950/

7/ Next steps include further validation & impact analysis. It could also be used in some clever analyses of RCTs to assess risk-based heterogeneity of treatment effect. Ultimately, this is a tool to inform patients with CAD of their prognosis: decisionaid.ca/cr-decide can be used for this now

6/ The full paper describing the development & (internal & external) validation: www.cjcopen.ca/article/S258...
Highlights:
-Good discrimination internally & in ISCHEMIA
-Slight under-prediction of MACE in ISCHEMIA trial, esp in highest-risk group
-Net clinical utility vs treating all as high risk

5/ We developed models with all predictors & "reduced", simpler models with fewer, readily-available variables. Importantly for the health status models, replacing patient-reported QoL with CCS/NYHA class substantially worsened predictions; there's no substitute for eliciting directly from patients!

4/ We developed the MACE model in a BC-based registry (n=24,990) & the health status models in the APPROACH registry in AB (n=13,312)🇨🇦. We considered known predictors of MACE & CAD-related health status, including history, labs, & coronary anatomy

3/ Unlike with primary CV prevention, there are no widely-used risk calculators in patients with existing CAD to guide decision-making. Everyone with CAD is considered "high risk" with various features used to qualitatively de-risk/up-risk patients.

2/ We set out to develop a set of clinical prediction models ("risk calculators") to predict major adverse cardiovascular events (MACE) @3yr & health status @1yr in patients with stable CAD to help facilitate shared decision-making re: revascularization & secondary prevention medications