Hacker News new | ask | show | jobs
by brookst 23 days ago
Did they though? Not accounting for confounding factors like “other drugs” seems to indicate it’s not about the risks of cannabis use so much as the risks from all sources that the average cannabis user faces. Using only cannabis might have zero impact or even be beneficial based on this evidence (if you hypothesize that most of the negative outcomes were cannabis+cocaine users, for instance).

Still good data, but I don’t think it’s predictive for what cannabis use leads to (unless you assume that taking up cannabis makes you proportionately more likely to also take up whatever the confounding factors were).
2 comments
Spooky23 23 days ago
With a metastudy based on EMR data, I’d only use this to advocate for studying the issue further.

Information like this get collected at a point in time and never goes away. People have EMR fatigue and click though the questions. Anecdotal point, several years ago I accidentally stated that I drank enough to be considered a severe alcoholic. Even after correcting it at my next visit, it never really goes away, I get asked lifestyle questions relating to alcoholism.

Similarly advocacy against drunk driving, a noble cause, juiced up the stats. If you run over and injure a guy on the sidewalk carrying a sealed bottle of liquor, it will be labeled “alcohol related” more often than not based on officer discretion. If it’s fatal, the autopsy will take that conclusion if any party has a 0.01 BAC.

link
mil22 23 days ago
They didn't find that cannabis use leads to cardiovascular disease. They found a strong association between cannabis use and cardiovascular disease in a very large study. Correlation isn't causation. The study itself acknowledges that. That doesn't mean it's an invalid or useless study that didn't add to the body of scientific knowledge and evidence about the relationship between cannabis use and cardiovascular disease - I think we agree there. That's how science works. Observational studies do not definitively prove causality.

After reading the study, should we update our posterior on the hypothesis that cannabis use causes cardiovascular disease to nudge it in the direction that it does? Yes - that's just Bayes' theorem. Does the probability go to 95%+? No, of course not; I'm not claiming otherwise. It's still useful research.

Also, worth noting that MI risk spikes several-fold within the first hour after cannabis use (and that's not caused by cocaine).

link