My friend, I am trusting academia, but I also know how flawed it can be. If anyone has a bias right now, it seems to be you and the bias is towards always trusting academic research without thinking for yourself.
> "If this is extractable to the general population, approximately 2.75 million (2.13-3.38 million) Americans may suffer annually from a phenomenon similar to CHS."
Well, it is most definitely not extractable, that's why they also wrote "if". The dataset is abysmally small, biased and the filters/conditions applied not nearly strict or valid enough to come to these conclusions.
They are also, at exactly this point, completely contradicting their own research and rationale, as well as your claims:
> 2127 patients approached for participation, 155 met inclusion criteria as smoking 20 or more days per month.
2127 patients that are already in the ER, for various possible reasons. Invalid dataset.
> 155 met inclusion criteria as smoking 20 or more days per month. Among those surveyed, 32.9% (95% CI, 25.5-40.3%) met our criteria for having experienced CHS
This would mean, by their and/or your logic:
- that out of a "general population" dataset, 7% smoke >= 20 days a month, which on a side note absolutely conflicts with other statistics on weed consumption
- even if that was the case, there is absolutely no way to infer from ER patients back to the general population
- even if that was a valid line of thinking, it would imply that: Out of 35 million cannabis smokers, there are 23 million that smoke weed more than 20 days a month (right..), and since they claim that 33% of them seem to have experienced CHS, this would imply that around 7.5 million of 35 million smokers have potentially experienced CHS.
You know what that is? 21% of all cannabis consumers. I don't think so. Their logic is flawed. ER patients are not a neutral slice of the population. You have absolutely no way of inferring back to the general population without knowing how and how much the dataset is skewed. This also shows very clearly that even if the dataset is at least somewhat valid, their pre-selection and applied conditions are absolutely bogus and not suitable.
> In fact, you're introducing your own concocted bias, because you've limited the dataset to the ER visits and not the population.
The dataset IS limited to a group of patients in the ER, you knucklehead. That's what was used to generate the dataset in the first place. A subset of people they found in an ER. Exclusively. That's a bias that almost certainly invalidates the whole calculation. You also have no way of inferring back to the population from it.
> You're now suggesting that these situations aren't actually marjiuana-related at all, but instead some other drug overdose that the ER staff totally missed.
Because I wrote "overdose", and not "THC overdose" or "cannabis overdose"? Sorry man, but.. if anyone is trolling here, it must be you.