Hacker News Reader

Compared with other factors like day of admission.

> Patients admitted on a Saturday and Sunday have a 10 per cent and 15 per cent higher risk of death than those admitted on a Wednesday.

But isn’t that because in weekend admits are rare and only done in serious situations where death is more likely?

Also worth considering publication bias: any result in the other direction will not get published.

They know a lot of people only read the headline, so how difficult would it have been to add five characters to the headline:

"Patients 0.2% ‘less likely to die’ if treated by a female doctor, study reveals"

It's borderline dishonest not to.

It will be statisticaly significant but if there's any confounders at all it's a correlation but not necessarily causation.