[tzs]

Note that when deciding if that is something to worry about you need to consider the base rate for the control group.

For example if I was choosing between an activity that killed 100 out of 400 people that did it and one that killed 121 out of 400 I'd pick the first. Picking the second would raise my death probability from 25% to 30.25%. That's over a 5% increase in my death probability.

On the other hand if I was choosing between something that killed 100 out of 100 million and something that killed 121 out of 100 million, I'd probably not really consider the differences in death probabilities. For the first I've got a death probability of 0.0001% and for the second I've got a death probability of 0.000121%. My death probability is only 0.000021% higher in the second activity.

In both cases the hazard ratio is 1.21, but with the second pair of events the probability of encountering the hazard is so low a 21% increase doesn't actually make enough of a difference for me to worry about it.

[riku_iki]

its reasonable note.

In the studies there were 170k participants with 13k disease cases during observed period, so looks like hazard rate was significant enough.