Know the probabilities of fatal and adverse side-effects and update them with evidence(Bayes' theorem mentioned above)

Update on all relevant evidence, even if you don't have empirical data.

I would add:
Make decisions based on cost/benefit analysis, not simply probabilities. For example, low probability treatments can make perfect sense to try if they are low risk, low cost.

Know that the failure to reject a null hypothesis is not proof of the null hypothesis. It does not establish the null hypothesis. A failure to reject is simply an epistemological failure.