In those treatment there is a metric that indicates success. For this it is not so clear, the main indicators saying "does not knowably prevent death and raises possibility of future death."

The same metric exists here. Various cancer survivor rates are measured by the average number of years someone survives after diagnosis, regardless of the rate of remission. How many years of life does treatment buy a patient, given the kind of cancer and the stage at which it is discovered?

Your argument is "more people who are suicidal enough to warrant involuntary commitment end up killing themselves than people who don't reach that point, therefore commitment causes (or at least does nothing to prevent) suicide," which is absolutely illogical.

This is beyond "I don't understand 'correlation != causation,'" and is into either "willful ignorance" or "I enjoy hearing myself talk" territory.