I would argue that the utility of a treatment also depends on the particular proximate genetic and/or environmental causes of the disease/illness/problem at hand.
Let's imagine two obese individuals, person A and B.
Person A's obesity can be attributed to some sort of genetic propensity to be eating more than the average person, e.g., having lower than average control of impulse, getting rewarded by high-calorie foods more than average, suffering more than average from exercising, etc.
Person B uses highly rewarding, high-calorie foods as a way to regulate negative emotions, in lack of a better way to cope. This might be a kind of behavior that he has learned as a child, either directly or by observing a parent, say.
Giving person A a hunger-reducing drug seems like a good idea, given that it will address some (not necessarily all) of the proximate causes of his obesity.
However, one could imagine that in the case of person B, while giving him the drug would solve his obesity problem, it would not solve his underlying problem of being unable to cope with negative emotions in a healthy way. Although an empirical question, one could worry that person B's coping mechanism would lose its viability. Could one imagine that he would instead resort to other unhealthy ways of coping with negative emotion, i.e., cutting, abuse of alcohol/drugs, etc. (would love to see research done on this)?
I am aware that even if this is the case for person B, losing weight by the drug could still be a net positive due to the health benefits of not being obese as well as making it less strenuous to exercise for further health benefits, and perhaps having a better perceived body image/other mental health benefits. The point I am trying to make is simply that, even if we grant that all mental and somatic problems can be seen as determined by genes and environments, there might be grave unexpected consequences of seemingly benign treatments, depending on the proximate causes of the particular problem, e.g., obesity.
Given the lack of side effects from psychotherapy, this should be the first choice of treatment, as is mandated in countries like Denmark by the Ministry of Health. I don't know how you fail to acknowledge this, in my eyes, fairly obvious conclusion.
There is also evidence that exercise, especially in groups, has similar effects on depression as those of SSRI's or psychotherapy. In the UK, exercise is prescribed as a treatment for depression. Again, this has the advantage of lacking side effects.
SSRI's should always be the last resort due to their side effects.
Of course, some patients, as you mention, are too sick to receive psychotherapy (or get out of bed to exercise), in which case SSRI's are the obvious choice.