I wonder how much of this is due to sampling bias.

Folks who're eg depressed tend to "do less" and people in pain or with other such chronic ailments (eg ME/CFS) may have less energy to begin with to "spend".

In a survey, if the effort required to answer isn't very close to zero, you would expect to see lesser responses from such people; however they may be "typically" represented in terms of suicides.

In contrast, otherwise/formerly "healthy" people would not have this issue causing them to be underrepresented in the survey.