Is a single-item measure of self-rated mental health useful from a clinimetric perspective?