The title is somewhat misleading; they looked for causal effects from subjective wellbeing to cardiometabolic health and vice-versa, but the only such effect that they found was in particular BMI-->subjective wellbeing (and with the sign you'd expect.)