Why would associations between cardiometabolic risk factors and depressive symptoms be linear?