Specific fatigue-related items in self-rating depression scales do not bias an association between depression and fatigue in patients with coronary artery disease.