Usefulness of impaired chronotropic response to exercise as a predictor of mortality, independent of the severity of coronary artery disease.