New formula for predicting long-term survival based on exercise treadmill testing

Jump to: navigation, search

New formula for predicting long-term survival based on exercise treadmill testing

December 31, 2007 By Benjamin A. Olenchock, M.D. Ph.D. [1]

Cleveland A new prospective study has evaluated a new mortality prediction rule based on clinical characteristics and treadmill testing in low-risk patients with normal ECGs and suspected coronary artery disease. The nomogram, published in the Annals of Internal Medicine, better identified patients with less than 3% 3-year mortality than did the Duke Treadmill Score.

The most commonly used clinical tool for prognosis based on exercise treadmill testing is the Duke prognostic treadmill score, which is derived from exercise time, maximum ST segment deviation, and exercise angina. Patients are classified as low, moderate, or high risk based on their score, with five-year survivals of 97%, 90%, and 65%, respectively.

Data for the new prediction nomogram was collected at the Cleveland Clinic, and validated at Kaiser Permanente Colorado in Denver, Colorado. Over 33,000 patients were included, mean age was 52, mean follow up of survivors was over 6 years. Patients were excluded if they had known coronary disease, heart failure, cardiomyopathy, valvular or congenital heart disease, pacemaker or defibrillator, end-stage kidney disease, or if they were taking digoxin. All had normal resting ECGs. A Bruce or modified Bruce protocol was used for exercise stress testing. The primary outcome was death, obtained from the Social Security Death Index. The statistical analyses were quite complicated; in essence, the investigators used Cox proportional hazard modeling to determine risk associated with a number of clinical and test variables, and used this data to construct a nomogram. They then evaluated the test accuracy, compared their test with the Duke treadmill score, and used the Kaiser data set for external validation.

The new prediction nomogram adds clinical characteristics and other treadmill parameters that have prognostic value. Points are tallied from age, gender, typical angina symptoms, diabetes, smoking history, hypertension, proportion of predicted METs achieved, ST-segment depression, test-induced angina, abnormal heart rate recovery, and frequent ventricular ectopy during recovery. Interestingly, in the Cox proportional hazard analysis, test angina and ST-segment depression—two of the three components of the Duke score—were not statistically associated with risk. Also, patients who experienced test angina had better survival in this study, so the nomogram actually assigns points for the absence of test angina. Points are then summed, and the total point value is used to determine probability of 3- and 5-year survival. The multivariable nomogram was excellent at predicting mortality, and was better than the Duke treadmill score at identifying patients with less than 3% 3-year mortality.

Limitations to this study include that it was a single-center analysis, and the exclusion of patients with known heart disease and abnormal electrocardiogram. The clinical utility is untested.

The nomogram is available online at

1. Lauer MS, Pothier CE, Magid DJ, Smith SS, Kattan MW. An externally validated model for predicting long-term survival after exercise treadmill testing in patients with suspected coronary artery disease and a normal electrocardiogram. Ann Intern Med. 2007 Dec 18;147(12):821-8.