How this instrument works
The Duke Treadmill Score folds three separate readings from a single exercise stress test — how long the patient exercised, how much the ST segment shifted on the ECG, and whether angina occurred — into one number with a direct link to long-term survival. It was derived by Mark, Hlatky, Harrell, Lee, Califf, and Pryor, who analyzed 2,842 patients undergoing both treadmill testing and cardiac catheterization, published in Annals of Internal Medicine in 1987. Rather than reading exercise time, ECG changes, and symptoms as three separate signals, the score combines them with fixed weights into a single prognostic figure.
The arithmetic subtracts two penalties from exercise time: five points for every millimeter of ST-segment deviation, and four points for an angina index of 0 (none), 1 (non-limiting), or 2 (exercise-limiting). A later validation study by Mark and colleagues in the New England Journal of Medicine, 1991, reported the resulting survival by risk band: a score of 5 or above is low risk, with about 99% four-year survival in that cohort; a score between −10 and 4 is moderate risk, near 95% four-year survival; and a score below −10 is high risk, closer to 79% four-year survival. Small differences in exercise time or ST deviation can therefore shift a patient across a genuinely meaningful boundary.
The score was derived and validated in outpatients referred for treadmill testing because of suspected coronary artery disease, using the standard Bruce protocol, and it assumes a genuinely diagnostic exercise test was completed. It says nothing about patients who could not exercise adequately, who have a resting ECG abnormality that makes ST-segment interpretation unreliable, or who present with an acute coronary syndrome rather than stable symptoms — those situations fall outside the population the original derivation and validation studies were built from, and the score should not be leaned on there.
- Enter Exercise time (minutes, Bruce protocol) — total treadmill time achieved before the test was stopped.
- Enter Maximal ST-segment deviation (mm) — the largest ST shift recorded on the exercise ECG.
- Set Angina during test to None, Non-limiting, or Exercise-limiting, matching what actually happened during the test.
- Read Duke Treadmill Score and compare it against the three risk bands: ≥5 low risk, −10 to 4 moderate risk, below −10 high risk.
- Treat the score as one input alongside symptoms, resting ECG, and overall clinical judgment, not as a stand-alone diagnosis.
Worked example — three treadmill results
10 minutes of exercise, 1 mm of ST-segment deviation, and no angina: DTS = 10 − (5 × 1) − (4 × 0) = 10 − 5 − 0 = 5. A score of exactly 5 sits right at the boundary of the low-risk category, associated with about 99% four-year survival in the original study population.
6 minutes of exercise, 2 mm of ST-segment deviation, and exercise-limiting angina (angina index 2): DTS = 6 − (5 × 2) − (4 × 2) = 6 − 10 − 8 = −12. That falls below −10, placing this result in the high-risk category, associated with about 79% four-year survival in the original cohort — a stark contrast built from a much shorter exercise time plus two compounding penalties.
12 minutes of exercise with no ST deviation and no angina: DTS = 12 − (5 × 0) − (4 × 0) = 12. With both penalty terms at zero, the score equals exercise time outright, landing well into the low-risk category rather than merely at its edge.
Questions
What do the three Duke Treadmill Score risk categories mean?
A score of 5 or higher is low risk, associated with roughly 99% four-year survival in the original validation cohort. A score from −10 to 4 is moderate risk, around 95% four-year survival. A score below −10 is high risk, around 79% four-year survival. These figures come from Mark and colleagues' follow-up validation published in the New England Journal of Medicine in 1991, built on the same derivation population as the original 1987 score.
What counts as an angina index of 1 versus 2?
An angina index of 0 means no chest pain occurred during the test. An index of 1 means angina occurred but did not force the test to stop — the patient kept exercising. An index of 2 means angina was severe enough to be the reason the test was terminated early. Each step up the index subtracts four points from the score, reflecting how much more prognostic weight limiting angina carries than angina alone.
Why does exercise time matter more heavily than the ST or angina penalties?
It doesn't inherently outweigh them — a poor showing on either penalty term can still pull a long exercise time down into a worse category, as the high-risk worked example above shows. What exercise time does capture directly is overall cardiovascular fitness and functional capacity, which the original Duke cohort found to be a strong independent predictor of survival on its own, before any ECG or symptom information is added.
Can the Duke Treadmill Score be used on a resting ECG that already shows abnormalities?
Not reliably. The score depends on measuring how much the ST segment shifts during exercise, which requires a resting baseline that itself is not already distorted by conditions like left bundle branch block, paced rhythms, or resting ST changes. In those situations the ST-deviation term loses its meaning, and stress imaging methods are generally preferred over a plain exercise ECG score.
Does a low-risk Duke Treadmill Score rule out coronary artery disease?
No. A low-risk score reflects a favorable prognosis — a low likelihood of dying over the next several years — rather than an absence of coronary disease. Patients with a low-risk score can still have measurable coronary artery narrowing found on catheterization; the score was designed to help decide who needs further invasive workup urgently, not to serve as a diagnostic test on its own.
References
- Mark DB, et al. Exercise treadmill score for predicting CAD prognosis. Ann Intern Med. 1987 (PubMed)
- Mark DB, et al. Prognostic Value of a Treadmill Exercise Score in Outpatients. N Engl J Med. 1991
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.