Background: The Duke Treadmill Score (DTS) is an established clinical tool for risk stratification of patients referred for exercise testing, but it does not consider age. We aimed to determine if age could improve the prognostic power of the DTS and if so, to modify the DTS nomogram to include age.
Methods: Of 1,959 patients referred for exercise testing from 1997 to 2006, 1,759 male veterans (age range 23-86 years) remained after exclusion of female and patients with heart failure. Cardiovascular mortality was the main outcome considered.
Results: Cox survival analysis was performed entering age and the DTS; both were significant (P <or= .002) with similar Wald Z values (5.4 and -3.1) and regression coefficients but opposite signs. The score: age-DTS yielded an area under the receiver operating characteristic curve of 0.80 compared with 0.76 for the DTS (P < .001). Using this equation, a nomogram was constructed by adding age to the original DTS nomogram. The point at which the age-DTS line intersects the drawing line from the DTS to the corresponding value for age indicates average annual cardiovascular (CV) mortality adjusted for age. For a DTS associated with a 2.5% annual CV mortality, an age of 30 compared with 70 decreased CV risk by a factor of 10 to less than 0.2% (P < .05, log-rank test).
Conclusions: We propose an age-adjusted DTS nomogram that improves the prognostic estimates of average annual CV mortality over the DTS alone. This nomogram requires external validation and extension to women.