Predicting premature termination of exercise during Bruce protocol stress echocardiography.

Abstract

Clinical guidelines recommend that the exercise protocol of a stress echocardiogram is selected to induce volitional exhaustion after a target duration of at least 8minutes. While the Bruce protocol is very commonly used for clinical stress tests, it is known to be "steep", and many patients therefore fail to reach 8minutes. We studied predictors of failure and developed a method for identifying patients not suitable for Bruce protocol which was accurate and yet simple enough to be used as a point-of-care decision support tool. We studied data out-patients undergoing Bruce protocol stress echocardiograms (n=11086) and analyzed predictors of inappropriate early termination (defined as test duration<8min as per current practice guidelines) using logistic regression. A prediction model was constructed as follows: .5 points were given for each of hypertension, diabetes, smoking, and E/e'>7.9 in the resting echocardiogram; .1 point was added for each 1-unit increment in body mass index; 1 point was added for patient age by decade; 2.0 points were subtracted for male sex (p for all<0.001). In tests on held-out validation data, the model was well calibrated (in plots of predicted vs actual risk) and discriminated failure versus non-failure well (C-statistic .86 for a score of 6.0 points; p<0.001). These data may help to standardize protocol selection in stress echocardiography, by identifying patients pre-hoc where Bruce protocol will be inappropriately steep.