Accurately assessing the prognosis of a patient is the "Holy Grail'' of diagnostic cardiology. Knowing whether a patient is at low or increased risk of experiencing an untoward outcome is tantamount to defining the short-term and long-term management of any patient with proven or suspected cardiac disease. The exercise test has a long tradition and reputation for assessing prognosis in a variety of clinical scenarios, including suspected coronary artery disease, postmyocardial infarction, and congestive heart failure. In this issue, Ghayoumi et al. add to this knowledge by including patients with chronic coronary artery disease to that list. They clearly demonstrated, in a large cohort of male veterans with established coronary artery disease, that age and exercise capacity were the strongest predictors of all-cause mortality over an average of over 5 years of follow-up. Exercise capacity is well known to be one of the most powerful predictors of prognosis in each of these three patient groups. 1 What is missing from their study is 1) external validation of patients from other institutions, 2) adequate consideration of exercise heart rate and heart rate recovery as prognostic predictors 2, 3) comparison with other prognostic scores, and 4) validation that other hard endpoints that are important to cardiologists (eg, nonfatal myocardial infarction) also track along with mortality. Despite their study's shortcomings, their principal findings are consistent with that found in other published studies.

The next point to consider is how prognostic information can be used to assist in clinical decision making. Although not intended for use in the clinical arena the equation or score developed by Ghayoumi et al. can assign patients with chronic coronary artery disease into groups with low intermediate or high probability of mortality. This assignment to one of three subgroups is popular among score developers relevant to risk stratification decision making. In principle we are attempting to separate those patients at low risk who require no further cardiac evaluation from those who require more evaluation. Those at the highest level of risk are small in number generally in need of timely evaluations (eg coronary angiography). The intermediate group is a larger group which shares many of the characteristics of the two extreme groups. From the standpoint of exercise testing they will need further evaluations (eg noninvasive imaging or perhaps coronary angiography). Given that patients at high risk are few relatively easy to identify the major purpose of these stratification schemes is to separate the more numerous patients at low intermediate risk.

This simple score of Ghayoumi et al. bears a resemblance to another score already in routine clinical use (ie the Duke treadmill score). 3 It also uses exercise capacity assigns patients into three risk groups. Despite the absence of a direct comparisonthe estimated 5-year survival using this new score (98% low 90% intermediate 80% high) lines up very well with that predicted by the Duke score. 4 On the other hand the Duke treadmill score differs from this new score in that it was derived in a different type of patient population (suspected rather than known coronary artery disease) had a slightly different prognostic endpoint (cardiovascular rather than all-cause mortality).

In comparing the two scores further one can look at the Duke treadmill score as exercise capacity adjusted for exercise symptoms ST-segment changes the new score as exercise capacity adjusted for age prior infarction. It has always been my contention that exercise capacity which is inversely correlated with age should be adjusted for age in any score of this sort. A report that has appeared only in abstract form confirmed these suspicions that an age-adjusted Duke treadmill score would perform better than the original unadjusted score. 5

The Duke score the new score are both interpreted using the raw score to define the three risk groups. However another aspect of risk stratification arises that is not considered by either score. Prior guidelines developed by a joint consensus of the American HeartAssociation the American College of Cardiology 6 suggest that before exercise testing patients should first be stratified by pretest probability into low intermediate high risk groups using clinical variables such as age sex symptoms. Although they were referring strictly to the diagnostic situation where coronary artery disease status is unknown the Bayesian principle of considering what is already known about a patient to the interpretation of an exercise test result should be applicable to those both with without known coronary artery disease. Prior studies have demonstrated that exercise test scores can be used in concert with pretest scores. 7 Interpretation of the exercise score within each pretest probability group differs significantly studies are ongoing to determine the practical significance of these differences.

This brings me to my concluding point. Comparing the study of Ghayoumi et al. to a recent review of the numerous existing exercise test scores, 8 we now know that the variables that predict risk are similar in patients with without artery coronary disease. To simplify matters for practicing clinicians not burden them with numerous scores for different situations score developers should strive to develop a single exercise score that is flexible enough to be accurate in stratifying diverse groups of patients. The "Holy Grail'' has always been considered to be a singular entity.

References