WHAT TEST CHARACTERISTICS ARE NEEDED FOR NEW TESTS TO EVALUATE CARDIOVASCULAR PREVENTION STRATEGIES?

Purpose: As new therapies for primary and secondary prevention of cardiac disease become available (and as previously available therapies become better validated), estimation of future cardiac risk has become increasingly important to determine for whom the benefit of using these therapies outweighs their inherent risks and costs. To this end, new tests are continuing to be developed in an effort to improve risk prediction and thus optimize treatment strategies. Although these tests are often evaluated as to whether they independently predict risk, their true value lies in how well they improve stratification compared to or in combination with existing strategies.

Method: We developed a Bayesian model utilizing the concept of the expected value of imperfect information to evaluate optimal testing strategies based on cost/benefit analysis. We then evaluated the use of a theoretical test in conjunction with cholesterol, high-sensitivity C-reactive protein, and coronary calcium scoring over different age groups to determine the minimum diagnostic odds ratio an additional test would need to justify its use. Diagnostic odds ratios were varied from 1.5 to 5.0 and age was varied from 40 to 80.  A ten year timeframe, similar to Framingham scoring, was utilized for this analysis.

Result: The diagnostic odds ratio required for the additional test to be included in the optimal stratification scheme varied by age. The odds ratios required for age groups 40, 50, and 60 were, respectively, 2.5, 2.0, and 2.5. At age 70 and higher, even the maximum evaluated odds ratio of 5 was insufficient for the additional test to be of utility. When this test was added, in most scenarios, a small fraction of the patients required the addition of this test to the baseline strategy.  Cost of the treatment strategy only marginally affected these results.

Conclusion: The above analysis demonstrates that not only must tests meet minimum criteria in order to be useful, but that these criteria can vary markedly with age. As new tests emerge, they should be evaluated in light of how much truly new information they add to current testing strategies.