We agree with Weinfurt and colleagues that diagnostic technologies such as the ACI-TIPI should be evaluated in prospective controlled clinical trials. As emphasized by the National Institutes of Health National Heart Attack Alert Program (NHAAP) Diagnostic Technologies Working Group report (1), not only should such technologies be shown to have good diagnostic performance in relevant patients, but also they should be demonstrated to be safe and effective in actual use in clinical care. The latter is crucial, because even if a test's diagnostic performance is good in isolation, this may not translate into better physician decision making (2, 3).

Unfortunately, even in an area of intense interest such as emergency department evaluation of chest pain for acute cardiac ischemia, assessment of the impact of diagnostic technologies on care has been rare (1). This may be because, unlike for drugs, the U.S. Food and Drug Administration does not require actual clinical trials of clinical outcomes for the impact of diagnostic tests on care, or because such testing is too expensive given the profits from such technologies, apparently much less than those of pharmaceutical agents. Indeed, such testing is time-consuming and expensive. We were fortunate to have funding for this and other similar trials from the Agency for Health Care Policy and Research (AHCPR), but no other agency has shown interest in funding these trials, and to do so in any comprehensive way would be vastly beyond the resources of that currently rather small agency. Indeed, although to our knowledge the ACI-TIPI Trial, including 10 700 patients, was the largest such prospective emergency department trial ever conducted, even at the cost per patient of less than one tenth of typical pharmaceutical trials, it cost several million dollars. The GUSTO trial (4) of thrombolytic therapy included more than 40 000 patients and cost about $55 million, but it was economically justified for the pharmaceutical companies that would benefit from its results. There appears to be no sufficient economic motivation for prospective clinical trials of impact of most diagnostic tests, especially one in the public domain, such as the ACI-TIPI. Thus, to ask that an evaluation of a diagnostic test's impact on care have the power of GUSTO is currently unrealistic.

In the case of the ACI-TIPI trial, the differences in adverse outcomes were only one or two patients per event that had a baseline event rate of about 1%. We agree with the results of the statistical testing that such a difference of one or two patients is not significant.

Finally, along with the NHAAP Technology Working Group, we join Weinfurt and colleagues in making the call for more support for clinical trials of such technologies.