The Need for an Outcomes Research Agenda for Clinical Laboratory Testing

"The frequency with which a laboratory test may be ordered varies from none to constant monitoring of all variables. What should be done? What constitutes good laboratory medicine? Should the physician order one of each available laboratory test on each patient once a lifetime, once a year, once an hour, or never? Many forces influence these decisions. Among them are availability of laboratory service, policies of technical instrument and reagent manufacturers, what the physician knows about the medical need for the test, and economic considerations, such as cost and who makes the profit or takes the loss. Laboratory tests should not be ordered without a plan for using the information gained. What will be done if the test result is normal? High? Low?"^{1 - 2}

I wrote those words in 1975 to begin the Toward Optimal Laboratory Use column in JAMA. This column introduced the use of algorithms for problems in which the use of the clinical laboratory would be important.³ The initial article ended, "Laboratory performance can be judged by the indexes of its products—data. Laboratory utilization performance is far more difficult to judge. Standards have not been established, but are urgently needed. The final questions include these: Were the laboratory tests appropriate? Were they beneficial? Were they economical? Were all loops closed? Was every laboratory test maximally useful?"^{1 - 2}

Sadly, the answer in 1998 is that we still don't know, not even in a research mode. We not only haven't gotten to first base, we haven't even picked our bat.

The 23 years since that column began have seen the greatest advances in technical laboratory testing in history. And yet, if one accepts the findings of the systematic review by van Walraven and Naylor⁴ in this issue of THE JOURNAL, we seem to have made very little progress toward the goal of using the clinical laboratory appropriately or even knowing whether we are doing so. A reasonable set of strategies for future research in appropriate use of laboratory tests is also proposed by van Walraven and Naylor.⁴

The study of the appropriate use of therapeutic interventions, albeit a transitional method largely dependent on one of the lowest-quality methods of evaluation, namely, expert opinion,⁵ has developed over the last 20 years. However, it has fallen into disfavor and is giving way to outcomes research. The good news is that because clinical laboratory appropriateness research has never gotten off the ground, perhaps the clinical laboratory community can leapfrog the appropriateness research quagmire and plunge right into serious outcomes research.

I call on the community of physicians who interface every day between the patient and the laboratory to join to define an agenda for determining whether given laboratory tests should be performed, when and how they should be performed, and whether such performance was beneficial or harmful or had no effect. The fact that this is so difficult should not be a deterrent to moving forward. The College of American Pathologists hosted a well-attended conference in 1997 on the subject of outcomes and accountability, from which came several articles in the Archives of Pathology and Laboratory Medicine.^{6 - 7} These help point the way.

A laboratory test on a human being is an intervention, although, strictly speaking, not a therapeutic intervention. An intervention of any kind (diagnostic or therapeutic) is appropriate only if it is more likely to benefit than to harm the patient and can be done at a reasonable cost and with reasonable risk. Recognizing that a cascade of therapeutic interventions commonly follows the performance of a screening or diagnostic test, David Wennberg and colleagues⁸ have demonstrated a tight positive relationship between total rates of cardiac stress tests and the rate of subsequent coronary angiography. Angiography, of course, often leads to coronary artery surgery or a coronary artery intervention of some kind. Although stress tests and coronary angiography are not clinical laboratory examinations, nonetheless the same type of cascade of therapeutic interventions often follows the performance of clinical laboratory tests at the front end of the process.

We have pointed out that the laboratory test loop begins with the conception of the desirability of a laboratory test, proceeds through processes of selection, ordering, identification, specimen procurement, transportation, preparation, analysis, and reporting.^{9 - 10} We have emphasized that at the end of this sequence, interpretation and action must be accomplished prior to actually completing a laboratory test loop. It is now clear that even that is not far enough. We clinicians and laboratorians should all be concerned about the effects of that laboratory test and whether the performance of it was useful for the patient or for the public's health.

Unfortunately, physicians order laboratory tests for many actual reasons in addition to the usual accepted purposes of diagnosis, monitoring, screening, prognosis, and confirmation of clinical opinion. These include pressure from patient, family, or peers; hospital or legal requirement; defensive or medical legal protection; curiosity, insecurity, or delaying tactics; reassurance of themselves, the patient, or family; profit for the hospital, managed care company, laboratory, or themselves; to establish a baseline, complete a database, or having frustration at nothing else to do; ease of performance with ready availability because somebody will pay for it; or, probably most frequent of all, habit.¹¹

The real reasons for laboratory testing should be to improve the outcome for the patient's or the public's health.¹² The time has come to develop a solid research agenda to measure and monitor these vitally important outcomes and to change practices when indicated by the results.