Changing Physician Behavior in Ordering Diagnostic Tests

Is it possible to change the behavior of physicians ordering diagnostic tests? The answer is yes. Is it difficult? Yes, in general, but it is easy to do if approached the right way. Although most people and organizations seem to not even try, those that do try often fail usually because they don't do it the right way. Physicians' behavior about ordering diagnostic tests can be changed if the laboratory director can confidently conjure up the wisdom about what to do, manifest the courage to do it, and sustain the backing of the organization's power structure.

Two articles in this issue of THE JOURNAL demonstrate how physician behavior can be changed and also point out what methods do not work. In a systematic critical review that cited 76 references, Solomon and colleagues¹ found that randomized clinical trials on improving physicians' use of diagnostic tests were few and conclusions were not very strong. Nonetheless, they report that exercises to develop consensus, description of guidelines, traditional education methods, utilization audits, and presentation of laboratory charges, when used alone, were ineffective; that environmental or administrative interventions, when carefully chosen, were very effective; and that combinations of interventions targeted at many behavioral factors were most likely to succeed. In a separate, large population-based intervention in Ontario, van Walraven and coworkers² report that individual actions and combinations of guideline dissemination, laboratory requisition form modification, and payment deletions produced rapid, profound, and lasting changes in utilization of certain common laboratory tests. These 2 articles offer remarkably clear support for the experience that some of us old hands in the clinical laboratory business have known for a long time.

Seventeen years ago, John Eisenberg, MD, suggested that there are 6 ways to change physicians' behavior: education, feedback, financial rewards, financial penalties, administrative changes, and physician participation.³ Five years ago, he updated these still-useful approaches.⁴ Many years ago in Los Angeles and Sacramento, I learned by trial and error and by trial and success that the best way to influence hospital physician behavior was to identify a problem and then involve the most respected staff physicians as a small committee in a given specialty area to address the issue and decide the right thing to do (ie, a patient focus committee).⁵ Having achieved wisdom in this manner, we made the administrative changes necessary and hung tight as it shook down, having received reasonable assurance of support by the administrative hierarchy. We wrote a book about the methods and results,⁶ which included abolishing "stats," eliminating obsolete tests, and organizing laboratory tests based on guaranteed turnaround time. By changing laboratory request forms, we eliminated improper but frequently used laboratory tests or sharply curtailed their use. By changing turnaround time, we promoted the use of good inexpensive tests, grouped into clinically useful clusters, not just using the groups the instrument manufacturer pushed to make money. By training pathology residents (on call always) to guide clinical residents about how to use complex tests properly, we guided the use of the toxicology and therapeutic drug testing laboratory, the blood bank and coagulation laboratory, and, of course, always surgical pathology and cytopathology.

It is very easy to get physicians to order certain laboratory tests and not order others. The control points are the order forms,⁷ the laboratory users' manual, specimen collection ease, turnaround time to results, and the use of economics. If the laboratory staff makes tests easy, fast, cheap, and good, there will be many orders. If tests are made hard to obtain, slow, poor, and expensive, far fewer will be ordered. By using these methods, we changed the ordering patterns of thousands of physicians for dozens of tests for many years. Of course, we do not know anything about whether these changes affected clinical outcomes.

I believe (and data and experience support) that the best ways to change physician behavior regarding the ordering of diagnostic tests are: