Article

Although there are multiple contributors to US health care costs, few factors exert greater influence in aggre-gate than the decision of a physician or related practitio-ner to order a test. The autonomy enjoyed by most providers in this arena has contributed to the unsustain-able increase in test use1 and has produced various attempts to assess the effectiveness of a particular test in a particular setting.2 Appropriateness as a criterion for the proper use of common diagnostic tests has gained momentum and credibility as professional societies such as the American College of Radiology have established evaluative criteria for test ordering,3 and the frequency with which those criteria are met is tested prospectively.4 In turn, quantification of appropriateness rates has led to renewed interest in the question of why tests that should not be ordered are ordered anyway5 and the compelling issue of what, if anything, can be done about it.6-10 Nonmedical reasons for test ordering have long been rec-ognized as important factors in medical use.11 The motives of the person ordering the test can be economic, such as the desire to maintain or increase remuneration12,13; medicole-gal, such as the fear of tort litigation14; or psychological, such as the requirement of a trainee or other subordinate to satisfy a supervising physician or the internal need to confirm a diagnosis that already was established.15 These explanations, however well understood and thoroughly catalogued,16 focus solely on medical culture and practitio-ner behavior and leave unexplored the potential influence of the characteristics of the patient receiving the test. The authors hypothesized that particular patient char-acteristics such as age, employment status, or type of insurance might be associated with an increase in inap-propriate test ordering. They focused on an echocardio-gram, a commonly ordered diagnostic cardiac test, because it has been reported to be widely overused,17 cri-teria by which to judge its appropriateness are well estab-lished,18 and appropriateness rates using those criteria have been estimated in a variety of settings.19-21 Th