Errors Made by Physicians When Ordering Tests

Selecting tests when there are no information systems to help has one major problem: it is a difficult procedure for physicians to complete successfully. Even when physicians agree to use specific, agreed-upon criteria based on history and physical examination to order tests selectively and thereby reduce routine testing, they still make a surprising number of mistakes when ordering tests. Approximately 30% to 40% of patients who should have certain tests (based on agreed-upon criteria such as those in Table 25-20 ) do not get such tests, and 20% to 40% of patients who should not have certain tests are nevertheless subjected to them. For instance, Blery and co-workers ^[128] examined 3,866 surgical patients in France. Even after medical personnel had been educated regarding which criteria indicated a need for which tests, 30% of the tests were ordered without need; another 22% of tests should have been ordered but were not. Thus, surgeons and anesthesiologists not only increased costs but also failed to obtain possibly valuable information.

These mistakes occur mainly because integrating the history, physical examination, and indications for laboratory tests is not an easy process. Even when criteria for testing have been previously agreed upon by surgeons and anesthesiologists, the number of variables one must remember makes arriving at correct conclusions a complex task. As an example, let us consider how many mistakes are made regarding one commonly used preoperative test, the chest radiograph.

Charpak and co-workers^[270] examined the value of preoperative screening chest radiographs for 3,849 patients. Surgeons and anesthesiologists agreed that any of the following findings on history or physical examination would warrant ordering a chest radiograph: any lung or cardiovascular disease; any malignant disease; current smoking by patients over 50 years of age; major surgical emergencies; immunodepression; and, for immigrants, absence of prior health examination. Surgeons made their decisions regarding ordering of chest radiographs after seeing the patient. Even with this agreement on criteria, of 1,426 chest radiographs that should have been ordered for this group of 3,849 patients, 271 were ordered but not warranted, and 596 were not ordered but should have been. Although clinical judgment may account for some of these discrepancies, most of these lapses appear simply to be errors. If so many errors occurred for a single test, even more errors would be likely if patients were subjected to multiple testing.

Data from studies performed by our group confirm this rate of error.^[269] We tested the hypothesis that for the period 1979 through 1988, physicians voluntarily and substantially reduced the ordering of preoperative tests not justified by history and physical examination. Reviewing 2,093 medical records from every other year of that period (and studying four operations at each of three cities), we investigated the indications for, and the performance of, preoperative tests. During this period, the incidence of unwarranted laboratory tests obtained preoperatively decreased from 32.2% to 25.8%. This decrease was irregular and varied from operation to operation, from test to test, and from city to city.

Furthermore, an unexpected 12% decrease (from 92.9% to 80.9%) in the ordering of indicated preoperative tests occurred. Overall, 66.9% of tests obtained preoperatively in 1979 were not warranted, decreasing to 60.1% in 1987. If the possible benefit of ordering only appropriate tests outweighed the possible harm of not ordering a needed test, the net result would still be a benefit to society. Unfortunately, however, the possible benefit of performing a needed test is probably more than twice the possible harm of performing an unnecessary test.

We concluded that the pressures to order tests more optimally and to do assessments more quickly have not been accompanied by changes in practice patterns that ultimately benefit the patient. In order for the net benefit to accrue to the patient, a better information system is needed for obtaining the truly necessary tests and for not ordering the unwarranted ones. On the other hand, punitive measures to reduce testing may save money but may impair health care. This may be the approach the HCFA is attempting, which now requires the use of diagnosis-related codes (International Classification of Diseases, edition 9 [ICD-9]) for reimbursement of testing.

Can physicians do better at preoperative evaluation than a 5- to 15-minute history-taking prior to induction of anesthesia for outpatient or "come-and-stay" patients (those to be admitted after surgery)? They can, and should, for their patients' sake and their own. The British have reached the same conclusion.^[271] A change in the system of obtaining patient histories and ordering tests has been advocated even for internists.^{[272] [273] [274]} ,

Just as the need for a more effective theoretical system of preoperative evaluation became evident, the study just

973