| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
a author for correspondence: fax 314-362-1461, e-mail mscott{at}labmed.wustl.edu
Today's healthcare environment mandates cost-effective utilization of ancillary services, and routine laboratory test panels are often suggested as a target for cost reduction. Over the last 30 years, large multitest profiles have become routine in medical practice and are used to "screen" patients for disease (1)(2), presumably identifying conditions not part of a patient's clinical presentation (3)). However, many studies over the last 20 years have suggested that the use of test profiles provides little benefit toward identification of unknown diseases (4)(5)(6)(7)(8)(9)). Indeed, it has been suggested that the widespread use of panels was a function of the available continuous flow analyzers (10)(11)) and the increased profits afforded laboratories before the establishment of diagnosis-related groups.
Reagent cost savings from elimination of test panels would be trivial in the overall cost of healthcare operations. However, substantial savings might be realized by decreasing the follow-up of slightly "abnormal" results, which yields few new diagnoses (12)(13)(14)). In the US, the number of "panels" and the number of tests per panel are being minimized to some extent as a result of the new Healthcare Financing Administration-mandated panels and the documentation of medical necessity now required for some forms of reimbursement (15)).
We had included a test for total creatine kinase (CK) in a 12-test routine admission panel before May 1, 1996, after which we eliminated several tests from this panel, including CK. One reason for removing CK from this panel was an earlier study that had shown that admission CK tests did not lead to new diagnoses (16)). Here we asked whether the prior presence of CK in the panel facilitated the diagnosis of myocardial infarction (MI) for inpatients.
All inpatient admissions to Barnes-Jewish Hospital during the 1-year period before May 1, 1996, were examined. The hospital information system was queried to obtain the following data: admission identifier, admit and discharge dates, associated ICD 9 diagnosis codes, and all CK, CK-MB, and troponin I (TnI) values, all times for test orders and all times for test result entry. The institutional review board approved the data query.
We identified patients whose admission CK may have led to a new diagnosis of MI by identifying admission profiles with an increased CK (>200 U/L), selecting those patients whose first test order time for a CK-MB or TnI followed the result time for the increased total CK within 24 h, further selecting those whose CK-MB or TnI was also increased, and finally selecting those for whom a diagnosis of acute MI was made during the admission. These medical records were reviewed to determine if the increased total CK in the test profile was responsible for the additional work-up that led to the diagnosis of MI. If the CK-MB or TnI order was preceded by classic MI symptoms, an electrocardiogram indicative of acute MI, or other suspicion of MI, it was presumed that the increased CK result did not facilitate the diagnosis of MI and that the CK-MB or TnI would have been ordered anyway. If none of the above three factors was present, the total CK value was considered to have facilitated the diagnosis of MI.
The results of CK, CK-MB, and TnI testing on inpatients during the
study period are shown in Fig. 1
. There were 38 635 distinct inpatient admissions.
Sixty-two percent of these (23 770) had at least one total CK ordered
as part of the admission profile 12. Of these 23 770 patients, 7463
also had at least one CK-MB or TnI ordered.
|
Twenty-two percent of the inpatients with a total CK in the Profile 12 had at least one value >200 U/L (5102 patients). Of these, 2220 patients also had a CK-MB or TnI performed, and 563 had their first CK-MB or TnI ordered after the result time for the increased CK from the profile 12. These 563 represented patients whose screening CK may have triggered the order for the CK-MB or TnI analysis. Of these 563 patients, 183 had a CK-MB or TnI that was increased and 39 of these 183 patients had acute MI as one of their discharge diagnoses. The 39 MI patients had a median maximum percentage of CK-MB that was 5.0% of their total CK value, whereas the 144 non-MI patients with an increased CK-MB had a median maximum CK-MB that was 1.6% of their total CK value. Furthermore, 23 of the 39 MI patients had TnI testing performed, and the median maximum TnI value was 8.5 µg/L. In contrast, 53 of the 144 non-MI patients who had a TnI ordered had a median maximum result below the detection limit of 0.6 µg/L.
Medical records were available for 37 of the 39 patients with MI. In 29
patients, it was clear that the total CK value was not a contributory
factor leading to the diagnosis of MI because either a diagnostic
electrocardiogram or symptoms of chest pain were noted before the
result time for the total CK. In five patients (Table 1
, patients 1–5), it was clear that the total CK was
contributory from both the timing of results and subsequent orders as
well as physician notes. Interestingly, three of these five patients
were unable to communicate (Table 1
, patients 1–3). Finally, we were
unable to determine if the total CK was contributory in three patients
(Table 1
, patients 5–8) after reviewing the medical records; however,
it was interesting that in all three of these patients there was a
history of diabetes, a population prone to "silent MI".
|
In summary, of >23 000 admissions with routine CK tests, no more than 8 new diagnoses of MI resulted from the tests. Interestingly, most of the patients in whom the increased total CK was the first clinical clue of an MI were either incapable of communicating or had a history of diabetes, and it is difficult to conclude from the small number whether these subpopulations would benefit from screening.
Of 563 patients who had their first CK-MB or TnI tests ordered following an increased total CK, 67% (380) of these patients never had an abnormal cardiac marker. Our findings would support contentions that unnecessary follow-up testing and evaluations often occur as a result of slightly abnormal values from tests performed as part of a profile (10)(13)). Although we cannot determine what percentage of these "more expensive" tests were a direct result of the initially increased CK, it is likely that a large number would not have been ordered without the profile CK result. Taken together with earlier studies, our current findings suggest that limiting the number of tests performed on asymptomatic patients will likely decrease costs across the entire healthcare system, not just the relatively minor direct laboratory costs, with another possible example being orders for bone imaging following slightly increased alkaline phosphatase test results.
The large number of abnormal values in healthy individuals is not completely unexpected. One reason is that reference ranges are defined as the central 95% distribution of healthy subjects, and thus the probability (P) of an abnormal value occurring in a perfectly healthy patient increases with the number of tests (n) performed [P = 1 - (0.95)n]. Thus, if 12 tests are performed, at least 1 test will be abnormal in 46% of patients. In addition, by their very nature, screening tests are performed on "low-prevalence" populations who are not selected by clinical presentation, which diminishes the positive predictive value (12)(14)).
In addition to the frequent use of profiles, many profiles contain redundant tests, thus further increasing the possibility of an abnormal value from a healthy patient. For example, among the following pairs of tests, it is unlikely that the second test provides additional information for most asymptomatic patients: creatinine and blood urea nitrogen, alkaline phosphatase and bilirubin, Na+ and Cl-, albumin and total protein, alanine aminotransferase and aspartate aminotransferase, and hematocrit and total hemoglobin. Nevertheless, redundant tests like these are still part of the new federally mandated laboratory profiles.
Acknowledgments
This work was supported in part by the College of American Pathologists.
Footnotes
Washington University School of Medicine, Division of Laboratory Medicine, Box 8118, 660 S. Euclid Avenue, St. Louis, MO 63110
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
