Background: Appropriate Use Criteria (AUC) are complex documents which provide great detail on which patients are likely to benefit from a test. For providers who order a low volume of cardiac tests, regular application and navigation through the AUC may be difficult. Audit and feedback (A&F) has been proposed as an effective method for encouraging de-implementation of unnecessary testing. We hypothesized that the volume of myocardial perfusion imaging (MPI) ordered by most providers would make A&F ineffective.
Methods: We performed a secondary analysis of data gathered for the evaluation of a program to reduce rarely appropriate MPI. Patients who underwent nuclear myocardial perfusion imaging (MPI) were rated using the AUC. Appropriateness rating was performed retrospectively by a nurse trained in application of the AUC. We extracted data on both the patient and the ordering provider from the medical record which accompanied the visit where the MPI was ordered. Patients were compared in two cohorts: those with rarely appropriate indications for MPI, and all others. We performed logistic regression to determine which patient and provider characteristics were associated with rarely appropriate MPI. We ranked the most common indications for rarely appropriate testing and calculated the mean number of MPI ordered by our providers.
Results: A total of 889 cases over six months were included; 22 MPI (2.5%) were rated as rarely appropriate. A minority of patients had a prior myocardial infarction (n=127, 143%) or coronary revascularization (n=291, 32.7%). Chest pain was more common among appropriate or maybe appropriate MPI (n=528, 61.1% vs. n=1, 4.5%, P<0.0001) while lack of symptoms was more common among rarely appropriate MPI (n=18, 81.8% vs. 107, 12.4%, P<0.0001). In logistic regression, lack of symptoms was retained in association with rarely appropriate testing (odds ratio 31.6, 95% confidence interval 10.5-95.2, P<0.0001). The majority (59.1%) of rarely appropriate tests were accounted for with only 3 indications (#8, #71, and #67). A total of 219 providers ordered MPI during the study period, with an average of 4.1 MPI per provider, less than one MPI per provider per month. The greatest number of rarely appropriate tests from any single provider was 3.
Conclusions: Within the setting of an educational program to reduce rarely appropriate MPI, the number of rarely appropriate MPI per provider was low. Lack of symptoms and indications including preoperative testing and screening in low risk patients were common drivers of rarely appropriate MPI. These data raise doubt about efficacy of a broadly applied program which provides individualized A&F. Efforts to increase awareness of common reasons for rarely appropriate testing may have greater impact.