The first pediatric appropriate use criteria (AUC) address the use of initial transthoracic echocardiography in outpatients by all ordering providers. The aim of this study was to appraise the performance of the AUC across pediatric cardiologists, noncardiologist subspecialists, and primary care providers (PCPs). A further aim was to describe the variations in ordering patterns of different groups of practitioners, which could serve as the basis for targeted quality improvement activities.Methods:
Electronic health records for Seattle Children's Hospital and its four regional sites were retrospectively reviewed for initial transthoracic echocardiographic studies performed on patients aged ≤18 years. A sample of 1,000 consecutive studies and a sample of 1,514 studies in which studies ordered by noncardiologists were enriched were reviewed. The ordering provider type, study indication, and findings (normal, incidental, or abnormal) were classified. Indications mapped to three categories: appropriate (A), may be appropriate (M), and rarely appropriate (R). If multiple indications were documented, the highest level of appropriateness was used.Results:
In the consecutive sample, pediatric cardiologists ordered 81%, noncardiologist subspecialists 13%, and PCPs 5% of the total studies. In the enriched sample, only 4% were unclassifiable by the AUC. Abnormal findings were identified in 23% of A, 13% of M, and 9% of R studies (P = .03). Appropriateness varied among the three groups of providers (P < .001). For pediatric cardiologists, 67% of studies were indication category A, 13% M, and 14% R. Noncardiologist subspecialists ordered the highest percentage of A studies (88%) and the lowest percentage of R studies (1%). PCPs had the highest percentage of R indications (18%), and 23% could not be fully classified, because of insufficient order information. Yield of abnormal findings was highest for subspecialists (23%), intermediate for cardiologists (19%), and lowest for PCPs (15%; P = .03).Conclusions:
The AUC performed well across all provider types, as measured by the low percentage of unclassifiable indications and the observed relationship between greater appropriateness and higher yield of abnormal findings. The three provider types differed in appropriateness rates and had distinct ordering patterns, which could form the basis for future targeted quality improvement efforts.