Methods

This multicenter prospective quality improvement project was conducted during 2 separate study periods: phase I (from April 1, 2014, to September 28, 2014) before the release of the pediatric AUC document and phase II (January 1, 2015, to April 30, 2015) after the release of the document. The pediatric AUC document was published online on September 29, 2014. A 3-month gap was purposefully kept between the 2 phases to assure that providers had enough time to be aware of the release. The 6 participating centers included Emory University School of Medicine and Children’s Healthcare of Atlanta, Children’s Hospital at Montefiore, Mayo Clinic Rochester, Massachusetts General Hospital, Nicklaus Children’s Hospital, and NewYork-Presbyterian, Morgan Stanley Children’s Hospital. This quality improvement project did not require the approval of the institutional review boards of most of the participating centers except for NewYork-Presbyterian, Morgan Stanley Children’s Hospital, where the study was reviewed and approved. During phase I, physicians were made aware of the data collection but were unaware of the contents of the AUC document. A week before initiation of data collection for phase II, the site investigators sent out an e-mail with the AUC document to ensure that all providers were aware of the release of this document. Besides this, no other active educational intervention was performed during the study period. Data were collected from patients ≤18 years who underwent initial TTE evaluation in the pediatric cardiology clinics at the centers. Patients with a known cardiac diagnosis, a previous TTE study in an inpatient or outpatient setting, or an evaluation by one of the study investigators were excluded. In addition, those referred for a TTE only by outside providers were excluded because of the lack of the clinical information that would be required to assign an indication based on the AUC document.

A data collection sheet including patient demographics, date of study, name of the ordering physician, and reason for TTE was filled out before obtaining the TTE. The reasons for TTE were divided into 10 broad categories: murmur, chest pain, syncope, palpitations, abnormal electrocardiogram (ECG), abnormal test other than ECG, systemic disorder, family history of cardiovascular disease, conditions that increase cardiovascular risk, and other (for those that did not fit into any of the earlier mentioned categories). In those patients with >1 indication for TTE, one was designated as the primary indication by the clinician, whereas the others were considered to be secondary indications. Results of TTE were reviewed by the site investigators and findings were classified into normal, incidental, or abnormal. Furthermore, the abnormal findings were categorized into minor, moderate, or severe based on their clinical relevance and need for follow-up as described previously. Minor findings were those that may require follow-up but no intervention. They included small atrial septal defect (ASD), small ventricular septal defect, small patent ductus arteriosus beyond the neonatal period, mild pulmonary stenosis, mitral valve prolapse, or bicuspid aortic valve without stenosis or insufficiency. Moderate findings were those that altered patient management but did not require urgent intervention (e.g., moderate to large secundum ASD, ASD other than secundum type, moderate to large ventricular septal defect, moderate to large patent ductus arteriosus, bicuspid aortic valve with stenosis or insufficiency, or anomalous aortic origin of a coronary artery). Severe findings were those that required urgent hospitalization or intervention (e.g., critical valvular stenosis, moderate or severe coarctation of aorta, severely reduced cardiac function, large pericardial effusion, or anomalous origin of a coronary artery from the pulmonary artery). The site investigators reviewed the data collection sheet and clinic notes to assign the AUC indication for the TTE and its corresponding rating (A, M, or R) based on the pediatric AUC document. If the clinical scenario did not fit into any of the indications listed in the document, it was considered unclassifiable. All de-identified data were entered into the REDCap System (Research Electronic Data Capture) that was maintained by the core site at Emory University School of Medicine and Children’s Healthcare of Atlanta. The individual sites maintained their own enrollment logs with their corresponding REDCap-assigned identification numbers.

The primary outcome measure was a change in the proportion of TTEs ordered for indications rated R during phase II. Secondary outcome measures included any change in the diagnostic yield of TTE and differences in the change in appropriateness rating among sites from phase I to II. Statistical analyses were performed using SAS, version 9.4 (Cary, North Carolina). Statistical significance was assessed at the 0.05 level unless otherwise noted. Descriptive statistics were calculated for all variables of interest and included medians with twenty-fifth to seventy-fifth percentiles or counts with percentages when appropriate. Normality of continuous variables was assessed using histograms, normal probability plots, and the Anderson-Darling test for normality. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to compare the proportion of abnormal findings in different patient subgroups using generalized linear mixed models. These models accounted for the clustering of multiple studies from the same site over both phase I and II. When assessing the impact of TTE rating on the odds of an abnormal finding, indications rated R were treated as the reference group and compared with those with indications rated A or M. Therefore, an OR >1 indicates that a group or risk factor is positively associated with an A or M rating, whereas an OR <1 indicates the risk factor is negatively associated with an A or M rating. In a multivariable model, we examined the impact of phase and age on the outcome abnormal finding, while controlling for clustering of patients within sites.