Background
Echocardiography is often used in the evaluation of pediatric chest pain, but the incidence of cardiac pathology is low. In 2014, the American College of Cardiology published appropriate use criteria (AUC) for echocardiography including recommendations for pediatric chest pain. We evaluated the frequency and diagnostic yield of echocardiograms performed for each AUC indication and cost associated with echocardiograms performed for indications meeting the “rarely appropriate” criteria.
Methods
Retrospective, single-institution study of all patients 18 years or younger undergoing an initial evaluation for chest pain by a pediatric cardiologist (2014-15). We categorized the appropriateness of indications for echocardiograms on the basis of the AUC. We used multivariable logistic regression to determine factors associated with performance of an echocardiogram with the “rarely appropriate” indication. Excess costs associated with nondiagnostic echocardiograms meeting the “rarely appropriate” criteria were estimated using the Healthcare Bluebook to estimate a fair market price.
Results
The cohort included 539 patients, median age 13 years (range, 3-18) and 51.0% female. With retrospective application of the AUC, echocardiogram indications were classified as “appropriate” (304/539, 56.4%), “maybe appropriate” (68/539, 12.6%), and “rarely appropriate” (167/539, 31.0%). Echocardiograms were performed in 70.5% (380/539) of patients overall and in 35.9% (60/167) of patients with “rarely appropriate” indications. Of those undergoing echocardiography, abnormal findings were present in 5.0% (19/380) and incidental findings in 2.6% (10/380); however, only one echocardiogram (0.3%) led to a diagnosis considered to be contributory to the patient’s chest pain. There were no abnormal findings in the “rarely appropriate” subgroup. Provider use of echocardiography for “rarely appropriate” indications varied widely from 0 to 75% across 15 providers ( P = .004). In multivariable analysis, provider clinical experience of ≥20 years was associated with a lower rate of echocardiograms for “rarely appropriate” indications (odds ratio, 0.21 [95% CI, 0.09-0.47] vs. providers with <10 years’ experience, P < .001). There was no significant association between race, ethnicity, age, sex, payer status, or total number of patients seen and performance of an echocardiogram meeting the “rarely appropriate” indications. Echocardiograms with “rarely appropriate” indications resulted in $47,578 in excess costs over the 1-year study.
Conclusions
Echocardiogram use in patients meeting the “rarely appropriate” indication criteria is of little diagnostic utility and contributes to additional cost to the patient and health care system.
Highlights
- •
Echocardiography is frequently utilized to evaluate pediatric chest pain.
- •
Utilization rates for echocardiograms vary significantly by provider.
- •
Echocardiograms meeting the ‘Rarely Appropriate’ criteria have low diagnostic yield.
- •
Echocardiograms meeting the ‘Rarely Appropriate’ criteria contribute to excess cost.
Chest pain is a common presentation at pediatric cardiology offices, yet cardiac causes are rare. A recent review found that only 1.2% of patients presenting to a major pediatric cardiology center for chest pain had identifiable cardiac etiologies. Chest pain is often evaluated with echocardiography by pediatric cardiologists due to its availability and noninvasive nature. The use of echocardiography has grown significantly over the past 15 years in both pediatric and adult settings in parallel with an increase in the use of all diagnostic imaging in the United States. Increased imaging use is thought to contribute significantly to the rising cost of U.S. health care.
Given the increasing utilization of pediatric echocardiography, in 2014 the American College of Cardiology published appropriate use criteria (AUC) guiding the performance of echocardiography for chest pain and other common outpatient complaints, with a goal of ensuring excellent patient care while optimizing resource utilization. These guidelines divided patients presenting with chest pain based on clinical characteristics and electrocardiogram findings into groups for whom echocardiography has “appropriate,” “maybe appropriate,” or “rarely appropriate” indications. The application of these criteria by pediatric cardiologists has been minimally studied, and no prior study has analyzed the downstream effects of deviations from these guidelines.
This study was designed to retrospectively review the application of the 2014 AUC for patients presenting with chest pain to pediatric cardiac clinics at a single major tertiary pediatric cardiology center. The goals were to (1) determine the number and findings of echocardiograms performed in each AUC category, (2) determine factors associated with ordering echocardiograms in patients meeting the “rarely appropriate” criteria, and (3) determine excess costs of nondiagnostic echocardiograms for “rarely appropriate” indications.
Methods
Identification of Subjects
A retrospective cohort study was conducted of all patients age 0 to 18 years of age presenting from June 1, 2014, to June 1, 2015, for initial outpatient evaluation of chest pain by 14 pediatric cardiologists at a single tertiary-care academic medical center. A web-based data query tool was used to identify all initial patient encounters with an International Classification of Diseases, 9th edition, visit diagnostic code for chest pain (786.5), specifically including “Chest pain, unspecified (786.50),” “Precordial pain (786.51),” “Painful respiration (786.52).” and “Other chest pain (786.59).” Chart review was subsequently performed by physicians (R.C.C., J.H.P., S.B.) to extract relevant information into a secure Research Electronic Data Capture (REDCap) study database. Patients were excluded if the visit was not confirmed as an initial consult ( n = 29), documentation of chest pain was lacking ( n = 16), an echocardiogram was obtained prior to initial visit ( n = 5), or there was inadequate documentation to apply AUC ( n = 7). Approval from the local Institutional Review Board was obtained with a waiver of informed consent prior to conducting the study.
Data Extraction and Study Definitions
Medical records were reviewed with data extracted for demographic features, clinical characteristics, cardiac imaging results (echocardiogram, computed tomography [CT], and magnetic resonance imaging), exercise stress testing, prolonged event monitoring, cardiac catheterizations, and cardiology follow-up. The recently published AUC delineate 11 indications for echocardiography in patients with chest pain for which a rating of “appropriate,” “maybe appropriate,” and “rarely appropriate” were applied. These indications for transthoracic echocardiography were retrospectively assigned by review of documented patient history and exam, family history, and electrocardiogram results. Indications were rated “appropriate” for documented exertional chest pain, abnormal electrocardiogram, or family history of sudden unexplained death or cardiomyopathy. “Maybe appropriate” indications had documented chest pain with other symptoms or signs of cardiovascular disease, recent onset of fever, recent illicit drug use, or family history of premature coronary artery disease. “Rarely appropriate” indications had documented nonexertional chest pain with a normal electrocardiogram or no recent electrocardiogram, reproducible chest pain, or chest pain without other symptoms or signs of cardiovascular disease. For the purpose of this study, multiple presenting complaints (i.e., syncope and chest pain) were classified as “appropriate” and exertional chest pain was conservatively defined as any chest pain reported with activity. Family history of early coronary artery disease was defined as onset before age 50. Unblinded results of all echocardiograms obtained were classified as normal, incidental (findings that require no further follow-up or monitoring), or abnormal. Abnormal results were classified based on severity of findings as mild, moderate, and severe based on the approach of Sachdeva et al. as well as whether the results were potentially causative of chest pain. Any uncertainty in echocardiogram result classification was reviewed and clarified by consensus of two board-certified pediatric cardiologists (M.J.C., K.D.H.). Follow-up encounters were reviewed for determination, if any, of underlying cardiac etiology for chest pain if not clearly established in the initial encounter. The cost of echocardiograms with “rarely appropriate” indications and associated downstream studies (i.e., those ordered to confirm or clarify any echocardiographic findings) as well as follow-up clinic visits were estimated from the payer perspective using the Healthcare Bluebook, which provides a “fair market price” that accounts for both technical and professional fees.
Statistical Analysis
Outcomes of interest included the frequency of abnormal echocardiogram findings stratified by all three AUC indications, frequency of additional imaging, cost of testing, provider echocardiogram ordering variability in encounters classified with “rarely appropriate” indications, and factors associated with performance of an echocardiogram for those patients with a “rarely appropriate” indication. Discrete variables were expressed as counts and proportions and compared between groups using Pearson’s chi-squared test or Fisher’s exact test where appropriate. Continuous variables were reported as median or mean values with ranges. Comparison of distribution of continuous variables was by two-tailed Wilcoxon rank-sum test. To evaluate the association between provider and patient characteristics and the performance of an echocardiography among encounters classified as “rarely appropriate,” we performed multivariable logistic regression analysis including the following covariates identified a priori: age (continuous), sex, race, ethnicity, payer status, provider years of experience (<10 years, 10-19 years, ≥20 years), and number of patients seen per year with initial complaint of chest pain (<50 vs. ≥50). Cutoffs for provider years of experience were selected in tertiles to ensure an adequate number of providers in each group for statistical analysis. For number of patients seen per year with initial complaint of chest pain, the cutoff was selected to represent a low patient encounter frequency (<1 patient per week). To account for the clustered nature of the data by provider, we used robust (sandwich) variance estimators in the multivariable regression. Data were analyzed using Stata 14.0 (StataCorp, College Station, TX) and IBM SPSS Statistics version 23 (IBM, Armonk, NY). Statistical significance was defined as P < .05.
Results
Cohort Demographics
The final cohort included 539 patients presenting with chest pain for initial evaluation by a pediatric cardiologist. Cohort demographics are presented in Table 1 . The median age at clinic visit was 13 years (range, 3-18 years). The majority of patients were Caucasian (37.7%) or African American (26.4%), with most identifying as non-Hispanic (65.9%). Nearly half of patients were insured by Medicaid (46.4%).
| N = 539 | |
|---|---|
| Male | 266 (49.4) |
| Age in years, median (range) | 13 (3-18) |
| Race | |
| Caucasian | 203 (37.7) |
| African American | 142 (26.4) |
| American Indian | 7 (1.3) |
| Multiracial | 11 (2.0) |
| Other | 40 (7.4) |
| Unavailable | 136 (25.2) |
| Ethnicity | |
| Hispanic | 39 (7.2) |
| Non-Hispanic | 355 (65.9) |
| Unavailable | 145 (26.9) |
| Insurance | |
| Medicaid | 250 (46.4) |
| Private | 259 (48.1) |
| Other | 30 (5.6) |
| AUC rating | |
| Appropriate | 304 (56.4) |
| Maybe appropriate | 68 (12.6) |
| Rarely appropriate | 167 (31.0) |
AUC and Diagnostic Yield
As demonstrated in Table 1 , indications for echocardiography were retrospectively classified based upon application of the AUC as “appropriate” (304/539, 56.4%), “maybe appropriate” (68/539, 12.6%), and “rarely appropriate” (167/539, 31.0%). An echocardiogram was obtained in 380/539 (70.5%) patients. This included 268/304 (88.2%) patients meeting the “appropriate” indication criteria, 52/68 (76.5%) meeting the “maybe appropriate” indication criteria, and 60/167 (35.9%) meeting the “rarely appropriate” indication criteria ( Figure 1 ).
Abnormal findings were noted in 19/380 (5.0%) echocardiograms and are summarized in Table 2 . Of those with abnormal echocardiographic findings, 17/19 (89.5%) were in patients meeting the “appropriate” indication criteria and 2/19 (10.5%) were in patients meeting the “maybe appropriate” indication criteria. The most common abnormal finding was a bicuspid aortic valve in 7/19 (36.8%) of patients who had an echocardiogram. Only one patient out of the 380 evaluated by echocardiography (0.3%) had an abnormal finding that was considered to potentially contribute to cardiac chest pain (an anomalous right coronary artery arising from the left anterior descending coronary artery). None of the 60 echocardiograms obtained in patients meeting the “rarely appropriate” indication criteria demonstrated an abnormal finding and none of these patients with “rarely appropriate” indications had a cardiac cause for chest pain identified through any imaging or further evaluation.
| Total, N = 380 | Rarely appropriate, n = 60 | Maybe appropriate, n = 52 | Appropriate, n = 268 | |
|---|---|---|---|---|
| Abnormal, n (%) | 19 (5.0) | 0 (0.0) | 2 (3.8) | 17 (6.3) |
| Patent ductus arteriosus | 1 | — | — | 1 |
| Atrial septal defect | 1 | — | — | 1 |
| Ventricular septal defect | 2 | — | — | 2 |
| Bicuspid aortic valve ∗ | 5 | — | — | 5 |
| Valvar regurgitation | 3 | — | 1 | 2 |
| Anomalous right coronary artery | 1 | — | — | 1 |
| Abnormal aortic origin of coronary † | 2 | — | — | 2 |
| Partial anomalous pulmonary venous return | 1 | — | — | 1 |
| Aortic dilation | 2 | — | — | 2 |
| Situs inversus | 1 | — | 1 | — |
∗ No significant stenosis or regurgitation.
† Reviewed by cardiac imaging specialists; aortic origin of coronary artery just superior to coronary cusp and felt to be nonpathologic.
Evaluation of “Appropriate” and “Maybe Appropriate” Echocardiogram Indications and Additional Testing
Exertional chest pain was the most common presenting complaint for patients meeting the “appropriate” indication criteria (180/304, 59.3%) and was also the most common indication for echocardiography in those who had an abnormal echocardiogram finding (10/17, 58.8%). Chest pain with other signs of cardiovascular disease was the most common indication category for patients meeting the “maybe appropriate” criteria (34/68, 50.0%). Additional imaging for these patients included CT angiography ( n = 2), cardiac magnetic resonance imaging ( n = 2), and stress echocardiography ( n = 1). Cardiology follow-up was scheduled for 13.4% (36/268) of patients who had an echocardiogram with “appropriate” indications and 17.3% (9/52) of patients who met “maybe appropriate” indications.
Factors Associated with Performance of an Echocardiogram with “Rarely Appropriate” Indication
As demonstrated in Figure 2 , the proportion of echocardiogram ordering for “rarely appropriate” indications varied widely among providers from 0.0% to 75.0% ( P = .004). In multivariable analysis, providers with 20 or more years of experience ordered fewer “rarely appropriate” indicated echocardiograms, with an odds ratio of 0.21 (95% CI, 0.09-0.47; P < .001), when compared with providers with less than 10 years of experience. There was a tendency toward decreased ordering frequency in providers who saw less than 50 patients per year with an initial complaint of chest pain (odds ratio, 0.38 vs. providers seeing ≥ 50 patients per year), but this did not meet statistical significance ( P = .09). There was no significant association between race, ethnicity, age, sex, or payer status and performance of an echocardiogram meeting “rarely appropriate” indication criteria. Multivariable analysis was repeated to evaluate frequency of ordering an echocardiogram across all AUC categories. Findings were similar, with only provider experience of 20 years or more being significantly associated with decreased ordering frequency of echocardiograms, with an odds ratio of 0.49 (CI, 0.28-0.88; P = .018), when compared with providers with less than 10 years of experience.
