Background
No data exist regarding the temporal trends in utilization of transthoracic echocardiography (TTE) in an outpatient pediatric cardiology setting. This study evaluates the trends in utilization of TTE for common diagnoses known to have low diagnostic yield and the factors influencing these trends.
Methods
Patients evaluated at our pediatric cardiology clinics from January 2000 to December 2014 and discharged with final diagnoses of innocent murmur, noncardiac chest pain, benign syncope, and palpitations were included. Variables collected retrospectively included patient age, sex, insurance type, distance from clinic, and ordering physician’s years of experience since fellowship.
Results
Of the 74,881 patients seen by 35 physicians, 36,053 (48.1%) had a TTE. The TTE rates increased from the beginning of 2000 to the end of 2004 (5.2% per year; P < .001) and then steadily declined until the end of 2014 (1.6% per year; P < .001). Utilization for noncardiac chest pain remained the highest, and use in infants increased significantly during the study period ( P < .001). After adjusting for all other factors, the following variables were associated with higher TTE utilization: younger age, males, Medicaid insurance, increased distance from clinic, and being seen by less experienced physicians. Temporal trends persisted after adjusting for all these factors.
Conclusions
After an initial surge in TTE utilization from 2000 to 2004, there was a steady decline. This study identifies some important factors influencing these trends. This information could help design quality interventions, but additional factors need to be explored since the trends persist despite adjusting for these factors.
Transthoracic echocardiography (TTE) is the most common imaging modality used in outpatient evaluation of pediatric patients presenting to cardiology clinics. There is a wide variability in the TTE ordering behavior of physicians for evaluation of common indications such as murmurs, chest pain, syncope, and palpitations. Previous studies have shown that TTE may be a high-cost and low-yield test for these common indications. There could be several physician- and patient-related factors driving the TTE utilization, but these factors have not been studied well in pediatric patients.
In the current era, there is an increasing emphasis by payers and administrators to curb the cost of unnecessary testing. In response to a significant increase in the use of noninvasive imaging in adult cardiology between 2000 and 2005, the appropriate use criteria (AUC) were developed. In addition, the Deficit Reduction Act of 2005 was implemented in 2007, which reduced the reimbursement for imaging services provided in physician offices. The data resulting in these changes were based on Medicare spending for adult patients. This action is understandable since imaging in pediatric patients probably constitutes a much smaller proportion of health care spending. Since opportunities may exist to improve patient care in a cost-effective manner, the importance of evaluating the temporal trends in TTE utilization and the factors affecting these trends in the pediatric population cannot be ignored.
The recently published first pediatric AUC addresses initial outpatient TTE evaluation. An implementation study using this document reported that the common indications rated “rarely appropriate” were related to innocent murmurs, syncope, palpitations, and chest pain. While there are studies from the adult cardiology population reporting temporal trends in utilization of noninvasive imaging, no such data exist for pediatric patients. The purpose of this study was to determine the temporal trends in utilization of TTE for outpatient evaluation of patients who were discharged with a final diagnosis of innocent murmurs, noncardiac chest pain, benign syncope, and palpitations in a large pediatric cardiology practice over the past 15 years and to determine the patient and physician factors influencing these trends.
Methods
This retrospective study was approved by the Institutional Review Board of Children’s Healthcare of Atlanta. Patients ≤18 years old who underwent initial evaluation in an outpatient setting at our pediatric cardiology clinics from January 2000 to December 2014 and were discharged with a final diagnosis of innocent murmur, noncardiac chest pain, benign (vasovagal or neurocardiogenic) syncope, and palpitations were included. This information was obtained from a database maintained by our billing office that keeps a record of the Current Procedural Terminology codes used to determine the final diagnosis. Our center operates 20 pediatric cardiology clinics throughout the metro Atlanta area and other cities in Georgia. All clinics have a similar infrastructure and staffing. The TTEs are interpreted by the cardiologist evaluating the patient in the clinic and not by the reader in the echocardiography laboratory at the main hospital. All TTEs are performed and interpreted prior to discharging the patient from the clinic.
Physicians with fewer than 500 patients over the study period (average of 2–3/month) and observations related to these physicians were excluded. The rationale behind this exclusion was that the physicians that saw an average of 2–3 patients per month were thought to represent those who did not conduct the clinic on a routine basis. In addition, 500 patients over the study period was chosen as the threshold for inclusion to allow for stable physician-level estimates when using our mixed models, as some of these physicians were only represented in a year or two of data.
Data collected included whether or not a TTE was performed, patient age, sex, insurance type (private, Medicaid or others), distance of their residence from clinic (<5 miles, 5–9 miles, 10–19 miles, 20–49 miles, ≥50 miles), ordering physician, and his or her experience in years since graduation from fellowship (arbitrarily chosen cut points of <5 years, 5–9 years, 10–19 years, ≥20 years). Age was stratified using 5-year intervals for chest pain, syncope, and palpitations since these symptoms are quite infrequent in those <5 years old. However, for murmur it was further stratified into those <5 years into ≤1 month, 1 month to 1 year, and >1 year to 5 years owing to high prevalence of murmurs in those <5 years and significant differences in the proportion of innocent versus pathologic murmurs in these categories. For a descriptive purpose, the distance from clinic was stratified using cut points of 5, 10, and 20 miles roughly based on the 25th, 50th, and 75th percentiles, respectively, and 50 miles was chosen as a cutoff since it is a rough approximation of the radius of the greater Atlanta metro area and also represented the 95th percentile.
TTE results from a sample of 100 randomly selected patients (50 with TTE and 50 without TTE) were reviewed to see whether the final discharge diagnosis matched with that in the database maintained by the billing office.
Statistical Analysis
Descriptive statistics were calculated using counts and frequencies or medians and interquartile ranges (IQRs) for patient, physician, and clinic characteristics. These characteristics were compared between patients who had a TTE and those who did not using χ 2 -tests and Wilcoxon rank-sum tests, as appropriate. An interrupted time series regression was performed to quantify the effect of changes starting 2005. The interrupted time series regression was performed using segment regression analysis on TTE utilization summarized by year quarters. The change in rate of TTE utilization was assessed by comparing the linear trend between time and TTE rate before and after January 2005. TTE rates over time were stratified and compared by a number of factors including indication, physician experience, and patient age, sex, insurance type, and distance from clinic. Comparisons within an individual year were done using χ 2 -tests, and overall trends were compared in a generalized linear mixed model. A fully adjusted mixed model was created to determine the association between factors and TTE rates. Variables included in the model were patient age, sex, insurance type, distance from clinic, physician experience, and year of evaluation. The distance from clinic and physician experience were treated as a continuous variable in the multivariate model. To account for physician variations in TTE utilization, random intercepts and slopes by physician were used. Diagnosis-specific multivariable mixed models using the same methods as described previously were used to assess the effect of age on TTE in each of the four diagnoses.
Results
During the study period 78,787 patients were seen by 50 physicians and discharged with a final diagnosis of innocent murmur, noncardiac chest pain, benign syncope, or palpitations. Fifteen physicians and 3,906 patients seen by these physicians were excluded since they saw fewer than 500 patients during the study period. Of the remaining 74,881 subjects who were analyzed, 36,053 (48.1%) had a TTE. In the randomly selected sample of 100 patients, the final discharge diagnosis matched with that recorded in the billing database in all but two patients, who were billed as innocent murmur. The clinic notes from these two patients documented an innocent murmur, but in addition to that, one patient had a tiny, hemodynamically insignificant patent ductus arteriosus and another a tiny midmuscular ventricular septal defect. Both of these patients were discharged from further follow-up.
Overall, the median age of patients was 8.6 years (IQR, 3.0–14.1), with 53% males, 52.5% with private insurance, and 36.5% with Medicaid. The median distance of a patient’s residence from clinic was 10.5 miles (IQR, 5.6–18.0). The median experience of the 35 physicians who were included in the study was 18 years (IQR, 7–30). Table 1 shows the overall differences among patient and physician factors in patients who had a TTE and those who did not. The median age of those who had a TTE was not significantly different from those who did not. Among those who had a TTE, the highest utilization was in those <1 year of age. Over the 15-year study period, after adjusting for other factors in the multivariable model, younger patients, males, Medicaid, increased distance from clinic, and less physician experience were associated with higher odds of utilization of TTE ( Table 2 ). Based on the multivariate model, there was a significant effect of year, and the temporal trends persisted after adjusting for all other factors. Compared with 2000, the odds of TTE utilization were higher in the years 2002–2011.
| TTE (n = 36,053), median (IQR) or n (%) | No TTE (n = 38,828), median (IQR) or n (%) | P value | |
|---|---|---|---|
| Patient age (years) | 8.9 (2.1–14.2) | 8.0 (3.1–13.8) | .219 |
| Sex | |||
| Male | 19,957 (55.4%) | 19,673 (50.7%) | <.001 |
| Female | 16,043 (44.6%) | 19,099 (49.3%) | |
| Insurance status | |||
| Private | 18,783 (52.1%) | 20,569 (53.0%) | <.001 |
| Medicaid | 13,791 (38.3%) | 13,515 (34.8%) | |
| Other | 3479 (9.7%) | 4744 (12.2%) | |
| Distance from clinic (miles) | 11.0 (6.2–19.4) | 10.4 (5.5–17.6) | <.001 |
| Physician experience (years since fellowship) | 17 (7–27) | 22 (9–30) | <.001 |
| Odds ratio | 95% CI | P value | |
|---|---|---|---|
| Patient Age | |||
| <1 month vs >10 years | 4.68 | (4.15–5.27) | <.001 |
| <1 year vs >10 years | 1.62 | (1.54–1.71) | <.001 |
| 1–10 years vs >10 years | 0.64 | (0.62–0.66) | <.001 |
| Sex, female vs male | 0.78 | (0.75–0.80) | <.001 |
| Insurance status | |||
| Medicaid vs private | 1.08 | (1.04–1.12) | <.001 |
| Other vs private | 0.93 | (0.88–0.99) | .014 |
| Distance from clinic | |||
| Per 10 miles | 1.01 | (1.00–1.01) | <.001 |
| Per 100 miles | 1.09 | (1.05–1.13) | <.001 |
| Physician experience (per 5 years since fellowship) | 0.86 | (0.78–0.94) | .002 |
| Year of evaluation | |||
| 2014 vs 2000 | 1.21 | (0.86–1.70) | .263 |
| 2013 vs 2000 | 1.30 | (0.93–1.81) | .121 |
| 2012 vs 2000 | 1.33 | (0.96–1.84) | .087 |
| 2011 vs 2000 | 1.51 | (1.09–2.09) | .013 |
| 2010 vs 2000 | 1.57 | (1.13–2.17) | .007 |
| 2009 vs 2000 | 1.57 | (1.14–2.17) | .006 |
| 2008 vs 2000 | 1.90 | (1.38–2.63) | <.001 |
| 2007 vs 2000 | 2.33 | (1.68–3.24) | <.001 |
| 2006 vs 2000 | 2.65 | (1.91–3.68) | <.001 |
| 2005 vs 2000 | 2.74 | (1.97–3.82) | <.001 |
| 2004 vs 2000 | 2.51 | (1.79–3.53) | <.001 |
| 2003 vs 2000 | 2.41 | (1.70–3.41) | <.001 |
| 2002 vs 2000 | 1.67 | (1.18–2.38) | .004 |
| 2001 vs 2000 | 1.17 | (0.82–1.67) | .388 |
Temporal Trends in Utilization of TTE
Interrupted time series regression ( Figure 1 ) showed an increasing trend of TTE use from the beginning of 2000 to the end of 2004 (increase of 1.3% per quarter, 5.2% per year; P < .001). Following this trend, a steady decline occurred in TTE use from the beginning of 2005 to the end of 2014 (decrease of 0.4% per quarter, 1.6% per year; P < .001).
The age- and sex-adjusted TTE rates grew from 402/1,000 persons in 2000 to 655/1,000 persons in 2005, for a rate increase of 63%, but declined to 485/1,000 persons in 2014. There was no statistically significant difference in TTE rate at the beginning and end of the study period ( Table 2 ).
Patient Factors and TTE Utilization
The trends in TTE utilization for each of the final diagnoses are shown in Figure 2 A. The utilization was highest among those with noncardiac chest pain, followed by innocent murmur, benign syncope, and palpitations. These trends continued throughout the study period. Based on patient age, overall the TTE utilization was the highest in those <1 month old and lowest for those between 1 and 10 years old ( Figure 2 B). The TTE utilization for those <1 year old significantly increased during the study period ( P < .001). However, TTE rates in those older than 1 year were similar in 2014 to those in 2000, despite an initial surge until 2005. Use of TTE based on patient age in each diagnosis category is shown in Table 3 . The odds of performing a TTE were significantly lower in those <10 years of age presenting with chest pain and palpitations when compared with those >10 ( Table 4 ). The odds were significantly higher for those presenting with syncope at age <5 years compared with those >10 years old. In addition, infants with a murmur had significantly higher odds compared with those >10 years old.
| TTE | No TTE | |
|---|---|---|
| Chest pain | ||
| <5 years | 150 (49.8%) | 151 (50.2%) |
| 5–10 years | 2,072 (54.2%) | 1,753 (45.8%) |
| >10 years | 6,941 (62.0%) | 4,250 (38.0%) |
| Syncope | ||
| <5 years | 212 (53.5%) | 184 (46.5%) |
| 5–10 years | 560 (39.7%) | 850 (60.3%) |
| >10 years | 3,341 (40.0%) | 5,005 (60.0%) |
| Palpitation | ||
| <5 years | 95 (17.9%) | 246 (72.1%) |
| 5–10 years | 638 (31.9%) | 1,364 (68.1%) |
| >10 years | 2,006 (34.9%) | 3,735 (65.1%) |
| Murmur | ||
| <1 month | 1,457 (79.6%) | 374 (20.4%) |
| 1 month to 1 year | 5,563 (59.1%) | 3,852 (40.9%) |
| >1–5 years | 5,543 (37.6%) | 9,200 (62.4%) |
| >5–10 years | 3,995 (42.6%) | 5,392 (57.4%) |
| >10 years | 6,245 (58.3%) | 4,463 (41.7%) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
