Infants and young children frequently have difficulty remaining still for an echocardiographic examination, potentially leading to poor study quality, increasing the likelihood of diagnostic errors. Sedation is believed to improve echocardiographic quality, but its effectiveness has not been demonstrated. The aim of this study was to test the hypothesis that sedation would improve study quality and reduce diagnostic errors.
Outpatient echocardiograms from children aged ≤36 months obtained from January 2008 to June 2009 were examined. Variables related to image quality, report completeness, and sedation use were collected. Diagnostic errors were identified and categorized. Multivariate analysis identified the odds ratios (OR) and 95% confidence intervals (CI) for risk factors for potentially preventable diagnostic errors and the impact of sedation on these errors.
Among 2,003 echocardiographic examinations, sedation was used in 498 (25%). The overall diagnostic error rate was 6.5%. Most errors (66%) were potentially preventable. Multivariate analysis identified the following risk factors for potentially preventable errors: precardiac procedure (OR, 2.19; 95% CI, 1.05–4.59; P = .04), moderate anatomic complexity (OR, 3.91; 95% CI, 2.25–6.81; P < .001), and high anatomic complexity (OR, 8.36; 95% CI, 3.57–19.6; P < .001). Sedation was independently associated with lower odds of potentially preventable diagnostic errors (OR, 0.47; 95% CI, 0.27–0.80; P = .006). Echocardiographic examinations with sedation had fewer image quality concerns (22% vs 60%) and fewer incomplete reports (3% vs 20%) ( P < .001).
Most echocardiographic diagnostic errors among infants and young children are potentially preventable. Sedation is associated with a lower likelihood of these diagnostic errors, fewer imaging quality concerns, and fewer incomplete reports.
Heart defects are among the more common types of congenital malformations in infants and remain a leading cause of death. Echocardiography is the primary imaging modality used to evaluate infants and young children for heart disease. Often, infants and young children cannot lie still during echocardiographic examinations, which may lead to incomplete examinations, suboptimal image quality, and diagnostic errors. Delayed or inaccurate diagnoses may place children with heart disease at risk for adverse outcomes. Procedural sedation is believed to improve echocardiographic study quality. Although the safety of various procedural sedation strategies for echocardiography has been extensively studied and reported, their impact on image quality and diagnostic error has not been systematically examined. As a result, there are few data to inform pediatric and congenital cardiologists if there is an actual impact of sedation on echocardiographic image quality or diagnostic accuracy. In this study, we sought to examine the effect of sedation on diagnostic errors, the frequency of incomplete examinations, and the frequency of image quality concerns in infants and young children presenting for outpatient echocardiography.
The Scientific Review Committee of the Department of Cardiology and the Institutional Review Board at Boston Children’s Hospital approved this study.
This study was performed in infants and young children undergoing outpatient echocardiography in a large academic pediatric and congenital echocardiography laboratory that performs approximately 22,000 echocardiographic examinations annually. Trained pediatric sonographers, pediatric cardiology fellows, and pediatric echocardiography staff cardiologists performed the studies, which were electronically stored. The pediatric echocardiography staff cardiologists interpreted all studies and issued reports, which are part of patients’ electronic medical records.
We examined outpatient echocardiograms obtained with and without sedation in infants and young children aged ≤36 months from January 1, 2008, through June 30, 2009. Only one echocardiogram was included per patient. For patients with multiple studies during this time period, only the first examination was included, whether it occurred with or without sedation. The study period was selected to allow ≥3 years between performance of the examination and our review. This time interval was chosen to increase the likelihood of discovery of any diagnostic error through subsequent investigations or procedures.
Echocardiography Sedation Methods
The patient’s primary cardiologist decided whether to request an echocardiographic study under sedation. A patient was considered eligible for outpatient nursing sedation if aged ≤36 months and without significant airway obstruction, lung disease, or severe cyanosis (oxygen saturation < 70% in room air). Patients not fulfilling these criteria or having other morbidities (e.g., severe pulmonary hypertension) were referred for sedation by the cardiac anesthesia service. As long as patients sedated by the cardiac anesthesia service were able to undergo their procedures on an outpatient basis, they were included in this study. Patients referred for echocardiography under sedation were screened by the pediatric cardiac nursing staff and were examined by the prescribing attending echocardiographer. Nursing sedation consisted of a single oral dose of chloral hydrate (typical dose 80 mg/kg, up to 1 g) according to institutional guidelines. Vital signs, including heart rate, blood pressure, respiratory rate, and pulse oximetry, were recorded throughout the procedure. As of April 2014, these sedation procedures have not changed substantially.
A diagnostic error was defined as a diagnosis that was unintentionally delayed, wrong, or missed as judged from eventual appreciation of the existing data or more definitive information. Diagnostic errors were identified by examining the echocardiographic reports for discrepancies compared with diagnostic findings from all other sources (e.g., cardiac catheterization, magnetic resonance imaging, operative observations, subsequent echocardiographic examinations, outpatient clinic records, and autopsy when available).
Diagnostic Error Categorization
After the identification of diagnostic errors, the primary echocardiographic images were reviewed to confirm errors and categorize them according to type, preventability, severity, and contributing factors, according to previously published methods. For cases without diagnostic errors, only reports, not primary image data, were reviewed. Errors were categorized as false-negatives, false-positives, or discrepant diagnoses. Error severity was categorized as minor (no effect on patient management), moderate (impact on patient management, may place patient at risk for adverse event), major (contributes to adverse event), or catastrophic (contributes to patient death). Contributing factors to error included the following categories: procedural or conditional, cognitive, technical, or patient or disease related. The specific anatomy involved in the error was also recorded. All diagnostic errors and relevant images were reviewed and either confirmed or excluded by a second investigator (O.J.B.), who adjudicated the error type, severity, preventability, and contributing factors.
Potentially Preventable Diagnostic Errors
Our focus was to examine diagnostic errors that were potentially preventable ( Table 1 ). Potentially preventable errors are errors for which echocardiography is considered the correct imaging modality to establish the diagnosis, which could have been made by a more complete examination or different imaging technique (e.g., the use of continuous-wave Doppler with proper alignment). We focused on potentially preventable errors because they represent diagnostic errors that might lend themselves to quality improvement actions, such as the administration of procedural sedation. Potentially preventable errors also represent the largest fraction of diagnostic errors.
|Potentially preventable||Diagnosis could have been made with a more complete examination or different imaging technique||Improper alignment of continuous-wave Doppler interrogation underestimates true aortic stenosis gradient|
|Preventable||Diagnosis is readily apparent on study images, but the diagnosis is not made||Pericardial effusion clearly demonstrated, but diagnosis of “no pericardial effusion” is made|
|Nonpreventable||Echocardiography could not be expected to establish the correct anatomic diagnosis||Failure to image a ligamentum arteriosum completing a vascular ring|
The other categories of preventability include preventable and nonpreventable diagnostic errors. Nonpreventable diagnostic errors are those in which echocardiography is the incorrect imaging modality and could not be expected to establish the correct anatomic diagnosis. Preventable diagnostic errors are errors in which the true diagnosis is readily apparent on study images but the diagnosis is not made (e.g., a clear demonstration of a large pericardial effusion, but a diagnosis of “no pericardial effusion” is made).
Echocardiography Quality Concerns and Incomplete Reports
An additional imaging outcome measure was the notation on the echocardiographic report of a concern about suboptimal image quality or inadequate data (e.g., “atrial septum not well imaged,” “incomplete color Doppler interrogation of the ventricular septum”). Another secondary outcome was the presence of an incomplete echocardiographic report. In our laboratory, an incomplete report is generated if a study yields few images or if the image quality is such that the study questions could not be reliably answered. The goal in our laboratory is to provide complete reports for all outpatient echocardiographic examinations.
The following demographic, clinical, and situational variables were recorded: sex, age, weight, height, presence of congenital heart disease, anatomic complexity, anatomic frequency, study location, and use of sedation.
Anatomic complexity was defined as follows: (1) low (no significant heart disease or a single, simple anomaly, e.g., atrial septal defect or single ventricular septal defect), (2) moderate (anomalies involving multiple lesions or diagnoses with common constellations of combined defects, e.g., complete common atrioventricular canal, tetralogy of Fallot, hypoplastic left heart syndrome), or (3) high (uncommon variants of moderately complex lesions or rare, complex anomalies, e.g., dextrocardia, superior-inferior ventricles with crisscross atrioventricular relations). Anatomic frequency was defined as follows, on the basis of the incidence observed in our echocardiography laboratory: (1) frequent (diagnosis is observed more than once per week, e.g., patent ductus arteriosus), (2) less frequent (diagnosis is observed more than once a month but less than once weekly, e.g., coarctation of the aorta), (3) rare (diagnosis is observed more than once per year but less than once monthly, e.g., inferior-type sinus venosus defect), or (4) very rare (diagnosis is observed less than once yearly, e.g., aortic–left ventricular tunnel).
Study locations included the following: main clinic (central echocardiography laboratory), satellite clinic (outreach clinics at community sites), recovery room (sedation room where examinations supervised by cardiac anesthesia are typically performed), and precardiac procedure. At our institution “precardiac procedure” is recorded as the study location when the study is performed with the purpose of being the final echocardiographic assessment before a planned cardiac catheterization or operation, regardless of the physical location where the study is performed.
Within the study population, we performed a univariate analysis to compare patient characteristics between the group that underwent procedural sedation and the group that did not using Fisher’s exact test. We then compared the above variables between those with the primary outcome of a potentially preventable diagnostic error and those without such errors. Variables with P values < .20 in univariate analysis were included in a multivariate logistic regression model to predict the odds of having a potentially preventable diagnostic error; P values < .10 were required for retention in the final model. Odds ratios (ORs) for having these diagnostic errors with 95% confidence intervals (CIs) were calculated. Procedural sedation was then added to the model; the logistic regression model allowed us to estimate the adjusted effect of procedural sedation on the odds of having a potentially preventable diagnostic error while controlling for other risk factors.
A total of 2,003 patients were included. The median age at echocardiography was 9.5 months (interquartile range, 2.3–22.1 months). Most (55%) of the study population had congenital heart disease. Among the 2,003 patients, 498 (25%) received procedural sedation ( Table 2 ).
( n = 2,003)
( n = 498)
|No sedation |
( n = 1,505)
|Male||1,020 (51%)||265 (53%)||755 (50%)|
|Female||983 (49%)||233 (47%)||750 (50%)|
|<6||832 (42%)||110 (22%)||722 (48%)|
|≥6 to <12||292 (15%)||126 (25%)||166 (11%)|
|≥12 to <24||415 (21%)||202 (41%)||213 (14%)|
|≥24 to ≤36||464 (23%)||60 (12%)||404 (27%)|
|Weight (kg) ( n = 1,982)||<.001|
|<5||506 (26%)||42 (9%)||464 (31%)|
|≥5 to <10||788 (40%)||288 (58%)||500 (34%)|
|≥10||689 (35%)||165 (33%)||524 (35%)|
|Location of testing||<.001|
|Main clinic||1,414 (71%)||416 (84%)||998 (66%)|
|Satellite clinic||475 (24%)||7 (1%)||468 (31%)|
|Recovery room||34 (2%)||34 (7%)||0 (0%)|
|Precardiac procedure||80 (4%)||41 (8%)||39 (3%)|
|Heart disease present||<.001|
|Yes||1,109 (55%)||374 (75%)||735 (49%)|
|No||894 (45%)||124 (25%)||770 (51%)|
|Frequent||1,441 (72%)||241 (48%)||1,200 (80%)|
|Less frequent||392 (20%)||179 (36%)||213 (14%)|
|Rare||151 (8%)||70 (14%)||81 (5%)|
|Very rare||19 (1%)||8 (2%)||11 (1%)|
Comparison of Echocardiographic Examinations with and without Sedation
As shown in Table 2 , only 22% of the sedated patients were <6 months of age, compared with 48% of the nonsedated group ( P < .001). The largest percentage (41%) of echocardiographic examinations with sedation were performed in patients aged 12 to 24 months. With respect to patient weight, only 9% of sedated patients weighed <5 kg, compared with 31% of nonsedated patients ( P < .001). Patients who underwent sedation were more likely to have congenital heart disease than the nonsedated group (75% vs 49%, P < .001) and were more likely to have moderate or high anatomic complexity ( Figure 1 ). Anatomic diagnoses in sedated patients were also less commonly seen compared with the nonsedated group.
All Diagnostic Errors
We identified a total of 131 diagnostic errors, for a diagnostic error rate of 6.5% in this cohort. Among these diagnostic errors, 78 (60%) were false-negatives, 17 (13%) were false-positives, and 36 (27%) were discrepant diagnoses. The most common primary contributors to diagnostic errors were incomplete examination of an anatomic structure (37%), underappreciation of a finding (17%), and poor imaging conditions (11%). In terms of categories of contributors to error, procedural or conditional factors were cited in 49%, cognitive errors in 30%, and technical or patient- or disease-related factors in 21%. Regarding severity, 41% were of moderate severity or greater, and the severity distribution was not significantly different between the sedated and nonsedated groups.
Diagnostic Error Preventability
Among 131 diagnostic error cases, 13 (10%) were not preventable and 32 (24%) were preventable. Potentially preventable errors represented the largest percentage (66%) of all diagnostic errors.
Potentially Preventable Diagnostic Errors
Most of the 86 potentially preventable errors were false-negatives (69%). This was followed by discrepant diagnoses (22%) and false-positive (9%) errors. Among the potentially preventable diagnostic errors, 40% were of moderate or greater severity. The largest contributor (65%) to potentially preventable diagnostic errors was related to procedural factors or study conditions (e.g., incomplete anatomic examination or difficult imaging conditions due to patient motion). This is in contrast to non–potentially preventable diagnostic errors, to which cognitive errors (e.g., errors committed at the time of study interpretation) were the largest contributor (56%) ( P < .001). Common anatomic segments involved were the atrial septum (27%), ventricles and ventricular septum (19%), patent ductus arteriosus (9%), pulmonary veins (8%), aortic arch (7%), and coronary arteries (6%).
Comparison of Potentially Preventable Error Cases to Cases without Such Errors
The findings of the univariate analysis focusing on potentially preventable errors are summarized in Table 3 . Compared with cases without potentially preventable errors, cases with such errors were more likely to occur in patients with structural heart disease, moderate and high anatomic complexity, and rare anatomic diagnoses. We also found that precardiac procedure patients had higher unadjusted rates of potentially preventable diagnostic error.