Author + information
- Received March 27, 2018
- Revision received May 17, 2018
- Accepted May 24, 2018
- Published online October 15, 2018.
- Oliver M. Barry, MDa,b,∗ (, )
- Kimberlee Gauvreau, ScDa,b,
- Jonathan Rhodes, MDa,b,
- Jeffrey R. Reichman, BAa,
- Laura Bourette, MSa,
- Tracy Curran, MSa,
- Julieann O’Neill, MSa,
- Jennifer L. Pymm, MSa and
- Mark E. Alexander, MDa,b
- aDepartment of Cardiology, Boston Children’s Hospital, Boston, Massachusetts
- bDepartment of Pediatrics, Harvard Medical School, Boston, Massachusetts
- ↵∗Address for correspondence:
Dr. Oliver M. Barry, Department of Cardiology, Boston Children’s Hospital, 300 Longwood Avenue, Farley Building, 2nd Floor, Boston, Massachusetts 02115.
Objectives This study quantified the incidence of arrhythmias during pediatric exercise stress tests (ESTs) and evaluated criteria to identify patients at risk of clinically important arrhythmias.
Background The incidence of clinically important arrhythmias during pediatric ESTs and criteria for identifying high-risk patients are poorly characterized.
Methods A retrospective review of ESTs performed from 2013 to 2015 was studied. Arrhythmias were categorized into 4 classes based on need for test termination and intervention. Risk factors evaluated included having an implantable cardioverter-defibrillator (ICD), cardiomyopathy, severe ventricular dysfunction, complex arrhythmia history, coronary disease with concern for ischemia, pulmonary hypertension, select poorly palliated congenital heart disease (CHD), and concerning symptoms. Negative predictive values (NPVs) were calculated.
Results During the study period, 5307 ESTs were performed. Median age of the subjects was 16 years (interquartile range: 13 to 24 years); 20% had complex CHD. At least 1 high-risk criterion was present in 507 tests (10%); having an ICD (37%) and cardiomyopathy (36%) were the most common criteria. Some arrhythmias were seen in 46% of tests, but only 33 events (0.6%) required test termination. Three events (0.06%) required cardiopulmonary resuscitation, all with high-risk criteria. Absence of a high-risk criterion had a 99.7% (95% confidence interval [CI]: 99.5% to 99.8%) NPV for an arrhythmia that required test termination and a 99.96% (95% CI: 99.85% to 99.99%) NPV for an arrhythmia that required intervention beyond test termination.
Conclusions Although self-terminating arrhythmias are common, dangerous arrhythmias are rare during ESTs in a high-volume pediatric cardiology program. Pre-defined high-risk criteria identified all patients with the most serious events. The absence of any criteria predicted a low risk for arrhythmias that required test termination. These data permitted informed choices regarding supervision of ESTs.
Exercise stress tests (ESTs) are an important testing modality that is commonly used by pediatric heart centers for the evaluation of functional capacity and exercise-induced arrhythmias to guide safe participation in physical activity and to assist in data-driven decisions regarding medical management (1–7).
Safety data regarding ESTs in children and young adults with congenital heart disease (CHD) are limited and largely extrapolated from adults with acquired cardiovascular disease (8-10). Previous pediatric studies have been limited by small testing volume and a trend toward milder disease severity (11-13). In particular, there are limited data on the safety of ESTs in pediatric populations who are identified as high risk based on institutional and American Heart Association guidelines (14). Due in part to the paucity of data in high-risk populations, there is considerable institutional variability in the willingness to undertake ESTs in some patient populations (14,15). In 1 of the largest single-center reviews of pediatric ESTs, the incidence of clinically important arrhythmias during ESTs was 3% (16).
To our knowledge, there have been no studies to report EST-related adverse events in pediatric populations that include substantial numbers of high-risk patients. Because patients with complex heart disease survive into adolescence and adulthood, and the role and use of ESTs has expanded, we believe that a better understanding of the risks associated with ESTs across a range of risk groups would add valuable knowledge to the field (17-29). Furthermore, although the American Heart Association recommends physician supervision for all ESTs, either directly in the laboratory or immediately available if needed (14), there is considerable variability in institutional practices regarding supervision (15). In an era of limited resources, defining the risks of serious adverse events associated with ESTs across a range of risk groups should permit more informed choices regarding patient selection and the potential for a tiered protocol for direct and indirect supervision of ESTs.
The practice pattern at Boston Children’s Hospital has long included regular ESTs in many potential high-risk groups. Approximately 90% of studies are performed with indirect supervision, including testing done at a satellite facility. We previously developed a set of a priori high-risk criteria to identify exercise tests with an increased risk of adverse events that would, by policy, require in-laboratory physician supervision. We undertook this study to review our experience with ESTs and their arrhythmic complications and to help fill the important previously described knowledge gaps.
Because we knew the incidence of dangerous arrhythmias was low, we hypothesized that the absence of high-risk criteria could predict the absence of a serious adverse event. This study had 2 main objectives: 1) to quantify the incidence of arrhythmias during ESTs in a pediatric and adult CHD cohort; and 2) to analyze the predictive value of high-risk criteria for arrhythmia events. A secondary aim was to provide data that would permit other heart centers and providers to make informed choices about referral for and supervision of ESTs.
This study was a retrospective review of all ESTs performed at Boston Children’s Hospital from January 2013 to December 2015. A waiver of consent for retrospective review was authorized by the Boston Children’s Institutional Research Board.
All exercise tests were performed by a trained exercise physiologist. Patients were tested using the Bruce protocol (treadmill) or a ramp protocol (cycle ergometer) (14). Cardiopulmonary data (metabolic cart) were collected on select patients, based on ordering physician preference, using a MedGraphics Breeze Ultima metabolic cart (MGC Diagnostics, Saint Paul, Minnesota). The physiologists screened records; patients with high-risk criteria had direct physician supervision by an attending or fellow cardiologist. The a priori high-risk criteria are outlined in Table 1. Those with an implantable cardioverter-defibrillator (ICD) had their device paired before the testing. Patients without high-risk criteria had indirect physician supervision by clinic staff members in close proximity to the exercise laboratory. For those with ICDs, tests were terminated if heart rates rose to within 20 beats/min of the criteria for therapy. In all patients, tests were terminated if the patient developed symptomatic supraventricular tachycardia, hemodynamic instability, pre-syncope or syncope, increasing chest pain, increasing ventricular ectopy at the discretion of the supervising technician or physician, or marked ST displacement (>2 mm) associated with chest pain and/or a history concerning for myocardial ischemia. In the absence of sustained arrhythmias or disqualifying physiology, patients were encouraged to continue until completion or fatigue and were monitored for 10 min of recovery or until heart rate was back to baseline.
Baseline demographic and anthropomorphic information was collected, including age, sex, height, and weight (Table 2). Baseline clinical findings were also collected, including the presence of a pacemaker and/or an ICD, baseline electrocardiogram, CHD, cardiomyopathy, and the presence of high-risk criteria.
CHD was categorized by lesion and severity using the categories outlined by the American College of Cardiology Adult CHD Task Force (30) (Table 3). Subjects who had multiple lesions were categorized as “other.”
Systemic ventricular systolic function data were collected from the most recent echocardiogram in our system, and systemic ventricular systolic function was characterized as none, mild, moderate, or severe based on the opinion of the interpreting cardiologist. For tests with no previous echocardiogram in our system and that showed no known structural heart disease, the systemic ventricular function was assumed to be normal. For tests with subjects who had CHD or cardiomyopathy but no available echocardiograms in our system, ventricular function was listed as unknown.
The primary outcome of this study was arrhythmia events during ESTs. All arrhythmias during testing were recorded and categorized, according to the maximal required clinical response, in the following manner: class I: nonsustained or isolated, not requiring test termination; class II: nonsustained, triggering test termination; class III: sustained, requiring vagal maneuvers or medications; and class IV: requiring cardioversion and/or defibrillation and/or cardiopulmonary resuscitation (CPR).
Patient and test characteristics are summarized using frequencies and percentages for categorical variables and medians and 25th and 75th percentiles for continuous variables. Negative predictive values (NPVs) are presented with 95% exact binomial confidence intervals (CIs). Because the prevalence of dangerous arrhythmias was low, we chose to focus on the NPV of the high-risk criteria or the percentage of subjects without arrhythmias who had a negative diagnostic test result (absence of high-risk criteria). Other metrics of test performance, including positive predictive value, sensitivity, and specificity, did not add clinical usefulness in this context. Tests with high NPVs were useful for identifying patients who were unlikely to have the disease in question, potentially eliminating unnecessary and costly treatments. Statistical analyses were performed using Stata (release 15, StataCorp., College Station, Texas).
There were 5,307 tests performed during the study period. Baseline demographics for the study population are shown in Table 2. There was a slight male predominance (53%). The median age was 16 years (interquartile range: 13 to 24 years; range 3 to 76 years). Testing was performed using a treadmill in 3,067 (58%) subjects, and expired gas analysis (metabolic cart) was performed in 2,840 (54%) subjects. Peak predicted heart rate was >85% in 68% of tests; the respiratory exchange ratios were >1.06 and >1.09 in 88% and 79% of tests, respectively.
The prevalence of CHD and cardiomyopathy in this cohort is illustrated in Figure 1. The most prevalent type of CHD was tetralogy of Fallot (555 tests; 10%), and there were 437 tests performed in patients with single-ventricle CHD (8%). The other types of CHD and their prevalences in this study are outlined in Table 3. Those with CHD were skewed toward those with more complex disease, with moderate or severe complexity present in 2,281 tests (87%). These included 1,237 tests (47%) with moderate complexity and 1,044 tests (40%) with severe complexity.
Sinus rhythm was seen on the baseline electrocardiogram for 4,628 tests (87%), and there was an ICD or pacemaker present in 399 tests (8%), with 198 tests with an ICD (4%) and 221 tests with a pacemaker (4%).
The systemic ventricular systolic function was normal in most of the tests (4,740 tests; 89%). Systolic dysfunction was mild in 410 tests (8%), moderate in 104 tests (2%), and severe in 48 tests (1%). Five tests had unknown systemic ventricular function before the EST.
The incidence of each individual high-risk criterion is listed in Table 1. One hundred six tests had 2 high-risk criteria, and 7 tests had >2 high-risk criteria. At least 1 high-risk criterion was identified in 509 tests (10%). The most prevalent high-risk criteria were ICDs (198 tests) and cardiomyopathy (186 tests), which were not independent.
The incidence and nature of the arrhythmias encountered during the study are summarized in Tables 4 and 5⇓⇓. Most of the arrhythmias (99.9%) were classified as Class I or Class II events and required nothing more than termination of exercise. Class III arrhythmia events occurred in only 2 tests (0.04%); both were supraventricular tachycardias that broke with Valsalva maneuvers. Class IV arrhythmia events occurred in 3 tests (0.06%). The individual details concerning the 3 class IV events are described in Table 6. There were no deaths. No neurological injuries or deficits occurred in any individual who had a Class IV event. When age was evaluated as a predictor of Class II to IV arrhythmia events, we found no significant association (p = 0.19). However, older age was significantly associated with the development of any arrhythmia (Class I to IV) during ESTs (p < 0.001).
Implication of class IV arrhythmia events
Each of the patients who required resuscitation had subsequent substantial and clinically important changes in their management plan based upon the results of the EST. Of the 2 patients with hypertrophic cardiomyopathy (HCM), 1 was referred for resection of subaortic stenosis, and the second patient, who had a hemodynamically marginal, relatively slow ventricular tachycardia (VT) at minimal exercise, was listed for and underwent successful cardiac transplantation. The patient with ischemic cardiomyopathy and VT secondary to Kawasaki disease was referred for urgent cardiac catheterization and was found to have an occluded coronary bypass graft that was successfully stented.
The frequency of serious events was low, limiting the value of traditional sensitivity, specificity, and positive predictive value calculations. Instead, the analysis focused on the likelihood of events not occurring. This NPV analysis was conducted for the study population, and the details are summarized in Table 7. For tests with no high-risk criteria, there was a NPV of 99.7% (95% CI: 99.5% to 99.8%) for any Class II, III, or IV events. There was a NPV of 99.96% (95% CI: 99.85% to 99.99%) for Class III or IV events, and a NPV of 100% (95% CI: 99.92% to 100%) for any Class IV event (Figure 2).
Experience in specific high-risk cohorts
During our study period, 147 tests were performed for patients with HCM (excluding patients with positive genetic testing and no clinical phenotype). Two of these tests (1.4%) had class IV adverse events that required CPR and defibrillation and/or cardioversion.
At our institution, patients with catecholaminergic polymorphic VT routinely undergo exercise stress testing, and this population represents most of the subjects with the high-risk criterion of “arrhythmia” (49 tests). Only 5 of these tests were terminated, all due to increasing ventricular ectopy. However, no subject had sustained VT or required any intervention beyond test termination.
There were 198 tests (4%) performed in patients who had an ICD. All 3 of the class IV arrhythmia events occurred in patients with an ICD.
Safety of ESTs
This study reported the experience of a single pediatric and congenital heart center with a busy exercise physiology laboratory. Our laboratory serves a large volume of patients with CHD, cardiomyopathies, arrhythmias, and other cardiovascular problems and has an established practice of testing high-risk patients and adults with CHD and some cardiomyopathies. To our knowledge, this was the largest number of ESTs reported from a single pediatric heart center, with the largest proportion of tests performed for patients with CHD.
These data confirmed the overall safety of ESTs for patients with CHD, cardiomyopathies, and other cardiovascular disorders commonly encountered in a pediatric heart center. Although serious adverse events did occur, they were rare, and high-risk groups could be accurately identified. Finally, our data permitted informed choices regarding referral for and supervision of ESTs.
High-risk criteria use
We encountered 3 serious events that required CPR and defibrillation and/or cardioversion. Fortunately, all 3 patients were quickly and effectively resuscitated and experienced no identifiable neurological injury. All 3 events led to important clinical management decisions for each individual patient. Each of these patients was accurately identified as high risk before they underwent ESTs. We believed that the ability to accurately identify patients at increased risk for adverse events permitted appropriate, effective, and efficient use of resources for supervision of ESTs. Specifically, our data indicated that for the 90% of patients without high-risk criteria, the incidence for serious arrhythmias during ESTs was extremely low. For these patients, our policy of indirect supervision, with physiologist review, and a cardiology team in the immediate vicinity seemed a rational use of resources. In contrast, serious arrhythmic events were only encountered in the 10% of patients with ≥1 high-risk criteria. Although the risk in this group remained low (it was highest in the HCM group and those with a previously identified risk of cardiac arrest that resulted in ICD therapy), direct in-laboratory physician supervision continued to be appropriate for most of these cases.
Clinically important arrhythmias
Ghosh et al. (16), in a recent, single-center retrospective review of >1,000 pediatric ESTs reported a 3% incidence of clinically important arrhythmias. However, none of these arrhythmias required acute medical therapy or cardioversion and/or defibrillation. Therefore, they corresponded to Class I or II arrhythmias under our categorization system. Ghosh et al. (16) did not encounter any class III or class IV arrhythmias. We believe that this discrepancy was due to the fact that our study population was medically more complex. The prevalence of HCM was >4-fold higher in our series, the prevalence of patients with moderate or severe ventricular dysfunction was 50% higher, the prevalence of CHD was almost twice as high, and the complexity of CHD was greater. The greater medical complexity in our cohort probably also accounted for the fact that we encountered arrhythmias much more often (46% vs. 28%) than did Ghosh et al. (16).
Our cohort included 11 tests that were terminated due to nonsustained VT (Class II events), as well as 313 tests with ventricular couplets and 100 tests with nonsustained VT that did not trigger test termination (Class I events).
Although the patients in the study by Ghosh et al. (16) who had clinically important arrhythmias would fall into our Class I and class II arrhythmia categories, we did not wish to imply that the arrhythmias so classified were unimportant. The decision about the importance of an arrhythmia for the patient could only be made within the context of a comprehensive clinical picture. The safety of obtaining that data was the focus of this analysis, and the 0.1% incidence of Class III and IV events was the most relevant statistic in this regard, whereas the 13% incidence of more than isolated ectopy that we observed was probably analogous to the 3% incidence of clinically important arrhythmias reported by Ghosh et al. (16).
Our experience highlighted patients with HCM as being high risk, and further research is ongoing to evaluate the clinical usefulness of ESTs in this population. Furthermore, in 2 of the 3 patients who required cardioversion and/or CPR, the life-threatening ventricular arrhythmia observed in the exercise laboratory was below their ICD detection criteria, and, in 1, it was refractory to standard cardioversion. That experience emphasized the unexpected, management-changing findings that could emerge when challenging physiology is studied in a controlled environment.
Patient selection and test supervision
Our data supported the safety of our institutional practice for supervision of exercise testing in our patient population. Because of the volume of ESTs at our center, it was not feasible to have direct supervision for all ESTs. Direct supervision for the high-risk tests (10%; 509 tests) required significant coordination within our multidisciplinary team but was feasible with our current resources. This was done by using a high-risk plan that was part of the morning care huddle, a cadre of ICD nursing staff who could pair the programmer (often combined with part of their scheduled check with the ICD), and a clearly defined sequence of exercise physiologists, fellows, and/or attending physicians. However, we recognized that our a priori criteria overlap, with the exception of HCM, clearly identified several cohorts in which the frequency of Class III or IV events was <1% and possibly much lower. The low incidence of adverse events limited our ability to make firm, data-driven recommendations regarding revisions to our high-risk criteria. Nevertheless, we believe that the data reported in this study will permit informed choices by other heart centers regarding referral for and supervision of ESTs.
These data also supported our practice of permitting testing of low-risk patients at a satellite facility covered by a code team but without onsite extracorporeal membrane oxygenation. However, high-risk patients were only tested at our main campus, where pacemaker nurses, advanced cardiac resuscitation, and additional staff were readily available.
Our data reflected the selection bias of clinical exercise testing in our practice. In addition to individual provider practice variation for the use of exercise testing, there was selection bias toward moderate to severe complexity CHD. At the same time, those with the most severe ventricular dysfunction and those with severe pulmonary hypertension were not systematically referred for exercise testing. It was also possible that some providers were less likely to obtain ESTs on patients with multiple high-risk criteria. Therefore, due to the selection bias, the risk associated with high-risk classification might be understated by our results. We recognized that the decision to obtain (or forgo) an EST was determined by the physician’s assessment of the risk/benefit ratio of the procedure, and that these considerations were not addressed by our study.
There were relatively frequent minor events that occurred during pediatric ESTs. These included neurally mediated syncope and near-syncope, exercise-induced atrioventricular block, and symptoms from exhaustion and/or panic attacks. None of these required significant interventions, and they were not effectively tabulated in this review.
Although we limited our formal review to a 3-year period, previous and subsequent experience was comparable, with events before the study period refining the protocol to include pairing ICDs. Like any complex protocol, it was impossible to firmly conclude exactly which parts of the process were most critical for the safety of our patients. We anticipate that some details will gradually evolve based on these data.
For patients with structurally normal hearts, this cohort was a reasonable pediatric and young adult cohort. The patients with CHD were systematically older and represented a more heterogenous group.
The overall incidence of serious arrhythmia events with ESTs in a cohort with pediatric and CHD patients was low. Using pre-defined a priori high-risk criteria, we effectively identified all of the patients who subsequently had the most serious (class IV) adverse events. The risks associated with ESTs appeared to be highest in patients with clinically apparent HCM. The absence of any high-risk criteria predicted a low risk of adverse events that required test termination. These data, combined with the overall low risk of ESTs, permitted informed choices regarding referral for and supervision of ESTs and limited direct supervision to ≤10% less of the test volume. Although a small number of patients will have life-threatening arrhythmias during ESTs, with appropriate supervision and resources, these arrhythmias can usually be managed effectively. The knowledge acquired from the ESTs can have important implications for the subsequent clinical management of the patients.
COMPETENCY IN MEDICAL KNOWLEDGE: ESTs are a valuable clinical tool with expanding roles for pediatric and CHD patients. This study demonstrated that at a large-volume pediatric heart center, clinically important arrhythmia events during exercise testing are rare. Patients at increased risk for arrhythmias can be accurately identified using pre-defined criteria.
TRANSLATIONAL OUTLOOK: Our data supported the use of exercise testing, with appropriate supervision, even in high-risk subgroups. A tiered approach to supervision of exercise tests, as we described in this study, might increase availability at other centers. Further research studies are needed to validate and refine these high-risk criteria as well as examine impact on availability. Further studies are also needed to analyze stress test risk for specific high-risk groups, including those with HCM, catecholaminergic polymorphic VT, and complex congenital heart disease.
Dr. Alexander has received royalties from Up-to-Date; and he has been an expert witness for medical malpractice cases of infant arrhythmia, amiodarone overdose in infant with CHD, alleged missed long QT syndrome, alleged delayed diagnosis of Lyme Disease, alleged missed diagnosis of myocarditis, as well as non-malpractice case involving assertion of latent long QT syndrome as cause for drowning. The other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- congenital heart disease
- confidence interval
- cardiopulmonary resuscitation
- exercise stress test
- hypertrophic cardiomyopathy
- implantable cardioverter-defibrillator
- negative predictive value
- ventricular tachycardia
- Received March 27, 2018.
- Revision received May 17, 2018.
- Accepted May 24, 2018.
- 2018 American College of Cardiology Foundation
- Pediatric and Congenital Electrophysiology Society (PACES), Heart Rhythm Society (HRS), American College of Cardiology Foundation (ACCF),
- et al.
- Goldberg D.J.,
- Avitabile C.M.,
- McBride M.G.,
- Paridon S.M.
- Kuebler J.D.,
- Chen M.H.,
- Alexander M.E.,
- Rhodes J.
- Roston T.M.,
- Vinocur J.M.,
- Maginot K.R.,
- et al.
- Takken T.,
- Giardini A.,
- Reybrouck T.,
- et al.
- Skalski J.,
- Allison T.G.,
- Miller T.D.
- Young D.Z.,
- Lampert S.,
- Graboys T.B.,
- Lown B.
- Rhodes J.,
- Barst R.J.,
- Garofano R.P.,
- Thoele D.G.,
- Gersony W.M.
- Paridon S.M.,
- Alpert B.S.,
- Boas S.R.,
- et al.
- Ghosh R.M.,
- Gates G.J.,
- Walsh C.A.,
- Schiller M.S.,
- Pass R.H.,
- Ceresnak S.R.
- Hasan B.S.,
- Lunze F.I.,
- Alvi N.,
- Shafer K.M.,
- Rhodes J.
- Hasan B.S.,
- Lunze F.I.,
- Chen M.H.,
- et al.
- Hasan B.S.,
- Lunze F.I.,
- McElhinney D.B.,
- et al.
- Priromprintr B.,
- Rhodes J.,
- Silka M.J.,
- Batra A.S.
- Miliaresis C.,
- Beker S.,
- Gewitz M.
- Lytrivi I.D.,
- Blume E.D.,
- Rhodes J.,
- Dillis S.,
- Gauvreau K.,
- Singh T.P.
- Massin M.M.
- Priromprintr B.,
- Silka M.J.,
- Rhodes J.,
- Batra A.S.
- Buber J.,
- Rhodes J.
- Rhodes J.
- Warnes C.A.,
- Liberthson R.,
- Danielson G.K.,
- et al.