Author + information
- Received November 12, 2019
- Revision received January 24, 2020
- Accepted January 29, 2020
- Published online March 25, 2020.
- Daniel J. Friedman, MDa,∗ (, )
- Sean D. Pokorney, MD, MBAb,c,
- Amer Ghanem, PhDd,
- Stephen Marcello, MDe,
- Iftekhar Kalsekar, PhDd,
- Sashi Yadalam, PhDd,
- Joseph G. Akar, MD, PhDa,
- James V. Freeman, MD, MPH, MSa,
- Laura Goldstein, MPHf,
- Rahul Khanna, MBA, PhDd and
- Jonathan P. Piccini, MD, MHSb,c
- aSection of Cardiac Electrophysiology, Yale School of Medicine, New Haven, Connecticut
- bDuke Clinical Research Institute, Durham, North Carolina, cElectrophysiology Section, Duke University Hospital, Durham, North Carolina
- cElectrophysiology Section, Duke University Hospital, Durham, North Carolina
- dJohnson and Johnson, Medical Device Epidemiology, New Brunswick, New Jersey
- eJohnson and Johnson, Medical Safety, New Brunswick, New Jersey
- fJohnson and Johnson, Franchise Health Economics and Market Access, Irvine, California
- ↵∗Address for correspondence:
Dr. Daniel J. Friedman, Section of Cardiac Electrophysiology, Yale School of Medicine, Courier P.O. Box 208017, New Haven, Connecticut 06520-8017.
Objectives This study identified factors associated with risk for cardiac perforation in the setting of atrial fibrillation (AF) ablation in contemporary clinical practice.
Background Cardiac perforation is an uncommon but potentially fatal complication of AF ablation. An improved understanding of factors associated with cardiac perforation could facilitate improvements in procedural safety.
Methods Logistic regression models were used to assess predictors of cardiac perforation among Medicare beneficiaries who underwent AF ablation from July 1, 2013 and December 31, 2017. Cardiac perforation was defined as a diagnosis of hemopericardium, cardiac tamponade, or pericardiocentesis, within 30 days of AF ablation.
Results Of 102,398 patients who underwent AF ablation, 0.61% (n = 623) experienced cardiac perforation as a procedural complication. Rates of cardiac perforation decreased over time. In adjusted analyses of the overall population, female sex (odds ratio [OR]: 1.34; 95% confidence interval [CI]: 1.14 to 1.58; p = 0.0004), obesity (OR: 1.35; 95% CI: 1.09 to 1.68; p = 0.0050), and absence of intracardiac echocardiography (ICE) (OR: 4.85; 95% CI: 4.11 to 5.71; p < 0.0001) were associated with increased risk for cardiac perforation, whereas previous cardiac surgery (OR: 0.14; 95% CI: 0.07 to 0.26; p < 0.0001) was associated with a lower risk for perforation. Patient risk factors for cardiac perforation were identical in the subset of patients in whom ICE was used (n = 76,134). A risk score was generated with the following point assignments: female sex (1 point); obesity (1 point); nonuse of ICE (5 points); and previous cardiac surgery (−6 points).
Conclusions Cardiac perforation is a rare complication of AF ablation; incidence has decreased over time. One of the strongest predictors of cardiac perforation in the contemporary era is a modifiable factor, use of intraprocedural ICE.
Cardiac perforation is the most common potentially fatal complication of catheter ablation of atrial fibrillation (AF). Although perforation rates were as high as approximately 6% in some early reports (1), larger studies over the past few years have documented improved safety (2–4), with a recent publication that reported that rates have decreased to 0.67% (5). Despite these improvements in safety, cardiac perforation rates are coming under increased scrutiny because AF ablation is increasingly being performed outside of the highly specialized centers in which it was developed, and an inverse relationship between center volume and perforation rate has been reported (4,6). Accordingly, the Heart Rhythm Society has endorsed using rates of cardiac perforation due to AF ablation as a specific electrophysiology performance measure (7). Additional research is needed to understand contemporary risk factors for cardiac perforation and facilitate continued improvement in patient safety in the context of increasing dissemination. To meet this critical need, we performed a retrospective nationwide analysis of fee-for-service Medicare beneficiaries to identify modifiable and static risk factors for cardiac perforation within 30 days of an AF ablation.
The Centers for Medicare & Medicare Services Medicare Standard Analytic Files data from January 2013 to February 2018 were used for the study. The database includes 100% of U.S. inpatient hospital and outpatient hospital claims for Medicare fee-for-service beneficiaries enrolled in Medicare Part A and Medicare Part B. The denominator file, which includes enrollment status for Medicare beneficiaries, was used to confirm enrollment and vital status.
Patients aged 65 years and older who underwent an ablation procedure with a primary diagnosis of AF in an inpatient or outpatient setting between July 1, 2013 and December 31, 2017 were identified. Patients were required to have 6 months of continuous Medicare enrollment before the index procedure to allow for identification of baseline characteristics and comorbidities, and 30 days of follow-up post-procedure to allow for endpoint ascertainment, unless <30 days of follow-up was due to death. Patients who were enrolled in a health maintenance organization or received care outside of the United States were excluded. The performance of catheter ablation for the treatment of AF was defined by Current Procedural Terminology code 93656 (intracardiac catheter ablation of AF by pulmonary vein isolation) (3). We excluded patients who had concomitant diagnosis of anomalous atrioventricular excitation or paroxysmal supraventricular tachycardia when the ablation occurred in an inpatient setting (3). The first occurrence of AF ablation during the study period was identified and classified as the index date.
All baseline characteristics were identified using Medicare data. Patient characteristics were ascertained using primary or secondary diagnosis codes or procedure codes during the 6 months preceding the ablation procedure (see Online Tables 1 and 2 for a list of billing codes). Characteristics of interest included region, year, and a diverse set of comorbidities (e.g., previous pacemaker/defibrillator implantation, diabetes, hypertension, cancer, cerebrovascular disease, chronic pulmonary disease, dementia, diabetes, heart failure, ischemic heart disease, peripheral vascular disease, renal disease, stroke, valvular disease, paroxysmal supraventricular tachycardia, paroxysmal ventricular tachycardia, atrial flutter). Comorbidities were considered individually and as part of validated risk scores [Charleston Comorbidity Index (8,9), and CHA2DS2-VASc score (10)].
The primary outcome for this study was occurrence of cardiac perforation within 30 days of ablation. Cardiac perforation was defined as the presence of at least 1 billing code for cardiac tamponade, hemopericardium, or pericardiocentesis (5).
Baseline characteristics of the study population by outcome group (cardiac perforation vs. no cardiac perforation) were described using proportions for categorical variables and means with SDs for continuous variables. Differences between groups were tested using the chi-square test for categorical variables and 1-way analysis of variance test for continuous variables.
Logistic regression analysis was used to examine the relationship between patient, procedure, and hospital characteristics and the outcome of cardiac perforation. Characteristics associated with the outcome at a significance level of p < 0.05 were used to construct an adjusted model. Procedure year was not included in the adjusted models to maximize clinical usefulness of the model. CHA2DS2VASc score was not considered in the final adjusted model due to co-linearity with individual factors (e.g., female sex) and history of cardiac surgery, particularly because the former variable was associated with increased risk for perforation and the latter variable was associated with decreased risk for perforation. Due to the magnitude of association between ICE use and incidence of perforation, we repeated the modeling process in the subgroup of patients (74.4%) who underwent ablation with intraprocedural ICE. This post hoc decision was made to ensure results were generalizable to the increasing number of ablation procedures performed with ICE guidance.
A weighted risks score was developed using the parameter estimates of significant multivariable predictors of cardiac perforation with rounding to the nearest whole number. Factors associated with increased risk were assigned positive point values, and factors associated with decreased risk were assigned negative point values.
Sensitivity analyses were performed using generalized estimating equations to assess for the potential impact of clustering by site.
All tests were 2-sided, and a p value ≤0.05 was considered significant. All analyses were performed using SAS Enterprise Guide version 7.1 (SAS Institute Inc., Cary, North Carolina).
A total of 102,398 patients 65 years of age and older who were enrolled in fee-for-service Medicare underwent catheter ablation of AF between July 1, 2013, and December 31, 2017 and met all study criteria. Table 1 details the patient characteristics at baseline. Overall, the median age was 71 years (25th to 75th percentile: 68 to 75 years), 43.8% were women, 94.2% were white, and 59.9% of patients had a CHA2DS2VASc score of ≥3. Nearly 20% of patients had heart failure, 7.3% had previous myocardial infarction, 9.2% had previous cardiac surgery, 6.2% had cerebrovascular disease, 17.9% had chronic obstructive pulmonary disease, 18.5% had diabetes without complications, and 3.6% of patients had diabetes with complications.
Cardiac perforation occurred in 0.61% (n = 623) of patients within 30 days of the index ablation procedure. Cardiac perforation was more common among women, among those with a CHA2DS2VASc of ≥3, and among those with hypothyroidism, obesity, and a fluid and electrolyte disorder (Table 1). Risk of perforation was not associated with age (Figure 1) and was less common over time (Figure 2). Cardiac perforation was less common among procedures that used intraprocedural intracardiac ultrasound (ICE), in patients who had previous cardiac surgery, and procedures performed later in the study period (Table 1).
Unadjusted logistic regression analyses demonstrated a significant association between several variables and risk for cardiac perforation: female sex (odds ratio [OR]: 1.61; 95% confidence interval [CI]: 1.38 to 1.89; p < 0.0001); CHA2DS2VASc score of ≥3 (OR: 1.24; 95% CI: 1.05 to 1.46; p = 0.0093); previous cardiac surgery (OR: 0.16; 95% CI: 0.08 to 0.29; p < 0.001); nonuse of ICE (OR: 4.87; 95% CI: 4.14 to 5.74; p < 0.0001); hypothyroidism (OR: 1.38; 95% CI: 1.14 to 1.68; p = 0.0010); obesity (OR: 1.37; 95% CI: 1.11 to 1.69; p = 0.0030); and fluid and electrolyte disorders (OR: 1.46; 95% CI: 1.17 to 1.81; p = 0.0006).
An adjusted logistic regression analysis demonstrated female sex (OR: 1.34; 95% CI: 1.14 to 1.58; p = 0.0004), obesity (OR: 1.35; 95% CI: 1.09 to 1.68; p = 0.0050), and nonuse of ICE (OR: 4.85; 95% CI: 4.11 to 5.71; p < 0.0001) were significantly associated with increased risk for cardiac perforation, whereas previous cardiac surgery (OR: 0.14; 95% CI: 0.07 to 0.26; p < 0.0001) was significantly associated with decreased risk for perforation (Table 2). The c-statistic of the overall model was 0.7303 and was similar when the nonsignificant variables (fluid and electrolyte disorder and hypothyrodisim) were removed (c-statistic of 0.727).
The association between center volume and risk for perforation was assessed after categorizing sites by tertile of AF ablation volume, based on the 6 months before the study period. Cardiac perforation was slightly more common at the lowest volume sites (0.71%) compared with middle (0.54%) and high (0.58%) volume sites (p = 0.015). Although unadjusted models demonstrated that, compared with low volume sites, odds of perforation were lower at middle (OR: 0.77; 95% CI: 0.63 to 0.93) and high (OR: 0.82; 95% CI: 0.68 to 0.99) volume sites, there was no association between site volume and odds of perforation in the adjusted model (Table 3). The c-statistic in the adjusted model that included volume (Table 3) was 0.733, which was similar to the c-statistic for the adjusted model that did not include procedure volume (c-statistic = 0.730) (Table 2).
Predictors of perforation were subsequently assessed among the subset of patients (n = 76,134) who underwent ablation with intraprocedural ICE. Consistent with the overall analysis, obesity and female sex were associated with increased risk for perforation, and history of cardiac surgery was associated with a decreased risk for perforation (Table 4). Notably, prediction of perforation was less robust in this cohort (c-statistic was 0.5902), which indicated that a significant proportion of the discriminatory power of the overall model was due to ICE status.
A weighted risk score for cardiac perforation at the time of AF ablation was generated, with higher point values indicating increased risk. Points were assigned for female sex (1 point), obesity (1 point), nonuse of ICE (5 points), and previous cardiac surgery (−6 points); possible point values could range between −6 and 7, but notably not all point values were achievable. Risk for perforation increased linearly with increasing point score, from 0.05% (score of −6) to 2.55% (score of 7). The Central Illustration depicts the distribution of risk scores and the incidence of perforation by risk score group; unachievable point values (−3, −2, 3, and 4) and infrequently achieved values (−4; n = 325) were removed.
Sensitivity analyses were performed using generalized estimating equations to assess the impact of clustering by center on the study results. The results of these sensitivity analyses were consistent with primary analyses (Online Tables 3 to 5 show the sensitivity analyses, corresponding to Tables 2 to 4, respectively).
This analysis delineated predictors of cardiac perforation due to AF ablation in a large, representative, nationwide cohort of older patients, and had several important findings. Reassuringly, rates of cardiac perforation continued to decrease over time. Notably, one of the strongest independent predictors of cardiac perforation in our study was a modifiable factor: nonuse of ICE, and this was associated with a 5× increased risk for perforation. Other independent predictors of cardiac perforation were female sex, obesity, and absence of previous cardiac surgery. Patient characteristics associated with risk for perforation did not differ based on use or nonuse of ICE.
Cardiac perforation can occur during AF ablation via a variety of mechanisms. Common causes of perforation include mechanical trauma of thinner walls of the left atrium myocardium (e.g., left atrial appendage, left atrial roof), inadvertent pericardial entry during the transseptal puncture, and in the setting of steam pops during radiofrequency energy delivery. Due to use of high-dose anticoagulation during AF ablation, even minor instances of trauma (e.g., with placement of a coronary sinus catheter) could result in bleeding into the pericardial space with hemodynamic compromise.
This nationwide cohort analysis confirmed declining rates of cardiac perforation in the context of AF ablation. The rate of cardiac perforation declined in our study from 0.67% in 2013 to 0.52% in 2017. Several procedural factors likely contributed to the observed decline in cardiac perforation. Winkle et al. (11) demonstrated that adoption of a radiofrequency needle over a standard needle for transseptal access reduced perforation rates from 0.92% to 0.00% (p = 0.031). Avoidance of high power linear ablation significantly reduced the risk of tamponade in another study (12). Technological improvements in ablation catheters, including the widespread availability of contact force sensing catheters beginning in 2014, might represent an important reason for the lower rates of cardiac tamponade in the contemporary era. In a prospective observational analysis, use of contact force sensing catheters was associated with a lower incidence of tamponade versus noncontact force catheters (0.0% vs. 3.3%, relative risk 0.67; 95% CI: 0.63 to 0.72; p = 0.021) (13). In a meta-analysis that examined the impact of contact force technology, Shurrab et al. (14) suggested a reduction in the risk of tamponade using a contact force catheter, although this analysis failed to achieve statistical significance. Some studies suggested using a cryoballoon might reduce the risk for cardiac perforation (15,16), although this remains controversial (4).
Although use of ICE for AF ablation has increased over time (17), evidence suggests that its adoption has been higher in North America compared with other regions of the world. In a survey of the writing group from the most recent AF ablation guidelines statement (1), 53% routinely used ICE, including 87% of members from the United States and Canada and 13% of those from other countries. In this setting, no guideline recommendations were made regarding intraprocedural ICE use (1). Routine ICE use has many theoretical safety advantages, particularly for precise anatomic targeting during transseptal puncture and ablation, assessing catheter contact and early signs of impending steam pop or char formation during ablation, identification of left atrial thrombus, and for the early detection of pericardial effusion. Despite these potential benefits, there are limited data on the topic to guide practice. A large survey of cardiac tamponade with AF ablation reported no apparent relationship between ICE use and rates of perforation, although many of these centers were highly experienced (6). One study of peri-procedural imaging reported an increased risk of major bleeding with ICE use, although the relative contribution of vascular hemorrhage versus pericardial hemorrhage was not reported (17). However, this report (17) serves as an important reminder that use of ICE can be associated with complications due to relative catheter stiffness and the need for large bore vascular access. In the present study, we reported a high rate of ICE use during AF ablation (74.3%) and observed that use of ICE was associated with a significantly lower rate of cardiac perforation. This study, which was the single largest to compare ICE use and risk for cardiac perforation, suggested that intraprocedural ICE use should be considered as a recommendation in the next iteration of the AF ablation guidelines. Based on the costs associated with ICE and its infrequent availability in resource limited settings, further research is warranted to determine in which contexts AF ablation might be safely performed without ICE. Further research is needed to examine if procedural success also improves using ICE during AF ablation.
The Heart Rhythm Society has endorsed cardiac perforation within 30 days of an AF ablation as a performance measure to ensure quality in the setting of increased procedural dissemination (7,18). In its current form, the measure would require reporting of cardiac perforation at the provider level overall and after stratification by age (younger than 65 years versus 65 years or older) and sex (18). Consistent with several studies, our study identified female sex as a risk factor for perforation, supporting the need for sex stratification. In contrast, our study did not identify age as a risk factor for perforation in an exclusively older cohort of patients. Another large study (which included patients of all ages) did identify older age as a risk factor for perforation (4). Taken together, these data supported age stratification as reasonable. Of note, the relatively modest ability to predict cardiac perforation in the setting of ICE use (c-statistic of 0.5902) might make it difficult to adjust this performance measure for case mix, if desired in the future.
Previous cardiac surgery has been described as protective against cardiac perforation due to the protective effects of pericardial thickening and adhesions between the parietal and visceral pericardial reflections. This association has been described by others (19) and was confirmed by our study, in which a history of cardiac surgery was associated with approximately 85% lower risk for perforation.
Although obesity has been associated with an overall increase in peri-procedural complications in AF ablations (20), to our knowledge, the association between obesity and perforation has not yet been reported. The reasons for this association may be multifactorial and could include lipomatous infiltration of the interatrial septum that complicates transseptal puncture, poorer quality fluoroscopic imaging, and more difficulty manipulating catheters via the femoral region with substantial adiposity. This risk factor is particularly important because obesity can make rapid pericardiocentesis more challenging, potentially increasing the morbidity of cardiac perforation. Although weight loss has become an increasingly recognized strategy for reducing AF burden and improving AF ablation efficacy, it is important to also consider the potential salutary effect on peri-procedural safety.
Our study had several limitations. Ascertainment of baseline characteristics and outcomes relied on billing codes that might have limited sensitivity and specificity. As such, it was possible that comorbidity burden and event rates might have been underestimated or overestimated. Use of ICE was defined using billing codes; because its use is separately billable in the context of AF ablation, this variable might be more accurate than comorbidities. Because we could not distinguish between pre-procedure and intraprocedural transesophageal echocardiography, we were unable to determine whether intraprocedural transesophageal echocardiography might be a reasonable substitute for ICE for reducing the risk of cardiac perforation. Information on the technology used for AF ablation (radiofrequency ablation vs. cryoballoon ablation) and ablation lesion sets were not available, again due to the limitations of claims data. Obesity was defined using billing codes; therefore, we were unable to determine the relationship between body mass index (across the continuous range) and risk for perforation. This study was observational in nature; although we performed statistical adjustment, we could not rule out the possibility of residual confounding. Importantly, ICE use was not randomized and was at the discretion of the treating physician; as such, we could not rule out the possibility of residual confounding due to differences in ICE use among advanced tertiary care centers versus regional medical centers, and among patients with different comorbidity burdens and/or severities that were incompletely assessed by claims data. The study population included older adults with fee-for-service Medicare, and therefore, the results might not be generalizable to younger patients or those with different insurance carriers. The c-statistics for the models in this study were likely an overestimate of discriminative ability because the estimates were generated from the same cohort from which the model was derived. The adjusted models that assessed association between institutional volume and perforation were necessarily based exclusively on Medicare fee-for-service procedures. Therefore, the accuracy of these results were influenced by variability in payor mix across sites. Although it appears reassuring that in the contemporary era (with frequent use of ICE) there was no relationship between center volume and perforation (even when accounting for clustering by site), these results should be considered hypothesis-generating and require confirmation via future studies.
In contrast to these limitations, there were several strengths of our analysis. Our use of nationwide Medicare claims data facilitated the generalizability of these results because the analysis included all AF ablations, regardless of site characteristics. Moreover, the use of Medicare data also allowed for the capture of outcomes data without reliance on site reporting of complications, regardless of where a patient might have presented after discharge but before 30 days. Finally, these data did not include the patient selection bias inherent to clinical trials and prospective studies.
In this nationally representative cohort of older individuals with AF who underwent ablation, obesity, female sex, and nonuse of ICE were associated with increased risk for perforation, whereas previous cardiac surgery was associated with decreased risk for perforation. It was notable that one of the strongest predictors of perforation in our study, nonuse of ICE, was modifiable. These data have important implications for maximizing the safety of AF catheter ablation in the setting of rapid growth and dissemination of the procedure.
COMPETENCY IN MEDICAL KNOWLEGDE 1: The incidence of cardiac perforation as a complication of AF ablation continues to decrease over time.
COMPETENCY IN MEDICAL KNOWLEGDE 2: Female sex, obesity, and absence of intracardiac echocardiography were associated with increased risk for cardiac perforation, whereas previous cardiac surgery was associated with a lower risk for perforation.
TRANSLATIONAL OUTLOOK: Based on the costs associated with intracardiac ultrasound and its infrequent availability in resource-limited settings, further research is warranted to determine in which contexts AF ablation may be safely performed without intracardiac ultrasound.
This study was funded by Johnson and Johnson. Dr. Friedman has received research support from Boston Scientific, Biosense Webster, and Abbott; has received educational grants from Boston Scientific, Medtronic, Abbott, and Biotronik; and has received consulting fees from Abbott and AtriCure. Dr. Pokorney has received research support from Bristol-Myers Squibb, Pfizer, Janssen Pharmaceuticals, the Food and Drug Administration, Gilead, and Boston Scientific; has received consulting/advisory board support from Medtronic, Boston Scientific, Philips, Janssen Pharmaceuticals, Bristol-Myers Squibb, Pfizer, Portola, and Zoll; and has received DSMB support from Medpace. Dr. Kalsekar holds stock in Johnson and Johnson. Dr. Freeman has received fees from Janssen Pharmaceuticals, Medtronic, Boston Scientific, and Biosense Webster; has received research support from the American College of Cardiology. Dr. Khanna holds stock in Johnson and Johnson. Dr. Piccini has received grants for clinical research from Abbott, American Heart Association, Boston Scientific, Gilead, Janssen Pharmaceuticals, National Heart, Lung, and Blood Institute, and Philips; and has served as a consultant to Abbott, Allergan, ARCA Biopharma, Biotronik, Boston Scientific, Johnson & Johnson, LivaNova, Medtronic, Milestone, Oliver Wyman Health, Sanofi, Philips, and Up-to-Date. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
The 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
- atrial fibrillation
- confidence interval
- intracardiac echocardiography
- odds ratio
- Received November 12, 2019.
- Revision received January 24, 2020.
- Accepted January 29, 2020.
- 2020 The Authors
- Gupta A.,
- Perera T.,
- Ganesan A.,
- et al.
- Piccini J.P.,
- Sinner M.F.,
- Greiner M.A.,
- et al.
- Bollmann A.,
- Ueberham L.,
- Schuler E.,
- et al.
- Michowitz Y.,
- Rahkovich M.,
- Oral H.,
- et al.
- Friedman D.J.,
- Al-Khatib S.M.
- Shurrab M.,
- Di Biase L.,
- Briceno D.F.,
- et al.
- Chun K.R.J.,
- Perrotta L.,
- Bordignon S.,
- et al.
- Cardoso R.,
- Mendirichaga R.,
- Fernandes G.,
- et al.
- Steinberg B.A.,
- Hammill B.G.,
- Daubert J.P.,
- et al.
- ↵(2017) Quality ID #392 (NQF 2474): HRS-12: Cardiac Tamponade and/or Pericardiocentesis Following Atrial Fibrillation Ablation – National Quality Strategy Domain: Patient Safety. Available at:. https://qpp.cms.gov/docs/QPP_quality_measure_specifications/Claims-Registry-Measures/2018_Measure_392_Registry.pdf. 722020.
- Elayi C.S.,
- Darrat Y.,
- Suffredini J.M.,
- et al.
- Padala S.K.,
- Gunda S.,
- Sharma P.S.,
- Kang L.,
- Koneru J.N.,
- Ellenbogen K.A.