Safety Outcomes After Percutaneous Transcatheter Closure of Patent Foramen Ovale

Abstract

Background and Purpose

We sought to evaluate the real-world rate of safety outcomes after patent foramen ovale (PFO) closure in patients with ischemic stroke or transient ischemic attack (TIA).

Methods

We performed a retrospective cohort study using administrative claims data on all hospitalizations from 2005–2013 in New York, California, and Florida. Using ICD-9-CM codes, we identified patients who underwent percutaneous transcatheter PFO closure within 1 year of ischemic stroke or TIA. Our outcome was an adverse event occurring during the hospitalization for PFO closure, defined as in prior studies as atrial fibrillation or flutter, cardiac tamponade, pneumothorax, hemothorax, a vascular access complication, or death. Crude rates were reported with exact confidence intervals.

Results

We identified 1,887 patients who underwent PFO closure after ischemic stroke or TIA. The rate of any adverse outcome during the hospitalization for PFO closure was 7.0% (95% CI, 5.9–8.2%). Rates of adverse outcomes varied by age and type of preceding cerebrovascular event. In patients >60 years of age, the rate of adverse outcomes was 10.9% (95% CI, 8.6–13.6%) versus 4.9% (95% CI, 3.8–6.3%) in patients ≤60 years of age. The rate of adverse outcomes was 9.9% (95% CI, 7.3–12.5%) in patients with preceding ischemic stroke versus 5.9% (95% CI, 4.7–7.1%) after TIA.

Conclusions

Approximately 1 in 14 patients who underwent percutaneous transcatheter PFO closure after ischemic stroke or TIA experienced a serious periprocedural adverse outcome or death. The risk of adverse outcomes was highest in older patients and in those with preceding ischemic stroke.

Paradoxical embolism through patent foramen ovale (PFO) may lead to ischemic stroke and transient ischemic attack (TIA).^1–3 Observational data suggest that percutaneous transcatheter closure of PFO for secondary stroke prevention is safe and effective^4–8; however, three recent randomized clinical trials failed to show a clear benefit in the intention-to-treat analysis of PFO closure over medical therapy.^9–11 Based on these data, the American Heart/Stroke Associations and the American Academy of Neurology currently do not recommended performing PFO closure for secondary stroke prevention outside of clinical trials,^12,13 but these recommendations may change with publication of long-term follow-up from existing clinical trials and new data from two recently completed randomized trials that found a benefit of PFO closure for secondary stroke prevention.^14,15 While PFO closure continues to be debated, the procedure is routinely performed in the community.^4,16 In fact, the Food and Drug Administration (FDA) recently approved the use of the Amplatzer PFO occluder device to reduce ischemic stroke recurrence in patients with stroke presumed to be from paradoxical embolism.¹⁷ The rate of adverse outcomes ranged from 1.5–6.7% in randomized trials,^9–11 but few data exist regarding the rate of adverse outcomes in real-world practice. Therefore, we sought to evaluate the rate of adverse outcomes after transcatheter PFO closure for secondary stroke prevention during routine clinical practice in a large, heterogeneous cohort of patients from three geographically and demographically diverse states.

Methods

Study Design

We performed a retrospective cohort study using administrative claims data on all discharges from nonfederal acute care hospitals in California from 2005–2011, New York from 2006–2013, and Florida from 2005–2013. These dates were chosen to ensure at least 1 year of follow-up data for all study patients. Trained analysts used standardized methods to collect data regarding discharges and reported these to state health agencies for regulatory purposes. After quality checking, these data were provided in a deidentified format to the Agency for Healthcare Research and Quality for its Healthcare Cost and Utilization Project (HCUP).¹⁸ A deidentified personal linkage number assigned to all patients allows them to be longitudinally followed across ED encounters and hospitalizations over multiple years.¹⁹ Each encounter contains up to 25 International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) discharge diagnosis codes and 22 ICD-9-CM procedure codes.

In performing this study, we adhered to the Report of Studies Conducted Using Observational Routinely-collected Health Data guidelines for claims-based analytical studies.²⁰ H.K. was responsible for dataset acquisition and data stewardship. The institutional review board at Weill Cornell Medicine approved this study and waived the requirement for informed consent.

Patients

We identified all patients at the time of their first recorded hospitalization for a percutaneous transcatheter PFO closure using ICD-9-CM procedure code 35.52 (repair of atrial septal defect with prosthesis, closed technique) as per previous studies.^21,22 In order to identify patients who underwent PFO closure for secondary stroke prevention, we included only patients who had a cerebrovascular event within the 1 year prior to and during the same hospitalization for PFO closure. As per both the CLOSURE I and PC Trial, we defined a prior cerebrovascular event as either a TIA or ischemic stroke.^10,11 Ischemic stroke was defined by ICD-9-CM codes 433.x1, 434.x1, or 436 in any hospital discharge diagnosis position without a primary discharge code for rehabilitation (V57) or an accompanying diagnosis of trauma (800–804 or 850–854) or ICH (431) or subarachnoid hemorrhage (430).²³ This algorithm has been validated to have a sensitivity of 86% and a specificity of 95% when compared to medical records.²³ TIA was defined by the presence of ICD-9-CM code 435.x in any hospital discharge diagnosis position; this algorithm has a positive predictive value of 72%.²⁴ ICD-9-CM code 35.52 may be used for the repair of a true atrial septal defect as well as PFO, so we excluded any patients with a previous or concurrent documented history of congenital heart disease as defined by ICD-9-CM codes 745.1–745.4, 745.6–745.8, 746.x and 747.x.

Measurements

Our primary outcome was a composite outcome of adverse events occurring during the index hospitalization for PFO closure. Adverse advents were defined according to a previously used definition in patients undergoing percutaneous transcatheter cardiac procedures: atrial fibrillation or flutter, cardiac tamponade, pneumothorax, hemothorax, a major vascular access complication (consisting of hemorrhage/hematoma and vascular complication requiring surgical repair) or death (Supplemental Methods).²⁵ As our cohort included patients who already had a recent ischemic stroke or TIA, these variables were not included in our composite outcome given the lack of a well-validated algorithm for identifying recurrent cerebrovascular events using claims data. In a secondary analysis, we used previously well-validated ICD-9-CM code algorithms to include the following additional adverse outcomes reported in at least one of the three major randomized trials comparing PFO closure to medical therapy: myocardial infarction, venous thromboembolism, ventricular arrhythmia, infective endocarditis, and sepsis (Supplemental Methods).^{9–11,25–31} Subgroup analyses were performed stratified by patients’ age, gender, race, type of preceding cerebrovascular event (ischemic stroke or TIA), and the number of Elixhauser medical comorbidities (categorized into quartiles). The Elixhauser comorbidity index represents a comprehensive set of comorbidity measures for use with large administrative datasets and is a surrogate for overall medical illness.³² To ensure that the cerebrovascular event was the indication for PFO closure and not a complication of the procedure, we conducted a sensitivity analysis in which we excluded patients who had a cerebrovascular event and PFO closure during the same hospitalization.

Statistical analysis

Continuous variables are presented as mean (SD) and categorical variables as counts and percentages. Comparisons between groups were performed using the t test for continuous variables and χ² test for categorical variables. Crude rates are reported with exact confidence intervals (CI). All analyses were performed by A.E.M. and G.G. using Stata/MP, version 13 (StataCorp, College Station, TX). The threshold for statistical significance allowed for an alpha error of 0.05.

Results

Patient Characteristics

We identified 1,887 patients who underwent percutaneous transcatheter PFO closure after ischemic stroke or TIA. Mean age at the time of closure was 54.2 (±14.6) years and 916 (48.5%) patients were women. Preceding cerebrovascular events consisted of 514 (27.2%) ischemic strokes and 1373 (72.8%) TIAs. Most patients were ≤60 years of age, but 650 patients (34.4%) were older than 60 years of age.

Primary Analysis

The rate of any adverse outcome occurring during the index hospitalization for PFO closure was 7.0% (95% CI, 5.9–8.2%). Patients with adverse outcomes were older (60.6 versus 53.7 years, P <0.001); more often had Medicare or Medicaid insurance; more often had congestive heart failure, peripheral vascular disease, and chronic obstructive pulmonary disease; and had a higher medical comorbidity index (Table 1). Atrial fibrillation or flutter was the most common adverse outcome, occurring in 3.7% (95% CI, 2.9–4.6%) of cases. The rate of in-hospital death during PFO closure was 0.3% (95% CI, 0.1–0.6%) (Table 2). Our results were unchanged in a sensitivity analysis in which we excluded patients with a cerebrovascular event during the hospitalization for PFO closure; the rate of any serious outcome was 6.1% (95% CI, 4.8–7.4%). In addition, there was no change in the rate of serious adverse outcomes between 2005 and 2013 (P = 0.66 for trend across years).

Table 1.

Characteristics of Patients, Stratified by Occurrence of Adverse Outcomes After Patent Foramen Ovale Closure

Characteristica	No Adverse Outcome (N = 1,755)	Adverse Outcome (N = 132)	P Value
Age, mean (SD), y	53.7 (14.5)	60.6 (14.7)	<0.001
Female	844 (48.1)	72 (54.6)	0.15
Raceb			0.57
White	1,212 (70.6)	88 (66.9)
Non-white	505(29.4)	42 (33.1)
Payment source			0.01
Medicare	458 (26.1)	49 (37.1)
Medicaid	125 (7.1)	14 (10.6)
Commercial self-pay or other	1,172 (66.8)	69 (48.9)
Hypertension	790 (45.0)	64 (48.4)	0.44
Diabetes	265 (15.1)	18 (13.6)	0.65
Coronary heart disease	283 (16.1)	28 (21.2)	0.13
Chronic obstructive pulmonary disease	50 (2.9)	11 (8.3)	0.001
Tobacco use	194 (11.1)	15 (11.4)	0.91
Alcohol use	173 (9.9)	11 (8.3)	0.57
Elixhauser comorbidities,c mean (SD)	1.4 (1.3)	2.1 (1.9)	<0.001

Abbreviations: SD, standard deviation.

^aData are presented as number (%) unless otherwise specified.

^bSelf-reported by patients or their surrogates. Numbers do not sum to group totals because of missing race/ethnicity data in 2.2% of patients.

^cNumbers represent the number of Elixhauser comorbid conditions, which comprise a comprehensive set of comorbidity measures for risk adjustment.

Table 2.

Risk of Adverse Outcomes During Hospitalization for Patent Foramen Ovale Closure

Characteristica	Rate of Adverse Outcomes (95% CI)
Total adverse outcomes	7.0% (5.9–8.2%)
Individual adverse outcomesb
Atrial fibrillation/flutter	3.7% (2.9–4.6%)
Vascular complication	3.0% (2.3–3.9%)
Hematoma/hemorrhage only	2.7% (2.0–3.5%)
Cardiac tamponade/perforation	0.5% (0.2–0.9%)
Death	0.3% (0.1–0.6%)
Pneumothorax/hemothorax	0.1% (0–0.3%)

Abbreviations: CI, confidence interval.

^aData are presented as % (95% CI), unless otherwise specified.

^bNumbers do not sum to group totals because some patients had multiple outcomes.

In subgroup analyses, older patients, those with preceding ischemic stroke as opposed to TIA, and those with higher numbers of medical comorbidities were more likely to develop an adverse outcome or death after PFO closure (Table 3). The overall rate of adverse outcomes in patients >60 years of age was 10.9% (95% CI, 8.6–13.6%) as compared to 4.9% (95% CI, 3.8–6.3%) in patients ≤60 years of age (P <0.001) (Table 3). All in-hospital deaths occurred in patients >60 years of age. The rate of adverse events was 9.9% (95% CI, 7.3–12.5%) in patients with preceding ischemic stroke as compared to 5.9% (95% CI, 4.7–7.1%) in patients with preceding TIA (P = 0.002). The rate of adverse events in patients with the highest quartile of medical comorbidities was 13.3% (95% CI, 10.0–17.3%) as compared to 5.1% (95 CI, 3.5–7.1%) in patients with the lowest quartile (P <0.001 for trend across quartiles).

Table 3.

Risk of Adverse Outcomes After PFO closure, Stratified by Subgroup

Characteristica	No Adverse Outcome	Adverse Outcome	P value
Age			<0.001
≤60 years	1,176 (95.1)	61 (4.9)
>60 years	579 (89.1)	71 (10.9)
Sex			0.15
Female	844 (92.1)	72 (7.9)
Male	911 (93.8)	60 (6.2)
Raceb			0.49
White	1,212 (93.2)	88 (6.8)
Other	505 (92.3)	42 (7.7)
Indication for PFO closure			0.002
Ischemic stroke	463 (90.1)	51 (9.9)
Transient ischemic attack	1,292 (94.1)	81 (5.9)
Medical comorbiditiesc			<0.001
0	580 (94.9)	31 (5.1)
1	536 (94.2)	33 (5.8)
2	327 (94.2)	20 (5.8)
≥3	312 (86.7)	48 (13.3)

Abbreviations: PFO, patent foramen ovale.

^aData are presented as number (%) unless otherwise specified.

^bSelf-reported by patients or their surrogates.

^cNumbers represent the number of Elixhauser comorbid conditions, which comprise a comprehensive set of comorbidity measures for risk adjustment.

Secondary Analysis

In a secondary analysis including myocardial infarction, venous thromboembolism, ventricular arrhythmia, infective endocarditis, and sepsis as outcomes, the rate of any adverse outcome or death was 8.9% (95% CI, 7.6–10.3%).

In subgroup analyses of our secondary outcome, 13.8% (95% CI, 11.2–16.9%) of patients >60 years of age developed an adverse event as compared to 5.7% (95% CI, 4.0–7.8%) of patients ≤60 years of age (P <0.001). The rate of adverse events was 16.0% (95% CI, 12.8–19.5%) in patients with preceding ischemic stroke as compared to 6.3% (95% CI, 5.0–7.7%) in patients with preceding TIA (P <0.001). The rate of adverse events was 19.9% (95% CI, 15.7–24.6%) in patients with the highest quartile of medical comorbidities as compared to 5.7% (95% CI, 4.0–7.9%) of patients in the lowest quartile (P <0.001 for trend across quartiles).

Discussion

In a large, heterogeneous group of patients who underwent percutaneous transcatheter PFO closure after ischemic stroke or TIA during routine clinical care, the rate of adverse outcomes was 7%, a rate comparable to those of the three recent randomized clinical trials evaluating the efficacy of PFO closure for secondary stroke prevention. Complications were more common in older patients and in those with preceding ischemic stroke.

Percutaneous transcatheter PFO closure was first introduced in 1992 as an alternative therapy for secondary prevention of ischemic stroke and has since become more widely utilized.^16,33 The rate of periprocedural adverse outcomes ranged from 1.5–6.7% among three recent randomized clinical trials evaluating closure of PFO after ischemic stroke or TIA.^9–11 This wide range may reflect differences in patient characteristics, the definitions of adverse outcomes, the experience levels of the trial proceduralists, and the specific device used for PFO closure. The adverse outcome rate of 7% in our study was derived from a large and geographically and demographically diverse group of patients using a widely accepted definition of adverse outcomes after percutaneous transcatheter cardiac surgery.²⁵ When we also included all other adverse outcomes that were reported in RESPECT, CLOSURE I, and PC Trial, we found that 8.9% of patients undergoing PFO closure for secondary stroke prevention experienced an adverse outcome. Since our study included older patients (>60 years of age) and those with multiple medical comorbidities—groups that were not well represented in the PFO closure trials—our results may be helpful to consider alongside the clinical trial results when considering transcatheter PFO closure.

The results of our study must be considered in light of its limitations. First, although we used previously defined ICD-9-CM code schemas to identify adverse events, ascertainment of outcomes after transcutaneous PFO closure may be incomplete and we may have underestimated the rate of adverse events. Second, we lacked the granularity of data that is available in clinical trials and prospective observational cohorts. Specifically, we lacked data on the extent of diagnostic evaluation prior to transcatheter PFO closure (to ensure that the PFO was indeed the most likely stroke etiology), the specific device used for PFO closure, and antithrombotic therapies (medications and dosages) prescribed after the procedure. We also lacked data regarding the size of PFO, the number of microbubbles passing through the PFO, and the presence or absence of an atrial septal aneurysm. Furthermore, we lacked data on certain characteristics of adverse outcomes, such as whether atrial fibrillation was paroxysmal or persistent, which may have guided clinical decision making. This limitation should be weighed against the benefit of our study’s heterogeneous cohort. Third, although we identified our cohort using the standard ICD-9-CM code for PFO closure for secondary prevention of stroke, this code lacks specificity, and it is possible that we may have inadvertently included patients who underwent closure of another form of congenital heart disease. This limitation is unlikely to substantially affect our results given the relative rarity of other forms of congenital heart disease as compared to PFO. Furthermore, we attempted to mitigate this limitation by only including patients who had an ischemic stroke or TIA within the preceding 1 year of closure and by excluding patients with a previous or concurrent diagnosis of congenital heart disease. Fourth, we lacked outpatient records and thus could not reliably ascertain whether patients diagnosed with TIA ultimately were assigned an alternative diagnosis. Fifth, as we were unable to evaluate recurrent stroke given the unreliability of claims data in identifying recurrence, our results may underestimate the true rate of adverse outcomes following PFO closure. On the other hand, our results are based upon PFO closure procedures performed in the United States between 2005–2013 and therefore do not reflect recent advances in closure devices, improved experience of the proceduralists, and potentially better patient selection, which may all lead to decreased rates of adverse outcomes. Finally, as this was not a randomized trial, we lacked a medical management control group and are therefore unable to comment on what the outcomes of patients with paradoxical embolism would have been if medical management rather than transcatheter PFO closure were pursued.

The management of PFO in patients with ischemic stroke or TIA remains an ongoing challenge for physicians. In a large, real-world cohort of patients, the rate of adverse outcomes following PFO closure for secondary stroke prevention was similar to those reported in the three recent randomized clinical trials comparing PFO closure to medical therapy, although adverse events were more common in older and sicker patients who were not well represented in randomized trials. Given the current lack of clear and consistent evidence from randomized clinical trials definitively establishing a benefit of PFO closure for secondary stroke prevention, clinicians and patients should be aware of this real-world rate of serious adverse outcomes when considering PFO closure after ischemic stroke or TIA.