The Number of Recalled Leads is Highly Predictive of Lead Failure: Results From the Pacemaker and Implantable Defibrillator Leads Survival Study (“PAIDLESS”)

Abstract: Objectives. The purpose of this study was to determine if implantation of multiple recalled defibrillator leads is associated with an increased risk of lead failure. Background. The authors of the Pacemaker and Implantable Defibrillator Leads Survival Study (“PAIDLESS”) have previously reported a relationship between recalled lead status, lead failure, and patient mortality. This substudy analyzes the relationship in a smaller subset of patients who received more than one recalled lead. The specific effects of having one or more recalled leads have not been previously examined. Methods. This study analyzed lead failure and mortality of 3802 patients in PAIDLESS and compared outcomes with respect to the number of recalled leads received. PAIDLESS includes all patients at Winthrop University Hospital who underwent defibrillator lead implantation between February 1, 1996 and December 31, 2011. Patients with no recalled ICD leads, one recalled ICD lead, and two recalled ICD leads were compared using the Kaplan-Meier method and log-rank test. Sidak adjustment method was used to correct for multiple comparisons. All calculations were performed using SAS 9.4. P-values <.05 were considered statistically significant. Results. This study included 4078 total ICD leads implanted during the trial period. There were 2400 leads (59%) in the no recalled leads category, 1620 leads (40%) in the one recalled lead category, and 58 leads (1%) in the two recalled leads category. No patient received more than two recalled leads. Of the leads categorized in the two recalled leads group, 12 experienced lead failures (21%), which was significantly higher (P<.001) than in the no recalled leads group (60 failures, 2.5%) and one recalled lead group (81 failures; 5%). Multivariable Cox’s regression analysis found a total of six significant predictive variables for lead failure including the number of recalled leads (P<.001 for one and two recalled leads group). Conclusions. The number of recalled leads is highly predictive of lead failure. Lead-based multivariable Cox’s regression analysis produced a total of six predictive variable categories for lead failure, one of which was the number of recalled leads. Kaplan-Meier analysis showed that the leads in the two recalled leads category failed faster than both the no recalled lead and one recalled lead groups. The greater the number of recalled leads to which patients are exposed, the greater the risk of lead failure. 

The Pacemaker and Implantable Defibrillator Leads Survival Study (“PAIDLESS”) was one of the largest single-center studies examining defibrillator lead failure among the major United States manufacturers.¹ PAIDLESS demonstrated a significant impact on mortality and lead failure based upon recalled lead status. There were 1507 recalled leads (Sprint Fidelis, Medtronic, Inc; Riata and Riata ST, St. Jude Medical) implanted among 3802 patients over approximately a 16-year period.¹ The impact of multiple (more than one) recalled leads implanted in the same patient has not been previously addressed. This substudy analyzed the association between the number of recalled leads and defibrillator lead failure.

Methods

PAIDLESS was a retrospective study that was approved by the Institutional Review Board of Winthrop University Hospital. A total of 3802 patients with 4078 defibrillator leads implanted between February 1, 1996 and December 31, 2011 at Winthrop University Hospital were analyzed in PAIDLESS.¹ In this substudy, the patients were stratified into three groups: no recalled leads, one recalled lead, and two recalled leads. All recalled leads were implanted before any safety advisory or recalled notice had been issued. 

The PAIDLESS database included information related to patient characteristics, implant information, lead status, and survival, as determined by the Social Security Death Index.² The database was de-identified based on the Health Insurance Portability and Accountability Act (HIPAA).³ Recalled leads in PAIDLESS included the Medtronic Sprint Fidelis leads, as well as the St. Jude Medical (now Abbott) Riata and Riata ST leads.¹ According to the Medtronic System Longevity Study, lead failure is defined as failure to capture, failure to sense, abnormal pacing and/or defibrillation impedance, increased pacing thresholds, insulation defect, fracture, dislodgment, extra-cardiac stimulation, and/or cardiac perforation.^1,4 Lead failure analysis was blinded to both patient and operator identifiers.

Continuous data were presented as mean ± standard deviation, and the categorical data as proportion. The main endpoints were lead failure and mortality. Patient characteristics were compared between the recalled groups (no recalled, one recalled, and two recalled) using Kruskal-Wallis and Fisher’s Exact tests. Survival estimates and cumulative event rates were compared by the Kaplan-Meier method using the time to event approach. The log-rank test was used to compare the Kaplan-Meier survival curves between different recalled groups. Sidak adjustment method was used to correct for multiple comparisons. Unadjusted Cox regression analysis was performed using time-dependent lead failure variable as the endpoint. Variables with unadjusted P-values of <.25 were considered for a multivariable model, with the exception of co-linear variables, which were not included in the multivariable model. A stepwise multivariable Cox proportional hazard model was constructed to determine risk factors for lead failure. Mortality rates were compared between recalled lead groups using Fisher’s exact test and P-values were adjusted using Bonferroni method to correct for multiple comparisons.

All calculations were performed using SAS 9.4 (SAS/STAT 13.1; SAS Institute) for Windows and results were considered statistically significant when the P-value was <.05.

Results

Baseline patient characteristics (n = 3802) are shown in Table 1. There were 2326 patients in the no recalled leads group with 2400 leads implanted (59% of all leads), 1447 patients in the one recalled lead group with 1620 implanted leads (40% of all leads), and 29 patients in the two recalled leads group with 58 leads implanted (1% of all leads). No patient received more than two recalled leads. In general, patients who received two recalled leads were younger (66 ±     14 years) than those with one recalled lead (71 ± 13 years) or no recalled leads (70 ± 12 years; P=.01). The patients who received two recalled leads also had a shorter follow-up time (2 ± 1 years) than the other two groups (4 ± 3 years; P<.001). Coronary artery disease and prior percutaneous coronary intervention were less evident in the two recalled leads group (P<.05). The two recalled leads group exhibited a higher percentage of ischemic and/or dilated cardiomyopathy (52% and 38%, respectively) than the other groups. Likewise, the two recalled leads group also had a greater occurrence of congestive heart failure, left bundle-branch block, left ventricular dysfunction, and sustained ventricular tachycardia (P<.001). 

Table 2 shows the recalled manufacturer and models in the 29 patients who received two recalled leads. Nearly two-thirds of these patients received the same recalled lead twice: 8 patients received only Riata leads (28%); 6 patients received only the Fidelis leads (21%); and 4 patients received only the Riata ST leads (14%).

Figure 1 shows the Kaplan-Meier curve comparing lead failures over time for the no recalled, one recalled, and two recalled leads groups. Comparing the three groups collectively, the two recalled leads group failed at a significantly faster rate than the no recalled leads group and the one recalled lead group (log-rank P<.001). Most importantly, there was a significant difference noted in lead failure when compared across the three lead groups. The no recalled leads group had 60 failures (2.5%) and the one recalled leads group had 81 failures (5%). Strikingly, the two recalled leads group had 12 failures (21%; P<.001). 

Unadjusted Cox regression analysis was performed comparing a specific subset of patients in the one recalled lead group (patients who received one recalled lead replaced by a non-recalled lead) to patients in the two recalled leads group (patients who received a recalled lead replaced by another recalled lead). Patients who received a recalled lead replaced by another recalled lead had more than four times higher risk of lead failure compared with those whose replacement lead was a non-recalled lead (hazard ratio [HR], 4.4; 95% confidence interval [CI], 2.2-8.7; P<.001). Figure 2 shows the Kaplan-Meier curve comparing the patients in the one recalled lead group (recalled replaced by non-recalled) and the two recalled leads group (recalled replaced by recalled). The patients who received two recalled leads had a significantly faster rate of lead failure over time as compared with the patients who received only one recalled lead (log-rank P<.001). 

Table 3 shows the lead-based multivariable analysis performed to determine the independent predictors of lead failure. Independent predictors of lead failure were younger age, recalled status (presence of recalled lead as well as the greater number of recalled leads), prior percutaneous coronary intervention, rhythms (baseline sinus rhythm as well as the presence of significant conduction disease), combined insulation coating, and dual coil leads. Using the Sidak adjustment method, a significant difference in lead failure was found when comparing the no recalled leads group to the one recalled lead group and two recalled leads group, respectively (P<.001 for both comparisons). When comparing the one recalled lead group with the two recalled leads group, the adjustment resulted in an insignificant difference between the two groups (P>.50).

All-cause mortality rate was also compared across the three groups. There were 671 deaths (46%) in the one recalled lead group, which was significantly higher than in the no recalled leads group (870 deaths; 37%; Bonferroni adjusted P<.001). However, mortality rate in the two recalled groups (8 deaths, 28%) was not statistically significantly different when compared with the no recalled leads and one recalled lead groups. It is important to note that the cause of death was largely unknown.

Discussion

This PAIDLESS substudy found that the number of recalled leads implanted was associated with an increased rate of lead failure. Those with one recalled lead had a lead failure rate twice as high as found in those with no recalled leads. The two recalled leads group failed at an even higher rate, approximately five times higher than the one recalled lead group and about eight times higher than those with no recalled leads. Furthermore, the leads in the two recalled leads group failed more rapidly than both the one recalled lead and no recalled lead groups. 

Multivariable analysis was performed in order to see if the number of recalled leads implanted was a predictor of lead failure. In the original PAIDLESS patient-based multivariable analysis, recalled lead status was found to be an independent predictor of lead failure (HR, 2.08; P<.001).¹ This substudy, unlike the prior publication, performed a lead-based multivariable analysis using a Cox regression model, which further demonstrated that the number of recalled leads implanted was a predictor of lead failure especially when compared with those who did not receive any recalled leads. There was a significant difference in the effect of implanting one recalled lead and two recalled leads against implanting no recalled leads. 

Although all-cause mortality data appeared significant among the different groups, the authors are uncertain as to the significance of this finding given the large difference in group size, as well as the uncertainty regarding the cause of death. Therefore, at present, we are unable to conclusively determine the implications of this finding.

There have been a number of studies that retrospectively analyzed lead failures in order to determine predictive factors for lead failure. Multiple studies revealed that female gender, small-diameter leads, younger age, pulse generator replacement, and vascular access method may be associated with increased lead failure rates.^5-10 Other studies have evaluated the time to lead failure. The Sprint Fidelis lead initially showed an exponential lead failure rate, which becomes linear 2.9 years post lead implant.¹⁰ Another more recent study compared Sprint Fidelis and Riata leads, demonstrating similar failure rates in both groups.¹¹ However, a search in the United States National Library of Medicine database of the National Institutes of Health (PubMed) failed to identify any implantable defibrillator lead failure studies that evaluated the effects of multiple recalled leads in a given patient on either lead failure or mortality. 

Study limitations. This study has several limitations. First, its design is retrospective and observational in nature. Second, the sample sizes of the three groups were disproportionate. Specifically, the two recalled leads group had very few patients (only 29) as compared with the no recalled leads and one recalled lead groups (which each had over one thousand patients). Third, the disparity in group numbers as well as the uncertain mortality etiology make it impossible to derive any useful mortality conclusion from our analysis. Finally, the authors advocate for a larger, prospective study to fully elucidate the true impact of multiple recalled leads on mortality as well as to confirm these conclusions with respect to the impact on lead failure.

Conclusion

This study was the first to show that the number of recalled defibrillator leads implanted in a given patient is highly predictive of lead failure. Multivariable Cox regression analysis using lead-based data confirmed the six predictive factors to determine lead failure as found in PAIDLESS. The higher the number of recalled leads implanted, the higher the risk of lead failure. Patients who received two recalled leads had a double jeopardy situation, which resulted in a higher risk of lead failure than those who received one recalled lead or no recalled leads.

Disclosures: The authors report no conflicts of interest regarding the content herein.

This article was reprinted with permission from the Journal of Invasive Cardiology. 2016;28(12):E198-E202.

References

1. Cohen TJ, Asheld WJ, Germano J, et al. A comparative study of defibrillator leads at a large volume implanting hospital: results from the Pacemaker and Implantable Defibrillator Leads Survival Study (“PAIDLESS”). J Invasive Cardiol. 2015;27:292-300. 
2. Social Security Death Index cross-referenced with manufacturer-supplied data to ensure up-to-date status of out of service leads that are due to the death of the patient (OOS-D) Social Security Death Index https://www.genealogybank.com/gbnk/ssdi/.
3. The United States Human and Health Services. Guidance regarding methods for de-identification of protected health information in accordance with the Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule. Washington, DC: Government Printing Office, 2012.
4. Medtronic Criteria for Cardiac Rhythm Disease Management (CRDM) and System Longevity Study. Medtronic, Inc. website. https://wwwp.medtronic.com/productperformance/content/method_for_estimating_leads.html. Accessed September 5, 2015. 
5. Rordorf R, Poggio L, Savastano S, et al. Failure of implantable cardioverter-defibrillator leads: a matter of lead size? Heart Rhythm. 2013;10:184-190.
6. Aizawa Y, Negishi M, Kashimura S, et al. Predictive factors of lead failure in patients implanted with cardiac devices. Int J Cardiol. 2015;199:277-281.
7. Cheung JW, Al-Kazaz M, Thomas G, et al. Mechanisms, predictors, and trends of electrical failure of Riata leads. Heart Rhythm. 2013;10:1453-1459.
8. Lovelock JD, Patel A, Mengistu A, et al. Generator exchange is associated with an increased rate of Sprint Fidelis lead failure. Heart Rhythm. 2012;9:1615-1618.
9. Feldman AM, Kersten DJ, Chung JA, et al. Gender-related and age-related differences in implantable defibrillator recipients: results from the Pacemaker and Implantable Defibrillator Leads Survival Study (“PAIDLESS”). J Invasive Cardiol. 2015;27:530-534. 
10. Cheung JW, Tobin-Hess A, Patel A, et al. Trends in Fidelis lead survival: transition from an exponential to linear pattern of lead failure over time. Circ Arrhythm Electrophysiol. 2012;5:906-912.
11. Fazal IA, Shepherd EJ, Tynan M, et al. Comparison of Sprint Fidelis and Riata defibrillator lead failure rates. Int J Cardiol. 2013;168:848-852.

Funding: This study was submitted to each of the manufacturers listed in the manuscript (Medtronic, Boston Scientific, and St. Jude Medical); however, the study was only partially funded by Medtronic and Boston Scientific.

Presented in Part at Venice Arrhythmias 2015 (October 2015).