Results: 
A total of 372 patients (69.5% male; 48.6 ± 14.4 years old) underwent S-ICD implantation during the study period. The most common indications for S-ICD implantation included primary prevention in patients with heart failure with reduced ejection fraction (39.3%), secondary prevention of ventricular tachyarrhythmias (31.1%), and primary prevention in patients with hypertrophic cardiomyopathy with high-risk features (17.2%). There were 22 (5.9%) patients (81.8% male; 52.1 ±13.2 years old) who underwent S-ICD extraction during a median follow-up period of 4.4 [2.0-6.5] years. The baseline characteristics of the study population are listed in Table 1. For all patients, no extraction tools were required to extract the S-ICD, highlighting that S-ICD extraction, when indicated, is uncomplicated.
The median length of time between implantation and extraction was 39.6 [8.3-64.6] months. The most common indications for S-ICD extraction were need for bradycardia pacing (frequency among extractions, 18.2%), infection (22.7%), and inappropriate shocks due to oversensing (22.7%) (Figure 1). Other common indications for S-ICD extraction included need for cardiac resynchronization pacing (9.1%), failure to shock (4.6%), heart transplantation (9.1%), and patient preference (13.6%). No patients required S-ICD extraction for ATP. Programming changes were unsuccessful in two of the five patients who presented with inappropriate shocks due to oversensing. For the other three patients, oversensing could not be overcome with changes in programming. One patient had lead coiling and significant retraction of the lead body, raising concern for Twiddler’s syndrome. For the second patient, the S-ICD electrode causing excessive noise was under the Boston Scientific advisory for premature electrode fracture; thus, the device was extracted. The third patient had already undergone revision of her S-ICD parasternal lead in years prior, but re-presented with inappropriate shocks due to oversensing of noise that was reproducible with isometric exercises in all three vectors. Univariate analysis demonstrated that a history of smoking (P=0.01), lower left ventricular ejection fraction (P=<0.001), and history of atrial fibrillation (P=0.02) were associated with S-ICD extraction (Table 1).
A multivariate analysis was conducted of all factors with P<0.05 on univariate analysis as well as BMI, given its established increased risk of shock failure in obese patients.16,17 This analysis revealed that both a history of smoking (p=0.02) and BMI (p=0.03) were independently associated with S-ICD extraction (Table 2). Specifically, patients whose devices were extracted smoked, on average, 13.9 pack years more than those patients who did not undergo device extraction (t (370)=3.99, P=0.0001). Patients whose devices were extracted had BMIs, on average, 3.91 kg/m2 higher than those patients who did not undergo device extraction (t (370)=2.31, P=0.02).
Subgroup analyses by indication for extraction were performed. Only variables with a priori clinical association with extraction for a particular indication were assessed to avoid multiple testing. In comparing those whose devices were extracted for all pacing needs (N=6) with those whose devices were not extracted, no significant differences in baseline PR interval (P=0.78) or QRS duration (P=0.39) were found on univariate analysis. Patients who underwent S-ICD extraction for inappropriate shocks (N=5) had higher BMIs than patients who did not undergo extraction (P=0.01). Finally, compared to patients who did not undergo device extraction, patients whose devices were extracted due to infection (N=5) were more likely to have a history of smoking (P=0.03), have a history of prior pocket infection (0.01), and have elevated BMIs (P<0.001) (Table 3). The bacteria found to be responsible for these new pocket infections included Stenotrophomonas maltophilia, Streptococcus agalactiae, Streptococcus mitis, Staphylococcus aureus, and Enterococcus.