Heart Failure Associated with Death Among Pediatric Patients Admitted With Existing Implantable Cardioverter Defibrillators

Background: There are no data regarding the incidence of death or factors associated with death among pediatric patients with implantable cardioverter defibrillators (ICDs). Out-of-hospital deaths are not tracked, however the nationally-representative Kids’ Inpatient Database (KID) may provide some insights. We hypothesized that death among patients with existing ICDs was most likely related to the presence of heart failure (HF) or cardiomyopathy rather than the underlying diagnosis of congenital heart disease (CHD), primary arrhythmia, in-hospital procedures or ICD complications during the admission. Methods: KID was used to identify all hospitalizations of patients less than 21 years of age admitted with an existing ICD for years 2000, 2003, 2006, 2009, 2012, and 2016. Hospitalizations with a procedure code for ICD implantation were excluded. Demographics and insurance status were collected. Using International Classification of Diseases 9th and 10th Revision Codes (ICD9 and ICD10), admissions were defined as primary cardiomyopathy disorders (e.g. dilated, hypertrophic), congenital heart disease (CHD), primary arrhythmia (e.g. Long QT, ventricular tachycardia) or other. Diagnosis of HF, ICD complications (e.g. device infection, mechanical complication, revision) during admission, and concomitant cardiac procedures (catheter or surgical) coded during admission were also collected. First, the number of hospitalizations of patients with existing ICDs per 100,000 pediatric hospitalizations by year was determined to evaluate trends in ICD admissions over time. Then independent variables were compared by univariable analysis using weighted totals for the primary outcome of hospital death. Mixed multivariable logistic regression analysis accounting for clustering by hospital, using stepwise backward elimination was then performed. Results: Of a weighted total 42,570,716 pediatric admissions, 3,966 had an existing ICD at the time of hospitalization, at a median discharge age of 18 years (IQR 15-20 years). Fifty-eight percent were male. The hospitalizations in patients with an existing ICD increased by five-fold from 2000 to 2016 (Figure 1, p < 0.0001). Among ICD hospitalizations, in-hospital death was noted in 55 (1.4%). By univariate analysis, variables associated with mortality were male sex, Black race, cardiomyopathy or CHD as a primary diagnosis, heart failure, or a cardiac intervention during the hospitalization. By multivariate analysis, only the presence of heart failure, adjusted OR 10.8 (95% CI 4.9-23.5, p<0.0001) remained significantly associated with mortality. Conclusions: The mortality rate among pediatric patients admitted with an ICD is low (1.4%). Among admissions with existing ICDs, mortality was significantly associated only with the presence of heart failure independent of age, race, sex, underlying diagnosis, or hospital-related procedure. This may suggest that among inpatients with ICDs, arrhythmias may not be the primary driver for death but rather heart failure, for which the ICD is not protective.