Relationship between mortality after ICD implantation and center volume in Belgium

Abstract Introduction In Belgium ICD implantation is restricted to 23 centers. A previous analysis of our group based on aggregated results per center showed that 3y mortality varied significantly between centers ranging from 7.5 to 23.4%. Multivariate analysis demonstrated that volume, infection rate and a higher proportion of implantations in primary prevention were predictors of 3y-mortality. These findings needed to be confirmed on a patient level since they could be caused by inter-patient rather than inter-hospital differences. Methods The QERMID-ICD database is a retrospective database of all patients implanted with an ICD in Belgium managed by the governmental health care institution (RIZIV/INAMI). Participation is mandatory for reimbursement. We analyzed data of 9896 new implantations performed between 2010 and 2016. Following patient characteristics were available: demographics (gender, NYHA class, primary vs secondary prevention, underlying heart disease, type of device, QRS duration, age and ejection fraction (EF)), comorbidities (atrial fibrillation, diabetes, COPD, neurological disease, oncological disease and renal failure), volume of center (low < median of 65 primo-implantations/year vs high >65 implantations/year) and the average income of the arrondissement in which the patient lived (low income < p25, median p25-p75, high > p75). The primary endpoint was 3y-mortality. Chi-squared test and Mann-Whitney U test with correction for multiple testing were used and multivariate logistic regression was performed to determine the corrected odds ratio for 3-year mortality. Finally, Kaplan-Meier survival analysis was performed. Results Low volume centers treated different patients than high volume centers. They implant more primary prevention (66.5 vs. 61.6%), more often patients with ischemic cardiomyopathy (49.8 vs 47.9%), less often arrhythmogenic heart disease (13.2 vs 16.6%) and patients with more co-morbidities and from communities with lower average income. High volume centers used more cardiac resynchronization therapy (26.8 vs 22.5%) despite no difference in QRS width. 1 and 3-y mortality were significantly higher in the low volume centers, respectively 5.6 vs. 4.4% and 16 vs. 11.1%. This was also confirmed in Kaplan Meier survival analysis. In multivariate logistic regression underlying heart disease, income, age, EF, NYHA class, CRT, indication and most comorbidities were significantly associated with mortality, but center volume remain