This is a single tertiary center retrospective study, evaluating the clinical outcomes, especially regarding inappropriate therapies and hospitalizations in patients with ICDs or CRT-Ds. We identified 160 patients who underwent an ICD or CRT-D implantation from January 2008 to May 2016 and excluded five patients due to follow-up loss and a lack of follow-up data, resulting in a total of 155 patients included in the study. For a comparison, we divided the patients into two groups: An ICD or CRT-D implanted before the year 2015 in which conventional ICD programming was applied (Group A, n = 94) versus that implanted after 2015 in which strategic programming with a long detection interval and higher rate cutoff was applied (Group B, n = 61).
Main difference after applying strategic programming was adopting monitoring without therapy in 165–180 beats per minute (BPM) zone in group B compared to group A. Also after strategic programming, default detection was prolonged to 30(or 24) out of 40(or 36) beats from 12 out of 24 beats in 181–200 BPM zone. This study was approved by the Institutional Review Board of Asan Medical Center (Seoul, South Korea; Institutional Review Board No. 2018-2261-0001).
The patient-related data included baseline characteristics such as the age and sex. Clinical characteristics including a history of arrhythmic events relating to sudden cardiac death (SCD), types of documented arrhythmias, underlying structural heart diseases, NYHA class, and a past medical history of hypertension, coronary artery disease, chronic kidney disease, and atrial fibrillation. In addition, the device-related parameters (type of the device, programming parameters, implanted and programmed dates, degree of prevention, and device-related complications) were gathered. Finally, follow-up information including the follow-up duration, occurrence of cardiac death or a heart transplantation, and whether appropriate/inappropriate therapy deliveries were given or not was reviewed. Reprogramming information, along with the changes in medications following events, was thoroughly reviewed.
Evaluation of ICD therapies and follow-up
The primary end point of this study was the number of inappropriate shock and hospitalization events before and after applying a long detection interval and higher rate cutoff for ventricular arrhythmias.
The definition of an inappropriate therapy was defined as any ICD therapy delivered when no VT (ventricular tachycardia) or VF (ventricular fibrillation) existed. The definition of appropriate shocks or therapies was defined as any ICD therapy delivered for VT or VF. All therapies were subdivided into ATP or shocks. If both ATP and a shock were delivered in a single episode, it was categorized into a shock episode. For the results, only the number of shock therapies was used for the inappropriate therapy analysis. Electrical storm(ES) events were defined by 3 or more sustained episodes of ventricular tachycardia, ventricular fibrillation, or appropriate shocks from an implantable cardioverter-defibrillator within 24 h. Investigators checked every implanted device on schedule. The interrogation schedule was as follows: on the day of the implantation, the day after the implantation, 1 month after the implantation, and every 3–6 months thereafter. If the patient had received an ICD therapy or was hospitalized for any reason, we performed the device interrogation on demand. During the interrogation of the device, the investigators reviewed whether a therapy was given or not, and if a therapy history existed, the stored EGMs were reviewed to determine their appropriateness.
Continuous variables were examined with the t test when appropriate and were expressed as the mean ± standard deviation. Continuous variables that were not normally distributed were described as the median ± interquartile range using a Mann–Whitney U test. We described the categorical variables using frequencies and compared them using a chi-square test or Fisher exact test as appropriate. To investigate the difference in the incidence rate of each outcome between Groups A and B, we used a negative binomial regression model, which is appropriate for count data with excessive zeros. To account for the different lengths of follow-up for the different patients, a log of the follow-up duration was included in the model as an offset term. To handle any slight overdispersion, we used a robust estimator of the standard error to calculate the 95% confident intervals and p-values. In the univariate analysis, we investigated the association between each candidate factor and the outcome. The factors that were statistically significant in the univariate analysis or that were deemed clinically important were entered into the multivariable analysis. The final multivariate model was determined by a backward elimination. There were no missing values in any of the factors or outcomes in the analysis. The statistical analyses were performed using R 3.5.1 statistical software (R Foundation for Statistical Computing, Vienna, Austria, 2018).