ABC | Volume 112, Nº5, May 2019

Original Article Warpechowski Neto et al Hospital readmission – implantable devices Arq Bras Cardiol. 2019; 112(5):491-498 and local hematoma subcomponents. There was no evaluation of CRT and ICD implants, with the sample being restricted to the conventional pacemaker implant. Nevertheless, the rates of infection and cable dislodgement were higher than that recorded in the present study. Infections An analysis of the PEOPLE 13 study (prospective cohort) evaluated 6,314 CIED implants at 44 centers. After 1 year, there were 633 deaths (10.1%), 548 (8.9%) non-infectious complications and 42 infections (0.56% in patients submitted to the first CIED implantation). In the multivariate analysis, the factors related to a higher risk of infection were the occurrence of fever in the 24 hours prior to the implantation, use of a temporary pacemaker and the need for early reintervention. Our sample found an incidence of pacemaker pocket infection in 2.5% of cases. The higher incidence, albeit in accordance with the literature, may be due to the fact that we did not evaluate the de novo implants and did not collect data regarding the use of a temporary pacemaker prior to the procedure. A Polish registry with 1,105 patients showed substantially lower infection rates when compared to the other recent studies: 0.1% at 2 months and 0.4% at the late follow-up of 2.4 years. 14 Although the antimicrobial prophylaxis is a well-established outcome factor of protection, 15,16 the registry differed from the others by the extended use of prophylaxis for a period of 5 days for surgical time >1h or immunosuppressive condition such as diabetes, chronic kidney disease, neoplasia or age >75 years. On the other hand, the percentages of device pocket hematoma were higher (6.1%), associated with platelet antiaggregation, triple therapy or, mainly, anticoagulation (present in 56% of patients). If, on the one hand, there is evidence that the occurrence of hematoma increases by 15-fold the risk of local infection, 13 prospective studies on the duration of anticoagulation do not show worse outcomes with their maintenance in the peri-implantation period: on the contrary, they show a decrease of events. 17,18 In a direct comparison between the two forms of stimulation, the ICD and CRT did not significantly differ, although absolute rates were higher in the ICD group. In the present study, there was a difference between the groups regarding the proportions of ischemic etiology, which was higher in the ICD group, and in the functional class, characterized by higher class I proportions in the ICD group and class III in the CRT group. Among the non-device-related hospitalizations, the difference in functional class did not translate into a statistical discrepancy regarding the percentage of unscheduled visits, either in absolute numbers or the specific causal etiology (Table 3). Although there were no recorded deaths, the ICD group showed a higher proportion of important events, such as chest pain in the ischemic scenario, HF decompensation and acute limb ischemia. Compared to predictors of cardiac mortality models in resynchronization therapy, 19,20 only 25% of the recorded deaths had an ejection fraction < 25% (specifically 28, 20, 58 and 29%) and 75% used loop diuretics at doses of 80mg/day or more. Right ventricular contractile dysfunction, an important factor associated with mortality, was not specifically analyzed in the present study. Inappropriate shocks, an important source of emergency consultations, were recorded in only 2 patients (14.28% of device-related visits) – both with cardioverter defibrillators. Compared with recent national data, 21 the present study displayed a large difference regarding pocket hematoma rates, showing a much lower percentage in our cohort, but a higher percentage of unplanned device-related readmissions (7% vs . 3.6%). It is important to emphasize that the current study did not account for the implantation of pacemakers without the ICD or CRT functions, situations that represented the majority of patients that were initially candidates for follow-up (Figure 1) and who, in fact, were included in a similar study, 21 hindering the direct comparison between the findings. Furthermore, it should be remembered that the two assessed populations showed very different percentages of patients in functional class I and II (84.8% vs . 31.5% in this study). Limitations Among the main limitations of the study is the small sample size when compared to the larger series, and the follow‑up period duration, which may have been short for some outcomes and prevents the direct comparison with the larger cohorts in the literature; the analysis of the MIRACLE-ICD study subgroup, 19 for instance, suggests that left ventricular cable dislodgements becomes more frequent in the long term, so our follow-up may have underestimated the occurrence of this complication in the CRT group. Also, we emphasize the fact that the data represent the practices and the results of a single cardiology center in the south of Brazil, with the limitations of unicentric studies on the extrapolation of results. Compared with recent local literature, 21 there are methodological differences regarding patient eligibility (mainly related to the type of device eligible for evaluation) and, consequently, considerable differences in baseline functional class and contractile function that limit the direct comparison between the studies regarding readmission predictors. The difference between etiology and functional class between the groups is also a factor to be remembered. Although not statistically significant regarding the percentage of unscheduled visits to the emergency unit, regardless of whether or not it is associated with the procedure, the population’s characteristics could lead to different results in the long-term follow-up, given the chronicity of the underlying diseases and their several forms of temporal evolution. Conclusion The results showed that patients submitted to CRT implantation, when compared to the ICD implantation cases, show a higher probability of mortality in the follow-up 496