The distribution of continuous variables was analyzed using the Kolmogorov Smirnov test and the stem-and-leaf plot. Normally distributed variables were presented as mean (standard deviation [SD]), and differences were assessed by t Student and Chi-square tests. Variables with a non-normal distribution were expressed as median (interquartile range) and compared by Mann Whitney U test if they were not normally distributed. Qualitative variables are presented as percentages, and differences were analyzed by analysis of variance (ANOVA) test. Survival analyses were performed after verifying the proportional risk assumption by the Schoenfeld residuals test.

The risk estimates for post-discharge HF could be affected by patients’ survival status. Therefore, the usual techniques for time-to-event analysis would provide biased or un-interpretable results due to competing risks, and the Kaplan-Meier estimation will overestimate the HF incidence. To avoid such effects, we applied the model introduced by Fine and Gray [7] to test the competing events. The incidence of HF is presented in cumulative incidence function graphs and results of the multivariate analysis as sub-hazard ratio (sHR) and corresponding 95% confidence intervals (CIs). Harrell’s C-statistic test was used to assess the model’s discrimination. Calibration was tested by the Gronnesby and Borgan test. Patients lost during follow-up were categorized as missing, as well as those who lacked any of the main variables for the analyses, although these were < 5%.

Recurrent hospitalizations were evaluated by determining the incidence rate ratio (IRR) by a negative binomial regression. Because an increase in HF hospitalizations is associated with an increased risk of subsequent death, it has been suggested that any analysis of recurrent admissions should also account for death as a terminal event. Thus, coefficients from this method were estimated by accounting for the positive correlation between the recurrent outcome and death as a terminal event by linking the two simultaneous equations (rehospitalization count and death) with shared frailty [8]. Thus, within the same model, we obtained estimates of risk for both endpoints. Covariate selection was performed based on previous medical knowledge. The covariates included in the final predictive model for both endpoints were age, gender, diabetes, hypertension, previous coronary artery disease, HF, stroke, revascularization, left ventricle ejection fraction, glomerular filtration rate, hemoglobin, and medical treatments at discharge. Harrell’s C-statistics and calibration of the model were 0.701 (95% CI 0.600–0.791) and 0.873, respectively.

The final study sample consisted of 111 patients. The mean age was 73.9 ± 9.6 years, and 50 patients (45%) were women. A total of 74 (66.7%) were on BB, and the other 37 patients (33.3%) were not on BB treatment.

At the time of AVN ablation, 95.5% of the patients had current or previous AF, 37.8% typical or atypical atrial flutter and eight (7.2%) had other supraventricular tachycardia. The mean left ventricular ejection fraction was 44.8 ± 16.14%. Eighty patients (72.1%) had previous HF. Previous HF hospitalization was more frequent among patients with reduced ejection fraction (97.3%) than in those with mildly reduced (85.7%) or preserved ejection fraction (70.7%).

Fifty patients had a single-chamber pacemaker (45.0%), 21 a dual-chamber pacemaker (18.9%) and 36 a cardiac resynchronization device (32.4%). In all cases, the programming was set in VVI or VVIR modes. Eighty-five had a pacemaker (76.6%) and 26 an implantable defibrillator device (23.4%).

Table 1 describes baseline characteristics of the study population stratified by BB therapy. Patients on BB had a higher proportion of previous coronary artery disease (36.5% vs 10.8%, p = 0.003), HF (83.8% vs 48.6%, p < 0.001), cancer (18.9% vs 5.4%, p = 0.040) and lower mean left ventricular ejection fraction (42.36% vs 51.9, p = 0.018). The medical treatment did not differ significantly, except for a higher proportion of statins (56.8% vs 29.7%, p = 0.007) and diuretics (68.9% vs 45.9%, p = 0.019).

There were no significant differences according to the type of device: 27.0% of Implantable cardioverter defibrillators among patients on BB versus 16.2% patients off BB (p = 0.205) and 39.2% resynchronization versus 16.2% patients off BB (p = 0.066).

Follow-up was available for all patients, and after a median follow-up of 45.5 months (interquartile range 19.4–70.1), 43 patients had died (38.7%) and 31 (27.9%) had had an HF episode requiring hospital admission—14 patients just one HF hospitalization and the remaining 17 had recurrent hospitalizations. The number of HF hospitalizations ranged from 0 to 12, and the number of days hospitalized ranged from 0 to 99.

Patients treated with BB had a non-significant trend to higher mortality rates (Fig. 1). Cumulative incidence of first HF hospitalization was statistically higher in patients treated with BB (Fig. 2). Recurrent HF hospitalization rate was 74/1000 patients/year, and it was almost 2.5-fold higher in patients treated with BB (Fig. 3). The multivariate analysis, adjusted by age, sex, creatinine, left ventricle ejection fraction, coronary artery disease and previous HF, revealed a higher risk of recurrent HF hospitalizations in patients treated with BB (IRR 2.23, 95% CI 1.12–4.44; p = 0.023) (Table 2).

The effects of BB in HF vary depending on the type of patient. In patients in sinus rhythm and with left ventricular dysfunction, there is solid evidence of the benefits of their use, whereas in patients with AF or in those with preserved ventricular function, their benefits are more questionable. The Beta-blockers in Heart Failure Collaborative Group, in a meta-analysis with individual patient data, showed that while BB use was associated with a 27% relative reduction in all-cause mortality in patients with sinus rhythm, patients with AF did not show any benefit from being treated with BB. These results were similar when cardiovascular mortality or readmissions for HF were evaluated [5]. In our study, most patients had AF, and as opposed to patients in sinus rhythm, those in AF have no better prognosis when heart rate is lower [9].

Interestingly, in patients with paroxysmal AF, pulmonary vein isolation ablation seems to restore BB benefits in patients with HF [10].

In the setting of HFpEF, the benefits of BB are even less clear. Although most studies show benefits of BB for patients with mildly reduced ejection fraction, most report lack of benefits for patients with ejection fraction above 50% [4]. Despite this lack of solid evidence, most patients with HFpEF are currently being treated with BB [11, 12]. There is not a unique explanation for this fact, but we speculate that the main reasons are the lack of a pharmacological treatment for these patients and the efficacy of BB for prevention and treatment of the main triggers of HF: hypertension, AF, and coronary artery disease.

Hospital admissions for HF are a major health problem. They affect both patients with HF and with AF, and are associated with a marked reduction in survival and quality of life and an increase in healthcare costs [13‐15].

Pacemaker dependence is defined as the absence of an adequate intrinsic heart rhythm. It is a complex phenomenon that includes a variety of heart rhythm diseases: patients with sinus bradycardia or sinus arrest, bradycardic AF, complete or high degree AV block.

Patients may take up to 10 years to develop pacemaker dependency [16], but it is variable among patients and studies, and in most patients, we should expect to find progressive disease of the conduction system. Pacemaker dependence has also been estimated in cohorts of patients with defibrillators, and it is estimated that between 10 and 20% of patients receiving an implantable defibrillator will become pacemaker dependent in the follow-up [17, 18].

Although most patients are dependent or not, there are some with intermittent pacemaker dependency [6, 16]. In some cases, the heart rate is driven by the pacemaker, but in others, it is mainly driven by the patient’s intrinsic rhythm.

Pacemaker dependency has been associated with increased cardiovascular and all-cause mortality [19], highlighting the importance of appropriate pharmacological therapy.

BB have proven to reduce intraventricular dyssynchrony in patients with dilated cardiomyopathy and normal QRS [20], but theoretically it could increase the percentage of ventricular pacing in patients with conventional pacemakers and thus produce paced-induced cardiomyopathy [21].

Previous studies that have evaluated heart rhythm in HF patients have excluded those with pacemakers, and when they have considered them, they have done so as if they were a homogeneous group. Patients with cardiac pacing devices have different percentages of pacing dependence, and many patients with bi- or tri-chamber devices may have sinus rhythm that is perfectly sensitive to the effects of the BB and use the pacemaker to assure the transmission of AV conduction. In this study, all patients had any type of atrial tachycardia and AV node ablation, nullifying the chronotropic effects of the drugs.

BB have proven benefits in terms of ventricular function, ventricular remodeling, HF symptoms, morbidity and mortality [4, 22]. However, the mechanisms behind these effects are not fully understood. Possible explanations are the reduction of heart rate, reduction of cardiac afterload, the alleviation of neurohormonal effects, the reduction of intraventricular dyssynchrony, improvement of coronary flow reserve, antiapoptotic effects or the ability to counteract metabolic disturbances caused by adrenergic hyperactivation [20, 23‐25]. Although we mainly use heart rate as the reference for BB indication, titration and follow-up, it is only one of the multiple effects these drugs have on the circulatory system.

Some studies have demonstrated an increase in the central blood pressure associated with the BB use, which increases cardiac afterload, and an additional increase in the ventricular load caused by the augmentation of the ventricular volume and pressure secondary to the increased filling interval caused by the bradycardic effects [26, 27].

The effect of BB on circulating brain natriuretic peptide (BNP) and N-terminal-pro brain natriuretic peptide (NT-proBNP) levels is again complex, as BB have proven to markedly reduce natriuretic peptide levels in HFrEF [28], while it increases them in patients with HFpEF [29].

In addition, and probably with more effect in this specific population, bradycardia is not associated with a better prognosis in AF patients, as it is in patients in sinus rhythm [9, 30], and it has been speculated that BB treatment may favor the appearance of significant pauses and pause-dependent ventricular tachycardias [31].

With the analysis of the published evidence and our own results, we hypothesize that patients with HF and AF with a pacemaker-dependent heart rate do not benefit from BB therapy for the following reasons: First, the chronotropic action of the drugs has no effect in these patients. Second, the hypotensive effect of BB may not be well tolerated in patients with HF and may limit the use of other drugs with proven benefit, such as angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers or neprilysin inhibitors. Third, although BB have shown benefit in patients with AF and also in patients with HF, interestingly, this benefit seems to disappear when both conditions are combined, as is the case of our patients, and the addition of the AV node ablation makes the use of bradycardia drugs even more futile. Last, in polymedicated patients, any drug that does not provide significant benefit may exert a negative effect in terms of drug interactions, side effects and reduced therapeutic adherence to other drugs with proven benefits.

This study has several limitations that need to be acknowledged. First is the inherent limitations of retrospective studies, which are prone to selection bias and confounders that may not be sufficiently adjusted with the multivariate analysis. The selection of symptomatic patients who have required AV node ablation may have excluded those patients medical treatment with BB has managed to stabilize. In addition, some relevant variables such as body mass index were not available for inclusion in the multivariate analysis. Second, another limitation is the lack of continuous evaluation of BB therapy, type of BB and doses, as well as ejection fraction after the AV node ablation. Third, the small sample size constrains statistical power, which may both reduce the probability of finding an effect and exaggerate the magnitude of the effect found. And last, our highly selected population restricts the extrapolation of our results. On the contrary, the study of this peculiar population allows us to differentiate the chronotropic effects from the rest of the cardiovascular effects of BB.