Age-Stratified Clinical Outcome in Patients with Known Heart Failure Who Receive Pacemaker, Resynchronization Therapy, or Defibrillator Implants

For patients with heart failure (HF), a cardiac implantable electronic device (CIED) can have various indications [1, 2]. A significant proportion of patients with HF also have intrinsic electrical conduction disturbances including fascicular block, bundle branch block, and atrioventricular (AV) block [3, 4]. The decision to implant a pacemaker or an implantable cardioverter defibrillator (ICD) in these patients depends on several factors such as the severity of HF, symptoms of bradycardia, other comorbidities, and patient preference. Choosing the wrong type of device may aggravate underlying HF and potentially have a negative impact on survival and risk of HF hospitalization, since right ventricular (RV) pacing introduces ventricular dyssynchrony [5]. On the other hand, using a more expensive device with a higher risk of complications, such as device infection and pocket bleeding, may not be cost-effective from a healthcare perspective and can be harmful in the short term from the individual patient’s perspective. The risk for pacemaker-associated HF increases with higher age, as does risk of other complications and competing morbidity/mortality risk [6]. A one-size-fits-all strategy is therefore not the best choice for selection of pacing devices in patients with prior HF. Age, patient preference and pre-existing comorbidities all need to be taken into account. A deeper understanding of the 5-year prognosis in this context may be helpful for choosing the best suited device for each patient. We therefore aimed to investigate age-adjusted survival and risk of HF hospitalization in a nationwide cohort of patients with prior HF, who had an indication to receive a de novo CIED.

From the nationwide Swedish Pacemaker Register, we identified all patients with prior HF diagnosis (ICD-10 code I50.X, I42.X, and I43.X) who had received a primary implant of a CIED during the period 2005–2018. The register includes all implanting centers in Sweden and covers >97% of all implants. Patients with leadless devices or without transvenous leads were excluded. Based on the unique Swedish personal identification number, data were then crossmatched with data from the Swedish Cause of Death Register and the Out-patient and In-patient Health Care Registers for diagnoses and mortality endpoints. The ICD-10 classification of the disease system was used throughout. The Registers and the Swedish system of crossmatching data using the personal identification number have previously been validated, and the positive predictive value of individual diagnoses has consistently been above 90% [7].

The primary study endpoint was all-cause mortality within 5 years post-implant, stratified for age intervals. In secondary time-dependent analyses, the risk of hospitalization for HF was evaluated, as was the mode of death with a particular focus on cardiac mortality (acute myocardial infarction, HF, or sudden cardiac death). Furthermore, the incidence of upgrades from pacemaker to cardiac resynchronization therapy (CRT) and/or implantable defibrillator was investigated.

Data are presented as median and (25–75th percentiles) for non-normally distributed continuous variables, which are compared using the Mann-Whitney U test. Categorical variables are presented as counts and percentages and compared using Fisher’s exact test or the χ² test as appropriate. Univariable Cox regression analysis was used to evaluate the hazard ratio for all-cause mortality in the whole cohort. All variables with significant associations to mortality were subsequently entered into a multivariable model. Five-year survival is presented as age-stratified Kaplan-Meier curves adjusted for relevant variables from the prior multivariable Cox regression analysis. These include age, prior atrial fibrillation, chronic ischemic heart disease, renal failure, diabetes, hypertension, chronic obstructive lung disease, malignancy, and cerebrovascular disease. In a pre-specified sensitivity analysis, the same tests were performed on patients with device indication “AV block,” to reduce some of the selection bias between groups. The assumption of proportional hazards was tested visually using log-log survival curves and was met in all cases. Data management was performed in SAS, and statistical analyses were performed using SPSS version 27.

From the total cohort of patients with a new-implanted device during the study period, 37,798 patients had a prior diagnosis of HF. Of these, 53 were excluded due to an unknown type of device. The remaining 37,745 patients were included in the study and split into groups depending on which type of device they had received (shown in Fig. 1). Upgrades to CRT were uncommon in the pacemaker groups (2.9–3.8%) but they were more common in the ICD groups (8.6–9.1%), as shown in Figure 2.

Baseline demographics and comorbidities are compared between groups and presented in Table 1. Patients receiving single-chamber pacemaker were older, had less ischemic heart disease, and almost all had a history of atrial fibrillation (93%). Similarly, in patients receiving single-chamber ICD, prior atrial fibrillation was more common (44%) than in the other device groups (p < 0.001). For other comorbidities, there were no major numerical differences between groups.

Within 5 years, 40% of the patients died, ranging from 17% for patients with dual-chamber ICD to 61% for patients with single-chamber pacemaker (Table 2). Cardiovascular mortality accounted for between 59 and 68% in all groups with only minor numerical differences, albeit significant in this large material. Death from HF was more common in those with CRT (23–25% compared to 15–20% for pacemakers, p < 0.001), whereas death from myocardial infarction was similar, around 9%. Sudden cardiac death classified as arrhythmia was more common in pacemaker patients (total 388 patients, 2%), but still very rare.

In Cox regression analysis, age and each of the included comorbidities were all independent predictors of all-cause mortality (Table 3). AV block as device indication was also independently associated with higher mortality (hazard ratio 1.17 [1.09–1.25], p < 0.001). In pairwise comparisons, patients with single-chamber pacemakers had significantly higher mortality compared to all other devices (all p < 0.001).

The risk of HF hospitalization at any time during the 5 years post-implant ranged from 28% (DDD pacemaker) to 39% (CRT-P). The vast majority of those hospitalized for HF only experienced one episode, and the median number of HF hospitalizations was 2 in all groups, whereas the mean number of hospitalizations showed slightly higher values in the CRT- and ICD-treated groups (Table 3).

Kaplan-Meier curves are presented per age group in the adjusted analyses stratified for device type (as shown in Fig. 3). Adjusted mortality was consistently highest for patients receiving single-chamber pacemakers, for all age groups, as shown in Kaplan-Meier curves in Figure 4. Adjusted risk of HF hospitalization within 5 years was highest for CRT recipients, patients with single-chamber pacemaker or ICD had medium risk, and those with dual-chamber pacemaker or ICD had the lowest risk.

Since different indications for CIED implant may confer selection bias not compensated for in the adjusted models, we also present age-stratified analyses focused on patients with indication “AV block” only. These patients are likely to receive a high percentage of ventricular pacing, regardless of CIED type, and therefore somewhat less subject to selection bias across groups. The Kaplan-Meier curves looked similar to those for all pacemaker indications, with single-chamber pacemaker patients having had the highest mortality (as shown in Fig. 4). This was the case for all age groups (as shown in online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000538529).

In this large nationwide cohort of patients with prior HF diagnosis, receiving pacemaker or defibrillator implants, we show that adjusted mortality was highest for patients treated with single-chamber pacemakers in all age groups. Five-year mortality ranged from 17% in patients with dual-chamber ICD to 61% in patients with single-chamber pacemaker, and death from HF occurred in all device groups. Similarly, the risk of new HF hospitalization was high (ranging from 28 to 39% within 5 years).

Patients with conventional pacemakers had the highest mean age, and the highest 5-year mortality, compared to those who received either ICD or CRT treatment. Mortality was closely associated with higher age, but nevertheless, age-adjusted mortality was highest for single-chamber pacemaker recipients in all age groups. On the contrary, the risk of new episodes of HF hospitalization within 5 years post-CIED implant was common in all groups, ranging from 28 to 36% in pacemaker-treated patients, 30–33% in ICD-treated patients, and 38–39% in CRT-treated patients. The finding that CRT treatment was associated with a higher risk of HF hospitalization but lower all-cause mortality compared to conventional pacemaker treatment may be regarded as surprising, since the major randomized controlled studies of CRT all showed reductions in HF hospitalizations, and it is unlikely that this improvement would be completely absent or even reversed in real-world data [8]. However, when looking more closely at mortality across groups, it is evident that while overall mortality was higher in the conventional pacemaker group, cardiovascular mortality was relatively higher in the CRT and ICD groups, mainly driven by an increase in HF mortality. These findings suggest that a CRT device was chosen for patients where HF symptoms were relatively more prominent, compared to the patients who received conventional pacing. The total mortality seems to be predominantly explained by the combined burden of all comorbidities, whereas cardiovascular and direct HF mortality correspond less to the number of comorbidities. It is known that one of the determinants of prognosis after CRT implant is NYHA class symptoms at the time of implant, and a limitation of this study is the absence of NYHA classification or other forms of HF symptom classification. Prior data from a large Swedish cohort of CRT-treated patients showed that 35–40% of patients were in NYHA class II, 54–59% in NYHA class III, and the rest in NYHA class IV, indicating that patients accepted for CRT treatment in Sweden have a significant burden of HF symptoms [9]. If patients in the CRT-treated groups had highly symptomatic HF, whereas the patients receiving conventional pacemakers had asymptomatic or well-controlled HF, then this difference may be a strong driver of cardiovascular morbidity and mortality and well explain the diverging outcome above.

Over time, symptoms may change, and for a patient with AV block who subsequently develops HF symptoms, it may be indicated to perform a change from conventional RV pacing to resynchronization therapy. However, upgrades from pacemaker therapy in this registry were rare, ranging from 2.9% (single-chamber pacemaker to CRT) to 9.1% (dual-chamber ICD to CRT). This would be in line with previous publications that suggest an underuse of ICD and CRT treatment in Sweden [10, 11]. Whether or not an upgrade is indicated depends primarily on the percentage of ventricular pacing, degree of HF symptoms, and overall frailty/perceived longevity for the patient. Typically, a high degree of ventricular pacing would be a prerequisite, and it is well known that ventricular pacing carries a risk of both de novo HF or deterioration of pre-existing HF due to pacing-induced or pacing-aggravated left ventricular dyssynchrony [8, 12]. In the present cohort, 87% of the patients with single-chamber device had an indication of AV block (mostly in combination with atrial fibrillation), suggesting a high need for ventricular pacing. Similarly, in the group of patients with dual-chamber pacemaker, 48% had AV block as an indication for the device, which associates with a high percentage of ventricular pacing. It is not possible to derive from the present data if more CRT upgrades would have had a positive impact on the prognosis for this cohort. However, HF hospitalization is a major cost for the healthcare system, and if episodes with hospitalization could be prevented, this would certainly be beneficial both from an individual and a societal perspective [13, 14].

Even though many patients without prior HF tolerate RV pacing for a long time without developing HF, prior studies have estimated the incidence of pacemaker-associated HF to be around 20% within 1–5 years post-implant [15]. The situation in patients who have already been diagnosed with HF is, however, quite different. These patients have already shown a propensity to develop signs and/or symptoms of HF and could potentially have an increased risk to aggravate their symptoms after induced ventricular dyssynchrony from RV pacing. There are no randomized trials to guide decisions in this group, and it is likely that there is substantial heterogeneity in the population. In this cohort, 82% (21,486 out of 26,178 eligible patients) of those who got a pacing device without a defibrillator (i.e., single chamber, dual chamber, or CRT pacemaker) did not get a resynchronization device. At the same time, overall mortality was highest for the patients receiving conventional pacemakers, suggesting that the clinicians may have chosen the type of device based on the presence of comorbidities and perceived longevity in the individual patient.

Whether or not the chosen device was the correct one in each individual case is not possible to deduce from registry data. But given that the number of patients with an a priori HF diagnosis at the time of the implant was substantial (representing 35% of all patients in this comprehensive national cohort), there is clearly a gap in evidence-based knowledge for guidance of choice of CIED in this setting. Both CRT and conduction system pacing have the potential to counteract the detrimental effects of ventricular pacing, and there may be a need for increased awareness about the risk of pacemaker-associated HF. Future trials are warranted to determine which type of CIED that should be used for which subgroup of patients with AV block and prior HF diagnosis.

This was a retrospective registry-based study, with all the inherent limitations of such a study design. We did not have a comparator group without pacemaker therapy, but since all patients had an indication for CIED therapy, we believe that this is a more relevant comparison than an actual control group. Even though we did our best to adjust for possible confounders and comorbidities, there will be residual bias which affects the results. Specifically, we did not have data regarding several variables known to influence prognosis in HF, such as HF symptom class, medication status, LVEF, or diastolic function. It is plausible that a proportion of the patients treated with conventional pacemaker had HF with preserved ejection fraction (HFpEF) phenotype, in which case CRT is not indicated [16, 17].

Registry data have limitations, but the national Swedish registers have all been validated and data quality is generally very good. Regarding the cause of death register, there is always an uncertainty and risk of underreporting of sudden cardiac death due to arrhythmia. This is an important endpoint whenever ICD discussions are brought up. Only a small minority of patients have their CIED interrogated after death has occurred, and even then, it is hard to tell whether a documented arrhythmia was the cause or a consequence of another lethal condition. The relatively low number of reported sudden cardiac deaths in this material suggests a risk of underdiagnosing the condition in the death registry. The comparison between groups of patients who receive different types of CIED for partly different indications has an impact on the validity of direct comparisons. Since there were notable differences in baseline demography between the groups, and the study is observational, it is not possible to make any direct inference that the choice of CIED type determines clinical prognosis. However, we believe that the nationwide size and high registry quality of the data in this material still make the analyses worthwhile and add value to the ongoing discussion regarding CIED choice.

In this large real-world cohort of CIED-treated patients with HF, demography and mortality data indicate that clinicians chose devices according to the overall status of the patient. Whereas all-cause mortality primarily was associated with higher age and burden of comorbidities, cardiovascular and HF mortality were higher in patients who received a CRT device. This finding likely reflects that CRT-treated patients had more symptomatic HF already at the time of implant. Regardless of CIED type, HF hospitalization episodes and mortality from HF were not negligible, and it is possible that there is an underuse of CRT in this setting.

This study protocol complies with the declaration of Helsinki and was reviewed and approved by the Swedish Ethical Review Authority, approval number (2021–05826-01). The need for informed consent was waived by the Swedish Ethical Review Authority in this decision (2021–05826-01).

R.B. has received speaker’s fees from Medtronic, Abbott, and Biotronik, has research grants from Boston Scientific, and is currently an employee of the non-profit Novo Nordisk Foundation. The other authors have no conflicts of interests.

This study was supported by ALF grants within the Swedish Health Care System (R.B.), Regional Funds in Region Skane (M.F.), Region Skane research grants (S.M.), and by grants from the Swedish Heart and Lung Foundation (R.B. and P.G.P.).

C.R. and R.B. drafted the manuscript. M.F., P.P., and S.M. are involved in managing the database from which the study material originated and were involved in conceiving the study idea, critical appraisal, and text revision. R.B. performed the statistical analyses and submitted the final manuscript.

The data that support the findings of this study are not publicly available on legal grounds, due to restrictions applied on extracted data from the National Death and disease registers in Sweden. Further inquiries can be directed to the corresponding author.