Prognosis of CRT-treated and CRT-untreated unselected population with LBBB in Stockholm County

Abstract

Aims

Left bundle branch block (LBBB) might be the first finding of cardiovascular diseases but also the prerequisite for cardiac resynchronization therapy (CRT) in heart failure (HF) with reduced ejection fraction (HFrEF). The prognosis for patients with LBBB and the implications of CRT in an unselected real-world setting are the focus of our study.

Methods and results

A central electrocardiogram (ECG) database and national registers have been screened to identify patients with LBBB. Predictors of HF and the use of CRT were identified with Cox models. The hazard ratios (HRs) of death, cardiovascular death (CVD), and HF hospitalization (HFH) were estimated according to CRT use. Of 5359 patients with LBBB and QRS > 150 ms, median age 76 years, 36% were female. At the time of index ECG, 41% had a previous history of HF and 27% developed HF. Among 1053 patients with a class I indication for CRT, only 60% received CRT with a median delay of 137 days, and it was associated with a lower risk of death [HR: 0.45, 95% confidence interval (CI): 0.36–0.57], CVD (HR: 0.47, 95% CI: 0.35–0.63), and HFH (HR: 0.56, 95% CI: 0.48–0.66). The age of over 75 years and the diagnosis of dementia and chronic obstructive pulmonary disease were predictors of CRT non-use, while having a pacing/defibrillator device independently predicted CRT use.

Conclusion

In an unselected LBBB population, CRT is underused but of great value for HF patients. Therefore, it is crucial to find ways of better implementing and understanding CRT utilization and characteristics that influence the management of our patients.

Graphical abstract

Open in new tabDownload slide

Introduction

In the general population, the prevalence of left bundle branch block (LBBB) is 0.19–0.82% and increases with age.¹ It is more common in the presence of hypertension, heart failure (HF), and coronary heart disease and is associated with an impaired prognosis.^2,3,4 Left bundle branch block further aggravates HF through delayed contraction of the left ventricle wall, but LBBB may also per se impair cardiac function over time and cause HF.⁵ Cardiac resynchronization therapy (CRT) partly restores the left ventricle activation pattern in LBBB and improves symptoms and prognosis. Therefore, CRT has a class I indication in current guidelines but is underused.⁶

The diagnosis of LBBB is easily obtained with routine 12-lead electrocardiogram (ECG) recording.⁷ We identified and characterized patients with LBBB with and without HF through a large central ECG database in Stockholm. We further identified patients with HF with reduced ejection fraction (HFrEF) with a CRT indication and compared outcomes in those with indications who were or were not treated with CRT and studied clinical predictors of use vs. non-use of CRT.

Methods

Data sources and study population

We queried the data from the central electronic ECG database MUSE (General Electric Healthcare) between 2000 and 2018 in an estimated population of almost 1 000 000 individuals. All ECGs from three hospitals (Karolinska Hospital-Solna, Karolinska Hospital-Huddinge, and Södertalje Hospital), from ambulances and elective ECGs in the larger Norrtälje area, were retrospectively considered. The ECGs from 432 108 unique adult subjects (>18 years of age) were retrieved and analysed by the MUSE system. Our selection criteria were LBBB with QRS duration ≥ 150 ms without right ventricular pacing or CRT. Our index ECG date was the first ECG fulfilling these criteria. Thus, 5359 patients were identified (Figure 1).

To obtain clinical characteristics, outcomes, and information about CRT or pacemaker (PM) treatment, we linked patients’ data with three national registries [the Swedish Pacemaker and ICD Registry (annual report accessible online http://www.pacemakerregistret.se), the Swedish National Patient Registry (including hospitalizations and outpatient visits), and the Cause of Death Registry] between 2000 and 2020 using the personal identification number used in all permanent residents in Sweden. New York Heart Association (NYHA) class, NT-proBNP, and HF medication at index ECG were not available.

Echocardiographic data were not stored digitally in the Stockholm area during this study period; therefore, such data were collected in patients with a history of HF or who developed HF by one investigator (P.G.) from electronic health records using written echo assessments of left ventricular ejection fraction (LVEF) and left ventricular end-diastolic diameter (LVEDD), when available, using a structured approach: the date of the first echocardiographic exam with LVEF ≤ 35% was considered if the LVEF was stable, below, or equal to 35% during follow-up or till CRT use; the last echocardiographic exam was considered if the LVEF was never observed below 35% or recovered over 35% during follow-up or till CRT use.

Timelines, definitions, and ethical considerations

Index ECGs were obtained from 1 January 2000 to 31 December 2018, and the ICD-10 codes were collected starting from one year before index ECG to the last available follow-up or the end of the study (31 December 2020).

We defined LBBB morphology as wide QRS, predominantly upright QRS with wide R and/or R′ in lateral leads (I and V6), and predominantly negative QRS in leads V1 and V2 with wide Q or S waves.

We defined HFrEF as a clinical HF diagnosis according to ICD-10 combined with stable LVEF ≤ 35%.

We estimated the date of class I indication for CRT (referring to the current 2021 EHRA/ESC pacing guidelines⁶) during follow-up of our cohort based on index ECG date and the first date of HF diagnosis with stable LVEF ≤ 35% (if present), whichever came last and, if not treated with CRT, under the condition of available follow-up of >12 months (survival of >12 months) and no previous history of permanent atrial fibrillation (AF) (persistent AF was not excluded).

The list of ICD-10 codes used to identify the clinical variables for the analysis is available in Supplementary material online, Table S1. The Charlson comorbidity index was calculated as follows: age 50–59 years = 1 point (pt), 60–69 years = 2 pts, 70–79 years = 3 pts, and ≥80 years = 4 pts; myocardial infarction (MI) = 1 pt; HF = 1 pt; peripheral vascular disease (PVD) = 1 pt; stroke or transient ischaemic attack (TIA) = 1 pt; dementia = 1 pt; chronic obstructive pulmonary disease (COPD) = 1 pt; musculoskeletal and connective disease (MSD) = 1 pt; peptic ulcer = 1 pt; liver disease = 2 pts; diabetes mellitus = 1 pt; chronic kidney disease (CKD) = 2 pts; history of cancer = 2 pts; hemiplegia = 2 pts; acquired immunodeficiency syndrome (AIDS); or human immunodeficiency virus (HIV) infection = 6 pts.

The local ethics committee approved the study protocol including data extraction and journal reading.

Statistical analysis

Categorical variables were presented as absolute numbers (%) and continuous variables as medians (interquartile ranges). Baseline characteristics were compared across groups by the Mann–Whitney test (if continuous) and by the χ² test (if categorical).

Univariable and multivariable Cox regression models were fitted to investigate patient characteristics independently associated with the development of HF after LBBB and CRT use after class I indication. Age was transformed into a binary variable using the integer number around the median (75 years old). Results were reported as hazard ratios (HRs) with 95% confidence intervals (CIs).

The outcomes considered coprimary endpoints were as follows: time to death for any cause, time to cardiovascular death (CVD), and time to first HF hospitalization (HFH).

Univariable and multivariable Cox proportional hazard regression models were fitted to compare the crude and adjusted risk of outcomes in patients with an indication for CRT, according to CRT use. Incidence per 1000 patient-years was calculated at 95% CI. A proportional hazard model for the subdistribution of the HR (SHR) was fitted to assess the cumulative incidence of CVD and HFH with all-cause death as a competing risk.

The proportional hazard assumption of the model was checked using the Schoenfeld residual method. Any violation of the proportional hazard assumption of the model was also considered and adjusted in the analysis splitting the time into two bands of 5 years. Data were censored on 31 December 2020 (end of study) or death.

All variables included in the multivariable regression models are reported in Supplementary material online, Table S2.

All analyses were performed using Stata/IC version 16.1. The level of significance was set to 5%, two-sided.

Results

A total of 5359 patients [1905 (35.6%) women, median age 76 (67–83) years] were included. Echocardiographic data were available in 3153 (86%).

Heart failure patients

In total, 3624 (68%) had a history of HF. Concerning index ECG date, 2158 (41%) subjects had a previous history of HF, 1466 (27%) developed HF after index ECG (Figure 2), and 1735 (32%) never developed HF.

Those who developed HF [1466 (27%)] did so after 416 (72–1237) days. They were younger and with less prevalence of comorbidities compared with those who had a previous history of HF but comparable to those who never developed HF (Table 1). The prevalence of HFrEF was higher in those who had a previous history of HF compared with those who developed HF after index ECG (58% vs. 43%).

At the time of index ECG, among patients without a previous diagnosis of HF, male sex, older age (>75 years old), and the diagnosis of COPD, diabetes, AF, PVD, CKD, and ischaemic heart disease were independent predictors of developing HF, while the presence of hypertension was an independent predictor of not developing HF (Figure 3); HRs are reported in Supplementary material online, Table S3.

Cardiac resynchronization therapy indication

As expected, most patients [4091 patients (76%)] did not meet the criteria for class I CRT indication (42% without HF diagnosis, 27% with an LVEF above 35%; 19% had a follow-up of <12 months, and in another 12%, it was not possible to confirm the indication) nor were they treated with CRT during the study period. In 214 patients who received CRT (4%), it was not possible to establish the class of indication because the echocardiographic data were missing or the available value of LVEF was over 35%. Of 1053 (20%) with a class I CRT indication, 630 (60%) were treated with CRT and 423 (40%) were not (Figure 1). The patient characteristics of those with class I CRT indication with or without CRT are reported in Table 2. Of those treated with CRT, 47% received CRT without defibrillation function (CRT-P) and 53% CRT with defibrillation function (CRT-D), and the median delay from indication to CRT use was 137 (35–378) days.

At the time of indication for CRT, among the variables tested in our models (see Supplementary material online, Table S2), the diagnosis of dementia, the age of over 75 years old, and the diagnosis of COPD were independent predictors of CRT non-use, while the presence of a PM or an implantable cardioverter defibrillator (ICD) was an independent predictor of CRT utilization (Figure 4); HRs are reported in Supplementary material online, Table S4.

Outcomes

In patients with a class I indication for CRT, after a median follow-up time of 4.0 (2.2–7.6) years from the date of indication, the incidence rate of all-cause of death, CVD, and first HFH for those not implanted with CRT compared with those implanted with CRT was 180/1000 person-years (95% CI: 161–203) vs. 62/1000 person-years (95% CI: 55–71), 113/1000 person-years (95% CI: 98–131) vs. 39/1000 person-years (95% CI: 33–47), and 553/1000 person-years (95% CI: 497–616) vs. 216/1000 person-years (95% CI: 197–238), respectively (Figure 5). After multiple adjustments, CRT use was associated with a lower overall risk of all-cause of death at 10 years from the indication [HR = 0.67 (95% CI: 0.45–0.99)]; CRT use was also associated with a reduced risk of CVD and first HFH at 5 years from the indication as individual endpoints [HR = 0.47 (95% CI: 0.35–0.63); HR = 0.56 (95% CI: 0.48–0.66)], but no significant risk reduction was observed at 10 years from indication. Crude and adjusted HRs are reported in Table 3.

Sensitivity analysis

In patients with a class I indication for CRT, considering non-CVD as a competing risk, the results were consistent: the use of CRT was associated with a lower risk of CVD and first HFH in the first 5 years from the indication (see Supplementary material online, Figure S1).

Discussion

In this analysis, we describe a large and unselected cohort of LBBB patients with a wide QRS duration (≥150 ms). As previously shown from several epidemiological studies, LBBB conduction abnormality is associated with comorbidities and worse prognosis and its prevalence increases with age and varies in accordance with cardiovascular disease profiles.^8,9 Furthermore, LBBB not only is easy to be diagnosticated but also has important therapeutic implications. In HF, electrical dyssynchrony further aggravates the HF condition.^10,11 Conversely, LBBB may cause left ventricular dysfunction and eventually HF over time, a condition that can be reversed by CRT.⁵

In terms of comorbidities, significant differences, as expected, were visible between those who had HF at the time of LBBB development compared with those who did not develop HF or developed HF after LBBB diagnosis.

In our cohort, 41% of patients had a previous history of HF. Although there are difficulties in comparison due to different populations’ selection criteria and definitions, these results are in line with previous findings in patients with LBBB.^1,9 Left bundle branch block is associated with HF and several comorbidities, and older age and male sex were associated with a higher risk of developing HF.¹²

As many clinical trials demonstrated the CRT benefit in terms of morbidity, mortality, and quality of life, it was also confirmed in this and other large unselected material, as many studies showed how CRT is still largely underused in Europe and Sweden. The magnitude of the problem varies based on eligibility criteria and population selection.^13,14 Lund et al.¹⁵ reported that CRT underutilization varies, depending on different consistency analyses, from 69 to 79% based on class I and IIa indications from 2013 EHRA. Linde et al.¹⁶ estimated that two-thirds of eligible HFrEF patients were not receiving CRT. In our study, the percentage of underuse was relatively lower (40%). This finding is mainly explained by our selection criteria that were limited, different from other studies, solely class I indication (HF with EF ≤ 35%, LBBB, and QRS ≥ 150 ms, without a history of permanent AF and survival of at least 12 months), based on the recent 2021 EHRA guidelines on cardiac pacing that had extended the indications in regard to symptoms but restricted the indications regarding morphology and QRS duration.⁶ Furthermore, the region is also well-supplied with hospital care and three CRT-implanting hospitals and our results cannot be extended to regions with worse healthcare access. We think that having limited the analysis to class I indication, a setting where CRT is proven to be beneficial without any doubt and that includes the majority of CRT device implantation in Europe,¹⁷ underlines, even more, that a big room for improvement is present in terms of therapy implementation. Notably, accurate data on NYHA class were not retrievable from our sources, and thus, all the patients with HF and LVEF ≤ 35% were considered symptomatic. However, although the recent 2022 American HF guidelines required NYHA class II-III or ambulatory IV to fulfil class I indication¹⁸ and less strength of evidence for CRT and NYHA class I is reported in 2021 EHRA/ESC guidelines on cardiac pacing⁶ and 2021 ESC/HFA guidelines on chronic HF,¹⁹ we assume that the majority of HF patients with LVEF ≤ 35% in our cohort, in accordance with NYHA class distribution in HFrEF, had NYHA class II or more.^20,21 Different from previous studies, where the indication was estimated based on a baseline time point data collection, the strength of our longitudinal data, which could partially explain the lower prevalence of CRT underuse, was the possibility to estimate the time when the indication was met during follow-up. This allowed us to restrict the indication only to those who lived >1 year and exclude those who had only a temporary indication. Furthermore, it was possible to calculate the delay from the indication to CRT use. A recently published large national perspective on the timing of CRT after HFH emphasized the association of total mortality and delay of CRT from the first HFH.²² Our results showed also how CRT implementation is affected by a substantial delay. Therefore, strong efforts have to be undertaken to timely adopt CRT when indicated. Interestingly, we evaluated the predictors of CRT utilization based on clinical characteristics at the time of indication; this might give more pragmatic evidence of what influences doctors’ decisions. First of all, the older age (>75 years old) was an independent factor of CRT non-use. It is important to underline how CRT in older age is associated with survival benefits irrespective of age.²³ Therefore, there is no evidence of holding back a beneficial therapy when indicated even in older patients. Thus, CRT is beneficial both for patients’ reported outcomes and for treatment facilitation, which is even more relevant in the elderly.²⁴ Dementia was also associated with CRT underuse, but perhaps this should not be the case, as a diminished risk for hospital care is relevant also in this patient group. On the other side, little evidence is available for patients with dementia; therefore, CRT has to be evaluated case by case, considering that patient acceptance might play a relatively bigger role. While many comorbidities were not associated with differences in CRT use in the multivariable analysis, COPD was still an independent factor associated with CRT underuse. In our analysis, males have a slightly higher likelihood of CRT use. Although not statistically significant, considering the different adherence to class I indication showed by Normand et al.¹⁷ and the results showed by Chatterjee et al.²⁵ on CRT-D, this finding could be relevant and not attributable to chance. The presence of a PM or an ICD was associated with CRT use. These findings can be explained by the fact that patients with a device had more frequent and specialized medical contacts such as with device follow-up clinics with higher awareness of CRT. However, it was not possible for us to adjust for this factor. Furthermore, the bias of patient acceptance for device therapy and its associated limitation has to be considered.²⁶ In the previous analysis of the Swedish HF registry, surprisingly, even after adjusting for NYHA class and organizational factors, a strong association was found between AF and a lower risk of CRT underuse.¹⁵ This finding was not confirmed in our analysis, where the history of AF was available at the time of CRT indication and patients with persistent and permanent AF were not considered having a class I indication for CRT if not implanted.

Outcomes

Patients with broad LBBB QRS morphology in sinus rhythm and symptomatic HFrEF are the perfect candidates for CRT, and although landmark clinical trials have been conducted more than two decades ago, CRT is still one of the most beneficial, with a lower number needed to treat medications available for these patients.²⁷ Our study showed that independent of age, sex, and comorbidities, CRT use is associated with lower all-cause mortality and cardiovascular mortality and less HFH in the first 5 years after the indication and, importantly, also after 10 years for the overall mortality. These results are consistent in the sensitivity analysis where non-CVD is accounted as a competing risk. Notably, as previously shown by Noordzij et al.,²⁸ models without the competing risk effect overestimate the outcome rate affected by the competing risk but the efficacy of CRT is still clearly present. Therefore, the competing risk of death on specific causes of death, which is relevant in the elderly and with constant improvement of medical treatment, has to be and is studied in clinical trials such as the RESET-CRT trial (NCT03494933) to determine whether the benefit overwhelms the risk of an expensive therapy like CRT-P vs. CRT-D.²⁹ These findings are crucial to stress furthermore the importance of not withholding a therapy when indicated and considering new strategies that may lead patients’ features to survival benefits like device remote monitoring.³⁰ Although some together with older age might be associated with lower CRT response, this should not lead to CRT underutilization as shown in the recent EHRA document by Mullens et al.¹⁴

Limitations

Our analysis is affected by several limitations. The observational design of the current study implies that causality could be extrapolated. Potential residual and unknown confounders despite a large adjustment have to be accounted for. Important information like medical treatment and socioeconomic characteristics were not available in our database, and we could not account for them, but we planned to implement and further analyse these aspects. Our results are not generalizable to non-LBBB or shorter QRS duration. Our study is limited to the Stockholm region, and the extrapolation of data like the CRT-P/CRT-D ratio may not be representative of other populations. The absence of medical treatment, NT-proBNP, and laboratory finding information is a major limitation since it was not possible to include them in the adjustments.

A selection bias in patients who received or did not receive CRT cannot be ruled out and may affect the outcome of our study. The definition of CRT indication and the characteristics were extrapolated by the available variable, and misclassification could not be completely excluded. Furthermore, it was not possible to exclude patients with persistent AF from class I indication for CRT because of the retrospective data and the overtime overlapping diagnosis with paroxysmal AF. The presence of undefined AF in our criteria for class I indication for CRT has to be considered since it implies a proportion of misclassification in accordance with the recent 2021 ESC-EHRA guideline indication. The expected relatively low numbers and the quality of the data sources should have unlikely impacted our results. The selection of a broad timeframe could increase the heterogeneity in CRT indications during the study period but, beyond allowing a longer follow-up, gives a broad perspective of CRT utilization. The succession of different CRT recommendations over the years limits the extrapolation of results to a single timeframe and affects the results. Therefore, a time-adjusted analysis was adopted. Many variables are based on ICD codes, and misclassification cannot be completely ruled out, but the longitudinal structure of our data allowed us to account for variation during the study period and reduce the misclassification and risk of missing data. Missing data were present, especially regarding echocardiographic parameters; under the assumption of random missingness, we think that this had a limited effect on the results.

Conclusions

Left bundle branch block is a marker of worse prognosis, but it is also a characteristic, in HFrEF patients, that could indicate that a very beneficial therapy is available (CRT). Our study and others indicate that CRT is beneficial but still affected by significant underuse. Therefore, it is crucial to find ways of better implementing CRT utilization and to better understand the characteristics that influence the management of our patients.

Supplementary material

Supplementary material is available at Europace online.

Funding

Unrestricted grant Medtronic, Institutional research funds.

Data availability

Data cannot be shared for ethical/privacy reasons.

References

Author notes