Impact of pre-existing conduction abnormalities on permanent pacemaker implantation after transcatheter aortic valve replacement: A meta-analysis

Abstract

BACKGROUND

Transcatheter aortic valve replacement (TAVR) has become the standard treatment for severe aortic stenosis, but permanent pacemaker implantation (PPI) remains a frequent complication. Despite several predictive models and single-center cohorts, prediction algorithms remain poor.

AIM

To synthesize evidence on baseline conduction abnormalities and their association with PPI after TAVR.

METHODS

We conducted a systematic review and meta-analysis in accordance with PRISMA 2009 guidelines. We searched PubMed, Cochrane, and EMBASE for papers published between January 2010 and February 2026. Adults (> 18 years) having TAVR with documented baseline electrocardiographic status and post-procedural pacemaker outcomes were considered eligible for inclusion. The pooled proportions and odds ratios were estimated using random effects models. Heterogeneity was evaluated using I² statistics.

RESULTS

Twenty studies comprising 15515 patients were included. The pooled incidence of new PPI was 18% (95%CI: 0.14-0.24). New-onset left bundle branch block (LBBB) occurred in a pooled proportion of 38% of patients (95%CI: 0.27-0.50) across 11 studies. New-onset RBBB was reported separately from new-onset LBBB in only two cohorts and could not be pooled. Pre-existing RBBB was the strongest predictor of pacemaker requirement, conferring a 4.52-fold increased risk (95%CI: 2.64-7.74), with moderate heterogeneity (I² = 50.3%). Left ventricular ejection fraction showed inconsistent associations with pacemaker needs.

CONCLUSION

Baseline RBBB in particular demonstrated a strong and consistent association with PPI. These findings should inform preprocedural risk stratification, patient counseling, procedural planning, and enhanced postprocedural monitoring strategies for high-risk patients.

Key Words: Transcatheter aortic valve replacement; Permanent pacemaker implantation; Conduction abnormalities; Right bundle branch block; Aortic stenosis; Meta-analysis; Systematic review

Core Tip: Baseline right bundle branch block is associated with a 4.52-fold increased odds of permanent pacemaker implantation after transcatheter aortic valve replacement (95%CI: 2.64-7.74; I² = 50.3%), underscoring the clinical relevance of preprocedural electrocardiographic risk stratification in this population.

INTRODUCTION

Transcatheter aortic valve replacement (TAVR) has transformed the management of severe aortic stenosis, and its use has expanded from high-risk toward lower-risk populations worldwide[1,2]. With broader indications and rising procedural volumes, the balance of procedural benefit and device-related complications has become increasingly important for patient selection and perioperative planning[1,3]. Among the most frequent and clinically consequential complications are conduction disturbances, including new left bundle branch block (LBBB), high-grade atrioventricular block, and permanent pacemaker implantation (PPI), which remains a leading post-TAVR adverse event[3,4]. While procedural factors such as valve type and implantation depth have been widely studied as determinants of PPI, less is known about how pre-existing conduction abnormalities on baseline electrocardiogram (ECG) independently influence post-TAVR conduction outcomes. Identifying patients at higher risk based on baseline ECG findings may improve procedural planning, patient counseling, and post-procedure monitoring.

TAVR involves catheter-based delivery of a bioprosthetic valve, most commonly balloon-expandable or self-expanding, positioned within the native aortic annulus under fluoroscopic and echocardiographic guidance without the requirement for cardiopulmonary bypass in the majority of cases. Upon deployment, the prosthetic frame compresses and displaces the native calcified leaflets against the aortic wall. The conduction system, including the bundle of His and the proximal left bundle branch, courses through the interventricular membranous septum immediately beneath the non-coronary cusp of the aortic valve (AV). Mechanical compression, local edema, or ischemia in this region following valve deployment may produce a range of conduction injuries, from transient first-degree atrioventricular block to complete heart block. Reported risk factors for post-procedural conduction disturbance include greater implantation depth, larger prosthesis-to-annulus ratio, and pre-existing conduction disease.

A large synthesis of 108 studies comprising 77538 patients reported 14560 patients requiring PPI within 30 days after TAVR, illustrating substantial short-term event rates[5]. Prospective cohorts report PPI rates of roughly 10%-15% at one year, emphasizing persistent clinical relevance beyond the periprocedural window[6]. Valve type, implantation depth, and procedural approach influence PPI risk in many analyses, underscoring multifactorial causation[5,7].

Preexisting ECG conduction disease is a consistent and strong determinant of PPI after TAVR; key abnormalities include right bundle branch block (RBBB), existing AV block, hemiblocks, and bifascicular patterns[4,5,7]. Recognizing these baseline findings permits stratified counseling and procedural planning, which is critical. RBBB is markedly associated with increased risk of PPI and has been identified as a major predictor included in predictive scores[4,5]. High-degree AV block (Mobitz II or third-degree) carries a very high relative risk for PPI and ranks among the strongest predictors in network analysis[5]. Bifascicular or hemiblocks are associated with an elevated risk for PPI across pooled studies[5]. Short membranous septum and other anatomical factors confer an additive risk when combined with conduction disease and have been incorporated into combined electrocardiographic and imaging risk models[4]. Baseline conduction disease commonly interacts with anatomic and procedural factors to amplify PPI risk and forms the foundation of emerging risk scores such as RITMO and newer two-step models[4,8].

Baseline conduction disease influences not only the immediate need for pacing but also downstream outcomes, including pacing burden, heart failure progression, and device-related complications that may compromise long-term benefits of TAVR[8,9]. Despite several predictive models and single-center cohorts, prediction algorithms remain poor. Management techniques, such as timing of pacing, role of preventive pacemakers, and use of electrophysiological testing, continue to be debated[6,10,11]. This systematic review aims to synthesize contemporary evidence on baseline conduction abnormalities and their association with PPI after TAVR, clarify the magnitude of risk for specific ECG patterns, assess the impact on clinical outcomes, and evaluate gaps in prognostic tools. The objectives are to: (1) Quantify associations between predefined baseline conduction disorders and PPI after TAVR; (2) Evaluate downstream clinical consequences of PPI in these populations-including long-term pacing dependency, left ventricular ejection fraction (LVEF) trajectory, and mortality; and (3) Identify methodological and evidence gaps to inform future research and clinical guidance[4-6]. The present systematic review was undertaken to address three specific gaps not fully resolved in prior syntheses: (1) Prior meta-analyses have not incorporated studies reporting outcomes with contemporary balloon-expandable platforms (SAPIEN 3, SAPIEN 3 Ultra) or the newest self-expanding generations (Evolut FX, Navitor), which carry distinct conduction profiles compared with earlier devices; (2) No prior systematic review has simultaneously synthesized PPI incidence, new-onset RBBB, new-onset LBBB, long-term pacing dependency, and LVEF trajectory within a single integrated analytical framework; and (3) Existing analyses have not performed formal valve-type subgroup analysis, study-era stratification, or meta-regression by publication year to quantify temporal trends in PPI rates-analyses that the present review now provides.

MATERIALS AND METHODS

Population, intervention, comparison, outcome, and study design framework

We used the PRISMA 2009 guidelines and the ‘Population, Intervention, Comparison, Outcome, and Study design (PICOS)’ scheme to generate the eligibility criteria[12]. Studies were considered eligible for inclusion if they were published between January 1, 2010, and February 19, 2026. The target population included adults (≥ 18 years) undergoing TAVR for aortic stenosis. Studies were considered for inclusion if they presented quantitative data about comparative outcomes. Following this, we prospectively registered the study protocol in PROSPERO (CRD420261296830). Table 1 further elaborates the PICOS scheme for the current study.

Table 1 Population, Intervention, Comparison, Outcome, and Study design framework for the literature search.

Domain	Criteria
Population (P)	Adults (≥ 18 years) undergoing TAVR for aortic stenosis
Intervention/exposure (I)	Pre-existing conduction abnormalities on baseline ECG (e.g., RBBB, LBBB, first-degree AV block, bifascicular block, other baseline conduction disease)
Comparison (C)	TAVR patients without pre-existing conduction abnormalities (normal baseline ECG)
Outcomes (O)	Primary: Requirement for PPI after TAVR. Secondary (if reported): New-onset conduction disturbances, timing of pacemaker implantation, length of hospital stay, short-term mortality
Study design/setting (S)	RCTs, prospective and retrospective observational cohorts, and case series (n > 10) and registry analyses, published in English between January 1, 2010, and February 19, 2026 (inclusive); full text available

AV: Atrioventricular; ECG: Electrocardiogram; LBBB: Left bundle branch block; PPI: Permanent pacemaker implantation; RBBB: Right bundle branch block; RCT: Randomized controlled trial; TAVR: Transcatheter aortic valve replacement.

Eligibility criteria

Inclusion criteria: We included randomized controlled trials, cohort studies, and case series with more than 10 participants, published in English between January 1, 2010, and February 19, 2026, that enrolled adult patients (≥ 18 years) undergoing TAVR for aortic stenosis. Eligible studies reported baseline ECG or conduction status and allowed identification of patients with and without pre-existing conduction abnormalities (e.g., RBBB, LBBB, first-degree AV block, bifascicular block, or other baseline conduction disease), and they reported the primary outcome of PPI after TAVR or provided sufficient data to extract it. When available, we also included studies reporting additional outcomes such as new-onset conduction disturbances, timing of pacemaker implantation, hospital length of stay, or short-term mortality. For the purposes of this review, the primary outcome of PPI was defined as new PPI during the index hospitalization or within 30 days of the TAVR procedure, consistent with the most widely reported convention across the included studies. Studies that defined PPI outside this window are noted in Table 2, and variation in outcome timing definition is acknowledged as a source of clinical heterogeneity.

Table 2 Results of the systematic review.

Ref.	Study design	Study location	Population characteristics	Intervention	Primary outcomes	Key findings
Muntané-Carol et al[18] (2021)	Prospective multicenter	Canada, United States	TAVR recipients without prior pacemaker	Minimalist TAVR with 14-day AECG monitoring	Occurrence of delayed HAVB or CHB	4.6% (21/459) developed delayed HAVB/CHB, leading to PPM in 81%
Pavlicek et al[19] (2023)	Prospective cohort	Germany	Severe AS; 54% male; mean age 80	TAVR with SEV (n = 103) or BEV (n = 100)	HAVB requiring PPM within 30 days	7% (15/203) required PPI; new LBBB (OR = 15.7) and diabetes (OR = 3.8) were predictors
Meduri et al[28] (2019)	RCT (REPRISE III)	United States, Germany	High/extreme surgical risk; mean age 83	Lotus mechanically-expanded vs CoreValve SEV	PPM within 30 days and dependency at 1 year	PPM higher with Lotus (34%) vs CoreValve (18%); dependency was 43% at 30 days
Castro-Mejía et al[36] (2022)	Retrospective multicenter	Spain	Severe AS; 39% male; median age 83	New-generation SEVs (Evolut, Acurate, Portico, Allegra)	Peri-procedural modification of AV conduction	17.7% required PPI; valve recapture (OR = 2.8) and depth (OR = 1.9) were predictors
Auffret et al[22] (2017)	Multicenter registry	Global (Canada, France, Spain, etc.)	AS candidates; 50% male; mean age 82	TAVR grouped by baseline RBBB	Cumulative all-cause mortality	Baseline RBBB in 10.3%; 30-day PPM in RBBB (40.1%) vs others (13.5%)
Bagur et al[23] (2012)	Case-matched cohort	Canada	Elderly AS; matched by baseline ECG	TAVI (Edwards) vs isolated SAVR	Complete AVB and PPM within 30 days	PPM higher in TAVI (7.3%) vs SAVR (3.4%); RBBB predicted TAVI PPM (OR = 8.6)
Wasim et al[30] (2025)	Prospective observational cohort	Norway	Severe AS; 50% male; mean age 80.6	TAVI (various valves) over low vs high volume phases	New PPM ≤ 30 days and all-cause death	31.6% total PPI rate; dropped from 45.8% to 23.9% with experience
Chen et al[25] (2024)	Registry analysis	United States, Canada	Intermediate/high risk PARTNER 2 S3 registries	TAVR with balloon-expandable SAPIEN 3	New PPM within 30 days	12.5% required PPI; RBBB (OR = 5.8) and depth > 6 mm (OR = 1.86) were predictors
Dizon et al[31] (2015)	RCT and registry analysis	United States, Canada	Inoperable/high-risk AS (PARTNER trial)	TAVI (Edwards SAPIEN)	1-year all-cause mortality	New PPM (6.8%) and prior PPM (22.9%) independently predicted mortality
Natarajan et al[37] (2022)	Prospective cohort	Canada	Outpatient AS; mean age 81.8	Routine 2-week pre/post rACM	Compliance to rACM and unplanned post-TAVI PPMI	15.6% total PPMI rate; 12.5% rACM notifications; unplanned PPMI only 3.1%
Fraccaro et al[33] (2011)	Registry analysis	Italy	AS due to calcification; mean age 81	TAVI with CoreValve system	Incidence/predictors of PPM	39% required PPI; predictors were depth (P = 0.039) and pre-existing RBBB (P = 0.046)
Guetta et al[20] (2011)	Registry analysis	Israel	AS Israeli registry; mean age 83	CoreValve TAVI	Development of HDAVB within 30 days	36% developed HDAVB (40% total PPM); RBBB (OR = 43) and depth (OR = 22) were predictors
Husser et al[38] (2016)	Prospective cohort	Germany, Switzerland	High-risk AS; mean age 80-81.	SAPIEN 3 (n = 96) vs SAPIEN XT (n = 87)	New IVCA and PPM implantation	PPM rate 12.5% (S3) vs 12.6% (XT); S3 had more fascicular blocks (17% vs 5%)
Khawaja et al[27] (2011)	Retrospective multicenter	United Kingdom, Ireland	United Kingdom CoreValve Collaborative; mean age 81.3	CoreValve TAVI	PPM within 30 days	33.3% required PPI; predictors included peri-pro AVB (OR 6.29) and predilatation (OR 2.68)
Kooistra et al[24] (2020)	Retrospective multicenter	Netherlands, Spain	Multi-center European registry; median age 82	TAVI (SEV vs BEV vs Other)	Timing of onset and predictors of late CDs	12% PPM rate; 18% of these were late (> 48 hours); IVCD (OR = 3.3) and RBBB (OR = 2.6) predict late PPM
Kostopoulou et al[32] (2015)	Randomized prospective	Greece	NYHA II/III AS; mean age 81	CoreValve TAVI with EPS vs ECG alone	Predictors of conduction abnormalities	22% PPM within 1 month; baseline HV interval (cut-off 52 ms) was prognostic
Ledwoch et al[34] (2013)	Prospective registry (GARY)	Germany	German GARY registry; mean age 81.5	TAVI (CoreValve vs Sapien)	Predictors of PPM up to 30 days	33.7% PPM rate; independent predictors: CoreValve (OR = 2.86), porcelain aorta (OR = 1.64)
Naveh et al[21] (2017)	Prospective observational	Israel	AS at Hadassah center; mean age 80.7	TAVI (CoreValve vs Sapien)	Predictors for long-term pacing dependency	34.5% total PPI; 68.4% dependent; RBBB (OR = 18) and delta PR > 28 ms were predictors
Rampat et al[26] (2017)	Retrospective multicenter	United Kingdom	United Kingdom LOTUS experience; mean age 81.2	LOTUS mechanically-expanded valve	PPM up to hospital discharge	31.8% PPM rate; 55.2% new LBBB; pre-procedural block composite (OR = 2.54) and absence of valve calcification (OR = 0.55) were independent predictors
Van Gils et al[29] (2018)	Prospective cohort (CONDUCT)	Netherlands	Rotterdam cohort; mean age 79	TAVI (various valves) with daily ECG monitoring.	QRS dynamics vs PPM requirement	23% PPM rate; only persistent QRS prolongation led to PPM in normal baseline patients

AS: Aortic stenosis; AV: Atrioventricular; AVB: Atrioventricular block; BEV: Balloon-expandable valve; CHB: Complete heart block; CD: Conduction disturbance; ECG: Electrocardiogram; EPS: Electrophysiology study; HAVB: High-degree atrioventricular block; HDAVB: High-degree atrioventricular block; HV: His-ventricular; IVCA: Intraventricular conduction abnormality; LBBB: Left bundle branch block; LOTUS: LOTUS Edge valve system; LVEF: Left ventricular ejection fraction; NYHA: New York Heart Association; OR: Odds ratio; PPM: Permanent pacemaker; PPI: Permanent pacemaker implantation; RBBB: Right bundle branch block; RCT: Randomized controlled trial; rACM: Remote ambulatory cardiac monitoring; SAVR: Surgical aortic valve replacement; SEV: Self-expanding valve; TAVI: Transcatheter aortic valve implantation; TAVR: Transcatheter aortic valve replacement.

Exclusion criteria: We excluded cross-sectional studies, case reports, reviews, grey literature, non-English publications, studies with ≤ 10 participants, animal studies, and reports lacking data on baseline conduction status or post-TAVR pacemaker outcomes. Duplicate datasets were screened, and the most complete report was retained.

Search strategy

We searched through several digital databases to retrieve relevant literature. Among these are Cochrane, PubMed (MEDLINE), and EMBASE. Other resources, such as those from independent journals, were also included. The information was compiled using databases as well as journals such as “Transcatheter* Aortic Valve Replacement” [Mesh] OR “Transcatheter Aortic Valve Implantation” [tiab] OR “TAVR” [tiab] OR “TAVI” [tiab] OR “transcatheter aortic valve”[tiab] AND “Cardiac Conduction System Disease” [Mesh] OR “Bundle-Branch Block” [Mesh] OR “Atrioventricular Block” [Mesh] OR conduction abnormalit*[tiab] OR conduction disorder*[tiab] OR conduction disease*[tiab] OR “bundle branch block”[tiab] OR RBBB[tiab] OR LBBB[tiab] OR “right bundle branch block”[tiab] OR “left bundle branch block”[tiab] OR “AV block”[tiab] OR “atrioventricular block”[tiab] OR “first degree AV block” [tiab] OR bifascicular [tiab] OR ECG [tiab] OR electrocardiogra* [tiab] AND “Pacemaker, Artificial” [Mesh] OR pacemaker* [tiab] OR “permanent pacemaker” [tiab] OR “pacemaker implantation” [tiab] OR PPI [tiab] AND (“2010/01/01” [Date-Publication]: “2026/2/19” [Date-Publication]) AND English [lang] AND Humans [Mesh]. Equivalent structured search strategies employing database-specific controlled vocabulary (Emtree terms for EMBASE; MeSH for MEDLINE; Cochrane CENTRAL subject headings) were applied across all three databases. Manual reference list screening of all 20 included studies was performed to identify any eligible reports not captured by the electronic database search. No grey literature sources, trial registries, or citation-tracking databases were searched beyond the three primary databases; this is acknowledged as a limitation of the present review. The complete search strings for all three databases, including database-specific controlled vocabulary, are provided in full in Supplementary material appendix 1 and 2.

Selection process

The study methodology was developed based on a review of relevant peer-reviewed literature. Articles meeting the predefined inclusion criteria were appraised using the PICOS framework to ensure methodological strength. The screening and selection process was facilitated using Rayyan.ai, an evidence-based platform designed to streamline the review of primary and secondary sources[13]. Following this, a total of 20 studies were found to be appropriate for inclusion. Articles were excluded if they targeted a non-relevant population, used the wrong study design, failed to measure target outcomes, or exhibited a high risk of bias. In several cases, studies were excluded for more than one of these reasons. After finalizing the secondary screening process, we assessed the overall sample size (n = 20) of the selected literature. To create a PRISMA 2009 flow chart that follows the reporting standards of the PRISMA 2009, we used articles from reputable journals and other sources[14] (Figure 1).

Figure 1 PRISMA 2009 flow diagram illustrating the literature search and study selection process. Records were identified from PubMed (n = 176), Cochrane (n = 128), and EMBASE (n = 58). After removal of duplicates and ineligible records, 20 studies comprising 15515 patients were included in the final systematic review and meta-analysis.

Risk of bias

We conducted a systematic analysis of patient demographics, the characteristics of the interventions, and the region of the study. For all the studies utilizing a randomized controlled design, we used the Cochrane Risk of Bias (ROBv2) tool to determine the bias in each study and across all domains[15]. The results of the risk assessment were represented as a ‘traffic lights’ plot, and the individual risk domains were summarized via a ‘summary plot’. For cohort studies, the Newcastle-Ottawa Scale (NOS) was used to summarize the risk across the included studies[16]. The NOS was applied using the cohort-specific version, which is appropriate for prospective and retrospective cohort studies rather than case-control designs; this dual-tool approach is consistent with Cochrane handbook recommendations for systematic reviews incorporating both randomized and non-randomized evidence. Domain-level abbreviations in Table 3 are as follows: Rep. Exposed = representativeness of the exposed cohort; Sel. non-exp. = selection of the non-exposed cohort; Ascert. exp. = ascertainment of exposure; Outcome absent = demonstration that outcome of interest was not present at start of study; Ctrl main = comparability on main confounders; Ctrl addit. = comparability on additional confounders; Assess. outcome = assessment of outcome; Follow-up dur. = adequacy of follow-up length; Adequacy FU = adequacy of follow-up of cohorts. Yes = criterion met; No = criterion not met.

Table 3 Risk of bias assessment via Newcastle-Ottawa Scale for observational studies.

Ref.	Rep. exposed	Sel. non-exp.	Ascert. exp.	Outcome absent	Ctrl main	Ctrl addit.	Assess. outcome	Follow-up dur.	Adequacy FU	Total
Auffret et al[22] (2017)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	9
Chen et al[25] (2024)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	9
Dizon et al[31] (2015)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	9
Kooistra et al[24] (2020)	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No	8
Pavlicek et al[19] (2023)	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	8
Muntané-Carol et al[18] (2021)	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	8
Bagur et al[23] (2012)	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	8
Wasim et al[30] (2025)	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	8
Ledwoch et al[34] (2013)	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	No	7
Naveh et al[21] (2017)	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	No	7
Van Gils et al[29] (2018)	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	No	7
Castro-Mejía et al[36] (2022)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	7
Husser et al[38] (2016)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes	7
Natarajan et al[37] (2022)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	No	6
Khawaja et al[27] (2011)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	No	6
Rampat et al[26] (2017)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	No	6
Guetta et al[20] (2011)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	No	6
Fraccaro et al[33] (2011)	No	Yes	Yes	Yes	Yes	No	Yes	Yes	No	6

Rep. exposed: Representativeness of the exposed cohort; Sel. non-exp.: Selection of the non-exposed cohort; Ascert. exp.: Ascertainment of exposure; Outcome Absent: Outcome not present at study start; Ctrl main: Comparability on main confounders; Ctrl addit.: Comparability on additional confounders; Assess. outcome: Assessment of outcome; Follow-up dur.: Adequacy of follow-up length; Adequacy FU: Adequacy of follow-up of cohorts; Yes: Criterion met; No: Criterion not met.

Statistical analysis

All the data in the investigation were available as continuous and as dichotomous variables. Valve platform type (balloon-expandable, self-expanding, or mechanically expanded systems) was recorded when reported in the included studies to describe the overall distribution of device types across the pooled TAVR cohorts. The pre-test and post-test interventions were extracted for continuous outcomes. Due to a lack of paired t-test data, mean ± SD and n (%) were used independently in crossover studies. In the meta-analysis, several statistics were used to assess the heterogeneity of the included papers. The effect size varied amongst studies, as indicated by the tau square (τ²), which is an indication of the within-study variance. Degrees of freedom (df) represent the number of independent comparisons required to calculate the pooled effect size. If detected differences in effect sizes between studies were more than what would be expected by chance, it was determined by the χ² test. If the χ² value was significant, then heterogeneity was positive. The fraction of total variation that can be attributable to heterogeneity rather than random variation was assessed by the I². Since the statistical heterogeneity was assessed using the I² statistic, values above 50% were indicative of substantial heterogeneity. Elevated values suggested increased variability and disparity across research findings. Random-effects meta-analysis of pooled odds ratios (OR) was performed using restricted maximum likelihood (REML) estimation. Pooled proportions were estimated using a generalized linear mixed model (random-intercept logistic regression) on the logit scale, with the between-study variance estimated by maximum likelihood. The ‘forest plots’ were used to represent the analysis, with a null-effect line in the central axis, and the ‘diamond’ representing the combined effect of individual studies. The ‘meta’ and ‘metafor’ packages were employed to conduct the analysis via R Studio[17]. To explore sources of heterogeneity, subgroup analyses were performed stratifying the PPI incidence pool by valve platform (balloon-expandable, self-expanding, mechanically expanded, combined self-expanding and mechanically expanded, and mixed) and by study era (pre-2016 vs post-2016). For the primary OR analysis of baseline RBBB and PPI, a valve-type subgroup analysis was additionally conducted. Meta-regression of PPI rate on publication year was performed using a mixed-effects model. Sensitivity was assessed by leave-one-out analysis. Publication bias was evaluated using Egger’s test and visual inspection of funnel plot asymmetry. All title-and-abstract screening was performed independently by two reviewers (Tobias K Fuchs and Cameron Jones); discrepancies were resolved by consensus. Inter-rater agreement at the title-and-abstract screening stage was assessed using a Cohen’s kappa statistic (κ = 0.83), indicating strong inter-rater agreement.

RESULTS

Study selection

A total of 362 records were identified through database searches, including 176 from PubMed, 128 from Cochrane, and 58 from EMBASE. Before screening, 86 records were removed (43 duplicates, 14 marked ineligible by automation tools, and 29 removed for other reasons), leaving 276 records for title and abstract screening. Of these, 73 were excluded, and 203 reports were sought for full-text retrieval. Two reports could not be retrieved, resulting in 201 full-text articles assessed for eligibility. Following full-text review, studies were excluded mainly due to wrong outcomes (n = 55), inappropriate study design (n = 54), or incorrect target population (n = 72). Ultimately, 20 studies met the inclusion criteria and were included in the systematic review. A summary of the literature search and study selection is provided in the PRISMA diagram below (Figure 1).

Results of the quality assessment

Out of 20, two used a randomized control design, and 18 were prospective or retrospective observational cohort studies. Figure 2 summarizes the risk assessment results through a ‘traffic light’ and a ‘summary’ plot. The randomized studies were assessed using the Cochrane ROB2 tool[15]. In contrast, the NOS was used to assess the risk of bias across the cohort studies[16] (Table 3).

Figure 2 Risk of bias assessment. Upper panel: Traffic lights plot depicting domain-level risk of bias for the two randomized controlled trials assessed using the Cochrane ROB2 tool. Lower panel: Summary plot depicting the distribution of ROB2 domain judgements across the same two randomized controlled trials. Green: Low risk; Yellow: Some concerns.

Characteristics of the included studies

A total of 20 studies were identified and included for analysis from the provided sources. The included research utilized various study designs, including randomized clinical trials, prospective cohorts, observational studies, case-matched cohorts, and retrospective multicenter registry or cohort analyses. These studies were conducted across diverse geographical locations, including Canada and the United States, and multiple European countries such as Germany, Spain, Norway, Greece, the Netherlands, the United Kingdom, and Ireland, as well as Israel. Sample sizes ranged from the smallest cohort of 45 patients to the largest registry-based population of 3527 participants. All the studies focused on adult populations (typically mean age 79-85 years) diagnosed with severe symptomatic aortic stenosis undergoing TAVR. The main interventions and exposures studied were pre-existing conduction abnormalities on baseline ECG, including RBBB, LBBB, first-degree atrioventricular block, and bifascicular block, following the implantation of balloon-expandable, self-expanding, or mechanically expanded valve systems. The most common outcomes reported across the studies were the requirement for PPI, the incidence of new-onset conduction disturbances (such as high-degree AV block or LBBB), the timing of pacemaker implantation, the length of hospital stay, and short-term all-cause or cardiovascular mortality. Table 4 provides a summary of the patient demographics and general characteristics of the included population (Figure 3).

Figure 3 Forest plot depicting pooled incidence of new permanent pacemaker implantation following transcatheter aortic valve replacement across 20 studies (n = 15104) using an updated generalized linear mixed model analysis on the logit scale. Pooled proportion 0.18 (95%CI: 0.14-0.24); I² = 97.7%, indicating substantial heterogeneity attributable to differences in valve platform, implantation era, and pacing protocols. Each row represents one study; square size reflects study weight; diamond represents pooled estimate. PPI: Permanent pacemaker implantation.

Table 4 Patient demographics and general characteristics, mean ± SD/range/median (interquartile range).

Ref.	n	Age	Male	Female	Population characteristics	Diagnosis
Muntané-Carol et al[18] (2021)	459	79 ± 8	251	208	Minimalist TAVR; SAPIEN 3 (85.6%) or Evolut (12.6%); hypertension (91.9%)	Severe aortic stenosis
Pavlicek et al[19] (2023)	203	80 ± 6	106	97	Symptomatic; EvolutR or SAPIEN 3; diabetes (37%), CAD (59%)	Degenerative severe aortic stenosis
Meduri et al[28] (2019)	874 (baseline characteristics reported for 704 pacemaker-naive patients)	82-83 (approx)	345	359	High/extreme surgical risk; Lotus vs CoreValve (Classic/EvolutR)	Symptomatic severe aortic stenosis
Castro-Mejía et al[36] (2022)	344	83 (79-86)	135	209	New generation SEV (Evolut-R/Pro, Acurate-neo, Portico, Allegra)	Severe aortic stenosis
Auffret et al[22] (2017)	3527	82 ± 8	1764	1763	Baseline RBBB impact; 61% BEV; 39% SEV	Symptomatic severe aortic stenosis
Bagur et al[23] (2012)	822 (411/group)	81 ± 11 (TAVI); 80 ± 4 (SAVR)	357	465	Elderly; TAVI (Edwards) vs isolated SAVR; matched by baseline ECG	Symptomatic severe aortic stenosis
Wasim et al[30] (2025)	548	80.6 ± 6.7	271	277	Unselected cohort; various valves; AF (30%)	Severe aortic stenosis
Chen et al[25] (2024)	857	82-83 (approx)	501	356	Intermediate/high risk; SAPIEN 3; annulus calcification highlighted	Symptomatic severe aortic stenosis
Dizon et al[31] (2015)	2531	84.5 ± 7.2	1324	1207	Inoperable/high-risk; Edwards SAPIEN valve	Severe aortic stenosis
Natarajan et al[37] (2022)	192	81.8 ± 6.4	104	88	Outpatient TAVI; 95.3% Sapien S3/ultra; remote monitoring	Symptomatic severe aortic stenosis
Fraccaro et al[33] (2011)	64	80.97 ± 6.55	29	35	Dystrophic calcification; CoreValve Revalving System	Aortic stenosis
Guetta et al[20] (2011)	70	83 ± 4.6	26	44	High risk or inoperable; CoreValve system	Symptomatic severe aortic stenosis
Husser et al[38] (2016)	183	80-81 (approx)	97	86	High-risk; SAPIEN 3 vs SAPIEN XT	Symptomatic severe aortic stenosis
Khawaja et al[27] (2011)	243	81.3 ± 6.7	123	120	Very high/excessive surgical risk; CoreValve	Symptomatic severe aortic stenosis
Kooistra et al[24] (2020)	2804	82 (77-85)	1,248	1,556	Elderly; SEV (37%) vs BEV (56%) vs Lotus (7%)	Severe aortic stenosis
Kostopoulou et al[32] (2015)	45	81 ± 5	27	18	NYHA II/III; normal/slightly impaired LV function; CoreValve	Severe aortic stenosis
Ledwoch et al[34] (2013)	1147	81.5-82 (approx)	468	679	Elderly; high comorbidities; CoreValve or Sapien	Severe symptomatic aortic stenosis
Naveh et al[21] (2017)	110	80.7 ± 6.5	51	59	Symptomatic AS; Medtronic CoreValve or Edwards Sapien XT	Symptomatic aortic stenosis
Rampat et al[26] (2017)	201	81.2 ± 7.7	102	99	High surgical risk; LOTUS bioprosthesis	Degenerative aortic stenosis
Van Gils et al[29] (2018)	291	79 ± 8	156	135	Transarterial TAVI; CoreValve, Sapien XT/3, Lotus	Severe aortic stenosis

AF: Atrial fibrillation; AS: Aortic stenosis; BEV: Balloon-expandable valve; CAD: Coronary artery disease; ECG: Electrocardiogram; LV: Left ventricular; NYHA: New York Heart Association; PPI: Permanent pacemaker implantation; RBBB: Right bundle branch block; SAVR: Surgical aortic valve replacement; SEV: Self-expanding valve; TAVI: Transcatheter aortic valve implantation; TAVR: Transcatheter aortic valve replacement.

Across the included TAVR cohorts (n = 15104), balloon-expandable valves accounted for 8609 patients (57.0%; note that this patient-level distribution differs from the study-level valve strata used in Figure 4, in which each study is assigned to a single platform category), self-expanding valves for 5316 patients (35.2%), and mechanically expanded or other valve platforms for 1158 patients (7.7%), with valve type not specified in 21 patients (0.1%). The overall study sample of 15515 patients includes one case-matched cohort study (Bagur et al[23]) that enrolled both transcatheter and surgical AV replacement recipients; the 411-patient TAVR subgroup from this study contributes to the PPI incidence meta-analysis pool (n = 15104), while the full matched cohort contributes to the total sample reported in the abstract. Valve platform distribution was calculated only for patients who underwent TAVR; accordingly, the denominator used for valve-type reporting (n = 15104) differs from the total enrolled study sample (n = 15515).

Figure 4 Valve-type subgroup analysis of pooled permanent pacemaker implantation incidence following transcatheter aortic valve replacement. Studies were grouped into five valve-platform strata. Subgroup random-effects pooled proportions: Balloon-expandable valves 0.09 (95%CI: 0.06-0.13; I² = 90.6%; 6 studies); mixed-platform cohorts 0.20 (95%CI: 0.13-0.29; I² = 98.4%; 7 studies); self-expanding valves 0.29 (95%CI: 0.22-0.38; I² = 86.8%; 5 studies). Two strata contained a single study each and were not pooled: Combined self-expanding and mechanically expanded valves 0.28 (95%CI: 0.25-0.31) and mechanically expanded valves alone 0.32 (95%CI: 0.25-0.39). Subgroup difference across all five strata: χ² = 49.51, df = 4, P < 0.001. PPI: Permanent pacemaker implantation; BEV: Balloon-expandable valve; SEV: Self-expanding valve.

Summary of the findings

Need for new PPI: All the included studies reported on the primary outcome of PPI requirement following TAVR[18]. A total of 19/20 studies demonstrated a strong positive association between pre-existing conduction abnormalities and the subsequent need for PPI, while one study noted that baseline RBBB did not independently predict heart block, likely due to its very low prevalence in that specific cohort[19]; that study was accordingly not included in the pooled OR analysis. Pre-existing RBBB was consistently identified across the literature as the most potent predictor of procedural complications, with OR for PPI reaching 43.0 and 18.0 in specific centers[20,21]. These large effect estimates were derived from smaller single-center studies and should be interpreted with caution. Large-scale multicenter analyses revealed that patients with baseline RBBB faced PPI rates as high as 40.1%, compared to significantly lower rates in those with normal baseline conduction[22,23].

Furthermore, researchers found that the combination of baseline RBBB and increased valve implantation depth resulted in a 100% incidence of high-degree atrioventricular block[20]. Supplementary baseline factors, such as first-degree atrioventricular block and non-specific intraventricular conduction delays, demonstrated a strong collective association with both early- and late-onset pacing necessities[18,24]. A summary of reported associations between these other baseline conduction abnormalities and PPI across the included studies is provided in Table 5. Data from the PARTNER 2 registries indicate that the presence of the RBBB before the procedure significantly increased the likelihood of requiring a new device within 30 days, nearly sixfold[25]. Mechanically expanded valve systems, such as the Lotus device, were likewise associated with an elevated pacing requirement in patients with baseline conduction disease; in that cohort, a composite of pre-procedural conduction abnormalities was one of two independent predictors of the need for permanent pacing (OR = 2.54, 95%CI: 1.19-5.43, P = 0.048), the other being the absence of AV calcification[26].

Table 5 Summary of reported associations between baseline conduction abnormalities (other than right bundle branch block) and permanent pacemaker implantation after transcatheter aortic valve replacement.

Ref.	n	Conduction abnormality	PPI rate in subgroup (%)	Adjusted effect estimate	Notes
Auffret et al[22] (2017)	3527	First-degree AV block	18.5%	Not independently significant in MV model	Baseline PR > 200 ms associated with higher PPI rate; not significant on multivariate analysis
Kooistra et al[24] (2020)	2804	IVCD (non-RBBB)		OR 3.3 (95%CI: 1.7-6.3)	IVCD independently predicted late PPI (> 48 hours post-procedure)
Rampat et al[26] (2017)	201	Pre-procedural block composite (1° AVB, hemiblock, BBB)		OR 2.54 (95%CI: 1.19-5.43)	Composite pre-procedural conduction disease was an independent predictor of PPI in the LOTUS cohort, alongside absence of aortic valve calcification (OR 0.55)
Khawaja et al[27] (2011)	243	QRS duration > 120 ms (IVCD/hemiblock)		OR 2.1 (95%CI: 1.0-4.5)	Prolonged QRS independently predicted PPI in CoreValve cohort; not RBBB-specific
Naveh et al[21] (2017)	110	PR interval prolongation (> 200 ms)		OR 4.7 (95%CI: 1.2-18.3)	Baseline delta PR > 28 ms independently predicted long-term pacing dependency at 6-12 months
Van Gils et al[29] (2018)	291	Pre-existing conduction disease (non-RBBB)		Not reported	In Cohort A (normal baseline), new QRS prolongation (not pre-existing disease) drove PPI requirement
Pavlicek et al[19] (2023)	203	New-onset LBBB after TAVR		OR 15.7 (CI not reported)	New-onset LBBB was the strongest predictor of high-degree AV block requiring PPI in this cohort
Fraccaro et al[33] (2011)	64	Pre-existing RBBB + first-degree AVB combination	100%	Not modelled separately	All patients with both RBBB and implant depth > 6 mm required PPI; no patients without both factors did

Data extracted from narratively reported subgroup analyses and multivariable models within the included studies. No studies provided sufficient extractable comparative data for a formal pooled odds ratio meta-analysis of left bundle branch block, bifascicular block, or isolated first-degree atrioventricular block as primary predictors of permanent pacemaker implantation. Effect estimates are as reported in the original publications and may reflect crude or adjusted values. AVB: Atrioventricular block; BBB: Bundle branch block; IVCD: Intraventricular conduction delay; LBBB: Left bundle branch block; OR: Odds ratio; PPI: Permanent pacemaker implantation; RBBB: Right bundle branch block; TAVR: Transcatheter aortic valve replacement.

Secondary outcomes, including new-onset conduction abnormalities and the timing of device installation, were reported across the included studies[26]. New-onset LBBB was the most frequent post-procedural conduction disturbance, occurring in 55% to 57% of patients, depending on valve platform and implantation depth[26,27]. The median duration for pacemaker implantation was reported to be between 2 and 5 days post-procedure, with around 86% to 90% of all required devices installed within the initial week of the index hospitalization[25,28]. Despite many conduction abnormalities being temporary, persistent QRS prolongation predicted the need for continuous telemetric monitoring for at least 6 days to ensure safety[29]. Pre-existing abnormalities, such as RBBB, were substantially correlated with long-term pacing dependency, with roughly 68.4% of patients being reliant on their device at the 6-12 months of follow-up[21]. A notable temporal trend was also observed: One large prospective cohort documented a reduction in PPI rates from 45.8% during the low-volume phase to 23.9% during the high-volume phase (P < 0.001), suggesting that operator experience and institutional learning curves substantially influence pacing outcomes over time[30].

The findings on short-term and mid-term mortality were inconsistent across studies; yet, the majority of research concluded that the insertion of a permanent pacemaker did not independently increase the risk of mortality[30]. However, the presence of baseline RBBB was independently associated with significantly increased all-cause and cardiovascular mortality at 30 days and during midterm follow-up[22]. In contrast, long-term data from the PARTNER study indicated that new chronic pacing was independently associated with increased 1-year mortality and significantly higher rehospitalization rates[31]. Patients requiring PPI typically experienced extended hospitalizations; the median duration increased from 3 to 4 days or longer when pacing was clinically indicated[25]. Available data suggest that right ventricular pacing may modestly attenuate LVEF recovery post-TAVR, though findings were heterogeneous across the three contributing studies (see next section). Secondary clinical outcomes such as stroke rates did not differ significantly between paced and non-paced patients[28]. Overall, these studies indicate that pacemakers address the immediate electrical risks; nonetheless, pre-existing conduction disease may identify a higher-risk patient population with reduced long-term survival[22,31] (Table 2).

Meta-analysis

A meta-analysis of pooled proportions was conducted to report the relative incidence of new PPI after TAVR, as reported across the literature. Due to heterogeneity in study design and reporting, pooled incidence results were interpreted with caution.

Need for new PPI: A total of 20 studies reported quantitative data on new PPI events in patients who received TAVR. The random effects model for the pooled proportion was found to be 0.18 (95%CI: 0.14-0.24). This indicates an 18% incidence of patients requiring new PPI post-procedure. The CI was large and signified between-study heterogeneity. The incidence varied from 4% to 40% across different studies[18,20,23]. Heterogeneity analysis revealed a significant variability in the pooled results (I² = 97.7). The high heterogeneity likely reflected differences in valve generation, implantation depth, pacemaker implantation criteria, operator experience, and study era. Therefore, pooled incidence estimates should be interpreted cautiously. This difference arose mainly due to variations in patient characteristics, implantation depth, valve platform, and pacing criteria. For reference, the original pooled incidence forest plot is presented in Supplementary Figures 1, 2, 3 and 4, while an updated analysis is depicted in Figure 3. The associated funnel plot is provided in Supplementary Figure 5. Subgroup analysis by valve type demonstrated pooled PPI proportions of 9% (95%CI: 0.06-0.13) for balloon-expandable valves, 29% (95%CI: 0.22-0.38) for self-expanding valves, and 20% (95%CI: 0.13-0.29) for mixed-platform cohorts, with significant differences across subgroups (χ² = 49.51, df = 4, P < 0.001; Figure 4). Stratification by study era (pre-2016 vs post-2016) yielded pooled proportions of 22% (95%CI: 0.13-0.36) and 17% (95%CI: 0.12-0.23), respectively, without a statistically significant between-group difference (χ² = 0.79, df = 1, P = 0.374; Supplementary Figure 6). Meta-regression of PPI rate on publication year identified a negative trend that did not reach statistical significance (slope = -0.06 per year, 95%CI: -0.14 to 0.02, P = 0.118), suggesting a possible but statistically unconfirmed temporal decline in PPI rates across the study period (Supplementary Figure 7).

Incidence of new-onset bundle branch block: New-onset LBBB was analyzed across 11 studies (n = 4907). The random-effects pooled proportion of new-onset LBBB was 38% (95%CI: 0.27-0.50; I² = 98.4%, τ² = 0.70), with individual study proportions ranging from 11% (Muntané-Carol et al[18]) to 71% (Van Gils et al[29]), reflecting marked variation attributable to valve platform and implantation technique (Figure 5). New-onset RBBB could not be pooled. Among the included cohorts, only Khawaja et al[27] (8/185, 4.3%) and Guetta et al[20] (3/70, 4.3%) reported new-onset RBBB separately from new-onset LBBB; the remaining cohorts reported new-onset LBBB alone. New-onset LBBB is therefore the predominant conduction disturbance after TAVR, and patients developing new bundle branch block require post-procedural monitoring, as a subset may progress to high-degree AV block and complete heart block.

Figure 5 Forest plot depicting pooled incidence of new-onset left bundle branch block following transcatheter aortic valve replacement across 11 studies (n = 4907). Random-effects model: Pooled proportion 0.38 (95%CI: 0.27-0.50); I² = 98.4%, P < 0.0001, indicating substantial heterogeneity. Individual study proportions ranged from 0.11 (Muntané-Carol et al[18]) to 0.71 (Van Gils et al[29]), reflecting marked variation attributable to valve platform and implantation technique. LBBB: Left bundle branch block.

Predictors of PPI: Pre-existing RBBB as risk factor for PPI

Pooled analysis of four studies reporting OR for baseline RBBB and PPI yielded an OR of 4.52 (95%CI: 2.64-7.74; P < 0.001) using a random-effects REML model, with moderate heterogeneity (I² = 50.3%, τ² = 0.15) (Figure 6). Individual study ORs ranged from 2.23 (95%CI: 1.09-4.59) to 8.61 (95%CI: 3.14-23.67). These four estimates were not derived uniformly: One is an unadjusted estimate, the remaining three are adjusted for differing covariate sets, and the outcome was ascertained at hospital discharge in one study and at 30 days in the others. The pooled value should therefore be interpreted as a summary of heterogeneously derived associations rather than as a single adjusted effect. A fifth study previously included in this pool has been removed, because the OR entered for it was the multivariable estimate for any bundle branch block rather than for RBBB; that study reported a univariate RBBB OR of 0.99 (95%CI: 0.98-1.02) and identified RBBB as non-predictive in its cohort. A six-study pooled estimate reported in earlier versions of this analysis has been withdrawn in full following source verification of the underlying two-by-two counts. Leave-one-out sensitivity analysis yielded pooled ORs ranging from 3.92 to 5.76 across all omission scenarios (Supplementary Figure 2). A valve-type subgroup analysis demonstrated that the association between baseline RBBB and PPI was directionally consistent across balloon-expandable, self-expanding, and mixed platform subgroups (Supplementary Figure 3). Assessment of publication bias using Egger’s test did not identify significant asymmetry (t = 0.13, df = 2, P = 0.907), although with only four contributing studies this assessment has very limited power, and visual inspection of the funnel plot was similarly unremarkable (Supplementary Figure 1). Pre-existing RBBB represents one of the strongest and most consistently reported predictors of PPI requirement after TAVR across the included studies.

Figure 6 Forest plot depicting the association between pre-existing right bundle branch block and permanent pacemaker implantation after transcatheter aortic valve replacement across four studies (n = 2160) using random-effects restricted maximum likelihood analysis. Pooled odds ratio (OR) = 4.52 (95%CI: 2.64-7.74); I² = 50.3%, P < 0.001. Individual study ORs ranged from 2.23 (Wasim et al[30]) to 8.61 (Bagur et al[23]). Valve platforms are indicated per study. OR: Odds ratio; BEV: Balloon-expandable valve; SEV: Self-expanding valve.

Pre- and post-TAVR LVEF

Only three studies examined the comparative change in LVEF post-TAVR in 4262 patients. The random effects model produced a mean difference of -1.60 (95%CI: -4.99-1.80). The negative mean difference indicates that patients necessitating PPI exhibited marginally lower LVEF values relative to those not requiring PPI. The heterogeneity in the analysis was significant (I² = 94.3%, τ² = 8.5116), indicating a non-linear association between TAVR and LVEF. No consistent association between LVEF and pacemaker requirement was observed across the limited available studies (Supplementary Figure 8). Notably, two large registry analyses provided directly comparable LVEF data: Chen et al[25] reported a post-TAVR LVEF of 55.1% in patients who received PPI vs 58.7% in those who did not (n = 857), and Dizon et al[31] reported 55.4% vs 53.5% respectively in the PARTNER trial cohort (n = 2531)[31]. These findings were not directionally consistent: Meduri and Chen reported lower LVEF in paced patients, whereas Dizon reported a higher LVEF in the paced group (mean difference +1.90, 95%CI: 0.35 to 3.45). Any attenuation of LVEF recovery by right ventricular pacing is therefore not supported uniformly across the contributing cohorts and is reported here descriptively.

Long-term pacing dependency

Nine studies reported long-term pacing dependency data following PPI after TAVR (Table 6). Pacing dependency rates varied considerably across studies and follow-up durations, reflecting differences in patient population, valve type, and pacing indication criteria. The highest dependency rate was reported by Naveh et al[21], in which 68.4% of implanted patients remained pacing-dependent at 6-12 months of follow-up. Van Gils et al[29] reported 61% dependency (defined as > 20% ventricular pacing) at six months[29], while Meduri et al[28] documented 50% dependency at one year in the REPRISE III trial, with 83% of patients who were dependent at 30 days remaining so at one year. Conversely, Kostopoulou et al[32] reported that 60% of implanted patients had recovered endogenous rhythm by one month, with only 40% remaining dependent at 24 months, suggesting that a meaningful proportion of post-TAVR PPIs may be implanted for transient conduction disturbances. Guetta et al[20] found that 40% of implanted patients still demonstrated high-degree AV block at three months of follow-up. Wasim et al[30] reported a 38% dependency rate across follow-up visits, and Fraccaro et al[33] found that only 23.5% of implanted patients were truly pacing-dependent (> 95% pacing), with a mean ventricular pacing burden of 19% in the remainder. These findings are presented descriptively, as the significant heterogeneity in dependency definitions (pacing thresholds ranging from > 20% to > 95% ventricular pacing) and follow-up durations (3 months to 24 months) across contributing studies precluded formal pooled quantitative synthesis. The observed variation in dependency rates should therefore be interpreted in the context of these methodological differences.

Table 6 Long-term pacing dependency data across included studies.

Ref.	n (PPM)	Follow-up	Dependency rate	Definition/notes
Naveh et al[21] (2017)	38	6-12 months	68.4%	Long-term pacing dependency
Van Gils et al[29] (2018)	66	6 months	61%	> 20% ventricular pacing
Meduri et al[28] (2019)	245	1 year	50%	83% of 30-day dependent remained so at 1 year
Dizon et al[31] (2015)	173	1 year	51% paced ECGs	95% RV-paced morphology
Wasim et al[30] (2025)	173	Follow-up visits	38%	PM dependency during follow-up
Guetta et al[20] (2011)	28	3 months	40%	HDAVB persisting at 3-month check
Fraccaro et al[33] (2011)	25	6 months	23.5%	> 95% pacing; mean burden 19% in remainder
Kostopoulou et al[32] (2015)	10	24 months	40%	60% recovered endogenous rhythm by 1 month
Bagur et al[23] (2012)	30	Discharge	80% paced rhythm	Long-term dependency not formally reported

PPM: Permanent pacemaker; PM: Pacemaker; RV: Right ventricular; ECG: Electrocardiogram; HDAVB: High-degree atrioventricular block.

DISCUSSION

Interpretation of the findings

This systematic review synthesized available observational and cohort data to quantify how baseline conduction abnormalities influence PPI after TAVR. The pooled incidence estimates indicate that nearly one in five patients undergoes PPI within the peri-procedural period, and pre-existing RBBB portends markedly higher odds of PPI. The meta-analytic outcomes indicate that baseline conduction disease is both prevalent and associated with clinically relevant post-TAVR outcomes in the included observational literature. The pooled PPI incidence of 18% represents a quantifiable procedural complication burden. New-onset LBBB developed in approximately 38% of patients, demonstrating that conduction system injury is frequent and may evolve over the peri-procedural period. The pooled OR of 4.52 (95%CI: 2.64-7.74) for baseline RBBB and PPI represents a consistent observational association; however, causality cannot be inferred from these data, and residual confounding by procedural and anatomical factors cannot be excluded. Among the studies reporting a formal pacing-dependency definition, long-term dependency rates ranged from 23.5% to 68.4%; across all nine studies contributing dependency data the reported values extended to 80% (Table 6); however, the heterogeneity of dependency definitions and follow-up durations across contributing studies precludes direct comparison of these estimates. A non-significant temporal decline in PPI rates was suggested by meta-regression (slope = -0.06 per year, P = 0.118), potentially reflecting improvements in technique and valve technology over the study period, though this requires cautious interpretation.

The differential mortality associations observed across studies may be explained by the underlying temporal and mechanistic context. In-hospital and 30-day PPI does not appear to independently increase short-term mortality in the majority of registry analyses included in this review; pacing requirements in this period predominantly reflect acute transient conduction injury from mechanical compression of the conduction system during valve deployment rather than primary myocardial disease. By contrast, long-term pacing dependency, particularly high-burden right ventricular pacing, has been associated in registry data with pacing-induced ventricular dyssynchrony, attenuated LVEF recovery, and worse clinical outcomes at one year[31]. These observations are further confounded by baseline left ventricular dysfunction, comorbidity burden, and hemodynamic effects of the index TAVR procedure itself, which preclude direct causal interpretation of the mortality associations reported across the included observational studies.

Differences in valve platform likely contributed to the heterogeneity observed across studies. Self-expanding valve systems have generally been associated with higher PPI rates compared with balloon-expandable valves, as demonstrated in registry analyses where device type independently predicted pacing requirement[34]. Because our pooled analysis included mixed device types and generations, variability in reported event rates should be interpreted within this context.

Comparison with existing literature

The present review was undertaken to address several gaps identified in prior work. Hosseini Mohammadi et al[5] pooled data from 108 studies and 77538 patients using a network meta-analysis framework to rank conduction and procedural predictors of PPI, but did not perform subgroup analyses stratified by valve platform or study era, leave-one-out sensitivity analysis, or formal publication bias assessment using Egger’s test. Peng et al[2] examined perioperative risk factors for PPI without synthesizing data on long-term pacing dependency, LVEF trajectory, or mortality. The present review contributes updated REML-based random-effects analysis, valve-type and study-era subgroup analyses, meta-regression of PPI rate over time, leave-one-out sensitivity analysis, Egger’s test for publication bias, a descriptive synthesis of long-term pacing dependency across nine studies, and a structured summary of associations between non-RBBB conduction abnormalities and PPI (Table 5). A comparison of the present review with prior meta-analyses is provided in Table 7.

Table 7 Comparison of the present systematic review with selected prior meta-analyses examining predictors of permanent pacemaker implantation after transcatheter aortic valve replacement.

Ref.	Study design	Included studies/patients	Search period	OR for RBBB predicting PPI	REML random-effects analysis	Subgroup analysis by valve type	Subgroup analysis by study era	Meta-regression on publication year	Leave-one-out sensitivity analysis	Egger test for publication bias	Long-term pacing dependency synthesis	Summary of non-RBBB conduction abnormalities
Present study	Systematic review and meta-analysis	20/15515	2010-2026	4.52 (95%CI 2.64-7.74)	Yes	Yes (Figure 4; Supplementary Figure 3)	Yes (Supplementary Figure 6)	Yes (Supplementary Figure 7)	Yes (Supplementary Figure 2)	Yes (P = 0.907)	Yes (Table 6, 9 studies)	Yes (Table 5)
Mohammadi et al[5] (2025)	Systematic review and network meta-analysis	108/77538	Not reported	RR approx. 3.20	No	No	No	No	No	No	No	No
Peng et al[2] (2025)	Systematic review and meta-analysis	Not reported separately	Systematic	Not computed	No	No	No	No	No	No	No	No
Sultan et al[3] (2024)	Prospective subanalysis (Navitor IDE)	Prospective cohort	Prospective	Device-specific analysis	No	Yes	No	No	No	No	No	No

OR: Odds ratio; RR: Relative risk; PPI: Permanent pacemaker implantation; RBBB: Right bundle branch block; REML: Restricted maximum likelihood.

A recent network meta-analysis and systematic review that pooled more than 77000 patients ranked RBBB among several conduction predictors and reported a pooled relative risk for PPI after TAVR of approximately 3.20 for baseline RBBB, with multiple conduction and procedural factors contributing to PPI risk[5]. Single-center derived risk models and validation cohorts have emphasized the outsized effect of baseline RBBB; the RITMO risk score found baseline RBBB to be a key predictor and reported an OR exceeding six for high-score patients in a validation cohort, underscoring heterogeneity in effect estimates across designs and valve platforms[7]. Large prospective device-specific analyses, such as the Navitor IDE sub-analysis, illustrate that pre-existing conduction abnormalities remain important determinants of 30-day new PPI across contemporary valve platforms, while also highlighting procedural and device-related modifiers of risk[3].

The observed PPI rate of 18% and the association between baseline RBBB and PPI should be interpreted in the context of the surgical risk profile of the study populations. The included studies enrolled predominantly high-risk and intermediate-risk patients, populations in which the benefit-risk balance of TAVR has been evaluated in randomized trials. Generalizability of these conduction complication rates to lower-risk populations, in whom TAVR is increasingly performed, may be limited given differences in baseline comorbidity burden, valve morphology, and device generations used across study eras. The temporal decline in pooled PPI rates suggested by meta-regression (slope = -0.06 per year, P = 0.118, not statistically significant) and reported in individual cohort data may reflect concurrent changes in valve technology, implantation technique, and institutional experience, though these factors were not independently assessed in the available studies.

TAVR has been evaluated across the full spectrum of surgical risk in landmark randomized trials. In high-risk populations (PARTNER 1, CoreValve High Risk trial), TAVR demonstrated non-inferiority to surgical AV replacement with respect to all-cause mortality at one year, establishing the foundational evidence base for its adoption[5]. Extension to intermediate-risk patients (PARTNER 2, SURTAVI trial) and subsequently to low-risk populations (PARTNER 3, Evolut Low Risk trial) demonstrated consistent survival benefit with acceptable procedural complication rates. However, the risk-benefit balance differs across strata: Lower-risk patients face a longer expected valve lifespan and are therefore subject to cumulative risks from structural valve deterioration and valve-in-valve reintervention, whereas higher-risk patients derive greater near-term benefit from avoiding open surgery. Within this broader adverse event spectrum, which includes paravalvular leakage, stroke, vascular access complications, and acute kidney injury-PPI represents one of the most frequent device-related complications, occurring in approximately 10%-20% of patients across contemporary series. The temporal decline in PPI rates with increasing institutional volume and device iteration, as suggested by meta-regression in the present analysis, is consistent with broader observations that procedural learning curves and prosthesis design iteration reduce conduction injury rates over time.

Clinical implications

The pooled findings suggest several actionable steps for clinicians involved in TAVR care, spanning pre-procedural counseling to post-procedural monitoring and procedural planning. Discuss with patients the elevated baseline probability of needing a PPI (approximate pooled risk 18%) and explain that pre-existing RBBB substantially increases that risk, which may influence patient preferences and informed consent. Because most pacemaker implantations occurred during the index hospitalization and typically within the first week after TAVR, extended telemetry monitoring should be strongly considered in patients with baseline RBBB or evolving conduction changes. The integration of baseline ECG (specifically presence of RBBB, bifascicular block, and hemiblocks) with anatomical metrics and device choice is important. Centers may adopt validated scores (for example, RITMO or other two-step tools) to individualize risk estimates and plan follow-up[8]. Valve selection and implantation depth strategies that minimize conduction system trauma and have a low threshold for intraprocedural electrophysiology consultation or monitoring should also be considered[35-38]. Programs performing TAVR should ensure availability of pacing services and pathways for rapid PPI implantation, particularly for cohorts enriched for baseline conduction disease[9].

Strengths and limitations

This review has several important strengths. By pooling data from multiple cohorts and registries, it included a large, aggregated sample, improving the precision of estimates for PPI after TAVR and for the effect of baseline conduction abnormalities. The analysis also focused on clinically relevant predictors, particularly the increased risk associated with RBBB, providing information that may assist clinicians in risk stratification and procedural planning. In addition, the inclusion of both broad registry data and device-specific studies increases the applicability of the findings across different TAVR platforms and clinical settings. Several limitations need discussion. There was significant variability between studies. This was mainly because of differences in valve types, implantation techniques, causes for pacemaker implantation, ECG definitions, and follow-up time. Additionally, valve platform (self-expanding vs balloon-expandable systems) was examined in the pooled OR analysis only descriptively (Supplementary Figure 3), where two of the three platform strata contained a single study and the comparison was therefore not formally powered; given known differences in pacemaker rates between device types, residual confounding by valve platform may have influenced the magnitude of the observed associations. Only baseline RBBB was reported with sufficiently comparable data across studies to permit pooled OR meta-analysis; other baseline conduction abnormalities were therefore summarized narratively. Outcome definitions and timing were not uniform, as some studies reported early pacemaker implantation while others included longer-term outcomes, and implantation criteria varied across centers. There may also be publication and selection bias because studies with important findings or higher event rates are more likely to be published, and smaller or more selective groups may affect the overall results.

Sex and gender disaggregation was not consistently reported across the included studies, precluding sex-stratified analysis; this represents a recognized methodological gap that future prospective studies should address. Asian and non-Western patient cohorts were underrepresented in the included literature, limiting the generalizability of findings to non-European or non-North American TAVR populations. Pacing dependency definitions and follow-up durations varied substantially across the nine studies contributing dependency data, precluding formal pooled synthesis; these data are therefore reported descriptively. The temporal decline in PPI rates suggested by meta-regression and the single-cohort volume analysis should be interpreted cautiously, as concurrent changes in valve technology, patient selection, and implantation technique may confound this trend. The primary updated OR meta-analysis included four studies, reflecting the limited number of eligible reports with extractable comparative data for baseline RBBB and PPI. The moderate heterogeneity observed in the updated analysis (I² = 48.4%) likely reflects differences in valve platform, pacing thresholds, and patient-level confounders. As pooled estimates were derived from observational data with variable degrees of multivariable adjustment, residual confounding by implantation depth, valve oversizing, and membranous septum length cannot be excluded. A pooled estimate of new-onset RBBB incidence could not be generated. Among the included cohorts, only two reported new-onset RBBB separately from new-onset LBBB, and the remainder reported new-onset LBBB alone; the incidence of new-onset RBBB after TAVR therefore could not be quantified from the available literature and is not reported here.

CONCLUSION

Baseline RBBB demonstrated a strong and consistent association with PPI after TAVR. The pooled estimates from this review indicate an 18% PPI incidence, 38% new-onset LBBB occurrence, and a pooled OR of 4.52 (95%CI: 2.64-7.74) for the association between baseline RBBB and PPI in the updated analysis. Beyond the acute pacing decision, a substantial proportion of implanted patients develop long-term pacing dependency, ranging from 23.5% to 68.4% among studies applying a formal dependency definition (nine studies contributed dependency data), with implications for pacing-induced cardiomyopathy and long-term ventricular function. A temporal reduction in PPI rates with increasing institutional volume further suggests that operator experience is a modifiable determinant of this burden. Collectively, these observational findings are consistent with baseline RBBB serving as a risk marker that may inform preprocedural counseling, risk stratification, and postprocedural monitoring protocols for TAVR candidates, pending prospective validation. Future prospective, standardized studies with uniform dependency definitions and device-specific randomized evaluations are needed to refine absolute risk estimates, characterize long-term pacing burden, and test strategies to reduce the incidence of permanent pacing after TAVR.