Article Text

Download PDFPDF
Original article
Impact of the permanent ventricular pacing site on left ventricular function in children: a retrospective multicentre survey
  1. Irene E van Geldorp1,2,
  2. Tammo Delhaas2,3,
  3. Roman A Gebauer4,
  4. Patrick Frias5,
  5. Maren Tomaske6,
  6. Mark K Friedberg7,
  7. Svjetlana Tisma-Dupanovic8,
  8. Jan Elders9,
  9. Andreas Früh10,
  10. Fulvio Gabbarini11,
  11. Petr Kubuš12,
  12. Viera Illikova13,
  13. Sabrina Tsao14,
  14. Andreas Christian Blank15,
  15. Anita Hiippala16,
  16. Thierry Sluysmans17,
  17. Peter Karpawich18,
  18. Sally-Ann Clur19,
  19. Xavier Ganame20,
  20. Kathryn K Collins21,
  21. Gisela Dann22,
  22. Jean-Benoît Thambo23,
  23. Conceição Trigo24,
  24. Bert Nagel25,
  25. John Papagiannis26,
  26. Annette Rackowitz27,
  27. Jan Marek28,
  28. Jan-Hendrik Nürnberg29,
  29. Ward Y Vanagt2,20,30,
  30. Frits W Prinzen30,
  31. Jan Janousek12,
  32. for the Working Group for Cardiac Dysrhythmias and Electrophysiology of the Association for European Paediatric Cardiology
  1. 1Department of Cardiology, Cardiovascular Research Institute Maastricht, Maastricht University Medical Center, Maastricht, The Netherlands
  2. 2Department of Pediatrics, Maastricht University Medical Center, Maastricht, The Netherlands
  3. 3Department of Biomedical Engineering, Cardiovascular Research Institute Maastricht, Maastricht University, Maastricht, The Netherlands
  4. 4Department of Pediatric Cardiology, Heart Center, University of Leipzig, Leipzig, Germany
  5. 5Sibley Heart Center, Children's Healthcare of Atlanta, Emory University School of Medicine, Atlanta, USA
  6. 6Paediatric Cardiology, University Children's Hospital, Zürich, Switzerland
  7. 7Devision of Cardiology, Hospital for Sick Children, Toronto, Canada
  8. 8Department of Cardiology, Children's Mercy Hospital, Kansas City, USA
  9. 9Department of Pediatric Cardiology, Universitair Medisch Centrum St. Radboud, Nijmegen, The Netherlands
  10. 10Department of Pediatric Cardiology, Oslo University Hospital, Oslo, Norway
  11. 11Department of Pediatric Cardiology, Ospedale Infantile Regina Margherita, Turin, Italy
  12. 12Cardiocentrum and Cardiovascular Research Center, University Hospital Motol, Prague, Czech Republic
  13. 13Department of Pediatric Cardiology, Children's Cardiac Center, Bratislava, Slovakia
  14. 14Department of Electrophysiology, Children's Memorial Hospital, Chicago, USA
  15. 15Department of Pediatric Cardiology, Wilhelmina Kinderziekenhuis, University Medical Center, Utrecht, The Netherlands
  16. 16Department of Pediatric Cardiology, Hospital for Children and Adolescents, Helsinki University Hospital, Helsinki, Finland
  17. 17Department of Pediatric and Congenital Cardiology, Clinique Universitaire Saint-Luc, Brussels, Belgium
  18. 18Department of Cardiac Electrophysiology, Children's Hospital of Michigan, Wayne State University School of Medicine, Detroit, USA
  19. 19Department of Pediatric Cardiology, Emma Kinderziekenhuis, Academisch Medisch Centrum, Amsterdam, The Netherlands
  20. 20Department of Pediatric Cardiology, University Hospital Gasthuisberg, Leuven, Belgium
  21. 21Department of Pediatric Cardiology, Children's Hospital University of Colorado, Denver, USA
  22. 22Department of Pediatric Cardiology, University Hospital Göttingen, Göttingen, Germany
  23. 23Department of Congenital Heart Disease, Hôpital Cardiologique du Haut-Lévéque, Bordeaux University, Bordeaux-Pessac, France
  24. 24Department of Pediatric Cardiology, Santa Marta Hospital, Lisbon, Portugal
  25. 25Department of Pediatric Cardiology, University Children's Hospital, Graz, Austria
  26. 26Department of Pediatric Cardiology, Mitera Children's Hospital, Maroussi, Greece
  27. 27Department of Pediatric Cardiology, Sophia Kinderziekenhuis, Erasmus University Medical Center, Rotterdam, The Netherlands
  28. 28Department of Cardiothoracics, Great Ormond Street Hospital, London, United Kingdom
  29. 29Department of Congenital Heart Disease and Pediatric Cardiology, Klinikum Links der Weser, Bremen, Germany
  30. 30Department of Physiology, Cardiovascular Research Institute Maastricht, Maastricht University, Maastricht, The Netherlands
  1. Correspondence to Irene E van Geldorp, Cardiovascular Research Institute Maastricht, Department of Cardiology, Maastricht University Medical Center, PO Box 616, Maastricht NL-6200 MD, The Netherlands; i.vangeldorp{at}


Background Chronic right ventricular (RV) pacing is associated with deleterious effects on cardiac function.

Objective In an observational multicentre study in children with isolated atrioventricular (AV) block receiving chronic ventricular pacing, the importance of the ventricular pacing site on left ventricular (LV) function was investigated.

Methods Demographics, maternal autoantibody status and echocardiographic measurements on LV end-diastolic and end-systolic dimensions and volumes at age <18 years were retrospectively collected from patients undergoing chronic ventricular pacing (>1 year) for isolated AV block. LV fractional shortening (LVFS) and, if possible LV ejection fraction (LVEF) were calculated. Linear regression analyses were adjusted for patient characteristics.

Results From 27 centres, 297 children were included, in whom pacing was applied at the RV epicardium (RVepi, n=147), RV endocardium (RVendo, n=113) or LV epicardium (LVepi, n=37). LVFS was significantly affected by pacing site (p=0.001), and not by maternal autoantibody status (p=0.266). LVFS in LVepi (39±5%) was significantly higher than in RVendo (33±7%, p<0.001) and RVepi (35±8%, p=0.001; no significant difference between RV-paced groups, p=0.275). Subnormal LVFS (LVFS<28%) was seen in 16/113 (14%) RVendo-paced and 21/147 (14%) RVepi-paced children, while LVFS was normal (LVFS≥28%) in all LVepi-paced children (p=0.049). These results are supported by the findings for LVEF (n=122): LVEF was <50% in 17/69 (25%) RVendo- and in 10/35 (29%) RVepi-paced patients, while LVEF was ≥50% in 17/18 (94%) LVepi-paced patients.

Conclusion In children with isolated AV block, permanent ventricular pacing site is an important determinant of LV function, with LVFS being significantly higher with LV pacing than with RV pacing.

  • Pediatrics
  • pacing
  • pacing site
  • LV function
  • atrioventricular block
  • pacemakers
  • congenital heart disease
  • paediatric cardiology
  • cardiac function
  • imaging and diagnostics
  • paediatric electrophysiology
  • paediatric arrhythmias
  • Fallot's tetralogy
  • echocardiography-fetal
  • transposition of the great arteries
  • echocardiography-paediatrics
  • paediatric echocardiology
  • QT interval
  • paediatric interventional cardiology
  • interventional cardiology
  • radiofrequency ablation (RFA)
  • myocardial disease
  • haemodynamics
  • cardiac resynchronisation therapy

Statistics from


In patients with bradycardia due to complete atrioventricular (AV) block, ventricular pacing is required to normalise heart rate. The pacing-induced activation pattern is characterised by a prolonged total activation duration and an abnormal sequence of activation (in both longitudinal and transverse directions). This abnormal electrical activation pattern may lead to dyssynchronous ventricular contraction, the degree of dyssynchrony varying with the site of pacing.1 Ventricular pacemaker electrodes are conventionally positioned either at the right ventricular (RV) endocardium or at the RV epicardium. However, RV pacing results in a left bundle branch block morphology and is associated with cardiac dysfunction and remodelling.2–7 The preservation of cardiac function during chronic ventricular pacing should be a high priority, especially in paediatric patients who are usually paced from an early age and may expect lifelong pacing. The main objective of this multicentre study was to investigate whether left ventricular (LV) pacing sites, in comparison with RV endocardial and RV epicardial pacing sites, have fewer adverse long-term functional and structural effects, and may prevent pacing-induced LV dysfunction in children with isolated AV block.

Materials and methods

Study population

From the institutional databases of the participating 27 centres, patients with a structurally normal heart and isolated advanced second-degree or complete AV block with chronic and permanent ventricular pacing for rate control (minimum of 1 year follow-up, minimum of 70% ventricular paced beats) were identified. All these patients were considered for inclusion. Exclusion criteria were postsurgical AV block, structural congenital heart disease and evident cardiomyopathy due to causes other than AV block. Study end points were reached whenever patients reached 18 years of age, underwent a change in pacing site, received cardiac resynchronisation therapy or cardiac transplantation for heart failure, or died. For those cases, data from the last echocardiography before the event were used.


Demographic data and clinical characteristics (gender, age, body surface area, aetiology of AV block, maternal autoantibody status, year of pacemaker implantation, duration of pacing) were collected. Parameters from the last echocardiography performed at a routine follow-up visit at age <18 years were reviewed. In all patients, LV end-diastolic diameter (LVEDD) and end-systolic diameter (LVESD) were assessed. The degree of LV dilatation was evaluated by adjusting LVEDD for body size, expressed as a z-score.8 As a measure of LV function, LV fractional shortening (LVFS) was calculated (LVFS=(LVEDD−LVESD)/LVEDD×100%). According to generally accepted criteria,8 9 we classified LVFS as ‘normal’ (LVFS ≥28%), ‘subnormal’ (LVFS <28%) or ‘depressed’ (LVFS <25%). For the subset of patients in whom end-diastolic and end-systolic LV volumes (LVEDV and LVESV, respectively) were assessed, LV ejection fraction (LVEF) was calculated (LVEF=(LVEDV−LVESV)/LVEDV*100%). Mitral regurgitation was scored on a scale from 0 (= no regurgitation) to 4 (= severe regurgitation).

Based on the location of the tip of the ventricular pacing electrode (the site of pacing), the cohort was divided into three groups: RV epicardium (RVepi), RV endocardium (RVendo), or LV epicardium (LVepi).

Statistical analysis

Comparisons between groups were performed with analysis of variance or χ2 tests, as appropriate. Linear regression analyses were used to examine whether LVFS and LVEDD z-scores differed between the groups (ie, between pacing sites). These analyses were adjusted for the following covariates: maternal autoantibody status (‘positive’, ‘negative’, or ‘unknown’), year of pacemaker implantation, age at implantation, participating centre and duration of pacing and body surface area at echocardiographic follow-up. The influence of pacing mode (VVI vs DDD) on LVFS was investigated only in the subset of patients for whom pacing mode at follow-up was reported.

Additionally, to carefully investigate the potential influence of maternal autoantibody status on LV function, linear regression analyses were performed on the study population grouped into ‘autoantibody positive’, ‘autoantibody negative’ and ‘unknown autoantibody status’. In these analyses, ‘pacing site’ was included as a covariate in addition to above-mentioned characteristics.

Group characteristics are expressed either as mean±SD, or as proportion (%). Mean differences in LVFS and LVEDD adjusted for covariates are expressed as effect sizes (β) with 95% CIs, p<0.05 = significant.


Study population: demographic data and clinical characteristics

A total of 297 children, from the institutional databases of the 27 participating centres, were included in the study. Pacing site distribution was: RVendo (n=113), RVepi (n=147), and LVepi (n=37). Maternal autoantibody status (anti-SSA (Ro), anti-SSB (La)) was reported for 201 (68%) of the patients. Maternal autoantibodies were present in 88 (44%) of these. Patient characteristics summarised for each pacing site are listed in table 1.With the above-mentioned numbers of patients in each group, it would be possible to detect differences in LVFS between the pacing site groups with a power >90% (μ=5% points; SD 8% points; α=0.05; unequal groups).

Table 1

Patient characteristics

Effect of pacing site on left ventricular fractional shortening and dilatation score

At routine follow-up, LVFS was significantly higher in children with LVepi pacing (39±5%) than in children with RVendo pacing (33±7%) and RVepi pacing (35±8%) (figure 1). Pacing site was the solitary significant determinant (p=0.001) of LVFS (maternal autoantibody status, p=0.266; duration of pacing, p=0.833; body surface area at follow-up, p=0.882; centre, p=0.560; year of implantation, p=0.182; and age at implantation, p=0.809). The mean LVFS differences adjusted for covariates (ie, ‘effect size β’) are reported in table 2.

Figure 1

Left ventricular fractional shortening after chronic ventricular pacing. Left ventricular fractional shortening (LVFS), a measure of LV function, was significantly (*) higher in children being paced at the LV epicardium (LVepi) than in children being paced at the RV epicardium (RVepi) or RV endocardium (RVendo). Box plots represent 25th, 50th and 75th centiles. Whiskers represent the minimal and maximal values within the range of (25th centile − 1.5×IQR) and (75th centile + 1.5×IQR), respectively. Dots display mean values.

Table 2

Mean differences in left ventricular fractional shortening and dilatation score

In the subgroup of patients for whom the pacing mode (either VVI or DDD) at follow-up was reported (n=242), pacing mode was not a significant determinant (p=0.209) while pacing site remained a significant determinant of LV function (p=0.002). Differences between pacing sites, adjusted for pacing mode in addition to the other covariates, were similar to the effect sizes reported in table 2.

LVEDD z-score was normal in all groups (RVendo 0.0±1.3; RVepi 0.4±1.1; LVepi 0.3±0.9, figure 2), and was not significantly influenced by pacing site (p=0.640), or by maternal autoantibody status (p=0.724) or any of the other covariates. The adjusted means were not significantly different as presented in table 2.

Figure 2

Left ventricular dilatation-score after chronic ventricular pacing. The z-score for left ventricular end-diastolic diameter (LVEDD z-score), a measure of LV dilatation, was not significantly influenced by the site of pacing. The LVEDD z-score was normal for all groups and differences between the pacing groups were not significant. Box plots represent 25th, 50th and 75th centiles. Whiskers represent the minimal and maximal values within the range of (25th centile − 1.5×IQR) and (75th centile + 1.5×IQR), respectively. Dots display mean values. LVepi, LV epicardium; RVendo, RV endocardium; RVepi, RV epicardium.

Depressed LV function in ∼10% of the chronically RV-paced patients

LVFS was subnormal (LVFS <28%) in 16 RVendo- (14%) and 21 RVepi-paced patients (14%). In more detail, LVFS was depressed (LVFS <25%) in 10 (9%) of the RVendo-paced and in 17 (12%) of the RVepi-paced patients. In contrast, LVFS was normal (LVFS ≥28%) in all patients in whom LV pacing was applied (p=0.049, χ2 test).

In the subset of patients for whom we could calculate LVEF (n=122), LVEF was <50% in 17/69 (25%) RVendo-paced and in 10/35 (29%) RVepi-paced patients, whereas this was only the case in one of the 18 (6%) LVepi-paced patients.

Mitral regurgitation

Distribution of mitral regurgitation scores was significantly different between the pacing site groups (p=0.032, χ2 test). Mitral regurgitation was mild (score 1) in 42 (29%) RVepi-paced and in 19 (17%) RVendo-paced patients, but only in three (8%) LVepi-paced patients. Score ‘2’ was reported for four patients (RVepi n=1; RVendo n=2, LVepi n=1), while none of the patients had moderate to severe mitral regurgitation (scores 3 and 4).

Influence of maternal autoantibody status

As shown by the analyses grouped by pacing site, LVFS was significantly affected by pacing site (p=0.001), while maternal autoantibody status did not affect either LVFS, or LVEDD z-score (p=0.266 and p=0.724, respectively). These results were confirmed when analyses were performed on the same population grouped by maternal autoantibody status. For the three groups, mean LVFS and LVEDD z-scores were comparable: 34±8% and 0.4±1.3 for ‘positive’, 35±7% and 0.1±0.9 for ‘negative’ and 34±8% and 0.1±1.3 for ‘unknown’ maternal autoantibody status (LVFS and LVEDD z-score, respectively). Maternal autoantibody status was not a significant determinant of either LVFS (p=0.386) or LVEDD z-score (p=0.901), while pacing site remained a significant determinant of LVFS (p=0.013).


This multicentre study retrospectively surveys LV function and dimensions in 297 children with structurally normal hearts and chronic ventricular pacing for isolated AV block. In these patients, requiring lifelong chronic ventricular pacing, preservation of cardiac functional and structural integrity is a major challenge. This study indicates that pacing site significantly influences LVFS, with better LVFS in LVepi-paced patients than in RVepi-paced or RVendo-paced patients.

Pacing site influences left ventricular function

In this survey, LV function was found to be subnormal (LVFS <28%) in 14%, and depressed (LVFS <25%) in ∼10% of the RVepi- and endo-paced patients. Our findings on LVFS are supported by those in the subset of patients for whom we were able to calculate LVEF: the proportion of patients with subnormal LVEF (LVEF <50%) was for each pacing group comparable to the proportion of patients with subnormal LVFS. Several other studies report that, within less than a decade of pacing, 7–10% of chronically RV-paced patients develop heart failure and that up to 13% have depressed LV function combined with LV dilatation.2 10 11 Chronic RV pacing, rather than the aetiology of AV block, has been identified as an independent risk factor for development of LV dilatation and dysfunction.11 12 This study indicates that it is predominantly the pacing site that affects LV function, as reflected by (1) a higher LVFS in the LV-paced group than in the RV-paced groups; (2) pacing site being the only significant factor influencing LVFS and (3) dissimilar incidence of patients with LVFS<28% between the pacing site groups. Moreover, maternal autoantibody status did not significantly influence LV function or LV dilatation score.

Though it appears that there are relevant proportions of patients who do not tolerate RV pacing, the reasons are not elucidated in this retrospective study. Chronic LV pacing, however, seems to be well tolerated by all patients, as suggested by absence of LV dysfunction in the LV-paced group. From the latter finding we hypothesise that pacing-induced LV failure in children with structurally normal hearts might be prevented by LV pacing.

Potential reasons for preservation of LV function by LV pacing

We postulate that, above and beyond synchrony (reflected by the total duration of activation), the sequence of activation is a major determinant of cardiac pump function.1 13 In LV pacing the pattern of activation and mechanical contraction pattern may be more favourable than patterns induced by RV pacing. During LV (free wall) pacing, the total duration of activation is prolonged similarly to that during RV free wall pacing, reflected by a similar QRS duration.14 However, in contrast to RV pacing, LV pacing activates the LV lateral wall before the septum and RV lateral wall, preventing paradoxical septal movement and resulting in better haemodynamic performance than with RV pacing.1 15 Furthermore, LV apical pacing induces physiological apex-to-base activation, which results in synchronous electrical activation and contraction around the circumference of the left ventricle.16 17 These remarks are supported by the observation of Gebauer et al, that LV apical pacing preserves septal-to-lateral LV mechanical synchrony as well as systolic function.18 Also, experimental studies have shown that LV apical pacing is better than pacing at other sites and that it maintains cardiac function at a normal level.19 20

Single-site left ventricular pacing versus biventricular pacing

Biventricular pacing is often used to resynchronise electrical activation in patients with either intrinsic or pacing-induced dyssynchrony and LV dysfunction. Biventricular pacing (following chronic RV pacing) improves pump function and reverses ventricular remodelling in children at least as effectively as in adults with heart failure.21–23 Although single-site LV pacing and biventricular pacing have not yet been compared in children, single-site LV pacing in adults with heart failure results in the same improvement in LV function as acute or chronic biventricular pacing.24–26 Also, animal experiments have indicated that single-site LV apical and LV septal pacing maintain cardiac function and efficiency at least as well as biventricular pacing.19 The use of a single optimal ventricular pacing site provides advantages over biventricular pacing, such as lower pacemaker battery usage and a reduction in the number of surgical access routes required and consequent scar tissue formation.

Clinical implications of this study

Since paediatric patients with AV block are usually paced from an early age and require lifelong pacing, preservation of cardiac function during chronic ventricular pacing is important. This study indicates that LV pacing may be better than RV pacing if LV function is concerned. In the individual patient, depressed LV function (as seen in ∼10% of the RV-paced patients) may indicate that chronic pacing is not well tolerated and that there may be a higher risk for pacing-induced heart failure. The median follow-up of pacing in this study, as in other studies, was less than a decade, which is a mere fraction of the lifelong follow-up expectancy of a child receiving ventricular pacing for complete AV block. The (very) long-term outcome of either RV or LV pacing beginning in childhood is still unknown. Considering the findings of this and other studies,14 27 28 we suggest the use of a single LV apex or LV free wall site for chronic ventricular pacing in children with AV block and structurally normal hearts. Unfortunately, as each ventricle brings a continuum of possible pacing sites, and accuracy of (retrospective) determination of the precise site is limited, this survey has not provided exact data to reliably test which spot on the left ventricle would be the best, or whether certain sites within the right ventricle might be better than others.7

It is important not to simply extrapolate the results of this study to children with structural congenital heart disease. Nevertheless, results are likely to be applicable to patients with a systemic left ventricle, without intrinsic RV activation delay.

Practical considerations

Surgical access for LV pacing via a left lateral thoracotomy is minimal, easy and safe, though invasive.29 In small children, the LV apex can easily be reached by a sub-xiphoidal approach, thereby avoiding a lateral thoracotomy, and with reasonable cosmetic results. Acknowledging the potential surgical complications, we particularly suggest implantation of electrodes at the LV epicardium if there are also other indications for a surgical approach.30 In larger children, single-site epicardial LV pacing may also be achieved by a transvenous approach via the coronary sinus. In the near future, endocardial pacing in the systemic ventricle may become feasible through the application of ‘wireless pacing’. However, in practice, the routine transvenous approach seems justifiable in young adults, as RV apical pacing is well tolerated by most patients.10 31

Regular follow-up with echocardiography to detect LV deterioration at an early stage is warranted in all paced patients receiving chronic ventricular pacing. Changing to either biventricular or single-site pacing at the systemic ventricle should be considered when echocardiography discloses signs of progressive ventricular dysfunction.21 22 32–34

Study limitations

The retrospective design of the study is a disadvantage, mainly because it hampers the use of sophisticated echocardiographic parameters. The shortening fraction is a limited marker for systolic function, and it may be affected by intraventricular asynchrony. Nevertheless, it was chosen as a main outcome parameter, as it is the most consistently measured variable in this population. Data on LVFS in the entire study population were supported by data on LVEF in those patients for whom LV volumes were available. The possibility of unintended bias for the effect of pacing from the various sites cannot totally be excluded since measurements were not performed blinded to the pacing site. However, at the time of the echocardiographic examination, there was no intention to compare the effects of different pacing sites. Each centre identified patients eligible for the study by systematically reviewing the institutional database. The structure of the database affected the time span of inclusion, and therefore also the number of patients from each centre. The number of LV-paced patients included in this study is relatively small in respect to the number of patients in the RV-paced groups, since LV pacing (for first implantation) is only used at a minority of centres. However, patients from several different centres are included for each pacing group (nine centres for the LV-paced group). Besides, ‘participating centre’ was entered as a covariate in the analysis to correct for potential bias arising from centre-based differences.

Future studies with larger numbers of LV-paced patients, more sophisticated echocardiographic indices and longitudinal follow-up are needed to confirm the conclusions of this survey, and to explore the effect of pacing site on parameters other than LV function.


In children with normal cardiac anatomy and AV block, the site of pacing is an important determinant of LV function, with LVFS being significantly higher in children with chronic LV pacing than in children with chronic RV pacing. LVFS was subnormal (LVFS <28%) in 14% of the RV-paced children, whereas LVFS was normal in all LV-paced children.



  • Funding JJ was supported by the research project of University Hospital Motol MZOFNM2005.

  • Competing interests None.

  • Ethics approval Maastricht University Medical Center.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.