Article Text


Original research article
Surgical and percutaneous pulmonary valve replacement in England over the past two decades
  1. Signe H Larsen1,2,
  2. Konstantinos Dimopoulos1,3,
  3. Michael A Gatzoulis1,3,
  4. Anselm Uebing1,3,
  5. Darryl F Shore1,3,
  6. Rafael Alonso-Gonzalez1,3,
  7. Aleksander Kempny1,3
  1. 1 Adult Congenital Heart Centre, Royal Brompton Hospital, London, UK
  2. 2 Department of Cardiology, Aarhus University Hospital, Aarhus, Denmark
  3. 3 National Heart and Lung Institute, Imperial College, London, UK
  1. Correspondence to Dr Signe H Larsen, Adult Congenital Heart Centre, Royal Brompton Hospital, London SW3 6NP, UK; signe_holm_larsen{at}


Objective Pulmonary valve replacement (PVR) is often required in patients with congenital heart disease. We aimed to describe temporal trends in PVR in a nationwide English cohort between 1997 and 2014, survival and the need for re-PVR.

Methods Patients were identified in the Hospital Episode Statistics Database. Survival data were retrieved from the UK Office for National Statistics.

Results A total of 2733 patients underwent PVR (2845 procedures) over the study period. Median age at first procedure increased from 20.1 years in 1997–2005 to 24.7 years in 2006–2014. The annual number of PVRs increased from 23 in 1997 to 251 in 2014. Homografts were the most common choice in the early years, but the use of xenografts increased after 2005. During a median follow-up of 5.8 years, 176 patients died and 108 required redo PVR. Early (30 day) survival was 98% for all PVRs and was similar for all types of prostheses but longer-term mortality dropped to 92% at 10 years and 90% at 15 years. Age >16 years and percutaneous PVR were risk factors for death. The cumulative incidence for re-PVR at 10 years was 8% for all PVRs and 11% at 15 years. Risk factors for re-PVR were complex diagnosis, male gender and black ethnicity.

Conclusion There was a significant increase in the number of PVRs performed in England over the last two decades and a significant change in the type of prosthesis employed. While early mortality was low across the board, longer-term mortality was not negligible in this young population.

  • congenital heart disease surgery
  • congenital heart disease
  • pulmonic valve disease
  • epidemiology

Statistics from


Pulmonary valve dysfunction is common in many patients with congenital heart disease, including repaired tetralogy of Fallot, pulmonary valve disease or pulmonary atresia. Surgical pulmonary valve replacement (PVR) can be part of the primary repair or as subsequent intervention. It is well known that pulmonary valve dysfunction over time can lead to decreased exercise tolerance, arrhythmia, right ventricular failure or sudden death if left untreated.1 The optimal pulmonary valve prosthesis remains to be clarified. The majority of patients are treated with homografts or xenografts,2 3 with an ongoing debate on which is the best choice when considering haemodynamics and longevity.4 5 Some authors have defended the use of mechanical valves due to their long-term durability6 7 and recently percutaneous PVR has been introduced for anatomically suitable patients.8 9 In recent years, there has been an increase in PVR procedures performed worldwide, including in the UK.10

The aim of our study was to examine temporal trends in PVR over the past two decades in a nationwide patient cohort with congenital heart disease, with a particular focus on the type of prosthesis, survival and the need for re-intervention. Moreover, we examined the risk factors associated with mortality and re-intervention.


Patient population and data collection

The Hospital Episode Statistics Database11 is an administrative database with details of all admissions, outpatient appointments and accident and emergencies attendances at National Health Service hospitals in England from 1997 and onwards. The English healthcare system is a free tax-based system with the National Health Service hospitals covering the vast majority of public hospitals in England. We identified all patients undergoing PVR in England between 1997 and 2014 from the database, as well as information on age, gender, ethnicity, diagnosis, type of procedure and the need for re-intervention. There was no information on admissions or procedures performed prior to 1997. Data on survival were retrieved from the UK Office for National Statistics.

Every patient was assigned a primary diagnosis according to the International Classification of Diseases Revision 10. When a patient had more than one diagnosis, the most severe diagnosis was chosen. The complexity of the diagnosis was grouped into simple, moderate or severe, based on the Bethesda classification.12 Since patients with aortic valve disease were assumed to undergo a Ross operation, their diagnosis was coded as of moderate severity. Type of PVR was allocated into xenograft valves, homograft valves, mechanical valves and percutaneous valves, based on the UK surgical codes. Patients with a non-specified type of conduit or valve were excluded (n=1021) since it was not possible to determine if the valves were homografts or xenografts. The study period was divided into two eras (before and after 2006) based on the introduction of the percutaneous valve into the dataset.

Statistical analyses

Continuous variables were presented as mean ±SD or median (IQR), as appropriate. Student t-test or Kruskal-Wallis test was used to examine differences between groups, as appropriate. Categorical variables were presented as the frequency with percentages. Chi-square tests or Fisher’s exact tests were used to examine differences between groups, as appropriate. Kaplan-Meier curves with 95% CIs were used to describe survival. Time to first documented reintervention was described as cumulative incidence proportions, accounting for death as a competing event. One of the 2733 patients was excluded from survival analysis as the date of the procedure was listed as being after the follow-up date. Cox regression analysis was used to examine risk factors for death. Risk estimates were reported with HRs. Harrell’s C-statistics were calculated to provide information on the prediction accuracy of the Cox regression models. Fine and Grey competing risk regression analysis was used to identify potential risk factors for the first reintervention, with all-cause mortality as a competing event. Risk estimates were reported as sub-distribution HRs (SHR). All demographic variables (table 1) were considered for inclusion in the regression models. For multivariate models, stepwise inclusion and exclusion were used. Variables were included in the final model if the p value was ≤0.1. The assumption of proportional hazards was tested by log-log plots. Overall, a p value of ≤0.05 was considered statistically significant. Stata SE V.14 was used for the analyses.

Table 1

Patient demographics

Ethics and approvals

The project was approved by the Health and Social Care Information Centre of the National Health Service. As only pseudoanonymised data were provided and analysed, individual informed consent and ethical approval were not required (UK National Research Ethics Service guidance).

The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive licence (or non-exclusive for government employees) on a worldwide basis to the BMJ Publishing Group Ltd and its Licensees to permit this article (if accepted) to be published in HEART editions and any other BMJPGL products to exploit all subsidiary rights.


Patients and trends

In total, 2733 patients (59% male) underwent 2845 PVR procedures with 638 procedures (22%) being performed before 2006 and 2207 (78%) performed after. The majority 2625 (96%) underwent a single procedure, 104 patients (3.8%) underwent two and only 4 patients (0.2%) underwent three procedures. The most common diagnosis was tetralogy of Fallot, followed by pulmonary stenosis or regurgitation (table 1).

The median age at first documented operation was 23.9 years (IQR 16.2–35.9). A similar number of patients had a xenograft valve (n=1147, 40%) compared with a homograft valve (n=1052, 37%), while only 343 (12%) had a mechanical valve (table 1).

The diagnostic profile of the patients changed significantly over the years (figure 1). In the later era, the majority of the patients had tetralogy of Fallot (n=1042, 50%), whereas in the first era, pulmonary and aortic valve disease were more common diagnoses (n=103, 16% and n=187, 30%, respectively) (online supplementary table 1). The age of the population changed as well. In the first period, approximately one-third of the population (n=244, 38%) was children under 16 years of age, while in the second period only 20% (n=410) was children. The median age at first documented PVR increased from 20.1 years (IQR 11.8–33.1) in the first period to 24.7 years (IQR 17.5–36.6) in the second period (p<0.001).

Figure 1

Diagnoses of patients having a pulmonary valve replacement in England in the years 1997–2014 in absolute numbers (A) and in percent (B). AS/AR, aortic stenosis/regurgitation; ccTGA, congenital corrected transposition of the great arteries; DORV, double outlet right ventricle; Non-spec, non-specified; PA, pulmonary atresia; PS/PR, pulmonary stenosis/regurgitation; TGA, transposition of the great arteries; TOF, tetralogy of Fallot; Truncus, truncus arteriosus.

The number of PVRs performed increased from 23 per year in 1997 to 251 in 2014 (p<0.001) (figure 2A). While in the early era, homograft was the valve of choice (n=465, 73%), these decreased in the second era in favour of the xenografts (n=1065, 48%, p<0.001) (figure 2B and online supplementary table 1). The percutaneous valves were introduced in the latter era, but only comprised a small proportion (n=303, 14%) of all PVRs.

Figure 2

Pulmonary valve replacements in England in the years 1997–2014 in absolute numbers (A) and in per cent (B) according to prosthesis type.


During a median follow-up of 5.8 years (IQR 2.9–9.0 years), allowing for 17 816 patient-years, 176 patients died. Early postintervention mortality was low, with 30-day survival at 98% (97%–98%) for all PVRs, 98% (98%–99%) for xenografts, 98% (97%–99%) for homografts, 98% (96%–99%) for mechanical valves and 99% (96%–100%) for the percutaneous valves (figure 3). Survival at 10 years was 92% (91%–93%) for all PVRs, 91% (88%–94%) for xenografts, 94% (92%–95%) for homografts and 90% (84%–93%) for mechanical valves. For percutaneous valves, the cumulative survival was 87% (79%–92%) after 9.2 years. Survival at 15 years was 90% (88%–92%) for all PVRs, 87% (80%–92%) for xenografts, 92% (89%–94%) for homografts and 87% (79%–92%) for mechanical valves.

Figure 3

Survival according to type of prosthesis. Follow-up in years are shown at the x-axis, survival at the y-axis. Be aware that the y-axis starts at 0.5. PVR, pulmonary valve replacement.

The most common cause of death was cardiac (n=106, 60%), followed by cancer (n=35, 20%) (table 2).

Table 2


The median age at death was 47.4 years (IQR 28.2–63.4 years). Despite the increase in age at first operation over time, there was no significant difference in the median age at death between the two eras (45.6 years (IQR 26.2–61.6 years) vs 48.9 years (IQR 29.4–65.1 years), p=0.28).

For all regression analyses, the assumptions of proportional hazards were met. Factors associated with increased risk of death in the multivariate analysis (table 3) were age above 16 years (HR 1.63 (95% CI 1.02 to 2.59), p=0.002) and use of a percutaneous valve (HR 2.37 (95% CI 1.26 to 4.48), p=0.008).

Table 3

Risk factors associated with mortality


During the study period, 108 patients (4%) underwent redo PVR. The cumulative incidence of redo PVR at 5 years was 5.3% (4.4%–6.2%) for all patients, 5.1% for the xenograft (4.3%–6.0%) and homograft valves (4.3%–6.1%), 5.3% (4.4%–6.2%) for the mechanical vales and 5.9% (5.0%–6.9%) for the percutaneous valves. At 10 years, the overall cumulative incidence was 7.8% (6.6%–9.1%), 7.5% (6.4%–8.7%) for xenograft valves, 7.6% (6.5%–8.9%) for the homograft valves and 7.8% (6.6%–9.1%) for the mechanical valves. Among percutaneous valves, 7.2% (6.1%–8.4%) have had a re-PVR after 9.2 years of follow-up. Over the 15 years of follow-up, the cumulative incidence of redo PVR was 11.3% (8.7%–14.2%) for all patients, 9.5% (7.5%–11.8%) for the xenograft valves, 11.3% (8.7%–14.2%) for the homograft valves and 10.8% (8.4%–13.5%) for mechanical valves.

The most common prosthesis for the second documented PVR was a percutaneous valve (n=52, 48%). A similar number of patients received either a homograft (n=24, 22%) or a xenograft (n=19, 18%), while only 12% (n=13) received a mechanical valve. For the few patients having a third PVR, the most common prosthesis was a percutaneous valve (n=2, 50%). Only two patients received a homograft or a xenograft as their third PVR. The type of prosthesis chosen for the second or third PVR did not change over time (p=0.3 and p=1.0, respectively).

Factors associated with redo PVR within the study period were male gender (SHR 1.91 (95%CI 1.25 to 2.91), p=0.04), black ethnicity (SHR 3.88 (95%CI 1.76 to 8.54), p=0.001) and a greater diagnostic severity (SHR 1.91 (95%CI 1.25 to 2.91), p=0.003) (table 4). Neither age nor type of prosthesis was associated with re-PVR.

Table 4

Risk factors associated with re-intervention


Our study is the first to describe contemporary temporal trends in PVR in a nationwide cohort of patients with congenital heart disease. This is, to our knowledge, the largest population-based study on PVR to date. We found that there was a significant increase in the number of PVRs over the years with increasing use of xenografts. Despite a low risk of early mortality, the longer-term risk of death was not insignificant and life-long follow-up in specialist centres is imperative.

Over the last few decades, there has been an immense development in the treatment of congenital heart disease with improved diagnostics, surgical techniques and postoperative treatment. These advances have resulted in a significant increase in the number of young adults requiring PVR, including repaired tetralogy of Fallot. There is now a general acceptance of the adverse effects of longstanding pulmonary regurgitation in this cohort and the need for timely PVR.4 13 Despite this, the ideal type of prosthesis for the pulmonary position still eludes us. The substantial increase in xenograft use in recent years, with a concomitant decrease in homografts, is likely a reflection of the decreasing availability of homografts, rather than a concern with regards to the longevity of homografts.5 14 It has been suggested that homografts provide better haemodynamics and also a suitable substrate for future percutaneous valves,4 but several factors such as donor age,5 ABO incompatibility,5 longer donor ischaemic time,15 younger patient age,15 smaller homograft size16 and heterotopic position of the graft17 have been identified as risk factors for homograft failure. However, the majority of studies comparing homografts with xenografts have shown no obvious differences in outcome2–4 13 18 and this was true also in our study.

Early mortality following PVR was low in our study, similar to previous reports.3 4 13 14 19–22 Longer term survival remains, however, substantial for this young cohort, with no obvious improvement over time: in our study, cumulative mortality was 9% after 10 years from PVR, which corresponds to the majority of studies, reporting a survival between 83% and 97%.3 4 20 23 The most common mode of death in our study was cardiac. In fact, heart failure and arrhythmia have previously been identified as the leading causes of death in adults with congenital heart disease.24 25 Unfortunately, it was not possible to separate these in our study.

Older age was a significant risk factor for death, with older patients more likely to develop heart failure or other cardiac complications of congenital heart disease. Patients in our study died at a similar age as previous cohorts of tetralogy of Fallot.21 24 However, our cohort included patients with both less and more complex disease, making direct comparison difficult.

Percutaneous valves were associated with a higher risk of death in this study. While percutaneous PVR remains a relatively novel technique and close monitoring of patients for long-term complications (eg, infective endocarditis) is important, the increased mortality seen in our study is likely related to referral bias. Patients undergoing percutaneous PVR have, by default, had a PVR prior to entering our study and percutaneous approaches are the procedure of choice for patients with perioperative risk factors, such as ventricular dysfunction, renal failure and multiple previous surgeries. Overall, the long-term mortality after PVR is likely related to the underlying diagnosis of the patients and their comorbidities more than the PVR itself.

We found a low incidence of redo PVR during our follow-up period, with no difference in re-intervention rate between different types of valve. Few studies report on long-term follow-up after PVR, making comparisons difficult.3 21 26 However, our results appear to be similar to most previous studies reporting freedom from reintervention between 76% and 96% at 10 years.3–5 21 26 Few studies14 19 20 have reported higher reintervention rates, with freedom from reintervention between 52% and 70%, primarily in bioprosthetic valves. Nevertheless, these studies included younger populations19 20 or patients operated in a different era.14

There was no difference in outcome between patients receiving mechanical versus bioprosthetic valves,27 even though there are concerns regarding the risk of valve thrombosis or bleeding associated with lifelong anticoagulation. While several studies have shown mechanical valves to be safe when patients receive adequate anticoagulation,6 7 23 27 it is current practice to favour biological valves in the pulmonary position, in order to avoid the risks and nuisance of lifelong anticoagulation in young individuals.

In our study ethnicity was associated with increased risk of re-PVR, even after considering diagnosis and social deprivation. This finding is in accordance with previous studies from the USA reporting black ethnicity to be associated with increased risk of mortality,28 29 longer length of stay,28 higher risk of complications28 and increased risk of readmission30 after congenital heart surgery in both children and adults. It has been suggested that this difference might be due to unequal access to healthcare,29 30 even though the vast majority of patients in this study (95%) were National Health Service patients, with free access to healthcare. We cannot, however, exclude an impact of other socioeconomic factors, not identified in this study, on long-term outcome.


Our study has several limitations. First, the data come from an administrative database, thereby limiting the availability of clinical details such as precise indication for PVR, detailed heart pathology, valve size, subtype of xenograft, information on ABO or HLA type and clinical follow-up data such as echocardiography or MRI. There will always be a risk of misclassification of diagnoses and procedures and we cannot completely rule out that coding patterns can have changed over time or differed across institutions. An advantage is that the database covers the vast majority of all public hospitals in England, thereby providing information on a large unselected cohort. However, the generalisability of the results needs to be considered when comparing the results with other healthcare systems or populations. Second, information on operations prior to 1997 was not available. It is, therefore, not possible to assess the exact number and type of previous procedures, which is likely to affect the outcome of patients (eg, those with percutaneous valves). Third, the cause of death often lacked detail making it difficult to determine exact mechanism leading to death. Finally, patients with a non-specified valve/conduits were not included in the study since the type of conduit was not specified (homograft vs xenograft), making it impossible to distinguish between the two.


Over the last two decades, there has been a significant increase in the number of PVRs performed in England and a significant change in the type of prosthesis employed. There has been an increase in the use of xenografts versus homografts, and percutaneous valves are now an alternative to surgery in suitable patients. Early mortality is low, but the longer-term risk of death is not negligible. Although reintervention rates are low, these patients will almost certainly need further procedures, reinforcing their need for life-long tertiary care.

Key messages

What is already known on this subject?

  • Pulmonary valve replacement (PVR) is often required in patients with congenital heart disease. In recent years, there has been an increase in PVR procedures performed worldwide, including in the UK.

What might this study add?

  • This study describes a significant increase in PVR over the past two decades in a nationwide patient cohort with congenital heart disease.

  • Xenografts were increasingly used compared with homografts. Percutaneous valves were introduced and are now an alternative to surgery in suitable patients.

  • Early mortality is low, but the longer-term risk of death is not negligible with a 15-year mortality for all valves of 90%.

How might this impact on clinical practice?

  • By describing a historic nationwide cohort of PVR, we get a better understanding of how PVR has evolved over the years.

  • Although PVR many places is now considered a simple procedure in relatively simple patients, this study shows that challenges remain and there is an ongoing need to improve the long-term outcome for these patients.


View Abstract


  • Contributors SHL: design of study, data analysis, interpretation of data, drafting of manuscript, critical revision of manuscript, and approval of manuscript; responsible for overall content. KD, MAG, AU, DFS, RA-G and AK: design of study, data analysis, interpretation of data, critical revision of manuscript and approval of manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

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

  • Patient consent for publication Not required.

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.