The UK EndoVascular Aneurysm Repair (EVAR) randomised controlled trials: long-term follow-up and cost-effectiveness analysis

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 15/141/04. The contractual start date was in December 2012. The draft report began editorial review in October 2016 and was accepted for publication in May 2017. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Michael J Sweeting reports grants from the National Institute for Health Research during the conduct of the study. Jessica K Barrett reports grants from the Medical Research Council during the conduct of the study. Roger M Greenhalgh serves as a salaried director of BIBA Medical and has an equity interest in the company, and also serves as an expert witness on behalf of patients with vascular disease.

Copyright statement

© Queen’s Printer and Controller of HMSO 2018. This work was produced by Patel et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2018 Queen’s Printer and Controller of HMSO

Scientific background

Rationale for follow-up from 10 years up to 15 years

In July 1999, the UK EVAR-1 and EVAR-2 trials were commissioned by the NHS research and development (R&D) HTA programme [now known as the National Institute for Health Research (NIHR) HTA programme]. The trials were funded for 4 years (from July 1999 to 2003), followed by a 2-year extension, which was granted to ensure that recruitment targets were met. Subsequently, long-term follow-up support was awarded for a further 5 years of follow-up until July 2010 (up to 10 years of follow-up). The trial objectives were to assess the safety and efficacy of EVAR against the established open surgical treatment in the management of large aneurysms of at least 5.5 cm diameter as measured using CT. Two trials were instigated: EVAR-1 would compare EVAR against OR in patients who were considered fit and suitable for both procedures and EVAR-2 would compare EVAR against no intervention for patients who were considered suitable for EVAR but unfit for OR. The primary outcome for both trials was mortality, with secondary outcomes of complications, reinterventions, HRQoL, adverse events, costs and cost-effectiveness.

In follow-up until December 2009, 27 sac ruptures had occurred after EVAR, carrying a 67% mortality rate. Type I endoleak, type III and type II with sac expansion, kinking and migration were identified as the ‘cluster’ of complications associated with these late EVAR secondary sac ruptures. 111 It is possible that the early aneurysm-related mortality benefit of EVAR is lost as a result of these late sac ruptures and death. If this trend were to continue, after 15 years, OR would have a lower aneurysm-related mortality than endovascular repair, even with every attempt made to intervene to correct known causes of sac expansion to reduce the risk of secondary rupture in EVAR patients. The possible future projection is shown graphically for EVAR-1 in Figure 1. The rationale for continuing follow-up to 15 years in EVAR-1 was to test this hypothesis. The primary outcome measure for the 15-year follow-up was aneurysm-related mortality and total mortality.

FIGURE 1.

Projected aneurysm-related survival after 8 years of follow-up in EVAR-1 based on observed trends. NEJM, New England Journal of Medicine.

Our secondary outcomes included reinterventions (time to first reintervention, to first reintervention for a life-threatening problem and to first serious reintervention).

Other secondary outcomes included complications, cost and cost-effectiveness. The evaluation of cost-effectiveness of OR and EVAR after a 15-year time period is particularly pertinent because EVAR-related complications continue to occur many years after the original procedure. This objective required knowledge of continuing graft-related complications and use of hospital resources to the end of 2014.

Based on the 27 sac ruptures reported by Wyss et al. ,111 in 2010, it was hypothesised that sac growth after EVAR is associated with an increased risk of rupture and the presence of a significant complication or endoleak. It was therefore assumed that patterns of sac growth might predict the need for reintervention, even in the absence of a defined endoleak. A further outcome was therefore to investigate any association between sac growth (Figure 2) and secondary rupture, complication, endoleak and reintervention correction.

FIGURE 2.

Change in sac diameter of a random selection of 50 patients (post-EVAR measurements only, preoperative measurements excluded). Graphs by ID.

In 2012, there was a call with vignette (NIHR Evaluation, Trials and Studies Coordinating Centre – HTA) for addressing the question ‘What is the optimal management strategy for type II endoleaks?’ The then Director of the HTA programme, Professor Tom Walley, questioned whether or not the HTA-funded EVAR trials would produce data that could answer this question. The authors of the EVAR trials considered the Director’s question and considered that an individual patient data (IPD) meta-analysis of EVAR-1, the DREAM trial, the OVER trial and ACE would produce much useful information and among it some on type II endoleaks. We informed the HTA programme that we would be able to address the question ‘What is the optimal management strategy for type II endoleaks?’ without requiring further data collection.

An IPD meta-analysis of four multicentre randomised trials of EVAR compared with OR was conducted to a prespecified analysis plan, reporting on mortality, aneurysm-related mortality and reinterventions.

Data from the IPD meta-analysis have been used to address the question of the management of type II endoleaks following EVAR.

Participants

Men and women aged ≥ 60 years were enrolled into EVAR-1 between 1 September 1999 and 31 August 2004 from 37 hospitals in the UK. Twelve hundred and fifty-two patients whose maximum aortic aneurysm diameter was ≥ 5.5 cm, with aortic morphology compatible with endograft placement within the manufacturer’s instructions for use (IFU) and who were deemed to have an acceptable risk of postoperative death for both procedures were randomised to undergo either OR or EVAR.

EVAR trial 2 was designed to address the question of whether or not endovascular repair reduces the rate of death among patients who were considered to be physically ineligible for OR. During the same time period, patients of both sexes who were at least 60 years of age with large aneurysms (≥ 5.5 cm in diameter) and who were deemed unfit for OR, were randomly assigned to undergo either EVAR or no repair.

Interventions

In EVAR-1, patients were randomised to undergo either OR or EVAR. In EVAR-2, patients were randomly assigned to undergo either EVAR or no repair.

Objectives

The objective was to follow up the world’s first RCT of EVAR compared with OR with up to 15 years of follow-up; to reconvene the EVAR trial centres and brief them on the uncertainty of outcome versus OR in the very long term and to inform them of the risk of secondary rupture;111 to re-establish a link via the new EVAR trial manager and identify and brief the current lead vascular surgeon and radiologist at each of the 37 trial centres; with funding for clinical trial data collection and results of CT scanning or other imaging, to identify the contacts for these purposes; and then to be in a position to achieve the 10–15 years of additional clinical follow-up for all alive patients from 1 September 2009.

Outcomes (measures)

Our primary outcome was aneurysm-related mortality and total mortality. Aneurysm-related mortality was defined as all deaths from aneurysm rupture before repair, within 30 days of the primary procedure, within 30 days of any reintervention attributable to the aneurysm, from other aneurysm-related causes (including graft infection or fistula), or from secondary aneurysm rupture after repair. Our secondary outcomes included reintervention (time to first reintervention, first reintervention for a life-threatening problem and first serious reintervention). Further secondary outcomes included complications, sac growth and risk of late complications, costs and cost-effectiveness.

For the primary mortality outcome, patients were followed up from 1 September 1999 to 30 June 2015 [using record linkage from the Office for National Statistics (ONS), with death classification based on death certificates and clinical information provided to the Endpoint Committee]. Patients were followed up for death (total and aneurysm related), dates as above, and for graft-related complications and reinterventions from 1 September 2009 to 31 March 2015. For graft-related reinterventions between 1 September 2009 and 31 March 2015, follow-up was predominantly using record linkage to administrative data for hospital readmissions and reinterventions via Hospital Episode Statistics (HES).

Reinterventions, including incisional hernia repair throughout the trial and other operative procedures preceding death, were subsequently checked with the trial centres, with 89% concordance between administrative and clinical site data. Graft-related complications and reinterventions were also directly obtained from the trial centres with a new case record form for our follow-up between 1 September 2009 and 31 March 2015. The primary analysis compared rates of total mortality and aneurysm-related mortality to 30 June 2015; aneurysm-related mortality was adjudicated and confirmed by the Endpoint Committee.

Sample size

As of 1 September 2009, there were 711 patients with an average age of 80 years reported alive and under follow-up in EVAR-1 (357 and 354 in the EVAR and OR groups, respectively), and 96 patients with an average age of 82 years reported alive and under follow-up in EVAR-2 (50 and 46 in the EVAR and no-intervention groups, respectively). This gives 80% power at the 5% significance level to detect hazard ratios (HRs) of 1.25 and 1.83 during the extended follow-up periods in EVAR-1 and EVAR-2, respectively, assuming 10% to be still alive at the end of June 2015.

Randomisation and blinding

Randomisation was conducted using computer-generated sequences of randomly permuted blocks stratified by centre at the trial hub (Charing Cross Hospital). Patients in EVAR-1 were randomly allocated (1 : 1) to undergo either OR or endovascular repair. In EVAR-2, patients were randomly allocated (1 : 1) to undergo either EVAR or to have no intervention. There was no blinding in either trial.

Statistical methods

All analyses were performed according to a predefined statistical analysis plan and were based on the intention-to-treat (ITT) principle, with outcomes assessed from the time of randomisation. Cox regression modelling was used to compare total mortality, aneurysm-related mortality and time to first graft-related reintervention. Crude regression estimates were presented as well as ones adjusted for two sets of baseline covariates: primary adjustment for age, sex, aneurysm diameter, forced expiratory volume in 1 second, log-creatinine level and statin use; secondary adjustment for the primary covariates as well as body mass index (BMI), smoking status (current, past and never), systolic blood pressure and serum cholesterol level. For the graft-related reintervention analysis, additional secondary adjustment was made for top neck aortic diameter at the level of the lowest renal artery, neck length (distance between the lowest renal artery and the start of the aneurysm) and common iliac diameter (largest of both legs). The primary and secondary adjustment results were very similar and the latter are reported. Baseline data were almost complete, with 94% and 92% of patients in EVAR-1 and EVAR-2, respectively, having a complete set of covariates for the adjusted analyses. HRs were presented as the EVAR group relative to the OR group for EVAR-1 and the EVAR group relative to the no-intervention group for EVAR-2. Owing to non-proportional hazards during the first 8 years of follow-up,94,95 data were analysed by splitting follow-up into four time periods: 0–6 months, 6 months–4 years, 4–8 years and > 8 years. HRs presented in these time periods help to explain the non-proportionality, but must be interpreted conditional on surviving up to the time point in question and are therefore dependent on how the randomised groups fared in the follow-up prior to the time period. Deviations from the proportional hazards assumption were assessed overall and within these periods by regressing scaled Schoenfeld residuals against log of time. Regression estimates are presented both unadjusted and adjusted for baseline covariates. Kaplan–Meier estimates were used to show survival probabilities up to 15 years in each group. Tests of interaction were performed for total mortality and aneurysm-related mortality between randomised group and sex, age and aneurysm diameter (the last two as continuous variables).

In EVAR-1, two sensitivity analyses were performed to allow inclusion of patients with missing covariates in the adjusted models: first, the missing indicator method112 and, second, multiple imputation using chained equations that included terms for the event outcome and the estimate of the cumulative baseline hazard. 113–117

In addition, a per-protocol analysis was performed in EVAR-1 on data from patients who had undergone their randomly assigned treatment. This analysis excluded patients who did not undergo aneurysm repair, those who underwent emergency repair, those in whom the repair was abandoned during surgery (i.e. the aorta was left unrepaired) and those who did not undergo the randomly assigned procedure.

In the EVAR-1 analysis, time to first reintervention analyses were conducted separately for any graft-related reintervention, any serious reintervention (see two or three stars in Table 1) and any life-threatening condition (see four or five stars in Table 1). The criteria used to censor individuals are provided in Severity score for reinterventions. Further analyses were also undertaken in patients without any reintervention between randomisation and 2 years and without any reintervention between randomisation and 5 years.

All analyses were carried out in Stata version 13 (StataCorp LP, College Station, TX, USA). The Data Monitoring and Ethics Committee (DMEC) approved the statistical analysis plan.

Methods used for very long-term follow-up and changes to protocol

Patients in both trials were followed up comprehensively for rates of perioperative and late death, graft-related complications, reinterventions and resource use until September 2009. 94

By the end of the 10-year follow-up in September 2009, three of the participating centres had merged with other local participating centres into single NHS trusts (Leeds, Belfast and Glasgow). All the trial participants at four further centres that had recruited very small numbers of patients had died and so were also not included in late follow-up. A total of 31 centres therefore participated in late follow-up to 15 years. Ethics approval for extended patient follow-up was obtained on 16 February 2011 from the UK North West Multicentre Research Ethics Committee, which did not require patients to be reconsented for ongoing follow-up of the EVAR trials up to 15 years (reference number 98/8/26 for EVAR-1 and reference number 98/8/27 for EVAR-2). Patient consent for the ongoing audit of medical notes for follow-up was given by each patient before randomisation. The EVAR trials’ extended follow-up for both trial 1 and trial 2 was to be based on a review of hospital notes and continued flagging with the Medical Research Information Service for date and cause of death. As with the earlier follow-up, a lead clinician and a local co-ordinator were identified at each of the centres. The co-ordinator was to be remunerated for annual data extraction of clinical information from patient hospital notes relating to aneurysm-related complications and reinterventions from routine hospital visits for surveillance of the aneurysm repair for 2010–14.

Very long-term follow-up for reinterventions: October 2009 to March 2015

For the current phase of very long-term follow-up, funding was not achieved until 26 April 2012 and the trial manager was appointed on 1 December 2012. There was a delay in the signing of contracts between Imperial College London and the NIHR. The collaboration agreement was a four-way agreement that included the University of Cambridge (statistics) and the University of York (cost-effectiveness). Final contracts were not signed until 28 August 2013. In addition, local R&D approval was required from the 32 NHS trusts (30 in England, one in Scotland and one in Northern Ireland). First contact with each of the NHS trust R&D offices was made by the trial manager in December 2012. This process was not completed until 28 January 2014. A new case record form (see Appendix 1) was designed with the expectation of collecting data on complications and reinterventions retrospectively for 2010–12 from clinical notes review and then prospectively for a further 3 years. Imaging scans and reports were also gathered as well as information on adverse events such as strokes, myocardial infarctions (MIs), chronic renal failure and amputation.

Owing to the delays described above, prospective data collection did not start until late in 2013. After 1 year of data collection, it became clear that for EVAR-1 the majority of OR patients (72%) and some EVAR patients (32%) were no longer being followed up at the original trial centres and some hospital mergers had taken place leading to patients being followed up at non-trial hospitals. For EVAR-2, nearly all of the patients had died. At a meeting of the Data Management Subcommittee of the Trial Management Committee (TMC) on 4 November 2014, it was decided to obtain permission for record linkage follow-up for all English trial patients alive or lost to follow-up in September 2009, by obtaining administrative data for hospital readmissions and procedures for patients from EVAR-1 only. There were so few EVAR-2 patients surviving, that given the need to provide robust justification to obtain ethics approval for using administrative data, this option was not pursued for EVAR-2 patients.

Obtaining permission to do this was a two-stage procedure. As patients had not given consent for access to their routine hospital administrative data on enrolment during 1999–2004, we had to obtain ethics approval (section 251) from the Confidentiality Advisory Group (CAG) to enable disclosure of confidential patient information. This required a lengthy process to demonstrate our institution met the stringent information governance requirements. The application to the CAG was submitted on 23 September 2014 and approval was granted on 18 February 2015. The second part of the process involved an application for HES data, which was submitted on 19 February 2015. HES is a government data warehouse containing records of all patients, tracked by their unique NHS number, admitted to NHS hospitals in England and contains details of inpatient care, outpatient appointments and accident and emergency attendance records. HES data allowed us to obtain patient data on all hospital readmissions and procedures (with procedure coding) until 31 March 2015 for 655 of the 724 (90%) patients alive in September 2009, including 13 patients previously lost to follow-up. Only seven English patients evaded HES record linkage. All procedures identified via HES which were potentially related to the original aneurysm repair, including abdominal hernia repairs, were validated with the relevant trial centre or where possible with other non-trial hospitals. There was 89% concordance between administrative data and clinical site data.

Sixty-nine patients, including 62 in either Scotland or Northern Ireland, could not be followed up via HES, of whom 48 (70%) had local follow-up at the trial centre. These 48 patients included 8 out of 15 patients in Scotland, 38 out of 47 patients in Northern Ireland and two out of seven English patients who were unmatched by HES. There were no Welsh centres. This process of using complementary sources of information from record linkage with local validation increased the completeness of patient follow-up in EVAR-1 from 51% to 97%, with only 25 patients being lost to follow-up for reinterventions by 31 March 2015.

Given the time taken for this process, the NIHR agreed to the project end date being extended until 31 August 2016.

Severity score for reinterventions

There is no recognised reporting system for reintervention or reintervention severity, to facilitate allocation of resource use. Therefore, the grading of the severity of aneurysm-related reinterventions and the associated use of high dependency or intensive care (high-dependency unit; HDU/ICU) were obtained by a questionnaire that was sent to each of the principal investigators at the 32 trial centres. Thus the EVAR trials severity score obtained is shown in Table 1.

Reinterventions recorded during follow-up included those that took place during primary admission and subsequent admissions. Also included are previously ignored reinterventions for ORs [e.g. incisional hernias (see Table 1 for full list of reinterventions recorded)]. The date of any reintervention during the primary admission was taken to be the mid-point between the date of admission and the date of discharge or in-hospital death. The censoring time for reinterventions was determined as follows:

• For all patients from Scotland and Northern Ireland and those from England unmatched in HES with a known follow-up in 2014/15, the date of their latest follow-up was used for censoring.

• For alive patients in Scotland and Northern Ireland and those from England unmatched in HES without a follow-up in 2014/15, the date of last follow-up or the date of audit of their notes was used for censoring.

• For all patients in England tracked via HES, censoring was taken as 31 March 2015, unless the date of death was earlier.

• For dead patients without a follow-up in 2014/15, the date of death was used for censoring, providing it occurred within the year after their last follow-up or date their notes were audited, otherwise these latter dates were used for censoring.

Very long-term follow-up for complications: October 2009 to March 2015

As administrative data do not report complications, only reinterventions, this meant that it would no longer be possible to conduct a comparison of complication rates by randomised group. Complication data were obtained in satisfactory numbers only for analysis of complication rate by time period in the group randomised to EVAR.

Complications were only obtained from the trial centres using the new case record form for late follow-up (see Appendix 1). The new case record form was designed to capture all information about the cluster of complications identified by Wyss et al. 111 as preceding rupture as well as all reinterventions over the whole 5-year follow-up period. Trial centres were reminded that all patients should continue in regular follow-up (the protocol specified annual follow-up including clinical, imaging and serum creatinine level assessment) and all patients, including lapsed follow-ups, should be recalled for a final clinical and imaging follow-up in 2014. As previously, the management of aneurysm-related complications was left to the discretion of the trial centre.

Very long-term follow-up for mortality: October 2009 to June 2015

Follow-up for mortality was unchanged throughout the trial and data for the date, place and cause of death from the ONS were obtained from September 2009 to June 2015. These data, together with information about reinterventions, were assessed by the Endpoint Committee, who were blinded to randomised group and trial in which the patient participated. The committee considered all events from both EVAR-1 and EVAR-2.

All death certificates were reviewed by an Endpoint Committee to agree cause of death (Figure 3). Aneurysm-related mortality was defined as all deaths (1) from aneurysm rupture before repair, (2) within 30 days of the primary procedure, (3) within 30 days of any reintervention attributable to the aneurysm, and (4) other aneurysm-related causes (including graft infection or fistula) or from secondary aneurysm rupture after repair. Deaths for which the underlying cause was attributed to International Classification of Diseases, Tenth Edition codes I713–719 were also classified as aneurysm related.

FIGURE 3.

Classification of death codes assigned by the Endpoint Committee.

Imaging follow-up

The original EVAR trials protocol included annual imaging via a CT scanning to record the aneurysm sac diameter in EVAR patients and anastomosis diameter in OR patients. For the late follow-up, wherever CT scanning was performed as part of routine follow-up, images were copied for transfer to the trial co-ordinating centre. After 2010, many of the centres had reverted to using duplex ultrasound and the number of CT scans available was largely curtailed. Often with duplex scans the infrarenal maximum outer-to-outer anteroposterior diameter is recorded. For EVAR, crucially, maximum sac diameter is an essential finding. Endoleaks and other complications were noted from the imaging reports reflecting the opinion of the local NHS radiologist. All CT scans that were available were examined in the trial centre core laboratory by a skilled and trained vascular fellow to note known factors associated with eventual secondary sac rupture.

Additional 231 patients not included in earlier reports

After publication of the 30-day mortality results (not the primary outcome of the trial), 26 of the 37 centres remained in equipoise and wished to continue randomising patients until the primary outcome results (mortality after minimum 1-year follow-up from randomisation) were available. The chairperson of the DMEC approved this further recruitment, provided it was undertaken as a separate study, which could report on these later cases, with more experienced clinicians responsible particularly for EVAR. Two hundred and thirty-one additional patients were therefore randomised to a separate study between 1 September 2004 and the publication of mid-term results on 15 June 2005 and are being reported for the first time. One hundred and seventy-five patients were randomised to either OR or EVAR in EVAR-1 and 56 patients to either EVAR or no intervention in EVAR-2 (total 231 patients). Patient follow-up for mortality was the same as for the main EVAR-1 and EVAR-2 patients. In EVAR-1 this included a further 175 patients who have now been included in sensitivity analyses for mortality. Of these 175 patients, 88 patients were randomised to the EVAR arm of the study, of whom 70 were alive in September 2009. Eighty-seven patients were randomised to the OR arm of the study, of whom 66 were still alive in September 2009. For EVAR-2, 28 patients were randomised to EVAR, of whom 11 were alive in September 2009 and 28 patients were assigned to the no-intervention arm, of whom eight were alive in September 2009. The delays described above made it very challenging to obtain retrospective follow-up data for these 231 patients. Evaluation of data collection for these patients by the Data Management Subcommittee of the TMC on 30 June 2014 showed follow-up information on only one-third. Where available, we therefore focused on obtaining details of the primary operation for these additional patients. This information included (1) the preoperative diameter of the aneurysm; (2) the date of operation; (3) whether the patient had an EVAR or OR; and (4) if EVAR, which device was used. All of these additional 231 patients were registered with the Medical Research Information Service for mortality reporting. This allowed the 175 EVAR-1 patients to be included in sensitivity analysis for all-cause mortality, but little else, including the impact of later generation endografts on complication and reintervention rates.

Data management

All data transferred to the co-ordinating centre at Imperial College London were entered into the trial database by the trial manager based at Charing Cross Hospital. Data were handled in a secure way according to good clinical trial practice with due consideration for patient data confidentiality. The pre-existing database and new database for late follow-up were both stored securely on a dedicated, stand-alone (non-networked) computer system that was maintained in a locked office. The system was protected by strong passwords and advanced encryption standard (14 rounds and a 256-bit key). It was accessible only to the trial manager and the data manager. The trial manager was the only person responsible for data entry. Data entry errors were assessed by carrying out double data entry on a randomly selected 10% of patients. The case record form used for late follow-up is included in Appendix 1.

The approval for obtaining HES data was granted on the condition that all data received should be stored on the Imperial College Healthcare NHS Trust system accessible to only the trial manager. The reason for this stipulation was the lack of a satisfactory information governance toolkit assessment at Imperial College London which the NHS trust had in place. The information governance toolkit is a performance tool produced by the Health and Social Care Information Centre for the Department of Health. It draws together legal rules and central guidance and presents them in one place as a set of information governance requirements. Organisations are required to carry out self-assessments of their compliance against these requirements, which include management structures and responsibilities, confidentiality and data protection and information security. A realistic time scale for completion of a satisfactory toolkit assessment is therefore in the order of several months, and in 2015 very few universities had a satisfactory assessment in place. This requirement was the reason for the delay in obtaining HES data and was satisfactorily resolved as a result of our institutions partnership with the NHS trust.

Computed tomography images that were transferred anonymously to the trial co-ordinating centre were also treated with the same level of security and confidentiality. These were subsequently transferred to the dedicated research workstation in the core laboratory at Charing Cross Hospital.

Patient and public involvement

Patient and public involvement has been a prominent aspect in the EVAR-1 and EVAR-2, and never more so than when explaining to patients the difference of the methods offered at randomisation. During follow-up we were increasingly concerned that failure to carry out clinical and imaging annual assessments could be detrimental to patient outcomes. Thus, the trial manager visited every trial centre co-ordinator who was given a copy of the Wyss paper (2010)111 to stress the need for ongoing follow-up and for this to be explained by them to patients.

In February 2015, we were granted approval to obtain confidential data on hospital admissions and procedures via HES. In June 2015, data were obtained for patients who were participants in EVAR-1 at hospitals in England and who had been lost to aneurysm-related follow-up. In order to obtain approval from the CAG we interviewed (face to face), trial patients at routine hospital visits. We explained to the patients that confidential records were held in a government database which stores information about all patient hospital visits in England. We asked them their view on access to confidential data from NHS and government records. All interviewed patients stated that they would be happy for us to access information in this way.

Organisational structure and committees

The late follow-up of the EVAR trials was managed centrally by the principal investigator (Professor Roger Greenhalgh), the trial manager (Dr Rajesh Patel) and Professor Janet Powell (co-applicant), who are based at the Charing Cross Hospital site of Imperial College London. Statistical expertise was provided by Dr Michael Sweeting (University of Cambridge) and costs and cost-effectiveness expertise were provided by Dr David Epstein who related to Professor Mark Sculpher (University of York).

Trial Management Committee

This committee was concerned with the day-to-day running of the trials and related to both the DMEC and the Trial Steering Committee. It was chaired by Professor Roger Greenhalgh and members included Professor Janet Powell (vascular biology), Dr Michael Sweeting (statistics), who related to Professor Simon Thompson, Dr David Epstein (health economics), Mr Colin Bicknell (vascular surgeon), Miss Regula Von-Allen (vascular surgeon), Mr Thomas Wyss (imaging core laboratory) and Dr Nick Burfitt (interventional radiologist). Dr Rajesh Patel (trial manager) attended all meetings to update committee members on trial progress and highlight any problematic issues.

Data Monitoring and Ethics Committee

This committee was regularly updated on trial progress and the DMEC communicated with the Trial Steering Committee. The committee was chaired by Professor Gerry Fowkes and included Dr Robert Morgan from the British Society of Interventional Radiology and Professor Bruce Campbell from the National Institute for Health and Care Excellence (NICE).

Trial Steering Committee

This committee was chaired by Professor Richard Lilford (University of Birmingham) and included Roger Greenhalgh for the applicants and TMC, as well as Michael Wyatt (Newcastle), Simon Thompson (statistics) and Mark Sculpher (health economics). The role of the committee was to liaise between the DMEC and TMC and oversee any issues relating to the progress of the trials or needs for additional funding.

Endpoint Committee

This committee was chaired by Professor Janet Powell and included an independent vascular surgeon (Professor Alison Halliday) and a consultant cardiologist (Dr Simon Gibbs). All death certificates were centrally coded at the ONS and were reviewed by this committee in relation to any aneurysm-related procedures. All available data relating to the death and a primary underlying cause of death were classified according to the groupings presented in Figure 3, where death codes 1, 2 and 12 were classified as aneurysm related. Aneurysm-related deaths were defined as all deaths occurring within 30 days of the primary aneurysm repair or any reintervention for a graft-related complication unless over-ruled by post-mortem findings or a separate procedure (unrelated to the aneurysm) that took place between the aneurysm intervention and death (code 1); all deaths from rupture of an unrepaired aneurysm (code 2); and all deaths from rupture of a repaired aneurysm, usually endograft rupture (code 12). In addition, late complications of aneurysm repair, such as aortoduodenal fistula or bowel obstruction, were recorded as procedure-related deaths (code 1).

Patients

From 1 September 1999 until 31 August 2004, we recruited 1252 patients to participate in this trial, who were equally, and randomly, assigned to the two treatment groups. There were no differences in baseline characteristics between the groups, mean age 74 years, 1135 men (Table 2). 101

Baseline characteristics were compared between randomised groups and no differences were observed.

Patients were followed until 30 June 2015 (mean 12.7 years; median 12.4 years; minimum 1.8 years; maximum, 15.8 years); mean person-years observation to either death or end of the study was 8.0 years. By 30 June 2015 only four patients were lost to follow-up for mortality and 25 for reinterventions (this includes five patients in the EVAR group and 20 patients in the OR group), with data now available from record linkage for 13 of the 17 patients previously lost to mortality follow-up (Figure 4).

FIGURE 4.

CONSORT diagram for mortality and reinterventions. GP, general practitioner.

For 13 individuals, a cause of death was established based only on a death certificate. Annual clinical follow-up with CT scanning or duplex imaging reduced steadily over the period of the trial and was consistently lower in the OR group (Table 3).

Over the course of follow-up, CT was carried out a median of six (interquartile range 3–8) times per patient in the EVAR group and three (interquartile range 1–6) times in the OR group. Of the patients who had not had death reported by 1 September 2009, 655 of 728 (90%) patients were tracked with HES, including 13 patients previously lost to follow-up, with local follow-up reported in 48 (70%) of the remaining patients (see Figure 4). After publication of the 30-day mortality results,100 26 of the 37 trial centres remained in equipoise and continued recruitment into a separate study from 1 September 2004 until 15 June 2005, when primary outcome results were published,101 with a further 175 patients not reported previously but now used in sensitivity analyses for mortality only. Operative characteristics are shown for the 1252 patients in the EVAR-1 in Table 4a and the additional 175 patients in Table 4b.

Aneurysm-related and total mortality

During 9968 person-years of follow-up, 910 deaths occurred, 101 of which were aneurysm related (Table 5). Overall aneurysm-related mortality was 1.1 deaths per 100 person-years in the EVAR group and 0.9 deaths per 100 person-years in the OR group [adjusted HR 1.31, 95% confidence interval (CI) 0.86 to 1.99; p = 0.21]. For total mortality, there were 9.3 deaths per 100 person-years in the EVAR group and 8.9 deaths per 100 person-years in the OR group (adjusted HR 1.11, 95% CI 0.97 to 1.27; p = 0.14). The results for the sensitivity analyses that included patients with missing baseline covariates were similar for aneurysm-related and total mortality (Table 6).

There was evidence of deviation from the proportional hazards assumption for aneurysm-related mortality (p < 0.0001), with an early benefit of EVAR during the first 6 months, counteracted by an increase in aneurysm-related mortality beyond 4 years, the difference being most marked after 8 years (adjusted HR 5.82, 95% CI 1.64 to 20.65; p = 0.0064) (see Table 5). There was also evidence of deviation from the proportional hazards assumption for total mortality (p = 0.0232), with an early benefit of EVAR during the first 6 months, similar mortality between the groups from 6 months to 8 years, but thereafter an increase in mortality in the EVAR group (adjusted HR 1.25, 95% CI 1.00 to 1.56; p = 0.0484) (see Table 5). Kaplan–Meier curves for aneurysm-related and total mortality are shown in Figure 5. Aneurysm-related mortality curves cross over between 6 and 8 years and total mortality curves diverge after 10 years. Median survival was 8.7 years and 8.3 years in the EVAR and OR groups, respectively (log-rank p-value 0.49). Sensitivity analyses, including the additional 175 patients from the separate 2004–5 study, yielded very similar results (Figure 6).

FIGURE 5.

Kaplan–Meier estimates for total and aneurysm-related survival over a maximum of 15 years’ follow-up.

FIGURE 6.

Kaplan–Meier estimates for total and aneurysm-related survival including the 175 additional patients randomised in a separate study from 1 September 2004 to 15 June 2005 (EVAR, n = 88; OR, n = 87) for sensitivity analysis.

The full list of causes of death by time period are given in Table 7. Overall, there were 31 deaths from rupture after aneurysm repair in the EVAR group and five in the OR group. Two patients in the OR group had ruptures in 2010 and 2012 having refused operation. Overall, there was no evidence of a difference in cancer-related mortality between the groups (adjusted HR 1.09, 95% CI 0.84 to 1.40; p = 0.53), although there was some evidence of an increase in the EVAR group after 8 years (adjusted HR 1.87, 95% CI 1.19 to 2.96; p = 0.007) (Figure 7 and Table 8).

FIGURE 7.

Kaplan–Meier estimates for cancer-related survival over a maximum of 15 years’ follow-up.

Details of the primary cancer sites underlying mortality after 8 years are shown in Table 9.

There was no evidence of significant interactions between the randomly assigned treatment and age, sex or aneurysm diameter for either aneurysm-related or total mortality (p > 0.10 for all comparisons).

Per-protocol analysis including 598 patients in the EVAR group who received elective EVAR and 567 patients in the OR group who received elective OR again showed strongly the benefit of EVAR during the first 6 months, counteracted by an increase in aneurysm-related mortality at all subsequent time periods (Table 10); the increase being proportionately greater than for the analysis by randomised group. Overall, aneurysm-related mortality was significantly higher in the EVAR group (1.0/100 person-years) than in the OR group (0.6/100 person-years) (adjusted HR 1.76, 95% CI 1.07 to 2.89; p = 0.0263). There was weak evidence that total mortality was higher in the EVAR group, at 9.1 per 100 person-years, than in the OR group, at 8.4 per 100 person-years (adjusted HR 1.14, 95% CI 0.99 to 1.31; p = 0.07).

Aneurysm-related reinterventions

During 9715 person-years of follow-up, there were 258 graft-related reinterventions performed in 165 patients in the EVAR group and 105 graft-related reinterventions performed in 74 patients in the OR group, with rates to first reintervention of 4.1 and 1.7 per 100 person-years, respectively (adjusted HR 2.42, 95% CI 1.82 to 3.21; p < 0.0001) (Table 11). The reintervention rate was significantly higher in the EVAR group for any reintervention and serious reinterventions in the first 4 years (Figure 8a, see Table 11) and for life-threatening reinterventions (including conversion to OR, repeat EVAR and treatment of graft infection), in the periods 6 months to 4 years and beyond 8 years (Figure 8b, see Table 11). Even after 2 years (Figure 8c) or 5 years (Figure 8d) without any life-threatening reintervention, new life-threatening reinterventions occurred at any time to 15 years of follow-up. The relative difference in reintervention rate between the groups was highest in the period 6 months to 4 years after randomisation, particularly for the most serious reinterventions (see Table 11). A similar pattern, by time period, was observed for second and subsequent reinterventions (Table 12).

FIGURE 8.

Kaplan–Meier estimates over a maximum of 15 years’ follow-up for (a) the time to first reintervention; (b) the time to first life-threatening reintervention; (c) the time to first life-threatening reintervention for individuals who have survived 2 years free of a life-threatening reintervention; and (d) the time to first life-threatening reintervention for individuals who have survived 5 years free of a life-threatening reintervention.

Discussion of EVAR trial 1

The most striking finding of these very long-term results is that both aneurysm-related and total mortality rates are greater in late follow-up for patients who were randomised to EVAR than those randomised to OR, although over the whole follow-up period the average total and aneurysm-related mortality rates were not significantly different between groups. The significant late divergence of the survival curves in favour of OR (see Figure 5) can be partly explained through greater contribution to late mortality from aneurysm-related and cancer deaths in the EVAR group, and the fact that these were elderly patients and the survival curves will inevitably start to converge in older ages.

Total and aneurysm-related mortality were lower in patients who received EVAR in the first 6 months, but increased after 6 months’ follow-up, leading to a significantly higher rate after 8 years’ follow-up in EVAR than in those who received OR. After the first 6 months, the increased aneurysm-related deaths in the EVAR group were predominantly from secondary sac rupture. Over the whole duration of follow-up, two aneurysm-related deaths followed reintervention, but 31 of the others were from secondary sac rupture, partly attributable to a failure to have corrected underlying causes of sac expansion from endoleak. 111 In those patients allocated to OR, there were five secondary ruptures, of which four were in patients who received EVAR and the remaining secondary rupture occurred more than 8 years after OR.

Reinterventions occurred in both groups throughout follow-up, including those who were free from reintervention after 2 or even 5 years, but the rate of reintervention was higher in the EVAR group at all time periods. These late reinterventions included those with high severity score, indicating that there was never a safe period to abandon follow-up for patients with EVAR. However, in this trial, some patients were discharged from surveillance and, therefore, lost the option of planned reintervention. With an average age of 74 years at randomisation, there could have been pressing clinical reasons not to reintervene for some patients with long-term follow-up because of age and frailty. A criticism in earlier reports from this trial, that not all incision-related reinterventions following OR were reported, was remedied in this long-term follow-up.

Limitations of EVAR trial 1

Limitations of this trial include that devices used were implanted between 1999 and 2004 and newer devices may be expected to have better results118 and imaging for sizing and placement of endografts has improved. The original trial protocol was for annual follow-up by CT, which was used in the early stages of the trial. However, in the later stages, many of the EVAR patients were followed up using ultrasonography. This change from CT to ultrasonography was influenced by increasing concern about radiation exposure. 119 Moreover, imaging follow-up declined over time, particularly for the OR patients. Consequently, reinterventions became less likely once surveillance ceased. Nor can we assume that follow-up practice is the same in the rest of the world as it is in the UK, where many patients were discharged from surveillance after several years. Since the EVAR group patients had more diligent follow-up than the OR patients, aneurysm-related mortality may have been underestimated in the OR group, although this does not affect our findings for total mortality. A further limitation is that, as a result of decreasing clinical follow-up at the original trial hospitals, the methodology to identify reinterventions changed after 2009 to predominantly using record linkage through the HES administrative data set, these reinterventions subsequently being validated at the trial hospitals. However, these data also captured data for patients whose care had moved to non-trial hospitals and recovered some patients previously lost to follow-up.

Patients prefer EVAR,120 and today it is the method of choice for repair of AAA. EVAR devices improve constantly, and sizing and imaging for deployment is better than between 1999 and 2004: a corollary is that experience in OR is declining. However, aortas with aneurysmal disease continue to dilate, and in time a good device can leak or migrate and even an OR can rupture. Challenges in the future to maintain the initially superior results of being in the EVAR group include the need to halt the dilating disease process as well as devices which allow for this inevitable dilating process over the years. It is fortunate that the long-term results of this study can act as a benchmark against which new endovascular technologies for aneurysm repair can be compared at each time point. In the meantime, surveillance must be addressed in clinical guidelines: it should be diligent, regular, easy and avoid CT scanning where possible, perhaps concentrating on the sac diameter after EVAR either by ultrasonography or novel implantable sensor devices. 121–125

Introduction

This chapter estimates the total cost per patient of each procedure over 15 years in patients who were suitable for EVAR and fit for OR. The costs for each patient include the index procedure, aneurysm-related reinterventions and surveillance, as received in EVAR-1. Other potentially related categories of acute and chronic health-care costs, such as MI, stroke, amputation, renal failure and cancer, are not included in this analysis, as data on health-care use for these reasons were not collected after 2009.

As this chapter presents a description of the data, rather than an economic evaluation, costs were not discounted. A full economic evaluation showing discounted costs and quality-adjusted life-years (QALYs) is presented in Chapter 5.

Methods

All resources in the current analysis were costed at 2014/15 prices. The follow-up is divided into two periods: (1) 1 September 1999–31 August 2009; and (2) 1 September 2009–31 March 2015. The method of costing is different for the two periods. During the first period (1999–2009), detailed hospital resource use data were available for the index procedure and aneurysm-related reinterventions for each patient from the case record forms. The same methodology is used here as in the previous publication,94 the only difference being that, in the previous publication, these resources were costed at 2008/9 prices. In the current analysis, these same resources have been recosted using largely new unit costs at 2014/15 prices (Table 13).

Detailed hospital resource use data were not widely available from case record forms after 1 September 2009 because of loss to follow-up, especially in patients in the OR group. To supplement the trial data and information from hospital notes, data on hospital episodes for each patient from 1 September 2009 to 31 March 2015 were obtained from national NHS HES (see Chapter 2 for the definition of the censoring date for reinterventions). Each episode was assigned a Hospital Resource Group code127 and matched with the published national average cost of that episode. 126 A Hospital Resource Group is similar to a Diagnostic Resource Group. HES data did not include time in critical care. The TMC estimated the likely time in ICU and HDU for each type of procedure during the follow-up via a survey of all the centre principal investigators (see Table 1). These estimates were used to calculate the costs of critical care for each episode in the follow-up after 1 September 2009 using published national average costs. 126

Surveillance was undertaken through attendance at vascular outpatient consultations, CT scanning and ultrasonography. These data were available for each patient in EVAR-1 over the entire period of follow-up (1999–2015) and these were costed at 2014/15 prices using national average costs. 126

Hospital notes were searched over the full trial period for incisional hernia repair procedures. These are almost exclusively carried out after OR. These reinterventions were not included in previous publications,94,101 but are included in this analysis and were costed by Hospital Resource Group at 2014/15 prices using national average costs. 126 Analysis is by ITT. Costs were converted to US dollars at purchasing power parities (£0.70 = US$1.00). 128 All data management and analysis was carried out in Stata version 13.

Results

Mean costs over 14 years estimated using inverse probability weighting

The cost results for the IPW method are shown in Table 14 and Figure 9. In the EVAR group there were 626 randomised patients, of whom 160 were alive on 30 June 2015. In the OR group, there were 626 randomised patients, of whom 181 were alive on 30 June 2015. Eleven patients in the EVAR group and 20 patients in the OR group did not receive any aneurysm repair and were assigned zero cost for the initial procedure. The overall difference in mean cost between the groups at 1 year is £2194. This difference increases gradually over 14 years to £3798 (95% CI £2338 to £5258). Most of this difference is as a result of the initial procedure cost and subsequent reinterventions.

TABLE 14 - Mean costs per patient of procedures and surveillance undertaken in EVAR-1, estimated using IPW

Note

Total mean cost may not correspond exactly to the row sum of procedures only and surveillance only because of the differences in the IPW in each case.

Year	Procedures and reinterventions (£)			Surveillance and imaging (£)			Total mean cost (£)
	OR	EVAR	Difference	OR	EVAR	Difference	OR	EVAR	Difference
1	13,725	15,660	1935	218	405	187	13,857	16,051	2194
2	168	153	–15	188	200	12	334	340	5
3	0	281	281	168	177	9	152	445	293
4	45	382	337	142	154	12	173	525	352
5	106	214	108	113	128	15	205	332	127
6	92	160	68	100	113	13	179	265	87
7	69	123	53	66	88	22	129	206	77
8	241	112	–129	49	81	32	280	188	–92
9	101	288	187	35	77	43	130	357	226
10	125	199	73	17	62	45	140	255	115
11	56	320	264	12	55	43	66	362	296
12	43	166	123	12	41	29	53	198	144
13	78	0	–78	14	26	12	88	25	–63
14	81	114	32	12	14	3	89	126	37
Total mean	14,931	18,171	3240	1147	1622	475	15,876	19,674	3798
SE			666			23			745
95% CI			1934 to 4546			430 to 521			2338 to 5258

FIGURE 9.

Evolution of mean cost (£) per patient in each treatment group over time. Mean cost over 14 years estimated using multiple imputation.

Table 15 shows the estimates from the regression of total cost on treatment group, using Rubin’s rules to aggregate coefficients over the M = 15 data sets. The estimated difference in mean cost per patient between the groups is £3757 (95% CI £2332 to £5182).

TABLE 15 - Regression of total cost on treatment group using multiple imputation to handle missing values

	Mean	SE	t	p-value	95% CI
Treat	3757	726	5	0.000	2332 to 5182
Intercept	15,891	513	31	0.000	14,884 to 16,897

Discussion

This chapter has estimated mean costs per patient for EVAR and OR over 14 years follow-up. The estimate of the difference in mean costs between the treatment groups is robust to the method used for handling missing data. The difference in cost is £3798 (95% CI £2338 to £5258) under IPW and £3757 (95% CI £2332 to £5182) under multiple imputation.

The methods make different assumptions about the missing data mechanism. The IPW method assumes data are missing completely at random, that is, the probability of an observation being missing does not depend on observed or unobserved measurements. This may be reasonable if the data are administratively censored, where the censoring arises because of the (random) date of recruitment into the trial, rather than unplanned loss to follow-up. The multiple imputation method assumes that data are missing at random, that is, given the observed data, the probability of an observation being missing does not depend on the unobserved data. The high degree of similarity between the two estimates suggests that the missing completely at random assumption is reasonable in this case.

It is perhaps unsurprising that the two methods give similar results, because the number of missing data is modest in this study. This is because there is complete follow-up for reinterventions for 11.5 years (the end of randomisation until the closure of the trial) and data for reinterventions, were obtained from HES (and verified from hospital notes and trial documents).

Introduction

This chapter presents a cost-effectiveness model of EVAR compared with OR for aneurysm repair, in patients fit for OR and suitable for EVAR. Parameters are mainly based on the outcomes of EVAR-1. Nevertheless, 24% of the OR group are alive at 15 years; hence some assumptions about the rate of mortality beyond follow-up are needed.

Methods

Results

Table 17 shows the results of the model. The base case assumes that there is no difference in aneurysm mortality beyond the end of the trial and no difference in reinterventions beyond the end of the trial. Figure 10 shows the predicted survival curves under these assumptions. The curves cross at about 4 years for any-cause mortality. EVAR is not effective or cost-effective over the lifetime under the base-case assumptions, with greater overall health benefit, measured in QALYs, but also greater cost. The incremental cost-effectiveness ratio (ICER) is £202,776 per QALY.

FIGURE 10.

Predicted survival curves from the base-case model (deterministic implementation).

Figure 11 shows the distribution of simulations from the probabilistic implementation of the base-case model. The difference in QALYs is highly skewed, with a long tail. This arises from the wide CIs around the HRs for aneurysm-related deaths. The probability that EVAR is cost-effective is 0.06 at a threshold of £20,000 per QALY and 0.19 at a threshold of £30,000 per QALY.

FIGURE 11.

Cost-effectiveness plane (base-case model, probabilistic implementation).

The first sensitivity analysis, scenario 1, assumes that the HR for aneurysm mortality observed at the end of the trial continues beyond the RCT. This makes EVAR somewhat less effective than the base case, but the results are not substantially different. The second sensitivity analysis, scenario 2, assumes that the HR for aneurysm mortality after 4 years is equal to the lower limit of the 95% CI for the mean (rather than the central value). Under this scenario, the ICER reduces to £34,026 per QALY. Under the per-protocol analysis, scenario 3, EVAR is more effective in the short term than in the base case, but there are more aneurysm-related deaths in the long term, and EVAR becomes less effective and more costly than OR over the lifetime. Scenario 4 applies the model input values used in the NICE final appraisal document. EVAR is considerably more effective than OR in this scenario and the ICER reduces to £11,088 per QALY. The probability that EVAR is cost-effective is 0.70 at £20,000 per QALY and 0.76 at £30,000 per QALY. Figure 12 shows the cost-effectiveness acceptability curve for the base-case and sensitivity analyses.

FIGURE 12.

Cost-effectiveness acceptability curves for the base-case and sensitivity analyses. Note that the cost-effectiveness acceptability curve for scenario 1 is the same as the base case.

Discussion

The base-case model uses EVAR-1 data to inform outcomes within the trial follow-up and assumes that the rates of aneurysm mortality and reinterventions are the same as OR after the end of the trial. Under this model, EVAR showed greater overall health benefit but higher costs than OR. The initial benefit after EVAR was attenuated by late mortality and reinterventions, which are associated with a decrement in HRQoL. Over the lifetime of the patient, these two factors outweigh the initial gain in operative mortality and faster recovery from surgery. Lifetime costs are greater for EVAR because of the price of the stent graft, the need for surveillance and the need for reinterventions. The ICER in the base case is > £200,000 per QALY, considerably above the threshold of £20,000–30,000 per QALY that conventionally applies in the UK.

The base-case model is based closely on EVAR-1 during the first 15 years, which found that the initial gains in operative mortality were gradually lost over the longer term and that EVAR was associated with greater costs. EVAR might be considered cost-effective in certain scenarios. Scenario 2 assumes that late AAA mortality is lower than the base case. This might be plausible, but it is selective of the evidence. Given the trial data, it is equally likely that the ‘true’ HR for aneurysm mortality could be at the upper limit of the 95% CI, rather than at the lower limit.

The results shown here are similar to earlier estimates based on the 8-year data from EVAR-1. 132 The DREAM trial, based in the Netherlands, concluded that EVAR was not cost-effective even after 1 year. 137 The OVER trial, based in the USA, found that EVAR was effective and cost-effective at 2 years,138 but this was in a lower-risk group and did not take account of late mortality or reinterventions.

The NICE appraisal of endovascular stent grafts (technology appraisal number 167) was conducted in 2009 when the 4-year results of EVAR-1 were known. The committee acknowledged the high internal validity of the EVAR trial but were concerned that the study did not properly reflect the performance of the latest generation of endovascular stents, which were thought to have improved since the trial recruitment period 1999–2004. Hence, NICE assumed more favourable outcomes for the later generation of EVAR devices than were reported in the trial. In particular, it assumed lower aneurysm-related mortality, fewer reinterventions, fewer resources used in the hospital stay and less cost for surveillance than observed in EVAR-1. Using these values in the current model, this scenario produced an ICER of £10,929 per QALY, which would be highly cost-effective at conventional thresholds used in the UK. These assumptions, as yet, are not supported by evidence from RCTs. Nevertheless, this modelling does show how EVAR would need to improve in order to be considered cost-effective. For example, EVAR might be cost-effective if ways could be found to monitor sac expansion to trigger more timely and better targeted reintervention and reduce late aneurysm deaths.

Introduction

Open repair of AAA was first introduced by Dubost et al. in 1951. 75 In the 1990s the less invasive EVAR was introduced and the first multicentre randomised trial of EVAR compared with OR was started in 1999 (EVAR-1), in the UK. 100 This was soon followed by other multicentre trials in Europe (the DREAM and ACE trials) and the OVER trial in the USA. 98,99,104

Each of the randomised trials of EVAR compared with OR recruited patients (suitable for either OR or EVAR) with slightly different entry characteristics with respect to age, sex, aneurysm morphology and other demographics. The EVAR-1, DREAM and OVER trials all showed an early survival benefit for EVAR, whereas the ACE trial did not. This early survival benefit for EVAR was lost within 1–3 years for EVAR-1 and the DREAM trial, but not until later for the OVER trial. 94,98,104 This ‘catch-up’ in mortality has been noted in many other studies, including a Cochrane review and analyses of the Medicare database,108,139 but no satisfactory explanation for this phenomenon has emerged (the Cochrane review was limited by not being able to report aneurysm-related mortality or subgroup analyses). 139 Each trial individually has been too small to investigate the reasons for this ‘catch-up’ mortality in the EVAR groups or to answer the much-discussed question of whether or not younger and fitter patients (or other subgroups) should be offered OR, which is considered more durable than EVAR. 140,141 This ‘catch-up’ in mortality needs to be avoided if EVAR is to outperform OR in the longer term. To try to address some of these issues, the four randomised trials agreed to pool their data for an IPD meta-analysis.

Methods

In July 2013, MEDLINE, EMBASE and clinical trial databases were searched for randomised trials comparing OR and EVAR of AAAs. The search terms used were AAA, aneurysm, endovascular, stent, open repair and randomised trial. From 275 reports, we identified four eligible trials reporting mid-term follow-up. The methods for these four multicentre trials included in this meta-analysis have been published previously. 94,102,105,106 EVAR-1 randomised 1252 patients (91% male), with an aneurysm diameter of > 5.5 cm, between September 1999 and August 2004 in the UK. The DREAM trial randomised 351 patients (92% male), with an aneurysm diameter of ≥ 5 cm, between November 2000 and December 2003 in the Netherlands and Belgium. The OVER trial randomised 881 patients (99% male), with an aneurysm diameter of ≥ 5.0 cm, between October 2002 and April 2008 at Veteran Affairs hospitals in the USA. The ACE trial randomised 306 patients (99% male), with an aneurysm diameter of > 5.0 cm, between March 2003 and March 2008 in France (seven patients withdrew consent before discharge). All patients were considered fit for open surgery under general anaesthesia and all trials used approved devices for EVAR, predominantly within the manufacturers’ IFU, and followed up patients for a minimum of 3 years. Summary baseline characteristics for patients by trial are shown in Table 18.

The four data sets were merged based on fields available in the case record forms of the largest trial (EVAR-1), range checks were conducted and queries resolved with the individual trial co-ordinating centres. The common baseline variables across the trials were age, sex, history of smoking, diabetes, coronary artery disease (defined as previous stable or unstable angina or MI), BMI, maximum aneurysm diameter, proximal aortic neck length and diameter, ankle–brachial pressure index (ABPI) and creatinine level, used for estimated glomerular filtration rate (eGFR),142 but without information on ethnicity. Each trial also contained some data about hypertension, which was included in a modified Wilkins cardiovascular survival risk score143 (Table 19). The reporting of both drug use (including antiplatelet and lipid-lowering agents) and reinterventions was very different in the four trials (particularly intestinal and wound-related reinterventions following OR). The postoperative surveillance protocol was identical for both randomised groups in all trials, except for the DREAM trial, in which, after 2 years, surveillance was relaxed for the OR group. However, for complications, only endoleaks after EVAR were reported similarly across the trials; laparotomy-related complications were not.

Statistical analysis

The primary analyses considered the groups ‘as randomised’ within each trial. Mortality after randomisation was assessed at 30 days, in hospital and then at three defined time periods: 0–6 months, 6 months–4 years and > 4 years after randomisation. Aneurysm-related mortality included death from (1) primary aneurysm rupture, (2) within 30 days of aneurysm repair or any reintervention; and (3) rupture after repair. Given the different times between randomisation and aneurysm repair in the four trials (Table 20), aneurysm-related mortality was also assessed at 30 days, 31 days–3 years and > 3 years after aneurysm repair. Kaplan–Meier survival curves by randomised group were generated from the combined data from all four trials and the restricted mean life-years up to a certain time estimated by the area under the curve up to that time. Logistic regression was used to compare operative (30-day) and in-hospital mortality among patients who underwent repair, and Cox proportional hazards regression was used to compare total and aneurysm-related mortality and time to reintervention. A two-stage IPD meta-analysis was performed. First analyses were conducted separately within each trial and then pooled time-period-specific estimates were calculated using random-effects meta-analysis with between-study heterogeneity estimated using the method of DerSimonian and Laird. 144 The proportion of between-trial variability beyond that expected by chance was quantified using the I²-statistic. 145 All analyses were then repeated adjusting for the following baseline covariates: age, sex, maximum aneurysm diameter and log-creatinine level.

The subgroups age, sex, eGFR, coronary artery disease, ABPI, the modified Wilkins cardiovascular survival risk score, maximum aneurysm diameter, proximal aneurysm neck diameter and neck length were assessed for differences in the effect of the EVAR and open strategies by including an interaction term between the subgroup and randomised group in a Cox regression model. Except for sex and coronary artery disease, all measures were entered as continuous variables to assess effect modification. Each interaction term was pooled across the trials using random-effects meta-analysis and its statistical significance assessed using a Wald test, taking the 5% level as significant. For presentation purposes only (and not for assessing significance), HRs are shown by dichotomising continuous measures at chosen cut-off points: age (72 years), eGFR (68.4), ABPI (0.9), cardiovascular risk score (2 major), maximum aneurysm diameter (5.9 cm), neck diameter (2.3 cm), neck length (2.5 cm), estimating the HRs within each subgroup and pooling these across studies.

The hazard of reintervention following aneurysm repair was analysed using an Anderson–Gill multiple failure time model. 146

All analyses were performed using Stata statistical software, version 13.

Results

A total of 2783 patients, with 14,245 person-years of follow-up, were included in this meta-analysis: their baseline characteristics are shown in Table 18, with significant intertrial differences in all variables. Patients in EVAR-1 were older and had larger aneurysms than patients in the other trials. Nearly all patients in the OVER and EVAR-1 trials had a history of smoking, compared with about half the patients in the DREAM and ACE trials. Nearly all patients in the OVER and EVAR-1 trials had a history of smoking, compared with about half the patients in the DREAM and ACE trials, and patients in the OVER trial had the highest BMI and highest proportion of patients with diabetes. Summary post-randomisation characteristics of the trials also are given in Table 18. The median follow-up was 6.0, 6.0, 5.4 and 3.1 years, for EVAR-1, the DREAM, OVER and ACE trials, respectively, 5.5 years for the pooled data and compliance with randomised allocation was 93% or higher in all trials.

Total mortality

Over the follow-up of all four trials there were 481 deaths in the EVAR groups and 482 in the OR groups. Kaplan–Meier curves by randomised group for total mortality across all four trials are shown in Figure 13. Overall, there was no difference in total mortality over the follow-up period of the trial (pooled HR 0.99, 95% CI 0.87 to 1.13) (Figure 14a and see Table 20). Between 0 and 6 months mortality was lower for the EVAR groups, with 46 deaths, than for OR, with 73 deaths (HR 0.61, 95% CI 0.42 to 0.89), with no evidence of heterogeneity between the trials. The early survival advantage of EVAR in the first 6 months was largely attributable to the lower 30-day operative mortality for EVAR than with OR (unadjusted pooled odds ratio 0.40, 95% CI 0.22 to 0.73) (Figure 14b). After this, the early advantage of the EVAR group was lost and the HRs moved (non-significantly) in the direction of OR. Adjusted HRs were similar. By 5 years the estimated survival was 73.6% (95% CI 71.1% to 75.9%) in both the EVAR and OR groups with an expected 0.06 additional life-years in the EVAR group, corresponding to 23 days (95% CI –16 to 61 days; p = 0.246). The causes of death by each time period are shown in Table 21.

FIGURE 13.

Kaplan–Meier survival curves for overall total mortality, by randomised group, for all four trials combined.

FIGURE 14.

(a) Total mortality, overall and at 0–6 months, 6 months–4 years and > 4 years since randomisation, unadjusted HRs; and (b) mortality within 30 days of operation, showing odds ratio.

TABLE 21 - Causes of death by randomised group and time since randomisation

Cause of death	EVAR-1, n (%)		DREAM, n (%)		OVER, n (%)		ACE, n (%)
	EVAR (n = 26)	OR (n = 45)	EVAR (n = 6)	OR (n = 10)	EVAR (n = 11)	OR (n = 17)	EVAR (n = 3)	OR (n = 1)
0–6 months
AAA related	14 (54)	30 (67)	3 (50)	10 (100)	5 (45)	14 (82)	3 (100)	1 (100)
MI/other cardiac	4 (15)	4 (9)	0 (0)	0 (0)	2 (18)	0 (0)	0 (0)	0 (0)
Stroke	0 (0)	1 (2)	0 (0)	0 (0)	1 (9)	0 (0)	0 (0)	0 (0)
Other vascular	2 (8)	2 (4)	1 (17)	0 (0)	3 (27)	0 (0)	0 (0)	0 (0)
Cancer, lung	1 (4)	0 (0)	1 (17)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
Cancer, other	2 (8)	0 (0)	0 (0)	0 (0)	0 (0)	3 (18)	0 (0)	0 (0)
Pulmonary	0 (0)	5 (11)	1 (17)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
Renal	2 (8)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
Other	1 (4)	3 (7)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
Unknown	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
	EVAR (n = 125)	OR (n = 116)	EVAR (n = 33)	OR (n = 25)	EVAR (n = 73)	OR (n = 78)	EVAR (n = 13)	OR (n = 10)
6 months–4 years
AAA related	12 (10)	8 (7)	1 (3)	1 (4)	2 (3)	0 (0)	4 (31)	0 (0)
MI/other cardiac	27 (22)	25 (22)	8 (24)	6 (24)	16 (22)	15 (19)	2 (15)	4 (40)
Stroke	11 (9)	6 (5)	1 (3)	2 (8)	3 (4)	1 (1)	1 (8)	0 (0)
Other vascular	7 (6)	6 (5)	0 (0)	3 (12)	3 (4)	4 (5)	0 (0)	0 (0)
Cancer, lung	19 (15)	20 (17)	1 (3)	7 (28)	7 (10)	12 (15)	2 (15)	2 (20)
Cancer, other	20 (16)	29 (25)	12 (36)	2 (8)	18 (25)	15 (19)	2 (15)	1 (10)
Pulmonary	9 (7)	15 (13)	2 (6)	2 (8)	4 (5)	8 (10)	1 (8)	1 (10)
Renal	4 (3)	1 (1)	0 (0)	0 (0)	0 (0)	1 (1)	1 (8)	1 (10)
Other	15 (12)	6 (5)	4 (12)	2 (8)	12 (16)	9 (12)	0 (0)	0 (0)
Unknown	1 (1)	0 (0)	4 (12)	2 (8)	8 (11)	13 (17)	0 (0)	1 (10)
	EVAR (n = 109)	OR (n = 103)	EVAR (n = 19)	OR (n = 25)	EVAR (n = 62)	OR (n = 51)	EVAR (n = 1)	OR (n = 1)
> 4 years
AAA related	10 (9)	2 (2)	2 (11)	0 (0)	3 (5)	3 (6)	0 (0)	0 (0)
MI/other cardiac	28 (26)	26 (25)	6 (32)	9 (36)	15 (24)	10 (20)	0 (0)	0 (0)
Stroke	11 (10)	11 (11)	1 (5)	2 (8)	4 (6)	2 (4)	0 (0)	0 (0)
Other vascular	8 (7)	6 (6)	0 (0)	0 (0)	5 (8)	2 (4)	0 (0)	0 (0)
Cancer, lung	9 (8)	9 (9)	0 (0)	0 (0)	7 (11)	7 (14)	0 (0)	0 (0)
Cancer, other	14 (13)	20 (19)	4 (21)	7 (28)	7 (11)	11 (22)	0 (0)	0 (0)
Pulmonary	14 (13)	17 (17)	3 (16)	2 (8)	9 (15)	11 (22)	0 (0)	0 (0)
Renal	4 (4)	2 (2)	0 (0)	1 (4)	0 (0)	0 (0)	0 (0)	0 (0)
Other	11 (10)	10 (10)	3 (16)	1 (4)	5 (8)	1 (2)	0 (0)	0 (0)
Unknown	0 (0)	0 (0)	0 (0)	3 (12)	7 (11)	4 (8)	1 (100)	1 (100)

Aneurysm-related mortality

The findings for aneurysm-related mortality were similar in direction, with relative benefit for the EVAR groups 0–6 months after randomisation (25 aneurysm-related deaths vs. 55 in the OR groups; pooled unadjusted HR 0.44, 95% CI 0.26 to 0.76). In later time periods the results move in the direction of OR, 6 months–4 years and > 4 years, with pooled HRs of 1.43 (95% CI 0.61 to 3.34) and 2.29 (95% CI 0.49 to 10.85), respectively (Figure 15). For those who received aneurysm repair, analysis by time from repair showed a strong relative advantage for the EVAR group in the first 30 days, between 30 days and 3 years there was no difference between the groups, but after 3 years there was a significant relative advantage for the OR group, with three aneurysm-related deaths compared with 19 deaths in the EVAR groups (HR 5.16, 95% CI 1.49 to 17.89; p = 0.010) (Table 22).

FIGURE 15.

Aneurysm-related mortality, overall and at 0–6 months, 6 months–4 years and > 4 years since randomisation. Unadjusted HRs.

TABLE 22 - Unadjusted and adjusted HRs for aneurysm-related mortality by time since operation for those who underwent surgery

*0.01 < p < 0.05, **0.001 < p < 0.01, ***p < 0.001.

Too few events to estimate a HR.

Adjusted for age, sex, maximum aneurysm diameter and log-creatinine level.

Aneurysm-related mortality	EVAR-1 (n = 1216) n deaths/N (rate/100 person-years)	DREAM (n = 343) n deaths/N (rate/100 person-years)	OVER (n = 868) n deaths/N (rate/100 person-years)	ACE (n = 297) n deaths/N (rate/100 person-years)	Pooled (n = 2724) n deaths/N (rate/100 person-years)
All patients
EVAR	31/614 (0.9)	6/170 (0.7)	9/439 (0.4)	7/150 (1.7)	53/1373 (0.8)
OR	32/602 (1.0)	10/173 (1.1)	13/429 (0.6)	1/147 (0.3)	56/1351 (0.8)
Time since operation
0–30 days
EVAR	11/614 (22.0)	2/170 (14.3)	1/439 (2.8)	2/150 (16.4)	16/1373 (14.2)
OR	26/602 (53.7)	5/173 (35.5)	8/429 (22.9)	1/147 (8.3)	40/1351 (36.5)
31 days–3 years
EVAR	7/603 (0.4)	2/168 (0.4)	5/438 (0.4)	4/148 (1.1)	18/1357 (0.5)
OR	4/576 (0.3)	5/168 (1.1)	4/421 (0.3)	0/146 (0.0)	13/1311 (0.4)
> 3 years
EVAR	13/498 (0.8)	2/140 (0.5)	3/380 (0.3)	1/78 (2.3)	19/1096 (0.6)
OR	2/484 (0.1)	0/146 (0.0)	1/352 (0.1)	0/72 (0.0)	3/1054 (0.1)
Unadjusted HR
All patients	0.94 (0.57 to 1.54)	0.61 (0.22 to 1.68)	0.68 (0.29 to 1.59)	6.86 (0.84 to 55.78)	0.89 (0.51 to 1.56)
Time since operation
0–30 days	0.41 (0.20 to 0.83)	0.40 (0.08 to 2.08)	0.12 (0.02 to 0.97)	1.97 (0.18 to 21.76)	0.41 (0.22 to 0.74)**
31 days–3 years	1.69 (0.50 to 5.77)	0.40 (0.08 to 2.08)	1.20 (0.32 to 4.47)	a–	1.07 (0.49 to 2.36)
> 3 years	6.35 (1.43 to 28.15)	a–	3.18 (0.33 to 30.64)	a–	5.16 (1.49 to 17.89)*
	EVAR-1 (n = 1211)	DREAM (n = 331)	OVER (n = 868)	ACE (n = 280)	Pooled (n = 2690)
Adjusted HRb
All patients	0.97 (0.59 to 1.59)	0.44 (0.15 to 1.30)	0.71 (0.30 to 1.67)	a–	0.81 (0.55 to 1.21)
Time since operation
0–30 days	0.42 (0.21 to 0.86)	0.21 (0.02 to 1.85)	0.13 (0.02 to 1.05)	a–	0.36 (0.19 to 0.67)**
31 days–3 years	1.82 (0.52 to 6.36)	0.32 (0.06 to 1.65)	1.13 (0.29 to 4.39)	a–	0.98 (0.38 to 2.55)
> 3 years	6.58 (1.48 to 29.21)	a–	3.19 (0.33 to 31.26)	a–	5.30 (1.52 to 18.46)**

Total mortality by subgroups

There was no significant effect of age or sex on the relative effectiveness of EVAR in preventing deaths in any time period, including the first 6 months following randomisation (Figure 16). There were two subgroups of patients who appeared to have no early benefit (to 6 months) under EVAR compared with OR: patients with moderate renal dysfunction and those with coronary artery disease. For those with above-median eGFR, the pooled HR of 0.42 (95% CI 0.21 to 0.84) was significantly in favour of EVAR compared with the less favourable and non-significant pooled HR of 0.68 (95% CI 0.43 to 1.08) for those with worse renal function; therefore, the interaction between eGFR measure and treatment group was significant (interaction p = 0.024) (see Figure 16).

FIGURE 16.

Unadjusted HRs for total mortality by subgroups of age, sex and eGFR: overall and at 0–6 months, 6 months–4 years and > 4 years since randomisation. Interaction p-value for age and eGFR calculated using the continuous measures (median eGFR 68.4). Not all trials contribute to the subgroup analyses or every time point.

Similarly, patients with coronary artery disease gained no early advantage of being in the EVAR group, in comparison with patients without prior coronary artery disease (interaction p = 0.047) (Figure 17). None of the morphological aneurysm characteristics, smoking, diabetes or BMI was associated with mortality (Figures 18 and 19).

FIGURE 17.

Unadjusted HRs for total mortality by subgroups of history of angina or MI, ABPI and cardiovascular risk score: overall and at 0–6 months, 6 months–4 years and > 4 years since randomisation. Interaction p-values for ABPI and cardiovascular risk score calculated using the continuous measures. Not all trials contribute to the subgroup analyses or every time point.

FIGURE 18.

Unadjusted HRs for total mortality by subgroups of maximum AAA diameter, neck diameter and neck length: overall and at 0–6 months, 6 months–4 years and > 4 years since randomisation. Interaction p-values for maximum AAA diameter, neck diameter and neck length calculated using the continuous measures. The number of trials contributing to each analysis is shown.

FIGURE 19.

Unadjusted HRs for total mortality by subgroups of history of diabetes, BMI and smoking status: overall and at 0–6 months, 6 months–4 years and > 4 years since randomisation. Interaction p-values for BMI calculated using the continuous measures. The number of trials contributing to each analysis is shown.

Baseline ABPI was not available for the ACE trial. In the other trials, patients with peripheral arterial disease (low ABPI of < 0.9) had a similar early survival advantage from being in the EVAR group as those with a ABPI of ≥ 0.9. However, in the 6 months to 4 years time period, for those with a ABPI of < 0.9, the OR group had the survival advantage (HR 1.67, 95% CI 1.12 to 2.49), in comparison with patients with a ABPI of > 0.9 (interaction p = 0.022). During the 6 months to 4 years time period, for those with a ABPI of < 0.9, total mortality was 9.6 and 5.7 per 100 person-years in the EVAR and OR groups, respectively, compared with 5.1 and 5.8 per 100 person-years, respectively, in the higher ABPI group. The cause of death in the two ABPI groups by time period is shown in Table 23. The operative mortality by subgroup shows that the highest mortality was in those with low ABPI (Table 24) and this group has higher aneurysm-related mortality throughout. Finally, a cardiovascular risk score was not discriminatory at any time period.

TABLE 23 - Causes of death by categorisation of baseline ABPI and time since randomisation

Cause of death	ABPI < 0.9, n		ABPI ≥ 0.9, n
	EVAR (n = 13)	OR (n = 23)	EVAR (n = 29)	OR (n = 43)
0–6 months
AAA related	7 (54)	18 (78)	15 (52)	30 (70)
MI/other cardiac	0 (0)	1 (4)	6 (21)	3 (7)
Stroke	0 (0)	0 (0)	1 (3)	1 (2)
Other vascular	4 (31)	2 (9)	2 (7)	0 (0)
Cancer, lung	1 (8)	0 (0)	1 (3)	0 (0)
Cancer, other	0 (0)	0 (0)	2 (7)	3 (7)
Pulmonary	0 (0)	2 (9)	0 (0)	0 (0)
Renal	1 (8)	0 (0)	1 (3)	0 (0)
Other	0 (0)	0 (0)	1 (3)	3 (7)
Unknown	0 (0)	0 (0)	0 (0)	0 (0)
	EVAR (n = 62)	OR (n = 39)	EVAR (n = 149)	OR (n = 161)
6 months–4 years
AAA related	6 (10)	3 (8)	7 (5)	6 (4)
MI/other cardiac	15 (24)	10 (26)	33 (22)	31 (19)
Stroke	3 (5)	1 (3)	12 (8)	7 (4)
Other vascular	5 (8)	3 (8)	5 (3)	5 (3)
Cancer, lung	6 (10)	5 (13)	20 (13)	30 (19)
Cancer, other	8 (13)	10 (26)	40 (27)	34 (21)
Pulmonary	6 (10)	3 (8)	7 (5)	20 (12)
Renal	1 (2)	0 (0)	3 (2)	2 (1)
Other	7 (11)	2 (5)	17 (11)	14 (9)
Unknown	5 (8)	2 (5)	5 (3)	12 (7)
	EVAR (n = 35)	OR (n = 36)	EVAR (n = 144)	OR (n = 131)
> 4 years
AAA related	3 (9)	4 (11)	12 (8)	1 (1)
MI/other cardiac	9 (26)	14 (39)	34 (24)	25 (19)
Stroke	2 (6)	4 (11)	13 (9)	10 (8)
Other vascular	4 (11)	3 (8)	8 (6)	4 (3)
Cancer, lung	0 (0)	0 (0)	14 (10)	16 (12)
Cancer, other	3 (9)	3 (8)	22 (15)	35 (27)
Pulmonary	7 (20)	3 (8)	19 (13)	23 (18)
Renal	0 (0)	2 (6)	4 (3)	1 (1)
Other	3 (9)	0 (0)	15 (10)	12 (9)
Unknown	4 (11)	3 (8)	3 (2)	4 (3)

TABLE 24 - Operative mortality by subgroup level, for individuals who underwent an operation

Subgroup	Operative mortality, n/N (%)
Age (years)
< 72	12/1311 (0.9)
≥ 72	44/1413 (3.1)
Sex
Female	6/150 (4.0)
Male	50/2574 (1.9)
eGFR
< 68.4	35/1341 (2.6)
≥ 68.4	20/1360 (1.5)
Angina/MI
No angina/MI	35/1721 (2.0)
Angina/MI	21/1003 (2.1)
ABPI
< 0.9	19/455 (4.2)
≥ 0.9	29/1813 (1.6)
Cardiovascular risk score
< 2 major	43/2052 (2.1)
≥ 2 major	13/668 (2.0)
Maximum AAA diameter (cm)
< 5.9	18/1330 (1.4)
≥ 5.9	37/1383 (2.7)
Neck diameter (cm)
< 2.3	23/1191 (1.9)
≥ 2.3	33/1521 (2.2)
Neck length (cm)
< 2.5	29/1284 (2.3)
≥ 2.5	27/1431 (1.9)
History of diabetes
No	47/2317 (2.0)
Yes	9/399 (2.3)
BMI (km/m2)
< 30	44/2057 (2.1)
≥ 30	11/649 (1.7)
Smoking status
Current	14/814 (1.7)
Ex/never	42/1906 (2.2)

Complications and reinterventions

Complications (apart from endoleaks after EVAR) and reinterventions were reported heterogeneously across the four trials. The overall rates of reinterventions reported were higher in the EVAR group than in the OR group for all trials (see Table 18). The risk of reintervention by time period following aneurysm repair is shown in Figure 20, with substantial heterogeneity between trials for reinterventions recorded between 31 days and 3 years.

FIGURE 20.

Hazard ratio for any reintervention by time period following aneurysm repair for 2718 subjects undergoing aneurysm repair (seven subjects missing follow-up for reinterventions after aneurysm repair). Note that the number of reinterventions reported here may differ slightly from those reported by the individual trial publications, particularly for the OVER trial where more reinterventions after EVAR-1 were included in their trial report. This difference is mainly attributable to the different ways in which reinterventions were categorised across the trials. For instance, amputations were included in reinterventions for the OVER trial, but placed in a separate category for all the other trials and this meta-analysis. Similarly, below knee reconstructions were included as OVER trial reinterventions, but were not included for the meta-analysis unless there was clear evidence that these procedures were aneurysm related.

There was no indication that complications following EVAR decreased with the year in which the trial commenced: these rates, together with types and numbers of complications, are reported in Table 25. The most common reported complication after EVAR was type II endoleak, which overall was reported 435 times in 325 of 2783 (12%) patients, with corrective reintervention being performed in 99 of 435 (23%) detected type II endoleaks. The second most common type of complication was type I endoleak, for which 79 of 120 (66%) patients received an early reintervention. Similarly, early correction of other serious EVAR-related complications was attempted only in less than two-thirds of cases. Secondary sac rupture was reported in 37 patients – 33 in the EVAR randomised groups (2.4% of patients) and four in the OR groups (0.3% of patients); although all four of these patients were treated with EVAR, for those with secondary rupture following treatment with EVAR, the median time to rupture was 3.5 years (Figure 21). Of these patients, 11 (30%) had a type I endoleak, of which seven were treated, seven (19%) had a type II endoleak, of which three were treated, two (5%) had a type III endoleak, of which one was treated, and nine had known graft migration. Nineteen (51%) patients had no endoleaks detected before secondary sac rupture and one further patient had a thoracic endograft for proximal aortic dissection 3 days earlier. The mean time between detection of the first endoleak and rupture was 1.8 years. The 30-day mortality rate following rupture was 62% (n = 23).

TABLE 25 - Complications by trial, focusing on EVAR-related complications

SD, standard deviation.

After 2 years only complications with reinterventions were reported.

Total number during primary admission and after discharge, allowing for patients to have recurring endoleaks. Generally, reinterventions during the primary stay were not linked to the complications of graft migration, graft thrombosis and graft infection, so that the total number of these complications receiving treatment cannot be reported.

	EVAR-1 (n = 1252)	DREAMa (n = 351)	OVER (n = 881)	ACE (n = 299)	Total (n = 2783)
Follow-up for complications, years mean (SD)	5.3 (2.5)	5.2 (2.2)	5.2 (2.1)	2.8 (1.1)	5.0 (2.4)
Rate of complications, n/person-years (rate per 100 person-years)
EVAR	315/3381 (9.3)	125/906 (13.8)	209/2334 (9.0)	103/419 (24.6)	752/7040 (10.7)
OR	27/3309 (0.8)	7/932 (0.8)	13/2276 (0.6)	8/408 (2.0)	55/6925 (0.8)
Type of complication,b n (n treated)
Type I endoleak	56 (35)	24 (17)	25 (19)	15 (8)	121 (79)
Type II endoleak	146 (39)	73 (12)	139 (38)	77 (10)	435 (99)
Type III endoleak	25 (15)	5 (1)	7 (7)	0 (0)	37 (23)
Type IV endoleak	0 (0)	0 (0)	5 (2)	0 (0)	5 (2)
Type V endoleak	0 (0)	0 (0)	11 (3)	5 (0)	16 (3)
Secondary rupture	27 (10)	2 (2)	5 (4)	3 (3)	37 (18)
Rate of secondary rupture after EVAR per 100 person-years	0.7	0.1	0.2	0.7	0.5

FIGURE 21.

Time from repair to secondary rupture, in those who were treated with EVAR.

Discussion

These four randomised trials, in Europe and the USA, provide the best evidence for the early survival advantage offered by EVAR rather than OR. Patients prefer the less invasive method of AAA repair, EVAR, which has been widely adopted, and the majority of elective repairs are now performed using EVAR. 108,120,147 This meta-analysis, over a 5-year time horizon, confirms that overall there is an early survival advantage under EVAR, which is lost within 3 years of randomisation, so that the life-years saved from EVAR over a 5-year time period are minimal. Between 0 and 6 months after randomisation, total and aneurysm-related mortality were lower for the EVAR group, mainly because of the 2.5-fold lower operative mortality in this group. However, after this time period, the early EVAR group advantage was eroded progressively. By 3 years after aneurysm repair, aneurysm-related mortality was five times higher in the EVAR group (at this stage mainly because of secondary rupture or reinterventions), and this is likely to contribute to the ‘catch-up’ in mortality.

The next investigations focused on whether the early survival advantage was either maintained or lost in subgroups of patients categorised by their preoperative characteristics. Over a 5-year time horizon, there was no convincing evidence that being randomised to EVAR or OR resulted in differential survival between any subgroups of the population. This does not support the suggestion that younger and fitter patients with aortic morphology suitable for EVAR are likely to benefit from OR over 5 years. 140 However, differential effect modification was suggested in some subgroups (renal dysfunction, coronary artery disease) in the first 6 months and for those with peripheral arterial disease in the later 6 months–4 years time period, possibly caused by frailty effects.

Low ABPI was introduced as a measure of peripheral atherosclerosis148 and is a marker of generalised atherosclerosis. 149 This subgroup had the highest pooled operative mortality, although the relative early advantage of EVAR was maintained. However, between 6 months and 4 years, fortunes reversed in favour of a survival advantage in the OR group, a pattern which indicates that those with low ABPI are another contributor to the ‘catch-up’ in mortality phenomenon.

All trials enrolled patients with evidence of moderate renal dysfunction, 35% of the overall enrolment having eGFR < 60 (chronic kidney disease stage 3 or above). For these patients, there was no evidence of a benefit from being randomised to EVAR (vs. OR), even in the first 6 months. These data are in agreement with an earlier observational study showing a high early postoperative event rate for patients with a low eGFR. 150 Whether or not renal function deteriorates more rapidly after either elective EVAR or OR of an AAA has been debated fiercely and less attention has been focused on improving the perioperative care. 151 Similarly, patients with known coronary artery disease had no evidence of an early survival advantage from being randomised to EVAR. Given that EVAR is less invasive than OR, these findings for the moderate renal dysfunction and coronary artery disease are surprising. Perhaps the stress of EVAR in these subgroups has been underestimated. It also is possible that those randomised to EVAR received less stringent preoperative evaluations, resulting in better perioperative care for the OR group with these comorbidities.

This study has several limitations that restrict its scope. First, all endografts were implanted between 1999 and 2008 using general anaesthesia, and today newer endografts are used, often implanted under local anaesthesia. Second, reporting standards for baseline characteristics (e.g. smoking, drugs) differed between trials. Third, the reporting of complications and reinterventions (aneurysm related and other cardiovascular) was very heterogeneous across the trials. Moreover, today it is recognised that type II endoleaks with sac enlargement can be dangerous111 and that a type II endoleak might even hide a type I or type III endoleak. Fourth, at the time these trials recruited (1999–2008), no trial had a clear policy for reintervention in the presence of endoleaks after EVAR and even serious complications such as type I endoleak were not always corrected quickly, which may have contributed to the increasing aneurysm-related mortality rate in the EVAR group at ≥ 3 years after aneurysm repair.

The reintervention rate was consistently higher in the EVAR groups (see Table 18), although the data are heterogeneous and the largest trial did not report incision-related complications after OR. It would be reassuring to learn that by using more recent EVAR devices within IFU, coupled with more rigorous surveillance, the continuing aneurysm-related mortality in the EVAR group could be attenuated to minimise the ‘catch-up’ in mortality. To rely entirely on the introduction of new devices to prevent aneurysm-related mortality in the EVAR group, especially without adequate surveillance, may be unwise. Recent analyses of the Medicare database support this caution. 108

In summary, this meta-analysis confirms the advantage of lower mortality in the EVAR group in the first 6 months and provides some new insight of how this early advantage of the EVAR group is eroded (aneurysm-related mortality and inclusion of those with peripheral arterial disease). Additionally, two subgroups were identified who do not have lower mortality under EVAR at any time, to suggest that these groups (moderate renal dysfunction and established coronary artery disease) may benefit from improved perioperative care, especially for EVAR. Surveillance must focus on reducing aneurysm-related deaths in the mid-term and longer term, particularly deaths resulting from reinterventions and secondary ruptures after EVAR.

Individual patient data meta-analysis

This is described in Chapter 6. In the merging of the data from the four RCTs, for complications only endoleaks after EVAR were similar across the trials.

The hazard of reintervention following aneurysm repair was analysed using a multiple failure time model. In addition, for patients who received EVAR, mortality HRs were investigated in individuals without and with a detected/treated type II endoleak.

Results on type II endoleak from the individual patient data meta-analyses

For patients who received EVAR (including patients who were randomised to OR but crossed over), mortality HRs were investigated in individuals without and with a detected/treated type II endoleak. These analyses are no longer by randomised groups and are thus observational in nature. To investigate this, a three-level time-dependent categorical variable was created for each individual based on information from their first detected type II endoleak. Specifically, follow-up for each individual was split into periods in which (1) no type II endoleak had been detected since randomisation; (2) the first type II endoleak had been detected but as yet was untreated; and (3) the first type II endoleak had been detected and treated. A Cox regression model was fitted from randomisation with this time-dependent exposure to assess the HRs of detected untreated and detected treated type II endoleaks in relation to no detected type II endoleaks.

Table 26 shows data from EVAR-1, DREAM, OVER and ACE, in terms of number of deaths and number of aneurysm-related deaths, and also pooled data. As can be seen in Table 26, there is no difference between patients with no detected type II endoleak and either untreated or treated type II endoleaks, unadjusted or adjusted for age, sex, aneurysm diameter and log-creatinine.

Overall, there was no evidence that a detected type II endoleak (either treated or untreated) was associated with worse overall survival, although results differed by trial (see Table 26). In EVAR-1, patients who had been treated for their type II endoleak had a higher hazard of mortality than patients with no detected type II endoleak, whereas patients with untreated type II endoleaks had a lower hazard. The converse was seen in the OVER trial, with the untreated type II endoleak patients having a higher mortality rate. These differences may be attributable to different decisions regarding who to treat in the two trial populations.

Further analyses from the individual patient data meta-analysis restricting to endoleaks detected within the first 3 years

Further analyses were conducted to assess whether type II endoleaks detected within the first 3 years of randomisation were associated with a higher rate of mortality. The following results are based on the EVAR-treated patients only, therefore are subject to potential confounding (non-randomised) comparisons. Data on treated and untreated type II endoleaks from the four RCTs, restricted to endoleaks detected within the first 3 years, were used to define a time-dependent variable in a Cox regression model.

Figure 22 shows survival beyond 3 years for each trial, separated by those with and without detected type II endoleak. There were no significant differences between the two groups in terms of overall survival in any of the trials. The HRs for a detected type II endoleak (within the first 3 years) for both all-cause and aneurysm-related mortality are shown in Table 27. As with the individual trial results, the pooled estimates show no evidence of a substantial increase in either all-cause or aneurysm-related mortality following a detected type II endoleak within the first 3 years, for individuals who survive beyond 3 years. However, because of the small numbers aneurysm-related mortality could be assessed only within EVAR-1.

FIGURE 22.

All-cause mortality for EVAR-operated patients surviving beyond 3 years, by whether or not a type-II endoleak is detected within the first 3 years. (a) ACE; (b) DREAM; (c) EVAR-1; and (d) OVER.

There was no overall evidence that type II endoleak (either treated or untreated) had a dramatic effect on subsequent survival.

Discussion

The pooled estimates show no evidence of a substantial increase in either all-cause or aneurysm-related mortality following a detected type II endoleak within the first 3 years, for individuals who survive beyond 3 years. However, as a result of small numbers, aneurysm-related mortality could be assessed only within EVAR-1.

As previously shown, type II endoleak as part of the ‘cluster’ of complications is associated with secondary rupture. However, this suggests that it is other complications that are listed in the cluster that are important and not type II endoleaks on their own. The ‘cluster’ did include type II endoleak with sac expansion in its definition and it seems that sac expansion is the important measure here.

Additionally, it is often difficult to be sure exactly where an endoleak is occurring. With improved experience and imaging quality, the certainty that a type II endoleak actually is one and nothing more serious should be improving. The big fear is that a type II endoleak is ‘a type I endoleak in disguise’. This is perhaps why the increase of sac diameter is such an important indicator of functionally important endoleak of one form or another.

Introduction

Following elective EVAR surgery, the aneurysm sac should gradually shrink over time as the endograft diverts blood flow away from the aneurysm. It is important, however, to continue monitoring the sac over time, through repeat imaging, and the position of the endograft to allow early detection of serious complications such as endoleaks, which can lead to secondary rupture. In the EVAR trials, the follow-up protocol suggested CT imaging of the sac and EVAR device location initially at 1, 3 and 12 months postoperatively and then annually thereafter. In this chapter we investigate whether or not variables that are easily measured on a CT scan or ultrasound, such as the diameter of the sac, are (1) associated with the onset of a complication or secondary rupture and (2) can be used to accurately predict the occurrence of the event so that an appropriate imaging protocol can be set up or a reintervention can be planned. To achieve this it is first important to model and better understand the trajectory profile of sac diameters postoperatively. A suitable model of the sac trajectory over time can then be utilised to provide predictors of outcomes, such as the current predicted sac diameter or the current rate of growth (see Figure 23 for an illustration).

FIGURE 23.

Illustration of a modelled sac trajectory (solid curve) using CT measurements taken over time (crosses). The model can be used to relate important aspects of the sac trajectory at the landmark time of prediction, such as (1) the current predicted diameter (green cross) and (2) the current rate of sac growth (green line).

Data

A total of 1656 patients were randomised in the EVAR trials prior to September 2004 (EVAR-1, n = 1252; EVAR-2, n = 404) (Figure 24). Of these patients, 847 underwent elective EVAR without conversion to OR (EVAR-1, n = 623; EVAR-2, n = 224). After excluding individuals who underwent no postoperative CT (one individual), whose sac diameter measurements were all missing (58 individuals) or who received a straight graft (four individuals), the final analysis data set consisted of 784 individuals who underwent elective EVAR surgery, irrespective of their randomisation group.

FIGURE 24.

Flow diagram showing EVAR-1 and EVAR-2 patients used in the sac growth analysis.

Of these 784 individuals, 591 (75.4%) were from EVAR-1 and 193 (24.6%) were from EVAR-2. Six hundred and ninety-nine (89.2%) patients were male and the mean age was 75 years (range 54–94 years). Further patient characteristics are presented in Table 28.

Results: pattern of sac growth over time

Table 30 shows the average number of sac diameter measurements per individual up until the landmark time of interest, and the number of individuals and complication events that occur after the landmark time contributing to the survival analysis.

TABLE 30 - Numbers of sac measurements and events contributing to the mixed model and Cox model analyses

	Landmark time
	2 years	3 years	5 years
Number of patients contributing to mixed model	784	784	784
Mean number of sac measurements prior to landmark time	2.8	3.5	4.6
Number at risk for any complication at landmark time	666	604	479
Number experiencing any complication after landmark time	136	100	50
Number at risk for cluster complications at landmark time	666	607	482
Number experiencing any cluster complication after landmark time	144	114	80
Number at risk for secondary rupture at landmark time	665	608	501
Number experiencing a fatal or non-fatal secondary rupture after landmark time	36	32	21

A plot of the repeated sac diameter measurements over time for each individual is shown in Figure 25, stratified by those who did not or did experience a rupture during follow-up. The time trend appears to be non-linear, with an initial decrease in sac diameter followed by a flattening out for those who did not experience a rupture. The time trend in the group who ruptured is similar, but with an increase in sac diameter prior to rupture for some individuals. (Note that the time of rupture may not necessarily be immediately after the last sac diameter measurement as a result of some patients ceasing regular clinical imaging.)

FIGURE 25.

Repeated sac diameter measurements against time since operation for each individual. (a) Individuals who did not experience a rupture during follow-up; and (b) individuals who had a rupture event.

After comparing all models with a linear time trend and up to two fractional polynomial terms, our final mixed-effects model contained linear, quadratic and square root powers of time with both fixed- and random-effects terms for each time component. Figure 26 shows the population-averaged time trajectory for those with average or reference baseline covariates (see Table 31 for specification of reference baseline covariates). Sac diameter initially tends to decrease sharply postoperatively, followed by a period of flattening out and then a gradual growth.

FIGURE 26.

Population-averaged trajectory of postoperative sac diameter for an EVAR-1 individual with average preoperative AAA size, fitted with a uni-iliac, Cook/Zenith graft.

Table 31 shows the estimates for predictors of sac diameter and rate of growth from the final mixed-effects model (coefficients for the function of time and variance components are given in Table 40). Baseline age and sex were not included in the final model because they were not significantly associated with postoperative sac diameter after adjusting for preoperative AAA size. Larger preoperative AAA size was associated with larger postoperative sac diameters. Different rates of sac growth were observed for different graft shapes and graft manufacturers, with higher growth rates associated with the use of uni-iliac grafts and Medtronic/Talent^® (Medtronic CardioVascular, Santa Rosa, CA, USA) and GORE^®/EXCLUDER^® (W.L. Gore and Associates, Flagstaff, AZ, USA) grafts relative to Cook^®/Zenith^® (Cook Medical, Bloomington, IN, USA) grafts.

TABLE 31 - Modelling of sac diameter measurements over time

Note

Mean differences, 95% CIs and p-values are reported from a mixed model with fixed- and random-effect terms for the intercept, linear, quadratic and square root time, and fixed-effect terms for baseline covariates and their linear time interactions.

	Change in mean sac diameter (cm) (95% CI)	p-value
Preoperative AAA size, per cm increase	0.937 (0.891 to 0.984)	< 0.001
Graft shape
Uni-iliac	Reference	0.479
Bi-iliac/bifem	0.059 (–0.104 to 0.221)
Graft manufacturer
Cook/Zenith	Reference	0.799
Medtronic/Talent	–0.016 (–0.110 to 0.079)
GORE/EXCLUDER	–0.047 (–0.231 to 0.137)
Other	–0.074 (–0.231 to 0.084)
Trial
EVAR-1	Reference	< 0.001
EVAR-2	0.194 (0.095 to 0.293)
	Change in rate of sac growth (cm/year) (95% CI)	p-value
Graft shape
Uni-iliac	Reference	0.015
Bi-iliac/bifem	–0.116 (–0.210 to –0.023)
Graft manufacturer
Cook/Zenith	Reference	< 0.001
Medtronic/Talent	0.191 (0.135 to 0.247)
GORE/EXCLUDER	0.245 (0.139 to 0.351)
Other	0.016 (–0.075 to 0.107)

Figure 27 shows fitted sac diameter trajectories at the landmark times of 2, 3 and 5 years post operation for six patients with differing outcomes. Each trajectory has been fitted using only the sac diameter measurements available up to the landmark time. Also plotted are all observed sac diameter measurements, up to September 2009, for each patient. The patient who experiences a rupture after 12 years (see Figure 27a) has initially stable, then increasing, sac diameter measurements, which is reflected in the fitted trajectories at 2, 3 and 5 years. At 5 years the fitted curve detects the increase in sac diameter and upward trajectory in the observed measurements. Patients who experience stable or decreasing measurements tend to have similar fitted trajectories at the different landmark times.

FIGURE 27.

Postoperative sac diameter measurements for six individuals and their fitted trajectories using only past measurements at 2, 3 and 5 years. (a) Rupture at 12 years post operation; (b) type II endoleak at 6 years post operation; (c) type II endoleak at 0.1 years post operation; and (d–f) no post-operation complications.

Results: associations between aspects of sac trajectory and risk of future events

Any future complication

There were 315 first-time complications that occurred in 2846 person-years of observation (crude risk of 11.1 per 100 person-years, 95% CI 9.9 to 12.4). There were 666, 604 and 479 individuals still under follow-up at the landmark forecasting times of 2, 3 and 5 years post operation, respectively. In univariate survival analyses, at 2 and 3 years following operation, there was found to be an association between the following factors and a heightened risk of developing any future complication: larger preoperative AAA size; a Medtronic/Talent or other graft; larger current sac diameter; and a larger current rate of sac growth (both predicted from the linear mixed model) (Table 32). However, after adjustment, only graft manufacturer remained statistically significant. At 5 years post operation, after adjustment, there were no variables that were convincingly associated with the risk of a future complication taking place.

TABLE 32 - Modelling survival for any complications

Univariate analyses restricted to sample used for multivariate analyses.

HRs for current rate of growth are given per approximate population standard deviation, 0.2 cm/year.

Note

HRs, 95% CIs and p-values are reported from univariate and multivariate Cox proportional hazards models.

Landmark time	Univariate model,a HR (95% CI)	Multivariate model, HR (95% CI)
2 years post operation (n = 638; number of events = 133)
Age (years)	1.024 (0.996 to 1.052)	1.016 (0.988 to 1.045)
p-value	0.093	0.265
Preoperative AAA size (cm)	1.188 (1.003 to 1.407)	1.401 (0.837 to 2.345)
p-value	0.046	0.199
Graft shape
Uni-iliac	Reference	Reference
Bi-iliac/bifem	0.719 (0.397 to 1.302)	0.900 (0.483 to 1.677)
p-value	0.276	0.740
Graft manufacturer
Cook/Zenith	Reference	Reference
Medtronic/Talent	1.437 (0.982 to 2.102)	1.126 (0.719 to 1.764)
GORE/EXCLUDER	0.814 (0.352 to 1.883)	0.540 (0.220 to 1.329)
Other	2.420 (1.445 to 4.052)	2.158 (1.284 to 3.627)
p-value	0.004	0.008
Current sac diameter (cm)	1.226 (1.073 to 1.401)	0.834 (0.494 to 1.410)
p-value	0.003	0.498
Current rate of growthb	1.229 (1.054 to 1.433)	1.422 (0.915 to 2.211)
p-value	0.008	0.118
3 years post operation (n = 583; number of events = 98)
Age (years)	1.035 (1.002 to 1.069)	1.024 (0.990 to 1.059)
p-value	0.037	0.163
Preoperative AAA size (cm)	1.234 (1.022 to 1.490)	1.107 (0.679 to 1.807)
p-value	0.029	0.683
Graft shape
Uni-iliac	Reference	Reference
Bi-iliac/bifem	1.450 (0.589 to 3.567)	1.824 (0.720 to 4.620)
p-value	0.419	0.205
Graft manufacturer
Cook/Zenith	Reference	Reference
Medtronic/Talent	1.448 (0.934 to 2.245)	1.149 (0.694 to 1.904)
GORE/EXCLUDER	0.502 (0.156 to 1.616)	0.325 (0.096 to 1.101)
Other	2.124 (1.165 to 3.871)	1.828 (0.992 to 3.368)
p-value	0.024	0.038
Current sac diameter (cm)	1.295 (1.126 to 1.490)	1.123 (0.682 to 1.851)
p-value	< 0.001	0.648
Current rate of growthb	1.262 (1.060 to 1.502)	1.192 (0.707 to 2.011)
p-value	0.009	0.510
5 years post operation (n = 467; number of events = 49)
Age (years)	1.004 (0.959 to 1.052)	0.993 (0.946 to 1.043)
p-value	0.850	0.783
Preoperative AAA size (cm)	1.240 (0.957 to 1.608)	1.371 (0.906 to 2.074)
p-value	0.104	0.135
Graft shape
Uni-iliac	Reference	Reference
Bi-iliac/bifem	0.990 (0.308 to 3.188)	1.169 (0.353 to 3.867)
p-value	0.987	0.799
Graft manufacturer
Cook/Zenith	Reference	Reference
Medtronic/Talent	1.291 (0.700 to 2.380)	0.924 (0.486 to 1.757)
GORE/EXCLUDER	0.296 (0.040 to 2.192)	0.209 (0.028 to 1.566)
Other	1.394 (0.569 to 3.412)	1.287 (0.512 to 3.232)
p-value	0.436	0.420
Current sac diameter (cm)	1.239 (1.036 to 1.481)	0.939 (0.633 to 1.393)
p-value	0.019	0.756
Current rate of growthb	1.317 (1.024 to 1.695)	1.504 (0.919 to 2.463)
p-value	0.032	0.105

Cluster complications

There were 288 first-time cluster complications that occurred in 3457 person-years of observation (crude risk of 8.3 per 100 person years, 95% CI 7.4 to 9.4). Stronger associations were observed with cluster complications (Table 33). At 2 and 3 years post operation, after adjustment for other potential confounders, graft manufacturer remained significantly associated with future complications, with Gore/Excluder grafts showing a fourfold lower risk than Cook/Zenith grafts (HR at 3 years = 0.26, 95% CI 0.08 to 0.86) and ‘other’ manufacturers estimated to have nearly double the risk as the Cook/Zenith grafts (HR at 3 years = 1.93, 95% CI 1.08 to 3.46). No other variables significantly predicted cluster complications at these time periods.

TABLE 33 - Modelling survival for cluster complications

Univariate analyses restricted to sample used for multivariate analyses.

HRs for current rate of growth are given per approximate population standard deviation, 0.2 cm/year.

Note

HRs, 95% CIs and p-values are reported from univariate and multivariate Cox proportional hazards models.

Landmark time	Univariate model,a HR (95% CI)	Multivariate model, HR (95% CI)
2 years post operation (n = 639; number of events = 137)
Age (years)	1.023 (0.996 to 1.051)	1.013 (0.986 to 1.042)
p-value	0.096	0.342
Preoperative AAA size (cm)	1.285 (1.090 to 1.516)	1.416 (0.854 to 2.347)
p-value	0.003	0.178
Graft shape
Uni-iliac	Reference	Reference
Bi-iliac/bifem	0.912 (0.479 to 1.736)	1.161 (0.595 to 2.266)
p-value	0.778	0.661
Graft manufacturer
Cook/Zenith	Reference	Reference
Medtronic/Talent	1.454 (0.999 to 2.117)	1.085 (0.698 to 1.686)
GORE/EXCLUDER	0.910 (0.394 to 2.103)	0.549 (0.224 to 1.349)
Other	2.711 (1.637 to 4.489)	2.450 (1.474 to 4.072)
p-value	0.001	0.001
Current sac diameter (cm)	1.350 (1.184 to 1.540)	0.916 (0.549 to 1.529)
p-value	< 0.001	0.737
Current rate of growthb	1.313 (1.126 to 1.532)	1.451 (0.938 to 2.245)
p-value	0.001	0.094
3 years post operation (n = 587; number of events = 108)
Age (years)	1.025 (0.994 to 1.058)	1.008 (0.976 to 1.041)
p-value	0.117	0.648
Preoperative AAA size (cm)	1.315 (1.098 to 1.575)	1.232 (0.789 to 1.922)
p-value	0.003	0.359
Graft shape
Uni-iliac	Reference	Reference
Bi-iliac/bifem	1.610 (0.656 to 3.953)	2.216 (0.879 to 5.585)
p-value	0.298	0.091
Graft manufacturer
Cook/Zenith	Reference	Reference
Medtronic/Talent	1.325 (0.872 to 2.014)	0.852 (0.532 to 1.365)
GORE/EXCLUDER	0.519 (0.162 to 1.663)	0.256 (0.077 to 0.855)
Other	2.270 (1.276 to 4.035)	1.929 (1.076 to 3.459)
p-value	0.016	0.008
Current sac diameter (cm)	1.452 (1.268 to 1.663)	1.102 (0.699 to 1.738)
p-value	< 0.001	0.676
Current rate of growthb	1.439 (1.213 to 1.709)	1.493 (0.932 to 2.392)
p-value	< 0.001	< 0.001
5 years post operation (n = 470; number of events = 75)
Age (years)	1.014 (0.976 to 1.053)	0.987 (0.949 to 1.027)
p-value	0.481	0.523
Preoperative AAA size (cm)	1.399 (1.132 to 1.729)	1.012 (0.710 to 1.443)
p-value	0.002	0.948
Graft shape
Uni-iliac	Reference	Reference
Bi-iliac/bifem	1.210 (0.442 to 3.315)	1.482 (0.531 to 4.137)
p-value	0.711	0.453
Graft manufacturer
Cook/Zenith	Reference	Reference
Medtronic/Talent	1.183 (0.720 to 1.944)	0.709 (0.426 to 1.181)
Gore/Excluder	0.252 (0.035 to 1.839)	0.133 (0.018 to 0.977)
Other	1.583 (0.765 to 3.273)	1.334 (0.637 to 2.792)
p-value	0.279	0.083
Current sac diameter (cm)	1.629 (1.405 to 1.888)	1.586 (1.116 to 2.252)
p-value	< 0.001	0.010
Current rate of growthb	1.697 (1.389 to 2.072)	1.151 (0.774 to 1.712)
p-value	< 0.001	0.488

At 5 years post operation, the graft manufacturer was no longer associated with the onset of future cluster complications, whereas there was some evidence that a larger sac diameter at 5 years was an important predictor. After adjusting for current sac diameter, the rate of sac growth was no longer found to be an independent risk factor, as sac diameter and rate of growth were highly correlated by 5 years (ρ = 0.76) (Figure 28).

FIGURE 28.

Predicted current rate of change against predicted current sac diameter using past measurements at 2, 3 and 5 years. Prediction at (a) 2 years; (b) 3 years; and (c) 5 years.

Secondary rupture

There were 41 secondary rupture events that occurred in 5666 person-years of observation [crude rupture risk of 0.72 per 100-person years, 95% CI 0.53 to 0.98). Survival free of secondary rupture is shown in Figure 29. Owing to the small number of ruptures, only current sac diameter and rate of growth were investigated as potential predictors of rupture. Both were strongly associated with rupture events at all landmark times in univariate analyses, but only (modelled) rate of sac growth remained an independent risk factor when adjusting for each other (Table 34). For a 0.2 cm per year increase in growth rate the risk of secondary rupture increases substantially (HR at 5 years = 2.61, 95% CI 1.49 to 4.57), a finding that is consistent across follow-up.

FIGURE 29.

Survival free of death from secondary rupture for 785 EVAR-operated patients in EVAR-1 and EVAR-2.

TABLE 34 - Modelling survival for rupture events

HRs for current rate of growth are given per approximate population standard deviation, 0.2 cm/year.

Note

HRs, 95% CIs and p-values are reported from univariate and multivariate Cox proportional hazards models.

Landmark time		Univariate model, HR (95% CI)	Multivariate model, HR (95% CI)
2 years post operation (n = 636; number of events = 36)	Current sac diameter (cm)	1.673 (1.303 to 2.148)	1.160 (0.820 to 1.640)
	p-value	< 0.001	0.402
	Current rate of growtha	2.622 (1.847 to 3.723)	2.372 (1.559 to 3.611)
	p-value	< 0.001	< 0.001
3 years post operation (n = 586; number of events = 32)	Current sac diameter (cm)	2.033 (1.597 to 2.588)	1.270 (0.883 to 1.828)
	p-value	< 0.001	0.197
	Current rate of growtha	3.628 (2.550 to 5.163)	2.993 (1.897 to 4.722)
	p-value	< 0.001	< 0.001
5 years post operation (n = 486; number of events = 21)	Current sac diameter (cm)	2.407 (1.821 to 3.181)	1.516 (0.989 to 2.324)
	p-value	< 0.001	0.056
	Current rate of growtha	3.888 (2.625 to 5.758)	2.609 (1.488 to 4.574)
	p-value	< 0.001	0.001

The EVAR trial 2 patients and their follow-up

Between 1 September 1999 and 31 August 2004, 404 patients were recruited to participate in EVAR-2. A total of 197 patients were randomly assigned to the endovascular group and 207 were assigned to the no-intervention group. There were no significant differences between the two study groups with respect to baseline characteristics, mean age was 76.8 years and 347 (86%) were men (Table 41), but these patients were older and more physically frail than patients randomised to EVAR-1. 95

Patients were followed for mortality until 30 June 2015 (mean 12.3 years, median 12.2 years); mean follow-up to either death or end of the study was 4.2 years. Mortality was from regular reports from NHS Digital, which were assessed by the Endpoint Committee in a consistent manner for all deaths reported between September 2009 and June 2015. Out of 98 patients still under follow-up in September 2009, 45 (46%) had local follow-up at the trial centre. In contrast to follow-up of EVAR-1 patients, the EVAR-2 patients were not tracked using the administrative database of HES from NHS Digital. The Consolidated Standards of Reporting Trials diagram is shown in Figure 30.

FIGURE 30.

Consolidated Standards of Reporting Trials diagram: EVAR-2.

Long-term survival of patients randomised in the EVAR trial 2

After the 5- and 10-year results of EVAR-2 were known,95,107 it has become increasingly recognised that there is a cohort of patients who have such a limited life expectancy or such extensive comorbidities that repair of an AAA should not be considered. 161,162 The difficulty is in defining carefully this group of patients. In EVAR-2, the participating clinicians appeared to have had good judgement in selecting patients for this trial: by 4 years < 40% of patients remained alive and by 8 years this had reduced to < 20%. 95 Nevertheless, the approximately 20% of patients who remain alive after 8 years were perhaps physically the fittest of all the patients enrolled in EVAR-2 and it is of interest to compare their very long-term survival both by ITT and per-protocol analysis. This starts to enable description of the patients who, although physically frail and ineligible for OR, may have many life-years ahead and might benefit from EVAR, particularly if conducted under local anaesthesia. Such information could be important for updating the guidelines for the management of physically frail patients with large aneurysms.

Aneurysm-related and total mortality

During 1713 person-years of follow-up, 381 deaths occurred and 84 of which were aneurysm related. Overall aneurysm-related mortality was 3.2 deaths per 100 person-years in the EVAR group and 6.5 deaths per 100 person-years in the no-intervention group (adjusted HR 0.55, 95% CI 0.34 to 0.90; p = 0.018). Beyond 8 years of follow-up, the rate of aneurysm-related mortality was very low, with two aneurysm-related deaths in the EVAR group (at 8.0 and 13.2 years after randomisation) and one aneurysm-related death in the no-intervention group (at 9.3 years after randomisation), with survival curves remaining parallel (Figure 31). However, at 8 years only 10 of 38 patients in the no-intervention group remained with an intact unrepaired aneurysm compared with 0 of 32 patients in the EVAR group.

FIGURE 31.

Kaplan–Meier estimates for total survival and aneurysm-related survival up to 12 years of follow-up. Twelve-year survival probabilities are reported in the key.

For total mortality, there were 22.4 deaths per 100 person-years in the EVAR group and 22.1 deaths per 100 person-years in the no-intervention group. Beyond 8 years the number of deaths and mortality rate were similar in the two randomised groups. By 12 years of follow-up the estimated survival was 5.7% (95% CI 3.05 to 9.7%) in the EVAR group and 8.5% (95% CI% 5.2 to 12.9%) in the no-intervention group (see Figure 31). There was no significant difference in life expectancy (restricted to 12 years of follow-up), between the groups (4.2 years in both the EVAR and the no-intervention groups; p = 0.99).

Deaths according to time since randomisation are given in Table 42; deaths from any cause and aneurysm-related causes according to time since randomisation are given in Table 43; and the full causes of death by time period are given in Table 44. Overall, there were 16 deaths from aneurysm rupture in the EVAR group (13 procedure related and three secondary ruptures) and 53 in the no-intervention group (52 primary and one secondary rupture).

Aneurysm-related reinterventions

During 1663 person-years of follow-up, 54 graft-related reinterventions were performed in 38 patients in the EVAR group and 21 graft-related reinterventions were performed in 15 patients in the no-intervention group (Table 45). HRs for risk of a reintervention are shown in Table 46 by time epoch. There was a significantly higher rate of reinterventions in the EVAR group in the 6 months following randomisation. Beyond 4 years the rate of reintervention was similar between the two groups, as a result of a decreasing rate in the EVAR group and a slightly increasing rate in the no-intervention group as surviving individuals among those who underwent aneurysm repair.

Further analysis was carried out to compare baseline characteristics between those who did not survive > 8 years and the 70 (17%) patients who did (Table 47). The patients who survived ≥ 8 years were younger, with higher BMI, lower creatinine levels and better lung function than patients who survived < 8 years. There was also a small difference in aneurysm diameter, with slightly smaller aneurysms in the long-term survivors.

Discussion of EVAR trial 2

Findings from these EVAR-2 results have shown that the benefit of EVAR in terms of aneurysm-related mortality persists throughout follow-up. However, although some patients (< 10%) survive to 12 years, either with or without aneurysm repair, the majority of EVAR-2 patients had a limited life expectancy, and hence at no time does aneurysm repair confer an overall survival benefit. Given the rather limited clinical and imaging follow-up of these patients beyond 8 years, aneurysm-related mortality could have been underestimated during the final few years of follow-up and, therefore, is a less reliable outcome for EVAR-2 than for EVAR-1 and the magnitude of the difference in aneurysm-related mortality appears to be unchanged between 8 and 12 years after randomisation. Our earlier interpretation, that it is advisable to take time to optimise patient fitness before considering EVAR, continues to hold up. By the end of patient follow-up at least 71 of 207 (34%) patients assigned to no intervention had undergone aneurysm repair. Some patients had their aneurysm repair many years after randomisation and over 12 years of follow-up there was no difference in either survival or average life-years between the randomised groups.

Inspection of the baseline characteristics of the long-term survivors suggested ‘the survival of the fittest’ phenomenon. The long-term survivors had notably better baseline lung function than the remainder, which might have impacted on the ability of the long-term survivors to cope well with general anaesthesia. During the randomisation period, many of the EVAR procedures are likely to have been conducted under general anaesthesia, although today there is more widespread use of local anaesthesia in physically frail patients. The differences in creatinine levels may be of less significance, as with their younger age and higher BMI, the glomerular filtration rates might be quite similar for the long-term survivors and the remainder. Aneurysm diameter is an accepted indicator of the risk of cardiovascular mortality, so that the patients with lower baseline aneurysm diameter are likely to have been at lower risk of cardiovascular death. 163–165

Although longer follow-up shows benefits in terms of aneurysm-related mortality, primarily through prevention of late aneurysm rupture, patients in this trial had a limited life expectancy, regardless of whether the aneurysm was repaired or no intervention was performed. Life expectancy was estimated to be 4.2 years in both the EVAR and no-intervention groups. By 1 September 2009, only 98 patients had survived, and by 30 June 2015 only 23 patients were still alive. The clinical factors leading to the assessment that these patients were physically ineligible for OR seem likely to have contributed to a high subsequent rate of death from any cause. This rate was not influenced by assignment to EVAR.

Beyond 8 years of follow-up, there were no reinterventions recorded in the EVAR group and just three reinterventions in the group randomised to no repair. By September 2009 there were so few EVAR-2 patients surviving that, given the need to provide robust justification to obtain ethics approval for using administrative data, HES data were not pursued for EVAR-2 patients. In hindsight, this might have been a mistake. HES data may have allowed us to capture further primary aneurysm repairs as well as more reinterventions, including those carried out at non-trial hospitals, and better define the cause of late deaths. However, the main reason for the low number of reinterventions is the attrition as a result of high mortality in EVAR-2, leaving less time for graft-related complications to develop. Throughout the trial, one might have expected a lower reintervention rate than in EVAR-1, as EVAR-2 patients were frailer and less fit. However, up to September 2009 patient fitness did not seem to have influenced the decision to intervene, with a similar reintervention rate in both EVAR groups of EVAR-1 and EVAR-2. In late follow-up from September 2009, in those patients that underwent endovascular repair (including a considerable number of patients in the no-intervention group that had aneurysm repair), it is likely that reinterventions, even when deemed necessary, may not have been performed because of physical frailty.

Like EVAR-1, EVAR-2 used principally early-generation endografts; later iterations of grafts would now be the commonly used. The long-term durability of these later iterations of endografts has not been evaluated, but it is hoped that they would be associated with lower rates of complications and reinterventions.

It is perhaps unsurprising that long-term survival is associated with younger age, smaller aneurysm and better lung function at baseline. It is thus likely that future selection for EVAR in the relatively frail might likely be influenced by age, size of aneurysm and respiratory status.

Suggestions of possible implications on practice and local health-care delivery

EVAR trial 1 was performed in the UK as a collaborative venture between the two disciplines of vascular surgery and interventional radiology, which worked well for elective repair of the abdominal aorta by endovascular methods. Recruitment was ahead of schedule both because of this high-level national support for the EVAR trials and also because Sir Miles Irving linked NIHR funding of EVAR to subventions which were triggered only by randomisation with local freedom of choice of EVAR manufacturer device.

The results have been segmented into four periods: 0–6 months, 6 months–years, 4–8 years and > 8 years. From a benefit in mortality in favour of the EVAR group up to 6 months, beyond 4 years this was counterbalanced by an increase in aneurysm-related mortality, the difference being most marked after 8 years. In terms of total mortality, there was benefit in favour of EVAR during the first 6 months. Groups were similar between 6 months and 8 years, but after 8 years total mortality was significantly worse in the EVAR group than in the OR group.

Reinterventions were greater in the EVAR group throughout EVAR-1 in all periods, and there was never a time when there was evidence that the interventions were expected to be so low that follow-up surveillance could stop.

Despite warnings to the trial centres to recommend ongoing annual clinical follow-up and imaging, this fell away. Even though many centres switched from CT to ultrasound, this did not stop the reduction of enthusiasm to follow patients at this late stage.

In EVAR-2 after 10 years there was poor clinical follow-up and we did not use HES to track reinterventions. This was because the number of patients alive at that stage was very small. However, we had robust data on mortality and our evidence shows superior aneurysm-related mortality in favour of EVAR increasing with time, but there was no difference in the groups in terms of all-cause mortality.

The IPD meta-analysis confirms the advantage of lower mortality in the EVAR group in the first 6 months and provides some new insight of how this early advantage of the EVAR group is eroded (aneurysm-related mortality and inclusion of those with peripheral arterial disease). Additionally, we identified two subgroups in which mortality was not lower at any time, suggesting that these groups (patients with moderate renal dysfunction and established coronary artery disease) may benefit from improved perioperative care, especially if undergoing EVAR. Surveillance must focus on reducing aneurysm-related deaths in the mid-term and longer term, particularly deaths resulting from reinterventions and secondary ruptures after EVAR.

In the IPD meta-analysis there was no overall evidence that type II endoleak in itself is associated with a higher rate of mortality, although, as previously shown, type II endoleak as part of the ‘cluster’ of complications is associated with secondary rupture. However, this suggests that it is other complications that are listed in the ‘cluster’ that are sinister and not type II endoleaks alone. The ‘cluster’ included definition of the type II endoleak with sac expansion and it seems that sac expansion is the important factor here. The findings here on type II endoleak might suggest that the enthusiasm to correct type II endoleak when on its own is misplaced. On the other hand, when type II endoleak is associated the so-called ‘cluster’ it is far from a benign condition. In addition, type II endoleak can be wrongly diagnosed and a more severe endoleak, such as types IA or IB, can be at the root of the problem.

The mean difference in costs over 14 years was £3798 (95% CI of £2338 to £5258). Decision modelling based on EVAR-1 showed that the lifetime difference in cost was £3616 and the difference in QALYs was 0.018, with a cost per QALY of £202,776. The cost per QALY exceeds conventional thresholds used in the UK. The DREAM and ACE trials (conducted in the Netherlands and France, respectively) found similar results. If EVAR is to be considered a cost-effective use of NHS resources, it needs to demonstrate fewer reinterventions and fewer late aneurysm deaths than were observed in the EVAR trial.

Patterns of sac growth, if modelled correctly, can be useful predictors for future serious rupture events. Estimates of sac expansion should be derived from longitudinal modelling with a sac diameter trajectory, which accounts for all previous measurements and thus produces more accurate estimates of current sac growth. Naive estimates of growth based on the last two observed diameter measurements are too crude to be of critical use in a prediction model. The current sac diameter was not found to be an independent risk factor of rupture after accounting for the growth rate of the sac and the predictive accuracy of the model just using sac diameter was lower than a model that used sac growth.

Recommendations for future research

To pursue annual follow-up with appropriate imaging is the first priority, although a simple switch from annual CT scanning (with associated radiation exposure) to ultrasound is in itself not enough for follow-up to be continued adequately.

As EVAR is more costly over the patient’s lifetime, an earlier discharge from secondary care to simple follow-up in primary care might be a worthwhile aim in the future. In order for EVAR to be considered effective and cost-effective, an area of further research is to find better ways to target reintervention of patients who are at risk of secondary rupture and avoid reintervention in patients at very low risk. Our early findings suggest that an algorithm could be developed based on annual measurements of aortic sac diameter only. This might have excellent predictive value for future rupture. If effective, it would need substantial validation on a separate cohort of patients.

The long-term findings of EVAR-2 present an ethical problem for consideration. Our evidence is that endovascular repair does not alter mortality from all causes in any way, but EVAR can lower aneurysm-related mortality. It is entirely possible that greater enthusiasm to perform endovascular repair may re-emerge and a view on the attitudes to this will become important.

The determination of which frail patients merit aortic repair needs further determination. We see early encouragement that such a cohort could be defined. It would need to be clinically tested. This may be the only realistic way that EVAR could become effective and cost-effective.

Contributions of authors

Rajesh Patel drafted and compiled the report as trial manager, referring always to the other authors.

Janet T Powell critically reviewed the manuscript and added value at every stage.

Michael J Sweeting drafted all statistical methods as he performed all statistical analyses. He also drafted Chapter 8 with Jessica K Barrett and critically reviewed the manuscript.

David M Epstein performed cost and cost-effectiveness analyses, and drafted Chapters 4 and 5.

Jessica K Barrett performed analysis for Chapter 8 and co-drafted this chapter with Michael J Sweeting.

Roger M Greenhalgh had overall responsibility, co-ordination and supervision, with special responsibility for clinical issues, and he edited the report.

Publications

Brown LC, Epstein D, Manca A, Beard JD, Powell JT, Greenhalgh RM. The UK Endovascular Aneurysm Repair (EVAR) trials: design, methodology and progress. Eur J Vasc Endovasc Surg 2004;27:372–81.

EVAR Trial Participants. Comparison of endovascular aneurysm repair with open repair in patients with abdominal aortic aneurysm (EVAR trial 1), 30-day operative mortality results: randomised controlled trial. Lancet 2004;364:843–8.

EVAR Trial Participants. Endovascular aneurysm repair versus open repair in patients with abdominal aortic aneurysm (EVAR trial 1): randomised controlled trial. Lancet 2005;365:2179–86.

EVAR Trial Participants. Endovascular aneurysm repair and outcome in patients unfit for open repair of abdominal aortic aneurysm (EVAR trial 2): randomised controlled trial. Lancet 2005;365:2187–92.

Greenhalgh RM, Brown LC, Powell JT. High risk and unfit for open repair are not the same. Eur J Vasc Endovasc Surg 2007;34:154–5.

Brown LC, Greenhalgh RM, Howell S, Powell JT, Thompson SG. Patient fitness and survival after abdominal aortic aneurysm repair in patients from the UK EVAR trials. Br J Surg 2007;94:709–16.

Brown LC, Greenhalgh RM, Kwong GP, Powell JT, Thompson SG, Wyatt MG. Secondary interventions and mortality following endovascular aortic aneurysm repair: device-specific results from the UK EVAR trials. Eur J Vasc Endovasc Surg 2007;34:281–90.

Powell JT, Brown LC, Greenhalgh RM, Thompson SG. The rupture rate of large abdominal aortic aneurysms: is this modified by anatomical suitability for endovascular repair? Ann Surg 2008;247:173–9.

Epstein DM, Sculpher MJ, Manca A, Michaels J, Thompson SG, Brown LC, et al. Modelling the long-term cost-effectiveness of endovascular or open repair for abdominal aortic aneurysm. Br J Surg 2008;95:183–90.

Rodway AD, Powell JT, Brown LC, Greenhalgh RM. Do abdominal aortic aneurysm necks increase in size faster after endovascular than open repair? Eur J Vasc Endovasc Surg 2008;35:685–93.

Brown LC, Brown EA, Greenhalgh RM, Powell JT, Thompson SG, on behalf of the UK EVAR Trial Participants. Renal function and abdominal aortic aneurysm: the impact of different management strategies on long-term renal function in the UK EndoVascular Aneurysm Repair (EVAR) Trials. Ann Surg 2010;251:966–75.

Brown LC, Greenhalgh RM, Thompson SG, Powell JT. Does EVAR alter the rate of cardiovascular events in patients with abdominal aortic aneurysm considered unfit for open repair? Results from the randomised EVAR Trial 2. Eur J Vasc Endovasc Surg 2010;39:396–402.

Brown LC, Greenhalgh RM, Powell JT, Thompson SG. Use of baseline factors to predict complications and reinterventions after endovascular repair of abdominal aortic aneurysm. Br J Surg 2010;97:1207–17.

The UK EVAR Trial Investigators. Endovascular repair of aortic aneurysm in patients physically ineligible for open repair. N Engl J Med 2010;362:1872–80.

The UK EVAR Trial Investigators. Endovascular versus open repair of abdominal aortic aneurysm. N Engl J Med 2010;362:1863–71.

Wyss TR, Brown LC, Powell JT, Greenhalgh RM. Rate and predictability of graft rupture after endovascular and open abdominal aortic aneurysm repair: data from the EVAR Trials. Ann Surg 2010;252:805–12.

Brown LC, Thompson SG, Greenhalgh RM, Powell JT, on behalf of the EVAR Trial participants. Incidence of cardiovascular events and death after open or endovascular repair of abdominal aortic aneurysm: results from the randomised EVAR trial 1. Br J Surg 2011;98:935–42.

Epstein D, Sculpher MJ, Powell JT, Thompson SG, Brown LC, Greenhalgh RM. Long-term cost-effectiveness analysis of endovascular versus open repair for abdominal aortic aneurysm based on four randomized clinical trials. Br J Surg 2014;101:623–31.

Brown LC, Powell JT, Thompson SG, Epstein DM, Sculpher MJ, Greenhalgh RM. The UK EndoVascular Aneurysm Repair (EVAR) trials: randomised trials of EVAR versus standard therapy. Health Technol Assess 2012;16(9).

Wyss TR, Dick F, Brown LC, Greenhalgh RM. The influence of thrombus, calcification, angulation, and tortuosity of attachment sites on the time to the first graft-related complication after endovascular aneurysm repair. J Vasc Surg 2011;54:965–71.

Patel R, Sweeting MJ, Powell JT, Greenhalgh RM, for the EVAR Trial Investigators. Endovascular versus open repair of abdominal aortic aneurysm in 15-years follow-up of the UK endovascular aneurysm repair trial 1 (EVAR trial 1): a randomised controlled trial. Lancet 2016;388:2366–74.

Powell JT, Sweeting MJ, Ulug P, Blankensteijn JD, Lederle FA, Becquemin JP, Greenhalgh RM. Endovascular or open repair for abdominal aortic aneurysm over 5 years: Individual patient data meta-analysis from EVAR 1 DREAM, OVER and ACE trials. Br J Surg 2017;104:166–78.

Sweeting MJ, Patel R, Powell JT, Greenhalgh RM. Endovascular repair of abdominal aortic aneurysm in patients physically ineligible for open repair: very long-term follow-up in the EVAR-2 randomized controlled trial. Ann Surg 2017;266:713–19.

IMPROVE Trial Investigators. Comparative clinical effectiveness and cost effectiveness of endovascular strategy vs. open repair for ruptured abdominal aortic aneurysm: three year results of the IMPROVE randomised trial. BMJ 2017;359:4859.

Data sharing statement

Any requests for data should be sent to the corresponding author, Professor Roger Greenhalgh.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.