The flow of patients entering the trial is shown in our CONSORT diagram, with 1082 of the 1423 who were eligible by December 2003 consenting to randomisation into EVAR trial 1 (76%). Patients who refused randomisation (n=341, mean age of 74 years (SD 7), 89% male) had a median aneurysm diameter of 6.4cm [IQR 5.9-7.0], which was slightly greater than randomised patients, Mann-Whitney p=0.02. Patients who refused had a preference for EVAR (106), open repair (203), no intervention (26), or had unknown preference (6) [refusal summary]. Baseline characteristics of the two randomised groups were very similar, with mean age 74 years (SD 6), 91% males, and median aneurysm diameter 6.2 cm [IQR 5.8-7.0]. The median time from randomisation to surgery was 43 {IQR 28-70} days for the EVAR group and 36 {IQR 20-59} days for the open repair group, and by December 2004 the median follow-up was 2.9 years {IQR 1.9-4.0} with 6 patients lost to follow-up. Over 99% of endovascular repairs used commercially available devices; 51% Zenith (Cook, Copenhagen, Denmark), 33% Talent (Medtronic, Minneapolis, USA), 7% Excluder (Gore, Flagstaff, Arizona, USA), 4% AneuRx (Medtronic), 2% Quantum or Teramed (Cordis, Waterloo, Belgium). 90% of these grafts were bifurcated and the remainder were aorto-uni-iliac.

30-day operative mortality
An ‘intention to treat’ analysis showed that 30-day mortality and in-hospital mortality were two-thirds lower in the EVAR group and adjustment for baseline covariates did not alter the benefit of EVAR [30-day mortality summary]. There were 9 deaths within 30 days in the EVAR group, including one after emergency open repair for AAA rupture and two from AAA rupture after repair; two further in-hospital deaths occurred, including one after emergency open repair of AAA rupture. There were 25 deaths within 30 days in the open repair group, including one after emergency open repair for AAA rupture and one from AAA rupture after open repair; 8 further in-hospital deaths occurred after open repair, including one after emergency repair for AAA rupture.

All-cause and aneurysm-related mortality
Due to staggered recruitment, the proportion of patients who had the potential, by 31st December 2004, to be followed up for at least at 1, 2, 3 and 4 years was 100%, 70%, 47% and 24% respectively. There were 209 deaths, with 53 from aneurysm-related causes [causes of death table] and 68 from remaining cardiovascular disorders. All-cause mortality at four years after randomisation was similar in the two groups (around 28%) and the survival curves to 4 years showed no difference between the groups [Kaplan-Meier curves], hazard ratio 0.90 [95% CI 0.69-1.18], p=0.46. However, there was a persistent difference in aneurysm-related mortality, 4% for the EVAR group and 7% for the open repair group by four years [Kaplan-Meier curves], hazard ratio 0.55 [95% CI 0.31-0.96], p=0.04. The adjusted and crude hazard ratios were similar. There were no significant interactions, for either all-cause or aneurysm-related mortality, with age, sex, aneurysm diameter or creatinine (all p>0.2).

In a post hoc analysis, the follow-up was divided into the first 6 months after randomisation and the period after 6 months. The hazard ratios for aneurysm related mortality comparing the EVAR and open repair groups were 0.42 (95% CI 0.21 to 0.82) and 1.15 (95% CI 0.39 to 3.41) in the two time periods respectively, the latter having a wide confidence interval because of the few deaths included. The corresponding hazard ratios for total mortality were 0.55 (95% CI 0.33 to 0.93) in the first 6 months and 1.10 (95% CI 0.80 to 1.52) after 6 months. These results suggest that the initial benefit of EVAR is possibly followed by a longer term disadvantage.

Durability and secondary interventions
Analysis was conducted by intention-to-treat for the time from randomisation to first complication (Kaplan-Meier curves) and to first re-intervention (Kaplan-Meier curves). During the first 4 years of follow-up, the overall rates of complications and re-interventions appeared to diverge between groups. By 4 years, the proportion of patients with at least one complication following AAA repair was 41% in the EVAR group, compared to 9% in the open repair group. Overall rates of complications were 17.6 per 100 person years in the EVAR group and 3.3 per 100 person years in the open repair group, hazard ratio 4.9 [95% CI 3.5-6.8], p<0.0001. Similarly, the proportion of patients with at least one re-intervention by 4 years was 20% in the EVAR group and 6% in the open repair group. The rate of at least one re-intervention was 6.9 per 100 person years in the EVAR group and 2.4 per 100 person years in the open repair group, hazard ratio for at least one intervention 2.7 [95% CI 1.8-4.1], p<0.0001.

The types of post operative complications and number of re-interventions that occurred following EVAR and open repair according to treatment received are shown. By December 2004, 186 (35%) of all patients receiving EVAR reported one or more post operative complications of whom 81 (44%) required a secondary intervention, 19 of these during the primary hospital admission. Among the remaining 62 re-admissions, 2 patients (both presenting with graft rupture) died within 30 days of their secondary intervention. In total, there were 14 conversions to open repair following EVAR deployment; 4 during the primary theatre procedure, 2 more during the primary admission and 8 after initial discharge from hospital. In contrast, complications and re-interventions were rare in open repair patients.

HRQL
At baseline, the EQ5D scores appeared to be similar to age and sex matched population norms and were very similar between the two randomised groups. Although the open repair group had a diminished HRQL at 0-3 months, this had recovered by 3-12 months and at 12-24 months after randomisation there was no difference between the groups (HRQL over time). Secondary analyses based on time from surgery did not alter these findings (data not shown).