Randomized Trials: What is the Evidence?

Fig. 14.1
Flowchart of AJAX recruitment

The primary end point was the composite of death and severe complications at 30 days.

The main findings were that there was no difference in mortality at either 30 days (EVAR 21 % vs OSR 25 % p = 0.66) or 6 months (EVAR 28 % vs OSR 31 % p = 0.84) between open and endovascular repair. The AJAX trial also found no difference in severe complications (cardiac, bowel ischemia, reintervention, stroke, amputation, or cord ischemia) between the EVAR and OSR group at 30 days and at 6 months. Renal insufficiency was found to be significantly lower at 30 days and at 6 months in the EVAR when compared with OSR (11 % vs 31 %, respectively, at 30 days, p = 0.01; 11 % vs 31 %, respectively, at 6 months). Encouragingly AJAX also found that mortality for OSR was lower than expected in comparison with a relatively contemporaneous meta-analysis (mortality of OSR = 48.5 %) [7].

AJAX showed that mortality and severe morbidity were equal in EVAR and OSR at 30 days and at 6 months. The trial also presented lower than expected mortality, which was ascribed to the effect of performing surgery in centers with a high level of expertise, in patients with aortic anatomy suitable for endovascular repair (relatively long aortic neck). Limitations of the study include that only aorto-uni-iliac endoprosthesis was used with femoral to femoral crossover grafts in all patients, which is not representative of current practice. Secondly only 22 % of patients with rAAA in the trial region were included, most being excluded due to unsuitable anatomy (46 %). This is likely to mean that the overall mortality rate was underestimated as morphologically complex aneurysms were excluded. Due to the hypothesis of a higher mortality with OSR, the study may not have been sufficiently powered to discriminate a smaller difference of mortality between OSR and EVAR in this particular group of patients with relatively favorable anatomy.

ECAR (Ruptured Aortoiliac Aneurysm: Endo Versus Surgery)

The ECAR trial recruited from January 2008 to January 2013, and 107 patients were enrolled across 14 centers. 56 patients were assigned to the EVAR group and 51 to OSR (see Fig. 14.2) [3].


Fig. 14.2
Flowchart of ECAR recruitment

The primary end point of this study was 30-day mortality, with secondary end points of 30-day rates of cardiovascular, pulmonary, gastrointestinal, renal, and neurological morbidity. Time spent in ITU and volume of blood transfusion received were also end points.

The main outcome of the ECAR trial was similar to that of the AJAX trial, demonstrating that 30-day (18 % EVAR vs 24 % OSR) and 1-year mortality (30 % EVAR vs 35 % OSR) was not statistically different between the OSR and EVAR groups. Analysis of secondary outcomes observed that EVAR was associated with less severe complications. There were reduced rates of pulmonary complications in the EVAR group (15.5 % vs 41.5 % in OSR p = 0.05), reduced requirements for blood transfusion (6.8 units EVAR vs 10.9 OSR p = 0.02), and less intensive use of hospital resources (significantly shorter ITU stay in EVAR vs OSR; 7 vs 11.9 days, p = 0.01).

As with AJAX this study only addressed the question of EVAR vs OSR in a stable group of patients with particular anatomy that would allow EVAR to occur, and once again the trial may not have been sufficiently powered due to better than expected outcomes with OSR.


The IMPROVE trial recruited from September 2009 to July 2013, recruiting patients from 29 British centers and one Canadian center. During this time 613 patients were randomized to OSR or EVAR (see Fig. 14.3), with randomization occurring at a clinical diagnosis of AAA and often before CTA to determine anatomical suitability for EVAR. These patients had varying levels of hemodynamic stability, but clearly moribund patients were excluded. The trial design meant that IMPROVE did include some patients without a diagnosis of ruptured aneurysm, but specified analyses defined that this was not a cause of bias in the trial.


Fig. 14.3
Flowchart of IMPROVE recruitment

Due to the differences in study design, IMPROVE was better able to assess the efficacy of an endovascular first strategy and was more generalizable to the entire ruptured aneurysm patient population. In addition the heterogeneity of trial hospitals more accurately represented the variety of sites that a patient may present to as opposed to only selected expert centers.

Although the AJAX and ECAR trials may appear similar to the IMPROVE trial, they are in fact answering very different questions (see Table 14.1 for a comparison between the three studies). ECAR and AJAX as described above only address a small percentage of the target population, i.e., those that are hemodynamically stable and have undergone a CT which documents rupture and has favorable anatomy for EVAR.

Table 14.1
A comparison between AJAX, ECAR, and IMPROVE




Number of randomized




Number of study sites




Randomized before or after CT



Before (mostly)

Primary end point

30-day composite of death and severe complications

30-day mortality

30-day mortality

Secondary end points

Length of hospital and ITU stay, duration of intubation/ventilation, use of blood products

30-day cardiovascular, pulmonary, gastrointestinal, renal, and neurological morbidity; time spent in ITU and volume of blood transfusion

Reintervention, hospital discharge, health-related quality of life, cost, quality-adjusted life years, cost-effectiveness

Mortality of EVAR vs OSR, 30 days

EVAR 21 %

OSR 25 %

EVAR 18 %

OSR 24 %

EVAR 35 %

OSR 37 %

Mortality EVAR vs OSR, 1 year

(6 months) EVAR 28

(6 months) OSR 31

30 % EVAR

35 % OSR

41 % EVAR

45 % OSR

Analysis of Findings from the IMPROVE Trial

30-Day Outcomes

The 30-day results from the IMPROVE trial demonstrated that the mortality for patients randomized to either EVAR or OSR was not statistically different. The trial observed that women benefit more from endovascular repair, and this was thought in part to be because women had a higher risk of mortality than men for OSR (57 % vs 32 %) when compared to EVAR in which they were more comparable (37 % vs 35 %).

EVAR also demonstrated a trend toward better outcomes for the most seriously ill and older people, which was thought to be due to the minimally invasive nature of the surgery placing less of a physiological strain on these patients. Similarly local anesthesia was also shown to have better outcome than general anesthesia. Many have suggested that permissive hypotension may have a beneficial effect in patients with rAAA in the same way that it has been demonstrated in those with hypovolemic shock following trauma. Contrary to this, permissive hypotension with a systolic blood pressure of under 70 mmHg was associated with increased mortality (51 % if lowest recorded <70 vs 34 % >70) [12]. This finding should be interpreted with some caution as it is well known that patients who develop profound and refractory hypotension are less likely to survive than those in which this does not occur. Similarly, the finding that suggested a better outcome in patients treated under local anesthesia should be interpreted cautiously.

Hospital stay was reduced in the EVAR patients when compared to the OSR group (17 days vs 26 days), and significantly more patients were discharged home after their intervention in the EVAR group. This was part of the reason that IMPROVE also showed that an endovascular strategy was fiscally beneficial, with EVAR calculated to give an incremental net benefit of £3877 compared to OSR at 1 year, despite an allowance of £4000–10,000 per device used and the possible increase in operative theater staff and other additional intervention costs.

IMPROVE did attempt to also include the cost of readmission and reintervention associated with EVAR, but this was performed using a health services questionnaire.

The IMPROVE investigators identified time of presentation as a factor that affected mortality with out-of-hours presentation being higher (odds ratio 1.47 p = 0.048). Efficacy of EVAR vs OSR was unchanged however when randomized to in or out of hours [12].

1-Year Follow-Up Data

The recently reported 1-year follow-up of the IMPROVE trial [14] has continued to reflect the 30-day data. There was no difference observed in all-cause mortality at 1 year (41.1 % EVAR vs 45.1 % OSR) and also no difference in aneurysm-related mortality (33.9 % in EVAR vs 39.3 % in OSR). Of note half of these deaths occurred within 24 h and the majority of the rest within 30 days.

In the longer follow-up, patients randomized to EVAR had an improved health-related quality of life and reduced costs and were more likely to be discharged home sooner. A cost-effectiveness analysis demonstrated benefit to the EVAR first strategy.


Due to the randomization procedure in the IMPROVE study mostly occurring before CTA, the effect of morphology on mortality could be studied [8]. The effect of six morphological parameters (maximum aortic diameter, aneurysm neck diameter, length and conicality, proximal neck angle, and maximum common iliac diameter) was studied to see the effect on a 30-day mortality and reintervention. There were no significant correlations between the six morphological variables seen.

Analysis of these data demonstrated that by far the greatest predictor of mortality was aneurysm neck length and that each 15-mm increase in neck length equated to a reduction in 30-day mortality of approximately 20 %. This relationship could also be seen in the 24-h period, during which half deaths seen occurred. It was also the most likely factor to preclude EVAR.

The importance of neck length as a key factor may also go some way in explaining why randomization after CT makes EVAR look more favorable in terms of early mortality in contemporary cohort studies, as without a suitable neck length, EVAR would not be technically possible. It also explains why to some extent women are seen to benefit so greatly from EVAR, as they often have shorter aneurysm necks [8]. One benefit of EVAR in women is that the distal landing zone is often in common iliac instead of requiring extension into the external iliac which reduces potential morbidity. This is due to a lower incidence of aortoiliac aneurysms in women as well as the smaller average diameter of the distal common iliac [8].

Based on these findings it is clear that the relationship between neck length and mortality significantly confounds the influence of choice of repair strategy on early survival after rAAA.

Individual Patient Meta-analysis of the Randomized Controlled Trials

The three RCTs discussed have been incorporated into an individual patient meta-analysis [15]. Due to the difference in design between AJAX and ECAR when compared with IMPROVE, only the patients that were amenable to EVAR from IMPROVE were included in this analysis.

Part of the rationale of this study was that the three trials assumed higher rates of mortality in the OSR group than observed, meaning perhaps they were insufficiently powered to discriminate a smaller difference in mortality observed in comparison with EVAR. In addition it was felt that advances in perioperative care techniques could potentially have improved 30-day survival in comparison with earlier reports [1], and this was the reason the primary outcome of the meta-analysis was 90-day mortality.

Only gold members can continue reading. Log In or Register to continue

Nov 11, 2017 | Posted by in ABDOMINAL MEDICINE | Comments Off on Randomized Trials: What is the Evidence?
Premium Wordpress Themes by UFO Themes