Survival After Endovascular vs Open Aortic Aneurysm Repairs

Importance  To our knowledge, long-term outcomes of open and endovascular (EVAR) repairs of abdominal aortic aneurysms (AAAs) have not been studied on a population level outside a controlled trial setting.

Objective  To determine long-term outcomes of EVAR vs open repair on a population level.

Design, Setting, and Participants  Analysis of the longitudinally linked California Office of Statewide Health Planning and Development inpatient database from 2001 to 2009. Median follow-up was 3.3 years.

Exposures  Endovascular vs open repairs.

Main Outcomes and Measures  Mortality and complications at 30 days, as well as long-term mortality and complications up to 9 years.

Results  In this observational study, a total of 23 670 patients were studied, with 52% receiving EVAR. Endovascular repair was associated with improved 30-day outcomes (all-cause mortality, readmission, surgical site infection, pneumonia, and sepsis), as well as significantly improved survival until 3 years postoperatively. After 3 years, mortality was higher for patients who underwent an EVAR repair. No significant difference in long-term mortality was observed for the entire cohort on adjusted analysis (hazard ratio, 0.99; 95% CI, 0.94-1.04; P = .64). Endovascular repair was found to be associated with a significantly higher rate of reinterventions and AAA late ruptures.

Conclusions and Relevance  The survival advantage for EVAR repair in a statewide population is maintained for 3 years. After 3 years, EVAR repair was associated with higher mortality; however, these mortality differences did not reach statistical significance over the entire study period. Reintervention and late AAA rupture rates are higher after EVAR repair.

Clinical trials of elective endovascular (EVAR) repair of abdominal aortic aneurysms (AAAs) vs open aneurysm repairs (OARs) have consistently shown a lower 30-day mortality for EVAR patients, approximately 1% for EVAR repair compared with 4% for open operation. However, by 1-year follow-up, trials observed that this survival advantage had disappeared. For example, Brown et al,¹ in an analysis of 2 large UK randomized clinical trials of EVAR repair vs OAR concluded that “EVAR offers a clear operative mortality benefit over open repair…but this early benefit is not translated into a long-term survival advantage.” Similarly, a Veterans Affairs Cooperative Study Group trial concluded that EVAR and open repairs resulted in similar long-time survival.²

We reasoned that patients randomized into these studies were required to be fit enough for either EVAR repair or OAR, resulting in selection of lower-risk patients than one would encounter in practice outside the controlled clinical trial setting. Further, if both groups were relatively healthy, having passed cardiac, pulmonary, and renal vetting, this could minimize overall long-term differences.

To our knowledge, the long-term survival and outcomes of EVAR repair and OAR have rarely been studied on a population level, outside a controlled trial setting. Prior population-level studies have relied on Medicare and, therefore, only included patients older than 65 years.³ A state-level database provides the opportunity to study patients of all ages. Therefore, the objective of this study was to analyze the survival and outcomes of patients in California undergoing EVAR repair or OAR to determine whether there is a difference between the statewide population and trial results.

An analysis of the California Office of Statewide Health Planning and Development statewide database from 2001 to 2009 was performed. The California Office of Statewide Health Planning and Development is one of 13 departments within the California Health and Human Services Agency, which maintains the database for all nonfederal hospitals in California. Using encrypted identifiers, this database tracks all patient admissions longitudinally throughout California. Additionally, this database is linked to the Social Security Death Index for mortality records.

Inclusion criteria were patients with International Classification of Diseases, Ninth Revision procedure codes of 38.34, 38.36, 38.44, 38.64, 39.52, or 39.71 in any record position, with International Classification of Diseases, Ninth Revision diagnosis code of 441.4 to indicate nonruptured status. Exclusion criteria included patients with concomitant thoracic aneurysm repairs (procedure code 39.73) or diagnosis of syphilitic (diagnosis code 093.0), traumatic (diagnosis code 901.0 or 902.0), thoracoabdominal (diagnosis code 441.6 or 441.7), ruptured (diagnosis code 441.3), or unspecified (diagnosis code 441.9) aortic aneurysms.

The primary outcome variable was all-cause mortality. Secondary outcomes included 30-day all-cause mortality and all-cause readmissions; 30-day readmissions for wound dehiscence, surgical site infection, pneumonia, and sepsis; and long-term outcomes including repeat AAA repair and AAA rupture. All-cause mortality outcomes were censored at the end of the study on December 31, 2009. All-cause readmission outcomes were censored at death, at the time of any elective admission, or at the end of the study. Reintervention outcome was censored at death or at the end of the study. Other clinical outcomes were censored at death, at the end of the study, or at the time of any subsequent admission because events occurring after subsequent admissions would be difficult to attribute to the initial operation. Cox proportional hazards analyses were performed, controlling for repair type (open vs EVAR), age, race/ethnicity, sex, insurance types, Charlson Comorbidity Index score, calendar year, admission type (scheduled vs unscheduled), and hospital type (teaching vs nonteaching based on the presence of general surgery residency, based on National Resident Matching Program data).

The study was deemed by the University of California–San Diego institutional review board to be exempt from review because the database was previously collected by a government agency and there was no direct patient contact.

A total of 23 670 patients were included in the study, for a median follow-up of 3.3 years (interquartile range, 1.4-5.7 years). Endovascular repair was done in 51.7% of patients. Patient demographic data for the study cohort, as well as unadjusted comparisons between open and EVAR repair patients, are presented in Table 1. Endovascular repair patients were significantly older, less likely to be white, more likely to be men, more likely to have scheduled admissions, and more likely to present to teaching-status hospitals.

Outcomes for the entire cohort, and unadjusted comparison between open and EVAR repair patients, are presented in Table 2. All-cause 30-day mortality was higher for open repair patients, while 30-day all-cause readmission rates were similar between the 2 groups after 30-day deaths were appropriately removed from the denominator. Open repair patients also had higher rates of wound dehiscence, a slightly lower rate of surgical site infections, similar rates of pneumonia, and slightly higher rates of sepsis at 30 days. Beyond 30 days, EVAR repair patients had lower all-cause mortality until year 3, beyond which EVAR repair patients had higher all-cause mortality rates. Endovascular repair patients consistently had higher rates of reintervention, as early as 6 months after the operation, while open repair patients consistently had higher rates of incisional hernia repair.

All-cause mortality trends over time are further presented in Figure 1. Figure 1A shows cumulative totals, which, as described previously, were significantly higher for open repair patients until year 3, after which EVAR repair patients had significantly higher mortality risk. To better appreciate mortality risks at different points, the slope of the cumulative mortality curves were plotted, as shown in Figure 1B. It shows that mortality risks for open repair patients decreased below that for EVAR repair patients before the end of year 1, after which EVAR repair patients consistently had significantly higher mortality risks.

Cumulative time-to-event curves for all-cause readmissions are presented in Figure 2 for 30 days and 1 year. There were several points of crossing-over. Notably, all-cause readmissions were significantly higher for EVAR repair patients up to approximately 17 days after hospital discharge, after which they converged. The top primary diagnoses on 30-day readmission were postoperative infections, congestive heart failure, complications related to the graft, complications related to the gastrointestinal track, pneumonia, and acute renal failure. Beyond 30 days, the cumulative totals for readmissions were slightly higher for open repair patients until approximately 120 days, after which the totals for EVAR repair patients became higher. These data suggest that readmission risks are higher for EVAR repair patients very early in the postoperative course and several months later during the recovery.

Reintervention data are presented in Figure 3A. Reintervention curves are significantly different throughout the entire study for EVAR repair patients, reaching 10% at year 7. Rates of AAA ruptures are presented in Figure 3B. A total of 38% of all patients who had a repeat AAA repair underwent an open re-repair. Of the patients who originally had EVAR repair, 31% required an open intervention subsequently compared with 62% of patients who originally had an open repair.

Unadjusted stratified analysis in patients older than 65 years, and in patients younger than 65 years, are presented in Figure 4 for all-cause mortality and reintervention. They showed no significant difference in all-cause mortality and all-cause readmissions between procedure types in the younger-than-65 population. And the difference in reintervention rates between procedure types is much smaller, albeit significant, in the younger-than-65 population.

On adjusted analyses with the Cox proportional hazards method, controlling for age, race/ethnicity, sex, insurance types, Charlson Comorbidity Index score, calendar year, scheduled vs unscheduled admissions, and teaching vs nonteaching hospital type, there was no significant difference in overall mortality curves (hazard ratio, 0.99; 95% CI, 0.94-1.04; P = .64), reflecting the crossover at year 3. Additionally, open repair patients had significantly lower risk for reintervention (hazard ratio, 0.26; 95% CI, 0.22-0.32; P < .001) and AAA rupture (hazard ratio, 0.14; 95% CI,0.08-0.24; P < .001).

As expected, we found 30-day mortality to be higher for OAR than EVAR repair in California patients, agreeing with the randomized trials’ results. Mortality at 30 days for EVAR repair patients in the statewide population was 1.54%, only slightly higher than that for EVAR repair in the clinical trials, in which it ranged from 0.5% to 1.2%.^4,5 Malas et al⁶ also found that the 30-day mortality after EVAR repair in a National Surgical Quality Improvement Program database analysis was 1.3%, very similar to our finding, and attributed the small increase over trial data to the exclusion of high-risk patients from the clinical trials. Similarly, mortality after OAR in the statewide population was 4.74%, slightly higher than the rates reported in clinical trials, such as the 4.7% reported in the EVAR Trial 1⁷ and 4.6% in the Dutch Randomized Endovascular Aneurysm Management Trial,⁵ with the exception of 3.0% in the Open Versus Endovascular Repair Trial.⁴ Also, a lower mortality rate of 3.7% was reported from the National Surgical Quality Improvement Program registry.⁶ Our findings were more similar to that reported from Medicare, with 1.2% 30-day mortality after EVAR repair and 4.8% after OAR.⁸

However, unlike the clinical trial results, we found the survival advantage for EVAR repair to be maintained until 3 years postoperatively. Given that the major risk factor for AAA is smoking, this advantage would inevitably erode as cardiovascular disease, emphysema, and pulmonary malignancy exact their toll.⁹ After 3 years, mortality was higher for patients who had EVAR repair. We believe this is explained by the willingness of the surgeon to undertake EVAR repair in older patients knowing that the less-invasive procedure is safer than OAR. Acceptance of high-risk patients was not possible in the clinical trials because every patient had to qualify for open surgery before randomization. Clearly, extension of life after either EVAR repair or OAR will require cessation of smoking and close management of chronic cardiovascular disease, emphysema, and renal dysfunction.

Although the readmission rate at 30 days was similar between EVAR repair and OAR, intervention for aneurysm-related complications, such as leak, was higher for EVAR repair at each interval from 6 months to 5 years, reaching 7% at 5 years. During this interval, computed tomographic angiography is done on an annual basis to detect endoleaks, which are usually type II and have an incidence initially of approximately 20%. As the significance of type II endoleaks is reconsidered, and if new devices prove effective, it is likely that fewer interventions will be made for this complication, reducing the major difference in readmissions for aneurysm-related complications.¹⁰

The advantage and uniqueness of this study was the ability to reliably track patients through hospitals and time to capture long-term outcomes. Capture of both inpatient and out-of-hospital mortality using statistical death data allows for accurate capture of survival without loss to follow-up. We included patients younger than 65 years not included in prior retrospective studies used of Centers for Medicare and Medicaid Services data.^3,8,11 Also, we examined outcomes from a variety of hospitals and surgeons and were able to demonstrate a more representative analysis of outcomes after AAA repair than has been reported in controlled trials or prior retrospective analyses.

Several disadvantages are inherent in the use of administrative databases, the most relevant of which is the ambiguity inherent in the International Classification of Diseases, Ninth Revision coding system. We selected the outcomes and variables used in risk adjustment carefully to avoid most of this uncertainty, studying more procedure-based outcomes, which are more likely to be coded accurately. Furthermore, focusing on long-term mortality as the primary outcome studied also circumvented this issue. The other relevant disadvantage of using a database was the inability to study the impact of clinical variables such as preoperative physical status, aneurysm size, endovascular device used, and other patient anatomic and operative information. Lastly, our study was limited to patients who underwent a AAA repair between 2001 and 2009 and therefore may not reflect more recent practices including a greater proportion of younger patients undergoing EVAR repair, newer devices, and greater experience with endovascular interventions and development of optimal surveillance protocols, which may have altered the outcomes studies.

Our analyses of a California statewide public health database showed a survival advantage until 3 years postoperatively for all patients undergoing AAA repair by an endovascular technique. After 3 years, the mortality rate of EVAR repair patients was higher; however, these mortality differences did not reach statistical significance on adjusted analysis over the entire study. The 30-day mortality in this real-world scenario was only slightly higher for EVAR repair than that reported in the clinical trials. Reintervention was higher from 6 months through 5 years in EVAR repair patients, which reflects the trends in management of endoleaks and technology available during the study.

Corresponding Author: Samuel E. Wilson, MD, Department of Surgery, University of California–Irvine Medical Center, 101 The City Dr, Bldg 53, Room 208, Rte 81, Orange, CA 92868 (wilsonse@uci.edu).

Accepted for Publication: May 18, 2015.

Published Online: September 2, 2015. doi:10.1001/jamasurg.2015.2644.

Author Contributions: Dr Chang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: Chang.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Chang, Parina.

Administrative, technical, or material support: Chang, Wilson.

Study supervision: Chang.

Conflict of Interest Disclosures: None reported.

Previous Presentation: This study was presented at the 86th Annual Meeting of the Pacific Coast Surgical Association; February 20, 2015; Monterey, California.