National outcomes for the treatment of ruptured abdominal aortic aneurysm: Comparison of open versus endovascular repairs
Article Outline
Objectives
Endovascular repair (EVAR) of ruptured abdominal aortic aneurysms (rAAA) has been shown to acutely decrease procedural mortality compared to open aortic repair (OAR). However, little is known about the effect of choice of procedure; EVAR vs OAR, or the impact of physician and institution volume on long-term survival and outcome.
Methods
Patients hospitalized with rAAA who underwent either OAR or EVAR, were derived from the Medicare inpatient dataset (1995-2004) using ICD9 codes. We evaluated long-term survival after OAR and EVAR in the entire fee-for-service Medicare population, and then in patients matched by propensity score to create two similar cohorts for comparison with Kaplan-Meier analysis. Annual surgeon and hospital volumes of EVAR (elective and ruptured), OAR (elective and ruptured), and rAAA (EVAR and OAR) were divided into quintiles to determine if increasing volumes correlate with decreasing mortality. Predictors of survival were determined by Cox modeling.
Results
A total of 43,033 Medicare beneficiaries had rAAA repair: 41,969 had OAR and 1,064 had EVAR. The proportions of patients with diabetes, hypertension, cardiovascular, cerebrovascular, renal disease, hyperlipidemia, and cancer were statistically higher in the EVAR than in the OAR group, whereas lower extremity vascular disease was higher in the OAR group. The initial evaluation of EVAR vs OAR, prior to propensity matching, showed no statistical advantage in EVAR-survival after 90 days. The survival analysis of patients matched by propensity score showed a benefit of EVAR over OAR that persisted throughout the 4 years of follow-up (P = .0042). Perioperative and long-term survival after rAAA repair correlated with increasing annual surgeon and hospital volume in OAR and EVAR and also with rAAA experience. EVAR repair had a protective effect (HR = 0.857, P = .0061) on long-term survival controlling for comorbidities, demographics, and hospital and surgeon volume.
Conclusion
When EVAR and OAR patients are compared using a reliable statistical technique such as propensity analysis, the perioperative survival advantage of rAAA repaired endovascularly is maintained over the long term. Institutional experience with rAAA is critical for survival after either OAR or EVAR.
Surgical repair of the ruptured abdominal aortic aneurysm (rAAA) has a perioperative mortality rate up to 70 percent,1 resulting in one of the highest surgical mortalities of all vascular emergencies. Between 1960 through 2000, meta-analysis has shown a gradual, albeit modest, reduction in mortality of 3.5 percent per decade, which is most likely the result of advances in medical care and surgical technique.2 Identifying factors contributing to this reduction may help to further reduce the mortality associated with repair of rAAA, especially in centers with high mortality rates.
In the past decade, the magnitude and invasiveness of surgical interventions has been dramatically changed by the advent of minimally-invasive procedures. Multiple publications have compared the short-term mortality of the open rAAA procedure to rAAA repaired endovascularly3, 4, 5 and have found differences similar to those reported contrasting EVAR to the elective open approach.6, 7, 8, 9 The lower hemodynamic shifts associated with endovascular repairs (EVAR) of AAA could potentially benefit survival after rAAA, especially in those patients that are already hemodynamically unstable. In fact, the reported short-term benefit of aortic ruptures repaired endovascularly when compared to open aortic repair (OAR) may be partially attributed to the different surgical approaches: femoral access vs midline laparotomy with complete aortic cross clamp, which confers different hemodynamic and physiologic challenges. However, there are explanations beyond the operative approaches that must be considered when comparing the outcomes of endovascular and open repair of rAAA. One of the most challenging obstacles in evaluating this patient population is the disparate nature of their presentation and the variations of their clinical comorbidities. We used propensity matching to standardize the EVAR and OAR cohorts. However, despite the use of propensity analysis, there still exist limitations in the analysis of administrative datasets. These limitations include certain variables, such as the time from arrival at the emergency room to aneurysm repair, the availability of endovascular imaging, and physiologic data are not captured and, therefore, cannot be controlled.
Improvements in surgical outcomes with increased surgeon and hospital volume have been observed in elective vascular procedures10, 11, 12, 13, 14 and many other surgical interventions, such as coronary artery bypass surgeries.15, 16, 17 However, a national volume-outcome analysis for rAAA has not been performed and the long-term effectiveness of endovascular repair of rAAA is unknown.
To describe the relationships between surgical technique, volume, and long-term rAAA mortality, we examined the Medicare database for the years 1995-2004.
Methods
Data sources and study population
Medicare In-patient Standard Analytical file (Medicare part A) from 1995-2004 was obtained from the Center for Medicare and Medicaid Services through Research Data Assistance Center (resdac@umn.edu). The file contains hospital-discharge abstracts on 100% Medicare reimbursed hospitalizations, except those delivered to beneficiaries enrolled in Medicare HMOs (approximately 10% patients). The data was supplemented with Medicare Denominator file which contain demographic, geographic, and vital status data on all Medicare beneficiaries.
Patients who underwent rAAA repair were identified through a combination of rAAA diagnoses (ICD-9-CM code 441.3 – aortic abdominal aneurysm, ruptured; any position) with primary or any secondary ICD-9-CM procedure codes for EVAR and OAR (Table I). All cases that had diagnoses of 441.1 thoracic aneurysm, ruptured or 441.6 thoracoabdominal aneurysm, ruptured were excluded from the analysis. The designated ICD-9 procedure code for EVAR was created in the year 2000, and only patients with this specific endovascular code were included in the EVAR cohort.
Table I. ICD9 Diagnostic and procedural codes for population selection
| Code | Description |
|---|---|
| Diagnostic | |
 441.3 | Aortic abdominal aneurysm, ruptured |
| Procedural | |
| 39.71 | Endovascular implantation of graft in abdominal aorta |
| 38.44 | Resection of abdominal aorta with replacement |
| 39.25 | Aorta-iliac-femoral bypass |
| 39.52 | Other repair of aneurysm |
| 38.34 | Resection of abdominal aorta with anastomosis |
| 38.64 | Other excision of abdominal aorta |
| 38.40 | Resection of vessel with replacement, unspecified site |
| 38.60 | Other excision of vessels, unspecified site |
We assessed the following comorbidities (primary and all secondary diagnosis): cardiac (coronary artery and valvular disease, congestive heart failure, and arrhythmia), diabetes, hypertension, chronic pulmonary, clinically significant lower extremity vascular disease, cerebrovascular, liver disease, renal atherosclerosis, renal failure, kidney transplant, neurological disorders, cancer, rheumatoid arthritis, and disorder of lipid metabolism. The list of ICD9 diagnosis codes for co-morbidities is provided in Table II, online only. The lack of “present on admission” flag in the Medicare dataset creates a challenge in the separation of comorbidities from complications. We identified as comorbidities the chronic conditions reported at the index hospitalization (hospitalization when rAAA repair occurred) and all comorbidities that were already present in the patient's previous hospitalizations.
Statistics
Kaplan-Meier analysis was used to analyze survival. Differences between groups were compared with the log-rank test. Mortality from all causes was included in the analysis. If a patient had several rAAA repairs, only the first surgery was included in the survival analysis. Cox proportional hazard regression was used to estimate covariate-adjusted hazard ratio of long-term survival. The following variables were included in the model: baseline co-morbidities, demographics, hospital and surgeon volume characteristics, and year when the surgery was performed. The final model included variables with a significance level of 0.05 or less. We used the Martingale methods to check the proportional hazard assumption.
To assess annual hospital and physician experience with AAA, all repairs (elective and ruptured) were included in the volume calculation. Hospital and surgeon volume were expressed as number of EVAR (ruptured and elective), OAR (ruptured and elective), or rAAA procedures (ruptured EVAR and ruptured OAR) for each year. In addition, we also created 5 categories (quintiles) for volume based on equal size groups. The number of surgeons who performed one rAAA procedure per year was disproportionally higher than the number of surgeons with higher annual volumes. Therefore, for the rAAA volume, the first and second quintiles that included surgeons with 1 annual rAAA repair were combined. Volumes were included as continuous variables in Cox models.
To estimate the difference in baseline patients' characteristics, univariate analyses were conducted using Student t tests for continuous variables and χ2 or the Fisher exact test, where it was appropriate, for dichotomous variables. The Cochran-Armitage test was used to analyze trends in the utilization of EVAR for rAAA by hospitals and surgeons.
Treatment selection bias was controlled by constructing propensity score.18 All baseline variables including age, gender, race, co-morbidities, year of surgery, the hospital, and surgeon volume in EVAR, OAR, and rAAA were included in the logistic regression model to predict probability that a patient would receive EVAR vs OAR. The fit of the propensity model to the data was assessed using the concordance index.19 Patients who underwent EVAR were matched to patients who had OAR, using individual propensity scores. We used 1:1 matching scheme without replacement.20 Additional testing was performed to ensure that there were no significant differences in demographics and co-morbid characteristics between matched groups. Paired t test was used for continuous and McNemar's test for categorical variables. Survival distributions for matched patients were compared using Kaplan-Meier curves and the log-rank test. Cox proportional hazard regression was used to identify risk factors of survival.
Statistical significance was expressed as both P values and 95% confidence intervals (CI). P values less than .05 were considered statistically significant. All statistical analysis was performed using the SAS system software version 9.1 (SAS Institute Inc, Cary, NC).
Results
Study population
From 1995 through 2004, 43,760 Medicare beneficiaries were identified who had repair of rAAA. Among them, 727 patients (1.66% of the cohort) were electively admitted with diagnoses of aortic abdominal aneurysm without mention of rupture and had rAAA diagnoses on their discharge summary. These patients were excluded from the study. The reason for exclusion was that these patients could have had interoperative rupture during elective repair. The final dataset contained 43,033 patients: 41,969 had open AAA repair and 1,064 had an endovascular procedure. We found 75 patients who had both endovascular and open repair during the same hospitalization. Seventy patients had the two procedures during the same day, while the remaining 5 patients first had EVAR with subsequent open repair within the next 1-5 days. Perioperative mortality for these patients was 44% (95% CI, 33.65-55.94%) and median survival 108 days. It appeared from the records that these patients had undergone an original attempt at EVAR which was not successful then requiring an open operation. These 75 patients were considered as failed EVAR and, therefore, included in the EVAR group. Patients treated with EVAR and OAR were different with regard to base-line characteristics (Table III). The EVAR group included slightly older patients (77 vs 76 years of age), more females (25.8% vs 22.9%), and a greater number of African American patients (6.7% vs 3.9%). The proportion of patients with diabetes, hypertension, cardiac arrhythmia, coronary artery, valvular, vascular intestine, cerebrovascular disease, renal atherosclerosis, renal failure, renal transplant, rheumatoid arthritis, hyperlipidemia, and cancer were statistically higher in the EVAR than in the OAR group. Despite significant variability in the comorbidities, only clinically significant lower extremity vascular diseases was more prevalent in the OAR group (6.10% vs 4.42%, P = .0233) as shown by Table III. To control for patient, hospital, and surgeon characteristics, a propensity model was developed and OAR and EVAR patients were matched by their propensity score. A total of 1,044 pairs were identified. The final propensity score model yielded excellent discriminative characteristics (concordance index = 0.892). Patients in these cohorts were similar in all variables (Table IV, online only).
Table III. Base-line characteristics of the patients who had open or endovascular ruptured aortic abdominal aneurysm repair
| Characteristic | Endo AAA group | Open AAA group | P value |
|---|---|---|---|
| (n = 1,064) | (n = 41,969) | ||
| Age (years) | 76.96 ± 0.45 | 75.9 ± 0.07 | <.0001 |
| Female gender (%) | 25.84% | 22.87% | .0228 |
| White race (%) | 91.35% | 93.68% | .0021 |
| Black race (%) | 6.39% | 3.86% | <.0001 |
| Asian race (%) | 0.66% | 0.47% | .3839 |
| Native American race (%) | 0.09% | 0.12% | .8295 |
| Hispanic ethnicity (%) | 0.75% | 0.51% | .2824 |
| Comorbidities (%) | |||
| Diabetes | 11.75% | 8.43% | .0001 |
| Hypertension | 57.33% | 45.16% | <.0001 |
| Chronic pulmonary | 40.22% | 37.51% | .0705 |
| Coronary | 35.24% | 27.53% | <.0001 |
| Congestive heart failure | 23.21% | 21.64% | .2176 |
| Cardiac arrhythmia | 32.80% | 28.91% | .0057 |
| Valvular diseases | 7.89% | 5.28% | .0002 |
| Clinically significant lower extremity | 4.42% | 6.10% | .0233 |
| Renal atherosclerosis | 2.63% | 0.66% | <.0001 |
| Vascular intestine disease | 0.66% | 0.24% | .0164 |
| Kidney transplant | 0.56% | 0.10% | .0009 |
| Renal failure | 9.21% | 6.45% | .0003 |
| Cerebrovascular | 9.02% | 6.04% | <.0001 |
| Other neurological | 5.36% | 5.00% | .6069 |
| Liver diseases | 1.12% | 1.27% | .6774 |
| Hyperlipidemia | 17.01% | 8.46% | <.0001 |
| Cancer | 6.20% | 4.25% | .0019 |
| Rheumatoid arthritis | 2.73% | 1.83% | .0324 |
The total number of patients treated with EVAR for rAAA increased over the study period – from 222 patients (5% of rAAA repairs) in 2001, and to 289 patients (9% of rAAA repairs) in 2004 (Fig 1, A). The number of hospitals as well as number of surgeons using EVAR for rAAA rose consistently during this time period (from 174 to 224 or 12-17% hospitals and from 204 to 263 surgeons or 7-11%) (Fig 1, B and C). Trends in number of surgeons, hospitals, and EVAR procedures were statistically significant.
Fig 1.
Percent of patients with ruptured aortic abdominal aneurysms repaired endovascularly (A), percent of surgeons using endovascular procedures for ruptured aortic abdominal aneurysm repair (B), and percent of hospitals using endovascular repair for ruptured aortic abdominal aneurysm (C) from 2001-2004. P values of trends shown in the parenthesis.
Mortality
Kaplan-Meier analysis of 1,044 pairs of matched patients, showed a survival benefit of EVAR over OAR, which persisted over 4 years of follow-up (log-rank P = .0042) (Fig 2). The median survival was 137 days in EVAR and 35 days in the OAR group. Prior to matching, including all 43,033 patients in the dataset, Kaplan-Meier analysis showed a statistically significant short-term benefit of EVAR vs OAR that persisted for only 90 days of follow-up, and an insignificant difference in long-term survival after 90 days and throughout the 4-year follow-up period (log-rank P = .2007).
Fig 2.
Kaplan-Meier analysis of survival of patients treated with endovascular (EVAR) and open (OAR) repair of ruptured aortic abdominal aneurysm. Cases were matched by patients baseline demographic (age, gender and race), comorbidities, annual hospital and surgeon volume and year of surgery using propensity score analysis.
The year the surgery was performed did not affect survival after OAR; however, we did observe an improvement in survival after EVAR with advancing years. Long-term survival after OAR and EVAR were the same in 2001-2002 (P = .8296), however in 2003-2004 the survival difference was statistically significant favoring EVAR (P = .0279).
Effect of volume on mortality
Perioperative mortality decreased with increasing surgeon experience for both procedures (Fig 3). For very low-volume surgeons, there was no significant difference in EVAR or open mortality (50.02% vs 54.49%, P > .05). The decrease in operative mortality rates between very low- and very high-volume surgeons was more substantial for EVAR, therefore, high-volume endovascular surgeons had lower operative mortality than high-volume open surgeons (28.37% vs 37.72%, P < .05) (Fig 3, A and B). High-volume endovascular surgeons had lower mortality when compared to high-volume open surgeons (28.37% vs 37.72%, P < .05) 30 days after rAAA repair. Perioperative survival significantly improved with surgeon rAAA volume: a 10% decline in 30-day mortality was observed with increasing annual volume from 1 to 3 or more rAAA procedures (Fig 3, C). Moreover, long-term survival also improved with increasing surgeon's experience in OAR (Fig 4, A), EVAR (Fig 4, B), and rAAA (Fig 4, C). The survival advantage persisted to the end of the follow-up period.
Fig 3.
Thirty-day mortality after ruptured aortic abdominal aneurysm surgery in relation to annual surgeon volume of aortic abdominal aneurysm repair by quintiles: A, open (ruptured and elective), B, endovascular (ruptured and elective), or C, ruptured (open and endovascular). *P <.05 for volume quintiles vs the first quintile.
Fig 4.
Kaplan-Meier survival analysis after ruptured aortic abdominal aneurysm surgery by surgeon volume. A - survival after open repair of ruptured abdominal aortic aneurysm in relation to annual surgeon volume of open aortic abdominal aneurysm repair (ruptured and elective) by quintiles. B - survival after endovascular repair of ruptured abdominal aortic aneurysm in relation to annual performing surgeon volume of endovascular aortic abdominal aneurysm repair (ruptured and elective) by quintiles. C - survival after repair of ruptured abdominal aortic aneurysm in relation to annual performing surgeon volume of ruptured aortic abdominal aneurysm repair (open and endovascular) by quintiles.
A similar analysis was performed focusing on annual hospital volume (Fig 5, Fig 6). Ruptured AAAs repaired in very low-volume hospitals (first quintile) had similar perioperative mortality rates for both open and endovascular procedures (56.42% vs 57.67%, P > .05), whereas there was a significant difference at the highest volume hospitals (fifth quintile) favoring EVAR: 38.43% after OAR vs 30.37% after EVAR (P < .05) (Fig 5, A and B). In addition, mortality declined from 53% to 40% with increasing annual hospital volume of rAAA repair from 1 to 7 or more procedures, respectively (Fig 5, C). Long-term survival improved with increasing hospital annual volume in OAR (Fig 6, A), EVAR (Fig 6, B), and rAAA volume (Fig 6, C).
Fig 5.
Thirty-day mortality after ruptured aortic abdominal aneurysm surgery in relation to annual hospital volume of aortic abdominal aneurysm repair by quintiles: A, open (ruptured and elective), B, endovascular (ruptured and elective), or C, ruptured (open and endovascular). *P <.05 for volume quintiles vs the first quintile.
Fig 6.
Kaplan-Meier survival analysis after ruptured aortic abdominal aneurysm surgery by hospital volume. A, survival after open repair of ruptured abdominal aortic aneurysm in relation to annual hospital volume of open aortic abdominal aneurysm repair (ruptured and elective) by quintiles. B, survival after endovascular repair of ruptured abdominal aortic aneurysm in relation to annual hospital volume of endovascular aortic abdominal aneurysm repair (ruptured and elective) by quintiles. C, survival after repair of ruptured abdominal aortic aneurysm in relation to annual hospital volume of ruptured aortic abdominal aneurysm repair (open and endovascular) by quintiles.
Predictors of mortality
Risk factors for decreased long-term survival after endovascular repair of rAAA identified with Cox model included: cerebrovascular disease, female gender, and patient age at the time of surgery. Hospital volume in rAAA and patient's hyperlipidemia had a protective effect on likelihood of survival (Table V). These two factors also had a protective effect on survival after open repair of rAAA (up to 10 years). Additional protective factors for OAR were hospital and surgeon volumes and hypertension. Multiple risk factors that negatively affected the survival after open rAAA included kidney transplant, chronic liver disease, renal failure, cancer, cerebrovascular, neurological, coronary, chronic pulmonary disease, renal atherosclerosis, diabetes, rheumatoid arthritis, female gender, patient age at the time of procedure, and white race (Table V).
Table V. Hazard ratios and 95% confidence intervals for Cox analysis for open and endovascular repair of ruptured abdominal aortic aneurysms, hazard ratio <1 is protective, hazard ratio >1 is harmful, hazard ratio = 1 is indifferent
| Predictor | Hazard ratio | 95% Confidence interval | P value |
|---|---|---|---|
| Endovascular repair (n = 1,064) | |||
| Cerebrovascular | 1.630 | 1.263-2.103 | .0002 |
| Female gender | 1.267 | 1.065-1.507 | .0075 |
| Patient age at time of surgery⁎⁎ | 1.030 | 1.019-1.041 | <.0001 |
| Ruptured AAA hospital annual volume⁎ | 0.972 | 0.950-0.995 | .0166 |
| Hyperlipidemia | 0.767 | 0.614-0.959 | .0200 |
| Open repair (n = 41,969) | |||
| Kidney transplant | 1.814 | 1.275-2.582 | .0009 |
| Liver diseases | 1.388 | 1.263-1.526 | <.0001 |
| Renal failure | 1.385 | 1.325-1.448 | <.0001 |
| Cancer | 1.243 | 1.179-1.310 | <.0001 |
| Cerebrovascular | 1.222 | 1.167-1.279 | .0001 |
| Other neurological | 1.185 | 1.128-1.245 | <.0001 |
| Female gender | 1.176 | 1.145-1.208 | <.0001 |
| Renal atherosclerosis | 1.161 | 1.014-1.329 | .0312 |
| Coronary | 1.112 | 1.084-1.142 | <.0001 |
| Diabetes | 1.109 | 1.065-1.155 | <.0001 |
| Rheumatoid arthritis | 1.103 | 1.017-1.196 | .0183 |
| White race | 1.084 | 1.034-1.137 | .0008 |
| Patient age at time of surgery | 1.041 | 1.040-1.043 | <.0001 |
| Chronic pulmonary | 1.025 | 1.002-1.050 | .0365 |
| OAAA annual hospital volume⁎ | 0.999 | 0.998-0.999 | <.0001 |
| Ruptured AAA hospital annual volume⁎ | 0.991 | 0.987-0.996 | .0002 |
| OAAA annual surgeon volume⁎ | 0.990 | 0.988-0.991 | <.0001 |
| Congestive heart failure | 0.941 | 0.916-0.967 | <.0001 |
| Hypertension | 0.899 | 0.877-0.921 | <.0001 |
| Hyperlipidemia | 0.780 | 0.744-0.817 | <.0001 |
⁎per one additional surgery. |
⁎⁎per one additional calendar year. |
After controlling for baseline characteristics, predictors of long-term survival included: EVAR (Hazard ratio [HR] = 0.857, P = .0061), hospital annual volume in rAAA surgeries (HR = 0.971, P = .0002). Risk factors found to decrease the likelihood of survival were cerebrovascular disease (HR = 1.543, P < .0001) and patient age at the time of surgery (HR = 1.037, P < .0001) as a continuous variable (Table VI).
Table VI. Risk factors of death after ruptured abdominal aortic aneurysms repair for patients matched by propensity score (n = 2,088), hazard ratio <1 is protective, hazard ratio >1 is harmful, hazard ratio = 1 is indifferent
| Predictor | Hazard ratio | 95% Confidence interval | P value |
|---|---|---|---|
| Endovascular repair | 0.857 | 0.768-0.957 | .0061 |
| Ruptured AAA hospital annual volume⁎ | 0.971 | 0.957-0.986 | .0002 |
| Patient age at the time of surgery⁎⁎ | 1.037 | 1.029-1.045 | <.0001 |
| Cerebrovascular diseases | 1.543 | 1.299-1.833 | <.0001 |
| Hyperlipidemia | 0.736 | 0.627-0.865 | .0002 |
| Hypertension | 0.888 | 0.792-0.995 | .0415 |
⁎per one additional surgery. |
⁎⁎per one additional calendar year |
Discussion
In endovascular repair of ruptured aortic aneurysms, laparotomy, aortic cross-clamping, and retroperitoneal dissection is avoided thereby reducing hemodynamic shifts as well as the likelihood of an inflammatory response.21 Significantly decreased mortality of endovascularly repaired rAAA at 30 days vs OAR have been demonstrated in three prospective trials.21, 22, 23 In one prospective multicenter study, no significant difference was noted in 30-day mortality between endovascular and open repairs (35% vs 39%, P = .78).24 As acknowledged by the authors, this result may be due to a larger percentage of sicker and hemodynamic unstable patients treated endovascularly. Also a single center study in which hemodynamically unstable patients had been excluded (31% vs 31% P = .98) reported no survival improvement after endovascular repair vs open repair.25 However, in this study there were differences in case mix among the patients treated with different procedures and there was no attempt to perform a propensity matching analysis to control for comorbidities. In fact, more patients treated endovascularly (50%) than patient treated with open surgery (32%) were considered at high risk for cardiac complications.
Review of our data shows that hypertension and hyperlipidemia are protective against death following EVAR. This seems counterintuitive since these factors are felt to predispose patients to the development of atherosclerosis. A possible explanation for this effect, which has been observed by other investigators, is that the medications used to treat these patients, such as beta-blockers and statins, may have a cardiovascular protective effect and once diagnosed and treated may decrease the instance of fatal myocardial events.26, 27
Hemodynamic determinants, such as blood pressure or cardiac output, are not variables in the Medicare dataset and, therefore, cannot be included in this analysis. Moreover, the Medicare database includes medical information on most US residents over the age of 65, but this does not ensure equal distribution of risk factors within its population. These are intrinsic problems when using large datasets that are not prospectively randomized. The use of propensity scores confers the advantage of having a single estimate available to control for multiple differences between groups in a matched-pair analytical design in order to create two similar cohorts for comparison.
We initially performed an analysis of all 43,033 patients. The unmatched survival analysis of the Medicare dataset showed a significant reduction in mortality for EVAR which disappeared at 90 days. Although these results, as well as those previously described, implied that rAAA repaired endovascularly benefited from improved short-term survival, it appeared that this advantage was not sustained. There was no statistical difference between the unmatched survival curves for EVAR and OAR, and an eventual higher death rate for rAAA patients selected to have EVAR was demonstrated.
The use of the Medicare dataset is not without limitations in that many factors that contribute to the morbidity and mortality of patients are not captured within the Medicare data. These factors include the hemodynamic stability of the patient at presentation (except emergent vs elective), duration of time from presentation to intervention within the same day, size, and location of the aneurysm, and aneurysm neck diameter and configuration. Recent studies have also shown that more unstable patients may in fact be offered endovascular repair preferentially to open repair in some centers.24 To further support this premise, in our study, patients that underwent EVAR had more comorbidities than those patients undergoing OAR (Table III).
In order to further investigate the impact of these factors, baseline characteristics of each group (endo vs open) were examined, hypothesizing that endovascular patients were likely sicker preoperatively and perhaps at greater risk of dying from disease unrelated to their rAAA. The endovascular cohort was comprised of significantly more females, was older, and had significantly higher rates of comorbidities compared to the open cohort (Table III). To control for the population differences, propensity score analysis was employed.
Repeated survival analysis of EVAR and OAR patients, matched by propensity score, demonstrated that the early survival benefit of EVAR is actually sustained throughout the 4 year follow-up period (Fig 2). Recently reported comparison of long-term survival after elective AAA repaired by EVAR or OAR showed that initial advantage of endovascular procedure disappear over time and survival curves crossed even after matching patients by propensity score.28 The explanation for this convergence is speculative and could relate to the possibility that patients who were previously felt to be too high-risk for OAR are now being treated with EVAR. These patients may have succumbed to an early mortality within several years of EVAR, not because of AAA-related complications, but rather due to their comorbidities.
The significant improvement in survival of rAAA repaired endovascularly in 2003-2004 may be due to a better distribution and variety of endovascular devices available, as well as greater surgeon experience. The learning curve associated with elective EVAR is vital and necessary to achieve optimal results29, 30, 31 including the repair of a ruptured AAA. In fact, survival rates for endovascular rAAA repairs improved with increased surgeon volume of EVAR procedures (elective and ruptured) (Fig 4, B). These findings were replicated for surgeon volume of OAR (elective and rupture) (Fig 4, A). Each additional OAR was found to be progressively protective in Cox analysis (Table V). These results suggest that increased surgeon exposure to AAAs improves technical skills and fosters better emergent clinical decisions when faced with a ruptured aortic aneurysm.
Successful management of a ruptured aortic aneurysm is dependent on many factors that are outside the surgeon's hands. The mobilization of the vascular team and surgical staff must be rapid and proficient, and immediate availability of CT resources is critical; increased experience with rAAA pre- and postoperative care will undoubtedly improve survival. The findings of increased survival of rAAA with increased total hospital rAAA volume (Fig 6, C) supports the hypothesis that high institutional experience improves management of the most treacherous vascular emergency from the emergency room to the ICU and ultimately results in better outcomes. These hospital volume-outcome improvements have been observed in elective vascular procedures10, 11, 12, 13, 14 and statewide analyses of ruptured AAA repairs,4 as well as coronary artery bypass, pancreatic cancer, and hip replacement surgeries.15, 16, 17 Cox analysis shows that each additional open repair per hospital (elective and rupture) was protective for rAAA repaired traditionally (Table V). While broadening experience and the constant progress of stent technology will certainly lead to a widespread adoption of the endovascular approach for the treatment of rAAA, the feasibility of EVAR in the context of this surgical emergency currently remains limited by adverse anatomy, severe hemodynamic instability, and device unavailability.22 The use of OAR of rAAA cannot be abandoned due to the limitations in interventional skill as well as many patients' aortic anatomy is unsuitable for current EVAR grafts.
Improved survival from rAAA, with high hospital and surgeon volumes, may lead to the conclusion that we should move towards regionalization and centers of excellence in order to decrease the mortality rate of repaired rAAA. This should be considered in those patients that are stable for transfer if the center is not accustomed to managing complex vascular emergencies or there is not an experienced surgeon available at the time of presentation. Unfortunately, a ruptured aneurysm is rarely diagnosed in the field, and only diagnosed at the time of presentation to the hospital. Indeed, the better outcomes of high-volume hospitals can also arise as the result of selective referral of hemodynamically stable patients to these hospitals. Although we were able to track patient information longitudinally, the Medicare dataset is subject to certain limitations. Missing from this database is information on patients' preoperative hemodynamics and vascular anatomy. Although diagnoses of comorbidities are reported, Medicare data does not capture the extent or severity of a disease; ejection fractions or creatinine levels are, unfortunately, unavailable. We cannot determine whether these unmeasured characteristics account for significant differences between the two cohorts of patients treated with OAR and EVAR. Although the list of potential confounders we used for propensity score matching was extensive, selection bias related to characteristics such as hemodynamic instability, aneurysm size, availability of endovascular imaging and inventory, which are not reported in the Medicare dataset, cannot be ruled out.
In conclusion, this study shows an increased survival after rAAA of patients who were treated with EVAR as compared to those treated with OAR over the course of 4 years. However institutional and surgeon experience were essential to fully benefit from the use of this procedure for the treatment of rAAA.
Author contributions
Tables (online only)
Table II, Online only. List of ICD-9-CM codes for comorbidities
| Comorbidity | ICD9 code |
|---|---|
| Index hospitalization | |
| Congestive heart failure | 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.91, 404.13, 404.93, 425.4, 425.5, 425.7, 425.8, 425.9, 428.0, 428.1, 428.20, 428.22, 428.30, 428.32, 428.40, 428.42, 428.9 |
| Cardiac arrhythmia | 426.0, 426.10, 426.11, 426.12, 426.13, 426.7, 426.9, 427.0, 427.1, 427.2, 427.3, 427.9, V45.0, V53.3 |
| Valvular disease | 093.2, 394, 395, 396, 397, 424, V42.2, V43.3 |
| Coronary disease | 412, 413, 414, 429.2 |
| Diabetes | 250 |
| Hypertension | 401, 402, 403, 404, 405 |
| Pulmonary diseases | 416, 417.9, 490, 491, 492, 493, 494, 495.0, 495.1, 495.2, 495.3, 495.4, 495.5, 495.6, 495.8, 495.9, 496, 500, 501, 502, 503, 504, 505, 506.0, 506.2, 506.4, 506.9, 508.1, 508.8, 508.9 |
| Clinically significant lower extremity vascular diseases | 440.22, 440.23, 440.24, 440.3, 444.22, V43.4 |
| Renal atherosclerosis | 440.1 |
| Vascular intestine disease | 557.1 |
| Renal failure | 403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 585, 588.0, V45.1, V56.0, V56.1, V56.2, V56.3, V56.8 |
| Other renal diseases | 582, 583.0, 583.1, 583.2, 583.4 |
| Kidney transplant | V420 |
| Liver disease | 070.22, 070.23, 070.32, 070.33, 070.44, 070.54, 070.9, 456.0, 456.1, 571, 572.1, 572.2, 572.3, 572.4, 572.8, 573.0, 573.1, 573.8, 573.9 |
| Cerebrovascular diseases and paralysis | 342, 344.1, 344.3, 344.4, 344.5, 344.9, 437.0, 438 |
| Other neurological diseases | 330, 331, 332, 333, 334.0, 334.1, 334.2, 334.4, 334.8, 335.0, 335.1, 335.2, 335.8, 335.9, 336.0, 336.2, 343, 344.0, 348.1, 348.3, 344.2, 344.6, 345, 437.3, 437.4, 437.5, 437.6, 437.7 |
| Hyperlipidemia | 272 |
| Cancer | 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 170, 171, 172, 174, 175, 176, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203.0, 238.6 |
| Rheumatoid arthritis | 446, 701.0, 710.0, 710.1, 710.2, 710.3, 710.4, 710.8, 710.9, 711.2, 719.3, 714, 720, 725, 728.5, 728.89 |
| Pre-index hospitalizations | |
| History of heart failure | 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.91, 404.13, 404.93, 425.4, 425.5, 425.7, 425.8, 425.9, 428 |
| Cardiac arrhythmia | 426, 427.0, 427.1, 427.2, 427.3, 427.4, 427.5, 785.0, 996.01, 996.04, V45.0, V53.3 |
| Valvular disease | 093.2, 394, 395, 396, 397, 424, V42.2, V43.3 |
| Coronary disease | 410, 412, 413, 414, 429.2 |
| Pulmonary | 415, 416, 417, 490, 491, 492, 493, 494, 495, 496, 500, 501, 502, 503, 504, 505, 506.0, 506.2, 506.4, 506.9, 508 |
| Clinically significant lower extremity vascular diseases | 440.22, 440.23, 440.24, 440.3, 444.22, 996.7, V43.4 |
| Renal atherosclerosis | 440.1, 445.81 |
| Vascular intestine disease | 557.1, 557.9 |
| Hypertension | 401, 402, 403, 404, 405, 458.0, 458.1, 458.8, 458.9 |
| Hyperlipidemia | 272 |
| Cerebrovascular diseases and paralysis | 342, 344.1, 344.3, 344.4, 344.5, 344.9, 362.30, 362.31, 362.34, 433, 434, 435, 436, 437.8, 437.9, 438, 784.3 |
| Other neurological diseases | 330, 331, 332, 333, 334.0, 334.1, 334.2, 334.3, 334.4, 334.8, 334.9, 336.0, 335.0, 335.1, 335.2, 335.8, 335.9, 336.0, 336.2, 340, 343, 344.0, 344.2, 344.6, 345, 348.1, 348.3, 430, 431, 432, 437.3, 437.4, 437.5, 437.6, 437.7, 780.3 |
| Diabetes | 250 |
| Renal failure | V45.1, V56.0, V56.1, V56.2, V56.3, V56.8, 403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 585, 586, 588.0 |
| Renal diseases | 582, 583.0, 583.1, 583.2, 583.4, 583.6, 583.7 |
| Liver disease | 070.22, 070.23, 070.32, 070.33, 070.44, 070.54, 070.6, 070.9, 456.0, 456.1, 456.2, 571, 572.2, 572.3, 572.4, 572.8, 573 |
| Cancer | 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 170, 171,172, 174, 175, 176, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203.0, 238.6 |
| Kidney transplant | V42.0 |
| Rheumatoid arthritis | 446, 701.0, 710.0, 710.1, 710.2, 710.3, 710.4, 710.8, 710.9, 711.2, 719.3, 714, 720, 725, 728.5, 728.89, 729.30 |
Table IV, Online only. Characteristics of patients who underwent open or endovascular repair of ruptured aortic abdominal aneurysm, hospital's and surgeon's annual volume in AAA repair after matching by propensity score
| Characteristic | Endo AAA group (n = 1,044) | Open AAA group (n = 1,044) | P value |
|---|---|---|---|
| Patients | |||
| Age (years) | 76.93 | 76.99 | .8462 |
| Female gender (%) | 26.15% | 26.34% | .9203 |
| White race (%) | 91.57% | 92.53% | .4263 |
| Comorbidities (%) | |||
| Diabetes | 11.59% | 10.92% | .6299 |
| Hypertension | 57.37% | 56.23% | .6008 |
| Chronic pulmonary | 39.94% | 39.94% | 1 |
| Coronary | 34.86% | 36.78% | .3573 |
| Congestive heart failure | 22.99% | 24.04% | .5741 |
| Cardiac arrhythmia | 32.76% | 33.72% | .6467 |
| Valvular diseases | 7.66% | 6.51% | .3035 |
| Clinically significant lower extremity | 4.50% | 6.23% | .0747 |
| Renal atherosclerosis | 2.20% | 2.49% | .6682 |
| Vascular intestine disease | 0.67% | 0.77% | .7963 |
| Kidney transplant | 0.57% | 0.49% | .763 |
| Renal failure | 9.20% | 7.76% | .2457 |
| Cerebrovascular | 9.10% | 10.63% | .2482 |
| Other neurological | 5.46% | 5.94% | .6381 |
| Liver diseases | 1.15% | 1.44% | .5637 |
| Hyperlipidemia | 17.05% | 17.72% | .6846 |
| Cancer | 6.23% | 6.80% | .5775 |
| Rheumatoid arthritis | 2.59% | 2.87% | .6911 |
| Annual Volume | |||
| Hospital volume in endo AAA repair | 23 | 23 | .6553 |
| Hospital volume in open AAA repair | 25 | 26 | .8118 |
| Surgeon volume in endo AAA repair | 8 | 7 | <.0001 |
| Surgeon volume in open AAA repair | 6 | 6 | .9062 |
References
- . Rupture in small abdominal aortic aneurysms. J Vasc Surg. 1998;28:884–888
- . A meta-analysis of 50 years of ruptured abdominal aortic aneurysm repair. Br J Surg. 2002;89:714–730
- . Endovascular management of ruptured abdominal aortic aneurysms: 6-year experience from a Greek center. J Vasc Surg. 2005;42:615–623discussion 623
- Outcomes of endovascular treatment of ruptured abdominal aortic aneurysms. J Vasc Surg. 2006;43:453–459
- . Endovascular repair of ruptured abdominal aortic aneurysms: logistics and short-term results. J Endovasc Ther. 2003;10:440–446
- . The impact of endovascular treatment on in-hospital mortality following non-ruptured AAA repair over a decade: a population-based study of 16,446 patients. Eur J Vasc Endovasc Surg. 2004;28:41–46
- A statewide experience with endovascular abdominal aortic aneurysm repair: rapid diffusion with excellent early results. J Vasc Surg. 2004;39:10–19
- . Open versus endo: early experience with endovascular abdominal aortic aneurysm repair beyond the clinical trials. Heart Surg Forum. 2004;7:E459–E461
- A randomized trial comparing conventional and endovascular repair of abdominal aortic aneurysms. N Engl J Med. 2004;351:1607–1618
- . Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals. Jama. 1989;262:503–510
- . Abdominal aortic aneurysm repair in Veterans Affairs medical centers. J Vasc Surg. 1996;23:191–200
- . Hospital vascular surgery volume and procedure mortality rates in California, 1982-1994. J Vasc Surg. 1998;28:45–56discussion 56-8
- . Hospital volume, calendar age, and short term outcomes in patients undergoing repair of abdominal aortic aneurysms: the Ontario experience, 1988-92. J Epidemiol Community Health. 1996;50:207–213
- . Dartmouth Atlas of Vascular Health Care review: impact of hospital volume, surgeon volume, and training on outcome. J Vasc Surg. 2001;34:751–756
- . Hospital volume influences outcome in patients undergoing pancreatic resection for cancer. West J Med. 1996;165:294–300
- . Coronary artery bypass surgery: the relationship between in-hospital mortality rate and surgical volume after controlling for clinical risk factors. Med Care. 1991;29:1094–1107
- . Relationship between mortality rates and hospital patient volume for Medicare patients undergoing major orthopaedic surgery of the hip, knee, spine, and femur. J Arthroplasty. 1997;12:235–242
- . Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127(8 Pt 2):757–763
- . Evaluating the yield of medical tests. Jama. 1982;247:2543–2546
- . Performing a 1:N case-control match on propensity score. Paper presented at: SUGI 29 Proceedings 2004;
- Ruptured abdominal aortic aneurysm: endovascular treatment. Abdom Imaging. 2005;30:263–269
- . Emergency endovascular repair for ruptured abdominal aortic aneurysms: feasibility and comparison of early results with conventional open repair. Eur J Vasc Endovasc Surg. 2005;30:632–639
- . Midterm outcome of endovascular repair of ruptured abdominal aortic aneurysms. J Vasc Surg. 2005;41:752–757
- Endograft treatment of ruptured abdominal aortic aneurysms using the Talent aortouniiliac system: an international multicenter study. J Vasc Surg. 2006;43:1111–1123discussion 1123
- . Endovascular repair versus open surgery in patients with ruptured abdominal aortic aneurysms: clinical outcomes with 1-year follow-up. J Vasc Surg. 2006;44:1148–1155
- . Optimizing the prediction of perioperative mortality in vascular surgery by using a customized probability model. Arch Intern Med. 2005;165:898–904
- . Cardiac risk stratification in patients undergoing endoluminal graft repair of abdominal aortic aneurysm: a single-institution experience with 365 patients. J Vasc Surg. 2003;38:56–60
- . Endovascular vs open repair of abdominal aortic aneurysms in the Medicare population. N Engl J Med. 2008;358:464–474
- . Cumulative sum failure analysis of the learning curve with endovascular abdominal aortic aneurysm repair. J Vasc Surg. 2004;39:102–108
- . The first 150 endovascular AAA repairs at a single institution: how steep is the learning curve?. J Endovasc Ther. 2002;9:269–276
- . Learning curve for endovascular abdominal aortic aneurysm repair: evaluation of a 277-patient single-center experience. J Endovasc Ther. 2002;9:262–268
Competition of interest: none.
Additional material for this article may be found online at www.jvascsurg.org.
PII: S0741-5214(08)01029-X
doi:10.1016/j.jvs.2008.06.036
© 2008 The Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved.
