Journal of Vascular Surgery
Volume 48, Issue 1 , Pages 29-36.e1, July 2008

Regionalization of abdominal aortic aneurysm repair: Evidence of a shift to high-volume centers in the endovascular era

  • Joshua S. Hill, MD

      Affiliations

    • Department of Surgery, University of Massachusetts Medical School, Worcester, Mass
    • ,
  • James T. McPhee, MD

      Affiliations

    • Department of Surgery, University of Massachusetts Medical School, Worcester, Mass
    • ,
  • Louis M. Messina, MD

      Affiliations

    • Department of Surgery, University of Massachusetts Medical School, Worcester, Mass
    • ,
  • Rocco G. Ciocca, MD

      Affiliations

    • Department of Surgery, Caritas St. Elizabeth's Medical Center, Brighton, Mass.
    • ,
  • Mohammad H. Eslami, MD

      Affiliations

    • Department of Surgery, University of Massachusetts Medical School, Worcester, Mass
    • Corresponding Author InformationReprint requests: Mohammad H. Eslami, Division of Vascular Surgery, Department of Surgery, 55 Lake Ave, Worcester, MA 01655.

Received 13 November 2007; accepted 19 February 2008.

Article Outline

Introduction

Since the early 1990s, many studies have shown lower mortality for abdominal aortic aneurysm (AAA) repair at high-volume centers compared with low-volume centers. The introduction of endovascular AAA repair (EVAR) also has changed the practice of AAA repair. The goal of this study was to determine if regionalization of AAA repair occurred in the United States. Etiologic factors were examined in addition to any reduction in operative mortality rates.

Methods

Patient discharges of nonruptured AAA repair were identified from the Nationwide Inpatient Sample between 1998 and 2004. Hospitals were stratified by yearly AAA surgical volume of low (≤17 cases), medium (18 to 50), and high (>50).

Results

A total of 46,901 patients underwent AAA repair (72.7% open vs 27.3% endovascular). The percentage of AAA repairs performed at both low-volume (36.2% to 24.3%) and medium-volume (51.0% to 44.8%) centers fell; whereas, the percentage performed at high-volume centers nearly tripled (12.9% vs 30.9%). In 1998 there were 10 high-volume centers; by 2004 this had increased to 26. The number of low-volume centers decreased, from 412 to 328. EVAR was more rapidly adopted by high-volume centers compared with low-volume centers. By 2004, 64.3% of AAA repairs at high-volume centers were done with endovascular techniques compared with 31.8% in low-volume centers. A concurrent reduction occurred in patient mortality, from 4.4% in 1998 to 2.5% in 2004 (P < .0001).

Conclusion

Between 1998 and 2004, a trend towards the regionalization of AAA repair to high-volume centers occurred. Nearly one-third of all AAA repairs were performed at high-volume centers. There was a concurrent increase in the frequency of endovascular AAA repair, especially at high-volume centers. During this period of regionalization of AAA repair to high-volume centers, patient mortality after AAA repair decreased by 23%. Thus, the observed regionalization of AAA repair and the reduction in short-term patient mortality for this operation may be explained by increased utilization of endovascular technologies at high-volume centers.

 

The risk of postoperative morbidity and mortality after complex operations is significantly lower at high-volume centers compared with low-volume centers.1, 2, 3 Although the underlying cause of this effect continues to be debated, the concept of a volume–outcome relationship has been shown to influence survival for elective abdominal aortic aneurysm (AAA) repair.4 In 2007 Killeen et al5 published an extensive review of the available literature on the volume–outcome relationship in elective AAA repair. This meta-analysis largely comprised studies done before the endovascular era. They found a 3% to 11% reduction in the relative risk of morbidity for patients operated on at high-volume centers compared with low-volume centers.5

Private insurers and health care advocacy groups have used these findings to stratify hospitals according to operative volume.6, 7 In 2000 the Leapfrog Group, a coalition of >150 large health care purchasers, defined high-volume centers for AAA repair as those that performed >50 elective AAA repairs per year. In 2003 the Leapfrog Group suggested that their member employers regionalize the care of elective AAA repair by referring only to high-volume centers.7

Concurrent with this dialogue, endoluminal therapy of AAA was introduced as an alternative to traditional open repair. The first reports of endoluminal therapies for the treatment of AAA were presented in the early 1990s.8, 9 In 1999 the United States Food and Drug Administration (FDA) approved two endoluminal devices for the elective treatment of AAA. Since that time, multiple other stent graft devices have been introduced and have received FDA approval.

Endovascular AAA repair (EVAR) has been shown to have lower mortality rates compared with open aortic repair in the immediate perioperative period; however, long-term outcomes appear to be equivalent.10, 11, 12, 13, 14 The adoption of EVAR has been rapid. Using the Nationwide Inpatient Sample (NIS), Nowygrod et al15 reported that 43.0% of AAA repairs were performed by EVAR in 2003. In 2006 Forbes et al16 suggested that the introduction of EVAR has centralized the care of elective AAA to larger-volume centers within Canada.

The aim of this study is to determine whether the regionalization of elective AAA repair has occurred through the referral of patients to higher-volume centers. The study was limited to elective AAA repair in order to limit any confounding effects introduced by the inclusion of ruptured AAA repair. We investigated the regionalization of AAA repair to high-volume centers and its influence on the operative mortality rate. Because the examined time frame included the introduction of endovascular technologies for AAA repair, we also determined the relative frequency and outcome of open vs EVAR and how this may have influenced these national trends. We propose that the early availability of EVAR techniques at high-volume centers may have further broadened their referral bases, thus accelerating the regionalization of AAA repair to high-volume centers.

Back to Article Outline

Methods 

This study used the NIS to perform a retrospective observational study for the years 1998 to 2004. Managed by the Healthcare Cost and Utilization Project, the NIS is the largest all-payer database of hospital discharges in the United States, providing a 20% stratified sample of all nonfederal hospitals, including academic and specialty hospitals. All patients discharged from member hospitals are included for the given year.

The NIS is a weighted stratified sample that allows for the calculation of national population estimates from the sampled data. The stratification schema of the NIS aims to provide a representative sample of the US population each year. The hospitals included in NIS are stratified using (1) geographic region, (2) urban vs rural, (3) teaching status, (4) for-profit status, and (5) bed size. The stratification criteria have not changed since 1998, thus comparisons across study years are valid.

NIS patient data are linked to hospital data provided by the American Hospital Association (AHA) annual survey of hospitals. This linkage provides demographic data for the hospital, including rural vs urban status, teaching vs nonteaching, and hospital size. Teaching status is defined as the presence of one of the following criteria: (1) Council of Teaching Hospitals and Health Systems membership, (2) a American Medical Association-approved residency program, or (3) an intern-plus-resident/bed ratio ≥0.25.

Patients with the primary diagnosis code of nonruptured AAA (441.4) were queried using codes from the International Classification of Disease, 9th Edition. Only adult records associated with operative treatment were analyzed (Table I).

Table I. Listing of International Classification of Disease, 9th Edition codes used for analysis
CodeDescription
Diagnosis
441.4Abdominal aortic aneurysm without mention of rupture
Procedure
38.34Resection of vessel with anastomosis, aorta
38.44Graft replacement (interposition), abdominal, aorta
39.71Endovascular repair of abdominal aortic aneurysm with graft

Although the sampling frame has not changed between 1998 and 2004, the actual hospitals that were sampled varied yearly. Thus, it is not possible to trend the operative volume of individual hospitals by using the NIS. Instead, the NIS provides a mechanism to evaluate national trends across multiple years by providing a validated stratified sample of US hospitals.17

Statistical analysis 

All analyses were performed with SAS 9.1 software (SAS institute, Cary, NC). The primary outcome measure was yearly number of AAA repairs performed per hospital. Yearly counts of the number of AAA repairs performed by individual hospitals were calculated using the AHA hospital information provided by the NIS. To account for hospitals with low case volumes and to delineate those with nonsuccessive years of sampling by the NIS, the performance of appendectomies was included as a control. Thus, if a hospital performed no AAA repairs and no appendectomies in a given year, then it was deemed to be not included in that year's NIS sample; however, if a hospital performed one or more appendectomies and no AAA repairs in that year, then the number of AAA repairs equaled zero for that year.

The yearly count of AAA repairs for both total operations and EVAR cases was used to assign hospitals to one of three Leapfrog Group volume categories: (1) low volume, zero to 18 AAA repairs, (2) medium volume, 18 to 49, and (3) high volume, ≥50. Volume assignments were provided on a yearly basis. The Leapfrog volume classifications were chosen for this study because the Leapfrog Group is one of the most commonly cited patient-advocacy groups. These cutoffs were developed before the endovascular era and thus do not provide direct evidence for the number of EVAR repairs that should be performed to maintain competency. To our knowledge, no published reports have described the volume–outcome relationship for EVAR. Thus, we believe that the Leapfrog strata provide the best approximation of the volume strata for open and endovascular AAA repair.

The number of hospitals assigned in each of the three volume categories and the number of operations performed by each volume strata were analyzed. A Mantel-Haenszel test of trend was performed to assess for statistical significance.

A survey-weighted multivariable logistic regression was performed. In-hospital mortality was the dependent variable, and covariates included patient sex, age group (<60, 60 to 79, and ≥80 years), and the presence of concurrent comorbidities such as diabetes mellitus, liver failure, renal failure, and congestive heart disease. The presence of these comorbidities was defined by previously validated software developed for the purpose of extracting medical comorbidities from national data sets such as the NIS.18, 19 Other patient level covariates included type of payer, type of operation performed (open vs EVAR) and year of operation. Hospital level variables such as teaching status, bed size, and Leapfrog volume assignment were also evaluated.

Back to Article Outline

Results 

Demographics 

The NIS included records (weighted, 230,615) of 46,901 patients (79% men) who underwent repair of nonruptured AAAs between 1998 and 2004. The mean age of all patients within the sample was 72.2 (standard error, 0.06; range, 18-102) years. Of these, 91.5% were white, 3.1% were black, 3.4% were classified as other races, and no racial information provided for 1.8%. A further comparison of the two groups is provided in Table II.

Table II. Patient characteristics for all patients undergoing repair of abdominal aortic aneurysm between 1988 and 2004
OverallHospital volume levelP
LowMediumHigh
Patients, No. (%)46,90113,214(28.2)20,021(42.7)13,666(29.1)
Patient characteristics
Age, mean (SEM) y72.2(0.06)72.0(0.08)72.2(0.08)72.5(0.16)
Patient sex, % <.0001
Male79.327.142.730.1
Female20.729.741.428.8
Patient race, %a <.0001
White91.526.142.931.0
African American3.134.540.125.5
Hispanic2.446.927.425.7
Asian1.152.830.217.0
Native American0.124.957.417.7
Payer, % .0008
Medicare76.527.042.630.4
Private/HMO20.228.842.328.9
Medicaid/self-pay2.035.936.727.3
No charge/other1.333.947.118.9
Hospital characteristics
Teaching status, % <.0001
Nonteaching44.168.246.018.6
Teaching55.931.854.081.4
Hospital bed size, % <.0001
Small6.414.94.80.9
Medium20.337.120.04.9
Large73.248.075.294.2

HMO, Health maintenance organization; SEM, standard error of the mean.

aNo information on race was provided for 1.8%.

Univariate analysis 

Demographics 

The perioperative mortality rate for women was higher than that of men by univariate analysis, at 5.0% vs 3.0% (P < .0001). African Americans had the highest mortality of any race, at 6.2%; with whites at 3.5%, Hispanics at 4.8%, Asians at 3.7%, and Native Americans at 4.6% (overall P < .0001). Age was a significant predictor of perioperative death. The mortality rate for patients aged ≥80 years was 6.2% compared with 0.8% for patients aged <60 years, and age 60 to 80 years had an intermediate mortality rate of 3.0% (overall P < .0001).

Operative characteristics 

Between 1998 and 2004, 34,102 patients (72.7%) undergoing AAA repair had an open procedure, whereas 12,799 (72.7%) underwent EVAR. Corresponding to both the adoption of EVAR technologies by an increasing number of surgeons and the creation of a separate ICD-9 code for this procedure, two different eras of AAA repair were observed within the study period. Before 2000, conventional open repairs were done in 99.9% of patients who underwent AAA repair. Beginning in 2000, 7.8% of patients were treated with EVAR, and by 2004, 52.0% of patients were treated with EVAR vs 48.0% with open repair (Fig 1). Women were more likely to be treated through conventional open repair than men (78.5 vs 71.2%; overall χ2, P < .0001). Patients who were treated by open repair were significantly younger than patients who underwent EVAR (71.7 vs 73.5 years; P < .0001). Open AAA repair was associated with a 4.4% risk of death compared with 1.2% for EVAR (P < .0001).

Hospital volume 

Hospitals performed an average of 13.6 AAA open and endovascular repairs per year between 1998 and 2004. Hospitals were stratified according to the Leapfrog Group volume criteria. The Leapfrog Group used the yearly case volume for each hospital between 1998 and 2004 to classify 75.0% of hospitals as low volume (≤17 AAA repairs per year), 19.5% as medium volume (18 to 49), and 5.5% as high volume (≥50).

For men, 27.1% underwent repair at low-volume centers, 41.3% at medium-volume centers, and 31.6% at high-volume centers; for women, the rate was 29.7% at low-volume, 40.1% at medium-volume, and 30.2% at high-volume centers (P < .001). A larger percentage of patients aged >80 years were operated on at high-volume centers compared with patients <60 years (33.3% vs 30.6%; P = .009). Whites (32.5%) were more likely to have AAA repairs performed at high-volume centers compared with African Americans (27.0%), Hispanics (26.0%), and Asians (17.0%; P ≤ .0001).

Most nonteaching hospitals were low- and medium-volume centers (87.4%); whereas, only 56.4% of teaching hospitals were low- and medium-volume centers (P < .0001). Hospital size was associated with case volume: smaller hospitals had lower case volumes than larger hospitals (P < .0001). Rural vs urban classification was highly predictive of volume status: Only 3.4% of rural hospitals were designated as high-volume centers, whereas 29.2% of urban centers achieved this status (Table II).

Yearly trends 

An increase was observed in the mean number of AAA repairs performed per hospital between 1998 and 2004, from 11.1 per year to 15.4. Mean hospital volume rose to a greater extent in high-volume centers compared with low-volume centers. Low-volume centers performed a yearly average of 5.1 operations in 1998 and 5.3 operations in 2004. This contrasted with high-volume centers, whose yearly operative volume for AAA repair increased from 74.8 in 1998 to 84.9 in 2004 (Table III).

Table III. Trends in number of hospitals, total cases, and mean number of cases per year for AAA repair performed at low-, medium-, and high-volume centers between 1998 and 2004
Volume level1998199920002001200220032004
Low
No. of hospitals412399377366347356328
Total cases2102210218931744178818521733
No. of cases, mean/y5.15.35.04.85.25.25.3
Medium
No. of hospitals10196891049492110
Total cases2962285324972945283027313203
No. of cases, mean/y29.329.728.128.330.129.729.1
High
No. of hospitals10182233333326
Total cases748128015342890237926792208
Cases, mean/y74.871.169.787.672.179.684.9
Overall
No. of hospitals523513488503474481464
Total cases5812623559247579699772107144
No. of cases, mean/y11.112.212.115.113.815.015.4

An examination of the proportion of operations performed at low-, medium-, and high-volume hospitals according to Leapfrog Group criteria demonstrated that a higher percentage of cases were performed at low- and medium-volume centers in 1998 (36.2% and 51.0%, respectively) compared with the percentage of total cases performed in 2004 (24.3% and 44.8%). Total cases performed at high-volume centers nearly tripled between 1998 and 2004, from 12.9% to 30.9% (Fig 2). This significance was tested by Mantel-Haenszel χ2 test (P < .001).

  • View full-size image.
  • Fig 2. 

    Percentage of abdominal aortic aneurysm repairs performed by hospitals of low- (black bars), medium- (gray bars), and high-volume (white bars) classification between 1998 and 2004.

The actual number of hospitals classified as low-, medium-, and high-volume centers differed between 1998 and 2004 (P < .0001). Between 1998 and 2004, the number of hospitals designated as high-volume centers increased from 10 to 26. There was a concomitant reduction, from 412 to 328, of low-volume centers performing AAA repair, a decrease of approximately 20%. The number of medium-volume centers remained about the same (Table III).

EVAR trends 

Utilization of EVAR was correlated with volume strata. Low-volume centers were less likely to adopt endovascular techniques than high-volume centers. In 2001, low-volume centers used endovascular techniques to perform15.6% of their AAA repairs. In comparison, EVAR comprised 46.9% of operations by high-volume centers at that time. By 2004, 62.9% of AAA repairs performed at high-volume centers were endoluminal, whereas only 34.2% of cases performed at low-volume centers were endoluminal (Fig 1). In 2000, no hospitals performed ≥50 AAA repairs using EVAR technologies; however, in 2001, 13 hospitals achieved this high-volume classification for EVAR techniques. The number of high-volume EVAR centers remained similar, from 13 in 2001 to 11 in 2004, but the average number of cases performed at these centers increased yearly from 71.2 in 2001 to 85.6 in 2004. The number of medium-volume centers for EVAR stabilized by 2002 at approximately 50 centers nationwide. A small increase in the number of AAA repairs performed using EVAR at these centers was observed between 2001 and 2004.

Multivariable analysis 

Mortality rate 

A multivariable Cox proportional hazards model was performed to define the factors independently influencing in-hospital mortality after elective AAA repair (Table IV). Increasing patient age was highly predictive of an increased risk of mortality (P < .0001). The mortality rate was 0.8% for those patients aged <60, 3.1% for individuals aged 60 to 79 years, and 6.4% for patients aged ≥80 years. The risk of mortality was higher in women than in men (5.1% vs 3.1%), with an adjusted odds ratio (OR) of 1.4 (95% confidence interval [CI] 1.3-1.6; P < .0001). The presence of liver and renal failure independently increased the odds of mortality (P < .0001 for each). Diabetes mellitus appeared to have a protective effect, as patients who were diagnosed with diabetes mellitus had improved in-hospital survival compared with those with no prior diagnoses of diabetes mellitus (P = .004). A listing of the relative frequencies of comorbidities used in the Cox regression can be found in the Appendix (online only).

Table IV. Multivariate Cox proportional hazard analyses of mortality after controlling for patient and hospital-level covariates
CovariatesHR (95% CL)P
Comorbidities
No diseaseReferent
Diabetes0.74(0.6,0.9).002
Liver failure3.4(2.4,4.9)<.0001
Renal failure4.7(4.0,5.5)<.0001
CHF11.8(8.1,17.0)<.0001
Patient sex <.0001
MaleReferent
Female1.4(1.3,1.6)
Age group, years <.0001
<60Referent
60-793.3(2.1,5.1)
≥807.3(4.7,11.4)
Payer .009
MedicareReferent
Medicaid/self-pay0.9(0.6,1.4)
No charge/other1.0(0.6,1.6)
Private/HMO0.7(0.6,0.9)
Teaching hospital .77
YesReferent
No0.9(0.8,1.1)
Hospital Size .30
SmallReferent
Medium1.2(0.9,1.5)
Large1.1(0.8,1.4)
Operative volume <.0001
LowReferent
Medium0.7(0.6,0.8)
High0.6(0.54,0.7)
Year1.0(1.0,1.0)<.0001
Type of repair <.0001
OpenReferent
Endovascular0.3(0.2,0.3)

CHF, Congestive heart failure; CL, confidence limits; HMO, health maintenance organization; HR, hazard ratio.

Patients with private insurance had improved survival of 2.0% compared with 3.9% for those with Medicare (adjusted OR, 0.7; 95% CI, 0.6-0.9). Endovascular repair was independently associated with significantly reduced risk of mortality compared with open repair (adjusted OR, 0.3; 95% CI, 0.2-0.3).

Hospital size was not independently predictive of in-hospital mortality in the Cox model, with large sized at 3.3% vs small at 4.1% (adjusted OR, 1.1; 95% CI, 0.8-1.4) and medium sized at 3.9% vs small 4.1%, (adjusted OR 1.2, 95% CI, 0.9-1.5). Hospital volume for AAA affected mortality. Compared to low-volume centers, the risk of death was significantly reduced at medium-volume centers (adjusted OR, 0.7; 95% CI, 0.6-0.8) and high-volume centers (adjusted OR, 0.6; 95% CI, 0.5-0.7).

High-volume centers adopted EVAR more quickly and to a greater extent than low-volume centers. The protective effects of EVAR attenuated the effect of year on mortality in the above Cox model. The univariate analysis showed a significant trend in the reduction of patient mortality between 1998 and 2004 (Mantel-Haenszel test of trend, P < .0001; Fig 3). However, there was no effect of year in the multivariate analysis (adjusted OR, 1.0; 95% CI 1.0-1.0).

  • View full-size image.
  • Fig 3. 

    The mortality rate in patients undergoing abdominal aortic aneurysm repair was significantly reduced between 1998 and 2004 (Mantel-Haenszel test of trend, P < .0001). Shown in heavy solid line is the overall mortality, light gray lines represent open repair, and dashed lines represent endovascular aneurysm repair (EVAR).

Back to Article Outline

Discussion 

This is the first published report, to our knowledge, to demonstrate that the regionalization of elective AAA repair has occurred within the United States. Using a large national data set, we showed that high-volume centers performed an increasing proportion of AAA repairs from 1998 to 2004 and that the number of high-volume centers rose with a concurrent fall in low-volume centers. It is unlikely that these findings are solely a result of low-volume and medium-volume hospitals increasing their operative volume. In 1998, 412 centers were in the low-volume strata; however, by 2004 this number was 328. This reduction of nearly 100 hospitals in the low-volume strata corresponded to the formation of only 16 new high-volume centers and an unchanging number of medium-volume centers. The underlying reasons driving the shift in cases from low-volume centers to high-volume centers are likely multifactorial and may include centralizing forces secondary to the introduction of EVAR or regionalization of care due to the relationship between hospital volume and outcome, or both.

The introduction and rapid adoption of endovascular aortic stents for the repair of AAAs occurred during our study period. The FDA approved the first device for EVAR in 1999, and a separate ICD-9 code was created in 2000. Therefore, most AAA repairs performed before 2000 can be assumed to be open repairs. Owing to the high initial start-up costs associated with endovascular repairs, we propose that low-volume centers were initially less likely to adopt endovascular techniques than high-volume centers. In 2004, high-volume centers were nearly twice as likely to perform a given AAA repair by endoluminal techniques as low-volume centers (62.9% of AAAs vs 34.2%).

Although a causal relationship cannot be proven, we propose that the introduction of EVAR technology greatly influenced the regionalization of elective AAA repairs more than any other factors because the growth in high-volume centers was temporally related to the introduction of endovascular techniques. Anderson et al20 reported in 2004 that the adoption of EVAR in New York state was first seen at large academic centers; however, they demonstrated a rapid diffusion of EVAR to nonacademic centers. In contrast, we found that that a greater proportion of academic centers compared with low-volume centers offered EVAR through the study period. Our findings are similar to Forbes et al,16 who showed a centralization of AAA patients within Canada; yet, our findings are the first of their kind to show a reduction in the number of low-volume centers.

The observed in-hospital mortality for AAA repair was 3.5%. Similar to prior reports,2, 4, 5, 21 we found a significant improvement in survival for patients who were treated at high-volume centers compared with medium- and low-volume centers. In our study, mortality was 4.9% at low-volume centers and 2.7% for high-volume centers. Starting in the late 1990s, the volume-outcome relationship became a topic of immense interest, and various publications examined the feasibility of the regionalization of AAA repair.22 Utilizing these data, Dudley et al23 analyzed AAA repair and 10 other conditions and concluded that approximately 600 deaths in California could be prevented through the regionalization of care.

First in 2001,6 and then again in 2003, the Leapfrog Group recommended to its members that complex operations should be performed at high-volume centers.7 McPhee et al24 concluded in 2007 that regionalization of pancreatectomy may have occurred between 1998 and 2003 due to calls for the centralization of this procedure. We found that the public pressure exerted by the development of the volume–outcome relationship may have contributed to the regionalization of AAA repair, but it appears that the adoption of EVAR technology has played a more significant role. Etiologic factors cannot be definitively established in this observational study.

Our data suggest that as the regionalization of AAA repair occurred, a simultaneous reduction took place in in-hospital mortality from 4.2% to 2.6%, an annual improvement of 9.1%. The lower perioperative mortality rate of 4.4% for EVAR compared with 1.2% for traditional open repair observed in this study agrees with prior reports11, 12, 13, 25 and was likely a factor in the temporal improvements in patient mortality after AAA repair. Specifically, as the number of AAAs treated by aortic stent grafts increased, the overall mortality rate for AAA repair would have dropped because the survival advantage for EVAR was more prominent.

Another factor leading to the observed reduction of patient mortality after AAA repair may have been regionalization of care to high-volume centers. The referral of patients to high-volume centers and the introduction of EVAR were intimately related. High-volume centers were more likely to perform EVAR compared with low-volume centers; thus, it is impossible to determine whether the improvements in survival demonstrated between 1998 and 2004 were secondary to more patients being treated at high-volume centers or due to an increased likelihood that patients would undergo endovascular repair.

We found, however, that an increasing proportion of patients were referred to high-volume centers and that these centers were more likely to perform endovascular repair. We propose that referring physicians may have referred additional patients to high-volume centers specifically for endovascular treatment. We further conclude that if patients are referred to higher-volume centers for the purpose of receiving novel types of treatment, such as EVAR, then a type of regionalization has occurred and that this regionalization subsequently improved patient survival.

We have shown that the regionalization of patients undergoing AAA repair was associated with a reduction in patient mortality; however, widespread regionalization may have some potential disadvantages. Forbes et al16 argue that the regionalization of elective AAA repair in Canada has led to long travel times for patients. Others have hypothesized that increased regionalization may lead to a reduction in patient survival secondary to worsening of patient access to needed treatments. Petersen et al26 demonstrated that patients followed up exclusively within the Veterans Administration (VA) system were less likely to undergo angiography compared with veterans who were followed at fee-for-service institutions. They concluded that this was due to the regionalization of the VA system and the subsequent lack of on-site angiography at most VA centers.26

Another area in which the regionalization of elective AAA repair could have a deleterious effect is in the care of ruptured AAA. If smaller hospitals are performing fewer elective AAA repairs, then the care of patients with ruptured AAA, who are unable to be transported to high-volume centers, may be deleteriously affected due to waning experience.

Our finding that women have significantly increased mortality after AAA repair corresponds to previously published work.25, 27 The finding that nonwhites were more likely to undergo operation at low-volume centers may help to explain historically poorer outcomes for nonwhites who have vascular or cardiovascular procedures.28, 29, 30 This inequity warrants further study, because reducing the number of nonwhites who undergo operation at low-volume centers may improve overall survival in this demographic group.

The protective effect of diabetes was unexpected. Possible factors contributing to this should be further investigated and are likely to be secondary to bias within the data set. One possibility is that patients with diabetes are closely monitored by physicians, leading to the observed survival benefit.

Our data also show that high-volume centers are treating a greater proportion of older patients. One possible explanation for this observation could be that surgeons at low-volume centers selectively refer older, more complex patients to larger-volume centers. The advent of EVAR may have exaggerated this tendency because the benefits of EVAR compared with open AAA repair are more pronounced in elderly and high-risk patients.

This study is by definition limited by the confines of the NIS. As with any administrative database, differentiating comorbid diseases from perioperative complications is difficult.31, 32 However, the comorbidity software used in this study has been previously validated for large data sets.18, 19 Ensuring equal case mix between volume clusters is also impossible. However, since 1998 the NIS has been weighted to provide a consistent mix of hospitals.17 The present study used in-hospital mortality as the main outcome measure. Long-term survival is unavailable from the NIS. Other important considerations, such as functional outcomes, are difficult if not impossible to determine from large national databases.

Despite these limitations, this study observed a trend of improved survival for elective AAA repair that is consistent with the introduction of EVAR technology and the regionalization of elective AAA repairs to higher-volume centers. We believe that our study suggests that the development of new technology at high-volume centers may create a mechanism for these hospitals to offer safer operations compared with low-volume centers. Furthermore, the development of new technology may provide an impetus for the selective referral of patients to these “cutting-edge” high-volume centers, possibly playing a more significant role in regionalization than factors such as the recommendations from groups within the health care system.

Back to Article Outline

Conclusion 

Examining the time frame between 1998 and 2004, this study is the first to demonstrate that regionalization of AAA repair has occurred in the United States. Whether this regionalization is secondary to the introduction of EVAR or a consequence of external forces promoting regionalization requires further research. We have shown that the regionalization of AAA repair was temporally related to a clinically significant improvement in patient survival. Continued regionalization of care or the development of newer technologies, or both, may provide a mechanism to further improve patient morbidity and mortality.

Back to Article Outline

Author contributions 


Conception and design: JH, LM, ME

Analysis and interpretation: JH, JM, LM, RC

Data collection: JH

Writing the article: JH, LM, ME

Critical revision of the article: JH, JM, LM, RC, ME

Final approval of the article: JH, ME

Statistical analysis: JH, JM

Obtained funding: Not applicable

Overall responsibility: ME

Back to Article Outline

Appendix (online only) 

Appendix (online only). Patient comorbidities for all patients undergoing repair of an abdominal aortic aneurysm between 1988 and 2004
ComorbidityOverallHospital volume level
LowMediumHigh
Diabetes mellitus, %10.39.410.510.8
Liver failure, %0.90.90.90.7
Renal failure, %3.73.73.83.4
Congestive heart failure, %0.40.50.40.4

Back to Article Outline

References 

  1. Begg CB, Cramer LD, Hoskins WJ, Brennan MF. Impact of hospital volume on operative mortality for major cancer surgery. JAMA. 1998;280:1747–1751
  2. Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA, Lucas FL, Batista I, et al. Hospital volume and surgical mortality in the United States. N Engl J Med. 2002;346:1128–1137
  3. Rouvelas I, Lindblad M, Zeng W, Viklund P, Ye W, Lagergren J. Impact of hospital volume on long-term survival after esophageal cancer surgery. Arch Surg. 2007;142:113–117
  4. Dimick JB, Stanley JC, Axelrod DA, Kazmers A, Henke PK, Jacobs LA, et al. Variation in death rate after abdominal aortic aneurysmectomy in the united states: Impact of hospital volume, gender, and age. Ann Surg. 2002;235:579–585
  5. Killeen SD, Andrews EJ, Redmond HP, Fulton GJ. Provider volume and outcomes for abdominal aortic aneurysm repair, carotid endarterectomy, and lower extremity revascularization procedures. Journal of Vascular Surgery. 2007;45:615–626
  6. Birkmeyer JD, Finlayson EVA, Birkmeyer CM. Volume standards for high-risk surgical procedures: Potential benefits of the leapfrog initiative. Surgery. 2001;130:415–422
  7. Birkmeyer J. Potential benefits of the new Leapfrog standards: effect of process and outcomes measures. Surgery. 2004;135:569–575
  8. Parodi J, Palmaz JC, Barone HD. Transfemoral intraluminal graft implantation for abdominal aortic aneurysms. Annals of vascular surgery. 1991;5:491–499
  9. Volodos' N. [A self-fixing synthetic blood vessel endoprosthesis] (Vestnik khirurgii im. I.I). Grekova. 1986;137:123–125
  10. Arko F, Lee WA, Hill B, Olcott C, Dalman RL, Haerris EJ, et al. Aneurysm-related death: Primary endpoint analysis for comparison of open and endovascular repair. Journal of Vascular Surgery. 2002;36:297–304
  11. Blankensteijn JD, de Jong SE, Prinssen M, van der Ham AC, Buth J, van Sterkenburg SM, et al. Dutch Randomized Endovascular Aneurysm Management (DREAM) Trial Group Two-year outcomes after conventional or endovascular repair of abdominal aortic aneurysms. N Engl J Med. 2005;352:2398–2405
  12. Greenhalgh RM, Brown LC, Kwong GP, Powell JT, Thompson SG 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;364:843–848
  13. Endovascular aneurysm repair and outcome in patients unfit for open repair of abdominal aortic aneurysm (EVAR trial 2): Randomised controlled trial. The Lancet. 2005;365:2187–2192
  14. Lee LK, Faries PL. Assessing the effectiveness of endografts: clinical and experimental perspectives. J Vasc Surg. 2007;45:A123–A130
  15. Nowygrod R, Egorova N, Greco G, Anderson P, Gelijns A, Moskowitz A, et al. Trends, complications, and mortality in peripheral vascular surgery. J Vasc Surg. 2006;43:205–206
  16. Forbes T, Lawlor DK, DeRose G, Harris K. Examination of the trend in canada toward geographic centralization of aneurysm surgery during the endovascular era. Ann Vasc Surg. 2006;20:63–68
  17. HCUP Nationwide Inpatient Sample (NIS). Healthcare cost and utilization project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality; 1998;www.hcup-us.ahrq.gov/nisoverview.jspAccessed Feb 19, 2008
  18. HCUP comorbidity software. Healthcare cost and utilization project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality; 2000;http://www.hcup-us.ahrq.gov/toolssoftware/comorbidity/comorbidity.jspAccessed Feb 19, 2008
  19. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27
  20. Anderson P, Arons RR, Moskowitz AJ, Gelijns A, Magnell C, Faries PL. A statewide experience with endovascular abdominal aortic aneurysm repair: Rapid diffusion with excellent early results. J Vasc Surg. 2004;39:10
  21. Manheim L, Sohn M, Feinglass J, Ujiki M, Parker M, Pearce W. Hospital vascular surgery volume and procedure mortality rates in California, 1982-1994.
  22. Urbach DR, Bell CM, Austin PC. Differences in operative mortality between high- and low-volume hospitals in Ontario for 5 major surgical procedures: estimating the number of lives potentially saved through regionalization. CMAJ. 2003;168:1409–1414
  23. Dudley RA, Johansen K, Brand R, Rennie D, Milstein A. Selective referral to high-volume hospitals: Estimating potentially avoidable deaths. JAMA. 2000;283:1159–1166
  24. McPhee JT, Hill JS, Whalen GF, Zayaruzny M, Litwin DE, Sullivan ME, et al. Perioperative mortality for pancreatectomy: A national perspective. Ann Surg. 2007;246:246–253
  25. McPhee JT, Hill JS, Eslami MH. The impact of gender on presentation, therapy, and mortality of abdominal aortic aneurysm in the United States, 2001-2004. J Vasc Surg. 2007;45:891–899
  26. Petersen LA, Normand SL, Leape LL, McNeil BJ. Regionalization and the underuse of angiography in the Veterans Affairs Health Care System as compared with a fee-for-service system. N Engl J Med. 2003;348:2209–2217
  27. Katz D, Stanley J, Zelenock G. Gender differences in abdominal aortic aneurysm prevalence, treatment, and outcome. J Vasc Surg. 1997;25:561–568
  28. Gornick M, Eggers PW, Reilly TW, Mentnech RM, Fitterman LK, Kucken LE, et al. Effects of race and income on mortality and use of services among medicare beneficiaries. N Engl J Med. 1996;335:791–799
  29. Feinglass J, Rucker-Whitaker C, Lindquist L, McCarthy W, Pearce W. Racial differences in primary and repeat lower extremity amputation: results from a multihospital study. J Vasc Surg. 2005;41:823–829
  30. Eslami M, Zayaruzny M, Fitzgerald G. The adverse effects of race, insurance status, and low income on the rate of amputation in patients presenting with lower extremity ischemia. J Vasc Surg. 2007;45:55–59
  31. Jencks SF, Williams DK, Kay TL. Assessing hospital-associated deaths from discharge data (the role of length of stay and comorbidities). JAMA. 1988;260:2240–2246
  32. Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heeren T. Comorbidities, complications, and coding bias (does the number of diagnosis codes matter in predicting in-hospital mortality?). [see comment] JAMA. 1992;267:2197–2203

 Competition of interest: none.

 Additional material for this article may be found online at www.jvascsurg.org.

PII: S0741-5214(08)00301-7

doi:10.1016/j.jvs.2008.02.048

Journal of Vascular Surgery
Volume 48, Issue 1 , Pages 29-36.e1, July 2008

Article Tools

* Image Usage & Resolution