Endovascular technology, hospital volume, and mortality with abdominal aortic aneurysm surgery
Article Outline
Objective
To determine whether the introduction of endovascular technology changed the relationship of hospital volume to mortality with abdominal aortic aneurysm repair.
Methods
Data from all hospitals in the United States that performed abdominal aortic aneurysm surgery on Medicare patients from 2001 to 2003 were obtained from the national Medicare database. The primary outcome variable was death ≤30 days of operation or before hospital discharge. We determined the effect of total hospital volume on operative mortality for all types of repair and for endovascular and open repair separately. All analyses were adjusted for patient risk using logistic regression.
Results
The proportion of abdominal aortic aneurysms repaired with an endovascular approach increased from 27% to 39% during the 3-year study period. Hospital volume was significantly related to operative mortality in all comparisons. Mortality rates were 80% higher at hospitals in the lowest vs the highest quartile of total volume (odds ratio [OR], 1.81; 95% confidence interval [CI], 1.62-2.04) when considering all types of repair together. A similar relationship between total hospital volume and mortality was found when separately examining open repair (OR, 1.52; 95% CI, 1.33-1.73) and endovascular repair (OR, 1.68; 95% CI, 1.32-2.22). Higher-volume hospitals were more likely to use the endovascular approach. The highest-volume hospitals used the endovascular approach 44% of the time compared with only 18% at the lowest-volume hospitals. This greater use of the endovascular procedure at high-volume hospitals accounted for 37% of the difference in mortality between high- and low-volume hospitals.
Conclusion
As the endovascular repair becomes more widespread, the relationship between hospital volume and operative mortality still remains. High-volume hospitals are more likely to use the endovascular approach, and this explains a significant portion of the observed impact of hospital volume on mortality.
The Unites States (US) health care system is presently undergoing an unprecedented change, with an increased focus on holding providers accountable for their outcomes. Patients are turning to the Internet and other sources to find information on hospitals and physicians.1 Payers are ramping up efforts to realign incentives to reward high quality.2, 3 Providers are scrambling to keep pace with these external pressures and are actively seeking input to ensure they are fair and make sense from a clinical perspective.
Elective abdominal aortic aneurysm (AAA) repair is often the focus of surgical quality assessment and improvement activities.2, 3 In large part, the interest in this operation comes from the simple fact that it is both common and high-risk, and ensuring optimal outcomes with this operation would avoid many preventable deaths. Much of the enthusiasm also comes from the large body of evidence showing variations in outcomes for this procedure, particularly between high- and low-volume providers. This evidence has not gone unnoticed by payers and policy makers. The Leapfrog Group, a coalition of health care purchasers, includes volume standards for AAA repair in its Evidence Based Hospital Referral initiative; and the Agency for Healthcare Research and Quality (AHRQ) includes information on mortality rates and hospital volume in its Inpatient Quality Indicators.2, 3, 4
Despite continued emphasis on volume as a proxy for quality, most of the evidence linking volume to outcome for AAA repair is outdated. Most important, very little data exist on the relationship of volume to outcome after the introduction of endovascular technology. In this article, we will update previous analyses and examine the relationship of hospital volume to operative mortality using the national Medicare population during the endovascular era.
Methods
Data source
We used 100% national analytic files from the Center for Medicare and Medicaid Services for years 2001 through 2003. Medicare Provider Analysis and Review (MEDPAR) files, which contain hospital discharge abstracts for all fee-for-service acute care hospitalizations of all US Medicare recipients, were used to create our main analytical data sets. The Medicare eligibility file was used to assess patient vital status at 30 days. Patients were included if they were aged 65 to 99 years.
We used the appropriate procedure and diagnostic codes to identify all patients in the data set who underwent either open or endovascular repair of an intact AAA during 2001 to 2003. Patients with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) primary procedure code for resection of abdominal aorta with replacement (ICD-9-CM 38.44) or endovascular implantation of a graft in the abdominal aorta (ICD-9-CM 39.71) were initially selected. We then selected for inclusion in the data sample those patients who also had a primary diagnostic code for AAA without mention of rupture (ICD-9-CM 441.4). We excluded patients with a diagnostic code for ruptured AAA (ICD-9-CM 441.3).
Mortality rates and hospital volume
We defined operative mortality as those deaths that occurred either ≤30 days of operation or before discharge. We examined the effect of hospital volume on operative mortality for overall repair and then stratified by the type of repair (open and endovascular). We calculated the annual hospital volume for each hospital during each year of the 3-year study period. For the present analyses, we converted the volume variable into a categoric variable by creating four equally sized patient groups (quartiles) for each volume measure (total, open, and endovascular volume). The volume thresholds for the total volume variable were low volume, ≤40; medium volume, 41 to 81; high volume, 82 to 154; and very high volume, 155 to 656. The volume thresholds for open volume were low volume, ≤24; medium volume, 25 to 49; high volume, 50 to 88; and very high volume, 89 to 405. The volume thresholds for endovascular volume were low volume, ≤23; medium volume, 24 to 47; high volume, 48 to 94; and very high volume, 96 to 430.
Statistical analysis
We compared patient characteristics for patients having open and endovascular repair using χ2 and t tests, where appropriate. We used multiple logistic regression analysis to study the relationship between our dependent (operative mortality) and exposure (hospital volume) variables after adjusting for other potentially confounding variables. In these analyses, we adjusted for patient characteristics, including age, sex, race, admission acuity (elective, urgent, or emergency), median income, and coexisting diseases. The coexisting diseases were obtained from the secondary diagnostic codes according to the methods validated by Elixhauser et al.5 We accounted for the non-independence of patients within hospitals by calculating robust variance estimates designed to deal with clustering of this nature.
We estimated the proportion of the hospital volume effect attributable to endovascular repair by running a logistic regression model with and without the variable for type of repair.6 The relative attenuation of the odds ratio (OR) was computed as [OR − ORtype] / [OR − 1], where OR is the OR for mortality with a given hospital volume without consideration of the type of repair, and ORtype is the ORfor mortality with a given hospital volume after adjustment for the type of repair; both ORs were adjusted for patient characteristics and other characteristics of the hospital. All statistical analyses were performed using STATA 8.0 software (StataCorp, College Station, Tex).
Results
Trends in the use of endovascular repair
During the 3-year study period, 2001 to 2003, 80,953 Medicare patients underwent AAA repair, and an endovascular approach was used in 26,750 (33%). Patients who had endovascular repair were older, more likely to be men, and more likely to be white (Table I). Although most admissions for both types of repair were elective, patients who had endovascular repair were more likely to be elective (Table I). During the study period, endovascular repair assumed a larger proportion of the market share, from 27% to 39% of repairs in Medicare patients, (P < .001). The increase in use of the endovascular approach was greater among older patients compared with younger patients (Fig 1). For patients aged >85, the use of an endovascular repair increased from 39% to 55% from 2001 to 2003 (P < .001). In contrast, for patients aged 65 to 75, the increase was only from 22% to 33% of total repairs (P < .001).
Table I. Characteristics of Medicare patients undergoing open and endovascular repair of abdominal aortic aneurysm from 2001 to 2003
| Characteristic | Open | Endovascular | P |
|---|---|---|---|
| Patient, total | 54,302 | 26,750 | |
| Patient age, mean (SD) years | 74 (5) | 76(6) | <.001 |
| Age >80 years, No. (%) | 11,219(20) | 8,297(31) | <.001 |
| Female sex, No. (%) | 16,917(31) | 4,674(17) | <.001 |
| Nonwhite race, No. (%) | 2674(5) | 827(3) | <.001 |
| Urgent/emergency repair, No. (%) | 12,654(23) | 3681(13) | <.001 |
| >2 coexisting diseases, No. (%) | 35,331(65) | 14,793(55) | <.001 |
Fig 1.
Increasing proportion of abdominal aortic aneurysm repairs performed by an endovascular approach (2001 to 2003). Age 65 to 75, squares; age 76 to 85, triangles; age >85, circles.
Of the hospitals included, 4% were categorized as very high volume, 8% as high volume, 7% as medium volume, and 75% as low volume (Table II). Higher-volume hospitals (based on total volume) were much more likely to use the endovascular approach than lower-volume hospitals (Fig 2). The highest-volume hospitals used the endovascular approach 44% of the time compared with only 18% at the lowest-volume hospitals (Fig 2). There was a strong correlation between total volume and both open volume (Spearman ρ = 0.96, P < .001) and endovascular volume (Spearman ρ = 0.80, P < .001). Most high-volume and very high-volume hospitals for total volume were also high or very high for both endovascular and open volume (Table II).
Table II. Relationship between total hospital volume and both open and endovascular volume
| Volume rating | Average annual total | Hospitals, No. (%) | Volume quartiles | |
|---|---|---|---|---|
| Open | Endovascular | |||
| Very high | 155-656 | 86(4) | 65 very high | 36 very high |
| 19 high | 37 high | |||
| 2 medium | 11 medium | |||
| 2 low | ||||
| High | 82-154 | 180(8) | 36 very high | 7 very high |
| 103 high | 58 high | |||
| 37 medium | 76 medium | |||
| 4 low | 39 low | |||
| Medium | 41-81 | 347(15) | 83 high | 8 high |
| 207 medium | 104 medium | |||
| 57 low | 218 low | |||
| Low | <41 | 1714(73) | 141 medium | 8 medium |
| 1547 low | 753 low | |||
Fig 2.
Proportion of abdominal aortic aneurysm repairs performed by an endovascular approach across total hospital volume categories (2001 to 2003).
Mortality rates for open and endovascular repair
Endovascular repair was associated with a much lower in-hospital mortality rate compared with open repair (6.6% vs 2.5%, P < .001). Older patients had much higher in-hospital mortality rate compared with younger patients for both types of repair (P < .001 for all comparisons; Fig 3). However, the difference in mortality rates in the oldest age group is much larger than in other age groups, at 4.5% for endovascular repair vs 14.3% for open repair (P < .001). Mortality rates did not change significantly during the 3-year period for either open (6.6% to 6.5%) or endovascular (2.7% to 2.4%) repair. Overall, patients undergoing urgent/emergency repair experienced higher mortality rates than those undergoing elective repair (3.9% vs 10.3%, P < .001). Mortality rates were lower for endovascular repair with both elective (2.1% vs 5.0%, P < .001) and urgent/emergency (5.1% vs 11.2%) operation.
Fig 3.
Relationship of patient age to in-hospital mortality after open and endovascular aneurysm repair. Age 65 to 75, light gray; age 76 to 85, medium gray; age >85, black.
Hospital volume and mortality
Fig 4 shows the relationship of total hospital volume to risk-adjusted operative mortality for all repairs, open repair, and endovascular repair. We found a strong inverse relationship between hospital volume and risk-adjusted mortality rates in all three comparisons (P < .001 for all comparisons). In this risk-adjusted analysis, mortality rates were nearly twofold higher at hospitals in the lowest vs the highest quartile of total volume when considering all types of repair together (OR, 1.81; 95% confidence interval [CI], 1.62-2.04). A similar relationship between total hospital volume and mortality was found when open repair (OR, 1.52; 95% CI, 1.33-1.73) and endovascular repair (OR, 1.68; 95% CI, 1.32-2.22) were examined separately. The relationship between operation-specific volume and mortality was similar to that observed for total volume with both approaches.
Fig 4.
Relationship of total hospital volume to operative mortality for abdominal aortic aneurysm repair.
Because high-volume hospitals perform more endovascular repairs, we conducted an analysis of volume and mortality adjusting for the type of repair. After this adjustment, the low-volume hospitals still had a 50% higher mortality rate compared with the highest-volume hospitals (OR, 1.52; 95% CI, 1.35-1.72). We can determine from this analysis that the higher use of the endovascular approach explains 37% of the apparent effect of hospital volume on operative mortality.
Discussion
We have demonstrated that the introduction of endovascular technology has not meaningfully altered the observed impact of hospital volume on operative mortality. We found a strong, inverse relationship between total hospital volume and operative mortality for all types of repair together and for both types alone. We also demonstrated that higher-volume hospitals are more likely to use the endovascular approach; and this difference in use of the lower-risk endovascular procedure accounts for 37% of the differences in mortality between high- and low-volume hospitals.
Although the relationship of hospital volume to mortality for AAA surgery is well documented, to our knowledge, we are the first to perform a systematic analysis since the introduction of the endovascular approach. This updated analysis is important because many existing quality assessment and improvement activities use volume as a proxy for quality for AAA surgery.2, 3 If the relationship no longer exists, then these efforts would be wasting time and resources that would ideally be put to another use to improve quality. It is apparent from the results of the present study that hospital volume is still a valid quality indicator for AAA surgery.
The present study also proves that changing referral patterns have not altered the apparent impact of volume on outcome. Many thought that endovascular technology would change referral patterns, altering case-mix distribution and potentially diminishing the apparent volume effect. If high-volume referral centers began repairing more complex pararenal and suprarenal aneurysms, they would likely have higher mortality rates and this would obscure any difference in mortality between high- and low-volume centers. But our data show that high-volume hospitals are doing fewer open repairs than they previously did—because they are performing a large proportion of endovascular repairs—and they still have very low mortality rates for open repair. These data clearly show that high-volume hospitals are performing well despite any changes in referral patterns.
It will still be important to continually investigate the relationship between volume and outcome as endovascular technology continues to take up a larger share of the market. It is possible that the relationship of volume to mortality observed in the present study is attributable, at least in part, to hospitals being at different positions on the learning curve. Some low-volume hospitals may be later adopters of endovascular technology, and they may still be early on their institutional learning curve. In contrast, some high-volume hospitals may have been early adopters of this new technology and are further along on the learning curve. As the data mature over the next 5 to 10 years, it will become evident whether the differences we are observing now are due to these relative differences in position on the learning curve or are true, steady-state volume/mortality relationships.
It is also necessary to continue to investigate the outcomes of this operation because it is possible that endovascular technology may change the existing relationship between volume and outcome for AAA repair. As a new technology is introduced and refined, it may actually change the learning curve and alter the effect of volume on outcome for an operation. For instance, patients who undergo endovascular repair are less likely to need the vast resources, such as intensive care units, high nurse–patient ratios, and other technologies that are thought to explain the effect of volume on outcome for open AAA surgery. If these resources are not needed, and the profession continues to move along a collective learning curve, the effect of volume on outcome for this operation may diminish over time.
The present study has several limitations. First, we used an administrative data source, which has limited ability to adjust for patient case-mix. Although this limitation is clearly present, this is a problem with almost all studies of volume on outcome. To obtain data on a broad range of hospitals, a population-based data set is needed. There is currently no registry for vascular surgery that includes a broad enough sample of hospitals to conduct such a study. Further, those few studies that use detailed clinical data for risk-adjustment still show a volume effect.7
A second limitation is that we used the Medicare population rather than an all-payer database. However, most patients undergoing AAA repair are elderly and included in the present study. In addition, previous studies on the volume-outcome effect using all-payer data sets and Medicare data sets yield entirely consistent results.7, 8
A final limitation is our focus on operative mortality as the main outcome variable in this study. Death represents only a single dimension of the quality of care for patients with AAA surgery. Of course, most would agree it is a very important one. Nonetheless, other aspects of care are important for a full accounting of quality. In particular, the reintervention rate would be an important outcome to consider, particularly when considering the quality of care for endovascular repair. Lifetime surveillance and repeated interventions are often needed after endovascular aneurysm repair. Unfortunately, this information is not available in the Medicare Part A claims data. Because this study includes both endovascular and open repair, and the reintervention rate is less important for open repair, we believe the operative mortality rate is a valid and useful quality measure.
Although the present study demonstrates that the volume effect is still present, it does little to resolve the ongoing debate about using volume as a proxy for quality. Many argue that hospital volume is a crude proxy for quality and should not be used as a performance measure.9, 10 This argument does have merit. Some low-volume hospitals may have good performance and some high-volume hospitals may provide poor quality care. Because the volume-outcome effect only holds true on average (for large groups of hospitals), volume is not good at discriminating between individual hospitals or surgeons.
Unfortunately, there are not many good alternative measures of surgical performance. Some argue we should directly measure outcomes using risk-adjusted mortality rates. However, most hospitals do not perform enough cases to measure mortality with precision—the so-called “big problem with small samples.”11, 12 Process measures are widely used to measure the quality of medical diagnoses, but the evidence base does not exist for AAA surgery. Until the evidence-base becomes more mature, using process measures is also not an option. Using hospital volume as a proxy for quality with AAA repair has several advantages compared with other approaches.2 Hospital volume is easily determined from readily available data sources. Hospital volume is also meaningful to patients. Unlike many complex quality measures used in some hospital report cards, patients can easily grasp the value of “experience” with a specific procedure.
Conclusion
Information on hospital volume can be used to improve the quality of care for abdominal aortic surgery in two ways. First, patients can be steered towards high-volume hospitals by using public reporting or various financial incentives. As the most visible of these efforts, the Leapfrog Group represents many large employers interested in creating incentives that result in directing more patients to the highest-quality hospitals.2, 3
Second, we can improve the quality of care at all hospitals through collaborative quality improvement. We know that hospital volume in and of itself does not produce better outcomes; underlying processes of care that differ between high- and low-volume settings are responsible for the differences in outcomes. If we can discover these details, we can disseminate them and improve the quality of care for all patients in need of elective abdominal aortic surgery.
It is not known which approach, selective referral or quality improvement, will yield the greatest results. We do know that it depends on the extent to which the details of care that lead to better outcomes can be determined. If discrete processes leading to the best outcomes can be measured and exported to all hospitals, quality improvement will be the most effective. But if the variations between high- and low-volume hospitals are due to skill and processes that cannot be measured, then selective referral would be most effective. But patients deserve the best care we can provide, whether it is delivered to them or they are delivered to it.
Author contributions
References
- . Publicly available hospital comparison web sites: determination of useful, valid, and appropriate information for comparing surgical quality. Arch Surg. 2007;142:863–868
- . Measuring surgical quality: what's the role of provider volume?. World J Surg. 2005;29:1217–1221
- . Potential benefits of the 2003 Leapfrog standards: effect of process and outcomes measures. Surgery. 2004;135:56975
- AHRQ quality indicators—guide to inpatient quality indicators: quality of care in hospitals—volume mortality, and utilization. Rockville, Md: Agency for Healthcare Research and Quality; 2002;AHRQ Publication 02-R0204
- . Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27
- . Surgeon volume and operative mortality in the United States. N Engl J Med. 2003;349:2117–2127
- . Is volume related to outcome in health care? (A systematic review and methodologic critique of the literature). Ann Intern Med. 2002;137:511–520
- Hospital volume and surgical mortality in the United States. N Engl J Med. 2002;346:1128–1137
- . Invited commentary: Surgeons, not General Motors, should set standards for surgical care. Surgery. 2001;130:429–431
- . The volume-outcome relationship: don't believe everything you see. World J Surg. 2005;29:1241–1244
- . Surgical mortality as an indicator of hospital quality: the problem with small sample size. JAMA. 2004;292:847–851
- . Identifying poor-quality hospitals (Can hospital mortality rates detect quality problems for medical diagnoses?). Med Care. 1996;34:737–753
Competition of interest: Dr Dimick has served as a paid research consultant for the Leapfrog Group, a coalition of private payers, who advocate selective referral to high-volume hospitals for abdominal aortic aneurysm repair and other high-risk operations.
PII: S0741-5214(08)00192-4
doi:10.1016/j.jvs.2008.01.054
© 2008 The Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved.
