Background

The Dartmouth Atlas Project provides ongoing population-based monitoring of the rates of use of medical and surgical services in the fee-for-service Medicare population. This report focuses on trends and regional variations in abdominal aortic aneurysm (AAA) repair.

Until the late 1990s, repair of abdominal aneurysms required a major intra-abdominal procedure that carried substantial risks of death and complications. On September 28, 1999, the Food and Drug Administration approved for widespread use an endovascular stentgraft that made it possible to carry out some AAA repairs without an open procedure. Initial randomized trials demonstrated that the endovascular approach -- at least in the setting of the clinical trials -- carried substantially lower short-term risks than the open procedure. Many observers hypothesized that the introduction of the lower-risk procedure would lead to a substantial increase in the overall number of procedures performed, as surgeons would be able to offer the new procedure to patients with aneurysms in need of repair who were felt to face unjustifiably high risks from the open procedure.

This report is the first of a series of studies that will examine the impact of the introduction of endovascular AAA repair on utilization and outcomes. The data are drawn from Medicare inpatient hospital discharge records (MedPAR files) and the Medicare Denominator Files from 1992 through 2003. The primary analyses are based on the standard analytic files that are made available on an annual basis, generally about six to nine months after the end of the calendar year.

National Trends in Utilization and Spending

Figure 1 presents national trends in the overall rate of AAA repair and the rates of open and endovascular procedures. Overall, utilization rates of AAA repair have been relatively stable throughout the decade, peaking in 1997. Even before the introduction of the endovascular procedure, rates of the open procedure had begun to fall. The introduction of endovascular repair at the end of 1999 was followed by a rapid increase in the use of the procedure in 2000 and 2001, with a rapid decline in population-based rates of open repair. The combined effect was a modest increase in overall rates of AAA repair in 2001, followed by a modest decline back to a level similar of the mid-1990s.

Figure 1

Figure 2 presents national trends in Medicare Part A inpatient payments for AAA repair. The introduction of the endovascular approach has not been associated with an increase in spending on AAA repair. Rather, since 2001, per-capita inpatient spending on AAA repair has fallen modestly.

Figure 2

Regional Variations in Rates of Abdominal Aortic Aneurysm Repair

The following figures and maps present data on regional variations in the rates of AAA repair procedures. Because rates are relatively low overall, many hospital referral regions have rates that are either below the threshold for suppression required to protect the privacy of Medicare beneficiaries or are statistically too imprecise to include. We therefore combined two years of data when describing these rates at the regional level.

Maps 1, 2, and 3 present the population-based rates of AAA repair in all U.S. hospital referral regions. The relationship between rates of open AAA repair and endovascular AAA repair is not at all obvious in the maps.

Map 1.

Ratio of Rates of Abdominal Aortic Aneurysm Repair (Open plus Endovascular) to the U.S. Average (2002-03). In 27 hospital referral regions, rates of AAA repair were at least 30% higher than the United States average of 1.0 per 1,000 Medicare enrollees. (more...)

Map 2.

Ratio of Rates of Open Abdominal Aortic Aneurysm Repair to the U.S. Average (2002-03). In 44 hospital referral regions, rates of open AAA repair were at least 30% higher than the United States average of 0.6 per 1,000 Medicare enrollees. In 40 hospital (more...)

Map 3.

Ratio of Rates of Endovascular Abdominal Aortic Aneurysm Repair to the U.S. Average (2002-03). In 51 hospital referral regions, rates of endovascular AAA repair were at least 30% higher than the United States average of 0.4 per 1,000 Medicare enrollees. (more...)

In fact, as shown in Figure 3, there was virtually no correlation between rates of open and endovascular AAA repair in 2002-03 (R² = 0.01). However, as shown in Figure 4, there was a strong relationship between the overall rates of AAA repair in 1996-97 and in 2002-03 (R² = 0.42). Regions that had high rates of AAA repair overall in 1996-97 continued to have high rates in 2002-03.

Figure 3

Figure 4

Figures 5 and 6, however, provide further insight into how the introduction of the endovascular repair occurred. Regions that had high rates of AAA repair in 1996-97 (all of which were open procedures), continued, in general, to have high rates of open repairs in 2002-03 (R² = 0.30). When one looks at changes in the rates of the open procedure between 1996-97 and 2002-03 (Figure 6), however, one sees that in general the regions with the greatest declines in rates of the open procedure (shown on the horizontal axis as a greater positive number when the decline was greater) were also the regions that tended to have the greatest increase in use of the endovascular procedure (shown on the vertical axis).

Figure 5

Figure 6

These data are consistent with the national data shown in Figure 1. In general, endovascular repair appears to have been introduced as a substitute for the open procedure. First, areas with high rates of AAA repair continued to have high overall rates of the two procedures (Figure 4). And having a high rate of procedures in 1996-97 predicted not only continued high rates of the open procedure in 2002-03 (Figure 5), but also predicted relatively high rates of the endovascular procedure (R² = 0.14).

Variations in utilization rates, however, persist. Figure 7 shows how rates of utilization of these procedures have changed over time and suggests an additional important insight: introduction of the endovascular procedure has increased the variability in procedure rates. This is consistent with the hypothesis that broadening of the indications (perhaps to patients at higher underlying risk from major surgery) has led to greater variability in clinical decision-making for this procedure.

Figure 7

Methods Overview

A hospital service area (HSA) is a collection of ZIP codes whose residents receive most of their hospitalizations from the hospitals within that area. Hospital referral regions (HRRs) are aggregations of HSAs and represent regional health care markets for tertiary medical care; each HRR contains at least one hospital that performs major cardiovascular procedures and neurosurgery.

The Medicare population in an area that was used as the denominator for the rates in these studies included those alive, age 65 to 99, and not enrolled in a risk bearing HMO. The numerator for the surgical rates presented was based on all individuals meeting these eligibility criteria who underwent the specified procedure during an inpatient stay within the given year, based on the ICD-9-CM procedure codes shown in the table. The numerator for Part A expenditures was based on the actual payments made by the Medicare program, including disproportionate share, medical education, and outlier payments.

Rates based on a count of fewer than 11 observed counts are not displayed for reasons of patient confidentiality. Rates with fewer than 26 expected events are reported in parentheses to indicate lack of statistical precision; the margin of error is greater than 20%. Rates were adjusted to the age, sex and race distribution of the national Medicare population as follows. The national event rate for each age-sex-race category was computed. These rates were then applied to the HSA and HRR populations to produce the expected number of events in the HSA or HRR; that is, the number of events that would have occurred in the HSA if its rate had been the same as the national event rate.

Table

Appendix Table.

Age-sex-race adjusted rates per 1000 Medicare beneficiaries in U.S. Hospital Referral Regions. Notes: When rates were based on observed counts of 11 or fewer individuals in the HRR, the rates are fully suppressed and the cell in this table is empty.When (more...)