The Epidemiology of Lower-Extremity Disease in Veterans With Diabetes

Hospital discharge records from FY1998 were obtained from the VA Patient Treatment File. The quality of this dataset has been validated previously (3). This file permits up to 5 codes for each operation and up to 11 diagnoses codes for the hospitalization. A scrambled social security number for each individual during the year allowed aggregation of all hospitalizations for an individual. Any person with a hospital diagnosis code for diabetes was identified for the entire year as having diabetes (4).

All hospitalizations for lower-extremity ulceration were identified and classified according to most severe diagnosis using the following hierarchy modified from Holzer (5): 1) chronic nonhealing ulcers (ICD-9-CM 707.1, 707.9); 2) cellulitis, abscess, or infected ulcer (ICD-9-CM 681.10, 681.11, 682.6, and 682.7); 3) osteomyelitis (ICD-9-CM 729.4, 730.x, 731.x); 4) gangrene (ICD-9-CM 785.4, 040.0, and 440.24); 5) surgical complications from a stump infection (ICD-9-CM 768); 6) orthopedic procedure (ICD-9-CM 997.6); or 7) complications from a prior vascular graft (ICD-9-CM 440.3, 996.62, 996.7, 996.74, and E878.2). We excluded hospitalizations with venous stasis ulcer codes and no additional specific codes for other lower-extremity conditions. We also excluded the 8,980 hospitalizations for decubiti ulcers (ICD-9-CM 707.0) without concurrent lower-extremity disease because the code provides no specific information on anatomic site.

All peripheral vascular procedures were noted from either the surgery file or the procedure file and classified according to the most invasive or proximal procedure during the hospitalization using the following hierarchy: 1) angioplasty (ICD-9-CM 39.50), 2) proximal vascular bypass (iliac-femoral bypass ICD-9-CM 39.25), and 3) distal bypass (“Other peripheral vascular bypass or shunt,” ICD-9-CM 39.29). The ICD-9-CM code 39.29 should not be used to identify arteriovenous shunts for renal dialysis or cardiac bypass harvest procedures, but to exclude the latter possibility, we eliminated procedures associated with codes for extracorporeal circulation (ICD-9-CM 39.61).

All lower-extremity amputations (ICD-9-CM 84.11–84.19) were identified. Procedures with the same ICD-9-CM code on the same day were considered to be a single procedure, even though they might represent bilateral amputations. They were assigned as one amputation at the most proximal level. Amputations were excluded if the diagnosis codes included cancer of the lower extremity (ICD-9-CM 170.7, 170.8, 172.7, and 173.7), traumatic amputation (ICD-9-CM 895.x-897.x), fracture or late effect of fracture (ICD-9-CM 82x, 905.4), dislocation (ICD-9-CM 835–838), or crush injury (ICD-9-CM 928–929). Lower-extremity amputations were grouped into toe (ICD-9 CM 84.11), transmetatarsal (ICD-9-CM 84.12), transtibial (ICD-9-CM 84.13–84.17), and transfemoral (84.18–84.19) amputations. Approximately 15% of hospitalizations recorded two or more amputations; thus, the highest level of amputation during the hospitalization was used for analysis.

The indication for either the peripheral vascular procedure or amputation was classified according to the most severe diagnosis using the following hierarchy:

1. Atherosclerosis without specified complications (atherosclerosis [ICD-9-CM 440.x and 44x.x]) and peripheral vascular disease (ICD-9-CM 443.8, 443.89, and 440.21).

2. Skin ulceration, comprised of chronic ulcers (ICD-9-CM 707.0, 707.1, 707.9, and 891–894); cellulitis and abscess (681.10, 681.11, 682.6, and 682.7); or deep infection, fasciitis, and osteomyelitis (ICD-9-CM 729.4, 730.x, and 731.x); and rest pain or ulcer due to atherosclerosis (ICD-9-CM 440.22 and 440.23).

3. Gangrene, nonspecific, gas gangrene, and due to atherosclerosis (ICD-9-CM 785.4, 040.0, and 440.24).

4. Complications, including a stump infection, amputation complications (ICD-9-CM 768 and 997.6), or complications from a prior vascular graft (ICD-9-CM 440.3, 996.62, 996.7, 996.74, and E878.2).

5. Miscellaneous conditions not listed above.

The diabetic user population was defined as any veteran with three or more ambulatory care visits in 1 year with at least one visit containing a diagnosis code of diabetes (ICD-9-CM 250.x). Each veteran was counted only once. This diabetic veteran user population was used as the denominator to calculate the age-specific rates of amputation.

The hospitalization was used as the unit of analysis for assessment of ulceration, bypass procedures, and amputation. Frequencies were computed to describe hospitalizations for lower-extremity ulceration, amputation, vascular bypass, and angioplasty. Age-specific rates were calculated using the outpatient diabetes denominator. Race-specific rates were age adjusted to the 1980 male population. Statistical significance was tested with t tests, χ², and Pearson’s χ².

During FY1998, the VA provided 18,476 hospitalizations for lower-extremity ulceration, 4,822 hospitalizations with peripheral vascular procedures, and 5,329 hospitalizations with lower-extremity amputation. Persons with diabetes comprised 57% of hospitalizations for ulceration, 34% of all hospitalizations with a peripheral vascular procedure, and 66% of hospitalizations with an amputation of the lower extremity. The overlap between the categories is noted below as each condition is described. Over 99.9% of the discharges were male, so no comparisons by sex are shown.

The features of veterans with diabetes discharged with lower-extremity ulceration are noted in Table 1. When hospital discharges were compared by diabetes status, those with diabetes had an older mean age (64.9 vs. 63.5 years, P < 0.0001), were less likely to be African American (20.0 vs. 22.1%, P < 0.01), and were more likely to be of other nonwhite or nonspecified race (9.7 vs. 7.7%, P < 0.01). Both groups were equally likely to have undergone a peripheral revascularization (4.5 vs. 5.0%, P = 0.10), but those with diabetes were more likely to also undergo minor amputation (13.1 vs. 5.1%, P < 0.01) or major amputation (7.6 vs. 5.9%, P < 0.01).

The features of hospital discharges with peripheral vascular procedure by diabetes status are shown in Table 2. Discharges with a diabetes diagnosis had an older mean age as compared with those without diabetes (66.4 vs. 64.2 years, P < 0.0001). Veteran discharges with diabetes diagnosis were more likely to be African American (14.6 vs. 11.0%, P < 0.01) and less likely to be white (53.5 vs. 61.4%, P < 0.01). The type of vascular procedure varied by diabetes status. Discharges with diabetes were less likely to have a proximal bypass procedure (with or without angioplasty) (10.9 vs. 23.9%, P < 0.01) and more likely to undergo distal bypass procedures (64.6 vs. 48.7%, P < 0.01). The indications for the vascular procedure differed significantly by diabetes status. Discharges with a diabetes diagnosis were less likely to have a diagnosis of atherosclerosis only (30.2 vs. 52.9%, P < 0.01) but were more likely to report an ulcer or osteomyelitis (25.5 vs. 15.9%, P < 0.01) or gangrene (23.8 vs. 7.3%, P < 0.01). Repeat procedures and amputation during the same hospitalization also varied significantly by diabetes status. Discharges with diabetes were more likely than those without diabetes to undergo a repeat vascular procedure (5.5 vs. 3.8%, P < 0.05), minor amputation (17.9 vs. 4.9%, P < 0.01), or major amputation (6.6 vs. 3.3%, P < 0.01).

The features of hospital discharges with lower-extremity amputation are shown in Table 3. When compared with discharges without diabetes, those with diabetes had a lower mean age (66.1 vs. 67.1 years, P = 0.002), were less likely to be African American (23.8 vs. 27.4%, P < 0.01), and were more likely to be other nonwhite or unknown racial categories (15.1 vs. 12.0%, P < 0.01). Hospital discharges of veterans with a diabetes diagnosis were more likely than those without diabetes to be married (48.6 vs. 40.2%, P < 0.01). Discharges with diabetes were more likely than those without diabetes to report the most proximal amputation level at the toe (41.0 vs. 25.4%, P < 0.01), transmetatarsal (10.5 vs. 6.0%), or transtibial level (28.4 vs. 26.4%) and much less likely at the transfemoral level (20.2 vs. 42.3%, P < 0.01).

When compared with those without diabetes, discharges with diabetes were less likely to report atherosclerosis alone (4.3 vs. 7.5%) but were more likely to have an indication diagnosis of ulceration or osteomyelitis (32.2 vs. 22.3%, P < 0.01) and equally likely to report gangrene (50.5 vs. 47.1%). When compared with those without diabetes, hospital discharges with amputation and diabetes diagnosis were more likely to undergo two or more amputation procedures (14.8 vs. 10.6%, P < 0.01) and less likely to undergo a peripheral vascular procedure (11.6 vs. 14.7%) during the same hospitalization.

The age-specific rates of discharge with ulcer, peripheral vascular procedures, and amputation per 1,000 diabetic persons are shown in Fig. 1. The rates of ulceration and amputation increased with age, while the rates for persons with diabetes and vascular procedures did not. Diabetic amputation rates from the National Hospital Discharge Survey for 1996 are shown for comparison in Fig. 2. Diabetic amputation rates in the U.S. were higher than VA rates in all three age-groups.

Although veterans with diabetes comprise ∼16% of the VA service population, they account for over half of all hospitalizations for lower-extremity ulceration, one-third of hospitalizations with a peripheral vascular procedure, and almost two-thirds of all hospitalizations with a lower-extremity amputation. There are few population-based reports by diabetes status for comparison with these VA rates. The 64% of VA amputations associated with diabetes is higher than the 50% reported from other U.S. populations and is probably related to the higher prevalence of diabetes in veterans and veterans treated at VA facilities as compared with the general population (1,2,6,7).

The proportion of lower-extremity amputations ascribed to diabetes in the VA is not only higher than rates reported in U.S. hospitals, but has increased over the last decade, from 59.3% of all amputations in 1989 to 66.4% in 1998 (2,8). The proportion of peripheral vascular procedures performed in persons with diabetes in VA facilities has also increased over the last decade, from 29.9% in 1989 to 34.2% in 1998. The increase in the proportion of these procedures reflects a decline in nondiabetic amputations and peripheral vascular procedures, while the number of diabetic amputations and vascular procedures has remained stable. The 10 trends in amputations and peripheral bypass procedures are described elsewhere (8).

The features associated with peripheral vascular procedures varied greatly by diabetes status. Vascular angioplasty and bypass were used less frequently for persons with diabetes. Discharges with diabetes were more likely to have severe indications (ulceration, osteomyelitis, and gangrene) as compared with those without diabetes (50 vs. 23%) and were more likely to come to amputation during the same hospitalization (23.5 vs. 8.2%). We were unable to locate population-based estimates of vascular bypass or ulceration by diabetes status for comparison with our data; therefore, we cannot comment on whether the utilization for these conditions and procedures is higher or lower than other populations. However, we note that only one-third of the discharges with bypass procedures were in persons with diabetes, while over half of all ulcerations and two-thirds of all discharges with amputations were in persons with diabetes. There are a number of reasons as to why the proportion of peripheral vascular procedures is dissimilar to the proportion of ulceration and amputation. First, the distribution of peripheral vascular disease in persons with diabetes tends to involve multiple sites in distal vessels of the leg, rather than occurring as a circumscribed lesion in a proximal artery (9). Thus, persons with diabetes may not have a disease distribution amenable to current vascular surgical management. It is also possible that persons with both peripheral vascular disease and diabetes do not perceive the classic symptoms of claudication or rest pain due to peripheral neuropathy and thus do not present for medical evaluation until they have developed advanced disease with complications of infection, osteomyelitis, or gangrene. Finally, it is possible that surgeons may not view persons with diabetes as suitable candidates for peripheral bypass procedures. Early studies noted poorer survival following peripheral bypass for persons with diabetes, but more recent studies have found the survival rates to be similar to those in nondiabetic populations.

The age-specific rates we calculated for the lower-extremity complications provide a means of adjusting for the changing numbers and age distribution of the population that receives care in VA facilities. The diabetes denominator we used was based on outpatient diagnosis codes first mandated in 1996. No accurate estimates of the entire VA diabetic population were readily available before 1998, so accurate trends in diabetic ulceration, vascular procedures, and amputation cannot be calculated. The diabetes denominator was based on a criteria of at least three outpatient visits to ensure that we had selected the population that received the majority of its care within the VA, but this criteria would also undercount noncompliant persons who present rarely for care yet are known to be at higher risk for the complications. Recent VA recruitment and accurate enrollment efforts may have also influenced the denominator. Surprisingly, the VA age-specific amputation rate per diabetic person is somewhat lower than the 1998 U.S. hospital discharge rates. No comparative data on diabetic vascular bypass or ulcer hospitalization rates were available for comparison.

In summary, persons with diabetes comprise a major proportion of veterans hospitalized for ulceration of the lower extremity, one-third of discharges with peripheral vascular procedures, and two-thirds of discharges with amputation. The toll exacted by these conditions among veterans is high.

This research was funded by the Veterans Affairs Rehabilitation Research and Development Program (grant no. A0806C) and the Veterans Affairs Epidemiologic Research and Information Center (ERIC), VA Puget Sound Health Care System, Seattle, Washington.