Peripheral arterial disease (PAD) is a major complication

Transcription

1 Peripheral Arterial Disease and Renal Transplantation Jon J. Snyder,* Bertram L. Kasiske, and Ross Maclean *Chronic Disease Research Group, Minneapolis Medical Research Foundation, and Department of Medicine, Hennepin County Medical Center, University of Minnesota Medical School, Minneapolis, Minnesota; and Bristol-Myers Squibb, Inc., Princeton, New Jersey It was hypothesized that peripheral arterial disease (PAD) is less frequent after kidney transplantation than among comparable patients who are on the deceased-donor waiting list. The cumulative incidences and risk factors for PAD were compared among 43,427 adult transplant recipients and 53,309 adults who were placed on the renal transplant waiting list between 1995 and All patients had Medicare primary insurance coverage, and Medicare claims were used to identify PAD. For patients with diabetes, the 3-yr cumulative incidence of de novo PAD was 24% on the waiting list versus 20% after transplantation. For patients without diabetes, the 3-yr cumulative incidence was 9% on the waiting list versus 5% after transplantation. The adjusted relative risk (RR) for PAD among patients without diabetes was 0.73 (95% confidence interval [CI] 0.66 to 0.80; P < ) in the transplant population versus the waiting list population, whereas among patients with diabetes, it was 0.88 (95% CI 0.82 to 0.96; P ). A diagnosis of PAD on the waiting list was associated with an almost three-fold increase in the RR for death for patients without diabetes (2.98; 95% CI 2.71 to 3.27; P < ) and with diabetes (2.92; 95% CI 2.71 to 3.15; P < ). After transplantation, de novo PAD increased the RR for death almost two-fold in patients without diabetes (1.92; 95% CI 1.63 to 2.26; P < ) and with diabetes (1.83; 95% CI 1.58 to 2.12; P < ). The incidence of PAD is higher on the waiting list than after transplantation and is associated with an increased risk for death among patients with and without diabetes. J Am Soc Nephrol 17: , doi: /ASN Received March 6, Accepted April 19, Published online ahead of print. Publication date available at The data reported here were supplied by the United States Renal Data System. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the US government. Address correspondence to: Mr. Jon J. Snyder, Chronic Disease Research Group, Minneapolis Medical Research Foundation, 914 S. 8th Street, Suite 5-253, Minneapolis, MN Phone: ; Fax: ; jsnyder@cdrg.org Peripheral arterial disease (PAD) is a major complication that results in increased morbidity and mortality and decreased quality of life among patients with ESRD. The incidence of PAD is very high in patients who have ESRD and are on maintenance dialysis (1 4). However, few studies have examined the incidence and risk factors for PAD after kidney transplantation using registry data. The incidence of ischemic heart disease (IHD) is lower after kidney transplantation than among comparable patients who are on the deceaseddonor transplant waiting list (5). Because IHD likely has many of the same risk factors as PAD, we hypothesized that the incidence of PAD would be similarly reduced for kidney transplant recipients compared with dialysis patients who were on the deceased-donor waiting list. Patients in the United States have been determined to be suitable candidates for kidney transplantation at the time they are placed on the United Network for Organ Sharing (UNOS) deceased-donor waiting list. Therefore, comparing patients who are on the deceased-donor waiting list with kidney transplant recipients should improve our understanding of similarities and differences in the incidence and risk factors for PAD in these two populations. We used data from the United States Renal Data System, which includes data from UNOS as well as Medicare claims, to describe the cumulative incidences, rates, and risk factors for PAD before and after kidney transplantation. We included all Medicare patients who were placed on the renal transplant waiting list or received their first kidney transplantation between 1995 and Materials and Methods Waiting List Patients All adult ( 18 yr of age) transplant candidates who were certified by Medicare to have ESRD and were first placed on the UNOS kidneyonly waiting list between 1995 and 2003 were included in the initial cohort (n 132,625). The population was restricted to patients with Medicare Part A and Part B as their primary insurance coverage when they were added to the UNOS waiting list (n 60,531 [46%]). This restriction was necessary because PAD events were identified by using Medicare claims data as detailed below. Medicare primary pay status was determined using the Medicare enrollment database in conjunction with Medicare claims for dialysis services. Patients who had received a previous organ transplant before being added to the waiting list were excluded (n 3504 [6%]). We also excluded 2542 (4%) patients with previous PAD on the End-Stage Renal Disease Medical Evidence Report Medicare Entitlement and/or Patient Registration form (CMS form 2728) and an additional 1176 (2%) patients who had PAD on the UNOS Transplant Candidate Registration Form. In all, 7% had evidence of PAD from these two sources at listing and were excluded. These exclusions left a final sample of waiting list patients of 53,309 (40% of the initial waiting list cohort). Copyright 2006 by the American Society of Nephrology ISSN: /

2 J Am Soc Nephrol 17: , 2006 Peripheral Arterial Disease 2057 and peripheral bypass. A patient was said to have PAD when one inpatient International Classification of Diseases, Ninth Revision (ICD-9) diagnosis code or any Current Procedural Terminology (CPT) or ICD-9 procedure code was found. In the absence of these criteria, a patient was said to have PAD when two outpatient ICD-9 diagnosis codes appeared at least 1 d apart and no more than 365 d apart, in which case the later date was as the date of diagnosis. This was done in an attempt to not capture suspect or rule-out cases of PAD for which a patient seeks outpatient care. In some cases, PAD was defined by ICD-9 diagnosis codes alone, some were defined by procedure codes alone, and some used a combination of both diagnosis and procedure codes. Specific codes used are in the Appendix. Figure 1. Cumulative incidence of peripheral arterial disease (PAD). DM, diabetes mellitus; NDM, no diabetes mellitus. Transplantation Patients All adult recipients of a first, kidney-only transplant in the United States between 1995 and 2003, with a known age at the time of transplant, were included in the initial cohort (n 97,139). Similar to the waiting list population, the transplant population then was restricted to recipients with Medicare Part A and Part B as their primary insurance coverage on the date that they received their transplant (n 46,471 [48%]). Medicare primary pay status was determined for transplant patients by using the Medicare enrollment database in conjunction with any Medicare claims that were associated with the transplant hospitalization. Patients with PAD documented on the 2728 form at the initiation of treatment for ESRD were excluded (n 1372 [3%]). Remaining patients with PAD indicated on the UNOS Transplant Candidate Registration Form also were excluded (n 1672 [4%]). In all, 7% had evidence of PAD from these two sources and were excluded. These exclusions left a final sample of 43,427 (45% of initial cohort). Identification of PAD Patients with clinical evidence of symptomatic PAD were identified by searching Medicare claims using the following diagnostic subcategories: Amputation; peripheral aneurysm; arterial embolism and thrombosis; peripheral angioplasty, atherectomy, or endarterectomy; Statistical Analyses The unadjusted cumulative incidence of PAD was estimated using the Kaplan-Meier method. Cumulative incidences were calculated as 1 minus the Kaplan-Meier survival estimate at time after transplantation for the transplant population and time after being placed on the waiting list for the waiting list population. For both populations, follow-up was censored at the first occurrence of death, loss of Medicare primary pay status, or December 31, In addition, waiting list follow-up was censored at transplantation, and transplant follow-up was censored at graft failure or 3 yr after transplant (as a result of loss of Medicare coverage, which typically occurs at 3 yr). Separate models were run for the populations with and without diabetes. A patient was considered to have diabetes when diabetes was the primary cause of renal failure or noted as a comorbid condition on the 2728 form at the time of ESRD certification. Unadjusted rates for PAD were calculated as the number of PAD events in the first 3 yr, divided by the total patient-years of observation, and multiplied by 100 to yield a rate per 100 patient-years of follow-up. Risk factors for PAD were assessed using Cox proportional hazards analyses, whereby follow-up was censored as described above for the Kaplan-Meier models. For the waiting list population, adjustment was made for age, race, ethnicity, gender, primary cause of renal failure, body mass index, previous dialysis time, and baseline comorbidity (congestive heart failure, hypertension, and ischemic heart disease/ cerebrovascular accident). For the transplant population, adjustment was made for the covariates above as well as donor type; receipt of an expanded-criteria donor kidney (as defined by UNOS); cold ischemia time; percentage of panel reactive antibodies; HLA mismatches; hepatitis C virus serology; baseline immunosuppressive regimen; and donor Table 1. Cumulative incidences of PAD events at 1 and 3 yr, waiting list versus transplant populations a Cumulative Incidence (%) PAD Component Waiting List Transplant No Diabetes Diabetes No Diabetes Diabetes 1Yr 3Yr 1Yr 3Yr 1Yr 3Yr 1Yr 3Yr Amputation Peripheral aneurysm Arterial embolism and thrombosis Peripheral angioplasty, atherectomy, or endarterectomy Peripheral bypass Any PAD (any of the above) a PAD, peripheral arterial disease.

3 2058 Journal of the American Society of Nephrology J Am Soc Nephrol 17: , 2006 Figure 2. PAD 3-yr event rates for waiting list and transplant patients without diabetes. TX, transplantation; WL, waiting list. age, gender, and race. Separate analyses were performed for patients with and without diabetes. We assessed the effect of transplantation on the risk for PAD using Cox proportional hazards analysis with transplantation as a timedependent covariate. These analyses included all patients as described in the waiting list population above (n 53,309) as well as patients who received a first, kidney-only transplant between 1995 and 2003, without ever being placed on the waiting list. There were 5036 patients who received a live-donor transplant without ever being listed, so the final sample was 58,345 patients. Patients who were never listed for a deceased-donor kidney contributed no time to the waiting list in the time-dependent Cox proportional hazards models. As above, patients were censored at death; December 31, 2003; loss of Medicare primary pay status; graft failure; or 3 yr after transplantation. Separate analyses were performed for patients with and without diabetes. All analyses considered time to first PAD diagnosis. Therefore, the relative risks (RR) comparing the transplant population with the waiting list population should be interpreted as the RR for a PAD diagnosis, assuming that the patient underwent transplantation before any PAD diagnosis while on the waiting list. The effect of PAD on subsequent mortality was assessed using Cox proportional hazards analyses in patients with and without diabetes separately. In these analyses, PAD and transplantation status both were time-dependent covariates. The interaction between transplant status and PAD diagnosis also was considered to allow the risk that was associated with PAD to vary by transplant status. All analyses were conducted using SAS version (SAS, Inc., Cary, NC). Results In patients with diabetes, the cumulative incidence of de novo PAD after 3 yr on the waiting list was 24%, compared with 20% at 3 yr after transplantation (Figure 1). Among patients without diabetes, the cumulative incidence after 3 yr on the waiting list was 9% compared with 5% after transplantation (Figure 1). Among patients without diabetes, angioplasties, atherectomies, and endarterectomies were the most common PAD events, whereas for patients with diabetes, amputations were most common (Table 1). There was an initial increase in the cumulative incidence of PAD among transplant recipients in the first few months after transplantation (Figure 1). Although we cannot be certain of the cause of this, it could be due to preexisting conditions that were not noted in the initial cohort construction or to conditions that were noted during a period of close physician supervision in the months immediately after transplantation. Subsequent to this initial increase, the cumulative incidence curves increase at a rate lower than that seen in the waiting list population. Similarly, unadjusted rates for new PAD events were higher on the waiting list than after transplantation. They were much higher for patients with diabetes compared with patients without diabetes, both on the waiting list and after transplantation (Figures 2 and 3). For patients who did not have diabetes either on the waiting list or after transplantation, the unadjusted rates of PAD events were highest for angioplasties, atherectomies, and endarterectomies compared with lower rates for amputations (Figure 2). In contrast, for patients with diabetes, amputations occurred at a much higher rate than angioplasties, atherectomies, or endarterectomies (Figure 3), although angioplasties, atherectomies, and endarterectomies still were more Figure 3. PAD 3-yr event rates for waiting list and transplant patients with diabetes.

4 J Am Soc Nephrol 17: , 2006 Peripheral Arterial Disease 2059 frequent among patients with than without diabetes. The exception was peripheral aneurysms (largely abdominal aortic aneurysms), which actually were less frequent among patients with diabetes than patients without diabetes (Figures 2 and 3). The adjusted RR of transplantation (compared with the waiting list) for PAD was 0.73 (95% confidence interval [CI] 0.66 to 0.80; P ) among patients without diabetes (Figure 4). Among patients with diabetes (Figure 5), the adjusted RR of transplantation for PAD was 0.88 (95% CI 0.82 to 0.96; P ). Notably, the estimated risk that was associated with each individual component of the PAD definition was reduced after transplantation (Figure 3). Older age increased the risk for PAD on the waiting list (Table 2) and after transplantation (Table 3), but the effect of age seemed to be greater for patients without rather than with diabetes. For both patients with and without diabetes, on the Diabetic Nondiabetic Amputation RR = 0.40 Peripheral Aneurysm 0.62 Aterial Embolism and Thrombosis 0.64 Peripheral Angioplasty, Atherectomy, or Endarterectomy Peripheral Bypass Figure 5. Adjusted RR for PAD in transplant patients versus waiting list patients with diabetes. Shown are results of six separate Cox proportional hazards analyses. A risk 1.00 indicates that transplantation was associated with a lower risk than on the waiting list. Failure of 95% CI to cross 1.00 indicates statistical significance. Each analysis was adjusted for age, race, ethnicity, gender, primary cause of kidney failure, BMI, previous time on dialysis, and baseline comorbidity (data not shown). Any PAD (any of the above) Adjusted Relative Risk (RR; TX:WL) Figure 4. Adjusted relative risks (RR) for PAD in transplant patients versus waiting list patients without diabetes. Shown are results of six separate Cox proportional hazards analyses. A risk 1.00 indicates that transplantation was associated with a lower risk than on the waiting list. Failure of 95% confidence intervals (CI) to cross 1.00 indicates statistical significance. Each analysis was adjusted for age, race, ethnicity, gender, primary cause of kidney failure, body mass index (BMI), previous time on dialysis, and baseline comorbidity (data not shown) waiting list or after transplantation, nonwhite patients had a lower risk for PAD (Table 2). Gender was not associated with PAD on the waiting list, but after transplantation, men were more likely than women (among both patients with and without diabetes) to have PAD (Table 2). It is interesting that patients who were on the waiting list and had hypertension as the primary cause of ESRD were less likely than patients with ESRD from glomerulonephritis to develop PAD (Table 2). In contrast, transplant patients who had hypertension as the primary cause of ESRD were more likely than patients with ESRD from glomerulonephritis to develop PAD (Table 3). Diabetes as a cause of ESRD was associated with a higher risk for PAD than diabetes that was not considered to have caused ESRD, both on the waiting list and after transplantation (Table 2). Patients who had been on dialysis for at least 2 yr before being placed on the waiting list had an increased risk

5 2060 Journal of the American Society of Nephrology J Am Soc Nephrol 17: , 2006 Table 2. Risk factors for PAD in waiting list patients a Characteristic Patients without Diabetes (n 34,202) Patients with Diabetes (n 19,107) %of Patients HR 95% CI P %of Patients HR 95% CI P Age (yr) 18 to Reference Reference 35 to to to to to to to to Race white Reference Reference black to to other to to Ethnicity Non-Hispanic Reference Reference Hispanic to to Gender female Reference Reference male to to Primary cause of renal failure diabetes 0 NA NA NA to hypertension to to glomerulonephritis Reference Reference cystic kidney disease to to other/unknown to to BMI to to to Reference Reference 25.0 to to to to to to to to unknown to to Previous dialysis time 1 yr Reference Reference 1to 2 yr to to to 3 yr to to yr to to Baseline comorbidity CHF to to hypertension to to IHD/CVA to to a BMI, body mass index; CHF, congestive heart failure; CI, confidence interval; HR, hazard ratio; IHD/CVA, ischemic heart disease/cerebrovascular accident; NA, not applicable. for PAD (Table 2). Similarly, patients who had been on dialysis for at least 3 yr (2 yr in the case of patients with diabetes) had a higher risk for PAD after transplantation (Table 3). A diagnosis of PAD greatly increased the risk for subsequent all-cause mortality (Figure 6). For patients with and without diabetes and on the waiting list, a diagnosis of PAD was associated with an almost three-fold increase in the risk for death. The RR was similar in patients without and with diabetes, being 2.98 (95% CI 2.71 to 3.27, P ) and 2.92 (95% CI 2.71 to 3.15; P ), respectively. Among transplant patients, PAD increased the RR for death almost two-fold, in patients without and with diabetes, being 1.92 (95% CI 1.63 to 2.26; P ) and 1.83 (95% CI 1.58 to 2.12; P ), respectively. Conversely, transplantation was associated with a reduced risk for death among patients without and with diabetes. Among nondiabetic patients with de novo PAD on the waiting list, transplantation was associated with a 53% reduction in the risk for death (RR 0.47; 95% CI 0.40 to 0.56; P ). Among diabetic patients with de novo PAD on the waiting list, transplantation was associated with a 64% reduction in the risk for death (RR 0.36; 95% CI 0.31 to 0.41; P ). Transplantation was associated with less reduction in mortality among waiting list patients without PAD (Table 4). Discussion There were several important findings in our study: (1) The incidence of PAD is high in patients who have ESRD and especially those who have diabetes, both on the waiting list and after kidney transplantation; (2) transplantation is associated with a lower rate of PAD compared with the waiting list; (3) a number of risk factors identify patients who are at particularly

6 J Am Soc Nephrol 17: , 2006 Peripheral Arterial Disease 2061 Table 3. Risk factors for PAD in transplant patients a Characteristic %of Patients Patients without Diabetes (n 32,009) Patients with Diabetes (n 11,418) HR 95% CI P %of Patients HR 95% CI P Age (yr) 18 to Reference Reference 35 to to to to to to to to Race white Reference Reference black to to other to to Ethnicity non-hispanic Reference Reference Hispanic to to Gender female Reference Reference male to to Primary cause of renal failure diabetes 0 NA NA NA to hypertension to to glomerulonephritis Reference Reference cystic kidney disease to to other/unknown to to Donor type deceased (non-ecd) Reference Reference deceased (ECD) to to living to to Cold ischemia time (h) Reference Reference 12 to to to to to unknown to to BMI to to to Reference Reference 25.0 to to to to to to to to unknown to to Previous dialysis time none (preemptive) to to yr Reference Reference 1to 2 yr to to to 3 yr to to yr to to PRA 50% Reference Reference 50% to to HLA mismatches Reference Reference to to to to to to high risk for PAD, including age, gender, race, ethnicity, duration of previous dialysis, and a history of cardiovascular disease; and (4) PAD is associated with an increased risk for all-cause mortality. We found that the cumulative incidence of PAD was 20 and 5% in patients with and without diabetes, respectively, at 3 yr after transplantation. In a study of 706 patients who received a transplant between 1976 and 1991 (28% had diabetes), the 15-yr cumulative incidence of PAD (amputations and revascularizations) was 15% (6). In a more recent study of 351 patients who received a transplant between 1998 and 2003 and did not have pretransplantation diabetes, the cumulative incidence of PAD

7 2062 Journal of the American Society of Nephrology J Am Soc Nephrol 17: , 2006 Table 3. Continued Characteristic Patients without Diabetes (n 32,009) Patients with Diabetes (n 11,418) %of Patients HR 95% CI P %of Patients HR 95% CI P to to to to to to unknown to to Hepatitic C serology negative/unknown Reference Reference positive to to Baseline immunosuppression regimen tacrolimus and MMF Reference Reference Neoral and MMF to to MMF alone to to none listed to to tacrolimus and rapamycin to to other to to Baseline induction antibodies none Reference Reference IL to to to other to to Donor age (yr) 0 to to to to Reference Reference 35 to to to to to to to to unknown 5 NE NE NE to Donor race white Reference Reference black to to other/unknown to to Donor gender female Reference Reference male to to unknown 5 NE NE NE 6 NE NE NE Baseline comorbidity CHF to to hypertension to to IHD/CVA to to a ECD, expanded-criteria donor; MMF, mycophenolate mofetil; NE, not estimable; PRA, panel reactive antibodies. at 40 mo of follow-up was 3% (7). In another study, the incidence of PAD (amputations or revascularizations) was three (8%) of 36 among kidney-pancreas recipients and 10 (11%) of 88 among kidney-only transplant recipients after a median of 45 mo of follow-up (8). Similarly, Woeste et al. (9) reported that 19 (9.5%) of 200 kidney-pancreas recipients who received a transplant between 1995 and 2001 underwent amputations an average of 18.7 mo after transplantation. Therefore, the incidence of PAD after kidney transplantation among patients with and without diabetes in these single-center studies is comparable to that found in our study. The cumulative incidence of PAD in this study was 24 and 9% in adult patients with and without diabetes, respectively, 3 yr after being placed on the waiting list for a first kidney transplant. In an analysis of the Medicare ESRD population between 1991 and 1994, Eggers et al. (1) reported rates of lower extremity amputations of 4.8 to 6.2 per 100 patient-years among unselected hemodialysis patients in 1991 to 1994 and 11.8 to 13.8 among patients who had diabetes and were on hemodialysis. Similarly, in a study by O Hare et al. (3), the amputation rate was 4.3 per 100 patient-years for all patients with ESRD and 13.8 per 100 patient-years for patients with diabetes. In our

8 J Am Soc Nephrol 17: , 2006 Peripheral Arterial Disease 2063 Figure 6. Effect of a PAD on subsequent mortality among waiting list and transplant recipients. Shown are results of four separate Cox proportional hazards analyses. A risk 1.00 indicates that transplantation was associated with a lower risk than on the waiting list. Failure of 95% CI to cross 1.00 indicates statistical significance. Each analysis was adjusted for age, race, ethnicity, gender, primary cause of kidney failure, BMI, previous time on dialysis, and baseline comorbidity (data not shown). analysis, rates of amputations were 5.5 and 0.4 per 100 patientyears in adult patients with and without diabetes on the waiting list. However, it is not surprising that rates on the waiting list are lower than those for the larger hemodialysis population, because patients who are on the waiting list are selected to be healthier. Differences in the rates of PAD events in patients without and with diabetes are in keeping with potential differences in the pathogenesis of PAD in these two populations. For example, amputations are especially common among patients with diabetes, whereas the rate of peripheral aneurysms (mostly large-artery aneurysms) is higher in patients without diabetes compared with patients with diabetes. This is in keeping with the notion that diabetic PAD often is caused by disease of small blood vessels, whereas PAD in patients without diabetes more often affects large blood vessels. In this study, the incidence of PAD was substantially lower after kidney transplantation than among comparable patients on the waiting list. Immediately after transplantation, there was an initial increase in the incidence of PAD, but thereafter the cumulative incidence was lower than that among patients who were on the waiting list (Figure 1). An analysis that adjusted for demographic characteristics, primary cause of kidney disease, and time on dialysis before transplantation confirmed that the RR for PAD for transplantation (compared with the waiting list) was 27% lower for patients without diabetes (Figure 4) and 12% lower for patients with diabetes (Figure 5). Eggers et al. (1) previously reported that the odds for amputations were 68% lower with a functioning transplant, compared Medicare patients who were on dialysis and had not had a previous transplant. This is not surprising, because patients who are on the waiting list are likely to be healthier than hemodialysis patients who are not on the waiting list. Nevertheless, some caution should be taken in interpreting these results. Although we adjusted the analysis for the time on dialysis before transplantation and other variables, it still is possible that patients who received a transplant were not strictly comparable to patients who were on the waiting list. We also examined a number of possible risk factors for PAD, both on the waiting list and after transplantation and both for patients without and with diabetes (Tables 2 and 3). Older age greatly increased the risk for PAD, both on the waiting list and after transplantation, but the increased risk that was associated with older age seemed to be greater in patients without than with diabetes. This possibly is a result of the frequent occurrence of small-vessel PAD among younger patients with type 1 diabetes. The lower RR for PAD among black (compared with white) and Hispanic (compared with non-hispanic) patients was similar in waiting list and transplant patients, as well as patients with and without diabetes (Tables 2 and 3). Eggers et al. (1) also reported a lower amputation risk for black compared with white patients with ESRD. Similarly, O Hare et al. (2) reported higher odds of PAD in white compared with nonwhite patients. Why black and Hispanic patients have a lower risk for PAD is unclear, but we also found that black and Hispanic patients have a lower risk for acute myocardial infarctions after being placed on the deceased-donor kidney transplant waiting list (5). Men had a higher risk for PAD after transplantation (Table 3), but gender was not associated with PAD among patients who were on the waiting list (Table 2). In contrast, O Hare et al. (2,3) reported that male hemodialysis patients were at increased risk for amputations. Similarly, Eggers et al. (1) reported that male patients with ESRD were more likely to undergo lower extremity amputations. It is not readily apparent Table 4. Effect of transplantation on the risk for mortality among patients with and without a diagnosis of PAD a Without Diabetes With Diabetes RR 95% CI b RR 95% CI b Patients without a diagnosis of PAD to to 0.62 Patients with a diagnosis of PAD to to 0.41 a RR, relative risk. b All P

9 2064 Journal of the American Society of Nephrology J Am Soc Nephrol 17: , 2006 why gender did not influence the risk for PAD in waiting list patients in our study. For patients who were on the waiting list, the risk for PAD was lower for patients with kidney disease that was caused by hypertension, cystic kidney diseases, or other causes compared with kidney disease that was caused by glomerulonephritis (Table 2). In contrast, Eggers et al. (1) found that amputations among patients with ESRD were more likely when kidney disease had been caused by hypertension than glomerulonephritis. We also found that after transplantation, the risk associated with kidney disease that was caused by hypertension was greater than that associated with glomerulonephritis (Table 3). There are reasons to believe that patients with either hypertensive nephrosclerosis or glomerulonephritis would be more likely to have PAD. In the latter case, factors that are associated with glomerulonephritis (e.g., inflammation, nephrotic proteinuria, use of corticosteroids) could increase the risk for PAD. It is noteworthy that the duration of time on dialysis, either before being placed on the waiting list or before transplantation, was a risk factor for PAD (Tables 2 and 3). O Hare et al. (3) also reported that for dialysis patients without diabetes, the duration of dialysis, or vintage, increased the risk for amputations. Similarly, we found that duration of previous dialysis was a risk factor for myocardial infarctions after placement on the transplant waiting list (5). Altogether, these data suggest that vascular disease is worsened by time on dialysis and argue for transplantation as soon as possible in patients with stage 5 chronic kidney disease (CKD). We also found that baseline congestive heart failure, IHD, and cerebral vascular disease were risk factors for PAD on the waiting list and after transplantation (Tables 2 and 3). It was reported previously in a single-center study that both IHD and cerebral vascular disease were strong risk factors for posttransplantation PAD (6). Similarly, congestive heart failure and cerebral vascular disease were risk factors for myocardial infarctions after being placed on the transplant waiting list (5). O Hare et al. (2,3) also found coronary artery disease and cerebral vascular disease to be risk factors for amputations in hemodialysis patients. Therefore, vascular disease at one site is a strong risk factor for vascular disease at another site. Curiously, a baseline history of hypertension was associated with a lower risk for PAD both on the waiting list and after transplantation (Tables 2 and 3). We also found a baseline history of hypertension to be associated with a lower risk for myocardial infarctions after placement on the deceased-donor waiting list (5). Similarly, O Hare et al. (2) reported that hemodialysis patients with higher BP were less likely to undergo amputations. It is plausible that this reverse association between BP and PAD results from the fact that patients with lower BP may already have advanced cardiovascular disease. Finally, we found an increased risk for all-cause mortality associated with a diagnosis of PAD. Among waiting list patients, a diagnosis of PAD was associated with a nearly threefold increase in the risk for death (Figure 6). For transplant patients, a diagnosis of PAD was associated with a nearly two-fold increase in the risk for death within the first 3 yr after transplantation (Figure 6). An analysis by Gill et al. (10) found a 41% increased risk for all-cause mortality in the first year after transplantation in patients with a history of PAD. Similarly, in a single-center study, the risk for IHD in transplant recipients with posttransplantation PAD was 4.58 times greater than that in patients without posttransplantation PAD (6). Others have reported similar results (11,12). These analyses clearly implicate PAD as a strong risk factor for further cardiovascular morbidity and mortality. In interpreting the results of this study, some limitations that are inherent to registry analyses should be considered. We excluded patients with documented PAD on the CMS 2728 form, but comorbidities on the 2728 form are known to be underreported (13). Our analysis used UNOS data, as well as the 2728 form data, to eliminate patients with preexisting PAD. However, we likely included in the analysis patients who had preexisting PAD. This may artificially inflate our cumulative incidence of de novo PAD. If we assume that baseline PAD was underreported by 50%, then this would indicate that we included in the analyses approximately 3000 patients who had preexisting PAD. If we assume further that these 3000 patients were found to have subsequent PAD, then this would reduce the cumulative incidence at 3 yr by approximately 4%. Furthermore, given that for transplant patients and waiting list patients we used the same method to eliminate preexisting PAD, the bias should affect all groups evenly, preserving the validity of comparisons between waiting list and transplant populations. Perhaps more important than the identification of patients with baseline PAD is the sensitivity and the specificity of our claims-based definition of de novo PAD events. No previous analyses have validated this claims-based definition of PAD, but similar methods have been validated for identifying diabetes (14) and myocardial infarctions (15) using Medicare claims data. By necessity, this analysis was restricted to patients who had Medicare as their primary payer, and approximately 48% of transplant patients and 46% of waiting list patients had Medicare as their primary insurance coverage. Although the population with Medicare as their primary insurance coverage is not a random sample of the entire population, we previously showed these populations to be similar (16). Given that the Medicare primary pay population is slightly older on average than the non-medicare primary pay population, one could hypothesize that our incidence figures are greater than what would be seen in the transplant population as a whole; however, we would not expect this to bias greatly the estimates of RR when comparing the waiting list with the transplant population. Conclusion The risk for PAD is very high among patients with ESRD. Transplantation seems to reduce this risk, and that duration of dialysis is an independent risk factor for PAD suggests that early transplantation and even preemptive transplantation are strategies that may help to reduce the risk for PAD. Aggressive treatment of traditional risk factors for PAD, including dyslipidemias, hypertension, cigarette smoking, and diabetes, should be pursued in patients with CKD, although data demonstrating

10 J Am Soc Nephrol 17: , 2006 Peripheral Arterial Disease 2065 the effectiveness of this strategy are lacking. Similarly, there is a high rate of failure of revascularization attempts in patients with stage 5 CKD (4). Therefore, it probably is important to intervene at earlier stages of CKD to prevent PAD events. Clearly, more research is needed in the treatment and the prevention of cardiovascular disease in CKD. Appendix The following tables show ICD-9 and CPT codes that are used to identify PAD along with their descriptions. We excluded ICD-9 codes that (1) were subjective (e.g., claudication); (2) may have included asymptomatic disease (e.g., discovered incidentally with radiographic imaging); (3) may have occurred for nonatherosclerotic vascular disease; or (4) were for arteries that supply the brain, upper extremities (possibly related to dialysis access), or coronary arteries: 1. Atherosclerosis 440.0, 440.1, , , , , , , , , , 440.8, Atherosclerosis of bypass graft of the extremities , , Other aneurysm of artery of upper extremity 442.0, or of neck , or of unspecified site Other peripheral vascular disease 443.0, 443.1, , , , , , , , and Arterial embolism and thrombosis of arteries of the...upper extremity or of unspecified artery Atheroembolism of...upper extremity or other site Polyarteritis nodosa and allied conditions to Other disorders of arteries and arterioles to Other shunt or vascular bypass 39.21, 39.22, 39.23, 39.27, and Endovascular repair or occlusion of head and neck vessels 39.72, or other vessels or 39.8 In addition, some ICD-9 procedure codes were excluded because they did not specify upper versus lower extremity. This may have undercounted procedures of native arteries that did not have a corresponding CPT code, but this was done to avoid counting upper extremity procedures that likely resulted from hemodialysis access procedures: 1. Angioplasty or atherectomy of noncoronary vessel Clipping of aneurysm or other repair of aneurysm Insertion of nondrug eluting, noncoronary stent(s) Similarly, to avoid including procedures that may have been done to attain, maintain, or improve a dialysis access, some CPT codes were excluded: : Embolectomy or thrombectomy...carotid, subclavian, or innominate artery, by neck incision : Embolectomy or thrombectomy...axillary, brachial, innominate, subclavian artery by arm incision : Embolectomy or thrombectomy...radial or ulnar artery, by arm incision , 35013: Direct repair of aneurysm, axillary-brachial artery : Direct repair of aneurysm, radial or ulnar artery : Thromboendarterectomy, axillary-brachial : Transluminal angioplasty, brachiocephalic trunk or branches : Transluminal angioplasty, percutaneous, brachiocephalic trunk or branches Acknowledgments This study was supported by a grant from the Bristol-Myers Squibb Company. We thank Susan Everson, PhD, for design and creation of the figures. Appendix 1. ICD-9 and CPT codes used to identify PAD a Condition ICD-9 Diagnosis Codes ICD-9 Procedure Codes a CPT Codesa Amputation NA 84.0X, 84.1X, , 24920, 25900, 25905, 25920, 25927, 27295, 27590, 27591, 27592, 27598, 27880, 27881, 27882, 27888, 27889, 28800, Peripheral aneurysm 441.X, to 442.3, to 34834, 34900, to 35103, 35131, 35132, Arterial embolism and thrombosis Peripheral angioplasty, atherectomy, or endarterectomy to , 444.1, , , , , , 35142, 35151, NA 34051, 34151, 34201, NA NA 35331, 35341, 35351, 35355, 35361, 35363, 35371, 35372, 35381, 35450, 35452, 35454, 35456, 35459, 35470, 35471, 35472, 35473, 35474, to 35483, 35485, to 35493, Peripheral bypass NA 39.25, 39.26, , 35531, 35533, 35541, 35546, 35548, 35549, 35551, 35556, 35558, 35563, 35565, 35566, 35571, 35583, 35585, 35587, 35621, 35623, 35646, 35647, 35651, 35654, 35656, 35661, 35663, 35665, 35666, a A patient was said to have the condition when one inpatient ICD-9 diagnosis code or any CPT or ICD-9 procedure code was found. In the absence of these criteria, a patient was said to have the condition when two outpatient ICD-9 diagnosis codes appeared at least 1 d apart and no more than 365 d apart, in which case the later date was as the date of diagnosis.

11 2066 Journal of the American Society of Nephrology J Am Soc Nephrol 17: , 2006 Appendix 2. Code descriptions Code Description ICD-9 diagnosis codes 441.X Aortic aneurysm and dissection to 442.3, Other aneurysm (excluding upper extremity, head and neck) to , to , , Arterial embolism and thrombosis (excluding upper extremity and unspecified artery) , Atheroembolism of lower extremity, kidney ICD-9 procedure codes Aorta-iliac-femoral bypass Other intra-abdominal, vascular shunt or bypass Other (peripheral) vascular shunt or bypass Endovascular implantation of graft in abdominal aorta 84.0X Amputation of upper limb 84.1X Amputation of lower limb Amputation, not otherwise specified CPT codes 24900, Amputation of arm through humerus 25900, Amputation of forearm Disarticulation through wrist Transmetacarpal amputation Disarticulation of hip 27590, 27591, Amputation, thigh, through femur Disarticulation at knee 27880, 27881, Amputation, leg, through tibia and fibula 27888, Amputation of the ankle 28800, Amputation of the foot Embolectomy or thrombectomy innominate, subclavian artery, by thoracic incision Embolectomy or thrombectomy renal, celiac, mesentery, aortoiliac artery, by abdominal incision Embolectomy or thrombectomy femoropopliteal, aortoiliac, by leg incision Embolectomy or thrombectomy popliteal-tibio-peroneal artery, by leg incision to Endovascular repair of abdominal aortic aneurysm Endovascular repair of iliac aneurysm to Direct repair of aneurysm, abdominal aorta 35131, Direct repair of aneurysm, iliac artery 35141, Direct repair of aneurysm, femoral artery 35151, Direct repair of aneurysm, popliteal artery Thromboendarterectomy, abdominal aorta Thromboendarterectomy, mesenteric, celiac, or renal Thromboendarterectomy, iliac Thromboendarterectomy, iliofemoral Thromboendarterectomy, combined aortoiliac Thromboendarterectomy, combined aortoiliofemoral Thromboendarterectomy, common femoral Thromboendarterectomy, deep (profunda) femoral Thromboendarterectomy, femoral and/or popliteal, and/or tibioperoneal Transluminal angioplasty, renal or other visceral artery Transluminal angioplasty, aortic Transluminal angioplasty, iliac Transluminal angioplasty, femoral-popliteal Transluminal angioplasty, tibioperoneal trunk and branches Transluminal angioplasty, percutaneous, tibioperoneal trunk or branches

12 J Am Soc Nephrol 17: , 2006 Peripheral Arterial Disease 2067 Appendix 2. Continued Code Description Transluminal angioplasty, percutaneous, renal or visceral artery Transluminal angioplasty, percutaneous, aortic Transluminal angioplasty, percutaneous, iliac Transluminal angioplasty, percutaneous, femoral-popliteal Transluminal atherectomy, open, renal or other visceral Transluminal atherectomy, open, aortic Transluminal atherectomy, open, iliac Transluminal atherectomy, open, femoral-popliteal Transluminal atherectomy, open, tibioperoneal trunk Transluminal atherectomy, percutaneous, renal or other visceral Transluminal atherectomy, percutaneous, aortic Transluminal atherectomy, percutaneous, iliac Transluminal atherectomy, percutaneous, femoral-popliteal Transluminal atherectomy, percutaneous, tibioperoneal trunk Bypass graft, vein, axillary-femoral Bypass graft, vein, aortoceliac or aortomesenteric Bypass graft, vein, aortofemoral or bifemoral Bypass graft, vein, axillary-femoral-femoral Bypass graft, vein, aortoiliac or bi-iliac Bypass graft, vein, aortofemoral or bifemoral 35548, Bypass graft, vein, aortoiliofemoral Bypass graft, vein, aortofemoral-popliteal Bypass graft, vein, femoral-popliteal Bypass graft, vein, femoral-femoral Bypass graft, vein, ileoiliac Bypass graft, vein, iliofemoral Bypass graft, vein, femoral-anterior tibial, posterior tibial, peroneal artery or other distal vessels Bypass graft, vein, popliteal-tibial, -peroneal artery or other distal vessels Bypass graft, in-situ vein, femoral-popliteal Bypass graft, in-situ vein, femoral-anterior tibial, posterior tibial, or peroneal artery Bypass graft, in-situ vein, popliteal-tibial, peroneal Bypass graft, other than vein, axillary-femoral Bypass graft, other than vein, axillary-popliteal or tibial Bypass graft, other than vein, aorto-bifemoral Bypass graft, other than vein, aortofemoral Bypass graft, other than vein, aortofemoral-popliteal Bypass graft, other than vein, axillary-femoral-femoral Bypass graft, other than vein, femoral-popliteal Bypass graft, other than vein, femoral-femoral Bypass graft, other than vein, ileoiliac Bypass graft, other than vein, iliofemoral Bypass graft, other than vein, femoral-anterior tibial, posterior tibial, or peroneal artery Bypass graft, other than vein, popliteal-tibial or -peroneal artery References 1. Eggers PW, Gohdes D, Pugh J: Nontraumatic lower extremity amputations in the Medicare end-stage renal disease population. Kidney Int 56: , O Hare AM, Hsu CY, Bacchetti P, Johansen KL: Peripheral vascular disease risk factors among patients undergoing hemodialysis. J Am Soc Nephrol 13: , O Hare AM, Bacchetti P, Segal M, Hsu CY, Johansen KL: Factors associated with future amputation among patients undergoing hemodialysis: Results from the Dialysis Morbidity and Mortality Study Waves 3 and 4. Am J Kidney Dis 41: , Jaar BG, Astor BC, Berns JS, Powe NR: Predictors of amputation and survival following lower extremity revascularization in hemodialysis patients. Kidney Int 65: , Kasiske BL, Maclean JR, Snyder JJ: Acute myocardial infarction and kidney transplantation. J Am Soc Nephrol 17: , 2006