When to refer patients with chronic kidney ... - Kidney International

http://www.kidney-international.org

original article

& 2007 International Society of Nephrology

When to refer patients with chronic kidney disease for vascular access surgery: Should age be a consideration? AM O’Hare1,2, D Bertenthal2, LC Walter1,2, AX Garg3, K Covinsky1,2, JS Kaufman4, RA Rodriguez5 and M Allon6 1

Department of Medicine, VA Medical Center and University of California, San Francisco, California, USA; 2VA San Francisco Research Enhancement Award Program (REAP), VA San Francisco, San Francisco, California, USA; 3Department of Medicine, University of Western Ontario, London, Ontario, Canada; 4Renal Section, VA Boston Healthcare System, Boston, Massachusetts, USA; 5Department of Medicine, San Francisco General Hospital and University of California, San Francisco, California, USA and 6Department of Medicine, University of Alabama Stata Statistical Software, Stata Corporation, College Station, Texas, USA

To determine whether age should inform our approach toward permanent vascular access placement in patients with chronic kidney disease, we conducted a retrospective cohort study among 11 290 non-dialysis patients with an estimated glomerular filtration rate (eGFR) o25 ml/min/ 1.73 m2 based on 2000–2001 outpatient creatinine measurements in the Department of Veterans Affairs. For each age group, we examined the percentage of patients that had and had not received a permanent access by 1 year after cohort entry, and the percentage in each of these groups that died, started dialysis, or survived without dialysis. We also modeled the number of unnecessary procedures that would have occurred in theoretical scenarios based on existing vascular access guidelines. The mean eGFR was 17.7 ml/min/1.73 m2 at cohort entry. Twenty-five percent (n ¼ 2870) of patients initiated dialysis within a year of cohort entry. Among these, only 39% (n ¼ 1104) had undergone surgery to place a permanent access beforehand. As compared with younger patients, older patients were less likely to undergo permanent access surgery, but also less likely to start dialysis. In all theoretical scenarios examined, older patients would have been more likely than younger patients to receive unnecessary procedures. If all patients had been referred for permanent access surgery at cohort entry, the ratio of unnecessary to necessary procedures after 2 years of follow-up would have been 5:1 for patients aged 85–100 years but only 0.5:1 for those aged 18–44 years. Currently recommended approaches to permanent access placement based on a single threshold level of renal function for patients of all ages are not appropriate. Correspondence: AM O’Hare, University of California San Francisco, VA Medical Center, San Francisco, 111J Nephrology, 4150 Clement Street, San Francisco, California 94121, USA. E-mail: Ann.O’[email protected] Received 8 August 2006; revised 9 November 2006; accepted 14 November 2006; published online 24 January 2007 Kidney International (2007) 71, 555–561

Kidney International (2007) 71, 555–561. doi:10.1038/sj.ki.5002078; published online 24 January 2007 KEYWORDS: elderly; permanent vascular access; hemodialysis; timing; Department of Veterans Affairs; chronic kidney disease

Among patients initiating chronic hemodialysis, use of a catheter rather than a more permanent form of vascular access such as a graft or fistula, is associated with excess morbidity, mortality, and cost.1–4 In order to reduce catheter use among incident dialysis patients, it is generally agreed that patients with chronic kidney disease (CKD) who are expected to begin chronic dialysis should be referred beforehand for surgery to create a permanent access. Ideally, this referral should occur in enough time to allow for fistula maturation and repeat attempts at fistula creation (if necessary) before initiation of dialysis.5 Although there is broad agreement on the importance of timely surgical referral for creation of a permanent access, most patients in this country nevertheless start dialysis with a catheter.6–10 A significant barrier to increasing the prevalence of permanent access among incident hemodialysis patients has been a lack of evidence to suggest exactly when patients with non-dialysis-dependent CKD should be referred for this surgery. In an individual patient, the expected time before initiation of dialysis is usually unknown and clinical practice guidelines on this topic are largely opinion-based and quite variable. For example, the 2000 Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines recommended that ‘patients should be referred for surgery to attempt construction of a primary AV fistula when their creatinine clearance is o25 ml/min, their serum creatinine level is 44 mg/dl (4352 mmol/l), or within 1 year of an anticipated need for dialysis.’5 In contrast, the 2006 Canadian Society of Nephrology guidelines recommend that patients be referred at a ‘creatinine clearance of 15 to 20 ml per min or serum creatinine of 3.4 mg/dl to 5.6 mg/dl (300 to 500 mmol/l), depending on the size and weight of the patient.’11 The recent 2006 KDOQI guideline 555

original article

AM O’Hare et al.: Vascular access surgery in CKD patients

recommends that patients be referred for fistula creation ‘at least six months before the anticipated need for dialysis’ and recommends that ‘patients with CKD stage 5 should bey strongly encouraged to allow the evaluation for and creation of a fistula for long-term access when appropriate.12,13 The question of when (and in whom) to place a permanent access in preparation for hemodialysis is a surprisingly complex one. Whereas the presence of a functioning graft or fistula is associated with improved outcomes among patients who begin dialysis, surgery to create a permanent access is clearly undesirable in patients who will never start dialysis, either because their CKD will never progress to end-stage renal disease (ESRD) or because they will die before needing dialysis. Placing permanent access in such patients subjects them to all of the risks of unnecessary surgery with no possible benefit. On the other hand, from a public health point of view, a certain number of unnecessary surgeries in patients who do not begin dialysis may be tolerated in order to maximize the proportion of patients who begin dialysis with a permanent access. As described above, many guidelines identify an ideal window of time before initiation of dialysis when permanent access should be placed (e.g. 46 months for 2006 KDOQI guideline, 41 year for 2000 KDOQI guideline). However, perhaps because clinicians can rarely predict who will need dialysis within a given time frame, most guidelines also identify a target level of renal function at which patients should be referred. This approach assumes a similar relationship between level of renal function and progression to ESRD among patients of all ages. However, at all levels of renal function, older patients are relatively more likely to die and less likely to start dialysis compared with younger patients.14–19 We therefore hypothesized that use of a single threshold level of estimated glomerular filtration rate (eGFR) to guide referral for permanent access would have different implications in older and younger patients. In the elderly, we predicted that the tension between the need to place a permanent access in patients who will begin dialysis and the need to avoid unnecessary surgery in patients who will not begin dialysis would be heightened. RESULTS Patients

There were 2 352 584 patients who underwent at least one serum creatinine measurement within the United States

Department of Veterans Affairs (VA) between 1 October 2000 and 30 September 2001. Among these, we excluded 11 061 patients who had already reached ESRD at the time of creatinine measurement. Among the remaining patients, we identified 19 342 whose initial eGFR measurement during the study period was o25 ml/min/1.73 m2. Because our study focused on permanent vascular access for hemodialysis (arteriovenous graft or fistula), we excluded 45 patients for whom initial ESRD treatment during follow-up was transplant and 433 patients whose initial dialysis modality was peritoneal dialysis rather than hemodialysis. Among the remaining 18 864 patients, we identified 11 290 who had a repeat eGFRo25 ml/min/1.73 m2 at least 3 months after cohort entry, or who started dialysis within 3 months of cohort entry in order to delineate a study population with clear evidence of a sustained eGFRo25 ml/min/1.73 m2. The mean age of the patient cohort was 70 years (95% confidence interval 69.8, 70.2), 2% (n ¼ 246) were female subjects, patient race was black in 22% (n ¼ 2466), white in 70% (n ¼ 7914), other in 6% (n ¼ 707), and unknown in 2% (n ¼ 203). Diabetes was diagnosed in 56%, coronary artery disease in 58%, congestive heart failure in 44%, peripheral arterial disease in 36%, cerebrovascular disease in 28%, chronic obstructive lung disease in 34%, and dementia in 7%. Patient average eGFR was 17.7 ml/min/1.73 m2 at cohort entry and was higher for older than for younger cohort members (Table 1). The median annual decrement in eGFR was 2.3 ml/min/1.73 m2 (25th to 75th percentile range 5.2 to 0.1 ml/min/1.73 m2) and was in general lower for older than for younger cohort members (Table 1). These calculations were based on follow-up creatinine measurements obtained a median of 530 days (25th to 75th percentile range 272–813 days) after cohort entry and 128 days (25th to 75th percentile range 83–254 days) before initiation of dialysis, death, or 30 September 2003, whichever came first. Overall, 15% (n ¼ 1728) of cohort patients initiated dialysis within 6 months, 25% (n ¼ 2870) within 1 year, and 40% (n ¼ 4502) within 2 years of cohort entry. However, these percentages varied considerably by age group: 32, 47, and 67%, respectively, for 18–44 year olds vs 8, 11, and 17%, respectively, for 85–100 year olds. The percentage of patients who died was 3% at 6 months, 9% at 1 year, and 22% at 2 years and also varied considerably by age group: 0.4, 4, and 6%, respectively, among 18–44 year olds vs 3, 17, and 41%, respectively, among 85–100 year olds.

Table 1 | Baseline and median rate of decline in eGFR by age group Age group (years) 18–44 45–54 55–64 65–74 75–84 85–100

% with outpatient nephrology visit by the end of follow-up (95% CI) 75.2 75.9 75.9 76.8 69.8 55.7

(69.9, (73.4, (73.9, (75.5, (68.4, (51.1,

80.5) 78.3) 77.8) 78.2) 71.3) 60.2)

Mean eGFR at cohort entry (95% CI), ml/min/1.73m2 15.3 15.7 16.0 17.9 18.6 18.8

(14.6, (15.4, (16.7, (17.7, (18.5, (18.4,

16.0) 16.0) 17.2) 18.0) 18.8) 19.2)

Median annual change in eGFR (25th to 75th percentile range), ml/min/1.73m2 3.4 3.6 3.2 2.2 1.9 1.3

(8.4, (7.6, (6.4, (5.0, (4.3, (4.0,

1.5) 0.9) 0.6) +0.2) +0.4) +1.1)

CI, confidence interval; eGFR, estimated glomerular filtration rate.

556

Kidney International (2007) 71, 555–561

original article


a

survived without dialysis – access not placed survived without dialysis – access placed died without dialysis – access not placed died without dialysis – access placed started dialysis – access not placed started dialysis – access placed

35

30

100%

20 80% 15 % in age group group

% access placed

25

10

5

0

40%

18–44 45–54 55–64 65–74 75–84 85–100 (n =254) (n=1165) (n=1789) (n =3691) (n=3931) (n=460) 20%

Age group

b

60%

50 0%

45

18–44 (n =254)

40

55 –64 (n =1789)

65– 74 (n=3691)

75–84 (n =3931)

85– 100 (n =460)

Age group

35 % access placed

45– 54 (n =1165)

Figure 2 | One year outcomes by age group.

30 25 20 15 10 5 0 18–44 (n=118)

45–54 (n=515)

55–64 (n =633)

65–74 (n=882)

75–84 (n =671)

85–100 (n =51)

Age group

Figure 1 | Permanent vascular access surgeries by age group. (a) Percent of all cohort patients who received pre-dialysis permanent access by the end of follow-up. Estimates are provided with a 95% confidence interval. (b) Percent of patients who initiated dialysis during follow-up that had undergone permanent access placement before initiation of dialysis. Estimates are provided with a 95% confidence interval.

By 1 year after cohort entry, only 16% (n ¼ 1919) of cohort patients had undergone pre-dialysis permanent access placement surgery. Among these, 30% (n ¼ 572) had already received a permanent access by the time of cohort entry and the remainder (n ¼ 1347) underwent pre-dialysis access surgery during the year after cohort entry. Seventy-three percent (n ¼ 8269) of patients had at least one outpatient nephrology encounter (as defined in the Materials and Methods section) within 1 year of cohort entry (Table 1). Twenty-five percent (n ¼ 2870) of patients started dialysis within 1 year of cohort entry. Among these, only 39% (n ¼ 1104/2870) had received a permanent access beforehand. The percentage of cohort patients who had received a permanent access by the end of 1 year follow-up decreased dramatically with advancing age (Figure 1a). However, among patients who initiated dialysis within 1 year of cohort entry, there were no significant differences by age group in Kidney International (2007) 71, 555–561

the percentage that had received a permanent access before dialysis initiation (Figure 1b), although the percentages were lowest at the extremes of age. In older age groups, the majority of patients either died without starting dialysis or survived without needing dialysis (Figure 2). Overall, low rates of permanent access placement in older patients appeared to be roughly commensurate with low rates of dialysis initiation at older ages. Rates of dialysis initiation decreased from the youngest to the oldest age group within each quartile of propensity score and regardless of whether patients had an outpatient nephrology encounter (results not shown). In theoretical scenarios where patients received permanent access at cohort entry based on a selection of different eGFR thresholds, the ratio of unnecessary to necessary procedures was always higher in older than in younger patients (Figure 3a–c). This pattern was most pronounced at higher threshold levels of eGFR. For example, if all patients with an eGFR o25 ml/min/1.73 m2 (i.e. all cohort patients) had been referred for access surgery at cohort entry, the ratio of unnecessary to necessary procedures even after 2 years of follow-up would have been almost 5:1 for patients aged 85–100 years but only 0.5:1 for those aged 18–44 years. Because the percentage of patients initiating dialysis increased with falling eGFR in all age groups, among older patients these ratios were markedly lower at lower eGFR referral thresholds. In contrast, among younger patients, the ratio of unnecessary to necessary procedures varied less between scenarios. DISCUSSION

In a large national cohort of VA patients with an eGFRo25 ml/min/1.73 m2, most patients who initiated 557

original article


Ratio

a 14 12 10 8 6 4 2 0 18– 44 45–54 55–64 65–74 75–84 85–100 (n=254) (n =1165) (n=1789) (n=3691) (n=3931) (n=460) Age group 6 months

Ratio

b

1 year 2 years

8 7 6 5 4 3 2 1 0 18 –44 (n=189)

45–54 55–64 65–74 75–84 85–100 (n=855) (n=1171) (n=2242) (n=2154) (n=246) Age group 6 months 1 year

Ratio

c

2 years

4 3.5 3 2.5 2 1.5 1 0.5 0 18– 44 45–54 55– 64 65–74 75–84 85–100 (n =116) (n=508) (n=617) (n=1003) (n=837) (n=96) Age group 6 months 1 year

2 years

Figure 3 | Ratio of unnecessary to necessary permanent access surgeries at different theoretical referral eGFR thresholds by age and length of follow-up. (a) Referral threshold eGFRo25 ml/min/ 1.73 m2. (b) Referral threshold eGFRo20 ml/min/1.73 m2. (c) Referral threshold eGFRo15 ml/min/1.73 m2.

dialysis over a 1-year period had not been referred beforehand for placement of permanent access. These findings corroborate earlier studies in other populations identifying pre-dialysis access placement as a priority area for quality improvement.8,9 At the same time, substantial age differences in rates of dialysis initiation in our cohort suggest that a uniform approach to permanent access placement in patients of all ages based solely on level of estimated renal function (implicit in many guidelines) may not be an optimal solution. Lower rates of dialysis initiation among elderly patients with CKD observed in the present study have been described in a variety of different populations.14–19 In this cohort, such age differences do not appear to be explained by confounding by other measured patient characteristics but may reflect the slower rate of eGFR decline, and higher competing risk of death among older patients in this cohort. In addition, the abbreviated Modification of Diet in Renal Disease equation was not developed nor has it been validated in an older population with severe CKD and thus may not provide accurate estimates of true GFR in this population. Goals of care may also be different among older patients, perhaps reflected in the lower rates of nephrology referral among cohort patients 75 years and older. Unfortunately, our data 558

sources were insufficiently detailed to determine whether age differences in rates of conservative therapy (i.e. no dialysis or transplant) for renal failure could account for lower rates of dialysis initiation in this cohort. Regardless of the underlying explanation, the group of patients who initiate dialysis (and thus stand to benefit from permanent access surgery) effectively becomes a ‘shrinking target’ at older ages. Our results demonstrate that use of a single eGFR (or presumably creatinine clearance) threshold to guide permanent access placement in patients of all ages has the potential to result in a disproportionately large number of unnecessary procedures in older patients. For example, among patients aged 85–100 years, placement of permanent access at an eGFRo25 ml/min/1.73 m2 (consistent with the 2000 KDOQI guideline) would have turned out to be necessary in only one in six patients, even after 2 years of follow-up. On the other hand, among patients aged 18–44 years with an eGFR o25 ml/min/1.73 m2, the procedure would have been necessary in one in three patients after only 6 months and one in 1.5 patients after 2 years of follow-up. Identifying which older patients are likely to start dialysis becomes particularly challenging over the relatively short periods of time recommended in some guidelines. For example, 2006 KDOQI guidelines recommend that access be placed at least 6 months before initiation of dialysis. However, even among 85–100 year olds with an eGFRo15 ml/min/1.73 m2, only one in four patients started dialysis within six months and only one in three patients started dialysis within a year. These findings are important because unnecessary surgeries are both costly and carry a risk to the patient with no benefit during the time frame expected. Current Medicare reimbursement for physician fees alone exceeds $500 for fistula placement. This figure does not include hospital costs related to the procedure or costs associated with repeat procedures when the initial procedure is unsuccessful. Although most practice guidelines implicitly favor a uniform approach to permanent access placement for patients of all ages, age did appear to influence placement practices in this ‘real world’ clinical setting. Low rates of permanent access placement in elderly compared with younger cohort members were commensurate with low rates of dialysis initiation at older ages. Overall, targeting of access procedures to patients who initiated dialysis within 1 year of cohort entry appeared to be no worse in older than in younger cohort members. Rates of pre-dialysis access placement among cohort members who initiated dialysis were uniformly low for patients of all ages. Low rates of pre-dialysis access placement among incident dialysis patients of all ages signal the need for more effective strategies to decrease catheter use among patients beginning dialysis. Our results suggest that a ‘one size fits all’ approach toward timing of permanent access placement may not be appropriate. In younger patients, broad efforts to increase permanent access placement at relatively high levels of eGFR (e.g. o25 ml/min/1.73 m2 as recommended in the 2000 KDOQI guideline) are probably appropriate. In this Kidney International (2007) 71, 555–561

original article


cohort, younger patients experienced a faster decline in eGFR, a lower competing risk of mortality, and (perhaps for these reasons) higher rates of dialysis initiation compared with older patients. Therefore, the ratio of unnecessary to necessary procedures would have been low in younger cohort members, even if all had been referred for access at an eGFRo25 ml/min/1.73 m2. On the other hand, in the elderly, efforts to improve targeting of permanent access placement to the subset of patients who are most likely to need dialysis in the expected time frame may be more appropriate. Such an approach might perhaps involve using a lower eGFR threshold for referral for access surgery in the elderly as modeled in this analysis (e.g. o15 ml/min/1.73 m2 as recommended in the 2006 KDOQI guideline). For example, referral of those aged 85–100 years at an eGFRo15 ml/min/ 1.73 m2 would have incurred only 1.5 unnecessary surgeries for every one patient that was necessary after 2 years of follow-up. However, even at such low levels of eGFR, the ratio of unnecessary to necessary procedures that would have occurred among the oldest age group is still higher than for the youngest age group if referred at an eGFRo25 ml/min/ 1.73 m2. Thus, we would argue that age should influence our approach to permanent access placement in patients with CKD. A uniform approach to permanent access placement may be appropriate in younger patients whose rates of dialysis initiation are relatively high even at relatively preserved levels of eGFR. However, in older patients, additional predictors of progression to dialysis beyond eGFR (e.g. proteinuria, change in eGFR, functional status) are needed to identify those patients most likely to benefit from permanent access placement. Limitations include: (1) this analysis only examines the timing of access placement surgeries. From the available data, we cannot determine whether an access was functional at the time of dialysis initiation. (2) We opted to define our cohort based on two creatinine measurements drawn at least 3 months apart within the VA. This approach allowed us to define a cohort of patients with evidence of a sustained eGFRo25 ml/min/1.73 m2 whose care was most likely to be captured in the VA and Medicare data sources available to us. However, this approach involved excluding patients who died within 3 months of cohort entry or did not have a repeat creatinine measurement for other reasons and patients whose repeat eGFR was 25 ml/min/1.73 m2 or greater. Thus, our results may not be completely generalizable to all VA patients who started dialysis during follow-up or to patients receiving care outside our system. However, we would argue that although selection bias may have influenced the absolute frequency of different outcomes among cohort members, such bias is unlikely to have greatly impacted the relationship between age and outcome frequency. (3) Our study was restricted to veteran health-care users and thus may not be generalizable outside the VA system and in particular to populations not well represented in our cohort (e.g. women and patients belonging to Hispanic or other racial-ethnic groups). However, although specific rates of permanent Kidney International (2007) 71, 555–561

access placement and death may differ for this cohort compared with others, age-related differences in outcomes are likely to be generalizable beyond the VA. Conclusion

Rates of pre-dialysis permanent access surgery among patients of all ages beginning dialysis are low and in need of improvement. However, dramatic differences between age groups in rates of chronic dialysis initiation suggest that a uniform approach to surgical referral based on a single threshold level of renal function among patients of all ages is not the optimal solution. Although such an approach may be appropriate in younger patients, our results call for a more targeted approach in the elderly directed at the minority that will begin dialysis within the desired time frame. MATERIALS AND METHODS Data sources We used the VA Decision Support System (DSS) Laboratory Results file to ascertain serum creatinine measurements associated with outpatient visits among cohort patients. We used the VA National Patient Care Database and Medicare denominator file and in-patient and outpatient claims to ascertain demographic and comorbidity information and to identify permanent access placement procedures occurring before or within 1 year after cohort entry. We used the VA Beneficiary Identification and Records Locator Sub-system (BIRLS) to ascertain date of death. These data were then linked to the United States Renal Data System (USRDS), a national ESRD registry, to exclude prevalent ESRD patients from the cohort and to identify new cases of treated ESRD occurring after cohort entry. Covariates Patient age was categorized as: 18–44, 45–54, 55–64, 65–74, 75–84, and 85–100 years. GFR at cohort entry was estimated using the abbreviated Modification of Diet in Renal Disease equation based on serum creatinine, age, race, and sex.20 Study outcomes To evaluate the impact of age on permanent access placement, patients were classified into six categories based on 1 year outcomes: whether they underwent pre-dialysis permanent access placement and whether they initiated dialysis or died before onset of dialysis within a year of cohort entry. We used both arteriovenous fistula and graft Current Procedural Terminology (CPT) and International Classification of Disease (ICD-9) procedure codes to identify permanent access procedures (ICD-9 procedure codes 39.27 and 39.93 and CPT codes 36800, 36810, 36821, 36825, and 36830) using both VA administrative and Medicare data. One year was selected for this analysis, because this was the maximum follow-up time available for vascular access procedures in the Medicare data available to us. In a separate analysis, we modeled theoretical scenarios defining the percentage of patients who would have received unnecessary vascular access procedures if vascular access had been placed at cohort entry based on different threshold levels of eGFR. Unnecessary procedures were defined as those performed in patients who did not start dialysis (and thus would not have used their access) within a given time frame. For these analyses, we varied the time frame over which procedures were classified as necessary or 559

original article

unnecessary from 6 months up to 2 years. We were able to use a longer follow-up time for this analysis than for the analysis described above, because we were not examining actual access placement procedures (for which we had more limited follow-up). The study was approved by the Institutional Review Board at the University of California, San Francisco and the Research and Development Committee at the VA San Francisco. Analysis All analyses were conducted using Stata version 8.1. (Stata Corporation, College Station Texas). We examined age differences in mean eGFR at cohort entry and in median rate of eGFR decline based on repeat creatinine measurements after cohort entry. For each patient we identified the most recent serum creatinine level at least 90 days after cohort entry and at least 60 days before onset of ESRD, death or 30 September 2003 (whichever came first). To calculate rate of eGFR decline, we then subtracted the eGFR at cohort entry from the most recent eGFR during this interval, and divided by the exact time between the two measurements. Because the distribution of rate of eGFR decline was skewed, we report median rather than mean annual decline in eGFR for each age group. Patients who did not have a follow-up creatinine measurement during the aforementioned time window (n ¼ 2273) were excluded from this analysis. To evaluate the impact of age on vascular access placement practices, we examined permanent access placement at different ages both among all cohort patients and among the subgroup that initiated dialysis within a year of cohort entry. For each point estimate, we calculated exact 95% confidence intervals based on the t-distribution. To understand the extent to which permanent access placement was targeted toward patients who started dialysis and away from patients who died without initiating dialysis during follow-up, we classified patients according to whether they initiated dialysis, died, or survived for 1 year without needing dialysis. Patients with each outcome were further classified according to whether or not they had been referred for pre-dialysis permanent access placement either before cohort entry or during 1 year follow-up. These analyses were stratified by age group. To determine whether our primary findings could be attributed to differences across age groups in patient characteristics other than age that were associated with mortality in this cohort, we developed a propensity score to capture each patient’s propensity to die within 1 year of cohort entry based on race, sex, eGFR at cohort entry and co-morbid conditions defined by ICD-9 and CPT codes before cohort entry in in-patient and outpatient encounters under the VA or Medicare. Specifically, we generated the propensity score using logistic regression analysis where the dependent variable was death within 1 year of cohort entry and the independent variables were race (black, white, other, missing), sex, and diagnosed diabetes, coronary artery disease, congestive heart failure, peripheral arterial disease, chronic obstructive lung disease, cerebrovascular disease, and dementia. Patients were then divided by quartile of propensity score and the primary analysis was repeated among patients with propensity scores in the lowest and highest quartiles, respectively. To determine whether lower rates of access placement in older patients were due to lower rates of referral to nephrology in the elderly, we repeated the primary analysis among patients who had and had not seen a nephrologist by 1 year after cohort entry. We defined nephrology referral as a visit to nephrology clinic within the VA or an outpatient encounter with a nephrology provider under 560


Medicare. We specified that the initial nephrology appointment must have occurred before the initial pre-dialysis access placement procedure among those who underwent this procedure and before initiation of dialysis among those who initiated dialysis without having undergone prior permanent access placement. In a separate analysis, we estimated the number of unnecessary procedures that would have occurred in each age group if all patients had received permanent access at cohort entry based on rates of death and dialysis initiation among cohort members. For this analysis, we varied both the level of eGFR at which access would theoretically be placed to correspond approximately with what is recommended in different guidelines (o25, o20, and o15 ml/min/ 1.73 m2). We also varied the follow-up time over which a procedure was judged necessary vs unnecessary (6 months, 1 and 2 years after cohort entry). ACKNOWLEDGMENTS

Dr O’Hare is supported by a Paul B Beeson Career Development Award in Aging (K23 AG28980-01) and was funded by a Research Career Development Award from the Department of Veterans Affairs Health Services Research and Development Service during a portion of the study period. Dr Allon is supported by NIDDK (K24 DK5981801). Dr Walter is supported by an Advanced Research Career Development Award from the Department of Veterans Affairs Health Services Research and Development Service. Mr Bertenthal is supported by the VA San Francisco Health Services Research and Development Program Research Enhancement Award Program. Dr Garg was supported by a Clinician Scientist Award from the Canadian Institutes of Health Research. Dr Covinsky is supported by a grant from the John A Hartford Foundation Inc. (no. 2003-0244). Disclosures Dr Allon serves as a consultant for Arrow International. Dr Kaufman receives grant support from Hoffmann – La Roche and Keryx Pharmaceuticals and serves as a consultant for Hoffmann – La Roche, Amgen, Genzyme, and Advanced Magnetics Inc. These funding sources had no involvement in the design or execution of this study.

REFERENCES 1.

Allon M, Daugirdas J, Depner TA et al. Effect of change in vascular access on patient mortality in hemodialysis patients. Am J Kidney Dis 2006; 47: 469–477. 2. Astor BC, Eustace JA, Powe NR et al. Type of vascular access and survival among incident hemodialysis patients: the Choices for Healthy Outcomes in Caring for ESRD (CHOICE) Study. J Am Soc Nephrol 2005; 16: 1449–1455. 3. Ortega T, Ortega F, Diaz-Corte C et al. The timely construction of arteriovenous fistulae: a key to reducing morbidity and mortality and to improving cost management. Nephrol Dial Transplant 2005; 20: 598–603. 4. Pastan S, Soucie JM, McClellan WM. Vascular access and increased risk of death among hemodialysis patients. Kidney Int 2002; 62: 620–626. 5. III. NKF-K/DOQI. Clinical Practice Guidelines for Vascular Access: update 2000. Am J Kidney Dis 2001; 37(Suppl 1): S137–S181. 6. Allon M, Robbin ML. Increasing arteriovenous fistulas in hemodialysis patients: problems and solutions. Kidney Int 2002; 62: 1109–1124. 7. Astor BC, Eustace JA, Powe NR et al. Timing of nephrologist referral and arteriovenous access use: the CHOICE Study. Am J Kidney Dis 2001; 38: 494–501. 8. Kausz AT, Guo H, Pereira BJ et al. General medical care among patients with chronic kidney disease: opportunities for improving outcomes. J Am Soc Nephrol 2005; 16: 3092–3101. 9. Lee T, Barker J, Allon M. Associations with predialysis vascular access management. Am J Kidney Dis 2004; 43: 1008–1013. 10. Avorn J, Winkelmayer WC, Bohn RL et al. Delayed nephrologist referral and inadequate vascular access in patients with advanced chronic kidney failure. J Clin Epidemiol 2002; 55: 711–716. 11. Jindal K, Chan CT, Deziel C et al. Hemodialysis clinical practice guidelines for the Canadian Society of Nephrology. J Am Soc Nephrol 2006; 17(Suppl 1): S1–S27. Kidney International (2007) 71, 555–561

original article


12. 13. 14.

15.

16.

Vascular Access Work Group. Clinical practice guidelines for vascular access. Am J Kidney Dis 2006; 48(Suppl 1): S248–S273. Vascular Access Work Group. Clinical practice guidelines for vascular access. Am J Kidney Dis 2006; 48(Suppl 1): S176–S247. Klein R, Klein BE, Moss SE et al. The 10-year incidence of renal insufficiency in people with type 1 diabetes. Diabetes Care 1999; 22: 743–751. Vikse BE, Bostad L, Aasarod K et al. Prognostic factors in mesangioproliferative glomerulonephritis. Nephrol Dial Transplant 2002; 17: 1603–1613. Drey N, Roderick P, Mullee M, Rogerson M. A population-based study of the incidence and outcomes of diagnosed chronic kidney disease. Am J Kidney Dis 2003; 42: 677–684.

Kidney International (2007) 71, 555–561

17.

18.

19.

20.

Evans M, Fryzek JP, Elinder CG et al. The natural history of chronic renal failure: results from an unselected, population-based, inception cohort in Sweden. Am J Kidney Dis 2005; 46: 863–870. Eriksen BO, Ingebretsen OC. The progression of chronic kidney disease: a 10-year population-based study of the effects of gender and age. Kidney Int 2006; 69: 375–382. Iseki K, Iseki C, Ikemiya Y, Fukiyama K. Risk of developing end-stage renal disease in a cohort of mass screening. Kidney Int 1996; 49: 800–805. Levey AS, Bosch JP, Lewis JB et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999; 130: 461–470.

561