Hemodialysis International 2014; 18:616–624
Association between vascular access type and patient mortality among elderly patients on hemodialysis in Canada Joyce C. ZHANG,1,2 Ahmed A. AL-JAISHI,1,3 Yingbo NA,4 Eric de SA,4 Louise M. MOIST1,2,5 1 Kidney Clinical Research Unit, London Health Sciences Centre, London, Ontario, Canada; 2Schulich School of Medicine and Dentistry, 5Department of Epidemiology and Biostatistics, Western University, London, Ontario, Canada; 3Institute for Clinical Evaluative Sciences, Ontario, Canada; 4Canadian Institute for Health Information and Canadian Organ Replacement Registry, Toronto, Ontario, Canada
Abstract Optimal vascular access in elderly patients requires consideration of the benefits and risks in a population with increased comorbidity and mortality. Our objective was to examine the association between vascular access type and patient mortality by age category among incident adult hemodialysis patients registered in the Canadian Organ Replacement Register between 2001 and 2010. We also describe the secular trend in incident and prevalent vascular access use. We used a Cox proportional hazards model to evaluate the overall mortality in patients aged less than 65, 65–74, 75–85, and greater than 85 years who initiated hemodialysis using a central venous catheter (catheter) or arteriovenous (AV)-access (fistula or graft) using an intention-to-treat approach. The cohort of 39,721 patients consisted of 42%, 27%, 26%, and 5% of patients aged 85, respectively. Patients who initiated hemodialysis using an AV-access constituted 21%, 22%, 20%, and 15% of each age category. AV access use was associated with lower adjusted mortality compared with catheter use in each age category (Hazard Ratios [HR], 0.67; 95% Confidence Interval [0.62–0.72]; HR, 0.76 [0.63–0.91]; HR, 0.77 [0.64–0.93], HR, 0.73 [0.56–0.96], respectively). In Canada, use of an AV-access is associated with lower mortality across all age categories, even in the very elderly. Further studies are required to understand the patient preference, complications, and resource use when selecting access type in the elderly. Key words: Canada, elderly, hemodialysis, vascular access
INTRODUCTION The incidence of end-stage kidney disease (ESKD) in patients over the age of 65 years has steadily increased over the last two decades in several countries, including Correspondence to: L. Moist, MSc, MD, FRCPC, Schulich School of Medicine, Western University; Kidney Clinical Research Unit, London Health Sciences Centre, Victoria Hospital, 800 Commissioners Rd E. London, ON, Canada N6A 5W9. E-mail: [email protected] Conflict of interest: None to declare.
Canada.1–4 Patients over the age of 65, considered the elderly, are more likely to choose hemodialysis as their primary mode of renal replacement therapy (RRT). In 2011, although the elderly patient population accounted for 42% of all existing ESKD patients in Canada, they comprised 56% of those receiving hemodialysis. The proportion of very elderly (75+) patients with ESKD also increased from 15% to 19% between 2001 and 2011, accounting for 32% of the total hemodialysis patient population in Canada.1 Clinical practice guidelines endorse the use of the arteriovenous fistula or graft (AV-access) over the central
© 2014 International Society for Hemodialysis DOI:10.1111/hdi.12151
616
Vascular access in the elderly
venous catheters (catheter) because of their association with improved patient survival and lower complication rates.5–13 The choice of vascular access is a more complex decision among elderly patients and is not specifically addressed by clinical practice guidelines.5,6,14 The potential benefits of decreased morbidity and mortality with AV-access must be balanced against the increased rates of failure to mature, complications, such as thrombosis and decreased patency rates.15,16 In addition, the risk of death before needing dialysis and anticipated survival on dialysis require consideration when selecting an incident vascular access type.17 Previous studies have demonstrated the superiority of using AV-access in elderly patients.18–21 However, a Canadian perspective is needed given the differences in health care systems, and vascular access use with Canada having one of the highest catheter rates in the world.4 Considering the extensive public policy recommendations and actions to promote AV-access use, we tested the association between type of vascular access use and overall mortality by age category. We also described the trends in vascular access use among incident and prevalent elderly patients starting hemodialysis during the same time period.
METHODS Study design We conducted a retrospective observational cohort study of consecutive adult patients (age ≥18 years at the start of RRT), registered in the Canadian Organ Replacement Register (CORR), who commenced hemodialysis as their first form of RRT between January 1, 2001 and December 31, 2010. Patients without documentation for initial vascular access type were excluded.
Data source and elements CORR is a national registry, administered by the Canadian Institute for Health Information that collects and records the incidence, prevalence, treatment changes, and outcomes of all chronic dialysis and solid organ transplant patients in Canada. Data were collected by voluntary completion of survey forms for each patient at dialysis initiation and annually every October 31. Clinical and demographical variables and comorbidities were collected by individual facilities at the start of dialysis. For this study, AV-access included both fistula and graft and central venous catheters (catheter) included temporary, permanent cuffed, and noncuffed catheters. Patients
Hemodialysis International 2014; 18:616–624
were grouped based on the following age categories: 85 years. The presence of comorbidities were documented as “yes”, “no”, or “unknown”. Patients with “unknown” or “no” were considered as not having the comorbidity of interest. CORR comorbidity data has recently been validated for accuracy.22 The associated risks of mortality using the CORR data or the medical record data used for validation were similar, suggesting that CORR data can be used in clinical research with minimal concern for bias.22 Detailed comorbidity and causes of death definitions can be found at Chronic Renal Failure Patients on Renal Replacement Therapy, Instruction Manual 2010.23 Late referral was defined as first visit to a nephrologist within 3 months of starting dialysis. Body mass index (BMI) was calculated using height and weight collected at the start of dialysis.
Analysis Categorical and continuous variables were compared using chi-square, Wilcoxon two-sample, or independent two-sample Student’s t test as appropriate.24 We calculated mortality rates per 10,000 person-years for each age group stratified by initial vascular access type used at hemodialysis start. We used a Cox proportional hazards model to assess the risk of death with an average follow-up time of 1103.21 days. Hazard ratios (HR) and corresponding 95% confidence interval were adjusted for patient variables in our cohort including: initial vascular access type, age group, gender, race, hemodialysis initiation year, province of treatment, primary cause of ESKD, late dialysis referral, BMI, last predialysis laboratory values (serum creatinine, albumin, and hemoglobin), and weighting of comorbidities (myocardial infarction, congestive heart failure, peripheral vascular disease, cerebral vascular disease, dementia, chronic lung disease, rheumatological diseases, peptic ulcer disease, diabetes, diabetes with complications, moderate/severe liver disease, metastatic disease, leukemia, and lymphoma) on the basis of a validated ESKD comorbidity index.25 We imputed missing data for BMI (kg/m2), predialysis estimated glomerular filtration rate (eGFR; mL/minute/1.73 m2), hemoglobin (g/L), and albumin (g/L) using multiple imputation Markov chain Monte Carlo method to avoid excluding patients with missing values.26,27 Complete variables chosen to impute missing data were age, sex, race, comorbidity index, year of hemodialysis initiation, BMI, eGFR, hemoglobin, albumin, and early vs. late referral to a nephrologist. We censored patients at modality switch (from hemodialysis to peritoneal dialysis), kidney transplantation, lost to follow-up, or withdrawal from dialysis. We performed two
617
Zhang et al.
sensitivity analyses: (1) excluding patients who used arteriovenous grafts from the AV-access group, and (2) excluding patients who used temporary catheters from the catheter group, to estimate the effect of AV-access use on the risk of mortality compared with catheter use. All tests were two-sided and P-values less than 0.05 were considered statistically significant. SAS version 9.2 (SAS Institute, Cary, NC, USA) was used to conduct all statistical analyses.
RESULTS Baseline characteristics In our study, 42,117 patients initiated hemodialysis between January 1, 2001 and December 31, 2010, with 2396 (5.7%) excluded because of missing incident vascular access information. Among patients included (n = 39,721), 42% were 85 years of age. AV grafts accounted for 9% (n = 725) of all incident AV-access use, whereas temporary catheters accounted for 47% (n = 14,825) of all incident catheter use. Of each age category, 21%, 22%, 20%, and 15% of patients initiated hemodialysis with an AV-access, respectively. Table 1 shows the patients’ baseline characteristics by age category and vascular access type. In general, across all age categories, female patients, patients with increased comorbidities, lower BMI, hemoglobin, serum albumin, eGFR, and late referrals were more likely to use catheter than AV-access.
Mortality by vascular access and age category Compared with patients using a catheter, patients using an AV-access at hemodialysis initiation had lower unadjusted absolute mortality rate per 10,000 patient-years for all age groups (Table 2). The Kaplan–Meier survival estimate at 5 years were 69.7%, 48.6%, 35.1%, and 24.5% for patients using an AV-access vs. 53.6%, 33.5%, 23.3%, and 15.0% for patients using a catheter in each age category (Figure 1). Unadjusted and adjusted HR for patient survival by age category are reported in Table 3. Use of an AV-access at hemodialysis initiation was associated with a 33%, 24%, 23%, and 27% reduction in adjusted hazard rate of death compared with the use of a catheter for each age category, respectively (HR, 0.67; 95% confidence interval [0.62, 0.72]; HR, 0.76 [0.63, 0.91]; HR, 0.77 [0.64, 0.93], and HR, 0.73 [0.56, 0.96]) (Table 3). Excluding patients who used arteriovenous grafts from the AV-access group and patients who used temporary cath-
618
eters from the catheter group did not result in any significant change from the primary analysis (Table 3). We explored prespecified interaction between age categories and vascular access type. When comparing patients across age categories, age did not modify the reduction in overall mortality risk when patients initiated hemodialysis with AV-access compared with those who started with catheter (Table 3). Other variables associated with the risk of death in these patients are presented in Table 4.
Secular trends of incident and prevalent vascular access use by age category The use of AV-access among incident hemodialysis patients remained consistently low from 2001 to 2010, on average accounting for 19.4%, 20.8%, 19.1%, and 14.3% of all vascular access types in each age category, with no significant change in trend overtime (Figure 2A). Over the same period, the use of AV-access among prevalent hemodialysis patients decreased significantly for all age categories (P < 0.0001), with each group experiencing percentage drop of 14.6%, 14.3%, 13.3%, and 13.0%, respectively (Figure 2B).
DISCUSSION Older age is often considered a barrier to the creation and use of an AV-access for hemodialysis. Given the lengthy and complicated process of establishing a successful AV-access, clinicians are faced with the uncertainty if older patients can realize the associated survival benefits cited in literature.5–13 However, we found that the association between using an initial AV-access and reduced mortality persisted across age categories, even after adjustment for baseline variables. In those over 65 years of age, risk reduction was consistent among elderly and very elderly. Our study largely agrees with findings from other countries reporting the association of increased mortality with incident catheter use elderly hemodialysis patients.18–21 In a similar population in the United States (>67 years), Xue et al.20 reported a 70% associated increase in the 1-year mortality with catheter use compared with fistula use. Similarly, Ocak et al.18 identified a 54% increase in allcause mortality among incident hemodialysis patients over 65 years using catheters compared with AV-access in Netherlands. Unlike these studies, we divided the elderly patient population further into graded age categories to closely examine the association with mortality. Similar to the United States Renal Data System study,19 we found a persistent association with increased survival for
Hemodialysis International 2014; 18:616–624
Hemodialysis International 2014; 18:616–624 46.1 53.8 68 (56,77) 41.3 75.1 5 3.5 11.8 4.6 9.8 2.1 35.3 19.8 2.1 3.7 13 14.2 12.6 35.4 20.2 14.5 13 13.9 28.3 79.4 13.9 26.1 (22.7, 30.6) 98.1 ± 17.5 31 ± 7 10.1 ± 6.8 48.2
51.1 49 68 (56,76) 34.9 75.7 5.8 2.8 11.2 4.6 12.1 2.2 38.6 20.1 7.6 3.9 5.2 10.3 10 31.3 17.8 12.7 10.9 12.3 18.5 86.3 12 27.3 (23.7, 31.8) 107.3 ± 16.3 35 ± 5.9 9.7 ± 4.3 3.3
Catheter n = 31,502
3.5
7.9 21.1 13 8 4.6 8.2 12.3 85.3 16.7 28.4 (24.2, 34) 106.9 ± 16.8 35.3 ± 6.3 9.1 ± 4.2
16.8 2.5 40.6 8.1 12.6 4.4 6.4 8.7
71.8 4.7 3.4 15.8 4.3
52.1 47.8 54 (46, 60) 33.2
AV-access n = 3466
49.2
9.8 23.4 15.9 9.4 7.9 9.2 21.6 77.2 19.6 26.9 (23.1, 32.2) 95.7 ± 18.2 30.5 ± 7.4 9.6 ± 7.4
13.9 2.6 40.4 9.4 3.4 3.7 14.7 11.8
68.1 4.8 5.3 17.6 4.2
45.0 55.0 54 (44, 60) 39.8
Catheter n = 13,189
85 n = 2057
Coronary artery disease (CAD) is defined as the presence of history of at least one of the following: angina, myocardial infarction, coronary artery bypass graft, percutaneous coronary intervention. AV-access = arteriovenous access (include fistula and graft); BMI = body mass index; Catheter = central venous catheter, COPD = chronic obstructive pulmonary disease; CVD = cerebral vascular disease; eGFR = estimated glomerular filtration rate, calculated using the modification of diet in renal disease equation,28 in mL/minute/1.73 m2; IQR = interquartile range; Late referral = seeing a nephrologist within 3 months of hemodialysis initiation; PKD = polycystic kidney disease; PVD = peripheral vascular disease; RVD = renal vascular disease.
a
Hemoglobin (mean ± SD) (g/L) Albumin (mean ± SD) (g/L) eGFR (mean ± SD) (mL/minute per 1.73 m2) Late referral (%)
Year period of dialysis initiation (%) 2001–2005 2006–2010 Age (median, IQR) Female sex (%) Race (%) Caucasian Asian Black Other Unknown Primary renal diagnosis (%) Glomerulonephritis Drug related Diabetes RVD PKD Pyelonephritis Other Unknown Comorbidities (%) Diabetes mellitus CADa PVD CVD Malignancy Lung disease (COPD) Pulmonary edema Hypertension Smoking BMI (median, IQR) (kg/m2)
AV-access n = 8219
All n = 39,721
Table 1 Baseline characteristics of adult patients (≥18) at hemodialysis initiation by age category and vascular access type, 2001 to 2010
Vascular access in the elderly
619
Zhang et al.
Table 2 Unadjusted all-cause mortality rate per 10,000 patient-years Age category (years)
Catheter
AV-access
3.52 6.25 8.26 10.76
1.95 3.99 5.43 6.78
85
AV-access = arteriovenous access (include fistula and graft); Catheter = central venous catheter.
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
B
AV access
Survival probability
Survival probability
A
catheter
0
365
730
1095
1460
AV-access across all age categories with little change in the HR across age categories. Most recently, an updated United States Renal Data System study by DeSilva et al. examining the effect of first vascular access on mortality continues to support the association between catheter use and increased mortality in patients over 67 years of age.29 Despite this evidence and numerous endorsements from clinical guidelines and initiatives to alter practice patterns, we found that AV-access use at hemodialysis
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
1825
AV access
catheter
0
365
Days after dialysis start N at risk catheter 13187 AV 3466
5460 1994
3757 1443
2546 1023
1657 680
D
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
AV access
catheter
0
365
730
1095
1460
8354 2375
4850 1868
4664 1529
3069 1130
1951 796
1195 496
3356 1461
2310 1089
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
1460
1825
975 510
catheter 0
1825
1550 745
AV access
365
730
1095
1460
1825
Days after dialysis start
Days after dialysis start N at risk catheter 8201 AV 2065
1095
Days after dialysis start N at risk catheter AV
Survival Probability
Survival probability
C
7841 2683
730
693 305
N at risk catheter AV
1745 309
918 214
512 153
274 91
157 53
78 23
Figure 1 Kaplan–Meier survival curves for arteriovenous (AV)-access vs. catheter in hemodialysis patients aged (A) 85. Results are adjusted for age, race, gender, era of dialysis initiation, end-stage renal disease comorbidity index, primary renal diagnosis, serum albumin, estimated glomerular filtration rate, province of treatment, and late referral.
620
Hemodialysis International 2014; 18:616–624
Vascular access in the elderly
Table 3 Vascular access type and all-cause mortality in adult hemodialysis patients by age category Sensitivity analysis 1a Hazard ratio (95% confidence interval)
Full model Hazard ratio (95% confidence interval) Unadjusted
AV-access vs. catheter 85
0.57 0.65 0.66 0.65
(0.56, (0.63, (0.64, (0.60,
Adjusted
0.60) 0.68) 0.68) 0.70)
0.67 0.76 0.77 0.73
(0.62, (0.63, (0.64, (0.56,
Sensitivity analysis 2b Hazard ratio (95% confidence interval)
Adjusted
0.72) 0.91) 0.93) 0.96)
0.66 0.74 0.76 0.73
(0.64, (0.72, (0.74, (0.68,
Adjusted
0.69) 0.77) 0.79) 0.79)
0.69 0.78 0.79 0.80
(0.67, (0.75, (0.76, (0.74,
0.72) 0.81) 0.81) 0.87)
a
Exclude patients using arteriovenous grafts, bExclude patients using a temporary central venous catheter. Results are adjusted for age, race, gender, era of dialysis initiation, end-stage renal disease comorbidity index, primary renal diagnosis, serum albumin, estimated glomerular filtration rate, province of treatment, and late referral. AV-access = arteriovenous access (include fistula and graft); Catheter = central venous catheter.
A
100 90 80
Percent (%)
70 60 50 40 30 20 10 0 2001
2002
2003
85
100 90 80
Percent (%)
70 60 50 40 30 20 10 0 2001
2002
85
Figure 2 (A) Incident and (B) prevalent adult (>18) hemodialysis patients using initial arteriovenous access by age category, Canada, 2001 to 2010.
Hemodialysis International 2014; 18:616–624
621
Zhang et al.
Table 4 Association between clinical and demographic variables and all-cause mortality in adult hemodialysis patients Variable Vascular access Catheter AV-access Age (years) 85 Sex (male) Race Caucasian Asian Black Other Unknown Dialysis year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Province of residence Ontario Alberta British Columbia Manitoba New Brunswick Newfoundland Nova Scotia Quebec Saskatchewan Primary renal diagnosis Glomerulonephritis Diabetes Renal vascular disease Polycystic kidney disease Other Unknown Adapted Charlson comorbidity index 0 1 2 3 4 5 6 7+ Late referral BMI Normal (18.5–24.9) Under weight (30) Predialysis laboratory values Hemoglobin (g/L) Albumin (g/L) eGFR (mL/minute)
Adjusted HR (95% CI)
P-value
Reference 0.67 (0.62, 0.72)
Association between vascular access type and patient mortality among elderly patients on hemodialysis in Canada Joyce C. ZHANG,1,2 Ahmed A. AL-JAISHI,1,3 Yingbo NA,4 Eric de SA,4 Louise M. MOIST1,2,5 1 Kidney Clinical Research Unit, London Health Sciences Centre, London, Ontario, Canada; 2Schulich School of Medicine and Dentistry, 5Department of Epidemiology and Biostatistics, Western University, London, Ontario, Canada; 3Institute for Clinical Evaluative Sciences, Ontario, Canada; 4Canadian Institute for Health Information and Canadian Organ Replacement Registry, Toronto, Ontario, Canada
Abstract Optimal vascular access in elderly patients requires consideration of the benefits and risks in a population with increased comorbidity and mortality. Our objective was to examine the association between vascular access type and patient mortality by age category among incident adult hemodialysis patients registered in the Canadian Organ Replacement Register between 2001 and 2010. We also describe the secular trend in incident and prevalent vascular access use. We used a Cox proportional hazards model to evaluate the overall mortality in patients aged less than 65, 65–74, 75–85, and greater than 85 years who initiated hemodialysis using a central venous catheter (catheter) or arteriovenous (AV)-access (fistula or graft) using an intention-to-treat approach. The cohort of 39,721 patients consisted of 42%, 27%, 26%, and 5% of patients aged 85, respectively. Patients who initiated hemodialysis using an AV-access constituted 21%, 22%, 20%, and 15% of each age category. AV access use was associated with lower adjusted mortality compared with catheter use in each age category (Hazard Ratios [HR], 0.67; 95% Confidence Interval [0.62–0.72]; HR, 0.76 [0.63–0.91]; HR, 0.77 [0.64–0.93], HR, 0.73 [0.56–0.96], respectively). In Canada, use of an AV-access is associated with lower mortality across all age categories, even in the very elderly. Further studies are required to understand the patient preference, complications, and resource use when selecting access type in the elderly. Key words: Canada, elderly, hemodialysis, vascular access
INTRODUCTION The incidence of end-stage kidney disease (ESKD) in patients over the age of 65 years has steadily increased over the last two decades in several countries, including Correspondence to: L. Moist, MSc, MD, FRCPC, Schulich School of Medicine, Western University; Kidney Clinical Research Unit, London Health Sciences Centre, Victoria Hospital, 800 Commissioners Rd E. London, ON, Canada N6A 5W9. E-mail: [email protected] Conflict of interest: None to declare.
Canada.1–4 Patients over the age of 65, considered the elderly, are more likely to choose hemodialysis as their primary mode of renal replacement therapy (RRT). In 2011, although the elderly patient population accounted for 42% of all existing ESKD patients in Canada, they comprised 56% of those receiving hemodialysis. The proportion of very elderly (75+) patients with ESKD also increased from 15% to 19% between 2001 and 2011, accounting for 32% of the total hemodialysis patient population in Canada.1 Clinical practice guidelines endorse the use of the arteriovenous fistula or graft (AV-access) over the central
© 2014 International Society for Hemodialysis DOI:10.1111/hdi.12151
616
Vascular access in the elderly
venous catheters (catheter) because of their association with improved patient survival and lower complication rates.5–13 The choice of vascular access is a more complex decision among elderly patients and is not specifically addressed by clinical practice guidelines.5,6,14 The potential benefits of decreased morbidity and mortality with AV-access must be balanced against the increased rates of failure to mature, complications, such as thrombosis and decreased patency rates.15,16 In addition, the risk of death before needing dialysis and anticipated survival on dialysis require consideration when selecting an incident vascular access type.17 Previous studies have demonstrated the superiority of using AV-access in elderly patients.18–21 However, a Canadian perspective is needed given the differences in health care systems, and vascular access use with Canada having one of the highest catheter rates in the world.4 Considering the extensive public policy recommendations and actions to promote AV-access use, we tested the association between type of vascular access use and overall mortality by age category. We also described the trends in vascular access use among incident and prevalent elderly patients starting hemodialysis during the same time period.
METHODS Study design We conducted a retrospective observational cohort study of consecutive adult patients (age ≥18 years at the start of RRT), registered in the Canadian Organ Replacement Register (CORR), who commenced hemodialysis as their first form of RRT between January 1, 2001 and December 31, 2010. Patients without documentation for initial vascular access type were excluded.
Data source and elements CORR is a national registry, administered by the Canadian Institute for Health Information that collects and records the incidence, prevalence, treatment changes, and outcomes of all chronic dialysis and solid organ transplant patients in Canada. Data were collected by voluntary completion of survey forms for each patient at dialysis initiation and annually every October 31. Clinical and demographical variables and comorbidities were collected by individual facilities at the start of dialysis. For this study, AV-access included both fistula and graft and central venous catheters (catheter) included temporary, permanent cuffed, and noncuffed catheters. Patients
Hemodialysis International 2014; 18:616–624
were grouped based on the following age categories: 85 years. The presence of comorbidities were documented as “yes”, “no”, or “unknown”. Patients with “unknown” or “no” were considered as not having the comorbidity of interest. CORR comorbidity data has recently been validated for accuracy.22 The associated risks of mortality using the CORR data or the medical record data used for validation were similar, suggesting that CORR data can be used in clinical research with minimal concern for bias.22 Detailed comorbidity and causes of death definitions can be found at Chronic Renal Failure Patients on Renal Replacement Therapy, Instruction Manual 2010.23 Late referral was defined as first visit to a nephrologist within 3 months of starting dialysis. Body mass index (BMI) was calculated using height and weight collected at the start of dialysis.
Analysis Categorical and continuous variables were compared using chi-square, Wilcoxon two-sample, or independent two-sample Student’s t test as appropriate.24 We calculated mortality rates per 10,000 person-years for each age group stratified by initial vascular access type used at hemodialysis start. We used a Cox proportional hazards model to assess the risk of death with an average follow-up time of 1103.21 days. Hazard ratios (HR) and corresponding 95% confidence interval were adjusted for patient variables in our cohort including: initial vascular access type, age group, gender, race, hemodialysis initiation year, province of treatment, primary cause of ESKD, late dialysis referral, BMI, last predialysis laboratory values (serum creatinine, albumin, and hemoglobin), and weighting of comorbidities (myocardial infarction, congestive heart failure, peripheral vascular disease, cerebral vascular disease, dementia, chronic lung disease, rheumatological diseases, peptic ulcer disease, diabetes, diabetes with complications, moderate/severe liver disease, metastatic disease, leukemia, and lymphoma) on the basis of a validated ESKD comorbidity index.25 We imputed missing data for BMI (kg/m2), predialysis estimated glomerular filtration rate (eGFR; mL/minute/1.73 m2), hemoglobin (g/L), and albumin (g/L) using multiple imputation Markov chain Monte Carlo method to avoid excluding patients with missing values.26,27 Complete variables chosen to impute missing data were age, sex, race, comorbidity index, year of hemodialysis initiation, BMI, eGFR, hemoglobin, albumin, and early vs. late referral to a nephrologist. We censored patients at modality switch (from hemodialysis to peritoneal dialysis), kidney transplantation, lost to follow-up, or withdrawal from dialysis. We performed two
617
Zhang et al.
sensitivity analyses: (1) excluding patients who used arteriovenous grafts from the AV-access group, and (2) excluding patients who used temporary catheters from the catheter group, to estimate the effect of AV-access use on the risk of mortality compared with catheter use. All tests were two-sided and P-values less than 0.05 were considered statistically significant. SAS version 9.2 (SAS Institute, Cary, NC, USA) was used to conduct all statistical analyses.
RESULTS Baseline characteristics In our study, 42,117 patients initiated hemodialysis between January 1, 2001 and December 31, 2010, with 2396 (5.7%) excluded because of missing incident vascular access information. Among patients included (n = 39,721), 42% were 85 years of age. AV grafts accounted for 9% (n = 725) of all incident AV-access use, whereas temporary catheters accounted for 47% (n = 14,825) of all incident catheter use. Of each age category, 21%, 22%, 20%, and 15% of patients initiated hemodialysis with an AV-access, respectively. Table 1 shows the patients’ baseline characteristics by age category and vascular access type. In general, across all age categories, female patients, patients with increased comorbidities, lower BMI, hemoglobin, serum albumin, eGFR, and late referrals were more likely to use catheter than AV-access.
Mortality by vascular access and age category Compared with patients using a catheter, patients using an AV-access at hemodialysis initiation had lower unadjusted absolute mortality rate per 10,000 patient-years for all age groups (Table 2). The Kaplan–Meier survival estimate at 5 years were 69.7%, 48.6%, 35.1%, and 24.5% for patients using an AV-access vs. 53.6%, 33.5%, 23.3%, and 15.0% for patients using a catheter in each age category (Figure 1). Unadjusted and adjusted HR for patient survival by age category are reported in Table 3. Use of an AV-access at hemodialysis initiation was associated with a 33%, 24%, 23%, and 27% reduction in adjusted hazard rate of death compared with the use of a catheter for each age category, respectively (HR, 0.67; 95% confidence interval [0.62, 0.72]; HR, 0.76 [0.63, 0.91]; HR, 0.77 [0.64, 0.93], and HR, 0.73 [0.56, 0.96]) (Table 3). Excluding patients who used arteriovenous grafts from the AV-access group and patients who used temporary cath-
618
eters from the catheter group did not result in any significant change from the primary analysis (Table 3). We explored prespecified interaction between age categories and vascular access type. When comparing patients across age categories, age did not modify the reduction in overall mortality risk when patients initiated hemodialysis with AV-access compared with those who started with catheter (Table 3). Other variables associated with the risk of death in these patients are presented in Table 4.
Secular trends of incident and prevalent vascular access use by age category The use of AV-access among incident hemodialysis patients remained consistently low from 2001 to 2010, on average accounting for 19.4%, 20.8%, 19.1%, and 14.3% of all vascular access types in each age category, with no significant change in trend overtime (Figure 2A). Over the same period, the use of AV-access among prevalent hemodialysis patients decreased significantly for all age categories (P < 0.0001), with each group experiencing percentage drop of 14.6%, 14.3%, 13.3%, and 13.0%, respectively (Figure 2B).
DISCUSSION Older age is often considered a barrier to the creation and use of an AV-access for hemodialysis. Given the lengthy and complicated process of establishing a successful AV-access, clinicians are faced with the uncertainty if older patients can realize the associated survival benefits cited in literature.5–13 However, we found that the association between using an initial AV-access and reduced mortality persisted across age categories, even after adjustment for baseline variables. In those over 65 years of age, risk reduction was consistent among elderly and very elderly. Our study largely agrees with findings from other countries reporting the association of increased mortality with incident catheter use elderly hemodialysis patients.18–21 In a similar population in the United States (>67 years), Xue et al.20 reported a 70% associated increase in the 1-year mortality with catheter use compared with fistula use. Similarly, Ocak et al.18 identified a 54% increase in allcause mortality among incident hemodialysis patients over 65 years using catheters compared with AV-access in Netherlands. Unlike these studies, we divided the elderly patient population further into graded age categories to closely examine the association with mortality. Similar to the United States Renal Data System study,19 we found a persistent association with increased survival for
Hemodialysis International 2014; 18:616–624
Hemodialysis International 2014; 18:616–624 46.1 53.8 68 (56,77) 41.3 75.1 5 3.5 11.8 4.6 9.8 2.1 35.3 19.8 2.1 3.7 13 14.2 12.6 35.4 20.2 14.5 13 13.9 28.3 79.4 13.9 26.1 (22.7, 30.6) 98.1 ± 17.5 31 ± 7 10.1 ± 6.8 48.2
51.1 49 68 (56,76) 34.9 75.7 5.8 2.8 11.2 4.6 12.1 2.2 38.6 20.1 7.6 3.9 5.2 10.3 10 31.3 17.8 12.7 10.9 12.3 18.5 86.3 12 27.3 (23.7, 31.8) 107.3 ± 16.3 35 ± 5.9 9.7 ± 4.3 3.3
Catheter n = 31,502
3.5
7.9 21.1 13 8 4.6 8.2 12.3 85.3 16.7 28.4 (24.2, 34) 106.9 ± 16.8 35.3 ± 6.3 9.1 ± 4.2
16.8 2.5 40.6 8.1 12.6 4.4 6.4 8.7
71.8 4.7 3.4 15.8 4.3
52.1 47.8 54 (46, 60) 33.2
AV-access n = 3466
49.2
9.8 23.4 15.9 9.4 7.9 9.2 21.6 77.2 19.6 26.9 (23.1, 32.2) 95.7 ± 18.2 30.5 ± 7.4 9.6 ± 7.4
13.9 2.6 40.4 9.4 3.4 3.7 14.7 11.8
68.1 4.8 5.3 17.6 4.2
45.0 55.0 54 (44, 60) 39.8
Catheter n = 13,189
85 n = 2057
Coronary artery disease (CAD) is defined as the presence of history of at least one of the following: angina, myocardial infarction, coronary artery bypass graft, percutaneous coronary intervention. AV-access = arteriovenous access (include fistula and graft); BMI = body mass index; Catheter = central venous catheter, COPD = chronic obstructive pulmonary disease; CVD = cerebral vascular disease; eGFR = estimated glomerular filtration rate, calculated using the modification of diet in renal disease equation,28 in mL/minute/1.73 m2; IQR = interquartile range; Late referral = seeing a nephrologist within 3 months of hemodialysis initiation; PKD = polycystic kidney disease; PVD = peripheral vascular disease; RVD = renal vascular disease.
a
Hemoglobin (mean ± SD) (g/L) Albumin (mean ± SD) (g/L) eGFR (mean ± SD) (mL/minute per 1.73 m2) Late referral (%)
Year period of dialysis initiation (%) 2001–2005 2006–2010 Age (median, IQR) Female sex (%) Race (%) Caucasian Asian Black Other Unknown Primary renal diagnosis (%) Glomerulonephritis Drug related Diabetes RVD PKD Pyelonephritis Other Unknown Comorbidities (%) Diabetes mellitus CADa PVD CVD Malignancy Lung disease (COPD) Pulmonary edema Hypertension Smoking BMI (median, IQR) (kg/m2)
AV-access n = 8219
All n = 39,721
Table 1 Baseline characteristics of adult patients (≥18) at hemodialysis initiation by age category and vascular access type, 2001 to 2010
Vascular access in the elderly
619
Zhang et al.
Table 2 Unadjusted all-cause mortality rate per 10,000 patient-years Age category (years)
Catheter
AV-access
3.52 6.25 8.26 10.76
1.95 3.99 5.43 6.78
85
AV-access = arteriovenous access (include fistula and graft); Catheter = central venous catheter.
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
B
AV access
Survival probability
Survival probability
A
catheter
0
365
730
1095
1460
AV-access across all age categories with little change in the HR across age categories. Most recently, an updated United States Renal Data System study by DeSilva et al. examining the effect of first vascular access on mortality continues to support the association between catheter use and increased mortality in patients over 67 years of age.29 Despite this evidence and numerous endorsements from clinical guidelines and initiatives to alter practice patterns, we found that AV-access use at hemodialysis
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
1825
AV access
catheter
0
365
Days after dialysis start N at risk catheter 13187 AV 3466
5460 1994
3757 1443
2546 1023
1657 680
D
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
AV access
catheter
0
365
730
1095
1460
8354 2375
4850 1868
4664 1529
3069 1130
1951 796
1195 496
3356 1461
2310 1089
1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0
1460
1825
975 510
catheter 0
1825
1550 745
AV access
365
730
1095
1460
1825
Days after dialysis start
Days after dialysis start N at risk catheter 8201 AV 2065
1095
Days after dialysis start N at risk catheter AV
Survival Probability
Survival probability
C
7841 2683
730
693 305
N at risk catheter AV
1745 309
918 214
512 153
274 91
157 53
78 23
Figure 1 Kaplan–Meier survival curves for arteriovenous (AV)-access vs. catheter in hemodialysis patients aged (A) 85. Results are adjusted for age, race, gender, era of dialysis initiation, end-stage renal disease comorbidity index, primary renal diagnosis, serum albumin, estimated glomerular filtration rate, province of treatment, and late referral.
620
Hemodialysis International 2014; 18:616–624
Vascular access in the elderly
Table 3 Vascular access type and all-cause mortality in adult hemodialysis patients by age category Sensitivity analysis 1a Hazard ratio (95% confidence interval)
Full model Hazard ratio (95% confidence interval) Unadjusted
AV-access vs. catheter 85
0.57 0.65 0.66 0.65
(0.56, (0.63, (0.64, (0.60,
Adjusted
0.60) 0.68) 0.68) 0.70)
0.67 0.76 0.77 0.73
(0.62, (0.63, (0.64, (0.56,
Sensitivity analysis 2b Hazard ratio (95% confidence interval)
Adjusted
0.72) 0.91) 0.93) 0.96)
0.66 0.74 0.76 0.73
(0.64, (0.72, (0.74, (0.68,
Adjusted
0.69) 0.77) 0.79) 0.79)
0.69 0.78 0.79 0.80
(0.67, (0.75, (0.76, (0.74,
0.72) 0.81) 0.81) 0.87)
a
Exclude patients using arteriovenous grafts, bExclude patients using a temporary central venous catheter. Results are adjusted for age, race, gender, era of dialysis initiation, end-stage renal disease comorbidity index, primary renal diagnosis, serum albumin, estimated glomerular filtration rate, province of treatment, and late referral. AV-access = arteriovenous access (include fistula and graft); Catheter = central venous catheter.
A
100 90 80
Percent (%)
70 60 50 40 30 20 10 0 2001
2002
2003
85
100 90 80
Percent (%)
70 60 50 40 30 20 10 0 2001
2002
85
Figure 2 (A) Incident and (B) prevalent adult (>18) hemodialysis patients using initial arteriovenous access by age category, Canada, 2001 to 2010.
Hemodialysis International 2014; 18:616–624
621
Zhang et al.
Table 4 Association between clinical and demographic variables and all-cause mortality in adult hemodialysis patients Variable Vascular access Catheter AV-access Age (years) 85 Sex (male) Race Caucasian Asian Black Other Unknown Dialysis year 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Province of residence Ontario Alberta British Columbia Manitoba New Brunswick Newfoundland Nova Scotia Quebec Saskatchewan Primary renal diagnosis Glomerulonephritis Diabetes Renal vascular disease Polycystic kidney disease Other Unknown Adapted Charlson comorbidity index 0 1 2 3 4 5 6 7+ Late referral BMI Normal (18.5–24.9) Under weight (30) Predialysis laboratory values Hemoglobin (g/L) Albumin (g/L) eGFR (mL/minute)
Adjusted HR (95% CI)
P-value
Reference 0.67 (0.62, 0.72)
