Hemodialysis International 2014; 18:686–694
Geographic disparities in arteriovenous fistula placement in patients approaching hemodialysis in the United States Alexander S. GOLDFARB-RUMYANTZEV,1,2 Wajih SYED,3 Bhanu K. PATIBANDLA,2 Akshita NARRA,2 Ranil DESILVA,2 Varun CHAWLA,2 Tammy HOD,4 Yael VIN5 1
Division of Nephrology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA; 2Transplant Institute, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA; 3Metrowest Medical Center, Framingham, Massachusetts, USA; 4Division of Nephrology and Center for Vascular Biology Research, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA; 5Section of Interventional Radiology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA
Abstract Arteriovenous fistula (AVF) is the preferred vascular access for hemodialysis (HD). Several factors associated with AVF placement have been identified (e.g., age, sex, race, comorbidities). We hypothesized that geographic location of patient residence might be associated with the probability of AVF placement as the initial access. We used the data from the United States Renal Data System (USRDS) database (2005–2008) linked to Medicare claims (2003–2008). Logistic regression was used to estimate specific characteristics of population associated with the AVF as first access placed or attempted for HD initiation. Our primary variable of interest was the geographic location, and the multivariate model was adjusted for age, sex, race, body mass index, primary cause of end-stage renal disease (ESRD), duration of pre-ESRD nephrology care, comorbidities, employment status, substance abuse, and income. Geographic location was determined using the data collected by the RUCA project and divided population into metropolitan, micropolitan, and rural categories. Patients (n = 111,953) identified from the USRDS database with linked Medicare claims were examined. Rates of fistula placement in the metropolitan, micropolitan, and rural population were 18.5%, 22.4%, and 21.6%, respectively. In comparison, patients who received catheter as the first access were 81.5%, 77.6% and 78.4%, respectively. The odds ratio of AVF placement as a first HD access in the rural and metropolitan population compared with the micropolitan population were 0.96 (0.90–1.03; P = 0.26) and 0.80 (0.76–0.84; P < 0.001), respectively. Our results indicate the presence of geographic disparities in AVF placement with decreased rates of AVF as the first access created in the metropolitan (but not rural) populations compared with the micropolitan communities. Key words: Geographic disparities, arteriovenous fistula, outcome, hemodialysis, access
Correspondence to: A. S. Goldfarb-Rumyantzev, MD, PhD, Division of Nephrology, Beth Israel Deaconess Medical Center, 185 Pilgrim Road, FA-832, Boston, MA 02215, USA. E-mail: [email protected]
INTRODUCTION There are significant benefits to patient survival associated with starting hemodialysis (HD) with an arteriovenous fistula (AVF) compared to a central venous catheter.1
© 2014 International Society for Hemodialysis DOI:10.1111/hdi.12141
686
Geographic disparities in HD vascular access
These benefits include greater long-term AVF patency, superior flow rates, and fewer complications.2–6 The factors affecting the choice of initial vascular access and the presence of functional AVF at the initiation of HD have been discussed in literature.7–9 Studying predictors of lack of AVF at the dialysis initiation may help to develop interventions that can increase AVF placement rates and may decrease the impact of vascular access maintenance on the total cost of end-stage renal disease (ESRD) care. We undertook a population-based study of HD patients, identified using data from the United States Renal Data System (USRDS), to study potential geographic disparities and to determine the size of the patients’ residential community (e.g., macropolitan, micropolitan, and rural) as an independent predictor of the placement of the AVF as an initial access for HD. The concept of geographic disparity is not new as it has been described in access to care and clinical outcome.8,9 In particular, the geographic disparities related to HD access used for the first dialysis treatment have been demonstrated in the past.9 The crucial difference of this project is that we examined the vascular access that was placed initially (first vascular access placed), as opposed to the vascular access that was used for the first HD treatment. Although significant number of vascular access types created in predialysis period are different from the types used for the first dialysis treatment, our approach reflects the disparities and barriers in creation of AVF (as opposed to the survival of AVF from being created to being used in dialysis). The rationale of this approach is based on the fact that “Fistula first” initiative recommends AVF to be the first access attempted. As considerable number of AVFs fail to mature before initiation of dialysis, to measure true geographic disparities in access to health care, the outcome needs to be defined based on the first vascular access procedure.
METHODS Study design and data source This is a retrospective cohort study using national sample of the dialysis patient’s data supplied by the USRDS. Data of patients with ESRD who were initiated on HD between January 1, 2005 and December 31, 2008 in the USDRS system were merged with Medicare institution claims that contain claims for each incident HD year and two years prior to incident year. Inclusion criteria include incident HD patients age ≥67, with the presence of fistula or catheter as the first access placed in preparation for HD. Exclusion criteria included patients not on HD (n = 7784, i.e.,
Hemodialysis International 2014; 18:686–694
patients on peritoneal dialysis), patients with missing access information (n = 18,122), as illustrated in Figure 1. We wanted our population to be uniformly eligible to Medicare to derive vascular access placement information from the Centers for Medicare and Medicaid Services (CMS) billing records. Since billing data are available 2 years prior to initiation of dialysis, we decided to enroll patients 67 years old or older. Geographic population distribution divided into metropolitan, micropolitan, and rural areas (see the definition below) was determined by the Rural-Urban Commuting Area (RUCA) database linked to USRDS by the ZIP code of patient’s residence. In addition, we used information from
Figure 1 The flowchart representing the inclusion and exclusion criteria applied to the United States Renal Data System (USRDS) followed by merging with the Medicare claims to obtain our final study cohort. 1 Unrealistic data probably represent data entry error, e.g., patients who died prior to initiation of dialysis should have been excluded since they do not have information of the first dialysis access used, yet there were 36 such patients identified in the dataset. 2 Among those initiating hemodialysis (HD) with a central venous catheter (CVC) (from Form 2728), patients were excluded from the catheter group if the CPT-4 codes indicated first placement of an arteriovenous fistula (AVF) or an arteriovenous graft (AVG), a maturing AVF or AVG at the time of initiation of HD, or complications of an AVF or AVG that could not be ascertained to be the first access placed, or those whose CVC was removed within 30 days without other permanent access (to avoid including acute kidney injury, rather than end-stage renal disease [ESRD] patients).
687
Goldfarb-Rumyantzev et al.
the US Census Bureau of median income stratified by race also linked to the study dataset by patients’ ZIP codes.
Definition of the study groups We merged the above dataset with the Medicare institutional claims from January 1, 2003 through December 31, 2008 to derive the predialysis access information. We identified patient records (n = 79,411) in the merged database that had information from both the Medicare claims and the USRDS. Patient records were excluded if the first (in chronological order) vascular access billing code was a vascular access complication code (which means that the billing code for actual access placement is missing). After this exclusion, there were 27,702 patients, of whom 21,436 patients had AVF as the first access placed, 3472 patients had arteriovenous graft (AVG) as the first access placed, and 2794 patients had catheter as the first access placed. Furthermore, we found that a substantial number of catheter placements are not reported by using USRDS data linked to Medicare claims. In other words, we found that the number of patients who likely had catheter as the first access did not have an appropriate CPT code in the Medicare data. To address this issue, we defined the catheter group as follows. First, we identified patients who initiated HD with catheter (n = 128,378). Assuming that some of them might have had AVF or AVG not used for the first dialysis treatment (due to stenosis or failure to mature), we excluded those with CPT code identifying AVF or AVG placement (predialysis), CPT code identifying treatment of AVF or AVG complications (predialysis or as a first CPT code after dialysis initiation), or CPT codes for catheter removal 30 days after the initiation of HD (indicating the possibility that immature AVF was possibly present prior to catheter placement). Patients who started dialysis with the catheter and had maturing AVF or AVG (based on the data from CMS Form 2728) were also excluded from this group. The final group of patients identified as starting HD with the catheter included 90,517 subjects. The final study cohort consisted of 115,425 patients (21,436, 3472, and 90,517 patients with AVF, AVG, and catheter as the first access placed, respectively) (Figure 1).
Potential predictors and outcome definition The primary variable of interest in this project was geographic location. Each patient in the cohort was assigned
688
a RUCA code (based on the ZIP code of residence), which classifies geographic areas using measures of population density, urbanization, and daily commuting. The RUCA system uses smaller geographic units and is more precise than the county-based systems. Thus, suburban areas with low population density but close employment associations with large urban centers are classified as urban. The RUCA system includes codes from 1.0 (most urban) to 10.6 (most rural). For this analysis, 3 independent RUCA groups were selected: metropolitan (RUCA 1.0–3.9), which includes cities with a population of more than 50,000 and the associated suburban areas; micropolitan (RUCA 4.0–6.0), for smaller cities or towns of 10,000– 49,999; and rural or small town (RUCA 7.0–10.6), for populations of less than 10,000. The outcome variable was the placement of AVF as first HD access prior to HD initiation compared with catheter (AVG was not included in the analysis). The model was adjusted for the variables that were considered to be potential confounders, specifically age, sex, race, primary cause of ESRD, employment status, and comorbidities. Some of the variables potentially mediating the effect of geographic location on access selection were also included in the model to isolate the effect of true disparities rather than potential objective barriers. We included duration of pre-ESRD nephrology care, body mass index (BMI), substance abuse, and income. The information regarding median annual income stratified by race and ZIP codes was obtained from the US Census Bureau. The comorbidities included in our model were those available in the USRDS database and had previously been identified as important predictors of mortality, general health status, or other outcomes in HD patients. Specifically, we included the history of diabetes, congestive heart failure, peripheral arterial disease, malignancy, and cerebrovascular disease.
Statistical analysis Means and standard deviations were used to summarize continuous variables with normal distribution. Categorical variables were summarized as percent of total. To compare between groups, we used analysis of variance or t test for continuous variables and chi-square for the categorical variables. Logistic regression was used to estimate the independent association of patient’s geographical location with the initial access being AVF (as opposed to central vein HD catheter). The model was adjusted for variables indicated earlier. Data were analyzed using the SAS software (SAS Institute, Cary, NC).
Hemodialysis International 2014; 18:686–694
Geographic disparities in HD vascular access
RESULTS
DISCUSSION
Baseline characteristics
In this project, we studied the association between geographic area of residence and the first HD access placement in patients of 67 years of age or older. A NHANES III study10 has suggested that 6.6 million people older than 60 years have chronic kidney disease (CKD). Also, the older fraction of dialysis population is quickly growing. The number of octogenarians and nonagenarians starting dialysis has nearly doubled, rising from 7054 persons in 1996 to 13,577 persons in 2003.11 The 1- and 2-year survival among patients older than 75 years who have stage 5 CKD has also increased to 84% and 76%, respectively. Addressing the need for optimizing care of these patients, The National Kidney Foundation developed and released clinical practice guidelines and has recommended that an AVF be the preferred first access over an AVG in patients initiating HD.1 This is because the AVF has better patency rates, less need for corrective procedures and complications, and subsequently lower morbidity and cost.7 Long-term use of tunneled catheters is discouraged due to higher infection rate, lower blood flows, less efficient HD, and an increased risk of sepsis, endocarditis, and metastatic infections.2–6 At the same time, there are certain barriers to the creation of an autologous AVF. It requires an adequate arterial and venous vasculature, which can be a challenge in patients who are older, diabetic, obese, and/or female,12 or in those who may have relatively smaller or compromised vascular anatomy. Vessel diameter has also been shown to be an important predictor of AVF survival.13 In other instances, patients who are new to dialysis may have presented too late for the establishment of an AVF before the need for dialysis. Previous published reports have demonstrated that African American population undergoes more AVG placement, suggesting the existence of racial differences in vascular anatomy, increased failure rate or subconscious bias leading to disparities in vascular access strategy. Furthermore, patients belonging to the low socioeconomic status may have limited access to health care,14 potentially resulting in later referral to a nephrologist.14 Other contributing factors may include limited patient education15,16 and failure of AVF to mature. In this project, in addition to the above-mentioned barriers and disparities to AVF placement, we hypothesized the presence of geographic disparities. We used patient-level data from the USRDS, a comprehensive registry of all patients with ESRD, with information such as date of onset of ESRD, demographic, employment status; as well as comorbid conditions at the initiation of dialysis.
Our study cohort included 111,953 patients identified in the USRDS linked with Medicare claims. Of those, 21,436 patients underwent AVF as the first access and the remaining 90,517 patients had catheter used for the initiation of HD. The study population characteristics are presented in Table 1. The age of onset of ESRD of patients in the rural, micropolitan, and metropolitan categories is 76.5, 76.7 and 77.1, respectively. Women represented 46.5%, 47.1%, and 46.9% in each group, whereas there was larger representation of blacks (21.6%) in the metropolitan areas compared with the micropolitan (14.6%) and rural (14.7%) regions (Table 1). Diabetes was the predominant risk factor associated with ESRD across all groups; 34.2% of patients in the rural areas received no pre-ESRD nephrology care compared to 32.1% in the micropolitan and 31.4% in the metropolitan areas.
Association between geographic area and first HD access We used logistic regression model to estimate the odds of AVF as first access compared to catheter in the metropolitan and rural population with the micropolitan population as a reference. The total number of patients in the metropolitan areas was 15,857 in the AVF group and 69,687 in the catheter group. Similarly, the total number of patient in the micropolitan population in the AVF group was 2752 compared to 9533 in the catheter group. In addition, 18.5% of metropolitan patients received AVF as first access compared to 22.4% of patients living in micropolitan areas (P < 0.001). In the metropolitan population, 81.5% patients received catheter compared to 77.6% patients in the micropolitan population and 78.4% patients in the rural areas. These difference in AVF as a first access placed between metropolitan and micropolitan areas remained significant after adjusting for other independent variables described earlier (Table 2). In particular, patient residing in metropolitan areas have lower odds of having AVF as a first dialysis access compared to those living in micropolitan areas (odds ratio [OR] 0.80; P < 0.001). Patients living in rural areas did not demonstrate significant difference with those living in micropolitan areas (reference group) in terms of AVF being the initial HD access (OR 0.96; P = 0.26).
Hemodialysis International 2014; 18:686–694
689
Goldfarb-Rumyantzev et al.
Table 1 Baseline characteristics of the study population at the time of ESRD onseta
First access type AVF fistula Central catheter Age at ESRD onset Age categories (y) 67–79 80–89 >90 Sex Male Female Race White Black Asian Native American Others Body mass index (kg/m2) BMI categories
Geographic disparities in arteriovenous fistula placement in patients approaching hemodialysis in the United States Alexander S. GOLDFARB-RUMYANTZEV,1,2 Wajih SYED,3 Bhanu K. PATIBANDLA,2 Akshita NARRA,2 Ranil DESILVA,2 Varun CHAWLA,2 Tammy HOD,4 Yael VIN5 1
Division of Nephrology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA; 2Transplant Institute, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA; 3Metrowest Medical Center, Framingham, Massachusetts, USA; 4Division of Nephrology and Center for Vascular Biology Research, Department of Medicine, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA; 5Section of Interventional Radiology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts, USA
Abstract Arteriovenous fistula (AVF) is the preferred vascular access for hemodialysis (HD). Several factors associated with AVF placement have been identified (e.g., age, sex, race, comorbidities). We hypothesized that geographic location of patient residence might be associated with the probability of AVF placement as the initial access. We used the data from the United States Renal Data System (USRDS) database (2005–2008) linked to Medicare claims (2003–2008). Logistic regression was used to estimate specific characteristics of population associated with the AVF as first access placed or attempted for HD initiation. Our primary variable of interest was the geographic location, and the multivariate model was adjusted for age, sex, race, body mass index, primary cause of end-stage renal disease (ESRD), duration of pre-ESRD nephrology care, comorbidities, employment status, substance abuse, and income. Geographic location was determined using the data collected by the RUCA project and divided population into metropolitan, micropolitan, and rural categories. Patients (n = 111,953) identified from the USRDS database with linked Medicare claims were examined. Rates of fistula placement in the metropolitan, micropolitan, and rural population were 18.5%, 22.4%, and 21.6%, respectively. In comparison, patients who received catheter as the first access were 81.5%, 77.6% and 78.4%, respectively. The odds ratio of AVF placement as a first HD access in the rural and metropolitan population compared with the micropolitan population were 0.96 (0.90–1.03; P = 0.26) and 0.80 (0.76–0.84; P < 0.001), respectively. Our results indicate the presence of geographic disparities in AVF placement with decreased rates of AVF as the first access created in the metropolitan (but not rural) populations compared with the micropolitan communities. Key words: Geographic disparities, arteriovenous fistula, outcome, hemodialysis, access
Correspondence to: A. S. Goldfarb-Rumyantzev, MD, PhD, Division of Nephrology, Beth Israel Deaconess Medical Center, 185 Pilgrim Road, FA-832, Boston, MA 02215, USA. E-mail: [email protected]
INTRODUCTION There are significant benefits to patient survival associated with starting hemodialysis (HD) with an arteriovenous fistula (AVF) compared to a central venous catheter.1
© 2014 International Society for Hemodialysis DOI:10.1111/hdi.12141
686
Geographic disparities in HD vascular access
These benefits include greater long-term AVF patency, superior flow rates, and fewer complications.2–6 The factors affecting the choice of initial vascular access and the presence of functional AVF at the initiation of HD have been discussed in literature.7–9 Studying predictors of lack of AVF at the dialysis initiation may help to develop interventions that can increase AVF placement rates and may decrease the impact of vascular access maintenance on the total cost of end-stage renal disease (ESRD) care. We undertook a population-based study of HD patients, identified using data from the United States Renal Data System (USRDS), to study potential geographic disparities and to determine the size of the patients’ residential community (e.g., macropolitan, micropolitan, and rural) as an independent predictor of the placement of the AVF as an initial access for HD. The concept of geographic disparity is not new as it has been described in access to care and clinical outcome.8,9 In particular, the geographic disparities related to HD access used for the first dialysis treatment have been demonstrated in the past.9 The crucial difference of this project is that we examined the vascular access that was placed initially (first vascular access placed), as opposed to the vascular access that was used for the first HD treatment. Although significant number of vascular access types created in predialysis period are different from the types used for the first dialysis treatment, our approach reflects the disparities and barriers in creation of AVF (as opposed to the survival of AVF from being created to being used in dialysis). The rationale of this approach is based on the fact that “Fistula first” initiative recommends AVF to be the first access attempted. As considerable number of AVFs fail to mature before initiation of dialysis, to measure true geographic disparities in access to health care, the outcome needs to be defined based on the first vascular access procedure.
METHODS Study design and data source This is a retrospective cohort study using national sample of the dialysis patient’s data supplied by the USRDS. Data of patients with ESRD who were initiated on HD between January 1, 2005 and December 31, 2008 in the USDRS system were merged with Medicare institution claims that contain claims for each incident HD year and two years prior to incident year. Inclusion criteria include incident HD patients age ≥67, with the presence of fistula or catheter as the first access placed in preparation for HD. Exclusion criteria included patients not on HD (n = 7784, i.e.,
Hemodialysis International 2014; 18:686–694
patients on peritoneal dialysis), patients with missing access information (n = 18,122), as illustrated in Figure 1. We wanted our population to be uniformly eligible to Medicare to derive vascular access placement information from the Centers for Medicare and Medicaid Services (CMS) billing records. Since billing data are available 2 years prior to initiation of dialysis, we decided to enroll patients 67 years old or older. Geographic population distribution divided into metropolitan, micropolitan, and rural areas (see the definition below) was determined by the Rural-Urban Commuting Area (RUCA) database linked to USRDS by the ZIP code of patient’s residence. In addition, we used information from
Figure 1 The flowchart representing the inclusion and exclusion criteria applied to the United States Renal Data System (USRDS) followed by merging with the Medicare claims to obtain our final study cohort. 1 Unrealistic data probably represent data entry error, e.g., patients who died prior to initiation of dialysis should have been excluded since they do not have information of the first dialysis access used, yet there were 36 such patients identified in the dataset. 2 Among those initiating hemodialysis (HD) with a central venous catheter (CVC) (from Form 2728), patients were excluded from the catheter group if the CPT-4 codes indicated first placement of an arteriovenous fistula (AVF) or an arteriovenous graft (AVG), a maturing AVF or AVG at the time of initiation of HD, or complications of an AVF or AVG that could not be ascertained to be the first access placed, or those whose CVC was removed within 30 days without other permanent access (to avoid including acute kidney injury, rather than end-stage renal disease [ESRD] patients).
687
Goldfarb-Rumyantzev et al.
the US Census Bureau of median income stratified by race also linked to the study dataset by patients’ ZIP codes.
Definition of the study groups We merged the above dataset with the Medicare institutional claims from January 1, 2003 through December 31, 2008 to derive the predialysis access information. We identified patient records (n = 79,411) in the merged database that had information from both the Medicare claims and the USRDS. Patient records were excluded if the first (in chronological order) vascular access billing code was a vascular access complication code (which means that the billing code for actual access placement is missing). After this exclusion, there were 27,702 patients, of whom 21,436 patients had AVF as the first access placed, 3472 patients had arteriovenous graft (AVG) as the first access placed, and 2794 patients had catheter as the first access placed. Furthermore, we found that a substantial number of catheter placements are not reported by using USRDS data linked to Medicare claims. In other words, we found that the number of patients who likely had catheter as the first access did not have an appropriate CPT code in the Medicare data. To address this issue, we defined the catheter group as follows. First, we identified patients who initiated HD with catheter (n = 128,378). Assuming that some of them might have had AVF or AVG not used for the first dialysis treatment (due to stenosis or failure to mature), we excluded those with CPT code identifying AVF or AVG placement (predialysis), CPT code identifying treatment of AVF or AVG complications (predialysis or as a first CPT code after dialysis initiation), or CPT codes for catheter removal 30 days after the initiation of HD (indicating the possibility that immature AVF was possibly present prior to catheter placement). Patients who started dialysis with the catheter and had maturing AVF or AVG (based on the data from CMS Form 2728) were also excluded from this group. The final group of patients identified as starting HD with the catheter included 90,517 subjects. The final study cohort consisted of 115,425 patients (21,436, 3472, and 90,517 patients with AVF, AVG, and catheter as the first access placed, respectively) (Figure 1).
Potential predictors and outcome definition The primary variable of interest in this project was geographic location. Each patient in the cohort was assigned
688
a RUCA code (based on the ZIP code of residence), which classifies geographic areas using measures of population density, urbanization, and daily commuting. The RUCA system uses smaller geographic units and is more precise than the county-based systems. Thus, suburban areas with low population density but close employment associations with large urban centers are classified as urban. The RUCA system includes codes from 1.0 (most urban) to 10.6 (most rural). For this analysis, 3 independent RUCA groups were selected: metropolitan (RUCA 1.0–3.9), which includes cities with a population of more than 50,000 and the associated suburban areas; micropolitan (RUCA 4.0–6.0), for smaller cities or towns of 10,000– 49,999; and rural or small town (RUCA 7.0–10.6), for populations of less than 10,000. The outcome variable was the placement of AVF as first HD access prior to HD initiation compared with catheter (AVG was not included in the analysis). The model was adjusted for the variables that were considered to be potential confounders, specifically age, sex, race, primary cause of ESRD, employment status, and comorbidities. Some of the variables potentially mediating the effect of geographic location on access selection were also included in the model to isolate the effect of true disparities rather than potential objective barriers. We included duration of pre-ESRD nephrology care, body mass index (BMI), substance abuse, and income. The information regarding median annual income stratified by race and ZIP codes was obtained from the US Census Bureau. The comorbidities included in our model were those available in the USRDS database and had previously been identified as important predictors of mortality, general health status, or other outcomes in HD patients. Specifically, we included the history of diabetes, congestive heart failure, peripheral arterial disease, malignancy, and cerebrovascular disease.
Statistical analysis Means and standard deviations were used to summarize continuous variables with normal distribution. Categorical variables were summarized as percent of total. To compare between groups, we used analysis of variance or t test for continuous variables and chi-square for the categorical variables. Logistic regression was used to estimate the independent association of patient’s geographical location with the initial access being AVF (as opposed to central vein HD catheter). The model was adjusted for variables indicated earlier. Data were analyzed using the SAS software (SAS Institute, Cary, NC).
Hemodialysis International 2014; 18:686–694
Geographic disparities in HD vascular access
RESULTS
DISCUSSION
Baseline characteristics
In this project, we studied the association between geographic area of residence and the first HD access placement in patients of 67 years of age or older. A NHANES III study10 has suggested that 6.6 million people older than 60 years have chronic kidney disease (CKD). Also, the older fraction of dialysis population is quickly growing. The number of octogenarians and nonagenarians starting dialysis has nearly doubled, rising from 7054 persons in 1996 to 13,577 persons in 2003.11 The 1- and 2-year survival among patients older than 75 years who have stage 5 CKD has also increased to 84% and 76%, respectively. Addressing the need for optimizing care of these patients, The National Kidney Foundation developed and released clinical practice guidelines and has recommended that an AVF be the preferred first access over an AVG in patients initiating HD.1 This is because the AVF has better patency rates, less need for corrective procedures and complications, and subsequently lower morbidity and cost.7 Long-term use of tunneled catheters is discouraged due to higher infection rate, lower blood flows, less efficient HD, and an increased risk of sepsis, endocarditis, and metastatic infections.2–6 At the same time, there are certain barriers to the creation of an autologous AVF. It requires an adequate arterial and venous vasculature, which can be a challenge in patients who are older, diabetic, obese, and/or female,12 or in those who may have relatively smaller or compromised vascular anatomy. Vessel diameter has also been shown to be an important predictor of AVF survival.13 In other instances, patients who are new to dialysis may have presented too late for the establishment of an AVF before the need for dialysis. Previous published reports have demonstrated that African American population undergoes more AVG placement, suggesting the existence of racial differences in vascular anatomy, increased failure rate or subconscious bias leading to disparities in vascular access strategy. Furthermore, patients belonging to the low socioeconomic status may have limited access to health care,14 potentially resulting in later referral to a nephrologist.14 Other contributing factors may include limited patient education15,16 and failure of AVF to mature. In this project, in addition to the above-mentioned barriers and disparities to AVF placement, we hypothesized the presence of geographic disparities. We used patient-level data from the USRDS, a comprehensive registry of all patients with ESRD, with information such as date of onset of ESRD, demographic, employment status; as well as comorbid conditions at the initiation of dialysis.
Our study cohort included 111,953 patients identified in the USRDS linked with Medicare claims. Of those, 21,436 patients underwent AVF as the first access and the remaining 90,517 patients had catheter used for the initiation of HD. The study population characteristics are presented in Table 1. The age of onset of ESRD of patients in the rural, micropolitan, and metropolitan categories is 76.5, 76.7 and 77.1, respectively. Women represented 46.5%, 47.1%, and 46.9% in each group, whereas there was larger representation of blacks (21.6%) in the metropolitan areas compared with the micropolitan (14.6%) and rural (14.7%) regions (Table 1). Diabetes was the predominant risk factor associated with ESRD across all groups; 34.2% of patients in the rural areas received no pre-ESRD nephrology care compared to 32.1% in the micropolitan and 31.4% in the metropolitan areas.
Association between geographic area and first HD access We used logistic regression model to estimate the odds of AVF as first access compared to catheter in the metropolitan and rural population with the micropolitan population as a reference. The total number of patients in the metropolitan areas was 15,857 in the AVF group and 69,687 in the catheter group. Similarly, the total number of patient in the micropolitan population in the AVF group was 2752 compared to 9533 in the catheter group. In addition, 18.5% of metropolitan patients received AVF as first access compared to 22.4% of patients living in micropolitan areas (P < 0.001). In the metropolitan population, 81.5% patients received catheter compared to 77.6% patients in the micropolitan population and 78.4% patients in the rural areas. These difference in AVF as a first access placed between metropolitan and micropolitan areas remained significant after adjusting for other independent variables described earlier (Table 2). In particular, patient residing in metropolitan areas have lower odds of having AVF as a first dialysis access compared to those living in micropolitan areas (odds ratio [OR] 0.80; P < 0.001). Patients living in rural areas did not demonstrate significant difference with those living in micropolitan areas (reference group) in terms of AVF being the initial HD access (OR 0.96; P = 0.26).
Hemodialysis International 2014; 18:686–694
689
Goldfarb-Rumyantzev et al.
Table 1 Baseline characteristics of the study population at the time of ESRD onseta
First access type AVF fistula Central catheter Age at ESRD onset Age categories (y) 67–79 80–89 >90 Sex Male Female Race White Black Asian Native American Others Body mass index (kg/m2) BMI categories
