Impact of Surgeon Selection on Access Placement and Survival Following Preoperative Mapping in the ‘‘Fistula First’’ Era Kenneth L. Choi, Loay Salman, Gururaj Krishnamurthy, Carlos Mercado, Donna Merrill, Ian Thomas, Shukrat Artikov, Gabriel Contreras, Rao Ali Hashim Khan, Ali Warda, and Arif Asif Section of Interventional Nephrology, Division of Nephrology, University of Miami Miller School of Medicine, Miami, Florida
ABSTRACT According to the ‘‘Fistula First Initiative’’ surgeon selection should be based on best outcomes, willingness, and ability to provide access services. This analysis presents arteriovenous access placement and outcomes in 75 patients when surgery was performed by one of two dedicated high-volume vascular access surgeons (community [surgeon I] and academic medical center [surgeon II]). Preoperative vascular mapping was performed in all the patients. Demographic characteristics were similar except that patients referred to surgeon I (n = 40) were older (52.7 16.2 years vs. 45.4 13.7 years; p = 0.04) and tended to have more previously failed accesses (50% vs. 29%; p = 0.06) and black race (65% vs. 43%; p = 0.055) including a history of previously failed accesses (50% for surgeon I and 29% for surgeon II; p = 0.06). Similarly, there was no significant difference in the size of forearm ([surgeon I:
2.0 1.0 mm], [surgeon II: 1.9 0.8 mm]; p = 0.45) or upper arm veins (cephalic vein: surgeon I = 3.2 1.4 mm, surgeon II = 2.9 1.2 mm, p = 0.34; basilic vein: surgeon I = 5.0 1.2 mm, surgeon II = 4.7 1.3 mm, p = 0.25). Fistulae placement occurred in 98% vs. 71% (p = 0.001) for surgeon I and II, respectively. Characteristics predictive of fistula placement over an arteriovenous graft were surgeon selection (odds ratio [OR] = 19.52; p = 0.01) and no history of diabetes (OR = 7.61; p = 0.016). Kaplan–Meier analysis revealed 6 and 12 months overall access survival rates of 82%, 58% and 82% and 47% for surgeon I and II, respectively (p = 0.007). This analysis demonstrates that surgeon selection can have a significant impact on the rate of fistula placement and its overall survival despite similar findings on preoperative vascular mapping.
The National Kidney Foundation-Dialysis Outcomes Quality Initiative (NKF-DOQI) guidelines state that arteriovenous fistulae (AVF) are preferred over arteriovenous grafts (AVG) due to longer patency, decreased thrombosis, infection, and mortality (1–3). The Fistula First National Vascular Access Program Improvement Initiative states that surgeon selection should be based on best outcomes, willingness, and ability to provide access services (http://www.fistulafirst.org/). Recently, a Veterans Association study suggested that there were surgeon-specific characteristics which were associated with increased AVF placement such as center volume (>30 procedures per year) and specific surgeon clustering (4,5). A program in which the surgery was performed by a nephrologist demonstrated 100% initial primary AVF placement (4). Recent data have emphasized that
preoperative mapping results in increased AVF placement rates (6–9). However, there have been no studies evaluating the effect of surgeon selection on the creation and outcomes of an AVF after controlling for preoperative vascular mapping. The objective of this report was to determine whether surgeon selection was a predictor of AVF creation in patients who underwent preoperative vascular mapping and whether surgeon selection affected overall access survival. Methods A retrospective study was performed of all 148 patients who underwent vascular access education and upper extremity vascular mappings from August 2003 through December 2006 at the University of Miami Interventional Nephrology Laboratory. Vascular access education was provided by the interventional nephrology team. Types of vascular accesses and their complications were highlighted. In addition, morbidity and mortality associated with tunneled hemodialysis catheter, AVG and fistulae were presented. Patients were encouraged to request a fistula creation at the encounter
Address correspondence to: Arif Asif, MD, Section of Interventional Nephrology, Division of Nephrology, University of Miami Miller School of Medicine, 1600 NW 10th Ave #7168, Miami, FL 33136, or e-mail:
[email protected]. Seminars in Dialysis—Vol 21, No 4 (July–August) 2008 pp. 341–345 DOI: 10.1111/j.1525-139X.2008.00446.x ª 2008 Copyright the Authors. Journal compilation ª 2008 Wiley Periodicals, Inc. 341
342
Choi et al.
with the surgeon. In terms of vascular mapping, venous evaluation was performed by using venography and arterial examination was performed by palpating arterial pulses, recording the difference of blood pressure between the two arms and performing Allen’s test to demonstrate the patency of the Palmar arch. Venous mapping was performed by cannulating a vein in the dorsum of the hand and injecting low-osmolar contrast diluted with saline as previously described (10). Angiographic images were obtained using a fluoroscopy machine (GE 9800; GE Medical Systems, Salt Lake City, UT) with measurements calibrated with a radiopaque ruler. Measurements of forearm and upper arm veins were recorded. Patients were referred to two different dedicated highvolume vascular access surgeons (>30 vascular access surgeries per year) based on insurance funding. Surgeon I offered services in a private practice setting at a community hospital while surgeon II provided access care at an academic university medical center. Findings of vascular mapping and recommendations for AV access location were sent to the access surgeon prior to surgery. Exclusion criteria for analysis included use of an outside surgeon other than surgeon I or surgeon II, patient refusal of permanent arteriovenous access, death within a year of vascular mapping before access was created, lower extremity arteriovenous access surgery, patients awaiting access surgery, or lost to follow-up. Following surgery, management of the patients’ accesses was conducted by the interventional nephrology team including interventions for early fistula failure, abnormal physical examination, prolonged bleeding, low blood flow by ultrasound dilution technique and clotted access. Outcomes Access survival or cumulative patency of AV accesses was defined as the time of access placement to the time of access abandonment for a new arteriovenous access or dialysis catheter placement, regardless of the number of primary interventions or thrombectomy procedures. Primary access failure was defined as an access that never achieved adequacy for dialysis (7). AV accesses never used for dialysis were defined to have an access survival of 0 days. Primary unassisted patency for AV accesses was defined as the time from access placement until the time of first access intervention (angioplasty, declot, or surgical revision) to maintain access patency (7). Patients with primary access failure were excluded from the primary unassisted patency analysis. Statistical Analysis Comparisons of continuous and categorical baseline characteristics were made by nonpaired t-tests and chisquared tests, respectively. In a cross-sectional design, multiple variables logistic regression analysis was performed to assess the factors associated with AVF placement over AVG. The following variables were evaluated as possible predictors—age, race, sex, diabetes, prior AV access surgery, surgeon, forearm vein size, upper arm
cephalic and upper arm basilic vein size. Independently significant predictors were identified by adding covariates in a forward stepwise fashion, selecting the predictor that most improved model performance at each step and stopping when no remaining predictors significantly improved model performance (p < 0.05 for change in model v2). Survival statistics was used to estimate primary unassisted patency and access survival. The cumulative survival curves were derived by the Kaplan–Meier method and the differences between survival curves were compared by the log-rank test. In both, primary unassisted patency and access survival analyses, patients were censored because of death, transplantation, loss of follow-up, and at the end of the study period (May 31, 2007). In the primary unassisted patency analysis, patients with primary access failure were also censored. All tests of significance were two-sided and differences were considered significant when p £ 0.05. Data are reported as percentages, mean standard deviation, odds ratio and 95% confidence interval (CI) as appropriate. All statistical analyses were performed using excel (Microsoft), Minitab 12 (State College, PA) and the NCSS 2000 software package (Kaysville, UT). Local institutional review board approval was obtained for this study. All study procedures were carried out in accordance with the Declaration of Helsinki regarding research involving human subjects. Results A total of 148 patients received vascular access education and underwent vascular mapping from August 2003 through December 2006. Seventy-three patients were excluded from the analysis for the reasons listed above (Fig. 1). Eventually, 75 patients had arteriovenous access surgery by either surgeon I (n = 40) or surgeon II (n = 35), who constituted the population for this study. Of the patients seen by surgeon I, 98% had AVF placed, compared with 71% by surgeon II (p = 0.001). To determine whether there were differences in patients referred to the two surgeons, demographic characteristics were compared (Table 1). There were no differences in baseline characteristics except that patients referred to surgeon I tended to be older (52.7 16.2 years vs. 45.4 13.7 years, p = 0.04) and had a higher rate of previous history of failed accesses (50% vs. 29%, p = 0.06) compared with those referred to surgeon II. The percentage of patients with forearm AV access placed tended to be higher for surgeon I than surgeon II (38% vs. 20%, p = 0.10). Of note, there was no statistically significant difference in vein size between patients seen by the two surgeons (forearm vein: surgeon I = 2.0 1.0 mm vs. surgeon II = 1.9 0.8 mm, p = 0.45; upper arm cephalic vein: surgeon I = 3.2 1.4 mm vs. surgeon II = 2.9 1.2 mm, p = 0.34; basilic vein: surgeon I = 5.0 1.2 mm vs. surgeon II = 4.7 1.3 mm, p = 0.25). Additionally, there was no difference in arterial pulses and Allen’s test between the patients referred to the two surgeons. Moreover, the blood pressure difference between the two arms did not exceed 20 mmHg in the two groups.
343
SURGEON SELECTION AND FISTULA SURVIVAL
Fig. 1. Clinical outcomes of all patients who underwent upper extremity vascular mapping from August 2003 through December 2006.
TABLE 1. Baseline characteristics of patients by surgeon Characteristics
Surgeon I
Surgeon II
p value
35 31 40 29 43 45.4 71 1.9 2.9 4.7 20
0.8 0.58 0.99 0.06 0.055 0.04 0.001 0.45 0.34 0.25 0.10
TABLE 3. Forward stepwise logistic regression for prediction of AVF placement over AVG Characteristic
Number Sex: female (%) Diabetes (%) Prior AV access (%) Race: black (%) Age (years) Fistula (%) Forearm vein (mm) Cephalic vein (mm) Basilic vein (mm) Forearm AV access (%)
40 38 40 50 65 52.7 98 2.0 3.2 5.0 38
16.2 1.0 1.4 1.2
13.7 0.8 1.2 1.3
AV, arteriovenous.
Of the 75 patients, 64 patients had AVF placed and 11 patients had AVG placed. Surgeon I placed one while 10 grafts were inserted by surgeon II. There was no significant difference in baseline characteristics between the patients who had an AVF or AVG placed, except the surgeon selection, presence of diabetes, and vein size (Table 2). Patients with AVG placed were more likely to be diabetic (73% vs. 34%, p = 0.016) and have smaller angiographic vein size (forearm vein: 1.5 0.7 mm vs.
Surgeon I No history of diabetes
Odds ratio
95% CI
p value
19.52 7.61
2.03–187.6 1.45–40.0
0.01 0.016
AVF, arteriovenous fistula; AVG, arteriovenous graft; CI, confidence interval.
2.1 0.9, p = 0.045; cephalic vein: 2.0 1.2 vs. 3.1 1.4, p = 0.035; basilic vein: 4.0 1.2 vs. 5.0 1.2, p = 0.011) compared to patients with AVF. Using stepwise logistic regression analysis, both surgeon I (OR = 19.52, p = 0.010) and no history of diabetes (OR = 7.61, p = 0.016) were independent predictors of AVF placement over AVG placement (Table 3). Access survival rates were 85% and 58% and 85% and 47% for surgeon I and II, respectively, at 6 and 12 months (p = 0.007) (Fig. 2). There was a 5% AV access primary failure rate for surgeon I and 31% for
TABLE 2. Baseline characteristics of AVG and AVF patients Characteristics Number Sex: % female Diabetes (%) Prior AV access (%) Race: % black Age (years) Surgeon (%) Surgeon 1 Surgeon 2 Forearm vein (mm) Cephalic vein (mm) Basilic vein (mm)
Graft
Fistula
11 45 73 36 64 45.7 13.8
64 33 34 41 53 50.0 13.8
2 29 1.5 0.7 2.0 1.2 4.0 1.2
98 71 2.1 0.9 3.1 1.4 5.0 1.2
AVG, arteriovenous graft; AVF, arteriovenous fistula.
p value 0.42 0.016 0.79 0.52 0.40 0.001 0.045 0.035 0.011
Fig. 2. Overall access survival rate by surgeon from the time of access surgery to the time of access abandonment.
344
Choi et al.
surgeon II (p = 0.002). There was no difference in primary unassisted patency rates in AV accesses that were successfully used. There was no statistically significant difference in primary patency rates (p = 0.26 by log rank test) between surgeon I and surgeon II with primary patency rates of 63% and 78% (p = 0.16) at 6 months and 29% vs. 44% (p = 0.09) at 12 months, respectively. Discussion Surgeon specialty, as well as surgical center volume, is associated with increased AVF placement over AVG (4,11). To control for surgeon specialty and surgical center volume, all patients in the present study were seen by one of two surgeons, both of whom were dedicated vascular access surgeons averaging greater than 30 access procedures per year. Preoperative mapping has also been shown to increase AVF placement (6–9). To control for preoperative mapping, all patients in the current report had upper extremity vascular mapping. To our knowledge, this is the first study demonstrating that surgeon selection has a significant impact on the placement of a fistula over an AVG after controlling for preoperative mapping, surgeon’s specialty and center volume. Although not statistically significant, surgeon selection was also associated with increased creation of forearm fistulae (38% vs. 20%, p = 0.10). A comprehensive program for vascular access identifying surgeon outcomes and adjusting referral patterns based on these outcomes is necessary to increase AVF placement rates and achieve AVF prevalence rates as recommended by the Fistula First Initiative. To evaluate potential confounding factors which might explain the difference in access placement and survival between the two surgeons, differences in baseline characteristics were examined. Patients seen by surgeon I were older compared with those seen by surgeon II (52.7 16.2 vs. 45.4 13.7, p = 0.04). Older age has been associated with a decreased placement rate of AVF compared to an AVG in multiple prior studies with an OR of 0.92 (p = 0.21) per decade increase in Veterans Affairs (VA) patients and an OR of 0.82 (p =