A randomized pilot study to evaluate graft-first versus fistula-first vascular access strategy in older patients with advanced kidney disease: results of a feasibility study Pilot and Feasibility Studies

Background. Although older adults encompass 40% of patients with advanced chronic kidney disease (CKD), it remains unclear which long-term hemodialysis (HD) vascular access type, arteriovenous (AV) fistula (AVF) or AV graft (AVG), is optimal with respect to effectiveness and patient satisfaction. Clinical outcomes based on the initial AV access type have not been evaluated in randomized controlled trials. This pilot study tested the feasibility of randomizing older adults with CKD to initial AVF versus AVG vascular access surgery. Methods: Patients 65 years or older with pre-dialysis CKD or incident end-stage kidney disease (ESKD) and no prior AV vascular access intervention were randomized in a 1:1 ratio to undergo surgical placement of either AVF-first or AVG-first after providing informed consent. Trial feasibility was evaluated as (i) recruitment of ≥70% of eligible participants, (ii) ≥50% to 70% of participants undergo placement of index AV access within 90 to 180 days of enrollment, respectively, (iii) ≥80% adherence to study-related assessments, and (iv) ≥70% of participants who underwent index AV access placement will have a follow-up duration of ≥12 months after index surgery date. Results: Between September 2018 and October 2019, 81% (44/54) of eligible participants consented and were enrolled in the study; 11 had pre-dialysis CKD and 33 ESKD. After randomization, 100% (21/21) assigned to AVF-first surgery and 78% (18/23) assigned to AVG-first surgery underwent index AV access placement within a median (1st, 3rd quartile) of 5.0 (1.0, 14.0) days and 13.0 (5.0, 44.3) days, respectively, after referral to vascular surgery. The completion rates for study-specific assessments ranged between 40.0 and 88.6%. At median follow-up of 215.0 days, 5 participants expired, 7 completed 12 months of follow-up, and 29 are actively being followed. Assessments of grip strength, functional independence and vascular access satisfaction were completed by >85% of patients who reached pre-specified post-operative assessment time point. Conclusions: Results from post-operative assessment the intervention groups was noted and that it could introduce confounding on intervention effects. An unexpected improvement in grip strength over time was also noted. We surmise that a selective dropout process may be at play, which could bias the observed results. For the main trial, the frequency of assessments will be changed to reduce participant fatigability and rates of missing data. Time-points of post-operative outcome assessments will be set at definite intervals from the date of surgical intervention. In the main trial, statistical analyses to assess longitudinal changes in grip strength and other quantitative outcomes will treat death as a competing confounding outcome. Rigorous assessment of the effects of versus in with on and care

post-operative assessment between the intervention groups was noted and recognized that it could introduce confounding on intervention effects. An unexpected improvement in grip strength over time was also noted. We surmise that a selective dropout process may be at play, which could bias the observed results. For the main trial, the frequency of assessments will be changed to reduce participant fatigability and rates of missing data.
Time-points of post-operative outcome assessments will be set at definite intervals from the date of surgical intervention. In the main trial, statistical analyses to assess longitudinal changes in grip strength and other quantitative outcomes will treat death as a competing confounding outcome.

Introduction
Surgical creation of an arteriovenous fistula (AVF) is considered the optimal long-term hemodialysis (HD) vascular access strategy in patients with pre-dialysis chronic kidney disease (CKD) or dialysis-dependent end-stage kidney disease (ESKD). Surgical placement of an AV graft (AVG) is reserved for those who do not have vasculature suitable for placement of an AVF. The standardized practice of AVF-first vascular access placement stemmed from observational studies in younger patients and was bolstered by subsequent practice guidelines. [1][2][3][4][5] Prior retrospective studies showed a graded relationship between the type of vascular access placed (or utilized) for HD with access complications and patient survival. The lowest access complication rates and longest patient survivals were seen with an AVF. The highest access complication rates and shortest patient survivals were seen with tunneled central venous catheters (TCVC); intermediate results were seen with an AVG. [1][2][3][4] The benefits of AVF over AVG and TCVC in these studies led national dialysis and vascular access committees to develop clinical practice guidelines. These guidelines set an orderly approach to vascular access placement according to the 'Fistula First Catheter Last' initiative. [5] Nationwide initiatives to increase the use of AVF for HD with a formulaic approach led to rapid declines in AV graft placement. Therefore, many patients are forced to use a TCVC at dialysis initiation, as the success of AVF development is dependent on patient-specific factors. [6,7] Flaws exist in the current AVF-first practice strategy. 'Fistula First' practice guidelines are based on research performed in the 1990's, an era when older patients comprised less than 10% of the ESKD population. Remarkable changes in demographics (rising mean ages) and morbidities (rising prevalence of diabetes, cardiovascular disease and malignancy) have since occurred in incident and prevalent dialysis populations. Based on national registry data encompassing the dialysis population in the early period of 2010's, adults ≥65 years comprise more than 40% of incident patients with ESKD on HD. [8] Compared with their younger counterparts, older patients on dialysis have greater numbers of concurrent illnesses, polypharmacy, undertreated conditions, hearing impairment, and high prevalence of physical and mental disabilities. [9][10][11] The prognosis for older adults initiating HD is poor, 25-60% mortality rates are seen in the first year (compared with 5-15% in their younger counterparts). [12][13][14] Recent studies showed that frailty and morbidity index directly impact patient survival on dialysis, independent of the type of vascular access. [9,[15][16][17] Another limitation in existing vascular access guidelines relates to translating conclusions from observational data into clinical practice. A challenge in all retrospective studies is distinguishing whether the intervention (here, vascular access type) directly impacted clinical outcomes or merely accentuated the effects of other comorbidities leading to observed outcomes. Patients who undergo AVF placement and achieve a functional access may be healthier, yet this may not be reflected in their list of medical problems (e.g., perceived better prognosis, less severe comorbid conditions). [18,19] Residual confounders not only impact achievement of a usable AVF, but also affect patient survival.
In a large cohort from the US Renal Data System comprised of incident patients ≥67 years old, those who initiated HD with a TCVC and had a history of primary AVF failure (i.e., underwent AVF placement surgery and the AVF did not mature) experienced 34% lower mortality rates than those who initiated HD with a TCVC but did not undergo AVF placement surgery, despite both groups of older incident dialysis patients receiving HD via TCVC. [18] These results support the contention that patient factors affecting the outcome of fistula placement may also impact mortality on HD. The upcoming revised Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines (In Press) recognize these limitations and no longer state that all patients should have an AVF as a preferred vascular access. [20] Therefore, prospective unbiased data are urgently needed to compare patient outcomes between AVF and AVG in older patients initiating HD.
Apart from age-related differences in clinical outcomes, age is a modulator of vascular access outcome. Several studies highlighted the challenging task of achieving a functioning AVF in older patients. [21] In a cohort of 168 patients who underwent AVF placement, the 12-month primary assisted patency was 35% in older patients (≥70 years) group and 67% in younger patients (<70 years) group (P=0.002); secondary patency was 36% and 67%, respectively (P=0.004). [22] A meta-analysis of 13 studies concluded that older patients have 50-65% higher odds of primary AVF failure and 80% higher odds of secondary AVF failure compared with younger patients. [23] These results led to the emerging concept that AVG may represent a better "catheter-sparing" strategy than AVF, because AVG often permit faster transition from TCVC to AV access. This must be balanced against reports from prior retrospective analyses of shorter AV access lifespan of an AVG (median cumulative patency 7.0 months) vs. an AVF (median cumulative patency 15,0 months), [19] although one study reported similar AVF and AVG access survival in patients 75 years of age and older. [24] Given the paucity of clinical trials data in the field of AV vascular access placement, we conducted a randomized pilot study to investigate the feasibility of implementing randomization to AVF-first vs. AVG-first in older patients with advanced kidney disease.
This manuscript reports data that directly addressed the feasibility and acceptability of trial procedures. Effectiveness is not reported, because the pilot study was not powered to detect outcome differences between study arms. Feasibility and pilot studies are designed to evaluate trial feasibility, acceptability and safety, rather than test the effectiveness of planned main trial interventions. [25,26] Study objectives

1.
The primary objective of this pilot trial was to examine the feasibility of random assignment to AVF-first or AVG-first vascular access placement in older patients with advanced CKD with no prior AV access intervention. Participants were referred for surgical intervention for AV access creation by their nephrologist and had vascular anatomy suitable for placement of either an AVF or AVG. Study procedures and estimates of the rates of study eligibility, consent, randomization, and retention were evaluated.

2.
The secondary objective of this pilot trial was to analyze the feasibility of data collection related to assessment tools of physical function and activities of daily living, by observing completion rates, and missing data.

Methods
The design and rationale of the trial has been published previously. [27] The eligibility criteria and schedule of assessments were adjusted during the pilot phase of the trial. A brief description is presented below.

Setting and participants
Following research ethics approval by an Institutional Review Board (IRB), this pilot study was conducted at 16 outpatient dialysis units, one inpatient dialysis unit, and three nephrology practices affiliated with an academic tertiary hospital and a regional hospital in North Carolina, USA. The nephrology practices in this region serve an annual population of nearly 500 prevalent patients with ESKD on HD. All adult patients with pre-dialysis CKD or incident ESKD who were referred for AV access placement by their treating nephrologist were assessed for eligibility.

Study design
Eligible patients were randomized in a 1:1 ratio either to AVG-first placement (intervention group) or to AVF-first placement (control group). The randomization sequence was computer-generated, and allocation was managed by the Research Coordinator of the Extremities were also excluded if a pacemaker defibrillator with central venous leads was present on that side.

Recruitment
Potential participants were first screened at the time of initiation of HD or at referral for AV access placement, whichever occurred first. The decision for referral to vascular access surgery was made by treating nephrologists, independent from this study. Review of electronic clinical records at outpatient dialysis facilities, nephrology outpatient clinics, and nephrology inpatient service determined eligibility for the study. Eligibility for anesthesia and surgical placement of an AV access was determined as part of standard of care in each patient. Surgical suitability for placement of AV access was determined by the vascular surgeon using the results of the ultrasound vascular mapping of both upper extremities. Final eligibility criteria encompassed medical and surgical eligibility for placement of either type of AV access. Once all inclusion and exclusion criteria were met, eligible participants were approached and the study rationale and objectives were explained in simple terms. Written informed consent was obtained in person.

Randomization and concealment
Consenting participants underwent baseline study-specific assessments followed by randomization to either AVG-first or AVF-first access placement. Block randomization, with variable block size of 2, 4, and 8, was used with no stratification. Following randomization, each participant's assignment was conveyed to the vascular surgery team on the day of randomization and on the day prior to surgery. Blinding of the research team or the participant to the type of AV access placed was not possible. However, vascular access outcomes (i.e., primary AV access failure, access-related infections) were adjudicated by patients' medical and/or surgical team, independent of the study team and without consideration of participant assignment.

Trial intervention
Based on randomization, participants were scheduled to undergo surgical placement of AVG or AVF. Surgical creation of the AV access was performed by a vascular surgeon with experience in dialysis vascular access placement. All AVG were made of polytetrafluoroethylene. When suitable vasculature was present, preference was given to distal (forearm) over proximal (arm) AV access placement in both study groups to preserve limited future vascular access sites.

Data collection
Patient demographics, medical history, vascular anatomy and surgery details, and postoperative complications were extracted from the electronic medical record. Data were de-identified and collected by a trained study coordinator. When applicable, clinic notes, operative reports, procedure notes and discharge summaries were reviewed to ensure capture of all events. Study specific assessments to evaluate participant's upper extremity muscle strength, physical activity, level of independence, satisfaction with AV access, and health-related quality of life were performed as described. [27] Trial feasibility -Participant recruitment and retention rates Screen failure logs and enrollment logs were maintained for all patients screened and approached for participation. Reasons for eligibility failure were recorded and entered into an excel spreadsheet. Recruitment success was defined as ≥70% of eligible participants agreeing to be enrolled. Successful retention was considered as ≥70% of participants who underwent AV access placement will have a follow-up duration of ≥12 months after index access placement.

Trial feasibility -Barriers to the implementation of the intervention
Reasons for not undergoing AV access surgical intervention were recorded. Intervention success was defined as ≥50% to 70% of participants undergoing placement of index AV access within 90 to 180 days of enrollment, respectively.

Trial feasibility -Collection of outcome data
Inpatient (hospitalization records) and outpatient (dialysis units, office visits) electronic medical records were reviewed on a monthly basis to collect events of vascular access outcome (e.g., primary or secondary AV failure, date of first cannulation, date of successful cannulation, angioplasty, AV access infection, TCVC-related infection etc) and patient outcomes (e.g., hospitalization, surgical intervention, etc). All medical and/or surgical diagnoses and vascular access-related diagnoses documented in this study were made by the participant's treating physicians. Outcome measures based on intervieweradministered questionnaires and grip strength and gait speed measurements were obtained at four time points with in-person visits done by the study coordinator preoperatively on the day of surgery (baseline assessment) and pre-dialysis at the outpatient dialysis unit (follow-up assessments). The baseline assessment occurred at study enrollment, with further assessments performed 2 weeks after index AV access placement (follow-up 1), 6 weeks after first index AV access cannulation (follow-up 2), and 6 months after successful use of index AV access or 6 months after index AV access placement if primary AV access failure occurred (follow-up 3). Feasibility of data collection was defined as ≥80% participant adherence to study-specific assessments (grip strength, level of independence, satisfaction with vascular access, and health-related quality of life), absent of a condition that precluded assessment.

Trial feasibility -Adverse events
Participants who suffered an adverse event were immediately reviewed by the study team.
The adverse events were collected prospectively through monthly review of inpatient and outpatient electronic medical records. The event was recorded in detail and reported to the IRB, when appropriate. An independent Medical Safety Officer, unblinded to treatment assignments, reviewed participant clinical outcomes and vascular access-related events on a quarterly basis; sooner, if indicated.

Primary outcome assessment data
Grip strength was measured using a hand-held dynamometer on a non-dialysis day or predialysis during a dialysis day. A cut-off point of <16 kg in women and<26 kg in men was used to define muscle weakness. [28] Secondary outcome assessment data

Ethical considerations
Confidentiality of patient data was maintained throughout the study and case report forms were kept in locked cabinets. The electronic data extracted from the monitoring systems was anonymized, as was all data in the final reports.

Eligibility rate and reasons for exclusion
The eligibility rate was calculated by dividing the number of patients referred to vascular surgery for AV access placement by the number who met the inclusion criteria after ultrasound vascular mapping of both upper extremities. This equated to an eligibility rate of 70.1%. Out of the 77 older adults with advanced CKD who were referred to vascular surgery and underwent vein mapping, 14 (18.2%) did not have vasculature suitable for placement of AV access, and 9 (11.7%) were felt unlikely to require initiation of HD within 90 days.

Consent rate and reasons for not participating in the study
The overall consent rate was 81.5%. Of the 54 patients eligible for this study, 44 agreed and consented to participation. Nine (16.7%) declined to participate with reasons for refusal being apprehension to undergo placement of an AV access other than an AVF (n=7) or general wariness of study participation (n=2). The average enrollment rate was 3 patients per month (Figure 2). Baseline characteristics of the participants are listed in

Randomization procedures
Twenty-one participants were randomized to AVF placement and 23 to AVG placement. The distribution of demographic characteristics and coexisting comorbidities are shown in Table 1.

Intervention adherence
Of the 44 participants randomized and scheduled to receive index AV access placement, 34 underwent surgery as scheduled and 10 missed their first appointed surgical date (4 in AVF group and 6 in AVG group). Of those who missed initial appointments, 9 had their

Retention rate
Three (6.8%) patients, all in the AVG-first group, were withdrawn from the study (1 became medically unstable to undergo surgery and 2 had protocol deviation and received AVF placement) (Figure 1). At the time of data lock, 7 participants had completed 12 months of follow-up, 5 participants had expired, and 29 were actively followed. Among the 5 participants who expired, 2 were in the AVF-first group and expired after index AV surgery; and 3 were in the AVG-first group, of whom 1 expired before and 2 expired after index AV access surgery (Figure 1). The observed median time to death was 168.0 days (127.0, 210.2) from the date of enrollment. The feasibility benchmark for retention rate required that ≥70% of participants who underwent index AV access placement will have a follow-up duration of ≥12 months after index access placement. Based on these preliminary results, we observed an overall drop-out rate of 6.8% and mortality rate of 11.4% within a median follow-up of 215.0 days (111.3, 282.0). Overall, these results suggest that we are on target of meeting the feasibility aim for participant retention rate.

Adverse events
Five deaths were observed, 2 in the AVF group and 3 in the AVG group (Table 3). One patient randomized to AVG placement expired before surgical intervention (Patient D in Table 3). One patient with CKD who underwent AVF placement was not started on HD by the date of expiration (Patient B). Of the 4 participants who underwent index AV access placement and expired, none had AV access-related infection and one had a TCVC-related infection.

Lessons learned
This pilot trial was very informative. The investigative team learned a number of things that will help in conducting a full trial. The lessons gleaned from this pilot study, along with a summary of 14 methodological items that will be evaluated in a feasibility study are listed in Table 4. [35] Discussion Surgical referral for placement of an AV vascular access and type of vascular access in patients with advanced CKD remain clinical and scientific conundrums due to lack of highquality data guiding these practices. This is particularly true in older populations, where available data suggests different outcomes compared to younger patients. However, current practice does not consider age-related personalized care in HD vascular access type.
Health care interventions require rigorous evaluation and this is best achieved with randomized controlled trials. Feasibility studies are an important initial step in the phased approach to identifying problems that might occur in planned main clinical trials for complex interventions in complex populations. In this case, the trial would consider placement of dialysis AV vascular access in older adults with advanced CKD and multiple comorbidities. We therefore performed an important preparatory stage analysis by conducting a feasibility study to ensure methodological approaches applied in a future multisite clinical trial are streamlined. Herein, we report the feasibility results of a pilot study for a first of its kind randomized clinical trial investigating placement of AVF versus AVG in adults 65 years of age and older with pre-dialysis CKD or ESKD receiving HD via a TCVC.
This pilot study achieved a 100% screening rate. At initiation, one large academic group and 14 dialysis centers encompassed the recruitment sites. Six months into the study, participating sites were expanded to include 2 additional nephrology practices, 2 additional dialysis units, and one regional hospital. Logistical adaptations were promptly implemented such that study personnel were able to screen all older adults who initiated chronic HD and/or were referred for vascular access placement at all participating sites.
We observed two events of protocol deviation wherein participants underwent placement of AVF instead of randomization-based placement of AVG. To prevent the occurrence of similar protocol deviations in the full-scale trial, AV access assignment will be reminded to the vascular surgery team and conveyed to the operating room nursing personnel the day on the day of surgical intervention.
Eligibility rates inform whether those recruited are likely to be representative of the target population, with higher eligibility rates suggesting generalizability of the study. In this pilot trial, the eligibility rate was 70.1%. A trial parameter that indicates potential loss in precision, which can introduce bias and reduce statistical power, thereby affecting the generalizability, validity and reliability of results is loss to follow-up. It has been estimated that a ≥ 20% loss can threaten trial validity. [36] The retention rate in this pilot study was good, with 6.8% drop-out rate. However, completion rates of questionnaires and gait speed were suboptimal. Throughout the study, our research team discussed their observations about how participants acted during the assessment. Easy fatigability and refusal to perform certain tests were noted. These issues are likely due to the chronically ill and frail nature of participants. To remedy this issue, the order of assessments were reorganized by reducing the frequency and number of questionnaires.
Besides changing the frequency of post-operative assessments for the main trial, the time-points of assessment will be set at definite periods from the date of surgical intervention. In the design of the pilot trial, the time-points of second and third assessments (follow-up 2 and follow-up 3) depended on other outcomes (i.e., first AV access cannulation and successful use of index AV access). As a result of this pilot trial, we appreciate that differences in time to AV access cannulation between the intervention groups could introduce systematic confounding of outcome assessment. Therefore, assessment visits will be set at fixed intervals after the date of surgical intervention during the main trial to eliminate potential confounding between timing of measurement and treatment effect.. This pilot trial also notes a rather unexpected improvement in grip strength over time, with higher average grip strength recorded at follow-ups 2 and 3 compared to baseline grip strength. Survival bias due to healthier participants (i.e., fewer comorbidities, better baseline physical status and grip strength) being more likely to reach later assessment time points could skew the interpretation of grip strength outcomes. In the main trial, analyses will be run under the intention-to-treat principle, and the effect of missing data and informative censoring will be assessed using appropriate imputation techniques and sensitivity analyses. Death as a competing risk will also be assessed in addition to the proposed death-censored outcomes.
While the recruitment rate was acceptable considering the frail and ill population involved, we recognize it will take more than 4 years to reach the required sample size given rates of recruitment from the two participating sites. Originally, we planned to recruit patients ≥ 70 years of age with incident ESKD. The eligibility criteria were adjusted to include adults ≥ 65 years of age, as well as those with pre-dialysis CKD referred for AV access placement. We subsequently recognized that many vascular access-related outcomes (e.g., primary AV access failure, time to successful AV access cannulation) would not be assessed in a timely manner for patients who were not on HD at the time of AV access placement. Therefore, we plan to limit enrollment in the main trial to those with and AVF abandonment (HR 1.28, 95% CI 1.10-1.50) than males. [38] In the main clinical trial, randomization will be stratified by sex to ensure balanced representation in both arms. Exploratory analyses will include testing for interaction effects between AV access configuration (forearm vs. arm) and treatment assignment (AVF vs. AVG) to determine whether the treatment effect on the primary outcome is modified by AV access anatomical location.

Limitations
Our trial has many strengths and several limitations. The pilot was conducted in only two hospital centers (a large academic center and a regional hospital) across three nephrology practices and 16 dialysis units. Additional logistical limitations are expected to be encountered with a large multisite randomized clinical trial. Going forward, potential research sites will be evaluated to ensure resources will be available to timely address any logistical issues that may arise. Another limitation was the slow rate of enrollment, complicated by reductions in the number of potential participants due to high mortality in this chronically ill population. This observation will help us select the number of sites necessary to participate in a large-scale well-powered trial.
Implementation of the randomization process in the pilot study was not free from potential bias. Allocation to the two arms of surgical interventional was not concealed from the study coordinator at the time of approaching potential participants. However, patients were not informed of their randomization arm until informed consent was obtained. To reduce potential bias in the implementation of randomization process in a future main trial, randomization will be implemented using a web-based system that performs randomization once the informed consent was obtained. The random sequence will be coded, implemented and executed by a web-based system to provide a random assignment that cannot be changed. Blinding to the type of index AV access placed during the study was not feasible due to the visible nature of the intervention. This limitation will remain in the main clinical trial. However, clinical outcomes (e.g., primary AV access failure, access-related infections, and hospitalizations) are adjudicated by treating physicians and are unlikely to be influenced by knowledge of study participation. In addition, the main clinical trial will encompass oversight committees (i.e., Data Safety and Monitoring Board, Endpoint Adjudication, and Steering Committees) to address efficacy and/or safety endpoints in an unbiased manner.

Conclusions
The results from this pilot study provide the feasibility of conducting a randomized clinical trial to test the impact of two type of AV access placement on clinical outcomes, physical function and quality of life in older patients with ESKD on HD.

Ethics approval and consent to participate
The study was approved by the Wake Forest School of Medicine Institutional Review Board (IRB00050577). The study is registered at Clinicaltrials.gov (NCT03545113).

Competing interests
The authors declare that they have no competing interests.

Consent for publication
All the authors of this manuscript reviewed the manuscript and approved its content and submission for peer review and publication. Baseline data were collected at the time of patient enrollment. Data are presented as number of participants (percentage) o quartile). *Renal function parameters are reported in patients with CKD at the time of enrollment. Race or ethnic group was body-mass index is the weight in kilograms divided by the square of the height in meters. Estimated glomerular filtration rat with the CKD-EPI creatinine equation. Abbreviations: AVF, arteriovenous fistula; AVG, arteriovenous graft; CKD, chronic kidney disease; ESKD, end-stage kidney dis estimated glomerular filtration rate (calculated with CKD-EPI equation).    Cumulative recruitment rate.

Figure 3
Distribution of patients who underwent study-specific assessments

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download. CONSORT-2010-Checklist-MS-Word.doc