Abstract

Background. Access blood flow (Qa) measurement is the recommended method for fistula (AVF) surveillance for stenosis, but whether it may be beneficial and cost-effective is controversial.

Methods. We conducted a 5-year controlled cohort study to evaluate whether adding Qa surveillance to unsystematic clinical monitoring (combined with elective stenosis repair) reduces thrombosis and access loss rates, and costs in mature AVFs. We prospectively collected data in 159 haemodialysis patients with mature AVFs, 97 followed by unsystematic clinical monitoring ( Control ) and 62 by adding Qa surveillance to monitoring ( Flow ). Indications for imaging and stenosis repair were clinically evident access dysfunction in both groups and a Qa < 750 ml/min or dropping by >20% in Flow .

Results. Adding Qa surveillance prompted an increase in access imaging (HR 2.96, 95% CI 1.79–4.91, P < 0.001), stenosis detection (HR 2.55, 95% CI 1.48–4.42, P = 0.001) and elective repair (HR 2.26, 95% CI 1.16–4.43, P = 0.017), and a reduction in thromboses (HR 0.27, 95% CI 0.09–0.79, P = 0.017), central venous catheter placements (HR 0.14, 95% CI 0.03–0.42, P = 0.010) and access losses (HR 0.35, 95% CI 0.11–1.09, P = 0.071). In the Kaplan–Meier analysis, adding Qa surveillance only extended short-term cumulative patency ( P = 0.037 in the Breslow test). Mean access-related costs were 1213 Euro/AVF-year in Control and 743 in Flow ( P < 0.001).

Conclusions. Our controlled cohort study shows that adding Qa surveillance to monitoring in mature AVFs is associated with a better detection and elective treatment of stenosis, and lower thrombosis rates and access-related costs, although the cumulative access patency was only extended in the first 3 years after fistula maturation. We are aware of the limitations of our study (non-randomization and the possible centre effect) and that further, better-designed trials are needed to arrive at a definitive answer concerning the role of Qa surveillance for fistulae.

Introduction

Measuring access blood flow (Qa) is the most widely recommended method for detecting haemodynamically significant stenoses [ 1,2 ], but whether combining Qa surveillance with elective stenosis correction cost-effectively succeeds in reducing thrombosis rates and extending access survival in AVFs remains to be seen [ 1 ].

Most studies on the role of Qa surveillance are considered methodologically inadequate because they are observational with historical control groups. Half of them reported a lower [ 3–5 ] and the other half a comparable thrombosis rate [ 6–8 ] when using Qa surveillance by comparison with the historical control period (when surveillance was based on clinical monitoring [ 4,5 , 7,8 ], dynamic arterial [ 5 , 8 ] or venous dialysis pressure [ 3 , 5–8 ], access recirculation [ 7,8 ] or duplex ultrasound (DU) [ 7 ]). Only two of them evaluated access patency [ 5 , 8 ] and neither found any difference in AVF survival between Qa surveillance and control groups.

To our knowledge, only three controlled studies (two randomized) are available on the influence of Qa surveillance on outcome in AVFs [ 9–11 ]: two found a significant reduction in thrombosis rate when Qa surveillance was compared with clinical monitoring [ 10 ] or added to DU surveillance [ 9 ]. The one study that found similar thrombosis rates for Qa surveillance and clinical monitoring was reportedly not powerful enough to detect any differences [ 11 ]. Unfortunately, none of these studies considered the influence of Qa surveillance on AVF survival, or its cost-effectiveness.

Data on the economic impact of Qa surveillance are available from two observational studies with historical controls [ 5,6 ], one showing a significant reduction [ 6 ] and the other no change in access-related costs [ 5 ]. A simulated economic evaluation using the best data available in the literature predicted that Qa surveillance induces a modest increase in the cost of AVFs as compared with no monitoring [ 12 ].

More, methodologically adequate studies (e.g. randomized controlled or well-designed prospective trials with concurrent control groups [ 13 ]) are clearly needed to elucidate the value and cost-effectiveness of Qa surveillance in AVFs.

We report the results of a 5-year controlled cohort study aiming to establish whether adding Qa surveillance to unsystematic clinical monitoring (associated with elective stenosis repair) reduces the rates of thrombosis and access loss, and access-related costs in mature AVFs.

Subjects and methods

Study design

This is a retrospective analysis of data that were collected prospectively between January 2002 and December 2006. Based on the results of our controlled randomized trial evaluating the role of surveillance combined with preemptive stenosis repair on AVF survival [ 14 ], we hypothesized that Qa surveillance would halve the AVF thrombosis/failure rate, so, in calculating the sample size, we considered a hazard ratio of 2.0. We chose a study period of 5 years to compare our data on access-related costs with those calculated by Tonelli et al . [ 12 ]. Assuming a patient dropout of 15% and a mean access thrombosis/failure rate of 12% a year, we calculated that the study needed at least 120 subjects to achieve a power of 0.80 and an α error of 0.05. During the 5-year study period, 168 patients with mature AVFs were dialyzed at three satellite haemodialysis units in the Verona metropolitan area, at the Ospedale Policlinico, Verona (Unit A, 12 haemodialysis stations), the Ospedale Civile di Caprino Veronese (Unit B, 8 stations) and Ospedale di Valeggio sul Mincio (Unit C, 6 stations), respectively. All patients enrolled were similar in that they began treatment at the haemodialysis unit of the Division of Nephrology, Ospedale Civile Maggiore, Verona, and were subsequently transferred to the satellite unit closest to their homes once the access was considered mature (i.e. enabling successful cannulation at four or more consecutive haemodialysis sessions). The vascular access was fashioned by the same two surgical teams in all patients, one at the Ospedale Civile Maggiore (Surgical Team 1, comprising three vascular surgeons) and the other at the Ospedale Policlinico (Surgical Team 2, comprising two surgeons). The three satellite units differed in their access stenosis detection strategy, however: at two (Units B and C) the nursing staff implemented a unsystematic vascular access clinical monitoring program [frequent palpation before access cannulation, visual inspection for any arm swelling and aneurysms, recording cannulation difficulties, the inability to achieve the prescribed blood pump flow rate (Qb) and prolonged post-dialysis bleeding]; Unit A added systematic Qa surveillance (by the ultrasound dilution method) to clinical monitoring as of February 1998. Patients treated at Units B and C formed the Control group and those attending Unit A the Flow group.

During the study period, all endovascular and surgical procedures were performed by two interventional radiologists (Radiologist 1 at the Ospedale Civile Maggiore and Radiologist 2 at the Ospedale Policlinico) and two vascular surgeons (Surgeon 1 at the Ospedale Civile Maggiore and Surgeon 2 at the Ospedale Policlinico).

All patients alive as on December 2006 gave their informed consent to the study protocol.

Subjects

During the study period, 97 adult patients with a mature AVF were dialyzed at Units B and C (51 and 46, respectively) and 71 at Unit A. Nine patients from Unit A were excluded from the analysis, however: six because they were part of the control arm of a randomized control study on the role of surveillance and preemptive stenosis repair on AVF survival [ 14 ], and three because it was impossible to measure their Qa for anatomical reasons.

Therefore in the final analysis Control included 97 fistulae (from 97 patients) and Flow 62 (from 62 patients).

Monitoring/surveillance

Patients in Control were referred for access imaging (digital subtraction angiography or DU) only if there was clinical suspicion of stenosis, e.g. arm oedema, aneurysm onset or enlargement, palpation of stenotic segments, persistent (for at least three consecutive haemodialysis sessions) cannulation difficulties or inability to achieve the prescribed Qb, and excessive post-dialysis bleeding. Patients in Flow were referred for access imaging using the same criteria as in Control or a Qa <750 ml/min or dropping by >20%, whatever the absolute flow value. Whenever possible, the drop in Qa was compared with the mean Qa obtained from previous measurements. These Qa parameters were chosen for their high sensitivity in detecting stenosis in our experience [ 15 ]. Qa was measured by ultrasound dilution with a Transonic HD01 monitor (Transonic System Inc., Ithaca, NY, USA) every 1–4 months, as previously described [ 15 ]. Angiography was performed as described elsewhere [ 14 ] and DU using a PHILIPS HDI 5000 apparatus (Royal Philips Electronics, The Netherlands).

Intervention procedures

Any significant stenosis (>50% reduction in luminal diameter) identified by clinical criteria and warning of inadequate dialysis delivery was treated by angioplasty (PTA) or surgical revision in both groups. In Flow , significant stenoses with a Qa <500 ml/min or dropping by >20% were treated immediately [ 1 ], except for critical stenoses (≥90% diameter reduction) in mid-forearm and elbow/upperarm AVFs, which were treated even if the Qa was >500 ml/min and stable. In distal forearm AVFs, significant stenoses with a stable Qa >500 ml/min were surveilled more strictly with Qa measurements every 1–2 months and corrected if the Qa dropped below 500 ml/min. The repair procedure was chosen case by case for patent AVFs, at the discretion of and depending on the availability of the attending radiologist and vascular surgeon, with a view to correcting the stenosis without any major loss of venous capital available for puncture, as previously explained [ 16 ]. Elective PTA and surgical revision [creating a more proximal neo-anastomosis or inserting a short interposition poly-tetra-fluoro-ethylene (PTFE) graft] were conducted as described elsewhere [ 16 ]. It was our standard procedure to perform surgical thrombectomy and repair any stenosis of an occluded AVF, as previously described [ 17 ]. All clotted AVFs were evaluated by the vascular surgeon within 48 h of their detection. Some were considered unsuitable for revision due to the patient's poor clinical conditions or the extensive organization of the thrombus and/or inadequate veins. These accesses were consequently abandoned and a new permanent access created, i.e. a new proximal AVF in the same arm or a new AVF in the controlateral arm, a PTFE graft or a Tesio catheter. Patients with access loss were also given a temporary central venous catheter (CVC) to bridge the interval until a new vascular access was suitable for cannulation. PTA was performed as an outpatient procedure, while patients were hospitalized for all surgical procedures.

Outcomes

The primary outcomes were thrombosis, access loss and access-related cost, which are reported as individual rates and expressed as event/AVF-year. We also computed thrombosis-free patency, defined as the interval between enrolment and access thrombosis (regardless of any surgical and endovascular measures to maintain a functional and patent access) and access survival, defined as the interval from enrolment to access loss, including all intervening actions to maintain or restore access function after a thrombotic episode. Secondary cumulative patency (defined as the interval from the time of fistula maturation to access loss, including all intervening actions to maintain or restore access function after an episode of thrombosis) [ 18 ] was assessed in the subgroup of AVFs constructed from January 1998 onwards, when Qa surveillance became a routine in Unit A. Patients were censored because of death, transplantation, transfer to another unit or if they ended the study with a functional access. Secondary outcomes of the study were imaging procedure, stenosis detection, elective stenosis intervention (endovascular or surgical), temporary CVC placement and access-related hospitalization rates, and they were expressed as procedures/AVF-year. For comparison to other reports, population rates per access-year were also calculated for all outcomes.

Cost analysis

A direct access care-related cost was estimated for each procedure, including all expenses for detecting and correcting stenosis, thrombectomy, placement of a new access or a CVC and hospitalization. Procedural costs are the actual costs of personnel, equipment and material. The cost of access monitoring with the ultrasound dilution technique was based on the number of measurements taken a year and the working life of the machine (7 years) and included costs for personnel. Hospitalization costs were also actual costs based on an average day in hospital in a surgical ward. Specifically, the cost per procedure was established from the Azienda Ospedaliera di Verona financial data system as 2004 Euros [ 16 ].

Statistical analysis

Data are given as percentages, means ± standard devia- tion, means (95% confidence interval, CI) or medians (10th–90th percentile), as appropriate. Normally distributed continuous variables were analysed using Student's unpaired t- test, skewed variables using the Mann–Whitney U -test and categorical variables using Fisher's exact test.

Rates were calculated for each of the patients by dividing the number of events/procedures by the duration of follow-up in years. Population rates were calculated by dividing the total number of events by the total number of years of access follow-up. Fistula patency was analysed using the Kaplan–Meier method and differences between groups were evaluated by the Log-rank and Breslow tests. Hazard ratio (HR) was determined using the Cox's proportional hazard model. All tests were two sided, and differences were considered significant at P ≤ 0.05. All statistical analyses were performed using the SPSS software, version 11 (SPSS, Chicago, IL, USA).

Results

Patient and AVF characteristics are given in Table 1 . The two groups were well matched for baseline parameters and length of follow-up. None of the patients were treated with anticoagulants or anti-platelet agents due to access problems and the proportion of patients on anticoagulants and anti-platelet agents was similar in the two groups, i.e. 6.2% and 65% in Control versus 4.8% and 71% in Flow ( P = ns).

Table 1

Patient and AVF characteristics

 Control Flow P 
Number 97 62  
Males/females 62/35 34/28 0.319 
Patients’ age (years) 65.1 ± 14.1 63.4 ± 15.1 0.480 
Proportion with diabetes 18.6% 30.6% 0.087 
Proportion with cardiovascular disease 61.9% 69.4% 0.396 
Proportion of patients on dialysis on 1 January 2002 46.4% 46.8% 1.000 
AVF age (years) 2.4 ± 3.5 2.3 ± 3.5 0.830 
Proportion of AVFs created by surgical team 1 82.5% 71.0% 0.298 
AVF anastomotic site a [number (%)]:     
 Distal forearm 52 [54%] 26 [42%]  
 Mid-forearm 27 [28%] 23 [37%] 0.429 
 Elbow/upper arm 18 [18%] 13 [21%]  
Median [10th–90th percentile] follow-up (years) 1.75 [1.08–3.58] 2.08 [1.17–3.92] 0.592 
 Control Flow P 
Number 97 62  
Males/females 62/35 34/28 0.319 
Patients’ age (years) 65.1 ± 14.1 63.4 ± 15.1 0.480 
Proportion with diabetes 18.6% 30.6% 0.087 
Proportion with cardiovascular disease 61.9% 69.4% 0.396 
Proportion of patients on dialysis on 1 January 2002 46.4% 46.8% 1.000 
AVF age (years) 2.4 ± 3.5 2.3 ± 3.5 0.830 
Proportion of AVFs created by surgical team 1 82.5% 71.0% 0.298 
AVF anastomotic site a [number (%)]:     
 Distal forearm 52 [54%] 26 [42%]  
 Mid-forearm 27 [28%] 23 [37%] 0.429 
 Elbow/upper arm 18 [18%] 13 [21%]  
Median [10th–90th percentile] follow-up (years) 1.75 [1.08–3.58] 2.08 [1.17–3.92] 0.592 

a Depending on anastomotic site, AVFs were classified as distal forearm (anastomosis located in the lower 3rd of the forearm), mid-forearm (in the mid 3rd of the forearm up to 4 cm below the elbow crease) and elbow / upper arm (proximally in the elbow region or upper arm).

During the study, 45 patients died (24 in Control and 21 in Flow ) and 25 were transplanted or transferred to other units (16 in Control and 9 in Flow ).

Thirty-nine access-imaging procedures were performed in Control (30 angiograms and 9 DU) and 100 in Flow (87 angiograms and 13 DU). In Control , indications for imaging were inability to achieve the prescribed Qb (21), palpation of stenotic tracts (5), difficult cannulation (4), aneurysm (4), arm swelling (2) and post-repair evaluation (3). In Flow , indications for imaging were Qa <750 ml/min (47), a drop in Qa >20% (29), abnormal clinical findings (18 cases: inability to achieve the prescribed Qb in 8, difficult cannulation in 4, arm swelling in 3, excessive post-dialysis bleeding, aneurysm and palpation of stenotic tracts in 1) and post-repair evaluation (6). A total of 429 Qa measurements were taken in Flow , i.e. mean [95% CI] 3.82 [3.27–4.36] Qa measurement per AVF-year.

Thirty-two stenoses were identified in patent AVFs (from 23 patients) in Control and 62 (from 39 patients) in Flow . The positive predictive value (PPV) of clinical monitoring for stenosis (i.e. the proportion of AVFs with a positive test result that had an angiographically or DU-proven significant stenosis) was 89% in Control (32/36) and 67% in Flow (12/18) ( P = 0.067). PPV of Qa <750 ml/min was 74% (35/47) and for the drop in Qa by >20% was 52% (15/29).

In Control , 27 stenoses were electively repaired (21 by 22 PTAs and 6 by surgery), while 5 were not treated (3 associated with the presence of aneurysms and 2 with arm oedema), because they did not affect access cannulation or Qb delivery. None of the untreated stenotic AVFs thrombosed during the follow-up. In Flow , 46 stenoses were electively repaired (32 by 33 PTAs and 14 by surgery, 1 after failed PTA), 9 for clinical indication (5 were mid-forearm or elbow/upper arm AVFs with Qa >500 ml/min), 24 for a Qa <500 ml/min and 13 for a Qa ≥ 500 ml/min (associated with a >20% drop in Qa in 8, and with a critical stenosis in 5 mid-forearm or elbow/upper arm AVFs). The remaining 16 stenoses did not undergo treatment because Qa was >500 ml/min and stable. Three of these untreated AVFs (18.7%) thrombosed during follow-up. One stenosed AVF was deemed unsuitable to elective intervention and was substituted by a Tesio catheter. The proportion of AVFs that underwent elective surgery was similar in the two groups, i.e. 22% (6/27) in Control and 28% (13/46) in Flow ( P = 0.783). The initial anatomical (<30% residual stenosis) and functional success rate of elective PTA and surgery was 91% (20/22) and 83% (5/6) in Control and 91% (30/33) and 100% (14/14) in Flow .

In Control , 18/22 PTAs were performed by Radiologist 1 and 4/22 by Radiologist 2, who performed all the PTAs in Flow. The two radiologists had similar anatomical success rates of 89% (16/18) and 92% (34/37), respectively ( P = 1.000). In Control , elective surgical repair was performed by Surgeon 1 in 5/6 cases and by Surgeon 2 in 1/6 cases, who performed all the elective surgeries in Flow. Success rates were again similar between the two surgeons, i.e. 100% (5/5) and 93% (14/15), respectively ( P = 1.000).

Twenty AVFs thrombosed in Control and five in Flow . Thirteen thrombosed AVFs were abandoned in Control (three because the patients were dying and ten because the access was considered unsalvageable due to extensive organization of the thrombus or inadequate veins) and three in Flow (one because the patient was dying and two because of inadequate veins). Excluding the thrombosed AVFs in moribund patients, all nine accesses (seven in Control and two in Flow ) deemed suitable for surgical revision were successfully declotted and repaired, resulting in a salvage rate of thrombosed AVFs of 41% in Control and 50% in Flow ( P = 1.000). In Control , 14 thrombosed AVFs in non-moribund patients were evaluated by Surgeon 1 and 3 by Surgeon 2, who evaluated all 4 thrombosed AVFs in Flow . Salvage rates for thrombosed AVFs were 36% (5/14) for Surgeon 1 and 57% (4/7) for Surgeon 2 ( P = 0.397).

The Kaplan–Meier analysis showed that Flow had significantly higher thrombosis-free survival than Control (Figure 1 ).

Fig. 1

Unadjusted thrombosis-free survival. The graph shows the unadjusted thrombosis-free survival as of enrollment, according to the Kaplan–Meier analysis. Thrombosis-free survival was significantly better in Flow (open circles, continuous line) than in Control (closed triangles, dashed line). Note that the survival axis is truncated at 50%.

Fig. 1

Unadjusted thrombosis-free survival. The graph shows the unadjusted thrombosis-free survival as of enrollment, according to the Kaplan–Meier analysis. Thrombosis-free survival was significantly better in Flow (open circles, continuous line) than in Control (closed triangles, dashed line). Note that the survival axis is truncated at 50%.

Fourteen AVFs failed in Control and four in Flow. The Kaplan–Meier Log-rank analysis showed no difference between the two groups in access survival, but this was significantly higher in Flow than in Control at the Breslow test, suggesting that adding Qa surveillance helps to improve survival only in the early follow-up period (Figure 2 A). To identify the time frame for the beneficial effect of adding Qa surveillance, we also evaluated cumulative access patency in the subgroup of fistulae constructed from January 1998 onwards (Figure 2 B): the results were much the same and the visual inspection of the curves suggested that any advantage of Qa surveillance is limited to the first 3 years after access maturation.

Fig. 2

Unadjusted cumulative survival. The graphs show the unadjusted cumulative survival as of enrollment ( A ) and as of fistula maturation in the subgroup of AVFs placed after January 1998 ( B ). The Kaplan–Meier analysis showed that access survival was significantly better in Flow (open circles, continuous line) than in Control (closed triangles, dashed line) with the Breslow but not with the Log-rank test. Note that the survival axis is truncated at 50%.

Fig. 2

Unadjusted cumulative survival. The graphs show the unadjusted cumulative survival as of enrollment ( A ) and as of fistula maturation in the subgroup of AVFs placed after January 1998 ( B ). The Kaplan–Meier analysis showed that access survival was significantly better in Flow (open circles, continuous line) than in Control (closed triangles, dashed line) with the Breslow but not with the Log-rank test. Note that the survival axis is truncated at 50%.

Twenty-two temporary CVCs were placed in Control and four in Flow . The mean hospital stay was 1.66 [95% CI 0.76–2.48] days/AVF-year in Control and 0.65 [0.25–1.06] in Flow ( P = 0.640).

Table 2 shows the mean values of the individual rates and the adjusted HR [95% CI] for all outcomes. HR was adjusted for patients’ age and gender, diabetes and arteriovenous anastomosis site. Table 3 shows the population rates for all outcomes. The results of the cost analysis are shown in Table 4 . The total access-related cost was significantly lower in Flow than in Control ( P = 0.0001).

Table 2

Individual rates and adjusted Hazard ratio (HR) of study outcomes

 Control (mean [95% CI] event/AVF-year) Flow (mean [95% CI] event/AVF-year) HR [95% CI] flow versus control P 
Access imaging 0.24 [0.12–0.35] 0.88 [0.58–1.19] 2.96 [1.79–4.91] 0.0001 
Stenosis detection 0.19 [0.10–0.28] 0.59 [0.41–0.78] 2.55 [1.48–4.42] 0.001 
Elective stenosis repair 0.15 [0.06–0.23] 0.42 [0.24–0.61] 2.26 [1.16–4.43] 0.017 
Temporary CVC 0.35 [0.07–0.62] 0.02 [0.00–0.04] 0.14 [0.03–0.42] 0.010 
Thrombosis 0.30 [0.12–0.47] 0.03 [0.00–0.05] 0.27 [0.09–0.79] 0.017 
Access loss 0.20 [0.05–0.38] 0.02 [0.00–0.04] 0.35 [0.11–1.09] 0.071 
 Control (mean [95% CI] event/AVF-year) Flow (mean [95% CI] event/AVF-year) HR [95% CI] flow versus control P 
Access imaging 0.24 [0.12–0.35] 0.88 [0.58–1.19] 2.96 [1.79–4.91] 0.0001 
Stenosis detection 0.19 [0.10–0.28] 0.59 [0.41–0.78] 2.55 [1.48–4.42] 0.001 
Elective stenosis repair 0.15 [0.06–0.23] 0.42 [0.24–0.61] 2.26 [1.16–4.43] 0.017 
Temporary CVC 0.35 [0.07–0.62] 0.02 [0.00–0.04] 0.14 [0.03–0.42] 0.010 
Thrombosis 0.30 [0.12–0.47] 0.03 [0.00–0.05] 0.27 [0.09–0.79] 0.017 
Access loss 0.20 [0.05–0.38] 0.02 [0.00–0.04] 0.35 [0.11–1.09] 0.071 
Table 3

Population rates of study outcomes

 Control (event/AVF-year) Flow (event/AVF-year) 
Access imaging 0.172 0.660 
Stenosis detection 0.141 0.409 
Elective stenosis repair 0.119 0.304 
Temporary CVC 0.097 0.026 
Thrombosis 0.088 0.033 
Access loss 0.062 0.026 
 Control (event/AVF-year) Flow (event/AVF-year) 
Access imaging 0.172 0.660 
Stenosis detection 0.141 0.409 
Elective stenosis repair 0.119 0.304 
Temporary CVC 0.097 0.026 
Thrombosis 0.088 0.033 
Access loss 0.062 0.026 
Table 4

Cost analysis in Euro/AVF-year

 Unitary cost in Euro Control mean [range] Flow mean [range] P 
Qa surveillance 10 36 [0–160]  
Access imaging 54–99 19 [0–360] 77 [0–720]  
Elective angioplasty 571 64 [0–1142] 186 [0–2284]  
Elective surgery 1281 48 [0–1186] 202 [0–2617]  
Thrombectomy 1895 274 [0–5742] 21 [0–813]  
New vascular access 796–1946 288 [0–7535] 25 [0–439]  
Temporary CVC 183 64 [0–2153] 3 [0–125]  
Hospitalization 281 456 [0–6812] 193 [0–2882]  
Total cost  1213 [0–16139] 743 [0–5685] 0.0001 
 Unitary cost in Euro Control mean [range] Flow mean [range] P 
Qa surveillance 10 36 [0–160]  
Access imaging 54–99 19 [0–360] 77 [0–720]  
Elective angioplasty 571 64 [0–1142] 186 [0–2284]  
Elective surgery 1281 48 [0–1186] 202 [0–2617]  
Thrombectomy 1895 274 [0–5742] 21 [0–813]  
New vascular access 796–1946 288 [0–7535] 25 [0–439]  
Temporary CVC 183 64 [0–2153] 3 [0–125]  
Hospitalization 281 456 [0–6812] 193 [0–2882]  
Total cost  1213 [0–16139] 743 [0–5685] 0.0001 

Discussion

Our controlled cohort study shows that adding Qa surveillance to unsystematic clinical monitoring of mature AVFs is associated with a reduction in thrombosis rate and access-related cost, and an extension of access patency in the short term after maturation.

We confirmed that regular Qa surveillance is associated with significantly more access imaging, stenosis detection and elective repair, and significantly fewer temporary CVC placements.

Clinical monitoring revealed an excellent PPV for stenosis of 81% (89% in Control and 67% in Flow ), which was higher than the PPV of 39% reported by Maya et al . in fistulae [ 19 ], but similar to the one recently reported for physical examination by Asif et al. (a PPV of 84% for inflow and 91% for outflow stenoses) [ 20 ], and by Campos et al . (a PPV of 86%). These discrepancies are hardly surprising, however, given that clinical access assessment is subjective, depending on the operator. The PPV of Qa surveillance was 66%, which means that the Qa criteria we adopted to decide when imaging was needed were associated with a large number (approximately one of three) of unnecessary procedures. Had we used the Qa <500 ml/min threshold proposed by the K/DOQI guidelines [ 1 ], its PPV for stenosis would have been 81% (25/31), but we would have overlooked too many significant stenoses (26, 52%). Such a strategy would have had the advantage of reducing the number of imaging procedures and related costs, at the expense of a higher risk of thrombosis, as suggested by the finding that three of five thrombosed AVFs in Flow had a Qa >500 ml/min immediately before clotting. These observations raise the question of whether the Qa threshold for imaging and stenosis repair recommended by the K/DOQI should be adjusted and higher Qa thresholds should be used (as suggested by others [ 22 ]), especially for mid-forearm or elbow/upper arm AVFs (even if this risks increasing the number of unnecessary procedures).

Qa surveillance was associated with a higher elective stenosis repair rate in patent AVFs. As reported elsewhere [ 17 , 23 , 24 ], elective repairs were highly successful, with an overall immediate success rates of 95%. In most cases, stenosis was corrected by PTA, but we also made liberal use of elective surgical revision rather than PTA, whenever the lesions were thought unlikely to respond to angioplasty (critical or long stenoses) or in the case of juxta-anastomotic stenoses in distal forearm AVFs. Both elective treatments were equally successful in our hands.

By showing that adding Qa surveillance to unsystematic clinical monitoring led to a nearly 4-fold reduction in the risk of thrombosis, our study confirms the results of previous adequately powered prospective controlled trials on AVFs [ 9,10 ] and of a controlled randomized study conducted at our institution on stenotic AVFs [ 14 ]. Our population thrombosis rate during Qa surveillance (0.033 event/AVF-year) compares favourably with the lowest thrombosis rates reported in the literature using Qa surveillance, which range 0.04–0.10 thromboses/AVF-year [ 3 , 5–8 , 10,11 ]. The thrombosis rate of 0.088 event/AVF-year in our Control arm was also at the lower end of the range reported in observational studies with bedside clinical monitoring (0.03–0.17 thromboses/AVF-year) [ 3 , 5–8 , 10,11 , 25,26 ] and within the range reported by many studies using Qa surveillance [ 3 , 5–8 , 11 ]. This rise the issue of whether adding Qa surveillance (with the associated extra burden on dialysis staff and imaging procedures) is worthwhile for an access that achieves a low thrombosis rate with non-aggressive clinical monitoring alone. Of course, the decision to implement Qa surveillance in AVF may depend on other outcomes, such as an improvement in access functional life or cost containment.

In our study, adding Qa surveillance to clinical monitoring was not associated with any significant extension of the functional life of the access. Closer analysis of the data suggests, however, that Qa surveillance does help to improve access patency. First, the observed 0.35 HR of access loss with Qa surveillance and its 95% CI support the notion that Qa surveillance is effective in improving patency, despite the lack of statistical significance [ 27 ]. Second, the Kaplan–Meier analyses showing that AVF survival was significantly better in Flow than in Control using the Breslow test [ 28 ] indicate that short-term patency improved with Qa surveillance and suggests that this benefit may be limited to the first 3 years after access maturation, whereas Qa surveillance does not appear to offer any additional advantage over clinical monitoring later on.

Cost analysis showed that adding Qa surveillance reduces access-related expenses by a mean of 500 Euro per patient-year: the added cost associated with access surveillance, imaging procedures and elective stenosis correction in Flow was completely offset by the saving associated with lower thrombectomy, new access and temporary CVC placement rates. Our data are consistent with the observational study by McCarley et al . [ 6 ], but at variance with two other studies [ 5 , 12 ]. The difference between ours and Wijnen et al .'s study [ 5 ] is likely to stem from the difference in thrombosis rates with Qa surveillance between the two studies (0.033 versus 0.09 thromboses/AVF-year), while the discrepancy between our own and Tonelli et al. 's findings [ 12 ] may depend on the huge difference in cost for Qa surveillance, access imaging and endovascular procedures at our institution and in Canada.

We recognize that our study has important limitations. First, there is the lack of randomization and blinding, though this may be mitigated by the fact that some authors [ 13 ] do not believe that controlled studies with a cohort design (like ours) systematically overestimate the effect of treatment by comparison with randomized controlled trials. Moreover, there may be a bias due to centre effect, as the separation between groups also distinguishes between the different centres, different nursing staff, and different nephrologists and endovascular and surgical teams. The impact of such a centre effect should be minimal, however, since we found a similar positive predictive value of clinical monitoring for stenosis obtained by the different sets of nursing staff, the choice of elective stenosis treatment by PTA or surgery was comparable between the different medical teams and the initial success rates for elective endovascular and surgical stenosis repair and the salvage rates for thrombosed fistulae were likewise similar between the different endovascular and surgical teams. Finally, our study was possibly underpowered for the detection of differences in access survival, due to the unexpectedly low failure rate in Control . We are aware that our study cannot provide a definitive answer on the role of Qa surveillance in AVF and that further, better-designed trials need to be undertaken.

In conclusion, this controlled cohort study shows that adding Qa surveillance to unsystematic clinical monitoring of mature fistulae increases the rate of stenosis detection and the elective treatment, and reduces the need of temporary CVCs, containing thrombosis and access-related cost, and improving access patency rates in the first 3 years after fistula maturation.

We wish to thank the patients who took part in the study and the nursing staff of the dialysis units.

Conflict of interest statement . None declared.

References

1
National Kidney Foundation K/DOQI
Clinical Practice Guidelines and Clinical Practice Recommendations for vascular access 2006
Am J Kidney Dis
 , 
2006
, vol. 
48
 
Suppl S1
(pg. 
S176
-
S273
)
2
Bakran
A
, et al.  . 
Management of the renal patient: clinical algorithms on vascular access for hemodialysis
2003
 
3
Schwab
SJ
Oliver
MJ
Suhocki
P
, et al.  . 
Hemodialysis arteriovenous access: detection of stenosis and response to treatment by vascular access blood flow
Kidney Int
 , 
2001
, vol. 
59
 (pg. 
358
-
362
)
4
Branger
B
Granolleras
C
Dauzat
M
, et al.  . 
Hemodialysis arteriovenous fistula thrombosis frequency; contribution of Doppler and ultrasound dilution technique
Nephrologie
 , 
2004
, vol. 
25
 (pg. 
17
-
22
)
5
Wijnen
E
Planken
N
Keuter
X
, et al.  . 
Impact of a quality improvement program based on vascular access flow monitoring on costs, access occlusion and access failure
Nephrol Dial Transplant
 , 
2006
, vol. 
21
 (pg. 
3514
-
3519
)
6
McCarley
P
Wingard
RL
Shir
Y
, et al.  . 
Vascular access blood flow monitoring reduces access morbidity and costs
Kidney Int
 , 
2001
, vol. 
60
 (pg. 
1164
-
1172
)
7
Lok
CE
Bhola
C
Croxford
R
, et al.  . 
Reducing vascular access morbidity: a comparative trial of two vascular access monitoring strategies
Nephrol Dial Transplant
 , 
2003
, vol. 
18
 (pg. 
1174
-
1180
)
8
Shahin
H
Reddy
G
Sharafuddin
M
, et al.  . 
Monthly access flow monitoring with increased prophylactic angioplasty did not improve fistula patency
Kidney Int
 , 
2005
, vol. 
68
 (pg. 
2352
-
2361
)
9
Sands
JJ
Jabyac
PA
Miranda
CL
, et al.  . 
Intervention based on monthly monitoring decreases hemodialysis access thrombosis
ASAIO J
 , 
1999
, vol. 
45
 (pg. 
147
-
150
)
10
Roca-Tey
R
Gausch
RS
Ibrik
O
, et al.  . 
Vascular access surveillance with blood flow monitoring: a prospective study with 65 patients
Nefrologia
 , 
2004
, vol. 
24
 (pg. 
246
-
252
)
11
Polkinghorne
KR
Lau
KKP
Saunder
A
, et al.  . 
Does monthly native arteriovenous fistula blood-flow surveillance detect significant stenosis—a randomized controlled trial
Nephrol Dial Transplant
 , 
2006
, vol. 
21
 (pg. 
2498
-
2506
)
12
Tonelli
M
Klarenbach
S
Jindal
K
, et al.  . 
Kidney disease network: economic implications of screening strategies in arteriovenous fistulae
Kidney Int
 , 
2006
, vol. 
69
 (pg. 
2219
-
2226
)
13
Concato
J
Shah
N
Hoewitz
RI
Randomized, controlled trials, observational studies, and the hierarchy of research design
N Engl J Med
 , 
2000
, vol. 
342
 (pg. 
1887
-
1892
)
14
Tessitore
N
Lipari
G
Poli
A
, et al.  . 
Can blood flow surveillance and pre-emptive repair of subclinical stenosis prolong the useful life of arteriovenous fistulae? A randomized controlled study
Nephrol Dial Transplant
 , 
2004
, vol. 
19
 (pg. 
2325
-
2333
)
15
Tessitore
N
Bedogna
V
Gammaro
L
, et al.  . 
Diagnostic accuracy of ultrasound dilution access blood flow measurement in detecting stenosis and predicting thrombosis in native forearm fistulas for hemodialysis
Am J Kidney Dis
 , 
2003
, vol. 
42
 (pg. 
331
-
341
)
16
Tessitore
N
Mansueto
G
Lipari
G
, et al.  . 
Endovascular versus surgical preemptive repair of forearm arteriovenous fistula juxta-anastomotic stenosis; analysis of data collected prospectively from 1999 to 2004
Clin J Am Soc Nephrol
 , 
2006
, vol. 
1
 (pg. 
448
-
454
)
17
Lipari
G
Tessitore
N
Poli
A
, et al.  . 
Outcomes of surgical revision of stenosed and thrombosed forearm arteriovenous fistulae for hemodialysis
Nephrol Dial Transplant
 , 
2007
, vol. 
22
 (pg. 
2605
-
2612
)
18
Sidawy
AN
Gray
R
Besarab
A
, et al.  . 
Recommended standards for reports dealing with arteriovenous hemodialysis access
J Vasc Surg
 , 
2002
, vol. 
35
 (pg. 
603
-
610
)
19
Maya
ID
Oser
R
Saddekni
S
, et al.  . 
Vascular access stenosis: comparison of arteriovenous grafts and fistulae
Am J Kidney Dis
 , 
2004
, vol. 
44
 (pg. 
859
-
865
)
20
Asif
A
Leon
C
Orozco-Vargas
LC
, et al.  . 
Accuracy of physical examination in the detection of arteriovenous fistula stenosis
Clin J Am Soc Nephrol
 , 
2007
, vol. 
2
 (pg. 
1191
-
1194
)
21
Campos
RP
Chula
DC
Perreto
S
, et al.  . 
Accuracy of physical examination and intra-access pressure in the detection of stenosis in hemodialysis arteriovenous fistula
Semin Dial
 , 
2008
 
Feb 1. [Epub ahead of print]
22
Lopot
F
Neyedly
B
Valek
M
Vascular access monitoring: methods and procedures—something to standardize?
Blood Purif
 , 
2005
, vol. 
23
 (pg. 
36
-
44
)
23
Turmel-Rodrigues
L
Pengloan
J
Baudin
S
, et al.  . 
Treatment of stenosis and thrombosis in hemodialysis fistulas and grafts by interventional radiology
Nephrol Dial Transplant
 , 
2000
, vol. 
15
 (pg. 
2029
-
2036
)
24
Beathard
Litchfield T.
Physician Operators Forum of RMS Lifeline, Inc.
Effectiveness and safety of dialysis vascular access procedures performed by interventional nephrologists
Kidney Int
 , 
2004
, vol. 
66
 (pg. 
1622
-
1632
)
25
Konner
K
Hulbert-Shearon
TE
Roys
EC
, et al.  . 
Tayloring the initial vascular access for dialysis patients
Kidney Int
 , 
2002
, vol. 
62
 (pg. 
329
-
338
)
26
Ravani
P
Marcelli
D
Malberti
F
Vascular access surgery by renal physicians: the choice of native arteriovenous fistulas for hemodialysis
Am J Kidney Dis
 , 
2002
, vol. 
40
 (pg. 
1264
-
1276
)
27
Curran-Everett
D
Taylor
S
Kafadar
K
Fundamental concepts in statistics: elucidation and illustration
J Appl Physiol
 , 
1998
, vol. 
85
 (pg. 
775
-
786
)
28
Marubini
E
Valsecchi
MG
Marubini
E
Valsecchi
MG
A family of two sample rank tests
Analysing Survival Data from Clinical Trials and Observational Studies
 , 
1995
Chichester, UK
Wiley
(pg. 
99
-
103
)

Comments

0 Comments