Author + information
- Received January 25, 2002
- Revision received March 4, 2002
- Accepted April 17, 2002
- Published online July 17, 2002.
- Malte Kelm, MD*,* (, )
- Stefan M Perings, MD*,
- Thomas Jax, MD*,
- Thomas Lauer, MD*,
- Frank C Schoebel, MD*,
- Matthias P Heintzen, MD*,
- Christian Perings, MD* and
- Bodo E Strauer, MD, FACC, FESC*
- ↵*Reprint requests and correspondence:
Dr. Malte Kelm, Heinrich-Heine-Universität Düsseldorf, Medizinische Klinik und Poliklinik B, Klinik für Kardiologie, Pneumologie und Angiologie, Moorenstraβe 5, D-40225 Düsseldorf, Germany.
Objectives We sought to determine the incidence of arteriovenous fistulas (AVF), identify risk factors for AVF, and follow up the clinical outcome of femoral AVF.
Background Arteriovenous fistulas are a potential harmful complication of cardiac catheterization. Incidence and clinical outcome of iatrogenic AVF are unknown so far, although important for risk stratification and treatment.
Methods A total of 10,271 consecutive patients undergoing cardiac catheterization were followed up prospectively over a period of three years. Diagnosis of AVF was performed by duplex sonography.
Results The incidence of AVF was 0.86% (n = 88). The following significant and independent risk factors for AVF were identified: high heparin dosage (odds ratio [OR]) = 2.88), coumadin therapy (OR = 2.34), puncture of the left groin (OR = 2.21), arterial hypertension (OR = 1.86), and female gender (OR = 1.84). Within 12 months 38% of all AVF closed spontaneously. No signs of cardiac volume overload or limb damage were observed in patients with persisting AVF. None of the risk factors for AVF influenced the incidence or the rate of AVF closure. Only intensified anticoagulation showed a tendency to extend AVF persistence.
Conclusions Almost 1% of patients undergoing cardiac catheterization acquire femoral AVF, for which patient- and procedure-related risk factors could be identified. One-third of iatrogenic AVF close spontaneously within one year. Cardiac volume overload and limb damage are highly unlikely with AVF persistence. Thus, a conservative management for at least one year seems to be justified.
Recent years have witnessed an exponential increase in the number of diagnostic and interventional cardiac catheterizations with several million procedures performed worldwide annually. The most common site to access the arterial circulation is the femoral artery. Potential complications are arteriovenous fistula (AVF), pseudoaneurysm, hemorrhage, arterial thrombosis, and peripheral embolization. Previous studies addressing the incidence of these iatrogenic complications were based on retrospective analyses of cohorts referred to vascular surgery. The reported incidence of all vascular complications ranged from 1% to 9%, and the incidence of AVF varied from 0.006% to 0.14% (1–5). However, as most of these retrospective analyses of surgical data bases included only patients with operative repair of a post-catheterization complication, they are not suitable to prospectively determine the incidence and the clinical outcome of iatrogenic AVF.
The clinical significance of AVF may result from hemodynamic relevant left-to-right shunts. Potential side effects of AVF are a reduced blood flow in the downstream vascular bed and progressive heart failure due to cardiac volume overload (6). It appears crucial to identify possible risk factors for AVF, which may provide a rationale for effective risk stratification in patients undergoing cardiac catheterization. Furthermore, characterizing the clinical outcome of iatrogenic AVF may help to define therapeutic strategies for these patients. Thus, we conducted a three-year follow-up study to estimate the incidence of AVF, identify risk factors for AVF, and determine the clinical outcome of femoral AVF.
Within two years a total of 10,271 consecutive patients undergoing endovascular catheterization were included in this prospective three-year follow-up study. The study was performed in accordance with the regulations of the local ethics committee and the declaration of Helsinki. All patients were informed and gave oral/written consent for endovascular catheterization. Patients were excluded from this study for the following reasons: 1) the patient refused to grant oral/written consent for examination (n = 3); 2) the arterial approach was performed via the brachial artery (n = 43); or 3) the patient was transferred to another hospital before removal of the sheaths (n = 3).
Puncture of the femoral artery was performed by experienced investigators (>600 procedures per year) using a sheath size of ≤7 F for diagnostic catheterization (n = 7,186) and of ≥8 F for therapeutic procedures (n = 3,085). An additional 7-F venous sheath was inserted in the femoral vein for all patients undergoing interventional procedures or in whom simultaneous left- and right-heart catheterization was performed. In patients treated with coumadin, treatment was discontinued until an international normalized ratio (INR) <1.8 was reached. Unless contraindicated, all patients undergoing diagnostic catheterization were treated with <5,000 international units (IU) of heparin, whereas 500 mg acetylsalicylic acid and >12,500 IU of heparin in a weight-adjusted mode were applied during percutaneous coronary intervention (PCI). Glycoprotein receptor blockers (GP) were given only rarely at the time of study period (n = 7). The sheaths were removed approximately 1 h after diagnostic and 4 h after interventional procedures if activated clotting time (ACT) was <150 s. Local hemostasis was achieved by manual compression for at least 15 min. If necessary, heparin was re-administered 4 h after sheath removal. Thereafter, patients were kept down for 24 h before ambulating. Seventy-nine of 88 patients with AVF received long-term treatment with acetylsalicylic acid 100 mg/day.
Diagnosis and follow-up of AVF
All patients were routinely examined by the medical staff before and after catheterization including auscultation of both groins. Special attention was paid to the detection of a new continuous femoral bruit (systolic/diastolic) on the day after sheath removal. In case of clinical suspicion of a vascular complication (hematoma, pulsatile mass, atypical pain, vessel thrombosis, or a new femoral bruit), a duplex scan was performed. Duplex scans were acquired with a Hewlett-Packard Sonos 2500 color Doppler sonography system, which was also used to estimate shunt volume across femoral AVF in a subset of patients with sufficient conditions for quantitative ultrasound measurements (non-obese subjects, without groin pain or hematoma after cardiac catheterization, n = 22). An AVF was diagnosed if all of the following criteria were fulfilled: 1) a colorful “speckled” mass at the level of the fistula with turbulent flow in the arteriovenous connection; 2) increased venous flow more to the proximal than the contralateral side of the fistula, with a lack of respiratory variation and a pulsatile arterial component in the affected vein; and 3) decreased arterial flow distal to the suspected fistula. The study protocol did not include the assessment of the incidence of pseudoaneurysm or severe hematoma.
To estimate shunt volume, the diameter of the artery and the mean blood flow velocity directly proximal to the fistula were determined. Identical measurements were done on the opposite side. The shunt volume was calculated as follows: SV = BFAVF− BFnorm, where SV = shunt volume [ml/min], BFAVF= blood flow in the artery proximal to the AVF [ml/min], BFnorm= blood flow in the normal contralateral artery [ml/min]. Blood flow was calculated by the formula: BF = (Vm × π × d2)/4) × h, where Vm = mean blood flow velocity proximal to the AVF [cm/s], π = 3.14, d = diameter of the artery proximal to the AVF [cm], hr = heart rate [beats/min]. All patients in whom an AVF was diagnosed underwent routine follow-up examinations. They were scheduled for outpatient visits every three months, which included: 1) an interview for potential cardiac and peripheral symptoms of AVF and a physical examination; 2) a duplex scan of the left and right femoral artery and vein; and 3) an electrocardiogram (ECG) and an echocardiographic recording in those patients in whom appropriate scans of all cardiac chambers could be obtained.
Risk factors for incidence and persistence of AVF
At the end of the study period the control group was established by collecting data of 278 patients selected as a representative random sample survey by picking every 35th patient of the entire study population. The ratio of male to female patients was comparable in the AVF (65% to 35%) and the control group (74% to 26%). The following parameters were assessed: patient-related factors (age, gender, weight, height, body mass index [BMI], arterial hypertension) and procedure-related factors (number and size of used sheaths, puncture of right versus left groin, procedural heparin dosage, and ongoing coumadin therapy after catheterization). Single sheath denotes puncture of only one femoral vessel, and double sheath denotes puncture of both the femoral artery and vein. During follow-up, identical patient- and procedure-related parameters were entered into the computerized data base every three months. Immediately after diagnosis, patients were offered treatment by either surgical repair of AVF, so far considered as a standard procedure at our institution, or to enter follow-up of non-interventional observation.
The incidence of AVF is given as percentage of all included patients. The data of metric variables (age, weight, height, and BMI) are given as median and interquartile distance (I50) because age showed an asymmetrical distribution. To determine independent risk factors for AVF, the Fisher exact test for univariate analysis and a logistic regression model for multivariate analysis were performed. For this purpose, metric variables (age, BMI) and non-metric variables with more than two categories (sheath size and procedural heparin dosage) were transferred into dichotomized variables. The odds ratio (OR) with 95% confidence interval (CI) was calculated to determine the relative risk of each suspected risk factor. Statistical significance was considered when the 95% CI of the multivariate OR exceeded the value of 1. The incidence and rate of AVF closures were analyzed by Kaplan-Meier analyses and log-rank tests, respectively. Incidence of enlargement of the right atrium (>20 mm) and ventricle (>25 mm) were assessed by the McNemar test for dichotomized variables. The Wilcoxon matched pair test was used to calculate statistical differences of the dimension of the left atrium and ventricle before and after acquisition of AVF. Statistical significance was considered at a p value of <0.05. All statistical analyses were performed using the software package SPSS 7.5.2.G.
The characteristics of the study population are given in Table 1. In 88 of 10,271 patients an AVF was diagnosed. Thus, overall incidence of AVF amounted to 0.86%. Arteriovenous fistulas were slightly more frequent after interventional procedures than after diagnostic procedures (1.1% vs. 0.75%) (p = 0.12). In a subset, shunt volume was estimated to be in a range of 160 to 510 ml/min (median = 330 ml/min, n = 22). Shunt volume was 310 ml/min (250 to 350) in closed AVF and 350 ml/min (160 to 510) in persistent fistulas, respectively (p = NS).
To determine independent risk factors for AVF, the representative control group and the AVF group were compared, and the OR for patient- and procedure-related parameters were calculated (Table 2). Female gender and arterial hypertension significantly enhanced the risk for AVF, whereas age and BMI had no impact. Procedure-related risk factors for AVF were puncture of the left groin and the intensity of anticoagulation. High procedural heparin dosage (≥12,500 IU) and coumadin therapy were identified as independent procedure-related risk factors for AVF. In contrast, the size and number of sheaths used were of no significant relevance (see Table 2).
During follow-up, 38% of all AVF closed spontaneously, whereas AVF persisted in 62% of patients. Sixty-nine percent of the spontaneous AVF closures took place in the first four months, and all occurred within 12 months (n = 33), resulting in a median period to closure of three months (Fig. 1). None of the risk factors for acquisition of an AVF significantly influenced the incidence of closure (Table 3) or the time to AVF closure (Table 4). However, a trend to extended persistence of AVF became obvious with a high procedural heparin dosage (p = 0.065) and with coumadin therapy (p = 0.091). None of the patients with persistent AVF experienced clinical symptoms of limb damage such as reduced walking distance, soft pulse, pain, and restless or cold legs. Furthermore, no significant signs of right heart failure were observed, that is, occurrence of right bundle branch block, changes of heart axis in electrocardiographic recordings, or echocardiographic signs of right ventricular enlargement. The left heart dimensions before/after AVF acquisition were as follows: 42 ± 7 / 43 ± 8 for left atrium (p = NS), and: 53 ± 3 / 54 ± 9 for left ventricle (p = NS). The percentage distributions of right atrial and ventricular dilation before/after AVF acquisition were 9%/18% and 12%/9%, respectively (p = NS).
The present study is the first to investigate prospectively the incidence and clinical outcome of iatrogenic femoral AVF in a large number of consecutive patients. The major findings of this study are as follows: 1) the overall incidence of femoral AVF after cardiac catheterization approximates 1%; 2) independent patient- and procedure-related factors for AVF could be identified, facilitating risk stratification; 3) 38% of all AVF closed spontaneously within one year; 4) no severe local or systemic side effects were observed with AVF persistence, which may be of importance for future treatment of iatrogenic AVF.
Incidence of AVF
The incidence of iatrogenic AVF found in this large prospective scale study (0.86%) corroborates the data reported in two small prospectively designed studies. In accordance with our protocol, patients in the study of Kent et al. were routinely examined by the medical staff for a new femoral bruit after cardiac catheterization, followed by a duplex scan. The authors diagnosed six new AVF in 1,838 consecutive patients (0.3%) (7). Using primary duplex scanning, Kresowik diagnosed four new AVF in only 144 patients, resulting in an incidence of 2.8% (8). As expected, the incidence of AVF was slightly higher with primary duplex scanning than with primary clinical examination. Thus, in our study population the ultrasound-guided detection of AVF might have revealed a somewhat higher incidence of AVF. However, this could not be performed, because our several-fold larger sample size did not allow routine duplex scanning in every out of the 10,271 patients. Using the approach of combined clinical examination and consecutive duplex scanning, we could identify 88 patients with iatrogenic AVF, which is by far the largest sample size studied so far. Only this large sample size enables one to perform statistical analysis that will reliably identify risk factors for the acquisition of AVF.
Arterial hypertension and female gender could be identified as independent, patient-related risk factors. The driving force of elevated blood pressure together with the hypertension-associated increase of vascular stiffness might promote development of an AVF during puncture or sheath removal. A possible explanation for the higher risk of AVF in women is that their smaller vascular diameter could require more punctures to achieve arterial access, which in turn might lead to a higher incidence of AVF. However, we cannot prove this hypothesis definitely, because we did not document the number of attempts to puncture. An increased frequency of females and of arterial hypertension have also been reported for the incidence of other post-catheterization complications such as hemorrhage and pseudoaneurysm of the femoral artery (9). Aortic regurgitation is frequently associated with increased systolic pressure and altered mechanical properties of the arterial wall due to widening of the blood pressure amplitude. In our cohort, only two individuals of the AVF group had aortic regurgitation. Because of the limited number of patients, we cannot prove that aortic insufficiency may have led to an increased rate of incidence or persistence of AVF.
Procedure-related risk factors were the mode and intensity of anticoagulation and the site of access. A puncture of the left groin was associated with more than a twofold risk of acquiring an AVF. All of the investigators were right-handed and performed the procedures from the patient’s right-hand side. Thus, puncturing the left instead of the right groin alters the accustomed angle of needle to vessel and may provoke the simultaneous puncture of femoral artery and vein, facilitating the development of AVF. High heparin dosage and coumadin therapy increased risk of AVF by more than twofold. An intensified anticoagulation thus promotes development of AVF as was reported for post-procedural false aneurysm and hemorrhage (10). Hence using a lower INR threshold might reduce the incidence of this complication.
The patient’s age, obesity, size, and number of used sheaths did not predict for an increased frequency of iatrogenic AVF. This contrasts strikingly with previous observations indicating that these parameters represent significant predictors for the incidence of false aneurysm and uncontrolled hemorrhage (10). Thus, in the present study, specific risk factors for the acquisition of an iatrogenic AVF could be identified, whereas others could be ruled out, and these factors are different from those reported for local hematoma and aneurysm. Therefore, female individuals with arterial hypertension, a need for intensified anticoagulation, and puncture on the left groin have to be stratified as patients at high risk of acquiring an AVF during cardiac catheterization.
Clinical outcome of AVF
To date, no data exist that predict the outcome of iatrogenic femoral AVF, that is, the probability of spontaneous closure and the period until closure. In the present study, one-third of all AVF closed spontaneously within one year, and the majority closed even more rapidly—within the first four months after acquisition of AVF. Neither the incidence nor the rate of AVF closure was significantly influenced by the risk factors for AVF. Only with high heparin dosage and coumadin therapy was a trend toward prolonged persistence observed. This finding is most likely explained by a reduced thrombus formation, a key event in the closure of AVF. In none of the patients with persistent AVF were clinical signs of limb damage or right heart failure observed. Left-to-right shunts at the level of the atrial or interventricular septum deteriorate right heart function only, when shunt volume exceeds 30% of cardiac output. In a subset of our study population, shunt volume was estimated in the range of 160 to 510 ml/min. This is far below significant left-to-right heart shunts and considerably lower than dialysis shunts (11)and thus unlikely to exert detrimental effects on the heart. It has been postulated from dialysis studies that with decreasing flow volume (<500 ml/min) (11)the probability of shunt closure increases. Using ultrasound techniques we were able to reliably quantitate shunt volume only in some but not all patients with AVF. Thus, we could not specify the impact of shunt volume on the persistence of AVF.
In view of these findings, therapeutic approaches for patients with persistent iatrogenic AVF deserve re-evaluation. Three different therapeutic strategies have been recommended so far to close femoral iatrogenic AVF: surgical repair, implantation of covered stents, and ultrasound-guided compression. Although a rather low efficacy for successful compression of AVF has been reported (12,13), it should be applied as first-line therapy because of its non-invasive nature. Implantation of stents into the common femoral artery represents an approach currently under clinical investigation (14). The limited number of patients studied, the potential likelihood of restenosis, and stent failure due to repetitive movements in the groin region prevent a final evaluation of this therapeutic alternative so far (15,16). Based on studies of the natural history of significant inborn or traumatic fistulas, surgical repair has frequently been recommended for patients with iatrogenic AVF (3). This therapeutic approach is hampered by a considerable perioperative mortality and postoperative complication rate (17). Thus, in light of our findings, surgical repair should be recommended only cautiously, and patients with persistent femoral iatrogenic AVF should initially be managed non-operatively and followed up clinically with duplex scans.
We have no data documented regarding the number of previous catheterizations performed in each patient. Thus, we cannot rule out whether a pre-punctured groin might be more susceptible for AVF. Particularly, the formation of connective and scar tissue that results from repetitive puncture might lead to an atypical anatomy of artery and vein and thus promote AVF development.
In the present study, arterial puncture was performed by experienced investigators, usually distal to the inguinal ligament and proximal to the bifurcation of femoral arteries. In this area, the femoral artery and vein are clearly separated from each other. At more peripheral access sites, the deep femoral vein lies inferior to the artery (18). Thus, with more distal access sites, the probability of puncturing artery and vein, thereby simultaneously facilitating the development of iatrogenic AVF, increases (19,20). Because we did not perform duplex scanning of the groin routinely in all patients before cardiac catheterization, we can estimate neither the influence of access site along the common femoral artery nor the incidence of topographic abnormalities as a potential cause for the acquisition of a femoral iatrogenic AVF.
After the study period, a slight downsizing of arterial sheaths for cardiac catheterization occurred. Presently some catheterization laboratories are using ≤5 F and ≤7 F sheaths for diagnostic and interventional procedures, instead of ≤7 F and ≥8 F sheaths as we used. The size of sheaths may be an important determinant of the incidence and extent of post-procedural hemorrhage and false aneurysm. However, in our analysis the incidence and outcome of AVF was independent from the number and size of sheaths used, indicating that a further downsizing of sheaths would not necessarily have altered our results.
The majority of our patients were treated with acetylsalicylic acid, but only a minority with GP IIb/IIIa inhibitors. Therefore, the heparin dosages used were somewhat higher than those recommended currently. Whether or not intensified antithrombotic therapy in addition to ACT-adjusted anticoagulation with heparin, at least during PCI, may alter the incidence and persistence of AVF in the future, deserves further studies.
One-third of iatrogenic AVF close spontaneously within one year, and cardiac volume overload and limb damage are highly unlikely with AVF persistence; therefore, a conservative management for at least one year seems to be justified.
The technical assistance of S. Matern is gratefully acknowledged. We also thank Reinhart Willers, PhD, Department of Mathematics at the Heinrich Heine University Düsseldorf, for his expert advice and critique in statistical aspects of this study.
- activated clotting time
- arteriovenous fistula/fistulas
- body mass index
- confidence interval
- French (catheter size of 0.3 mm/F)
- interquartile distance
- international normalized ratio
- international units
- odds ratio
- percutaneous coronary intervention
- Received January 25, 2002.
- Revision received March 4, 2002.
- Accepted April 17, 2002.
- American College of Cardiology Foundation
- Popma J.J.,
- Satler L.F.,
- Pichard A.D.,
- et al.
- Schaub F.,
- Theiss W.,
- Heinz M.,
- Zagel M.,
- Schömig A.