Prior research has provided inconsistent data regarding the risk factors associated with complications from arterial cannulation. The goal of this study was to clearly define the incidence and risks factors associated with arterial cannulation complications.
After obtaining institutional review board approval, all patients requiring arterial line placement with documentation were included in this retrospective study between January 1, 2006, and December 31, 2012. Leveraging two robust data warehouses, the Perioperative DataMart and the Mayo Clinic Life Silences System, the authors cross-matched arterial line cannulation with a documented vascular consult, neurologic consult, infection, or return to surgery within 30 days in order to identify the initial patient population.
A total of 62,626 arterial lines were placed in 57,787 patients, and 90.1% of the catheters placed were 20-gauge catheters. The radial artery was cannulated in 94.5% of patients. A total of 21 patients were identified as having experienced vascular complications or nerve injuries, resulting in a complication rate of 3.4 per 10,000 (95% CI, 2.1 to 5.1). Cardiac surgery had the largest number of catheters placed (n = 15,419) with 12 complications (complication rate = 7.8 per 10,000; 95% CI, 4.0 to 13.6). The rate of complications differed significantly (P < 0.001) across the three most common catheter sizes (2.7 per 10,000 [95% CI, 1.5 to 4.4] for 20 gauge, 17.2 per 10,000 [95% CI, 4.7 to 43.9] for 18 gauge, and 9.4 per 10,000 [95% CI, 1.1 to 34.1] for 5 French).
In a large retrospective study, the authors document a very low rate of complications with arterial line placement.
In a series of 57,787 patients receiving arterial cannulation, 21 patients were identified as having experienced vascular complications or nerve injuries, resulting in a very low complication rate of 3.4/10,000. The rate of complications differed significantly (P < 0.001) across the three most common catheter sizes (2.7/10,000 for 20 gauge, 17.2/10,000 for 18 gauge, and 9.4/10,000 for 5 French). Given the low frequency of complications observed, the current study does not have sufficient statistical power to make definitive conclusions regarding the risk factors (listed in the appendix).
Prior research has provided inconsistent data regarding the risk factors associated with arterial cannulation.
The goal of the current study was to define the risks associated with arterial cannulation in order to prevent the potential development of vascular complications.
In a series of 57,787 patients receiving arterial cannulation, 21 patients were identified as having experienced vascular complications or nerve injuries, resulting in a very low complication rate of 3.4 per 10,000. The rate of complications differed significantly (P < 0.001) across the three most common catheter sizes (2.7 per 10,000 for 20 gauge, 17.2 per 10,000 for 18 gauge, and 9.4 per 10,000 for 5 French). Given the low frequency of complications observed, the current study does not have sufficient statistical power to make definitive conclusions regarding risk factors (listed in the appendix).
ARTERIAL line cannulation is frequently used in the operating room and intensive care unit settings to provide easy access for continuous and real-time systemic blood pressure measurements, blood gas analysis, and other laboratory measurements.1 Arterial cannulation provides invaluable hemodynamic information that aides practitioners in precise treatment of patients during the critical periods of care, but the benefit of the provided information does not come without risk. In humans, the use of arterial catheters can be traced back to 1949, with current widespread use throughout the United States and Europe totaling 8 million and 2.5 million placements, respectively.2 Surgical and anatomical considerations are the influencing factors when choosing the site for arterial cannulation. Possible sites available for cannulation include the radial, brachial, axillary, femoral, ulnar, dorsalis pedis, posterior tibial, and temporal arteries.3 The radial artery is the most common site for cannulation because it is thought to have a low rate of associated complications. This low rate of complication occurs as a result of extensive collateral circulation involving the ulnar artery and palmar arches with flow to the distal limb.4 Complications associated with the arterial cannulation include temporary vascular occlusion, thrombosis, ischemia, hematoma formation, and local and catheter-related infection and sepsis.2,3,5 Rarer complications associated with arterial cannulation include nerve damage and severe ischemia necessitating surgical amputation.6 There are many risk factors that may contribute to these vascular complications.
Prior research has provided inconsistent data regarding the risk factors associated with arterial cannulation.3,7–10 Due to the outdated and conflicting information provided by previous research, there is further need for a large-scale investigation into the risk factors associated with arterial line cannulation. The goal of our study was to clearly define the risks associated with arterial cannulation in order to prevent the potential development of vascular complications. Our hypothesis was that the factors such as diseases associated with arterial injury or atherosclerosis, catheter diameter size, catheter length, catheter composition or type, number of puncture attempts, and duration of arterial line in combination with arterial line cannulation are associated with an increased risk of an arterial and/or limb injury requiring consultation of a vascular surgeon, a neurologist, or need for surgery within 30 days.
Materials and Methods
After obtaining institutional review board approval, all patients requiring arterial line placement with documentation available in the Perioperative DataMart system at Mayo Clinic in Rochester, Minnesota, were included in this retrospective study between January 1, 2006, and December 31, 2012.
Since 2005, for all patients who had an arterial line placed in the operating room, a computerized form was filled out. This form included type of skin preparation, type of catheter placed and its size, site of catheter placement, patient position, use of ultrasound guidance and/or pressure transduction, catheter insertion depth, observations, and comments. The number of arterial line documents in the initial data pool was 77,388. The timeframe of data collection began at the time of arterial line insertion and ended at 30-day postline removal. Using two robust data warehouses, the Perioperative DataMart and the Mayo Clinic Life Silences System, we cross-matched arterial line cannulation with a documented vascular consult, neurologic consult, or return to surgery within 30 days in order to identify the initial patient population. After identification of a research population, chart reviewers (J.B., A.H., and S.K.) confirmed the accuracy of a diagnosis of vascular insult or neurologic compromise resulting from arterial cannulation. A review of 100 random negative charts was conducted in order to confirm the accuracy of the data pool. Patients who were identified with a vascular surgery consult or neurologic consult were subjected to a thorough chart review conducted to assess the presence of the above listed risk factors such as previous arterial injury, catheter diameter size, catheter length, catheter composition, number of puncture attempts, duration of arterial line, and coexisting diseases. To identify any infectious complications associated with arterial lines, we cross-matched our data with that in our infection surveillance program called Electronic Assisted Surveillance of Infection program.
Chart reviews were conducted only on those patients who provided a prior consent authorizing research to be performed on their medical records according to the Minnesota law. Patients under the age of 18 yr were also excluded (fig. 1). Patients who experienced line placement in a nonprocedure or nonoperating room area were excluded from our study. In total, 2,609 charts were individually reviewed for evaluation of a complication related to arterial line placement.
Patient characteristics were summarized by using mean ± SD for continuous variables and frequency counts and percentages for categorical variables. For patients who underwent multiple surgeries during the study period, only the first surgery was included. However, since some patients had multiple arterial lines placed during surgery, the total number of arterial lines was greater than the total number of patients. The frequency of arterial line complications was calculated and expressed as the rate per 10,000 patients and also as the rate per 10,000 arterial lines. Rates were calculated overall and separately according to the type of procedure and summarized by using point estimates and exact 95% CIs. To assess whether the complication rate differed between radial versus femoral (and radial vs. brachial) in the same patients, an exact McNemar test was performed for the subset of patients who had both types of arterial lines placed for the same surgery. Conditional logistic regression was used to assess the patient characteristics potentially associated with arterial line complications. For these analyses, procedure type was included as the stratification variable. Findings from the conditional logistic regression analyses are summarized using the point estimate for the odds ratio (OR) along with 95% confidence limits. Due to the limited number of arterial line complications observed, we supplemented the traditional logistic regression approach with a Bayesian approach using data augmentation.11,12 The Bayesian approach requires a “prior” assumption about the effect of each risk factor on the outcome. Because little evidence is available for the risk factors assessed, for each risk factor, we specified a normal prior for β with a mean of 0 and variance of 2.0. A mean of 0 and variance of 2.0 for the prior distribution of β correspond to a prior median for the OR (ORprior = eβ) of 1.0 indicating no association, and a prior 95% interval for the OR of that ranges from 0.0625 to 16.0. For the data augmentation approach, a prior data record for β is constructed so that fitting a no-intercept logistic regression model to this record would result in point and variance estimates that correspond to those specified for the prior. The prior data record is then added to the actual data records to construct the augmented data set. When this data set is analyzed by using logistic regression, the resulting OR estimate and corresponding 95% confidence limits are the approximate posterior median (ORpost) and corresponding 95% posterior limits.11,12 Data were analyzed by using SAS software (SAS version 9.3; SAS Institute Inc., USA).
A total of 77,388 arterial line documents were identified between January 1, 2006, and December 31, 2012, with 14,762 documents excluded as previously described (fig. 1). A total of 62,626 arterial lines were placed in 57,787 patients. The vast majority (90.1%) of the catheters placed were 20-gauge catheters. The majority of femoral catheters were either 15-cm 5-French catheters (44.7%) or 10-cm 18-gauge catheters (44.8%). Patients who had more than one line were included only once, and the first catheter placed was used. The demographics, disease comorbidities, and preoperative medications versus the site of arterial line placement are shown in table 1. The vast majority of arterial lines placed were in the radial artery (94.5%).
Upon review of the positive vascular consults, neurology consults, and returns to surgery documents in the 30-day postcatheter removal, a total of 21 complications were identified as related to arterial line placement. The majority of arterial lines were placed in cardiac surgery patients, and the majority of complications occurred in cardiac surgical patients (table 2). Of the related complications, 2 were in the brachial arterial lines (complication rate of 12.3 per 10,000 patients [95% CI, 1.5 to 44.4 per 10,000 patients]), 4 were in the femoral arterial lines (complication rate of 9.0 per 10,000 patients [95% CI, 2.4 to 22.9 per 10,000 patients]), and 15 were in radial arterial lines (complication rate of 2.7 per 10,000 patients [95% CI, 1.5 to 4.4 per 10,000 patients]) (table 3). These differences were not statistically significant. A listing of complication by site of arterial catheter is provided in the appendix. Of note, only 11 of the 21 patients identified had a complication that required an intervention. The rate of complications differed significantly (P < 0.001) across the three most common catheter sizes (2.7 per 10,000 [95% CI, 1.5 to 4.4] for 20 gauge, 17.2 per 10,000 [95% CI, 4.7 to 43.9] for 18 gauge, and 9.4 per 10,000 [95% CI, 1.1 to 34.1] for 5 French). To further assess the association between catheter size and complications, a conditional logistic regression analysis was performed with site (radial/brachial/femoral) as the stratification variable. From this analysis, the risk for complication was found to be significantly (P = 0.029) associated with catheter size (OR, 8.0; 95% CI, 1.7 to 37.1 for 18 gauge vs. 20 gauge but not for 5 French vs. 20 gauge, OR, 5.9; 95% CI, 0.5 to 63.8).
Among 3,091 patients who had both radial and femoral lines placed during the same procedure, there were 6 patients who developed complications (4 femoral arterial line complications vs. 2 radial arterial line complications; McNemar test exact P = 0.688). Among 357 patients who had both radial and brachial lines placed during the same procedure, there was 1 patient who developed a complication (1 radial line complication vs. 0 brachial line complications; McNemar test exact P = 1.0).
Patient characteristics potentially associated with arterial line complications are listed in table 4. From the standard conditional logistic regression analysis, male sex was protective (P = 0.046) and preoperative use of clopidogrel was associated with increased risk (P = 0.033). A history of connective tissue disease and cerebral vascular accident were associated with increased risk, but this was not statistically significant. When a Bayesian approach was used, the confidence bounds for the OR contained 1.0 for all of the characteristics assessed. This suggests that the statistically significant findings from the standard analysis may be questionable given the small number of complications observed.
There were 14 patients who developed the line-associated infection due to their arterial line, resulting in an overall infection rate of 2.4 per 10,000 patients (95% CI, 1.3 to 4.0 per 10,000 patients). Line-associated infections were positive catheter tip cultures on surveillance cultures. For the patients who developed infections, all of the arterial lines were placed in the radial artery. Five of the patients had a tracheostomy performed as their surgery, five of the patients underwent a laparotomy, and four of the patients underwent cardiac surgery.
The goal of our study was to clearly define the risks associated with arterial cannulation in order to prevent the potential development of vascular complications. The majority of the catheters placed were of only three types. From conditional logistic regression, stratified by catheter site, the likelihood of complications was found to be increased with the larger catheters versus the 20-gauge catheters. At our institution, we tend to put the larger catheters in the femoral site. Although not statistically significant, the radial arterial line site had the lowest rate of complications and the femoral had the highest. We found some evidence to suggest that female sex and preoperative use of clopidogrel were potential predictors of arterial line complications. However, given the low number of complications observed, and the confidence bounds for the OR from the Bayesian analysis, these findings should be considered exploratory. Overall, we document a low rate of complications associated with arterial line placement.
Prior research has provided inconsistent data regarding the risk factors associated with arterial cannulation.3,7–10 A study conducted by Hoencamp et al.7 focused on examining the possible damaging effects of arterial monitoring catheters on arterial functioning in critical care patients after elective surgical procedures. The study looked at 23 patients with an arterial catheter, using the noncannulated arm as a control and using the radial-to-ulnar ratio for the assessment of arterial functioning. The radial-to-ulnar ratio was defined as the radial artery systolic pressure divided by the ulnar artery systolic pressure. The study concluded that there is a significant change in the hemodynamic functioning of the cannulated arteries 1 day and 5 days after arterial line removal, with the significant change decreasing and disappearing 30 days after line removal. The authors’ conclusion was that arterial catheterization on a short-term basis is a safe procedure.
A study by Wilkins13 found that emboli originating from a proximal site, excessive trauma resulting from large-bore cannulae, prolonged shock, hyperlipoproteinemia, or prior vascular disease are contributing factors to the onset of tissue necrosis.
A study conducted by Slogoff et al.9 examined the safety of radial cannulation during a 4-month period of time on adult patients scheduled for elective operations requiring radial arterial cannulation at the Texas Heart Institute, 1,699 cardiovascular surgical patients and in 83 patients in whom cannulation was performed in another artery after failure at the radial site. The study indicated that the incidence of abnormal radial arterial flow was related neither to the duration of cannulation nor to the size or material of the cannula. Hematoma and female sex were the only factors found to significantly increase the incidence of abnormal radial flow. Abnormal flow was found early in the study in 21.2% of participants, but sensory abnormalities and other vascular complications only occurred in 0.6% of all cases. They showed, as do we, that arterial lines have a very low rate of complications and female sex is a risk factor for complications.
Lipira et al.8 published a case report and literature review in May 2010 discussing hand ischemia and axillary arterial catheter use. The case reported involved many factors including hypothermia, a female patient, and a trauma involving long-bone fractures. Frezza and Mezghebe14 performed a retrospective comparison of 4,392 in surgical and medical intensive care unit patients with arterial catheters. This study found that the most common complication were “vascular insufficiency,” bleeding, and infection. A prospective study by Scheer et al.3 found that axillary catheters have a lower rate of obstruction, ischemia, and thrombosis than radial catheters. Risk factors listed by Lipira et al. were the use of vasopressors, previous injury to the artery, duration of cannulation (over 48 or 72 h), hematoma, disseminated intravascular coagulation, reduced cardiac output, and the female sex. Our study was not able to address many of these risk factors except size of the catheter and female sex.
How does our complication rate compare with those in the cardiac catheterization laboratory because the cardiologists cannulate either the radial or femoral arteries for angiography and percutaneous coronary intervention (PCI)? There are two meta-analyses of randomized trials comparing radial to femoral access for angiography and PCI. Jolly et al.15 demonstrated a major bleeding in 0.05% for radial access and 2.3% for femoral access. Another study by Jolly et al.16 found major vascular complications in 1.4% of radial cannulations and 3.7% of femoral cannulations. Minor bleeding occurred in 2.9% of radial cannulations and 3.4% of femoral cannulations. It should be noted that the cannulas used for angiography and PCI are larger than those used in the operating room, and the patients frequently are on antiplatelet agents after PCI. Furthermore, cardiologists often use specific vascular closure devices, which may result in complications.
There are limitations to our study. Our initial intention was to look at the number of arterial cannulation attempts and the duration of catheter placement as predictors of adverse outcomes. We were not able to capture these data. Our outcome measures are designed to detect only major complications. We were not able to detect minor complications such as transient neurologic symptoms. We also did not look at the complications of arterial line placement outside of the operating room. Also, given the low frequency of complications observed, our study does not have sufficient statistical power to make definitive conclusions regarding the risk factors.
In a very large retrospective study of prospectively collected data, we found that larger arterial catheters had a higher complication rate than 20-gauge catheters. We also document a very low rate of complications overall with arterial line placement.
All financial and material support for the research and work described in this article is received from the Department of Anesthesiology, Mayo Clinic College of Medicine, Rochester, Minnesota.
The authors declare no competing interests.