- Question: What is the incidence and risk factor(s) of venous air embolism during endoscopic retrograde cholangiopancreatography (ERCP)?
- Findings: Venous air embolism occurred in 20 patients (2.4%) of the 843 that underwent ERCP; half of these resulted in hemodynamic alterations such as cardiac collapse.
- Meaning: There is a high incidence of venous air embolism in patients that undergo ERCP, some of which result in a serious adverse event. Clinicians are encouraged to use modalities such as a precordial Doppler ultrasound for detection and intervention.
Endoscopic retrograde cholangiopancreatography (ERCP) is an advanced endoscopic diagnostic and therapeutic procedure used to treat a variety of conditions such as obstructive jaundice, traumatic or iatrogenic damage to the bile ducts, and obstruction related to bile duct and pancreatic tumors. ERCP was initially utilized for diagnostic purposes; however, it is now widely available and has become a very valuable tool for the management of biliary and pancreatic diseases.1 ERCP is used for therapeutic interventions such as sphincterotomy, removal of biliary stones, dilation of strictures, and placement of plastic or metal stents. Today, over 500,000 ERCP’s are performed annually in the United States.2 Endoscopists are aware of common adverse events associated with ERCP such as pancreatitis, cholangitis, bleeding, postspincterotomy perforation, and sedation-related cardiopulmonary issues. However, less common, but potentially fatal complications include systemic venous air embolism (VAE).3 A compilation of different case studies in which VAE occurred is shown in Supplemental Digital Content, Table 1, http://links.lww.com/AA/C445. Air embolism is a rare adverse event that can cause fatal cardiopulmonary compromise. Patients who survive VAE can have long-term neurological effects.4 Immediate recognition and intervention is needed to prevent possible catastrophic consequences. ERCP involves the use of insufflation of air or carbon dioxide to dilate the lumen of the gastrointestinal tract, thereby improving visualization during the procedure. It is presumed that insufflation of gas coupled with disruption of a mucosal vascular barrier predisposes patients to a VAE.
Precordial Doppler ultrasound (PDU) has been shown to be an efficacious, inexpensive, and noninvasive monitoring device for the detection of VAE. Changes in the heart tones, usually referred to as “mill-wheel murmur,” are characteristic of VAE. The anesthesiologist has an important role in the detection of these characteristic sounds and subsequent intervention.5 As shown in Supplemental Digital Content, Table 1, http://links.lww.com/AA/C445, there have been numerous case reports, but no quantitative data have been published regarding the incidence and risk factors of VAE during ERCP. The primary objective of the present study was to assess the incidence of VAE in patients undergoing ERCP. A secondary objective was to assess which procedural factors were associated with VAE in patients undergoing diagnostic and therapeutic ERCP at a busy academic institution.
This was a prospective cohort study evaluating the incidence of VAE in patients >19 years scheduled for ERCP in the GI Endoscopy Suites of the University of Alabama at Birmingham Hospital. This study was approved by the University of Alabama at Birmingham’s Institutional Review Board and requirement for written informed consent was waived by the institutional review board. This manuscript adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Patients, ≥19 years of age, undergoing an ERCP during the study interval were included. Patients who had a failed ERCP were excluded from the analysis. A failed ERCP was defined as an incomplete procedure due to various causes; the primary cause being the inability to cannulate the ampulla or due to surgical changes making access to the ampulla impossible. Demographic and other clinical data such as type of anesthetic, positioning, gas used for insufflation, and therapeutic procedures performed were recorded for each patient. Type of stent was also recorded because metal stents have been proposed as a risk factor for VAE. Anesthetic management was left to the discretion of the attending anesthesiologist.
Detection of VAE
Monitoring and the detection of VAE were performed using a PDU (Medasonics Versatone, Model D8; Cooper Surgical, Trumball, IL). The face of the PDU probe was coated with acoustic gel and placed in the third intercostal space 1 cm to the right of the left sternal border adjusting position until clear heart tones were heard. The probe was then secured using an adhesive dressing. The majority of patients were then placed in the prone position for the procedure. A digital audio recorder was plugged into the output port of the PDU and activated at the beginning of the procedure. If a VAE was suspected based on distinct changes in the sounds detected by the PDU, the attending anesthesiologist and gastroenterologist were notified, and the event was noted in the anesthetic record. Interventions were performed at the discretion of the attending anesthesiologist in collaboration with the gastroenterologist. To assure that intraoperatively observed events were supportive of a VAE event, digital recordings were reviewed by investigators with experience in PDU monitoring, and anesthetic records were assessed for additional evidence of VAE such as hemodynamic changes or alterations in end-tidal gas monitoring. Hemodynamic changes were defined as a significant increase or decrease in heart rate and decrease in blood pressure resulting in an intervention (eg, cardiovascular collapse and severe bradycardia).
Statistical analyses were performed by using SAS version 9.4 (Cary, NC). Patient demographics and intraoperative data were summarized using means and ranges (for continuous variables) and frequencies and percentages (for categorical variables). The 2-sample t test and Fisher exact test were used to compare patients with and without VAE. Point estimates and exact binomial (Clopper-Pearson) CIs for the proportion of subjects having a VAE were calculated. Fisher exact test was used to evaluate the association between type of procedure and incidence of VAE. A P value <.05 was considered statistically significant.
Between January 2013 and April 2014, 948 patients were scheduled to undergo ERCP and were monitored for VAE using PDU. Of these, 105 had an incomplete or a failed procedure and were excluded. The majority of failed procedures were attributed to the inability to cannulate the ampulla or due to surgical changes making access to the ampulla impossible. Notably, the incidence of VAE in those 105 patients was 0. The demographics of the remaining 843 patients that underwent ERCP are shown in Table 1. Twenty subjects had a VAE identified, making the overall incidence proportion of VAE in subjects undergoing ERCP 2.4% (95% CI, 11.5%–3.6%). There were no significant differences observed in the demographics between the patients who developed VAE and those that did not.
Table 2 indicates the incidence of VAE in subgroups of patients who received various additional interventions. For each type of intervention, Fisher exact test was used to assess whether the incidence proportion of VAE for patients receiving that intervention significantly differed from those not receiving the intervention. Forty-nine (5.8%) of the procedures were purely diagnostic and 794 (94.2%) were therapeutic ERCP procedures. All VAE occurred in subjects who underwent therapeutic interventions in which there was a potential for physical disruption of a mucosal/vascular barrier. If one combines ERCPs done for diagnostic purposes only (n = 49) with those done in which stent removal was the only therapy performed (n = 151; both groups had a 0% incidence of VAE observed), then there is a statistically significant relationship between ERCPs associated with minimally invasive additional procedures and those associated with more invasive procedures (P = .006). There was also a significant relationship between type of procedure and the incidence of VAE. Two procedure types had significantly higher incidences of VAE: stent removal with replacement (4.4% vs 1.8%; P = .049) and cholangioscopy (9.1% vs 2.1%; P = .040). That is, patients receiving these procedures had higher incidences compared to those requiring other procedures. A metal stent was used in 152 cases, only 1 of which resulted in the development of a VAE suggesting that use of a metal stent is not a risk factor for VAE.
Hemodynamic changes (eg, increase or decrease in heart rate and decrease in blood pressure) were noted in only 10 (50%) VAE patients. There was cardiovascular collapse in 2 subjects (hypotension, hypoxia, end-tidal carbon dioxide measure reduced >50%, procedures terminated), severe bradycardia in 2 additional subjects, a ≥15 beat per minute increase in heart rate in 2 subjects and new-onset ectopic heartbeats in 2 subjects (1 of whom also had an increase in heart rate). The 2 cases associated with cardiovascular collapse were both associated with stent removal and replacement; in 1 case, the exchange was followed by necrosectomy. None of the embolism cases in which carbon dioxide was used as an insufflation agent had hemodynamic changes.
The present study is the largest prospective series performed to date assessing for the incidence of VAE in subjects undergoing ERCP. As evidenced by these results, VAE can be a serious complication of ERCP. In our series, the incidence of VAE was 20 (2.4%) in 843 patients. Interestingly, no VAE occurred during minimally invasive procedures (stent removal only and diagnostic ERCP) but occurred in all other interventional subgroups including stent replacement, biopsy, cholangioscopy, sphincterotomy, dilation, gallstone retrieval, and necrosectomy. Incidence rates of VAE by procedure are listed in Table 2.
During ERCP, air or CO2 is used for insufflation to distend the bowel lumen to allow sufficient visualization and manipulation of instruments within the duodenum. Air is introduced at relatively high pressures and can be introduced at a flow of 30 mL/s.6 The amount of air entrained that can cause a lethal event in an adult is estimated at 200–300 mL or approximately 3–5 mL/kg. The rate of air entrainment is also important as the pulmonary circulation and alveolar interface provide for dissipation of the intravascular gas.7 In those patients in which CO2 was used as the insufflating agent and embolism occurred, there were no adverse hemodynamic consequences.
While this is the largest series to examine the relationship between ERCP and VAE, there are several limitations to this study. The primary limitation is that only PDU was used for detection of VAE; there was not another confirmatory method such as transthoracic echocardiogram or transesophageal echocardiogram used. Of the 20 VAE cases, 10 remained subclinical but for the PDU monitoring—the caveat to that is the act of monitoring may have altered clinical practice in those 10 patients thus preventing hemodynamic deterioration.
Transthoracic echocardiogram would not be possible in a prone patient and transesophageal echocardiogram could not be accomplished in a patient with an endoscope in place. End-tidal nitrogen monitoring was not utilized during these procedures. Additionally, detection of VAE would not be accomplished if the PDU probe was placed incorrectly or if the probe moved during prone positioning. Once the correct position was confirmed, the probe was attached with an adhesive dressing. Correct position was once again confirmed after prone positioning.
The present study is restricted to experience at 1 institution and thus limits the ability to generalize findings to other centers. In this study, several gastrointestinal procedural physicians with advanced training in ERCP with a range of years of experience in performing ERCP conducted the procedures. This is both a limitation and strength, as physicians with varying degrees of experience may actually make the results generalizable; yet it also introduces confounding.
Our institution utilizes the anesthesia care team model of attending anesthesiologist supervising certified registered nurse anesthetists or residents. A variety of attending anesthesiologists and certified registered nurse anesthetists provided care during this study. They received training regarding correct probe placement and the sounds to listen for. They were instructed to alert the procedural physician and the attending anesthesiologist if there was suspicion of a VAE. It is possible that a subtle finding could have been missed over the course of the study. Finally, while volume status could potentially impact hemodynamic changes if an embolism occurred, fluid administration was appropriate for all cases in this study; it was not considered as an outcome variable or any part of this study.
Air was the most commonly used insufflating agent during the course of the study and as a result, our institution is now using CO2 as the insufflating agent for ERCP in almost all instances. CO2 is absorbed faster than nitrogen, the principal gas in room air. Therefore, some authors recommend its use as the insufflating agent during endoscopic procedures due to its rapid absorption should an embolic event occur.8
There is a significant incidence of VAE during ERCP, and therefore VAE should continue to be on the radar of anesthesia professionals. In this study, the occurrence of VAE is 2.4%; use of monitoring devices such as PDU may prove beneficial in detecting VAE. Consideration could be given to monitoring with PDU in all ERCP procedures even if a lower risk procedure is planned, as, at times, a low-risk procedure can become a high-risk procedure depending on clinical findings. Utilizing PDU monitoring can allow the detection of VAE. This can allow efforts to terminate the insufflation and immediately treat the patient thereby hopefully mitigating significant hemodynamic compromise or passage of air into the arterial system. Such monitoring should lead to improved patient outcomes after the occurrence of a VAE.
Name: Lubana K. Afreen, BS.
Contribution: This author helped collect the data and in interpretation of analysis, review the literature, and prepare the manuscript.
Name: Ayesha S. Bryant, MSPH, MD.
Contribution: This author helped with the interpretation of the data analysis and writing and editing of the manuscript.
Name: Tetsuzo Nakayama, MD.
Contribution: This author helped with the development of methodology, assisted with patient recruitment, data collection, and entry.
Name: Timothy J. Ness, MD, PhD.
Contribution: This author helped with the development of the study methodology, data interpretation, and writing of the manuscript.
Name: Keith A. Jones, MD.
Contribution: This author helped with the project oversight, develop the study methodology, data interpretation, and manuscript writing and editing.
Name: Charity J. Morgan, PhD.
Contribution: This author helped perform the data analysis and write the statistical sections of the manuscript.
Name: Charles M. Wilcox, MD.
Contribution: This author helped with the study methodology and facilitated manuscript writing and editing.
Name: Mark C. Phillips, MD.
Contribution: This author helped initiate the study, develop the methodology, recruit the patient, interpret the data, and write the manuscript.
This manuscript was handled by: Richard C. Prielipp, MD.
Acting EIC on final acceptance: Thomas R. Vetter, MD, MPH.