Racial and Ethnic Disparities in Operative Experience Among General Surgery Residents: A Multi-Institutional Study from the US ROPE Consortium.

OBJECTIVE: To determine the relationship between race/ethnicity and case volume among graduating surgical residents. BACKGROUND: Racial/ethnic minority individuals face barriers to entry and advancement in surgery; however, no large-scale investigations of the operative experience of racial/ethnic minority residents have been performed. METHODS: A multi-institutional retrospective analysis of the Accreditation Council for Graduate Medical Education case logs of categorical general surgery residents at 20 programs in the US Resident OPerative Experience Consortium database was performed. All residents graduating between 2010 and 2020 were included. The total, surgeon chief, surgeon junior, and teaching assistant case volumes were compared between racial/ethnic groups. RESULTS: The cohort included 1343 residents. There were 211 (15.7%) Asian, 65 (4.8%) Black, 73 (5.4%) Hispanic, 71 (5.3%) "Other" (Native American or Multiple Race), and 923 (68.7%) White residents. On adjusted analysis, Black residents performed 76 fewer total cases (95% CI, -109 to -43, P <0.001) and 69 fewer surgeon junior cases (-98 to -40, P <0.001) than White residents. Comparing adjusted total case volume by graduation year, both Black residents and White residents performed more cases over time; however, there was no difference in the rates of annual increase (10 versus 12 cases per year increase, respectively, P =0.769). Thus, differences in total case volume persisted over the study period. CONCLUSIONS: In this multi-institutional study, Black residents graduated with lower case volume than non-minority residents throughout the previous decade. Reduced operative learning opportunities may negatively impact professional advancement. Systemic interventions are needed to promote equitable operative experience and positive culture change.

Disparities in the Operative Experience Between Female and Male General Surgery Residents: A Multi-institutional Study From the US ROPE Consortium.

OBJECTIVE: To examine differences in resident operative experience between male and female general surgery residents. BACKGROUND: Despite increasing female representation in surgery, sex and gender disparities in residency experience continue to exist. The operative volume of male and female general surgery residents has not been compared on a multi-institutional level. METHODS: Demographic characteristics and case logs were obtained for categorical general surgery graduates between 2010 and 2020 from the US Resident OPerative Experience Consortium database. Univariable, multivariable, and linear regression analyses were performed to compare differences in operative experience between male and female residents. RESULTS: There were 1343 graduates from 20 Accreditation Council for Graduate Medical Education-accredited programs, and 476 (35%) were females. There were no differences in age, race/ethnicity, or proportion pursuing fellowship between groups. Female graduates were less likely to be high-volume residents (27% vs 36%, P < 0.01). On univariable analysis, female graduates performed fewer total cases than male graduates (1140 vs 1177, P < 0.01), largely due to a diminished surgeon junior experience (829 vs 863, P < 0.01). On adjusted multivariable analysis, female sex was negatively associated with being a high-volume resident (OR = 0.74, 95% CI: 0.56 to 0.98, P = 0.03). Over the 11-year study period, the annual total number of cases increased significantly for both groups, but female graduates (+16 cases/year) outpaced male graduates (+13 cases/year, P = 0.02). CONCLUSIONS: Female general surgery graduates performed significantly fewer cases than male graduates. Reassuringly, this gap in operative experience may be narrowing. Further interventions are warranted to promote equitable training opportunities that support and engage female residents.

Identifying barriers to resident robotic console time in a general surgery residency through a targeted needs assessment.

Robotic-assisted general surgery is experiencing exponential growth. Despite our institution's high volume, residents often graduate with inadequate console experience. Our aim was to identify the educational needs of residents and perceived barriers to residents' console time from both attendings and residents. Separate surveys were created and distributed to robotic surgery faculty and general surgery residents at our institution. Questions were a variety of modalities and focused on the robotic surgery experience at our institution, including barriers to resident console time from both attending surgeon and resident perspectives. Although residents' interest in robotic surgery exceeded that of open and laparoscopic surgery, confidence in their robotic skills was low compared to the other modalities. The top barriers to participating in robotic cases according to residents included minimal or no previous console time with the attending, lack of simulator time, and being required to perform bedside assistant duties. Faculty reported resident preparedness, prior robotic skill demonstration, simulator time, case complexity, and their own confidence as significant factors influencing resident console time. Using these results, we concluded that the design and implementation of a formal robotic surgery curriculum should incorporate simulation-based opportunities for residents to practice their skills, improve confidence, and increase console experience. In addition, simulation opportunities for faculty should also be considered to allow for improvement and maintenance of robotic surgical skills.

Evaluation of a new elderly trauma triage algorithm.

OBJECTIVES: Undertriage is common in patients 55 years and older and is even worse for those 65 and older. In 1999, the Florida legislature implemented a statewide trauma system, including a new Florida trauma triage algorithm (FTTA). This study examines how the new system affected prehospital triage in younger versus older patients. METHODS: A retrospective review of appropriate triage was conducted at a regional level 2 trauma center during a 1-year period. Patients were considered to have major trauma if they were FTTA positive or had an Injury Severity Score (ISS) of ≥ 16. An internal trauma review panel examined hospital discharge data to assess triage accuracy. Odds ratios (ORs) and confidence intervals (CIs) were calculated. RESULTS: A total of 49% of nontrauma patients 15 to 54 years old were seen at the trauma center compared with 83% of FTTA positive and 86% of patients with an ISS ≥ 16 (OR 2.88, 95% CI 2.44-3.41). For those with an ISS ≥ 16, the OR was 6.53 (95% CI 4.07-10.47). For patients 55 years and older, 52% of nontrauma patients were triaged to the trauma center versus 59% of FTTA positive and 64% of patients with ISS ≥ 16 (OR 1.03, 95% CI 0.93-1.15). Patients 55 years and older with an ISS ≥ 16 had only a slightly increased triage effect (OR 1.67, 95% CI 1.08-2.58) compared with those with an ISS 0 to 15 (OR 1.00, 95% CI 0.89-1.12). CONCLUSIONS: Whereas FTTA appropriately triaged patients 15 to 54 years old to the trauma center, those 55 years old and older were much less likely to be triaged correctly. The reasons for this finding remain unknown, and further studies are needed to investigate and improve elderly triage.

A multi-institutional study from the US ROPE Consortium examining factors associated with directly entering practice upon residency graduation.

BACKGROUND: There is concern regarding the competency of today's general surgery graduates as a large proportion defer independent practice in favor of additional fellowship training. Little is known about the graduates who directly enter general surgery practice and if their operative experiences during residency differ from graduates who pursue fellowship. METHODS: Nineteen Accreditation Council for Graduate Medical Education-accredited general surgery programs from the US Resident OPerative Experience Consortium were included. Demographics, career choice, and case logs from graduates between 2010 to 2020 were analyzed. RESULTS: There were 1,264 general surgery residents who graduated over the 11-year period. A total of 248 (19.6%) went directly into practice and 1,016 (80.4%) pursued fellowship. Graduates directly entering practice were more likely to be a high-volume resident (43.1% vs 30.5%, P < .01) and graduate from a high-volume program (49.2% vs 33.0%, P < .01). Direct-to-practice graduates performed 53 more cases compared with fellowship-bound graduates (1,203 vs 1,150, P < .01). On multivariable analysis, entering directly into practice was positively associated with total surgeon chief case volume (odds ratio = 1.47, 95% confidence interval 1.18-1.84, P < .01) and graduating from a US medical school (odds ratio = 2.54, 95% confidence interval 1.45-4.44, P < .01) while negatively associated with completing a dedicated research experience (odds ratio = 0.31, 95% confidence interval 0.22-0.45, P < .01). CONCLUSION: This is the first multi-institutional study exploring resident operative experience and career choice. These data suggest residents who desire immediate practice can tailor their experience with less research time and increased operative volume. These data may be helpful for programs when designing their experience for residents with different career goals.

Machine learning methods to predict mechanical ventilation and mortality in patients with COVID-19.

BACKGROUND: The Coronavirus disease 2019 (COVID-19) pandemic has affected millions of people across the globe. It is associated with a high mortality rate and has created a global crisis by straining medical resources worldwide. OBJECTIVES: To develop and validate machine-learning models for prediction of mechanical ventilation (MV) for patients presenting to emergency room and for prediction of in-hospital mortality once a patient is admitted. METHODS: Two cohorts were used for the two different aims. 1980 COVID-19 patients were enrolled for the aim of prediction ofMV. 1036 patients' data, including demographics, past smoking and drinking history, past medical history and vital signs at emergency room (ER), laboratory values, and treatments were collected for training and 674 patients were enrolled for validation using XGBoost algorithm. For the second aim to predict in-hospital mortality, 3491 hospitalized patients via ER were enrolled. CatBoost, a new gradient-boosting algorithm was applied for training and validation of the cohort. RESULTS: Older age, higher temperature, increased respiratory rate (RR) and a lower oxygen saturation (SpO2) from the first set of vital signs were associated with an increased risk of MV amongst the 1980 patients in the ER. The model had a high accuracy of 86.2% and a negative predictive value (NPV) of 87.8%. While, patients who required MV, had a higher RR, Body mass index (BMI) and longer length of stay in the hospital were the major features associated with in-hospital mortality. The second model had a high accuracy of 80% with NPV of 81.6%. CONCLUSION: Machine learning models using XGBoost and catBoost algorithms can predict need for mechanical ventilation and mortality with a very high accuracy in COVID-19 patients.

Predictors of mortality in trauma patients with intracranial hemorrhage on preinjury aspirin or clopidogrel.

BACKGROUND: The mortality risk in elderly patients who sustained head trauma resulting in intracranial hemorrhage (ICH) while taking the antiplatelet agents aspirin (ASA) or clopidogrel or both (Plavix) was evaluated. METHODS: A retrospective review identified trauma patients, age 50 or greater, who had computed tomography (CT) evidence of ICH and were taking ASA, clopidogrel, or a combination of both. Patient demographics, type of medication, mechanism of injury, Glasgow Coma Score (GCS), grading of head CT scans, and outcomes were characterized. RESULTS: One hundred nine patients including 61 men and 48 women were identified; the mean age was 77 years +/- 10 years. Injury was due to level fall (73), fall from height (21), motor vehicle crash (11), and other (4). Twenty (18%) patients died; age, gender, type of medication, and mechanism of injury were not predictive of death. The initial GCS for survivors was 14.2 +/- 1.9 versus 11.3 +/- 4.9 for nonsurvivors (p < 0.007). Deaths based on initial CT grade were: grade 1, 5 of 70; grade 2, 4 of 17; grade 3, 5 of 10; grade 4, 6 of 12 (p = 0.002). Follow-up CT scans were performed in 81 patients who were not taken to surgery and had grade 1 or 2 hemorrhage initially. Of 4 patients with hemorrhage progression, there was 1 death (25%) versus 6 deaths in 77 patients without progression (8%; p = 0.70). CONCLUSIONS: There is high mortality rate associated with ASA or clopidogrel or both in elderly patients who have head trauma resulting in ICH. The presenting GCS and initial grade of CT scan are most predictive of death. Progression of hemorrhage after admission is unusual. The risk of brain injury, particularly from falls, should be explained to elderly patients taking these medications.

Nutritional and metabolic consequences in the pediatric burn patient.

Nutritional support is one of the many important considerations when treating severe burn injury in children. The type, route, timing, amount, and monitoring of nutritional support have been widely investigated, yet many questions remain. This review will highlight the current state of knowledge and the important aspects of nutritional support in severe burn injury in children.

Spontaneous retroperitoneal hemorrhage.

BACKGROUND: We evaluated patients with spontaneous retroperitoneal hemorrhage for reliable predictors of early diagnosis and improved outcomes. METHODS: A retrospective chart review was done to determine patient demographic and laboratory findings, presenting symptoms, time to diagnosis, anticoagulant and/or antiplatelet agent use, transfusions, and patient outcome. RESULTS: One hundred nineteen patients were identified; 14 (12%) died (mean age 77 +/- 9 years vs. 74 +/- 10 years for survivors [P = 0.235]). All nonsurvivors were on anticoagulants: 8 of 89 (9%) were on heparin or warfarin alone, and 6 of 23 (26% [P = 0.028]) were on a combined anticoagulant-antiplatelet regimen. Symptom onset to computed axial tomography (CAT) scan averaged 1.3 +/- 1.3 days for nonsurvivors versus 1.5 +/- 1.9 days for survivors (P = 0.778). Hemoglobin was 9.07 +/- 3.35 for nonsurvivors versus 9.60 +/- 2.07 for survivors (P = 0.435). Eighty-eight patients were transfused, and 10 died; 31 patients had no transfusion, and 4 of these died (P = 0.821). CONCLUSIONS: A high index of clinical suspicion is necessary for diagnosis of spontaneous retroperitoneal hemorrhage because these patients present with a variety of symptoms. Prospective studies are necessary to determine whether earlier diagnosis combined with aggressive resuscitation can impact the high mortality rate seen in these patients.

Nurse driven protocol for head injured patients on warfarin.

The trauma quality improvement committee at our facility identified a significant number of patients on warfarin presenting to the emergency center after minor head trauma that subsequently expired from their intracranial hemorrhage prior to appropriate intervention. An analysis of this patient population identified multiple areas of delay. A collaborative effort between the emergency center nurses and the trauma service personnel resulted in a formal protocol to address each component of delay and expedite the process. Since implementation of this nursing driven protocol we have dramatically decreased the time to (1) Emergency Center Physician evaluation, (2) completion of head computerized tomography, (3) reversal of anticoagulation with fresh frozen plasma (FFP), and (4) most importantly, patient mortality rate. We conclude that this nursing driven protocol is effective in decreasing the mortality rate by eliminating diagnostic and therapeutic delays in this high-risk patient population.

Financial implication of proposed legislation on hospital reimbursement for the injured drunk driver.

The state of Michigan currently has no-fault automobile insurance with personal injury protection, providing anyone injured in motor vehicle collisions with unlimited medical and rehabilitation benefits and lost wage recovery. A new bill proposal, Michigan House Bill 5588, will eliminate hospital reimbursement for those who are found to be intoxicated at the time of a motor vehicle collision. These medical costs will be passed on to patients, which may result in a large reimbursement deficit for hospitals caring for these patients. This retrospective review examines the costs of caring for all intoxicated drivers who were admitted to a Level 1 trauma center after a motor vehicle collision over a 2-year period. Intoxicated drivers were younger (P = 0.0002), had a lower Glasgow Coma Scale (P = 0.0013), and were more likely to meet Level 1 trauma criteria (P = 0.0002). The sum of total charges for injured drunk drivers totaled $5.2 million. When taking into account fixed and variable costs of care, lost hospital net income would be $3 million (21.9%) over a 3-year span whether House Bill 5588 passes. In conclusion, the passage of House Bill 5588 will lead to a large financial burden for hospitals that treat intoxicated drivers.

Novel Educational Module for Subclavian Central Venous Catheter Insertion Using Real-Time Ultrasound Guidance.

OBJECTIVE: Given increasing evidence supporting a real-time ultrasound (US)-guided approach for subclavian vein (SCV) central venous catheter (CVC) insertion as compared with the traditional landmark approach, we sought to develop a standardized curriculum to offer healthcare providers a means to attain increased competency and confidence in US-guided SCV CVC insertion. DESIGN: Retrospective review of prospectively collected data. SETTING: Single institution's American College of Surgeons Level 1 Accredited Education Institute within an academic tertiary care center. SUBJECTS: A total of 77 residents and midlevel providers working in our surgical intensive care unit. INTERVENTIONS: Providers participated in a tiered educational module designed to teach safe US-guided SCV CVC insertion. The education consisted of a multimedia didactic presentation and a hands-on simulation session, including US anatomy on live subjects and anatomical model-based SCV CVC insertion. MEASUREMENTS AND MAIN RESULTS: Assessment of the effect of education included a written examination and confidence survey, administered pre- and postintervention, and videotaped simulation session graded by blinded expert evaluators. Of the 77 participants, 70 participants completed a posttest with a median 5-point increase in score compared with that of the pretest score (p < 0.0001). Confidence ratings based on a 5-point Likert scale demonstrated an increase in confidence in SCV CVC insertion (p < 0.0001), using the landmark approach (p < 0.0001), using US-guided approach (p < 0.0001), and in use of US to image the SCV (p < 0.0001). Postgraduate year-1 residents had lower mean global rating score (p = 0.010) than any other participants. CONCLUSIONS: This comprehensive hands-on teaching module-based curriculum enhanced learner knowledge of and confidence in US-guided SCV CVC insertion. This module can be implemented in simulation centers for teaching safe and successful SCV CVC insertion.

New approaches to managing congenital diaphragmatic hernia.

A number of new techniques have been studied for managing newborns with congenital diaphragmatic hernia and respiratory insufficiency. Among these have been the techniques of delayed approach to the repair of the diaphragmatic hernia; permissive hypercapnia; nitric oxide and surfactant administration; intratracheal pulmonary ventilation; liquid ventilation; perfluorocarbon-induced lung growth; and lung transplantation. These interventions are at various stages of development and evaluation of effectiveness. All, however, are being explored in the hopes of improving outcome in patients with congenital diaphragmatic hernia who continue to have significant morbidity and mortality in the newborn period.

Management of blunt splenic injury in patients with concurrent infectious mononucleosis.

Selective nonoperative management is appropriate for most blunt splenic injuries in adults and children, but the efficacy of this approach is unknown when injury occurs in patients with concurrent infectious mononucleosis. We have reviewed our experience during the past 23 years with the selective nonoperative management of blunt splenic injury in these patients. Medical record review identified nine patients with blunt splenic injury and infectious mononucleosis from 1978 to 2001, representing 3.3 per cent of our total trauma population with blunt splenic injury treated during that interval. Two patients underwent immediate splenectomy because of hemodynamic instability. Seven patients were admitted with the intent to treat nonoperatively. Five patients were successfully managed nonoperatively. Two patients failed nonoperative management and underwent splenectomy, one because of hemodynamic instability and one because of an infected splenic hematoma. Concurrent infectious mononucleosis does not preclude the successful nonoperative management of blunt splenic injury. This small subset of patients may be managed nonoperatively using the same criteria as for patients whose splenic injuries are not complicated by infectious mononucleosis.

Early colectomy may be associated with improved survival in fulminant Clostridium difficile colitis: an 8-year experience.

BACKGROUND: We hypothesized that colectomy for fulminant Clostridium difficile colitis (CDC) before organ failure would be associated with decreased mortality. METHODS: Data were retrospectively collected on patients operated on for CDC between 2000 and 2007. Variables examined included age, sex, immunodeficiency, recurrent CDC, vasopressor requirement, acute respiratory failure, acute renal failure, white blood cell count, and stress ulcer prophylaxis. Univariate and multivariate analyses were performed to identify predictors of mortality. RESULTS: During this period, 6,841 patients were diagnosed with CDC and 69 patients underwent surgery. Independent predictors of mortality were age >65 years (odds ratio [OR] 6.8, confidence interval [CI] 1.4-32.3, P = .016), acute respiratory failure (OR 5.4, CI 1.6-18.1, P = .007), and acute renal failure (OR 3.8, CI 1.1-13.1, P = .035). CONCLUSIONS: Colectomy before the development of organ failure is associated with decreased mortality in patients with fulminant CDC, especially in those >65 years old.

Practice patterns and outcomes of retrievable vena cava filters in trauma patients: an AAST multicenter study.

BACKGROUND: The purpose of this study is to describe practice patterns and outcomes of posttraumatic retrievable inferior vena caval filters (R-IVCF). METHODS: A retrospective review of R-IVCFs placed during 2004 at 21 participating centers with follow up to July 1, 2005 was performed. Primary outcomes included major complications (migration, pulmonary embolism [PE], and symptomatic caval occlusion) and reasons for failure to retrieve. RESULTS: Of 446 patients (69% male, 92% blunt trauma) receiving R-IVCFs, 76% for prophylactic indications and 79% were placed by interventional radiology. Excluding 33 deaths, 152 were Gunter-Tulip (G-T), 224 Recovery (R), and 37 Optease (Opt). Placement occurred 6 +/- 8 days after admission and retrieval at 50 +/- 61 days. Follow up after discharge (5.7 +/- 4.3 months) was reported in 51%. Only 22% of R-IVCFs were retrieved. Of 115 patients in whom retrieval was attempted, retrieval failed as a result of technical issues in 15 patients (10% of G-T, 14% of R, 27% of Opt) and because of significant residual thrombus within the filter in 10 patients (6% of G-T, 4% of R, 46% Opt). The primary reason R-IVCFs were not removed was because of loss to follow up (31%), which was sixfold higher (6% to 44%, p = 0.001) when the service placing the R-IVCF was not directly responsible for follow up. Complications did not correlate with mechanism, injury severity, service placing the R-IVCF, trauma volume, use of anticoagulation, age, or sex. Three cases of migration were recorded (all among R, 1.3%), two breakthrough PE (G-T 0.6% and R 0.4%) and six symptomatic caval occlusions (G-T 0, R 1%, Opt 11%) (p < 0.05 Opt versus both G-T and R). CONCLUSION: Most R-IVCFs are not retrieved. The service placing the R-IVCF should be responsible for follow up. The Optease was associated with the greatest incidence of residual thrombus and symptomatic caval occlusion. The practice patterns of R-IVCF placement and retrieval should be re-examined.

Treatment of trauma patients with intracranial hemorrhage on preinjury warfarin.

BACKGROUND: Preinjury warfarin anticoagulation has been shown to increase the mortality of traumatic intracranial hemorrhage. We have evaluated the impact on patient mortality of the rapid triage of patients at risk for warfarin associated traumatic intracranial hemorrhage. METHODS: A "Coumadin Protocol" was implemented in January, 2001 in the Emergency Department that expedited triage of anticoagulated trauma patients to immediate physician evaluation. Patient outcomes during a 2 year period were compared with a matched control group of similarly injured, anticoagulated patients who were treated before protocol initiation. RESULTS: Thirty-five patients were treated after implementation of the Coumadin Protocol. Mean time until warfarin reversal was 4.3 +/- 4.4 hours, and there was a 37% mortality. Twenty-two control patients had a mean time to reversal of 4.2 +/- 2.9 hours, with a 45% mortality (p = 0.610). Ten protocol patients were shown to have intracranial hemorrhage progression by computed tomography (CT) scan, with a 60% mortality rate. Seventeen patients had follow-up CT scan and showed no progression; only one of these patients (6%) died (p = 0.004). Hemorrhage severity based on the initial CT scan did not predict mortality or hemorrhagic progression. CONCLUSIONS: We conclude from these data that a trauma center protocol for rapid identification of intracranial bleeding without a concomitant therapeutic protocol does not improve survival in head injured patients on preinjury warfarin.

Rapid warfarin reversal in anticoagulated patients with traumatic intracranial hemorrhage reduces hemorrhage progression and mortality.

BACKGROUND: A prospective cohort study at our institution demonstrated a 48% mortality rate in warfarin anticoagulated trauma patients sustaining intracranial hemorrhage (ICH) compared with a 10% mortality rate in nonanticoagulated patients. Forty percent of patients demonstrated progression of their ICH, despite anticoagulation reversal, with a resultant 65% mortality rate. Seventy-one percent of these patients initially presented with a Glasgow Coma Scale (GCS) score > or = 14 and a 'minor' ICH. We postulated that early diagnosis of ICH and rapid anticoagulation reversal would reduce ICH progression rates and mortality. METHODS: All anticoagulated patients with known or suspected head trauma were entered into the Coumadin protocol. The protocol ensured immediate triage and physician evaluation, head computed tomography (CT) scan, and fresh frozen plasma administration in patients with documented ICH. RESULTS: Eighty-two patients were entered into the protocol with ICH documented in 19 (23%). Sixteen of 19 patients (84%) presented with GCS > or = 14. Median international normalized ratio (INR) for treated patients with ICH was 2.7 versus 2.5 for patients without ICH (p = 0.546). Mean time to initiate warfarin reversal was 1.9 hours for protocol patients versus 4.3 hours for preprotocol patients (p < 0.001). Two of 19 (10%) protocol patients with ICH died. However, both patients presented >10 hours after injury with a severe ICH. This 10% mortality rate is significantly less than the 48% mortality rate seen previously (p < 0.001) and is now consistent with that observed in similarly injured patients not on anticoagulation. CONCLUSION: Neither the initial GCS nor INR in anticoagulated trauma patients reliably identifies patients with ICH. Rapid confirmation of ICH with expedited head CT scan combined with prompt reversal of warfarin anticoagulation with fresh frozen plasma decreases ICH progression and reduces mortality.

Development of an artificial placenta: CO2 elimination and hemodynamics as a function of arteriovenous blood flow.

BACKGROUND: As a first step toward the development of an artificial placenta, we investigated the relationship between blood flow rate through an arteriovenous (A-V) circuit/oxygenator and both CO2 elimination and hemodynamic stability in a small animal model. METHODS: Male New Zealand rabbits (N = 10) with an average weight of 2.7 +/- 0.2 kg were anesthetized, paralyzed, and heparinized before carotid-jugular cannulation. A tracheostomy tube, an arterial catheter, and an aortic flow probe were placed. Arteriovenous flow through a custom-made, low-resistance, 0.5 m2 hollow fiber oxygenator was initiated. Oxygen sweep flow was maintained at 300 mL/min, whereas blood flow was controlled at 10 to 40 mL/(kg min). Ventilation was discontinued during each blood flow rate trial. Hemodynamic and preoxygenator and postoxygenator blood gas data were recorded 30 minutes after initiation of each flow rate. CO2 removal was the product of the oxygen sweep gas flow rate and the sweep flow exhaust CO2 content as determined by capnometry. Data were analyzed by analysis of variance with post hoc Dunnett's t test. RESULTS: CO2 removal increased and PaCO2 decreased as a function of A-V blood flow rate. Simultaneously, systolic blood pressure did not significantly change. CO2 removal was effective at device flows greater than 20% of cardiac output. CONCLUSION: In this rabbit model, A-V blood flows at 25% to 30% of cardiac output allow full gas exchange without hemodynamic compromise. This model raises the possibility of using A-V support and an artificial placenta in newborns with respiratory failure.