Development of a Clinical Risk Score for Vasoplegic Shock After Cardiopulmonary Bypass

T. J. Miles¹, M. T.. Guinn¹, O. Suero², T. K. Rosengart³, J. S. Coselli², R. K. Ghanta³, S. Chatterjee^2
1Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, Texas;, Houston, Texas ²Baylor College of Medicine, Houston, Texas ³Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, Texas, Houston, Texas

Purpose: Vasoplegia is a morbid complication after cardiopulmonary bypass. Identification of patients at risk for vasoplegia may improve outcomes but current assessment tools are limited by highly heterogenous and static criteria for defining vasoplegia. We developed a risk score derived from dynamic post-operative data that stratifies patient risk for vasoplegia.

Methods: Hemodynamic and medication data were extracted from the electronic medical record (EMR) for patients who underwent surgery with cardiopulmonary bypass at Baylor St. Luke’s Medical Center from 2017-2022. Transplant patients, mechanical circulatory support patients, and patients without pulmonary artery catheter data were excluded resulting in a final cohort of 2052 patients. Vasoplegia was identified using EMR data for each 60-minute interval over the first 24 post-operative hours. Criteria for vasoplegia included hourly average mean arterial pressure ≤ 65 mm Hg, cardiac index ≥2.2 L/min/m2, norepinephrine equivalents ≥0.1 µg/kg/min, and central venous pressure ≤15 mm Hg for at least one hour post-operatively. After merging EMR data with institutional Society of Thoracic Surgery (STS) registry data, statistically significant variables were included in a multivariate logistic regression model. Regression coefficients were used to assign integer points to the risk score. Score performance was assessed using area under the receiver operating characteristic curve (AUROC).

Results: The overall incidence of vasoplegia in the cohort was 16.7% [n=342]. The median [interquartile range] duration of vasoplegia was 3 [1-4] hours. Vasoplegic patients demonstrated significantly higher in-hospital mortality (6.4% [n=22] vs. 3.0% [n=52], p< 0.001) and major morbidity (51.5% [n=176] vs. 23.4% [n=400], p< 0.001). On multivariate logistic regression, older age (odds ratio [95% confidence interval]: 1.20 per decade [1.08 – 1.35]), male gender (1.41 [1.07-1.86]), pre-operative dialysis (1.68 [1.12-2.53]), liver disease (1.73[1.08-2.76]), heart failure (1.38[1.07-1.79), prior cardiac surgery (1.63[1.18-2.25]), pre-operative amiodarone (1.72[1.18-2.50]), pre-operative beta blockers (0.69[0.51-0.95]), non-elective surgery (1.51 [1.15-1.98]), CAB (1.55[1.12-2.14], CAB-Valve (3.21[2.10-4.90]), and cardiopulmonary bypass time (1.22 per hour [1.08-1.37] were independently associated with vasoplegia. A risk score, ranging from 0 to 15, consisting of the independent predictors of vasoplegia was developed. The median [IQR] risk score was 4 [3-6]. Patients were divided into risk categories according to total score. The probability of vasoplegia was 3.9%, 16.7%, and 34.2% in patients with low ( < 3), moderate (3-6), and high (>6) scores respectively . The AUROC of the risk score on internal validation was 0.70 and 0.68 in predicting vasoplegia and prolonged vasoplegia (>3 hours) respectively. The sensitivity and specificity of the risk score was 0.71 and 0.56 respectively.

Conclusion: Integrating dynamic post-operative EMR data with STS registry data allows for identification of risk factors for vasoplegia. A risk score incorporating these risk factors predicts vasoplegia with fair accuracy on internal validation. Our risk score may mitigate or even prevent vasoplegia by identifying high-risk patients and thereby expediting clinical intervention.

Identify the source of the funding for this research project: This work was supported by the National Institutes of Health (NIH), National Heart Lung and Blood Institute [T32HL139430].