303 - Physiological Changes Due to Bloodstream Infection in Intensive Care Unit Patients Differ According to Transplant Status

Background: Transplant recipients are at increased risk of bloodstream infection (BSI), which often leads to critical illness. Due to immunosuppression, BSI in these patients may manifest with different pathophysiology compared to non-transplant recipients. We aimed to identify different trends in the pathophysiology of critically ill patients with BSI based on transplant status.

Methods: We reviewed data from patients admitted to the medical and surgical/trauma intensive care units (ICUs) at the University of Virginia Medical Center from 2011 to 2015. We included both solid organ and hematopoietic stem cell transplant recipients. We performed univariate logistic regression modeling to evaluate trends in different physiological features in both transplant and non-transplant recipients in the 96 hours surrounding a positive blood culture. We then performed multivariate logistic regression modeling to identify features independently associated with a positive blood culture in the next 24 hours in transplant recipients.

Results: We analyzed 9,954 ICU patient-admissions (including 505 transplant recipients), with a total of 144 patient-years of physiological data, 1.3 million hourly measurements, and 15,577 blood culture instances. Of the 1,068 blood culture instances in transplant recipients, 125 (12%) were positive, compared to 1,051 of 14,509 (7%) blood culture instances in non-transplant recipients. Critically ill transplant recipients with BSI had greater abnormalities in vital signs, oxygen requirement, markers of organ damage, APACHE score, and Charlson Comorbidity Index (CCI) compared to non-transplant recipients (Figure 1). Trends in many of these features also differed based on transplant status. The multivariable logistic regression model of BSI in transplant recipients included, in decreasing strength of association: total bilirubin, systolic blood pressure, fraction of inspired oxygen, number of intravenous lines, and CCI. This model had an AUC of 0.75.Figure 1. Trends in pathophysiological abnormalities in 9,954 critically ill patients with BSI based on transplant status, 2011-2015. Each graph demonstrates the average value of the physiological variable over time relative to the acquisition of a positive blood culture. Blue curves depict trends in transplant recipients, while red curves depict trends in non-transplant recipients. We assessed 108 physiological features and show the 24 features with the greatest change around the time of blood culture.


Conclusion: Critically ill transplant recipients have a higher prevalence of BSI and different pathophysiological manifestations of BSI compared to non-transplant recipients. This may have implications regarding early detection and treatment of BSI in these patients.