Purpose: To investigate whether adrenal gland radiodensities alone or compared to the inferior vena cava (IVC) can correctly predict hospital mortality in patients in intensive care.
Methods: One hundred thirteen intensive care patients (76 males, age: 67.2 ± 14.0 years) with an acute clinical deterioration were included in this retrospective analysis. For the venous and the arterial phase CT attenuation (Hounsfield units) of adrenal glands and IVC was ROI-based evaluated by two radiologists separately. ROC analysis, combined with the Matthews Correlation Coefficient (MCC) as a classifier, was used to assess whether one of the parameters is suitable for predicting short and medium-term mortality and, if so, which parameter is most appropriate. Interrater agreement was assessed using the intraclass correlation coefficient.
Results: Twenty-one patients (18.6%) died within three days in the ICU. Measurements of the adrenal glands in the portal venous phase yielded the highest discriminative power (=AUC) to distinguish between deceased and survivors. A threshold ratio of >95.5 predicted 72-hour mortality with a sensitivity of 76.19% and a specificity of 92.39% (AUC = 0.84; p < 0.0001). The positive likelihood ratio was 10.1; the positive predictive value was 69%. The predictive power for 24-hour mortality was slightly lower. Venous adrenal-to-IVC ratios and arterial measurements as a whole were substantially less suitable. All intraclass correlation coefficients indicated a high interrater agreement.
Conclusions: In the portal venous phase, hyperattenuating of the adrenal glands on contrast-enhanced CT can predict short and intermediate ICU mortality quite well and may serve as a reproducible prognostic marker for individual patient outcomes.
Keywords: Adrenal enhancement; Hospital mortality; Mortality prediction, intensive care units; Prognosis; ROC curve.
Copyright © 2020 Elsevier Inc. All rights reserved.