Sepsis is a systemic inflammatory response syndrome with high mortality. There is an upward trend in sepsis prevalence and mortality worldwide. Early and accurate prediction of outcome in sepsis is important. There remains a great need to improve a reliable prognostic model for sepsis patients with widely available variables. The aim of this study was to explore the correlation between serum thyroid hormone levels and prognosis in sepsis patients.
Septic patients were identified from the Medical Information Mart for Intensive Care (MIMIC)-III database. Factors that were found to contribute to the outcome in the uni-variate analysis at P value <0.1 were included in the multivariate. Multivariate analysis was performed by binary logistic regression analysis, which allows adjust for confounding factors. We combined an assessment of thyroid hormone and some variables together, which improve the accurate prediction of outcome. The accuracy of the test was assessed measuring the area under the ROC curve (AUROC).
A total of 929 eligible septic patients were included in the data analysis. Seventy hundred and three patients had a good functional outcome, whereas 226 patients had a bad functional outcome. Thyroxin (T) level was significantly decreased in patients with an unfavorable functional outcome as compared to patients with a favorable functional outcome (P < 0.01). Binary logistic regression analyses revealed that lower thyroxin concentrations on admission were associated with a risk for poor outcomes (OR 0.556, 95% CI 0.41-0.75; P < 0.01). In addition, in ROC curve analysis, the combined model AUROC was 0.82 for ICU survival, which was significantly higher than the AUROCs of original fT (0.65 and 0.65), T (0.71 and 0.71) and SAPSII (0.70 and 0.72) (all P < 0.05).
Low serum thyroxin levels can be a predictive marker of short-term outcome after sepsis. A combined model (fT, T and SAPSII score) can add significant additional predictive information to the clinical score of the SAPSII.
Copyright © 2021 The Authors. Published by Elsevier Inc. All rights reserved.
About The Expert
Yiping Wang
Fangyuan Sun
Guangliang Hong
Zhongqiu Lu
References
PubMed