ETA2022 Poster Presentations COVID & Thyroid Disease (9 abstracts)
1Universitair Ziekenhuis Brussel, Diabetologie & Endocrinologie, Jette, Belgium; 2Vrije Universiteit Brussel (Vub), Research Group Biostatistics and Medical Informatics, Brussel, Belgium
Background: Subacute thyroiditis (SAT) is a destructive thyroiditis which is presumably caused by a viral infections or post-viral inflammatory response. Thyroid dysfunction evolves through a set of stages (hyperthyroidism - hypothyroidism - euthyroidism) and is usually temporary, although some patients may develop permanent hypothyroidism. The risk factors for permanent hypothyroidism remain unclear.
Methods: A retrospective analysis of patients with SAT of the University Hospital of Brussels from 2001 to 2020. Firstly, a description of the patient characteristics, inflammatory and thyroid parameters is provided. Secondly, a predictive model for the need to initiate thyroid hormone therapy (THT) is developed by means of a logistic regression analysis. The optimal model configuration was selected by a forward sequential analysis maximizing the classification accuracy while the prediction performance was validated by a leave-one-out cross validation.
Results: 35 patients were included, with a female predominance (ratio 4.8/1). Thyrotoxicosis was detected in the majority of patients (91%), while hypothyroidism developed in 71% (of which 27% subclinical). Thyroid replacement therapy was initiated in approximately half of patients (18/35). The discontinuation of THT was attempted in 10/18 patients, and was successful in most of these (8/10). The logistic regression model selected the predictors Age, Season and Regimen, and demonstrated a maximum accuracy of 86%, classifying 30 of 35 patients correctly in the outcome measure for what the model fit is concerned, while during a leave-one-out cross validation the accuracy is 74% predicting 28 of 35 patients correctly.
Conclusion: THT was initiated for hypothyroidism in half of the patients, which is higher than previously reported. A scoring model, applicable in clinical practice to identify patients who may benefit from THT, was developed.