BackgroundNon-thyroidal illness (NTI) is frequent in hospitalized patients. Its recovery is characterized by a raise in TSH levels. However, the clinical significance of high TSH levels at admission in hospitalized elderly patients with NTI remains uncertain.AimTo explore the relevance of baseline TSH evaluation in hospitalized elderly patients with NTI.MethodsWe examined the participants with NTI (n = 123) from our previous study (Sforza, 2017). NTI was defined as: low T3 (< 80 ng/dL) and normal or low total T4 in the presence of TSH values between 0.1 and 6.0 mU/L. Thyroid function tests were performed on day 1 and day 8 of the hospital stay. Positive TSH changes (+ ΔTSH) were considered when the day-8 TSH value increased more than the reference change value for TSH (+ 78%). Multiple logistic regression was used to evaluate the independent association of baseline TSH, sex, clinical comorbidities (by ACE-27) and medications with + ΔTSH.ResultsOut of 123 patients (77 ± 8 years, 52% female), 34 showed a + ΔTSH. These patients had a lower TSH at admission (p < 0.001) and intra-hospital mortality (p = 0.003) than the others. In multiple logistic regression, TSH > 2.11 mU/L at baseline was associated with reduced odds to show + ΔTSH [odds ratio (95 CI) 0.29 (0.11–0.75); p = 0.011] in a model adjusted by age, sex and ACE-27.DiscussionInappropriately higher TSH levels at admission in hospitalized elderly patients were associated with a reduced ability to raise their TSH levels later on. The present results confront the idea that TSH levels at admission are irrelevant in this clinical context.