Relationship between thyroid function and ICU mortality: a prospective observation study

INTRODUCTION Although nonthyroidal illness syndrome is considered to be associated with adverse outcome in ICU patients, the performance of thyroid hormone levels in predicting clinical outcome in ICU patients is unimpressive. This study was conducted to assess the prognostic value of the complete thyroid indicators (free triiodothyronine (FT3), total triiodothyronine; free thyroxine, total thyroxine, thyroid-stimulating hormone and reverse triiodothyronine) in unselected ICU patients. METHODS A total of 480 consecutive patients without known thyroid diseases were screened for eligibility and followed up during their ICU stay. We collected each patient's baseline characteristics, including the Acute Physiology and Chronic Health Evaluation II (APACHE II) score and thyroid hormone, N-terminal pro-brain natriuretic peptide (NT-proBNP) and C-reactive protein (CRP) levels. The primary outcome was ICU mortality. Potential predictors were analyzed for possible association with outcomes. We also evaluated the ability of thyroid hormones together with APACHE II score to predict ICU mortality by calculation of net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indices. RESULTS Among the thyroid hormone indicators, FT3 had the greatest power to predict ICU mortality, as suggested by the largest area under the curve (AUC) of 0.762±0.028. The AUC for FT3 level was less than that for APACHE II score (0.829±0.022) but greater than that for NT-proBNP level (0.724±0.030) or CRP level (0.689±0.030). Multiple regression analysis revealed that FT3 level (standardized β=-0.600, P=0.001), APACHE II score (standardized β=0.912, P<0.001), NT-proBNP level (standardized β=0.459, P=0.017) and CRP level (standardized β=0.367, P=0.030) could independently predict primary outcome. The addition of FT3 level to APACHE II score gave an NRI of 54.29% (P<0.001) and an IDI of 36.54% (P<0.001). The level of FT3 was significantly correlated with NT-proBNP levels (r=-0.344, P<0.001) and CRP levels (r=-0.408, P<0.001). CONCLUSION In unselected ICU patients, FT3 was the most powerful and only independent predictor of ICU mortality among the complete indicators. The addition of FT3 level to the APACHE II score could significantly improve the ability to predict ICU mortality.