Feasibility assessment of a multiple logistic regression model utilizing serum FT3, FT4, and TSH to predict prognosis of pulmonary infection in elderly patients.

BACKGROUND: Numerous studies have established a strong link between thyroid hormones and thyroid-stimulating hormone (TSH) and their prognostic significance in critical illness. This study aimed to evaluate whether serum FT3, FT4, and TSH levels are independently associated with prognosis in elderly patients with pulmonary infection and to determine whether a combined logistic regression model improves prognostic accuracy. METHODS: An observational study was conducted on 362 elderly patients hospitalized for pulmonary infection. The patients were categorized into two groups based on their prognosis: a mortality group (35 cases) and a survival group (327 cases). Serum levels of FT3, FT4, and TSH were measured, and their associations with prognosis were evaluated. Multiple logistic regression analysis was performed to determine the independent contributions of each hormone. Receiver operating characteristic (ROC) analysis was utilized to establish optimal thresholds and assess the predictive power of the hormones both individually and as a combined prediction model. RESULTS: The study revealed significant positive correlations between serum levels of FT3, FT4, and TSH and the prognosis of pulmonary infection. Each hormone exhibited a favorable area under the curve (AUC) value in the ROC analysis, suggesting their potential as prognostic markers. Optimal thresholds for FT3, FT4, and TSH were determined, providing cutoff values for risk assessment. Moreover, multiple logistic regression analysis demonstrated that all three hormones independently contributed to the prognostic significance. Integrating FT3, FT4, and TSH levels in a combined prediction model improved the accuracy of prognosis. CONCLUSIONS: Elevated serum levels of FT3, FT4, and TSH are associated with a more favorable prognosis in elderly patients after pulmonary infection. These thyroid hormones hold promising prognostic value and can serve as potential markers for risk assessment.