RESUMO
We aimed to determine prognostic indicators of PE patients with hemodynamic decompensation admitted to the ICU. PE patients with hemodynamic decompensation at ICU admission from Medical Information Mart for Intensive Care IV database were included. Least absolute shrinkage and selection operator with 2 specific lambdas were performed to reduce the dimension of variables after univariate analysis. Then we conducted multivariate logistic regression analysis and 2 models were built. A total of 548 patients were included, among whom 187 died. Lactate, creatine-kinase MB, troponin-T were significantly higher in death group. Eight common factors were screened out from first model statistically mostly in consistent with second model: older age, decreased hemoglobin, elevated anion gap, elevated International Standard Ratio (INR), elevated respiratory rate, decreased temperature, decreased blood oxygen saturation (SpO2) and the onset of cardiac arrest were significantly risk factors for in-Hospital mortality. The nonlinear relationships between these indicators and mortality were showed by the restricted cubic spline and cutoff values were determined. Our study demonstrated that age, hemoglobin levels, anion gap levels, INR, respiratory rate, temperature, SpO2 levels, the onset of cardiac arrest could be applied to predict mortality of PE patients with hemodynamic decompensation at ICU admission.
