A Prediction model for Pulmonary Embolism in Patients with Spontaneous Intracerebral Hemorrhage

Background and PurposeSpontaneous intracerebral hemorrhage (sICH) patients were susceptible to pulmonary embolism (PE), which is frequently neglected due to the absence of predictive tools for early detection. This study aimed to investigate the risk factors of PE in sICH patients and develop a efficient model for the prediction of PE in clinical practice. Materials and methodsWe conducted a retrospective study involving 1129 sICH patients. The enrolled patients were divided into training set, internal validation set and external set (n=525: 357: 247). The univariate and multivariate stepwise logistic regression analyses were employed to screen the independent risk factors of the PE in sICH patients. A nomogram model (Model P-P) was constructed based on R language and subsequently validated. For further evaluation, the Model P-P was compared with another similar model (Model A). ResultsThe analyses revealed that deep vein thrombosis, age, Glasgow coma scale score, fibrinogen degradation product, D-dimer, hemoglobin, hemorrhage volume, plasma osmolality, and surgical method were significant risk factors for PE in sICH patients (p<0.05). The Model P-P was established based on these factors, and its sensitivity, specificity were 84.2%, 94% and 0.910, respectively. Furthermore, the AUC of Model P-P (0.894) was significantly higher than that of Model A (0.785) (p<0.05). ConclusionThe Model P-P developed in this study exhibited reliable predictive efficiency, making it an applicable tool for the early evaluation and prediction of PE after sICH surgery. This model is potential to assist personalized treatment decisions in clinical practice, especially in the neurosurgical intensive care unit.