Background: The impact of hip fracture on older adults is significant, including increased mortality, reduced activity levels and abilities and reduced quality of life。 Hip fractures often occur in the elderly and increase the risk of death. The purpose of this study is to analyze the risk factors associated with 28-day mortality in elderly patients with severe hip fractures using two models, XG Boost and multivariate logistic regression, and to compare the predictive value of the two models.
Methods: MIMIC database is a powerful tool to provide clinical data to clinical researchers. The database was established in 2003 with funding from the National Institutes of Health by the Computational Physiology Laboratory at the Massachusetts Institute of Technology, Beth Israel Deaconess Medical Center (BIDMC) at Harvard Medical School, and Philips Medical. Patients with severe hip fractures in the elderly were included based on the MIMIC-IV database and were divided into a death group and a survival group based on the death 28 days after admission to the ICU. Baseline data differences between the two groups of patients were compared, and risk factors associated with 28-day mortality in severe elderly patients with hip fractures were analyzed using XG Boost and multivariate logistic regression models, respectively. The predictive power of the two models was compared using receiver operation characteristics curves.
Results: 287 elderly patients with severe hip fractures were included, including 43 cases (15.0%) in the death group and 244 cases (85.0%) in the survival group. Logistic regression analysis showed that advanced age, male, congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), high sepsis-related organ failure (SOFA) score, high heart rate, high white blood cell count, high creatinine, high mean arterial pressure, and high hemoglobin levels were associated with 28-day mortality after admission to the ICU, while the higher the mean arterial pressure and the hemoglobin level, the lower the risk of death. Although the rate of using mechanical ventilation and receiving blood transfusion in the death group was higher than that in the survival group, neither of them reached statistical significance. The XG Boost model shows that the top 5 factors associated with 28-day mortality are Sequential organ failure score (SOFA) score (31 points), chronic heart failure (20 points), chronic structural pulmonary disease (18 points), age (17 points), and male (15 points). The higher the mean arterial pressure and the hemoglobin level, the lower the risk of death. The area under the ROC curve predicted by the multivariate logistic regression model for mortality risk was 0.729 (95% CI: 0.701-0.783), and the Jordan index was 0.412. The area under the ROC curve predicted by the XG Boost model for mortality risk was 0.804 (95% CI: 0.720-0.837), and the Jordan index was 0.492.
Conclusion: The ability of the XG Boost model to predict the 28-day mortality risk in elderly patients with severe hip fractures is better than the multivariate logistic regression model, which will help healthcare professionals provide more support for elderly patients with hip fracture.