Constructing a Risk Predictive Model of Fractures by Falls for Elderly: A Retrospective Study Focus on Elderly Hospital Inpatient

Objective The purpose of this study is to create a nomogram to evaluate the risk of fractures by falls in elderly hospital inpatients. Methods The data of elderly patients who had sustained a fall were accessed from the hospital's adverse event reporting system and electronic patient records between January 2022 and April 2024. The collected data included general information, clinical data, laboratory examination results, and imaging findings. The Least Absolute Shrinkage and Selection Operator (LASSO) regression model and multivariate logistic regression analysis were conducted to develop a risk-predictive model for fractures. The C-index was used for the internal validation of the model. Results 103 patients > 55 years who had sustained a fall were identified and their mean age was 76.98 ± 7.917 years. The occurrence of fractures was 21.4% (22 of 103). The risk prediction nomogram for fractures was developed with 4 prognostic factors which included fall time (00:01–08:00, P = 0.04), gender (female, P = 0.02), serum potassium (> 5.5mmol/L, P = 0.003), serum calcium (1.97-2.11mmol/L, P = 0.001). The calibration results and the C-index values (0.87; 95% confidence interval: 0.82296–0.91704) showed that the nomogram was very reliable. Conclusion The prediction nomogram we developed is a simple and accurate tool for the early prediction risk of fractures by falls in elderly hospital inpatients, allowing for the timely initiation of appropriate preventive measures.