Predictive Values of Preoperative Characteristics for 30-Day Mortality in Traumatic Hip Fracture Patients

Abstract

Hip fracture patients have a high risk of mortality after surgery, with 30-day postoperative rates as high as 10%. This study aimed to explore the predictive ability of preoperative characteristics in traumatic hip fracture patients as they relate to 30-day postoperative mortality using readily available variables in clinical practice. All adult patients who underwent primary emergency hip fracture surgery in Sweden between 2008 and 2017 were included in the analysis. Associations between the possible predictors and 30-day mortality was performed using a multivariate logistic regression (LR) model; the bidirectional stepwise method was used for variable selection. An LR model and convolutional neural network (CNN) were then fitted for prediction. The relative importance of individual predictors was evaluated using the permutation importance and Gini importance. A total of 134,915 traumatic hip fracture patients were included in the study. The CNN and LR models displayed an acceptable predictive ability for predicting 30-day postoperative mortality using a test dataset, displaying an area under the ROC curve (AUC) of as high as 0.76. The variables with the highest importance in prediction were age, sex, hypertension, dementia, American Society of Anesthesiologists (ASA) classification, and the Revised Cardiac Risk Index (RCRI). Both the CNN and LR models achieved an acceptable performance in identifying patients at risk of mortality 30 days after hip fracture surgery. The most important variables for prediction, based on the variables used in the current study are age, hypertension, dementia, sex, ASA classification, and RCRI.

Predicting 1-Year Mortality after Hip Fracture Surgery: An Evaluation of Multiple Machine Learning Approaches

“…Of the variables analyzed, metastatic carcinoma, ASA classification, sex, CCI ≤ 4, age, dementia, congestive heart failure, hypertension, surgery using pins/screws, and chronic kidney disease were the most important for predicting mortality 1 year postoperatively. In contrast, a previous study from our group investigating 30-day postoperative mortality found that the most important predictors were age, hypertension, dementia, sex, ASA classification, RCRI, CCI, congestive heart failure, non-displaced cervical hip fractures, and cerebrovascular disease [ 8 ]. There was a significant overlap, with age, sex, ASA classification, CCI, dementia, congestive heart failure and hypertension being important for predicting both 30-day and 1-year mortality [ 8 ].…”

“…In contrast, a previous study from our group investigating 30-day postoperative mortality found that the most important predictors were age, hypertension, dementia, sex, ASA classification, RCRI, CCI, congestive heart failure, non-displaced cervical hip fractures, and cerebrovascular disease [ 8 ]. There was a significant overlap, with age, sex, ASA classification, CCI, dementia, congestive heart failure and hypertension being important for predicting both 30-day and 1-year mortality [ 8 ]. The increased importance of metastatic carcinoma and chronic kidney disease for predicting 1-year mortality is understandable as they are both chronic conditions with systemic, long-term effects; however, the remaining differences require further discussion.…”

“…Of particular note, the RCRI had previously demonstrated superiority over the CCI in predicting 30-day mortality [ 8 ]. However, while the CCI remained valuable for predicting 1-year mortality, the RCRI was significantly less useful.…”

“…As such, it is important to delineate other factors, which could be modifiable in the perioperative and postoperative period, that could aid in achieving better outcomes beyond the immediate postoperative period. For instance, while the RCRI does not excel at predicting 1-year mortality, it has been found to be predictive of, and associated with, worse short-term outcomes after emergency surgery [ 1 , 8 , 53 , 54 ]. This variable is readily available at the time of admission and for those with an increased cardiac risk, higher levels of care and cardioprotective medication such as beta-blockers are recommended by American Heart Association [ 55 ].…”

“…There is a potential utility in outcome prediction, particularly in better matching perioperative resources to this population’s needs. The logistic regression (LR) method of prediction is well-known to researchers and surgeons, and our recent work demonstrated that convolutional neural networks do not outperform LR in predicting 30-day postoperative mortality in hip fracture patients [ 8 ]. There are, however, other viable alternatives that should be explored.…”