E-submission
KOTA
Search
- Page Path
- HOME > Search
Original Article
- Risk factors of surgical complications after use of the femoral neck system: a random forest analysis
- Chul-Ho Kim, Hyun-Chul Shon, Han Soul Kim, Ji Wan Kim, Eic Ju Lim
- J Musculoskelet Trauma 2025;38(3):160-167. Published online July 23, 2025
- DOI: https://doi.org/10.12671/jmt.2025.00157
-
Abstract
PDF - Background
The femoral neck system (FNS), a novel fixation device for managing femoral neck fractures (FNFs), has gained popularity in recent years. However, analyses of the surgical complications and reoperation risks associated with the use of FNS remain limited.
Methods
This retrospective observational study analyzed 57 patients who had undergone FNS fixation for FNF at two university hospitals between July 2019 and February 2024. Demographic, perioperative, and outcome variables, including age, sex, fracture classification (Garden, Pauwels, and AO), implant characteristics, tip-apex distance (TAD), neck shortening, and neck-shaft alignment, were analyzed. In addition to univariate analysis, a machine learning analysis was conducted using a random forest classifier with stratified sampling (80% training, 20% testing). The accuracy, precision, recall, F1-score, and area under the receiver’s operating curve were calculated to assess model performance.
Results
Ten patients experienced osteonecrosis of the femoral head (n=6), implant cut-out or penetration (n=3), and peri-implant fracture (n=1). Univariate analysis revealed that the TAD in the complication group was significantly shorter than that in the control group (12.1 vs. 16.7 mm; P=0.012). Additionally, neck shortening in the complication group was greater than that in the control group (4.9 vs. 2.3 mm; P=0.011). The random forest model achieved an accuracy of 83.3% and identified postoperative neck-shaft angle (NSA) as the most important predictor of complications (feature importance, 0.161), followed by bolt length (0.102) and preoperative NSA (0.094).
Conclusions
Risk factor analysis conducted using a random forest model identified postoperative NSA as the most important feature associated with postoperative complications following FNS. Therefore, care should be taken to normalize the postoperative NSA during FNF surgery. Level of Evidence: III.
- 520 View
- 24 Download
First
Prev
