DEVELOPMENT OF A SURGICAL DIFFICULTY SCORE FOR OPEN REDUCTION INTERNAL FIXATION OF PILON FRACTURES

Definitive surgical management of pilon fractures varies from relatively straightforward to extremely challenging. Objective assessment and prediction of surgical difficulty would be useful to guide orthopedic surgeons with management or referral, help accurately schedule and bill for the required operative time, and track surgeon and institutional performance. Integrating machine learning (ML) into these predictions will allow for robust models and broaden their applicability. The purpose of this research is to 1) identify specific injury characteristics that contribute to surgical complexity in pilon fracture management, and 2) develop a ML Pilon Surgical Difficulty Score (PSDS) based on these factors. A retrospective cohort study of 100 pilon fractures managed definitively with operative fixation was conducted at an academic Level 1 Trauma Center. The following patient and injury characteristics were assessed: age, body mass index (BMI), osteoporosis, open fracture type, AO/OTA code, articular comminution, presence and location of articular impaction, articular displacement, metaphyseal comminution, diaphyseal extension, fibula fracture, syndesmosis injury, and delayed delay to surgery (days). Surgical difficulty for each case was measured using two outcomes: 1) perceived difficulty and 2) operative time. Perceived difficulty was determined by taking the average of 12 fellowship trained traumatologist grading on the perceived case difficulty from an ordinal scale from one to ten. Univariate analysis was performed to identify significant predictors of difficulty for each outcome. ML models including linear regression, random forest, and neural network were used to develop various PSDSs using recursive feature elimination and evaluated for predictive accuracy by five-fold cross-validation. The cohort included 8 43A, 31 43B, and 61 43C type fractures. The mean perceived difficulty was 5.06 (range: 1.50-9.25) and mean operative time was 4.98 (range: 1.45-11.07 hours). Significant predictors of perceived difficulty included AO/OTA classification (p=0.01) and delay to surgery (p There is a large range in the surgical difficulty and operative time required to perform definitive fixation for pilon fractures. A large proportion of the variance is predictable based on the severity of the fracture pattern and delay to definitive fixation. This project used machine learning to generate accurate PSDSs for both difficulty and operative time. Future work should aim to clinically validate these PSDSs so they may be used to accurately schedule patients, improve institutional performance and improve patient outcomes.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.