To explore the risk factors for developing lower extremity deep vein thrombosis (DVT) after open wedge high tibial osteotomy (OWHTO) using patient medical data and construct a predictive line graph model.

Clinical data were retrospectively collected from patients who underwent OWHTO for medial compartment knee osteoarthritis with inversion deformity at the Department of Orthopedics, 3rd Hospital of Hebei Medical University between May 2022 and May 2023. The population was divided into a DVT group and a non-DVT group by the presence or absence of DVT after surgery. Risk factors independently associated with the occurrence of DVT after OWHTO were identified by univariate and multifactorial logistic regression analysis, and predictive nomogram were constructed based on the results. The predictive efficacy of the model was further evaluated and validated using the area under the subject operating curve AUC, Poe's score, C index, Bootstrap 500 replicate sampling validation, and calibration curve and decision curve analysis.

372 patients were included in the study. Multifactorial logistic regression showed that a propped height ≥13 mm, high levels of fasting glucose, low levels of ALB, and D-dimer >1.07 mg/L were identified as independently associated with a high risk of developing DVT after OWHTO. The predictive nomogram constructed from the results of the multifactorial analysis performed well with an AUC of 0.775, a Brier score of 0.031, and a C-index of 0.775, which were validated by Bootstrap 500 replicate sampling with a corrected Brier score and C-index of 0.747 and 0.033, respectively, in addition to calibration curves showing that the model's predicted probability and the actual probability of DVT occurrence between with good agreement. The decision curve analysis showed that the nomogram model could guide clinical practice well.

The propped height ≥13 mm, high fasting glucose levels, low ALB levels, and D-dimer >1.07 mg/L were independent risk factors for postoperative DVT after OWHTO, and the nomogram model constructed accordingly had a good predictive performance.