Prediction of 90-Day Local Complications in Patients After Total Knee Arthroplasty: A Nomogram With External Validation

BACKGROUND: Local complications after total knee arthroplasty (TKA) significantly affect the patient’s prognosis. Nomograms can be a useful tool for predicting such complications. PURPOSE: To compare the preoperative and intraoperative factors of patients who underwent TKA with and without complications and to construct and validate a nomogram based on selective predictors of local complications within 90 days postoperatively. STUDY DESIGN: Case-control study; Level of evidence, 3. METHODS: The nomogram was developed in a primary cohort that consisted of 410 patients who underwent primary TKA at the authors’ institution between January 2015 and September 2018. Predictor variables included 4 major local complications that can occur within 90 days: reoperation (including implant revision or removal for any reason and manipulation under anesthesia), infection, bleeding requiring ≥4 unit transfusion of red blood cells within 72 hours of surgery, and peripheral nerve injury. The authors used least absolute shrinkage and selection operator (LASSO) regression analysis for data dimension reduction and feature selection. Multivariable logistic regression analysis was used to develop the nomogram. Performance of the nomogram was assessed using C-index, calibration plot, area under the receiver operating characteristic curve (AUC), and decision curve analysis (DCA). The model was subjected to bootstrap validation and external validation using a prospective cohort of 249 patients. RESULTS: Four significantly prognostic factors were incorporated into the nomogram: age-adjusted Charlson Comorbidity Index, American Society of Anesthesiologists score, tourniquet time, and estimated intraoperative blood loss. The model displayed good discrimination, with a C-index of 0.819 and an AUC of 0.819. The calibration curves showed optimal agreement between nomogram prediction and actual observation. A high C-index value of 0.801 could still be reached in bootstrap validation. Application of the nomogram in the validation cohort showed good discrimination (C-index, 0.731) and good calibration. DCA demonstrated that the nomogram was clinically useful. CONCLUSION: The authors developed and validated a novel nomogram that can provide individual prediction of local complications within 90 days for patients after TKA. This practical tool may be conveniently used to estimate individual risk and help clinicians take measures to minimize or prevent the incidence of complications.