Predict Postoperative Anemia of Patients: Nomogram Construction and Validation

Abstract

IntroductionThe loss of blood is a significant problem in Total Knee Arthroplasty (TKA). Anemia often occurs after such surgeries, leading to serious consequences, such as higher postoperative infection rates and longer hospital stays. Tools for predicting possible anemia can provide additional guidance in realizing better blood management of patients.Methods2,165 patients who underwent TKA from 2015 to 2019 in the same medical center were divided into training and validation cohorts. Both univariate and multivariate logistic regression analyses were performed to identify independent preoperative risk factors for anemia. Based on these predictors, a nomogram was established using the area under the curve (AUC), calibration curve (AUC), and the area under the curve (AUC). The model was then applied to the validation cohort, and decision curve analyses (DCA) were also plotted.ResultsThrough analysis of both univariate and multivariate logistic regression, five independent predictors were found in the training cohort: female, relatively low BMI, low levels of preoperative hemoglobin, abnormally high levels of ESR, and simultaneously two sides of TKA in the same surgery. The AUCs of the nomogram were 74.6% (95% CI, 71.35%–77.89%) and 68.8% (95% CI, 63.37%–74.14%) of training and the validation cohorts separately. Furthermore, the calibration curves of both cohorts illustrated the consistency of the nomogram with the actual condition of anemia of patients after TKA. The DCA curve was higher for both treat-none and treat-all, further indicating the relatively high practicality of the model.ConclusionFemale, lower BMI, lower levels of preoperative Hb, simultaneous bilateral TKA, and high levels of preoperative ESR were figured out as five independent risk factors for postoperative anemia (<9.0 g/dL) in patients undergoing TKA. Based on the findings, a practical nomogram was constructed to predict risk of postoperative anemia. The evidence level should be level 4 according to guideline.