Background  The incidence of postoperative cognitive dysfunction (POCD) is high in elderly patients, which increases the incidence of postoperative dementia and the mortality. Therefore, it is vital to identify aging patients with a high risk of POCD early and accurately.

  Objective  To retrospectively analyze risk factors and establish a risk predictive model of POCD in elderly patients undergoing total knee arthroplasty.

  Methods  Based on the Perioperative Database of Chinese Elderly Patients (PDCEP), a total of 199 elderly patients (age≥65 years) who underwent total knee arthroplasty from February 2020 to April 2022 at the First Medical Center of Chinese PLA General Hospital and other 17 third-grade class-A hospitals in different regions of China were selected. The participants’ cognitive function was assessed by Mini-Mental State Examination (MMSE). According to the results, patients were divided into POCD group and non-POCD group. The independent risk factors of POCD in elderly patients undergoing total knee arthroplasty were analyzed and a predictive model was established. The prediction efficiency of the model was evaluated by drawing the receiver operating characteristic curve (ROC), and the area under the curve (AUC) was calculated to evaluate the value of the predictive model.

  Results  Among the 199 patients included in this study, POCD occurred in 47 (23.6%) patients. Univariate analysis showed that the surgery duration, intraoperative blood transfusions, blood loss, infusion volume, the resting Numerical Rating Scale (NRS) at 1 day after surgery and the incidence of remedial analgesia were related to POCD (all P<0.05). Multivariate logistic regression analysis showed that the resting NRS at 1 day after surgery (OR=1.304, 95%CI: 1.047-1.624, P=0.018) was a risk factor of POCD. Intraoperative blood transfusion (OR=0.317, 95%CI: 0.144-0.700, P=0.004) and the incidence of remedial analgesia (OR=0.348, 95%CI: 0.158-0.768, P=0.009) were protective factors of POCD in elderly patients undergoing total knee arthroplasty. A predictive model was constructed according to the regression coefficient of each variable, and the ROC curve analysis was conducted. The AUC value was calculated to be 0.811 (95% CI : 0.731-0.891, P<0.00). The sensitivity was 84.1% and the specificity was 74.5%.

  Conclusion  Resting pain at 1 day after surgery is the independent risk factor, while the intraoperative blood transfusion and the incidence of remedial analgesia are protective factors of elderly patients undergoing total knee arthroplasty.