Development And Validation Of Delirium Risk Prediction Model In Elderly Patients With Hip Fracture

ObjectiveTo construct and validate a risk assessment model for post operative delirium in elderly patients with hip fracture based on systematic evaluation and Meta-analysis.Methods1.Meta-analysisRelevant studies on postoperative delirium risk factors in middle-aged and elderly patients with hip fracture were searched by computer from CNKI,Wanfang,VIP,China Biomedical Literature Service System,Pub Med,EMBase and Cochrane Library databases.The search period is from the establishment of the database to April 2022.Risk factors of postoperative delirium in elderly patients with hip fracture were systematically reviewed,and meta-analysis of risk factors and Odds Ratio of postoperative delirium in elderly patients with hip fracture was performed using Revman5.4 software.2.Establishment of risk prediction modelThe combined OR values of the risk factors selected by meta-analysis and the corresponding 95%Confidence Interval(CI)were used to calculate the regression coefficients and take them as the corresponding risk scores.The postoperative delirium risk prediction model in elderly patients with hip fracture was established and presented by a simple risk rating scale.3.Verification of risk prediction modelClinical data of elderly patients with hip fracture who met inclusion and exclusion criteria from a third-class hospital in Luoyang from April 2022 to September 2022 were collected to verify the model.Hosmer-Lemeshow test was used to evaluate the goodness of fit of the model,and a calibration curve was drawn.By comparing the sensitivity,specificity and Jorden index of models with different prediction tangential values,the optimal tangential values were determined.Area Under Curve,Decision Curve Analysis and Clinical Impact Curve evaluated model differentiation,clinical net benefit,and clinical impact.Results1.Meta-analysisA total of 28 studies involving 34,944 elderly patients with hip fracture were included in the meta-analysis,with a cumulative incidence of about 23%.The results of meta-analysis showed that:Age[OR=1.05,95%CI(1.04,1.06),P<0.05],history of diabetes[OR=4.41,95%CI(2.76,7.05),P<0.05],history of stroke[OR=4.50,95%CI(2.88,7.04),P<0.05],chronic comorbity[OR=8.61,95%CI(6.53,11.36),P<0.05],American College of Anesthesiologists grading(ASA grading)[OR=6.27,95%CI(4.61,8.54),P<0.05],preoperative albumin level[OR=1.75,95%CI(1.38,2.21),P<0.05],preoperative cognitive dysfunction[OR=3.83,95%CI(3.26,4.51),P<0.05],time between injury and surgery[OR=3.04,95%CI(2.06,4.50),P<0.05],operation duration[OR=1.99,95%CI(1.46,2.70),P<0.05],general anesthesia[OR=3.10,95%CI(2.12,4.54),P<0.05],perioperative blood loss[OR=3.51,95%CI(2.06,5.96),P<0.05],perioperative hypoxia[OR=5.09,95%CI(2.16,12.02),P<0.05].2.Establishment of risk prediction modelThe above risk factors were incorporated into the risk prediction model,and all risk factors were scored according to their weight:Age(0 points for<75,0.9 points for>75),chronic comorbidities(0points for no,2.2 points for no),history of diabetes(0 points for no,1.5 points for no),history of stroke(0points for no,1.5 points for yes),ASA classification(0 points for<3,1.8 points for≥3),preoperative cognitive impairment(0 points for no,1.3 points for yes),preoperative albumin level(0 points for>40g/L,0.6 points for≤40g/L),general anesthesia(0 points for no,1.1 points for yes),preoperative waiting time(0points for<3d,1.1 points for≥3d),surgical duration(0 points for<2h,0.7 points for≥2h),perioperative blood loss(0 points for<300ml,0.9 points for≥300ml),intraoperative hypoxia(0 points for no,1.6 points for yes),with the highest score of 14.3 points.3.Verification of risk prediction modelHosmer-Lemeshow test showed thatχ~2=7.875 and P value was 0.547,and the Brier score in the calibration curve is 0.084,indicating that the model has a good calibration degree.By comparing the sensitivity,specificity and Yoden index of the model with different tangent points,the optimal tangent point value was 5.150 points,and the AUC of the model was 0.950(0.921～0.983),and the sensitivity and specificity were 0.881 and 0.880,respectively.It is considered that patients with total score≥5.150 belong to the high risk group of postoperative delirium in elderly patients with hip fracture.The Decision Curve Analysis showed that using the model to predict POD and take preventive measures was more beneficial than intervention for all patients when the predicted probability was greater than 10%.The Clinical Impact Curve showed that when the risk threshold was greater than 60%,the positive estimate was closer to the actual number of cases.Conclusion1.Independent factors of postoperative delirium in elderly hip fracture patients included age,chronic comorbidity,history of diabetes,history of stroke,ASA grade,preoperative cognitive impairment,preoperative albumin level,general anesthesia,time from injury to surgery,duration of surgery,perioperative blood loss,and intraoperative hypoxia.2.By means of meta-analysis,a simple risk scoring system was selected to construct a risk prediction model for postoperative delirium in elderly patients with hip fracture,which can directly reflect the risk of postoperative delirium in patients,and facilitate early identification of high-risk groups for postoperative delirium in elderly patients with hip fracture according to the score.3 The model has good predictive efficacy and clinical practicability,which can assist medical staff to predict the occurrence of postoperative delirium in elderly patients with hip fracture,and also provide a reference for formulating relevant preventive measures for postoperative delirium.