Construction and effectiveness assessment of a Harvard cancer index-based predictive model for perioperative venous thromboembolism in elderly patients with femoral neck fracture
10.3760/cma.j.cn501098-20250303-00116
- VernacularTitle:基于哈佛癌症指数的老年股骨颈骨折患者围术期静脉血栓栓塞症风险预测模型构建及效能评估
- Author:
Yifeng GUO
1
;
Bingdu TONG
;
Xin GUO
;
Tingting GUO
;
Yuchen MA
;
Na GAO
;
Xuan WANG
;
Weinan LIU
;
Xiaopeng HUO
;
Yaping CHEN
Author Information
1. 中国医学科学院，北京协和医学院，北京协和医院骨科，北京 100730
- Publication Type:Journal Article
- Keywords:
Aged;
Femoral neck fractures;
Venous thromboembolism;
Risk factors;
Predictive models
- From:
Chinese Journal of Trauma
2025;41(5):501-509
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct a Harvard cancer index-based risk predictive model for perioperative venous thromboembolism (VTE) in elderly patients with femoral neck fracture and assess its predictive effectiveness.Methods:A retrospective cohort study was conducted to analyze the clinical data of 610 elderly patients with femoral neck fracture admitted to Peking Union Medical College Hospital between January 2013 and December 2022, including 193 males and 417 females, aged 60-99 years [(77.3±9.0)years]. The patients were divided into VTE group ( n=125) and non-VTE group ( n=485) according to occurrence of VTE during the perioperative period. The two groups were compared in terms of gender, age, body mass index, smoking status, alcohol consumption, time from fracture to admission, surgical waiting time, comorbidities, perioperative electrolyte disorders, past or present history of malignancy, past history of deep vein thrombosis (DVT) or pulmonary embolism (PE), and preoperative use of oral anticoagulants. Univariate analysis and multivariable stepwise Logistic regression analysis were conducted to evaluate and identify independent risk factors for perioperative VTE in elderly patients with femoral neck fracture. A perioperative VTE risk predictive model for elderly patients with femoral neck fracture was constructed using the Harvard cancer index: (1) assigning a risk score to each variable according to the corresponding conversion criteria of the Harvard cancer index and risk score, based on the magnitude of their ORs; (2) determining the exposure rate of each risk factor based on the population distribution observed in this study; (3) calculating the average population risk score; (4) computing the individual VTE risk score; (5) deriving the ratio (X) of each individual ′s VTE risk score to the population average. Based on the Harvard cancer index classification criteria for disease risk levels, individual VTE risk categories were determined. The predictive performance of the risk stratification was evaluated by comparing the incidence of VTE across different risk levels. The predictive performance of the model was evaluated based on sensitivity, specificity, and the area under the receiver operating characteristic (ROC) curve (AUC). The calibration of the model was assessed using the Hosmer-Lemeshow (H-L) test and internal validation was performed using the bootstrap resampling method with 1000 iterations. Results:Univariate analysis showed that gender, age, time from fracture to admission, surgical waiting time, previous cerebral infarction, stroke within the past month, Alzheimer′s disease, primary Parkinson′s syndrome, hysterectomy with bilateral adnexectomy, perioperative electrolyte disorders, history of DVT or PE, and preoperative use of oral anticoagulant drug were moderately associated with the occurrence of VTE in elderly patients with femoral neck fracture ( P<0.10). Multivariable stepwise logistic regression analysis demonstrated that female gender ( OR=2.26, 95% CI 1.34, 3.80, P<0.01), time from fracture to admission>1 day ( OR=3.70, 95% CI 2.24, 6.12, P<0.01), surgical waiting time>70 hours ( OR=2.06, 95% CI 1.29, 3.30, P<0.01), previous cerebral infarction ( OR=3.78, 95% CI 1.04, 13.76, P<0.05), stroke within the past month ( OR=11.57, 95% CI 1.21, 110.44, P<0.05), Alzheimer′s disease ( OR=3.26, 95% CI 1.12, 9.49, P<0.05), primary Parkinson ′s syndrome ( OR=3.47, 95% CI 1.22, 9.85, P<0.05), previous hysterectomy with bilateral adnexectomy ( OR=4.75, 95% CI 2.09, 10.80, P<0.01), perioperative electrolyte disorders ( OR=2.73, 95% CI 1.39, 5.35, P<0.01), and preoperative oral anticoagulant use ( OR=3.86, 95% CI 1.18, 12.67, P<0.05) were significantly associated with the occurrence of perioperative VTE in elderly patients with femoral neck fracture. Based on the above 10 risk factors, a perioperative VTE risk predictive model for elderly patients with femoral neck fracture was constructed with the Harvard cancer index. The formula was as follows: X=[10×(female gender)+25×(time from fracture to admission>1 day)+10×(surgical waiting time>70 hours)+25×(previous cerebral infarction)+50×(stroke within the past month)+25×(Alzheimer′s disease)+25×(primary Parkinson′s disease)+25×(previous hysterectomy with bilateral adnexectomy)+10×(perioperative electrolyte disorders)+25×(preoperative use of oral anticoagulant drug)]/33. Individualized VTE risk was classified into five levels: very low, low, moderate, high, and very high, with corresponding VTE rates of 4.8%, 11.8%, 14.9%, 32.3%, and 73.5%, respectively ( χ2=87.71, P<0.01). The VTE risk predictive model demonstrated an AUC of 0.74 (95% CI 0.69, 0.79, P<0.01), with a sensitivity of 63.2% and specificity of 74.8%. The H-L goodness-of-fit test indicated satisfactory model calibration ( P>0.05). The internal validation with the bootstrap method confirmed that the AUC remained 0.74. Conclusions:Female gender, time from fracture to admission>1 day, surgical waiting time>70 hours, previous cerebral infarction, stroke within the past month, Alzheimer′s disease, primary Parkinson′s syndrome, hysterectomy with bilateral adnexectomy, perioperative electrolyte disorders, and preoperative use of oral anticoagulant drug are independent risk factors for perioperative VTE in elderly patients with femoral neck fracture. Based on these factors, the perioperative VTE risk predictive model constructed using the Harvard cancer index demonstrates good clinical predictive value. Individualized VTE risk stratification can effectively identify high-, intermediate-, and low-risk populations, providing a valuable reference for tailoring anticoagulant prophylaxis strategies and enhancing postoperative surveillance.
