Establishment and validation of risk prediction model for postoperative venous thromboembolism in elderly patients with gastric cancer
Objective To explore the influencing factors of postoperative venous thromboembolism(VTE)in eld-erly patients with gastric cancer and the construction of its risk prediction model.Methods Clinical data of 520 elderly gastric cancer patients admitted to the general surgery department of Qilu Hospital(Qingdao)of Shandong Universi-ty from January 2016 to June 2023 after surgery were collected,and randomly assign them to the training and valida-tion sets in a 7∶3 ratio.Self-designed clinical data questionnaire,cancer-related fatigue,Barthel's index,nutritional risk screening,Pittsburgh Sleep Quality Index Scale,and frailty index were used to evaluate the patients and screen out the risk factors,and then logistic regression was used to construct and validate a predictive model for postopera-tive VTE risk in elderly gastric cancer patients.Results The univariate results showed that spouse,smoking history,self-care ability,cancer-related fatigue,comorbidity type,preoperative neoadjuvant therapy,frailty status,patho-logical type,perioperative blood transfusion history,admission D-dimer,D-dimer on the day after surgery,and ad-mission carcinoembryonic antigen were the influencing factors for postoperative thrombosis in patients(P＜0.05);The multivariate results showed that having no spouse,smoking history,inability to take care of oneself,severe cancer-related fatigue,comorbid circulatory system diseases,severe frailty,pathological type of adenocarcinoma,perioperative history of blood transfusion,high D-dimer levels upon admission,and the day after surgery were risk factors for postoperative blood clots in patients(P＜0.05);A nomogram model for postoperative VTE in elderly gastric cancer patients was established based on factors such as spouse,smoking history,self-care ability,cancer-re-lated fatigue,comorbidity type,frailty status,pathological type,perioperative blood transfusion history,admission D-dimer,and D-dimer on the day after surgery.The Bootstrap results showed a good fit between the correction curve and the ideal curve.The area under ROC curve of the training set was 0.884(95％CI was 0.843-0.924),and the C-index index was 0.884,and the area under ROC curve of the validation set was 0.891(95％CI was 0.833-0.950),and the C-index index was 0.892,indicating that the line graph model had good discrimination.The goodness of fit test for the training set was Hosmer Lemeshow x2=9.216,P=0.410,and that for the valida-tion set was Hosmer Lemeshow x2=9.526,P=0.300,indicating that the column chart model had good accuracy.The DCA results showed that the model was beneficial in clinical practice when the threshold probability was greater than 20％.Conclusion Those who have no spouse,smoke,can't take care of themselves,severe cancer-related fa-tigue,comorbidity circulatory system disease,weakness,pathological reason of adenocarcinoma,perioperative blood transfusion,high D-dimer on admission and the next day after surgery are the risk factors for postoperative thrombo-sis.The establishment of related nomogram prediction model can improve the diagnostic or nursing skills of medical staff for elderly patients at risk of VTE after gastric cancer surgery.
NETL
NSTL
万方数据
维普
