Construction Of A Risk Forecast Model Of Shoulder Joint Dysfunction For Patients With Postoperative Breast Cancer

Objective To investigate the risk factors of shoulder joint dysfunction for patients with postoperative breast cancer and construct a risk prediction model for that,so as to provide a method for health care providers to screen high risk patients,and take targeted measures to these patients clinically and formulate a systematic,reasonable,and effective nursing intervention for preventing and reducing the incidence of shoulder joint dysfunction for these patients.Methods Demographics and general Information(such as age,height,weight,education level,marital status,place of residence,main hand,et al),disease information(such as tumor stage,tumor pathological type,axillary lymph),operation information(such as operation methods,type of incision,operation side,number of lymphatic dissection,complications of subcutaneous hydrops,lymphedema,lymphangitis,radiotherapy),functional exercise information(such as time to start exercise,time to start shoulder joint exercise,exercise method,exercise intensity,compliance,active exercise,attention to functional exercise,et al)and social support information(such as sources of social support,family support,family supervision)of breast cancer patients who underwent radical mastectomy were collected at the department of breast cancer,the First Affiliated Hospital of Zhejiang Chinese Medical University from July 2014 to July 2015 and analyzed retrospectively.Shoulder joint ROM measurement was chosen as a gold standard for diagnosing shoulder joint dysfunction.Patients were divided into shoulder joint dysfunction group and non-shoulder joint dysfunction group.First,a single factor logistic regression was used to analyze the clinical data.Those variables which had statistical significance between two groups were selected for multiple stepwise logistic regression and establish a prediction model so as to screen the risk factors of shoulder joint dysfunction for patients with postoperative breast cancer.In detail,the ROC curve analysis was adopted to evaluate the predictive efficiency of the model and identify the optimal cut-off point.And the dignostic test was adopted to evaluate the prediction accuracy,sensitivity,specificity,misdiagnosis rate,omission diagnostic rate,Youden index and kappa value.Results1.General data of shoulder joint dysfunction for patients with postoperative breast cancerThe degree of shoulder joint dysfunction:120 patients experienced radical mastectomy.Among them,62 cases(51.67%)were in the non-SJD group and 58 cases(48.33%)were in the SJD group.In the latter group,24 cases(20.00%)were in a slight degree,20 cases(16.67%)were in a moderate degree and 14 cases(11.67%)were in a severe degree.The time distribution of shoulder joint dysfunction:70 cases(58.30%)were three months to six months after operation.26 cases(37.14%)were normal,14 cases(20.00%)were in a slight degree,19 cases(27.14%)were in a moderate degree and 11 cases(15.71%)were in a severe degree.26 cases(21.70%)were six months to twelve months after operation.18 cases(69.23%)were normal,5 cases(19.23%)were in a slight degree,1 cases(3.85%)was in a moderate degree and 2 cases(7.69%)were in a severe degree.24 cases(20.00%)were more than twelve months after operation,18 cases(75.00%)were normal,5 cases(20.83%)were in a slight degree,0 cases was in a moderate degree and 1 cases(4.17%)was in a severe degree.2.Analyzing the influencing factors for postoperative breast cancer patients occurring shoulder joint dysfunctionThe simple logistic regression showed that eleven risk factors had significant relationship with occurig shoulder joint dysfunction for patients with postoperative breast cancer.They were operation methods,radiotherapy,time to start exercise,time to start shoulder joint exercise,the way of exercise,exercise intensity,compliance,active exercise,obtaining exercise knowledge initiatively,attention to functional exercise and family supervision.Multivariate logistic regression showed that time to start exercise,exercise methods,compliance,family supervision,operation methods were independent risk factors for postoperative breast cancer patients with shoulder joint dysfunction,and exercise methods,compliance,family supervision were protective factors for postoperative breast cancer patients with shoulder joint dysfunction.3.Construction of a risk forecast model and evaluation for postoperative breast cancer patients with shoulder joint dysfunctionThis study established a risk forecast model,which was consisted of five factors as time to start exercise,exercise methods,compliance,family supervision,operation methods;and the prediction model of logistic regression is P=eY/(1+eY),Y=1.190+1.180X1-1.118X2-1.222X3-1.075X4+1.391X5(X1:time to start exercise,X2:exercise methods,X3:compliance,x4:family supervision,x5:operation methods);Results of the ROC curve analysis showed that the area under the ROC curve of the model is 0.849,which had statistical significance;The optimal cut-off point of the model was 0.45,namely if P≥0.45,taking it as SJD,otherwise it is non-SJD;The corresponding accuracy,sensitivity,specificity,misdiagnosis rate,omission diagnostic rate,Youden index of the model are 79.17%、81.03%、77.42%、22.58%、18.97%、0.585 respectively.And it had a moderate consistency with shoulder joint ROM measurement standard(the Kappa was 0.584,P<0.01).Conclusion1.The main risk factors for predicting the incidence of shoulder joint dysfunction were operation methods,start to exercise 7 days after operation,single exercise method,incompliance or partial compliance,no supervision from family.Positive prevention which plays an important role in guiding further functional exercise need to be developed early and effectively for the controllable factors.2.This research used logistic regression to establish a risk prediction model for postoperative breast cancer patients with shoulder joint dysfunction,which showed certain practical value in screening population at high risk of shoulder joint dysfunction.The model needs more verification and improvement before it can be promoted and applied,as data sources were limited.3.The model can predict the incidence of shoulder joint dysfunction,which provide evidence for focusing on high risk population and working out prevention strategies.