The Risk Assessment Modelof Imminent Vertebral Fracture In Postmenopausal Women With Osteoporosis

Part Ⅰ Development and validation of risk assessment model for osteoporotic women with imminent vertebral fracturesObjective:Evaluating the risk of imminent vertebral fracture in patients with osteoporosis,and accurate and targeted treatment for high-risk groups can obtain the best social benefits.However,there are few tools available to assess imminent vertebral fractures in patients.Our research goal is developing and validating a new vertebral fracture prediction model that reliably estimates the risk of imminent vertebral fractures in osteoporotic women,with a view to providing guidance for stratified management of populations with different fracture risks.Methods:The multicenter,retrospective study included female patients with osteoporosis who underwent bone mineral density(BMD)examination in three hospitals in China in 2016.1.1-2016.12.31,2020.2.1-2022.1.10.Retrieval of clinical data through electronic medical records.The clinical data of the Nanfang hospital of Southern Medical University(NFH)were randomized at a 2:1 ratio to create a training cohort and a validation cohort,and the clinical data of the other two hospitals(People’s Hospital of Ganzhou(PHGZ),The First Affiliated Hospital of Gannan Medical University(FAHGN))were used for external validation.The variables were screened by the least absolute shrinkage and selection operator(LASSO)regression.The model was constructed by Logistic regression,the fitting effect of the model was evaluated by Hosmer-Lemeshow test,and the discrimination of the model was evaluated by receiver operating characteristic curve(AUC).Decision curve analysis(DCA),goodness of fit and clinical impact curve(CIC)were used to evaluate the clinical applicability of the model.The data analysis time is 2022.1.11-2022.4.10.Results:The study included 1064 women(median[IQR]age,65.7[59.1-71.2]years)and was randomly divided into model training cohort(709[67%])and validation cohort(355[33%]).By using the multivariable Logistic regression,LASSO regression,combined with a literature review,the model(NFLT)with the lowest BMD T value(LSD T),the model(NFOVF)with previous vertebral fracture history and the LSD T and the model(NF+age)with previous vertebral fracture history,LSD T and age were screened.The AUC of NFLT in the development and internal validation cohort were 0.741 and 0.747,respectively.The AUC of NFOVF in the development and internal validation cohort were 0.788 and 0.774,respectively.The AUC of NF+age in the development and internal validation cohort were 0.789 and 0.779,respectively.The AUC of NFLT on the external validation cohort were 0.707 and 0.657,respectively.The AUC of NFOVF on the external validation cohort were 0.790 and 0.741,respectively,while NF+age were 0.786 and 0.748,both performing well in the goodness of fit,DCA,and CIC.There was no statistical difference in the prediction efficiency between NFOVF and NF+age(P>0.05).The incidence of fractures during prospective follow-up was also basically in line with our expectations.Conclusions:Based on clinical data and BMD,this study constructed a predictive model for osteoporotic imminent vertebral fractures.For the risk prediction of osteoporotic vertebral fractures,the LSD T at the lumbar and hip positions can provide enhanced power.Models based on LSD T values and a history of vertebral fragility fractures are a valuable tool for vertebral risk classification.The personalized nomogram we proposed can be used to predict the risk of imminent vertebral fractures in female osteoporosis patients,which is helpful to provide guidance for osteoporosis patients,and provides a basis for future research.Part Ⅱ Preliminary comparison of NFOVF and FRAX modelsObjective:Based on the establishment of the NFOVF model,we are not clear about the differences between the NFOVF and FRAX models in classifying patient characteristics.Therefore,the purpose of our study was to compare the differences between NFOVF and FRAX models in baseline data,quality of life and medication of the same group of osteoporosis patients.Methods:This single-center,retrospective study collected female osteoporosis patients from the outpatient department of Nanfang Hospital,Southern Medical University.Through face-to-face inquiries with patients in the outpatient department and the hospital electronic medical record system,we extract the variables we need,such as age,quality of life,medication status,etc.The clinical data of all patients who met the inclusion criteria were filled with missing data by mean substitution method,and the normality was evaluated.The patients were grouped according to the high and low risk of FRAX and the high,medium and low risk of NFOVF.The data were expressed by median(quartile),and the non-parametric test was used to compare whether there were statistical differences in medication and quality of life among the groups.Results:The subjects included 91 eligible participants(median[IQR]age,65.0[59.0-68.0]years old).According to FRAX high-risk(38[42%]),low-risk(53[58%])and NFOVF high-risk(8[9%]),medium-risk(50[55%]),low-risk(33[36%])groups,non-parametric tests were used to test statistical differences between groups.Whether in the FRAX risk group or in the NFOVF risk group,there was no statistically significant difference in medication and fall risk between the groups.The ODI score of the FRAX high-risk group was 28.1[21.6,40.1],and that of the high-low-risk group was 23.4[15.3,29.7].The total score of SF-36 score 54.5[41.8,68.3],physiological function 55.0[30.0,80.8],emotional function 0.0[0.0,100.0]and psychological composition 46.1[37.8,77.8]in FRAX high-risk group were significantly lower than those in low-risk group.The activity ability 0.158[0.066,0.287]and daily activity 0.107[0.045,0.194]in EQ-5D-5L score of FRAX high-risk group were higher than those of low-risk group,and the total score 0.625[0.300,0.827]was lower than that of low-risk group,with statistical difference.There was no statistical difference in quality of life scores between NFOVF groups.Conclusions:Based on the baseline data,quality of life and medication of osteoporosis patients,this study preliminarily compared the NFOVF and FRAX models.FRAX model can distinguish the quality of life of patients with osteoporosis,and it is urgent to carry out graded treatment measures for people with different fracture risks of osteoporosis.