The Study Of Uterine Prolapse Logistic Regression Analysis Of Risk Factors And The Establishment Of The Prediction Model

Objiective:To explore the main risk factors may cause uterine prolapse,appropriate regression prediction model is established,and the accuracy of the model to predict the onset of uterine prolapse,specificity and sensitivity evaluation,through the forecast probability of high-risk women early intervention and treatment.Methods: In this paper,a retrospective analysis in December 2015 to January 2018 in the south China university first affiliated hospital of department of gynaecology clinic diagnosis of uterine prolapse women(n= 102)as the case group,and with clinic has not occurred during uterine prolapse women(n = 102)as a control group of clinical data.Single factor analysis with chi-square risk factors associated with uterine prolapse may,coarse sieve has a significant impact on independent factors,and through the multifactor unconditioned logistic regression analysis of parsing the independent factors,eliminate confounding factors,the end result is statistically significant predictors of uterine prolapse.Predictor based on these and the corresponding coefficient of regression to construct the Logistic regression forecasting model,assessing Hosmer-Lemeshow test goodness-of-fit of prediction model,and through the ROC curve drawing,to evaluate the accuracy of themodel for predicting the uterus disease son,sensitive and specific degrees.Results: Through the single factor analysis,meaningful include:patient’s age,BMI,professional,vaginal delivery times,history of Ⅱ or higher degree perineal laceration and prolonged second labor history,birth weight in babies,ovarian function decline,and chronic pelvic inflammation,the nine risk factors and prediction of the onset of uterine prolapse are closely related.Multivariate Logistic regression analysis was carried out for the nine significant risk factors above,and there were 5factors that were eventually included in the regression equation,which were sorted as follows:Delivery times X7(P < 0.001,OR = 17.866,95%CⅠ [6.034,51.975]),history of ⅡOR higher degree perineal laceration X8(P < 0.001,OR = 12.277,95% CⅠ [5.462,30.647]),age X1(P < 0.001,OR = 10.736,95% CⅠ [5.693,26.143]),professional X3(P = 0.003,OR =0.003,95% CⅠ(2.190,43.140)),ovarian function decreased X12(P<0.001,OR=6.104,95%CⅠ[2.520,14.785]).Establishing Logistic regression prediction model: logit(P)=In(P/1-P)=-3.380+2.620X1+2.274X3+3.085X7+2.849X8+1.809X12,predicting the probability model of uterine prolapse: P=1/(1+Exp(3.380-2.620X1-2.274X3-3.085X7-2.849X8-1.809 X12)).The correct index of this prediction model was 0.627,81.37%,and the sensitivity and specificity were 83.33% and 79.41% respectively.Conclusion: the pathogenesis of uterine prolapse may be related to the number of vaginal delivery,occupation,age,thistory of Ⅱ or higher degree perineal laceration and the decline of ovarian function.The Logistic regression prediction model established by this study has a great reference value for predicting the occurrence of uterine prolapse.