A Prediction Model About Acupuncture For Functional Diarrhea And Diarrhea-Predominant Irritable Bowel Syndrome

Objective: Acupuncture is effective and widely used for patients with functional diarrhea(FD)and diarrhea-predominant irritable bowel syndrome(IBS-D).In this study,a prediction model was developed to predict the efficacy of acupuncture for FD and IBS-D,and analysis the factors which influence the efficacy.The model could help the doctors select a more effective treatment.Methods: This study acquired data from a randomized controlled trial including 290 participants with FD or IBS-D from 8 Chinese provinces,and divided the participants into responders and non-responders.Then we used stratified random sampling method to divide the responders and non-responders into a training set and a test set.With the data of training set,a prediction model was developed.We used Student's t test to screen the 11 predictors for treatment success,and developed a prediction model based on the Logistic model with “enter” and “forward: LR” methods.The one with the minimum AIC and the maximum AUC would be selected as the best fit model.The test set was used to validate the model.The prediction accuracy of this model was evaluated using ROC curve and AUC,sensitivity and specificity.Results: 1.290 participants were included,and 232 in the training set and 58 in the test set.2.In the Student's t test,5 predictors were found statistically related to acupuncture treatment success(p<0.05),including gender,RBC count,HGB concentration,weekly stool frequency and the weekly number of days with normal defecation.3.Fitting of models developed by different methods: the AIC in the “forward: LR” method was 270.131,the AUC was 0.755.The AIC in the “enter” method was 270.701,and the AUC was 0.763.The best fit model was the one developed by “forward: LR” method.4.RBC count(OR,0.443 [95%CI,0.259 to 0.760];p=0.003)and weekly stool frequency(OR,0.865 [95%CI,0.820 to 0.912];p=0.000)were included in the final prediction model.5.The AUC in the test set was 0.747(95%CI,0.620 to 0.874).The sensitivity was 75% and specificity was 73.5%.Conclusions: 1.The probability of treatment success decreased with the increasing of RBC count in blood examination and stool frequency.The increasing of these 2 factors may relate to more severe diarrhea.2.The prediction model about acupuncture for FD and IBS-D is Logit(p)= log(p/(1-p))=5.612-0.813?RBC count(?1012/L)-0.145? weekly stool frequency 3.The prediction model may be helpful to predict the efficacy of acupuncture for IBS-D and FD.