Xiaohui Wu1, Huanjun Huang2, Hongyan Hou1, Guanxin Shen3, Jing Yu1, Yu Zhou1, Munyemana Jean Bosco1, Lie Mao1, Feng Wang1, Ziyong Sun1. 1. Department of Laboratory Medicine, Tongji Hospital, Tongji Medical College, Huazhong University of Science and TechnologyWuhan, China. 2. Department of Gastroenterology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and TechnologyWuhan, China. 3. Department of Immunology, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
Abstract
Background: The differentiation between intestinal tuberculosis (ITB) and Crohn's disease (CD) is a challenge. The aim of this study was to investigate a predictive model for differential diagnosis between ITB and CD. Methods: A total of 268 patients who were suspected of having ITB or CD were prospectively recruited between January 2013 and September 2016. The clinical, laboratory, radiological, endoscopic, and histological features were investigated and subjected to univariate and multivariate analyses. The final predictive model was developed based on the regression coefficients of multivariate logistic regression. To validate the model, the same regression equation was tested on the other group. Results: A total of 239 patients had a final diagnosis, including 86 ITB and 153 CD. Five variables (perianal disease, pulmonary involvement, longitudinal ulcer, left colon, and ratio of tuberculosis-specific antigen to phytohaemagglutinin) were selected for the predictive model to discriminate between ITB and CD. In the predictive model of the training data set, the area under the receiver operating characteristic (ROC) curve, sensitivity, specificity, and accuracy, with a cutoff level of 0.29, were 0.975 (95% confidence interval [CI], 0.939-0.993), 96.7%, 90.7%, and 92.8%, respectively. Application of the predictive model to the validation data set showed similar performance in distinguishing ITB from CD. The area under the ROC curve, sensitivity, specificity, and accuracy were 0.950 (95% CI, 0.871-0.987), 88.5%, 93.5%, and 91.7%, respectively. Conclusions: This 5-marker predictive model could be conveniently used by clinicians to draw a reliable differential diagnosis between ITB and CD in clinical practice. 10.1093/ibd/izy154_video1izy154.video15790725497001.
Background: The differentiation between intestinal tuberculosis (ITB) and Crohn's disease (CD) is a challenge. The aim of this study was to investigate a predictive model for differential diagnosis between ITB and CD. Methods: A total of 268 patients who were suspected of having ITB or CD were prospectively recruited between January 2013 and September 2016. The clinical, laboratory, radiological, endoscopic, and histological features were investigated and subjected to univariate and multivariate analyses. The final predictive model was developed based on the regression coefficients of multivariate logistic regression. To validate the model, the same regression equation was tested on the other group. Results: A total of 239 patients had a final diagnosis, including 86 ITB and 153 CD. Five variables (perianal disease, pulmonary involvement, longitudinal ulcer, left colon, and ratio of tuberculosis-specific antigen to phytohaemagglutinin) were selected for the predictive model to discriminate between ITB and CD. In the predictive model of the training data set, the area under the receiver operating characteristic (ROC) curve, sensitivity, specificity, and accuracy, with a cutoff level of 0.29, were 0.975 (95% confidence interval [CI], 0.939-0.993), 96.7%, 90.7%, and 92.8%, respectively. Application of the predictive model to the validation data set showed similar performance in distinguishing ITB from CD. The area under the ROC curve, sensitivity, specificity, and accuracy were 0.950 (95% CI, 0.871-0.987), 88.5%, 93.5%, and 91.7%, respectively. Conclusions: This 5-marker predictive model could be conveniently used by clinicians to draw a reliable differential diagnosis between ITB and CD in clinical practice. 10.1093/ibd/izy154_video1izy154.video15790725497001.