Assessment the relationship between reflux and body mass index with comparing different regression models

Aim: The aim of this study was to determine whether there is relation between body mass index and symptoms of gastro-esophageal reflux disease in our community using Logit, Probit and Complementary log-log models. Background: The most frequent statistical tool to address the relationship among a dichotomous response and other covariates is logistic regression. However logistic regression is familiar for researchers, some other models with similar results are favorable to assess such relationship like as Probit and Complementary log-log. Patients and methods: We carried out a population-based study to estimate the strength of the association between body mass and reflux symptoms. During interview, participants completed a valid Gastro-esophageal Reflux Questionnaire. The coefficients calculated by logistic, Probit and Complementary log-log regression with multivariate adjustments for covariates. Results: Reflux symptoms at least once a week over the last three months were reported by 522 (9.1%) of the 5733 interviewees. There were no evidence in all models to address the significant relation between Reflux and BMI and all models derived in same efficacy. Conclusion: Our results showed that the choice of the link function in this analysis is not the most important issue but still a user could also look forward to some other link functions like Probit and Complementary log-log. © 2011 RIGLD, Research Institute for Gastroenterology and Liver Diseases.