Application of negative binomial model for safety assessment in new drug

    LI Chanª²Juan, XIA Jieª²Lai, YAO Chen , Yu Liª²Li

    1Department of Health Statistics, School of Preventive Medicine, Fourth Military Medical University, Xi¬ðan 710033£¬China£¬2Department of Medical Statistics, General Hospital of PLA, Beijing 100853, China

    ¡¾Abstract¡¿AIM£º To apply negative binomial model for the safety assessment of a new drug, so as to analyze the occurrence of adverse events. METHODS: Using the data from one phaseª²III clinical trial of Rosiglitazone in treating typeª²2 diabetes, we fitted the poisson regression model and the negative binomial regression model, with the frequency of adverse events as response variable, and sex, age as well as many other factors as independent variables. The fitting effectiveness of the models and the statistical significance of the factors were analyzed. RESULTS: The negative binomial regression model proved satisfactory and its fitting effectiveness was better than that of the poisson model. Two factors, sex and treatment, were found significant and were included in the regression model. No significance was found in the factor of age, and it was not include in the regression model. CONCLUSION: It is a new and better choice to apply the negative binomial regression model to drug safety assessment, which not only analyzes the frequency of adverse events but also assesses the effects of covariates. ......