Analysis of Repeated Data of Multi-center Clinical Trial for Traditional Chinese Medicine by Hierarchical Linear Model / 中国中医药信息杂志

ABSTRACT

Objective To explore the applicability and feasibility of the hierarchical linear model in dealing with the repeated measurement data by applying it into multi-center clinical trials’ evaluation of new Chinese medicine. Methods The theoretical research described the basic concepts and principles of hierarchical linear model and compared the applying conditions of traditional statistical methods with hierarchical linear model. In the empirical study, the hierarchical linear model was used into a real multicenter clinical trial research of traditional Chinese medicine, with the TCM syndrome score as the analysis indicator. Results The hierarchical linear model added time, group and time×group as random variables into the model to get the final result. All of the three variables were significant in the result. The interaction (time×group) statistical result was t=2.65, P=0.008 1. During the whole treatment, the TCM syndrome mean score of trial group reduced 8.5 and the control group reduced 7.47. At the end of treatment, the TCM syndrome mean score of trial group was 2.46, and the control group was 3.31, which was higher than trail group. For this result, combining with the TCM syndrome score’s developing trend, we can see that the efficacy of trial group is not worse than the control group. Conclusion By comparing the data requirements, applying conditions and analysis results, the hierarchical linear model could be an effective method which can be used in multi-center trial to evaluate traditional Chinese drug.