Statistical analysis of random coefficients model and covariance pattern model for longitudinal data with time varying covariate

OBJECTIVE To apply random coefficients model and covariance pattern model to analyze longitudinal data with time varying covariate. METHODS An example of light, medium primary hypertension clinical trial was given to show the application of random coefficients model and covariance pattern model with time varying covariate-dose considering the dose at all-time points changing with illness and the MIXED procedure of the SAS system was used. RESULTS The results of the two models were very close. At 5% level of significance, there were no statistically significant differences between the groups, and the effects of time, age, dose and diastolic pressure before treatment were statistically significant (P﹤0.05). CONCLUSION It can get more objective results of medicine effect by using random coefficients model and covariance pattern model in longitudinal data because the models consider not only the data correlation and the effect of time varying covariate but also can handle the material with missing value.