Abstract: AIM: To estimate the treatment effect before and after patients stopped taking the drug, based on the combination of withdrawal randomized study and active extension trial design. METHODS: The short term treatment effect was estimated separately by the traditional method of using first stage data and the method of taking second or third stage data into consideration. And the treatment difference between long term and short term, also the treatment difference after patients discontinued from treatment were further assessed. The robustness of the result was tested by simulation assumed different scenario. RESULTS: The standard error (0.17) of the treatment effect estimation used more stage data was less than that of only used the first stage (Standard error: 0.19), besides much powerful for those treatment effect could be stabilized in the short period. CONCLUSION: Compared with the method only utilized the data of the first stage, the method developed here utilized second or third stage data. The utilization of more information leaded to decreasing of standard deviation and increasing of validity for the estimation of treatment effect. But the results will be influenced by the time required for the stabilization of the treatment effect, since the method was based on certain assumptions.

Key words: active extension trail, clinical trail, repeated measurements