Research On The Consistency Evaluation Method Of International Multi-regional Clinical Trials Based On Binary Data

Based on the dichotomous data,the consistency evaluation method of the international Multi-Regional Clinical Trial(MRCT)suitable for the dichotomous data is discussed.The main methods are:(1)Discount factor method:Approximate the effective rate as a normal distribution,calculate the test statistic of the target area with the data of the target area,and calculate the test statistic of the non-target area with the data of the non-target area.Then construct the weighted test statistic,use the discount factor D to calculate the weight coefficient,and then obtain the value of the weighted test statistic.(2)Effect size weighting method:The effect size weighted value was constructed by using the difference between the effective rate of the test group and the control group in the target area and the difference of the effective rate in the non-target area.The efficiency follows a beta distribution.Data were repeatedly extracted from the efficient beta distribution to calculate the difference in effect size,and the probability that the weighted value of the effect size was greater than 0 was used to judge whether the drug efficacy in the target area was significant.(3)Bayesian method:The non-target region sample information in MRCT is used as prior information,combined with the target region sample data in MRCT,the posterior distribution of the target region’s efficiency is constructed.Coefficients M are used to control the influence of non-target region information on the posterior distribution.The data were repeatedly extracted from the posterior distribution,and the corresponding values were calculated.If the χ2 mean value was greater than the critical value,the drug effect in the target area was considered to be significant.In the discount factor method,both the test power and type Ⅰ error increase with the increase of the discount factor D,the test power increases with the increase of the discount factor f,and the type Ⅰ error decreases with the increase of the discount factor f.Efficacy is the same as Type Ⅰ error.The test power and type Ⅰ error in the effect size weighting method decrease with the increase γ,and the weight coefficient has a greater impact on the type Ⅰ error than the test power.decreases with increasing.In the Bayesian method,the test power and type Ⅰ error increase with the increase of the coefficient M,the type Ⅰ error increases slowly with the increase of the coefficient M,the test power increases with the increase f,and the type Ⅰ error increases with and decrease f.The effect size weighting method based on the adjustment of dichotomous data is not only suitable for large samples,but also for small samples.The Bayesian method can make better use of the non-target area information,and while the test efficiency increases,the type Ⅰ error increases slowly.The three methods are all borrowing non-target area methods.Discount factor method and effect size weighting method have more type Ⅰ errors than Bayesian method due to borrowing information from non-target areas.The Bayesian method can borrow more non-target region information to increase the test performance while ensuring that the TypeⅠ error is small.