Group sequential trials

Our website uses cookies to enhance your experience. By continuing to use our site, or clicking "Continue," you are agreeing to our Cookie Policy Continue. Download PDF Comment. Article Information References. Effect of blinatumomab vs chemotherapy on event-free survival among children with high-risk first-relapse B-cell acute lymphoblastic leukemia: a randomized clinical trial.

Sample size calculation for a hypothesis test. Group sequential methods in the design and analysis of clinical trials. A multiple testing procedure for clinical trials. Discrete sequential boundaries for clinical trials. Design and analysis of group sequential tests based on the type I error spending rate function. A review of methods for futility stopping based on conditional power. Recommendations for data monitoring committees from the Clinical Trials Transformation Initiative.

Interpretation of clinical trials that stopped early. Save Preferences. Privacy Policy Terms of Use. Limit characters. Limit 25 characters. Conflicts of Interest Disclosure Identify all potential conflicts of interest that might be relevant to your comment. Err on the side of full disclosure. Yes, I have potential conflicts of interest. No, I do not have potential conflicts of interest. Limit characters or approximately words. The following information is required and must be completed in order to submit a comment:.

Thank You. Your comment submission was successful. Please allow up to 2 business days for review, approval, and posting. This Issue. Views 11, Citations 0. View Metrics. Twitter Facebook More LinkedIn. Power can be calculated using the function getPowerRates. This function has the same arguments as getSampleSizeRates except that the maximum total sample size needs to be defined maxNumberOfSubjects and the Type II error beta is no longer needed. The getPowerRates as well as getSampleSizeRates functions can also be called with a vector argument for the probability pi1 in the intervention group.

This is illustrated below via a plot of power depending on this probability. For examples of all available plots, see the R Markdown document How to create admirable plots with rpact. Sample size calculation proceeds in the same fashion as for superiority trials except that the role of the null and the alternative hypothesis are reversed. Testing in non-inferiority trials is always one-sided. The probability under the null hypothesis can be specified with the argument thetaH0 and the specific alternative hypothesis which is used for the sample size calculation with the argument pi1.

In general, rpact supports both one-sided and two-sided group-sequential designs. Group sequential designs using Lan-DeMets error spending functions are proposed for historical control trials with time-to-event endpoints. Simulation results show that the proposed group sequential designs using historical controls preserve the overall type I error and power. Randomized clinical trials RCTs are considered the gold standard for clinical trials comparing treatment groups. However, historical control trials HCTs are an alternative to RCTs if randomization is not feasible because of ethical concerns, patient preference, limited patient populations, or regulatory acceptability.

The major benefit of HCTs is that all patients can receive the new treatment with historical data providing the information for the control arm. Therefore, HCTs are useful for studies with limited patient populations. HCTs are widely used in clinical research. However, Korn and Freidlin showed that these methods do not preserve power if uncertainty in historical control data is considered.

To overcome this challenge, Chang et al. HCTs are often monitored by interim analysis to stop accrual when outcomes of treated patients are worse than those in the historical control group. Monitoring clinical trials with historical controls poses the statistical issue of comparing two outcomes in a situation in which data from the experimental group are sequentially collected and all data from historical controls have been collected at each interim analysis.

Few studies have discussed the monitoring of clinical trials that use historical controls. Chang et al. However, the popular group sequential methodology that uses error spending functions to generate efficacy and futility boundaries for interim monitoring has not been applied to HCT designs. The advantages of the error spending function methodology are the flexible choice of the error spending functions and user-friendly software, such as East 5 and the R package available for the implementation of trial designs.

We first introduce the sequential log-rank tests for HCTs, followed by the number of events calculation for fixed-sample tests. We then discuss the information time for HCTs and provide a detailed method to determine the maximum number of events and decision boundaries. Finally, we use our proposed method with an example dataset to illustrate the HCT designs and conduct simulation studies. Consider a trial with two groups, a historical control group and an experimental group, designated as group 1 and group 2, respectively.

The survival distributions of the two groups satisfy the proportional hazards model. The hypothesis for testing improvement of the survival distribution of the treatment group compared with that of the historical control group can be expressed as.

The nonparametric log-rank test is often used for testing the above hypothesis. Therefore, the log-rank score at an interim look at calendar time t is given as follows:. The variance V t is given by. The sequential log-rank test at calendar time t is given by. For an interim analysis at calendar time t , the sequential log-rank test for HCTs uses all data from the historical control group but only data up to time t from the experimental group.

Furthermore, this test is a two-sample test between the historical control arm and experimental arm. The uncertainty in historical control data is taken into account by the two-sample sequential log-rank test. For RCTs, a well-known formula for the number of events is given as follows:. To design HCTs, we assume that no further follow-up is available for the historical control group or that the historical control data were frozen at the design stage; therefore, the number of events for the historical control is fixed during the trial.

Hence, only calculating the number of events for the experimental group is necessary, which can be calculated as follows:. This results in. For unbalanced RCTs, the traditional information time, defined as the proportion of observed number of events up to an interim look to the total number of events as planned, is not valid, because of an unbalanced number of events in each group.

Alternatively, the information time at calendar time t for unbalanced RCTs can be calculated as follows:. For HCTs, data accumulated up to an interim look from the experimental group are compared to all data from historical controls. Therefore, the information time for HCTs at calendar time t is given by. The information time for HCTs can then be calculated as. The test procedure is as follows:.

One drawback of this simultaneous computation is that the futility boundary cannot be overruled. If the test statistic crosses the futility boundary, the trial must be terminated, or the type I error may be inflated.

Nonbinding futility boundaries suggest that, if the test statistic crosses the futility boundary, the trial can continue without inflating the type I error. Sequential nonbinding efficacy and futility boundaries are constructed by using an algorithm similar to that given by East 5 and Chang et al.

For the first look, compute upper boundary u 1 as.