At a recent conference Adam Hamm, Director Biostatistics at Cytel, presented his thoughts on Best Practices and Operational Considerations for Adaptive Designs and Interim Analyses. In this blog, we share some highlights from his presentation, drawing out some of the key operational priorities from the biostatistician's point of view.
Data Integrity Clearly, data integrity must be maintained, and this starts with having robust clinical monitoring in place at the sites, to ensure that clean reliable data is available for decision making at the interim analysis. This is underpinned by the use of a reliable electronic data capture system to ensure that data is available in a timely fashion to the data management team.
Firewalls Appropriate systems must be in place to maintain blinding and data integrity and share information to the relevant parties. One example of such a system is Cytel’s ACES which was specifically designed to securely manage the interim analysis workflow. If the same CRO is delivering both blinded and unblinded statistical and/ or data management services for the clinical trial, then appropriate division must be put in place between the blinded and unblinded teams. These divisions can include separate physical locations for each team as well as electronic internal firewalls. There should be clear, robust documentation in place to guide the flow of data and reports- such documentation may include data transfer agreements. This documentation should detail the process for how data from outside vendors will be handled. From a data management point of view, it’s important to note that lab or PK vendors may have data that is potentially unblinding, and so blinded team members should not have access to this data.
Clear Documentation We have discussed on the blog previously the importance of robust documentation to help ensure the avoidance of operational bias within an interim analysis. What kinds of documentation can we expect to see in place to support the interim analysis?
Data Monitoring Committee Charter
The Data Monitoring Committee (DMC) Charter is clearly a central document in the interim analysis. It should outline the purpose of the data review committee and identify all of the stakeholders involved (Sponsor, Independent Statistical Center, Executive Committee, Data Monitoring/Review Committee, and their specific roles and responsibilities). Expectations should also be provided for the timings of the meetings. For studies with interim analyses that could change the design of the study (ex. Sample size re-estimation, dose selection/dropping), details of those analyses should only be made available to a select group of individuals (DMC, DMC statistician, unblinded statistician) so that operational bias can be avoided as much as possible. Finally, a communication plan will detail what decisions will be made by the Data Monitoring Committee and the key criteria, along with a list of personnel who will have access to those decisions.
Interim Analysis Plan An Interim Analysis plan is not always produced, but can be useful for more complex interim analyses. Similar to the Statistical Analysis Plan (SAP) it will contain details of the statistical methods to be used for the Interim Analysis, the scope of the data used, and the data cut-off criteria for data review. It would also include table and figure shells specific to the interim analysis, which have been reviewed by the Data Monitoring Committee, an unblinding plan and guidance for stopping rules, if applicable.
Data Review Recommendations Data Review Recommendations help document the DMC recommendations for how the study will proceed after the interim analysis. Depending on the trial design next steps could include trial continuation, trial modification, halting (delay) of enrollment, or stopping of the trial. The data review recommendations are provided to the executive committee or designated appointee through appropriate, secure channels.
Cytel's biostatistics and data management experts have extensive experience designing and implementing adaptive trials and supporting interim analyses. Read more about how we support clients in the suggested reading below: