I was very excited to attend the ASA Biopharmaceutical Section Regulatory-Industry Statistics (ASA-FDA) workshop in Washington D.C. in September. It reminded me why I love being a clinical trials statistician and particularly why I enjoy working at a small CRO. The purpose of the ASA-FDA workshop is to provide an opportunity for statisticians and researchers from the FDA, academia, and industry to gather and discuss statistical issues we face in clinical research. The workshop also provides a chance to socialize with fellow statisticians.
The first plenary session discussed ICH E9 R1 – the technical guideline supplementing the E9 document outlining the statistical principles used for clinical trials. It was a great way to set the stage for the remaining two days by reminding statisticians of the importance of design, the choice of the measurements of the treatment effect, consideration of the analysis populations, handling missing data, and the role of sensitivity analyses (see my colleague Benny’s blog about this topic!).
The second plenary session discussed the broad applications of Bayesian methods, a perennial favorite at statistical and clinical trials conferences. As a trained frequentist, earlier in my career I used to be a bit apprehensive about Bayesian approaches, but I have seen more designs and applications using Bayes in studies that I’ve supported and have grown more comfortable with the concepts. Apparently, so has the FDA. Beyond the plenary session, presenters rarely passed up an opportunity to plug an application where a Bayesian method could be implemented.
After the plenary sessions, I felt like a kid in the candy shop – too many great topics (sometimes overlapping) and only a day and half to cover them! (The conference was shortened this year due to the impending hurricane that was forecast to affect Washington D.C.) For the contributed sessions, I wanted to attend those on current or past issues that I’ve encountered in my work. I also tried not to miss sessions with speakers whom I’ve worked with or those whom I have heard speak and know will deliver a good session. Finally, I wanted to attend at least two sessions about a novel but relevant topic.
Among the sessions I attended was a discussion on master protocols (including the recently released FDA guidance), advancements in the methods for detecting and handling non-proportional hazards, design and analyses of studies in rare disease, and software tools and statistical approaches in centralized monitoring. I also attended other interesting sessions about the use of extrapolation and its application in pediatric studies, as well as a session discussing the use of real world data, missing data, and rare disease applications.
The last session I attended, in my opinion, could have been a plenary session – statistical leadership. While certain statistical tasks during the clinical trial are well established (sample size calculation, randomization scheme planning and execution, statistical analysis plan writing, data analysis), the statistician may not be seen as a leader. Sometimes we allow ourselves to be relegated to the technical corners of the process (development, analysis, fixing problems after they occur). The panelists, who represented industry and the FDA, shared their personal experiences and perspectives in growing influence as a statistician. While most other contributed sessions were heavily technical in their exposition and content, this session reminded us of the important voice that we as statisticians bring to clinical research and it encouraged us to develop those skills. This also brought it back full circle to the first plenary topic on ICH E9 R1 and the importance of the collaboration between clinicians and statisticians when developing the trial.
The ASA-FDA workshop not only offered insight into several current statistical and design issues facing industry and the regulatory bodies, it reminded me of why I really enjoy being a CRO statistician. At PROMETRIKA, we are a small CRO with strong statistical expertise. We have the opportunity to work in diverse indications with unique statistical challenges and our role can vary depending on the needs of the project. As CRO statisticians, we are widely versed in many areas of clinical trials from design and development to the handling and presentation of the data. We often see concepts, techniques, or approaches in one indication that can be translated to another. The FDA workshop was a refresher and a great opportunity to discuss and explore issues with other statisticians first hand. Within the first month after the conference, I have already referenced topics from the workshop.
I hope I get to go back to the candy shop next year!