Also, although the study might not have gone exactly as planned, a package can still be submitted successfully
and receive approval. To help prepare the regulatory submission in the type of situation described in the previous
section, it can be helpful to step back and look at the data and results as if you had no knowledge of or years of
experience with the product. What does the data tell you? What is clinically important? Are there any glaring
omissions or questions that you would ask about the results if you were at the regulatory authority?
EXAMPLE
The study fails to demonstrate statistical significance for the primary endpoint, and
the final analysis is different from that described in the protocol. Understandably, the
regulatory authority is concerned about the validity of the trial.
A company approached Clinipace Biostatistics for ad hoc statistical consulting shortly after
completing their pivotal phase III trial, in which they wanted to demonstrate non-inferiority to an
existing product using a composite outcome of three component outcomes. While their product
was actually superior to the comparator on two of these components, it was slightly worse for the
third component. Therefore, they did not conclusively demonstrate non-inferiority on the composite
primary endpoint.
They had an initial meeting with the FDA to discuss the regulatory path forward, which suggested
that they would need to conduct a second randomized, double-blind, concurrent-control pivotal
trial to resolve the issue. The company then asked Clinipace to assist in better understanding what
the FDA review team was looking for, designing the requirements of such a trial within a protocol,
and working with the FDA to approve the approach. Clinipace worked with the client to design a
single-arm open-label registry study, calculate a sample size, write an associated protocol, and
discuss this plan with the FDA. Importantly, this study design avoided the requirement of a second
randomized clinical trial and leveraged their existing device clearance in Europe by not requiring
solely US-based subjects.
Furthermore, failure to submit a full submission can result in the regulatory authority picking up a glaring omission
Instead of omitting data, address the situation up front by stating the limitations and issues and how they were dealt
with. An extra week preparing the package can avoid numerous review questions and a lengthy (three to four months)
review cycle (or cycles).
General Biostatistical Consulting Advice
Based on these experiences with clients and regulatory authorities, it is best to spend more time up front during the
study design to identify ways to minimize the risk of issues rather than chasing them during study conduct or, even
worse, at study completion. Additionally, this time at the beginning should be used to plan the strategies to address
potential issues. For example, identifying operational methods for avoiding or minimizing missing data at the trialplanning stage and during study conduct is always preferred to trying to apply statistical techniques to accurately
assess the impact of missing data at analysis or the submission stage. Methods to reduce the likelihood of missing
data include the use of rescue therapies for individuals who stop the assigned treatment, limiting participant burden
in terms of travel, providing frequent reminders of visits, and providing incentives to limit dropouts.3,5
11