top of page
particlesolutions

A Surprisingly Easy Way to Avoid Protocol Amendments


I realize I am stating the obvious when I say that Clinical trial results are critical to the success of a biotech. Study results are instrumental in garnering exits, financing and ultimately sales. If the clinical study execution does not deliver the analysis that supports the product thesis the company can be sunk.

Protocol amendments are often needed because statisticians and programmers uncover a disconnect between the story the company is trying to tell versus the that can be told due to data taken in the study. This disconnect can arise when statisticians and programmers begin to work toward the creation of analysis data sets, TFL’s etc. as ambiguities in the data become clear. These ambiguities often require protocol amendments resulting in reporting delays. Occasionally these ambiguities can result in an inability to for the company to tell an effective product story. In the following interview with Mike Willis of Tradecraft, Mike outlines some simple steps that when applied with the appropriate expertise and rigor to the protocol can help teams to avoid these pitfalls.

What percentage of the initiated protocols you are involved with, require protocol amendments to allow the analysis and reporting the team wants? What percentage of those protocols cannot be modified to deliver the analysis the team hoped for?

In my experience, 70% of the protocols we work on will require at least one amendment to address analysis and reporting issues. These amendments often cost $50 – 75k and take at least a month to execute. In addition, these changes are unnecessarily disruptive.

What are the top drivers of the analysis issues and therefore the amendments?

The issues fall into two buckets, protocol design and data collection. Protocol designs often do not collect the right descriptive information (metadata) that allows the understanding of the context of the data collected. Data formats can also be an issue. For example, the specific dates for first dose and last dose are important. We have experience with studies an end date for study participation was collected but not the specifics around dose discontinuation. There have also been studies that failed to capture the initial dosing date. In those cases our analysis was far less robust than the team would have preferred.

What do you see as the root cause to the issues you just shared?

Each stakeholder brings a different perspective to a protocol review. Clinical operations teams are typically focused on reviewing the protocol from a medical perspective. The root cause of the issues is the lack of timely detailed examination required to ensure the right data and metadata are collected. The expertise to perform this lies with your statistician and programmers.

What are some simple fixes, both technical and work practice that can help clinical teams to avoid these issues?

The simplest way to get started is to include stats and programming in the team from the time the draft protocol and CRF are available. They don’t need to be involved in every meeting but asking for their input early will allow them time to uncover potential issues. This is especially important in larger programs running multiple studies. A program-wide view from a stats and programming perspective ensures the team will be positioned to tell the product story effectively.

What do you think the average clinical team could save in time and money if they implemented these simple changes? What other benefits around team working and morale do expect teams to enjoy from the changes?

As I said before each protocol amendment can often cost between $50 and $70k. The delays for each amendment can be counted as a handful of months. If you have a couple of amendments it can really add up.

Understand that this does not happen often, there are instances when no matter what analysis gyrations we go through some analysis cannot be performed. In those cases, especially in late phase development, these issues can have a significant effect on the value of the product.

Keep in mind that we are not suggesting using more stats or programming resources. The recommendation is simply to apply the expertise earlier.

Recent Posts

See All

Commenti


bottom of page