All of us in this room are involved with clinical trials, writing, delivering, participating in clinical trials, and we know that often in those trials we need to make changes. So today I'm going to talk to you about to amend or not to amend. That truly is the question. So what's the problem right in our world? We need to be flexible, We need to be innovative. We need to be able to move and change our studies as the information comes
in that we require it. But what we've seen is that certainly within ROSH, we traditionally have had a high level of protocol amendments, even compare to our peers. Now, some of my colleagues would say, well, because we're on the forefront of innovation, a science demands that we are flexible. And that's true, But is there a case for some protocol amendments that are not needed, They don't need that process. Can we group some of those protocol amendments so you
only do it once? Can we learn as we're designing studies from the protocol amendments we've done in the past, so you don't make that mistake again. So the first thing we did was we started to think about what's the problem why should we change? When we looked at actually what is the cost and the burden of doing a protocol amendment, we really found some surprising data. Yes, there's the direct costs. You may need to pay fees
to submit a protocol amendment. You know that you need to put it through some approval process, and that takes time. But when we added up the amount of resource it took for the people in house, but importantly the burden it pays it placed on the sites that we're trying to implement our protocols, we found that the amount of both direct and indirect costs through the time, delays and the resources we used, we're in the region of up to half a million dollars per amendment. We showed that.
One of the first things we did in our change management process was we just formed a simple infographic that explained how many protocol amendments we were doing per study, what the cost of those amendments were, and how possibly we could redirect that cost, that time, and that effort into things that were going to add more value to patients. So then the question became, so what is driving these
protocol amendments? And I would often hear the conversation. Well, I would never do a protocol amendment that wasn't absolutely needed. My regulatory colleague might make me do it because this is a very regimented industry. Well, I would never do a protocol amendment that wasn't needed. It's a lot of works and a lot of process. But my safety colleague might think that it's necessary. I would never do a protocol amendment that's not needed. It's only for the safety
of the patient. But for some of these smaller issues, the quality department will say we need to do a protocol a member. So it really was the time to address this as an issue in how we work as an organization. So what did we do. The first thing we did was we got together a team of real experts cross functionally so that we could figure out what the key questions you need to ask of the change that you want to make in your protocol that will help you decide whether an amendment is needed or not.
It comes out into a pretty simple decision framework. We could implement that in a system that is able to pick up the metadata about that study easily, so the teams don't have a lot of burden in the process that they're following, it guides them through these questions. If they've got a particularly complex case, then it directs them to this team of experts who can all get round
a table. So you've got your regulatory input, your safety input, your quality input, legal if necessary, all round the table and can give that team advice In the moment. The decision whether they make a protocol amendment or not is still down to that study team, and so they can pull this through. By standardizing this amendment approach, it helped us to reduce unnecessary amendments and collect key data and insights so that we can learn from that in the future.
So you may ask, yeah, that's all very nice and pretty logical, but how do you change the mindset of people to actually make a difference. Well, it's important that this works for them. The features of this tool is that is integrated into their daily work pattern. They would normally have written up a brief of what their amendment would be and ask for some medical writing support and other support to put the work through the system. This feeds into that and the automation enables the process to
follow three freely. It creates a dashboard so that we can see the overview of all of the submissions the status and we can get insights from them, and we have this deddicated support and continuous improvement through our in house team. So did it work. We've established a huge amount of uptake across the organization, in particularly in our
late stage trials where this is focused. And if you look at the graphic in the middle, when we started this process in twenty twenty three, the blue blocks show you that there was a really high proportion of our studies would submit an amendment in any one year and
all of them were processed. When we started this initiative in twenty twenty four, we still saw a high proportion of our studies submitting an amendment, but we were able to say more than twenty five percent of them were not necessary and we avoided the teams going through that process. As we've got into twenty twenty five, we've actually seen a mindset shift. The teams are not coming to us requesting to do a protocol amendment in the first place.
They've learned from this behavior, they've learned what the guidance is, and we're not seen as many come through, and even of those that do come through, we're now seeing that we're avoiding about a third of those protocol amendments. So what does that mean, Well, if you want to put it in the sort of crudest terms, I suppose in terms of money, in terms of the avoided direct costs, indirect costs, and the time we've saved in getting our
trials to patients more quickly. We have saved money in the region of tens of millions of dollars that can get reinvested into places where it will really make a difference. And for our teams, they've found this process easy to use, intuitive, it goes with their normal workflow, and they can see the impact that it's having directly. So what's next. We've got lot of rich data. Now we are using AI to take all of these amendments and categorize them into
a standard categorization framework. We can then take all of those amendments from early and late stage development and learn from these amendments. We intend to pull that right back into the study design phase and so that really building literally quality by design. We can avoid these amendments in the first place by giving visualizations and data to the teams as they put pen to paper. So I thank you for your attention, and I don't know if there
are any questions. We've got it maybe a couple of minutes, Mary, But other than that, thank you.
Take one question, how are you handling amendments? How are you avoiding amendments under the EU CTR and the CTIS process.
So it's interesting some of the amendments that we do. We know that you can't avoid all amendments. There are some amendments that are required. What we're trying to do is be more thoughtful about it, and is there a way that we can group those the way that the time procesing it's taking for those amendments to go through. You don't want to be doing an amendment more than
once a year anyway. So what we've tried to do is figure out which the ones that are absolutely required versus those that can be handled in a different way, for for instance, via clarification letter. But that wouldn't apply.
To that Based on this the last part, when you were researching what kind of amendments were happening, do you have any examples that you saw the specific areas that came up a lot in your analysis.
Yeah, I think there's a couple of areas. One area was inconsistencies across a protocol document. So where we have the same topic discussed in various parts of the protocol document, if you've got inconsistencies, then we were needing to correct them. We've approached that in a couple of different ways. We've simplified our protocol template using the transcelerate as a model, and created a lot more standard content that we're autogenerating
so that we're removing those inconsistencies as we go. We're also using an AI check to provide consistency checks across the protocol as well.
M mm hmm.