Part II: Enrollment Projections - Scenario Planning with Real-Time Data

In Part I of this series, we went over spreadsheet-based models like Excel and Smartsheet that are commonly used in the biotech and pharmaceutical industry to project enrollment timelines. If you haven’t yet read Part I, check out the current industry trends based on spreadsheet trackers and commonly encountered pitfalls.

If you’re running clinical trials at your biotech company, your enrollment projections could have huge financial implications that determine the company’s long term success. You need to provide accurate, data-driven projections illustrating the best-case and worst-case scenarios for enrollment for your board and management to best plan for post data readout and figure out next steps.

Today, we will dive into how you can utilize the data that you already have to produce more robust, data-driven projections. We will also showcase how innovative biotech companies are conducting real-time projection modeling and scenario planning as part of their clinical trial management — for example, simulating how adding or removing trial sites can dynamically impact trial timelines.

‍

Importance of Scenario Planning

The primary reason behind numerous biotech companies prioritizing their clinical trial’s enrollment projections is to prepare the management team and the board for the best-case and worst-case scenarios. Many biotech companies are often working towards their first drug approval from the Food and Drug Administration (FDA), and they need to make the best use of their limited resources and limited runway.

Leading up to every board meeting, a biotech company’s Clinical Operations team often spends days and weeks preparing accurate forecasts for every board meeting. They need to present to their board with an accurate update with how the trial is progressing, what changes they are making to overcome enrollment bottlenecks, and how each of those changes are impacting the rate of recruitment.

It’s possible that your clinical trial may finish enrollment by the initially planned timeline. However, it’s a lot more likely that your trial will overshoot the initially projected timelines given that more than 80% of clinical trials are delayed because of problems in recruitment. In this case, it’s important to determine by when your trial can expect to complete in various scenarios that could play out.

‍

Levers to Impact Enrollment Timelines

When there are recruitment challenges leading to potential delays in clinical trials today, the typical response in the industry is to add more sites. At a high level, this makes sense — if you assume that each added site could help add “x” additional randomizations per site per month into your study and you bring on additional “y” trial sites, you’d be recruiting “x*y” additional subjects per month into your study.

How do you determine what the right “x” randomizations per month metric is for your trial sites? If you assume the mean or average randomization of your current trial sites, it could be wildly inaccurate — the randomization rates across your sites is most likely not a normal distribution. Rather, a handful of sites are probably making up the bulk of your study’s enrollment, so the randomization rates across sites could be heavily skewed.

If you then apply the median randomization rate to your projections, that may not be accurate either. There’s no guarantee of that all your sites will continue to perform at the same exact rates, and so you may need to account for your sites suddenly randomizing at the 25th percentile, 75th percentile, etc., in order to account for various scenarios that could play out.

It could also take a few weeks or months for the new sites to start up and activate before they can contribute meaningful randomizations into your study. To accommodate this, you need to properly account for the lag in new randomizations that would come from your new sites in your projection model.

In addition, to add new sites to your study, you may need to allocate your clinical trial budget. If you’re not able to expand your budget, you may need to examine which trial sites are unlikely to suddenly start recruiting at a rapid pace after lackluster performance thus far, especially if those sites are not where your Key Opinion Leaders (KOLs) are. However, how can you ascertain that removing those sites won’t negatively impact your enrollment projections?

‍

Leveraging Real-Time Data From Your EDC

Many biotech companies that we have seen are compiling all this information manually into their spreadsheet enrollment models. It needs to be maintained frequently and in an error-free fashion. This could work if there is a dedicated person on your team to dedicate hours each week to updating the spreadsheet tracker. But the spreadsheet model will only be accurate as of the last time that it was updated and lag behind your EDC system.

If you have programmers or software engineers at your organization, it’s not difficult to set up an automated pipeline. Check if your EDC system has Application Programming Interface (API) available. An experienced programmer should be able to quickly set up a data pipeline to fetch data from your EDC using the available API and automatically update your spreadsheet trackers. If you are using Smartsheet, they also have a robust set of API’s that you can use to programmatically access and manage your team’s Smartsheet.

Older EDC systems may not have API available, but you can also check if your EDC vendor allows you to use FTP/SFTP or other file-based transfer protocols to fetch updated enrollment information packaged up into CSV files. While it may not be as straightforward as using API’s, an experienced programmer should be able to set up a data pipeline to fetch updated CSV files from the FTP/SFTP server. There may be a few additional steps required to pre-process the data to ingest them properly from CSV files, but this shouldn’t be more than a few lines of code.

By setting up an automated pipeline to fetch enrollment data from your EDC, you can keep your spreadsheet projection model always up to date with the latest site performance.

‍

How Innovative Biotech Companies Simulate and Forecast Trial Timelines

Many innovative biotech companies are working with Miracle, where we provide our suite of integrations to clinical trial platforms to set up a data sync and get everything up and running in just a few days. For example, Miracle makes it easy to integrate with a wide variety of EDC systems from across the industry, including Medidata Rave EDC.

Thanks to our robust integrations, we’re able to provide an automated, interactive enrollment projections tool that overcomes the shortcomings of spreadsheet-based models. Because Miracle is able to track the performance of each site over time, we can visualize percentile ranks of randomization rates as confidence intervals for future projected enrollment.

‍

Users can simulate what would happen if they closed certain sites, got rid of sites with 0 randomizations, or added new sites. For instance, this is what the projection would look like if we were to add 20 new sites at 0.5 randomizations per month.

‍

They’re also able to see the impact on recruitment if it takes their research sites a few months to activate before they’re able to contribute meaningful enrollment numbers. If we change the “Number of Months to Activate Site” parameter to “5 months”, notice the kink in the line graphs kicking in at the 5-month mark.

‍

Once you have all your data in Miracle, creating an automated forecast model is simple. We can add any additional simulations based on your data, such as removing specific sites or visualizing how removing lowest performing sites would impact your forecast. By not only relying on mean and median values but also breaking down various percentile randomization rates of sites, Miracle illustrates different scenarios that could play out, providing insightful information into the potential best-case and worst-case scenarios for enrollment.

Here is a recap of how Miracle’s projection tool is helping biotech companies:

1. Dynamic Modeling: Transitioning to models that automatically update with new data inputs enables a more accurate reflection of the trial's current state, facilitating timely adjustments to strategies and operations.
2. Site-Specific Analysis: By analyzing performance data from individual sites, companies can move beyond the limitations of average rates to uncover actionable insights, allowing for more precise projections and interventions.
3. Regular Updates and Reviews: Establishing a routine for continuously updating projections ensures that any deviations from planned trajectories are quickly identified and addressed, maintaining the trial's alignment with strategic goals.
4. Stakeholder Communication: Real-time data facilitates transparent and timely communication with all stakeholders, including clinical operations teams, CROs, and the board. This transparency is crucial for collaborative decision-making and aligning expectations.
5. Board Meeting Prepartion: Clinical Operations teams often export the dashboards as images or take screenshots to simply insert into their board slides, saving them all the time it would have taken to work with spreadsheet models and producing graphs.

‍

If you’d like to try out Miracle, please get in touch. We’d be happy to show you a demo of the Miracle platform that is powering a number of clinical trials around the globe, ranging from smaller studies with just 10-20 sites recruiting for 100 patients to global, large-scale studies with hundreds of sites enrolling thousands of patients. In just a few days, see how Miracle can level up your clinical trial management, and ultimately, save time.