Five Critical Challenges in Investigator Site Selection and How Clinical Data Science Can Solve Them

Clinical development faces mounting pressure to deliver therapies quickly and cost-effectively. One way for sponsors to achieve this is to ensure they are selecting the optimal investigator sites for every trial. But this is more difficult than it seems, as the complexity of clinical development has increased––with traditional methods of selecting investigator sites often resulting in inefficiencies, high costs and increased patient burden.

How can researchers overcome major challenges in investigator site selection?

Main Takeaways

Poor investigator site selection is one of the main causes for difficulty during the drug development phase.
Researchers must solve for a variety of issues, including unreliable site performance predictions, a lack of comprehensive real-world data, and enrollment and diversity downfalls.
AI-based technology using clinical data science has the potential to provide solutions.

Analysis from Tufts reveals that poor investigator site selection can severely impact clinical trial success, with 11% of sites failing to enroll a single patient and 37% under-enrolling. Meanwhile, Phesi research of oncology trials found nearly 1 in 5 investigator sites enroll only a single patient. Based on a $40,000 site activation cost and $3,000 per month to manage a site, an investigator site running for 30 months with just one patient will cost $130,000. In contrast, the best performing sites operate at approximately $14,000 per patient for similar study durations – just over 10% of the cost per patient at single patient sites.

It's clear, then, that poor investigator site selection remains a significant driver of spiraling costs and inefficiencies in clinical development. But why exactly does investigator site selection so often fall short? Advances in clinical data science and analytics are starting to give us a clearer picture of the five most common challenges that sponsors face at this crucial stage of clinical development. The good news for sponsors is that clinical data science can also help to solve these problems:

1: Inaccurate feasibility assessments

Typically, sponsors assess investigator site nominations by sending feasibility questionnaires to hundreds, if not thousands, of potential sites. Writing these questionnaires and collating the responses requires a huge amount of time and resources from both sponsors and investigators. It also represents a huge duplication of effort for investigators, who might be asked the same questions multiple times by different sponsors. The quality of the responses is variable, given that feasibility assessments rely on self-reported questionnaires. Investigator sites can overestimate their ability to recruit patients, while critical factors like competing studies, access to essential facilities and patient population access are often overlooked, leading to underperforming sites, escalating costs and the need to rescue studies through the introduction of costly CRO change orders.

2: Unreliable site performance predictions

Sponsors struggle to predict investigator site performance accurately. Many rely on subjective criteria when making their selections––such as linear extrapolation from previous trial phase recruitment rates, prior relationships, institution prestige or simple proximity––rather than objective data analysis using quantifiable factors like an investigator’s trial experience, enrollment performance and site activation speed.

3: Lack of comprehensive real-world data

Without access to comprehensive real-world data (RWD), such as dynamically updated records from investigator sites across the world, study teams risk making decisions based on outdated data or may have to rely solely on CRO recommendations. As a result, sponsors could be missing out on high-potential investigator sites they are currently unaware of, or deploying sites with little or no access to relevant patient populations. They could also misjudge patient availability, leading to site saturation and recruitment delays.

4: Enrollment and diversity shortfalls

There is often misalignment between a trial protocol, the patients involved in a trial and the actual patient population of the indication under study. For example, an investigator treating Parkinson's disease patients to relieve related symptoms cannot identify and treat Parkinson's disease patients who undergo brain surgery and are followed by dopamine-releasing stem cell transplantation. Similarly, a doctor that routinely treats community-acquired pneumonia patients with antibiotics generally does not have access to pneumonia patients being rushed to emergency rooms or intensive care units.

Preparing for the Pharma C-Suite

This is typically the result of sponsors and CROs repeatedly relying on the same geographic regions and investigator sites for their clinical trials, or selecting investigators without the right level of experience or access to the desired patient population. This limits recruitment to populations that may be overrepresented in the disease area as a whole but are underrepresented in a specific region – and can also lower minority participation in clinical trials where it is needed.

5: Regional regulatory variability

Global trials face the challenge of variable regulatory and compliance environments across regions. Each country or region has different ethics committee requirements and regulatory approval timelines for starting an investigator site. For example, in the US it is possible to use a single ethics committee (or IRB) for an entire multicenter trial, while Germany requires a separate ethics committee review for each individual investigator site.

Navigating this variability is a complex task during site selection. Sponsors must consider not only where the patients and investigators are, but also how quickly those sites can be activated and whether they will adhere to protocol and regulatory requirements without major issues. A lack of data-led decision making in this area can have a significant impact on the length of the trial, given that site activation consumes a large proportion of the time needed to fully enroll a trial.

How clinical data science is providing solutions

To address these challenges, sponsors can harness sophisticated AI-driven clinical data science methods. With increasing access to historical clinical trial data and other RWD––such as observational studies, disease records, pharmacy records and claims data––combined with AI tools capable of collating and analyzing this data, sponsors can implement the following solutions to optimize country and investigator site selection:

These solutions enable sponsors to augment and quality-check their country and investigator site nomination lists, while greatly saving time for teams by rapidly eliminating investigator sites where there isn’t sufficient data to ensure performance.

Investing in data-driven methodologies for investigator site selection in this way can dramatically reduce trial costs, shorten cycle times and ease patient burden––ensuring efficient progress from clinical trial initiation through to market launch.

Many areas of clinical development are now benefiting from a greater understanding of the benefits of clinical data science, and investigator site selection should not be an exception. By embracing new ways to build efficiencies, lower costs and enhance accuracy, the sector can ensure that innovative treatments get to patients faster and at considerably lower costs.