Introduction
Developing pharmaceutical products is a daunting endeavor. Only 10- 20% of potential drugs ever receive marketing approval (Yamaguchi, Kaneko, and Narukawa; 2021) and for those successful products, clinical testing (i.e., phase 1 – 3 trials) took an average of seven years with an associated cost of approximately $1B (Schlander, et al.; 2021). Given the massive investment of time and money into the research of new drugs and the limited quantities of both within companies, it’s critical to deploy these resources as strategically as possible to maximize the chances of bringing products to market and achieving a return on investment.
While there are myriad reasons for clinical trials to fail and/or terminate prematurely, the largest drivers – accounting for 70% of trials – are poor enrollment and lack of efficacy (Pharmaceutical Technology Enrollment Issues are the Top Factor in Clinical Trial Terminations. 2018. Dec 05). Given these causes, it is of critical importance for any clinical study to optimize its chances for full enrollment and to ensure, assuming a viable drug, that the clinical study can attribute the efficacy to said drug.
Regarding enrollment, it is well known that a substantial portion of sites underperform (37% of sites under enroll and 11% of sites fail to enroll a single subject in a clinical trial (Tufts CSDD Impact Report. 2013. January/February. Volume 15 Number 1.)). Given this reality, it makes sense to embrace high-performing sites, as they give a study the best opportunity to meet its enrollment targets. However, overreliance on these high-enrolling sites can have unintended consequences. In particular, it can make potential efficacy claims no longer attributable to the study drug due to the introduction of a confounding factor. Statistically speaking, a confounding factor is the introduction of a different variable/reason that could be the cause for the observed clinical benefit. When studies rely too heavily on a subset of sites for enrollment, it introduces the possibility of the effect of treatment and study sites becoming confounded. This means that we can no longer tell which is causing the clinical benefit.
Inevitably, the question then arises, “How many subjectsares too many for a site?” Similarly, “Should I have an enrollment cap for siteinon my study?”
What Should the Enrollment Cap Be Set At?
When considering how many subjects are too many for a site, the answer is highly dependent on context (e.g., size of the trial, number of sites, subject population size, etc.). Ideally, each site would contribute to a study in roughly equal proportions. For example, a 100-subject study with 25 sites would suggest that four subjects be enrolled at each site. Realizing that equality in enrollment is unlikely to occur, a common rule of thumb is to take this optimal contribution and allow for 3-5x the value. In this example, 12-20 subjects at one siteares reasonable.
Pros of Enrollment Caps
Evaluating the broader therapeutic landscape (e.g., site geography, institution sizes, site expertise, subject population centers, etc.) and consulting with your statistician is a good first step to determining the site-levelel enrollment cap. However, the second question “Should I have an enrollment cap for sites on my study?” is not as easy to answer. When considering the implementation of an enrollment cap, there are broad positive and negative implications that should be weighed carefully.
Positively speaking, site enrollment caps mitigate the risk of confounding treatment and side effects, which is critical to ensuring that regulators accept the study as evidence of a drug’s efficacy. These caps also require more sites to participate in study enrollment, which can improve the chances of your study having a diverse pool of subjects. Putting these two concepts together leads to a potential timesaving benefit downstream when under regulatory review, as there may be fewer questions about the drug’s performance or the generalizability of the trial data to the subject pool at large. Lastly, site enrollment caps can help site audit readiness status as there are fewer subjects and/or documents to manage at each site. Being “audit ready” will not only make audits go more smoothly but can also de-risk the impact of any potential audit findings if a site’s data needs to be removed from analysis consideration.
From a logistical standpoint, when implementing a site cap, sponsors sometimes lean towards adding more sites to a trial than they would have otherwise. Having these additional sites may lead to a faster enrollment rate than originally predicted, which could shorten the overall study duration and save money. Some sites are more motivated to participate in a trial knowing that they have a better chance to contribute subjects to the research, which can justify their investment into your trial. Lastly, having a cap limits the number of subjects a site is responsible for and how much data a site can contribute. This allows sites to monitor the subjects more closely than they have enrolled and improve subject safety. This also inherently reduces the data entry workload at any given site, which may lead to more timely entry of data, higher data quality, and a faster data cleaning process.
Cons of Enrollment Caps
From a more negative standpoint, site enrollment caps introduce a host of potential logistical, financial, and regulatory challenges. Logistically speaking, some sites will likely under-enroll, and placing a cap on high-enrolling sites can result in a longer trial (particularly if the enrollment caps are low). Compared to a study with no cap, it also means that your trial will ultimately require more sites than otherwise planned. As the number of sites increases, the demands on the study sponsor and/or contract research organization (CRO) also increase. For example, more sites will require additional staff members to conduct activities such as site training and study monitoring. Similarly, with more sites involved in a study, the number of study documents, Institutional Review Board (IRB) submissions, and site personnel also iincrease Start-up times are often delayed because there are more budgets to negotiate, more contracts to sign, more review boards to engage, etc. There is also a higher likelihood of turnover amamongite staff, which increases the demand for Clinical Research Associates (CRAs) to support training and onboarding at the affected sites. From a vendor management standpoint, an increase in sites will likewise require extra planning to ensure sufficient equipment and trial supplies (e.g., lab kits and study drugs)aree available at each location.
Financially, site caps often lead to higher upfront costs. Considering the logistical considerations noted, the greater demands on operational resources, and the need for more clinical supplies, studies with enrollment caps tend to be more expensive initially. Furthermore, if the enrollment projections are slower than they would be without a cap, you may also face cost increases later in the trial due to a longer than-anticipated study duration.
From a regulatory perspective, caps are generally positive. However, they can be detrimental to the diversity of your trial if most of said diversity is coming from only a few capped sites. However, with proper planning of site locations, this risk can generally be overcome by ensuring diverse populations across a larger number of sites.
Lastly, when implementing an enrollment cap, it is important to manage your relationships with the study sites. Enrollment caps can cause frustration among high-enrolling sites if they have more subjects to offer to the trial. In these situations, the cap is viewed as a limit on the revenue the site can earn and the denial of a potentially valuable treatment option for their subjects. Similarly, if enrollment caps are added to the trial while ongoing and/or are deployed selectively, sites may become upset with the changing enrollment dynamics or the perceived inconsistent enforcement of the cap. If sites perceive that they are being slighted in this manner, it can lead to problems such as decreased enthusiasm for your product, the trial, or the company. To minimize the risk of these situations, open lines of communication and a consistent and transparent application of enrollment caps iarecritical. Another helpful risk mitigation strategy, if early in program development, is to offer affected sites/investigators the opportunity to participate in future trials within the program.
Case Study
Background: Two phase III anesthesiology studies were conducted in a staggered fashion with the objective of evaluating the efficacy and safety of a medication to allow for induction of general anesthesia.
Design: 400 subjects per study were competitively enrolled in a randomized, active-controlled trial taking place at approximately 40 sites.
Implementation: Given that each trial site/surgical center tended to focus on a subset of surgeries, an enrollment cap of 25 subjects per site was set to ensure that no site dominated the trial enrollment, allowing for a more diverse group of surgery indications to be represented in the study. Upon reaching the 25-subject threshold, sites could request permission for further subject enrollment. The site would then be evaluated to determine 1) how many surgeries were performed by indication, 2) what potential surgery types were being proposed, 3) how it fit into the broader enrollment picture, and 4) the quality and timely entry of the site’s data. If the proposed cap overage would further support the surgery diversity goal, additional subjects would be approved on a case-by-case basis.
Outcomes:
1. Study #1
a. The study had originally projected a < 6-month enrollment period. However, given site start-up delays, some under-enrolling sites and enrollment caps being enforced, the study ended up enrolling for 3x of the projected enrollment period.
b. Additional sites were required to join the study part-way to supplement subject recruitment to meet the study goals, resulting in an average start-up cost of $60K/per site.
c. Relationships with some investigators soured given the perceived inconsistent application of the enrollment cap. Meanwhile, some investigators declined to participate altogether given the prospect of having their enrollment capped.
d. Given the time delays and additional resources required, this trial resulted in approximately 30% higher costs than originally estimated due in large part to the enrollment cap.
2. Study #2
a. Learning from study initiation and general enrollment pains from Study #1 and some soured principal investigator (PI) relationships, additional sites with different investigators were brought on from the beginning of the study.
b. The revised site mix led to enrollment for the study being slightly faster than anticipated, resulting in a modest cost savings of < 10% compared to the original trial estimate.
3. Studies #1 and 2
a. Both studies achieved a diverse mix of sites and surgery types via the use of enrollment caps to broadly represent how the drug could potentially be utilized.
Conclusion
In conclusion, enrollment caps are a tool that can be used during a clinical trial to minimize the risk of confounding a side effect with that of the treatment under investigation. Other benefits include the potential for smoother regulatory review, better tracking of subject safety, higher quality data, and perhaps a faster overall timeline. Despite these positive traits, caps often carry a unique set of disadvantages commonly resulting in additional costs, resource needs, and more time for enrollment. When planning your next clinical trial, consulting with regulatory and statistical experts is advisable, as they can help navigate these tricky waters and determine if site enrollment caps make sense for your specific program’s development.
Author Details
Scott Mollan- Associate Director of Biostatistics, Rho
Scott Mollan has more than 18 years of experience in clinical and non-clinical statistics across the CRO and pharmaceutical industry. As the Associate Director of Biostatistics at Rho, a global CRO, he provides day-to-day technical and operational leadership to project teams and ensures success by disseminating industry trends and implementing best practices. During his career, Mollan has been responsible for all aspects of statistical support including study design, creation of statistical analysis plans, analysis execution, and results interpretation and communication.
Publication Details
This article appeared in Pharmaceutical Outsourcing:Vol. 25, No.4 Oct/Nov/Dec 2024Pages: 27-29