Optimal site selection has been a long running discussion among operational experts in clinical development. Often this process has many different stakeholders, not always driven by efficiencies, and it has been up to now somehow alchemic in nature. Indeed data, across the industry, suggest that more than 60 percent of clinical trials do not recruit within the originally forecasted timelines, and usually between 20-30 percent of selected sites fail to recruit patients. These failures are partially attributed to poor protocol designs, where unrealistic populations are targeted by regulators or medics within the industry.
Often, sites are selected with a fully defined protocol that includes various mitigation strategies. However, their performances consistently fail to meet their forecast. In summary, these inefficiencies are very costly to the industry. For example, Irecently evaluated that opening a site incurs an approximate cost of $50,000 USD. On a global phase III study with 120 sites, saving the opening of 20 percent of unproductive sites could result in more than $1 million USD.
Additionally, this simple calculation does not account for the better use of internal and external resources. It also doesn’t factor in the advantage of a more patient-centric approach adopted by sites that truly have the right patient population. A mixture of practices and new technologies can dramatically change these current shortfalls in clinical trials.
Site Selection a Complex and Multifactorial Process
The selection of sites should be considered a complex and multifactorial process. It is perhaps the most important part in the preparatory phase of a large clinical trial. Standard feasibility (usually provided by CROs and conducted through feasibility questionnaires) can provide some high level indication, but it is not sufficient ground for any specific selection of sites. In fact, standard feasibility mainly relies on the information provided by site staff and private investigators (PI), and does not account for several other factors.
This information should be collected and considered (in any format preferred) together with more comprehensive data, factoring in the other complex layers of the study. A good example comes from studies that require important internal laboratory results to be successful (e.g. the outcome of a bacterial sensitivity test). In such cases, it’s important to collect and analyze data from a laboratory network.
Additionally, in recent times the NHS, and similar organizations outside of the U.K., has been organizing themselves to provide in depth, free of charge feasibility data on sponsor protocols within their networks. While these services are still in an experimental stage, they already provide additional insights to a standard feasibility, and enrich the amount of information on which to base our site selection. Finally, in our matrices, always include feedback and information from all other stakeholders (e.g. Medical Affairs, Marketing, etc.). Their input should be analyzed with other available data to reach optimal site selection.
The Role of Big Data in Site Selection
Recently big data has been a punch line as a solution to many of the issues in clinical development. While, I believe this is an over-optimistic view, I also think big data can be extremely useful in helping solve some of the issues posed by feasibility assessments and site selection.
Today, large and easily consultable databases can give a large amount of historical information on past studies. This information should be collected and factored in with other data to obtain a more refined site selection process. Particularly, it is important that the companies owning these databases also provide analytical services for specific questions. This service can support sponsors to better define the appropriate sites for conducting a study and go into deeper analysis of protocols and their shortfalls.
Recently service providers, such as Phesi that own very comprehensive databases, offer more detailed and complete analytical services, including protocol analyses, site selection, and timeline validation. The companies owning these databases claim they are updated weekly and feed over hundreds of publicly available internet resources and some that are proprietary.
Using Analytical Services
I have found the analytical services extremely promising in terms of potential to truly improve efficiencies and precision of our feasibilities. The main added value is provided in these cases by the extensive and comprehensive analysis performed on all aspects of clinical trial feasibility. In fact, the analytical part of the service is the key point to make sense out of the large amount of available data. Theoretically, some of the data are accessible to all and the real difference is made by how the questions are asked and answered.
The new proposal for feasibility conceived by other major CROs should work in a similar way, where a large set of data is used to drive the next activities, and improve efficiencies and effectiveness.
While such database interrogations and analysis can be extremely informative, these data should be combined with data collected through the more traditional process as described above. An analysis of the overall information is important to avoid missing essential points in the complex process that feasibility is, and to capitalize on any competitive advantage your company might have.
This is very critical in today’s competitive environment where, according to recent data, there is an approximate 95 percent overlap in site usage among different pharmaceutical companies. How do you intend to differentiate yourself from your competitors? That is the question that can only be answered through the “softer” aspects of your strategy, which go far beyond the hard evidence of previous performance or EC/IRBs approval timelines.
In summary, feasibility is an extremely complex task and it has been underestimated for a long time. This has caused some major shortfalls in the actual performance of sites, with often slower than predicted recruitment, many inactive sites, and shifts in timelines and costs. Overall, this has resulted in missing efficiencies for pharmaceutical companies when conducting clinical trials. The solution to this issue, and to enhanced site feasibility and selection, is to be found in a combined approach between a systematic use of information collected in a more traditional way, and information collected through those systems regarded as big data.
The key point in feasibility is to recognize its complexity. Further, it’s important to analyze in depth all available data obtained from different sources and stakeholders, with or without the support of specialized service providers. Following such an approach can truly enhance feasibility accuracy and predictability, and ultimately contribute to more efficient and effective clinical trials.
Lastly, it’s worthwhile considering how the outcomes of feasibility inform improvements and changes to our plans (e.g. protocol, timelines, etc.). To use the words of the Prussian General Moltke the Elder: “No plan survives contact with the enemy.” In other words, feasibility should not be a one-way tool to justify or legitimize our protocols and strategies within the real world, but rather a two-way feedback loop, which might suggest changes to our protocols or strategies should these not be fit for purpose.
Dr. Yuri Martina
VP Clinical Operations
Aylin Sertkaya et al. Key cost drivers of pharmaceutical clinical trials in the United States. SCT. Volume: 13 issue: 2, page(s): 117-126. Article first published online: February 8, 2016; Issue published: April 1, 2016
Rajadhyaksha V. Conducting Feasibilities in Clinical Trials: An Investment to Ensure a Good Study. Perspectives in Clinical Research. 2010;1 (3):106-109
White paper – Clinical trials and their patients: The rising costs and how to stem the loss; online published Nov. 3, 2016
This article expresses the views of Dr. Yuri Martina and is not necessarily the views of Shionogi.
For any company name mentioned in this article, it is given as an example and it is not intended to represent advertising or any other form of appraisal (negative or positive). This includes the further readings section.