Using data analytics to evaluate clinical trial network performance: a look “under the hood”

Date: 10. 06. 2019

As the clinical research environment becomes more complex and competitive, pharmaceutical companies are increasingly relying on networks of physicians to conduct their critical trials. The appeal of these networks is understandable: rather than evaluate and assemble a team of individual investigators one-by-one, engaging an existing network of investigators can, theoretically, expedite the processes of protocol design and patient recruitment, thereby enhancing trial efficiency.

The network approach is particularly prevalent in the oncology category, where a handful of prominent industry-affiliated and independent networks are increasingly driving clinical trials. But the power of those networks is limited by the relevant experience of the individual physicians. Indeed, our database suggests that engaging a large, well-known network of investigators does not necessarily translate to clinical trial efficiency. In other words, before you hand the keys of your trial to a network driver, you need to look under the network’s hood, where you can find valuable information about membership and historical performance.

There are roughly 14,000 oncologists in the U.S., but only about 8,000 have been involved in oncology clinical trials. Whereas many oncologists belong to affiliated networks, their decision to join these networks may be based more on their clinical or academic orientation than on interest, experience, or expertise in conducting clinical trials. For example, one of the most prominent independent oncology networks in this country claims to have approximately 500 investigator-members, but only about 200 have actual clinical trial experience, according to the Phesi database.

Notably, none of those 200 investigators appears in a list of the top 10 enrolling trialists across eight of the most active arenas in oncology research: breast cancer, colon cancer, liver cancer, lung cancer, lymphoma, melanoma, prostate cancer, and renal cancer. Undoubtedly, this network’s membership includes numerous eminent physicians, some of whom may be among the leading experts in their respective tumor-type specialties. However, as a whole, the network has an underwhelming track record in terms of enrolling patients into oncology clinical trials. What that means in practical terms is that if a pharma company chose to partner with that network as a supplier of sites for its multicenter clinical trial, the company risks incurring significant delays as site performance starts to dip, the roster of sites expands, and the cost per patient escalates.

It may seem simplistic to reduce clinical trial success to an equation, but to a great extent, the following formula applies: a well-designed protocol plus a roster of top-performing sites equals faster delivery of a safe and effective medication to patients. Without either of the inputs to that equation, a clinical trial program may languish. Even when a trial’s protocol is sound, it is crucial to identify and assemble a roster of sites with relevant experience, a high level of interest, availability of applicable resources, and a track record of high performance in terms of recruiting and retaining patients. By the same token, a poorly designed or executed protocol may result in onboarding of too many sites, possibly sparking heightened competition with other trials for the same limited pool of patients.

But a one-time snapshot of network performance may not be a fully accurate yardstick; networks can evolve, and their performance can improve over time. Consequently, an automated, dynamic data analytics platform can enable continuous measurement of network performance over a given time period, yielding actionable insights without expending significant resources.

As clinical development programs progress, pharma companies may find that certain networks are evolving along with them. Some networks may be particularly well-suited for conducting post-marketing studies, but less so for earlier-stage trials.

The key for any sponsor of a multicenter trial is to have a partner who can offer advice on all aspects of a networking strategy, with insights on the potential impact of that strategy. To that end, data analytics can help answer a number of important questions, including:

  • How can we complete clinical trials quickly?
  • What is the value of a network?
  • Which network performance metrics are the most valuable and insightful?
  • Should we partner with an existing network, or build one of our own?
  • If we partner with an existing network, should we choose a commercial network, or one used by one or more of our competitors?
  • How does the performance of a given network compare to those of peer groups in our industry?
  • How does one build a network?
  • Is our network strategy evolving? How should it evolve?

Grappling with those questions need not be arduous or expensive. Phesi is the only company that has the data and predictive analytics capability to inform the right clinical trial design and investigator site selection. Don’t hesitate to ask us for help if you feel the need to look under the hood.

Related posts
04. 28. 2022
Why do nine out of ten clinical trials fail, and how can the industry learn from these mistakes? Read more
04. 01. 2022
What steps can clinical development organizations take to mitigate the risks after the war in Ukraine? Read more
03. 02. 2022
Phesi’s reflections of SCOPE 2022 Read more
12. 01. 2021
New diversity data from Phesi shows inclusion of Black and African American patients in clinical trials improving, but Asian, Hispanic and Latino groups significantly and consistently underrepresented Read more
12. 05. 2019
Delivering nash trials in a competitive world using clinsite Read more
08. 26. 2019
Pharma and biotech are waiting for 21st century clinical trial data analytics engines Read more