Report examines big data's value in patient recruitment
By Mia Burns (email@example.com)
Digital health company Tudor Reilly Health has released a new report detailing how big data has the potential to provide pharmaceutical companies with revealing new insights regarding clinical trial site locations to maximize patient recruitment. Clinical trial recruitment in the age of the epatient looks at three distinct therapy areas – schizophrenia, atopic dermatitis, and non-small cell lung cancer and identifies hot spots of online activity at state and city level throughout the United States. Associate Web Editor Mia Burns interviewed report co-author Peter Coë about the findings.
Q: Why did Tudor Reilly Health choose to focus on schizophrenia, atopic dermatitis, and non-small cell lung cancer in this new report?
A: We chose these three conditions as they are all in areas of unmet medical need; they are all a current focus of interest for clinical research; and all have their own characteristic problems in recruiting patients.
Q: What are some examples of the major discrepancies between where clinical trial sites are planned or up-and-running and where prospective trial participants may actually live? Why would this prove to become problematic?
A: Drug trials in schizophrenia are among the most challenging to recruit patients for. In Texas, for example, where search data indicates there’s a hot spot of epatient interest, the vast majority of late stage clinical trial sites (24 in all) are focused on just two cities: Dallas and Austin. Meanwhile, Houston and San Antonio (both of which have higher populations than Dallas or Austin) are the source of a high proportion of schizophrenia-related search, yet they have only five Phase III trials between them.
In other states, such as Ohio, we also found major mismatches between these epatient hot spots and where late-stage trials for schizophrenia treatments were actually located. Columbus and Toledo, for example, which between them accounted for more than a third of schizophrenia-related search activity in the state, had no Phase III trials underway. And we found similar mis-matches of trial sites to patients in experimental treatments for atopic dermatitis and non-small cell lung cancer.
If you fail to locate trial sites near to where patients actually live, you are bound to exacerbate the recruitment problem. This is particularly true in trials for schizophrenia treatments, which – by the nature of the condition – are among the hardest to recruit for.
Q: In your view, why has there been a lack of big data within clinical trial planning? What are some of the potential advantages of including big data within this process?
A: One reason - which we heard more than once from those we interviewed for the report - is a culture in clinical research of trial sites and patient recruitment targets being decided on with barely any data at all on patients (let alone big data) to back up those decisions. Historic patient information can be hard to access or there’s too little of it to be of much use. Site feasibility studies are often inadequate. And sometimes these decisions are not patient-centric at all but based on relationship-building between trial sponsors and their investigators, and at which centers they happen to be based. So a patient recruitment target, for example, may be based on a clinician’s best guess of how many patients he or she thinks they might recruit, rather than hard evidence. It’s not surprising then that the recruitment failure rate is so high.
Bringing big data to bear on these decisions takes professional relationships out of the equation and significantly increases the trial planner’s analytical power; not only to determine where to locate trial sites, but also how and where to target recruitment initiatives, such as traditional and online advertising.
Posted: December 2013