Real World Data in Clinical Trials | MIT Technology Review & Medidata Solutions

Written in partnership with MIT Technology Review 

Real-world data has the potential to further refine the clinical trial process by making it faster, cheaper, and more accurate, and extending it to post-market surveillance to verify whether new medications and procedures fulfill the promise of their clinical trials.

Medidata is exploring such possibilities with health tech company Datavant, which works with health systems and electronic-claims clearinghouses to collect and link data on patients while preserving their privacy as required by federal law. In this way, researchers can study the health information that’s gathered over time on a single individual, or do advanced analytics on a group of patients that share certain characteristics.

“We’re trying to link data from those clinical trial cohorts to the rest of their real-world data, but in a privacy-preserving way,” says Jason LaBonte, chief strategy officer at Datavant. “So, if you want to understand more about the patients that were in the trial, you’re not stuck if you didn’t collect the data.”

Datavant works with about 400 health systems and other providers and has access to a vast trove of insurance claim data. LaBonte estimates that among its various data sources, the company has at least some health and other information on about 300 million people in the United States.

Health systems, companies, and organizations that work with Datavant run its software on their patient databases to create a de-identified version of the data. The set of data that identifies someone as a unique individual is replaced with a “linking token.” Each institution’s data is encrypted so that no other institution can identify individual patients.

Datavant’s secret sauce is a complex method of identifying the tokens for the same patient from multiple sources. In that way, a researcher doesn’t know the identity of a subject but does know all the subject’s diagnoses and treatments over time—across multiple physicians, hospitals, pharmacies, labs, and even insurers. The researcher can track what happens to an individual clinical trial participant after the trial is over—a type of long-term follow-up that’s typically expensive and not always practical.

“If I have a subject in a clinical trial, and I want to connect all of her lab records and her electronic medical record, and her insurance claims, data linking via tokenization allows us to link all that data,” says Arnaub Chatterjee, senior vice president for products, Acorn AI at Medidata. “We can pull together all those records that exist out in the ether.” Clinical trial patients, however, need to give explicit consent for their data to be used in this manner.

“Everybody who uses our software retains full control of their data and can say no to anything they don’t want to participate in, but the data is safe to share under HIPAA,” LaBonte says. The sheer size of the database allows researchers to assemble a study population that includes almost any set of characteristics.

LaBonte predicts that real-world data will be used increasingly for “pragmatic trials” such as the studies done in 2020 establishing that the malaria drug hydroxychloroquine was not effective against COVID-19.

“That was a good pragmatic trial candidate because doctors were using it in practice, so people could go through their data, find the patients where it was tried, and then find a matching set of patients who weren’t given hydroxychloroquine and call it the control arm,” LaBonte says. “So, you’ve got a trial without ever actually enrolling patients.”

For mid-sized and emerging biopharma companies, Medidata’s 20+ years of expertise, dedicated support, and continuous learning with customers and trials of all sizes helps you stay on the path to success.

Download our white paper with MIT Technology Review, Clinical Trials are Better, Faster, Cheaper with Big Data, for a comprehensive guide to external control arms in clinical trials.