Looking for a ride? Or a dinner reservation? Chances are, there’s an app on your phone that will give you a real-time solution. Consumer industries are being revolutionized as applied data science allows them to match availability with demand. However, if you’re looking for a clinical trial for cancer patient, you may not be so lucky. But why is it so hard to match patients with appropriate clinical trials?
Here’s an example from my own practice: I recently saw a young woman with metastatic colorectal cancer. Her tumor harbored a rare but important mutation in the BRAF gene that made it more likely to grow faster and less likely to respond to conventional treatment. Whenever possible, colorectal cancer patients who have this mutation should be treated on a clinical trial.
Because my cancer center did not have a suitable trial open, I searched for colorectal cancer trials on ClinicalTrials.gov, the national clinical trials registry. I found one listed as open and enrolling, just 8 miles away at Chapel Hill. I emailed colleagues at UNC that evening. But later the following day, they responded that the trial was full and no longer accepting patients.
Why is it so hard to match patients with appropriate clinical trials?
None of this happened in real-time. Ideally, this information should be at the physician’s fingertips as part of the electronic health record (EHR). In reality, it took days to get a definitive answer, partly because the available information was outdated. Although there are many reasons only a small proportion of cancer patients are treated on clinical trials, barriers like these are part of the problem. Trial spots go unfilled, trials close without reaching their goals for patient enrollment, and patients and providers may be unaware of a trial enrolling literally next door.
How do we disrupt this system to better connect patients with trials and improve care? Here are 3 key steps:
- Merge the data streams. Today, EHR data, genomic data, and trial data all exist independent of each other. Data science can provide ways to merge these data streams and give doctors the information that will help patients find the right trial.
- Modernize the ClinicalTrials.gov repository. Although information on ClinicalTrials.gov “is updated throughout the study,” it can still be out of date, as I discovered. What’s more, not all trials are included, much of the descriptive information is free text instead of structured data, and genomically targeted trials can be hard to find (this is especially important as cancer therapeutics increasingly utilize targeted treatment). Well-defined data fields and less reliance on free text would substantially improve search functionality.
- Make it real time. This is the most challenging step, but it’s essential for a fully functional repository. First, trial teams should be provided access to their trial’s page on the registry, and then required to update trial information in real time. This will increase burdens on already-busy teams, but the payoff is avoiding misinformation down the road. Second, EHRs must be leveraged to support this effort, with genomic data automatically entered into the EHR in discrete fields, then linked to discrete genomic eligibility criteria in ClinicalTrials.gov, and then presented in an “available trials” tab that the provider can review.
Importantly, none of these steps require new technology—it’s a matter of using data science to merge existing sources of information. Sadly, our patients (and the drug development programs that could help them) are suffering because healthcare is only now engaging with concepts that our colleagues in consumer service industries have already taken to a sophisticated level. It’s past time for our clinical trials data structure to catch up to the 21st century.