Insurance data reveals drug benefits years faster than traditional trials

A patient with diabetes sits in a waiting room. Their doctor mentions a new medication that might help their heart, but it's not officially approved for that use yet. The evidence exists — scattered across thousands of insurance claims, pharmacy records, real patient outcomes — but it's locked away in databases, unusable. Meanwhile, a formal trial to prove the same thing might take 15 years.

Harvard Medical School researchers have found a way to unlock that data. Instead of waiting for expensive, time-consuming clinical trials, they're using insurance claims and health records to answer questions that traditional research can't afford to ask.

The project, called RCT-DUPLICATE, does something deceptively simple: it mimics the structure of gold-standard randomized trials using data that already exists. "There are very tangible use cases where we would love to have trials, but we'll never have them," says Sebastian Schneeweiss, who co-directs the work. "And at the same time, we're sitting on this huge pile of longitudinal and patient-level data, of which only a very small proportion is used today."

The team, led by Schneeweiss and Shirley Wang at Harvard Medical School, has already validated this approach. Instructor Nils Krüger used real-world evidence to show that certain diabetes drugs provide cardiovascular benefits beyond what their original FDA approvals claimed. The findings matched what traditional trials would likely show — but emerged in months rather than years.

This matters because some drugs will never get formal trials for new uses. The economics don't work. A pharmaceutical company won't spend hundreds of millions to test whether a diabetes medication helps the heart if there's no patent protection or market incentive. Patients with rare conditions face the same wall. But the data exists anyway. Every prescription filled, every hospital visit, every outcome gets recorded somewhere.

The real question isn't whether this evidence is useful — it's whether regulators will trust it. "Do we wait 15 years for more trials to be completed, or can we get more expedited evidence that is well grounded?" Wang asks. The FDA is watching. Early conversations suggest openness to using real-world evidence alongside traditional trials, not instead of them.

For now, the researchers are focused on a different goal: saving lives with information we already have. A patient running out of options doesn't have 15 years to wait.