By Mun Keat Looi

Each year, tens of millions of people around the world are diagnosed with diabetes, high blood pressure, or depression. You’d expect that by now, doctors would have settled on a few standard ways to treat these diseases. But you’d be wrong.

A new analysis of common treatments for these three conditions, using a database of 250 million patients’ records from four countries, has found that at least one in 10 patients received a course of drugs that no other patient with the same condition did. In other words, more than 25 million people were essentially being treated by guesswork.

Though that may sound surprising, a lot of the time trial and error is, in fact, how medicine works. The progression of a disease depends on many factors such as age, weight, and genetics. Doctors can’t always know precisely how a patient will respond to a given treatment. The decision about which drug to use, and what the patient will accept, is complicated, influenced by everything from official medical guidelines to what health insurance will cover to what the patient has read on the internet. In many situations, the best a doctor can do is to make a calculated guess, see how you respond, adjust, and repeat.

But the simple fact that such massive analyses are now possible might help reduce this kind of guesswork. In the study, published on June 6 inthe Proceedings of the National Academy of Sciences, researchers led by a group from Columbia University found that 24% of patients being treated for hypertension had unique treatment regimens: That is, no other patient underwent the same sequence and dosing of up to 18 drugsfrom start to finish for three years.

Even depression, notoriously complex to treat, seems to have a more standardized drug regimen than hypertension: About 11% of the patients were treated in a unique way. Diabetes patients are more consistent, with 75% of them starting on metformin, generally the recommended front-line drug. Nearly 30% of those stick to just that drug for their whole treatment. Yet, in the end, even among diabetics, 10% got unique treatment regimens.

The findings drew on 250 million electronic health records, pharmacy records, and insurance claims from Japan, South Korea, the UK, and the US in a newly created collection of databases called OHDSI (pronounced “Odyssey”). In total, OHDSI contains 52 databases holding 682 million patient records—equivalent to a tenth of the world’s population. That makes it a powerful tool for studying how chronic diseases are treated over time in different countries. More analyses like these would mean doctors could know more about what patients are being prescribed and how they respond.

There are limitations to the approach; the information from electronic health records and insurance claims databases isn’t always consistent and can be incomplete. But it’s the first time anyone’s combined such vast amounts of detailed data from different countries. This proof-of-principle study, the researchers say, could be expanded, to improve the way we evaluate treatments, and thus develop better ones.