Our pathbreaking research

Randomized controlled clinical trials are considered the gold standard in healthcare research, but even their findings leave big gaps in the evidence for treatments’ relative effectiveness and safety. In the case of hypertension, a problem for hundreds of millions of people worldwide, doctors prescribe dozens of different medications, even among patients with similar medical profiles. 

Typically, a trial is designed to assess the effectiveness and safety of one drug compared to one other. What if instead we could evaluate an entire class of drugs, comparing each medication against all the others simultaneously? 

The best option would emerge from the vast data pool like a buoy bobbing to the surface of a lake. “The concept is, let the data do the talking,” says Kristin Kostka, the OHDSI Center’s director. 

To get answers, OHDSI members launched the LEGEND study in 2018. Taking hypertension as their initial challenge, they collected and standardized—using OHDSI’s common data model, known simply as OMOP—datasets from the U.S., Germany, Japan, and South Korea containing records for 4.9 million patients. Then they applied advanced analytical methods for assessing treatments and outcomes, including heart attack, hospitalization for heart failure, and stroke. 

And lo, a surprise. As reported in The Lancet in 2019, the most popular drug given to first-time users, ACE inhibitors, was by no means the best. Thiazide diuretics had far fewer side effects and got better results longterm, says co-author David Madigan, a renowned biostatistician who is now Northeastern’s provost. “We found that more than 3,100 major cardiovascular events might have been avoided, had the first-time users taken a thiazide diuretic instead.” 

LEGEND investigators have since applied their analytical framework to studies of depression and COVID-19, and is looking at Type 2 diabetes medications. In 2020, the LEGEND team laid out 10 design principles for evidence generation in the Journal of the American Medical Informatics Association. 

The idea is to minimize causal, publication, and other biases, avoid risk distortions, be fully transparent, and invite reproduction, Madigan says. “For OHDSI and the quest for reliable real-world evidence that physicians can use to guide decision making, LEGEND is a watershed in scientific excellence.” 

For OHDSI and the quest for reliable real-world evidence that physicians can use to guide decision making, LEGEND is a watershed in scientific excellence. 

David Madigan
OHDSI investigator; Provost and Senior Vice President for Academic Affairs, Northeastern University