The opioid crisis, with an escalating number of dead and destitute Americans, was declared “a national emergency” last week.

Researchers continue to drill into the foundations of how the still-unfolding national tragedy began. A Princeton University team now points toward the training of the doctors prescribing painkillers – the gatekeepers for many of the life-destroying addictions.

“We find that where a doctor received his/her initial training matters in terms of predicting how likely they are to prescribe opioids,” they write. “Physicians trained at the lowest-ranked U.S. medical schools prescribe nearly three times as many opioids per year as physicians trained at the top medical school.”

The paper is published online by the National Bureau of Economic Research, and is authored by Molly Schnell, a doctoral candidate in economics, and Janet Currie, a professor at Princeton’s Woodrow Wilson School of Public and International Affairs.

The two investigators transposed two sets of data: the prescribing data, and the rankings of the nation’s medical schools.

The prescription data came from QuintilesIMS, a public company that provides pharmaceutical market statistics, spanning from 2006 to 2014. That dataset also incorporates information from the American Medical Association: the doctors’ specialty areas, current practices, and medical schools. The rankings of those medical schools came from the widely accepted U.S. News and World Report annual publication; that included 92 ranked U.S. medical schools, 55 unranked domestic institutions, and hundreds more abroad.

The analyses were split into all physicians, and general practitioners (who write roughly half the opioid prescriptions in America), they write.

Harvard Medical School was the top medical school, and the data showed the general practitioners who trained there wrote an average of 180.2 opioid prescriptions annually. The lowest-ranked medical schools (such as the University of South Carolina School of Medicine, the West Virginia University School of Medicine, and the Michigan State University College of Human Medicine) produced doctors who wrote more than triple that amount (550) per year, they find.

The other results showed a gradient in prescribing. For instance, the schools ranked second to fifth wrote an average of 233 prescriptions per year – still less than half of the schools ranked at the lowest tier. (The schools were Johns Hopkins University School of Medicine, the Perelman School of Medicine at the University of Pennsylvania, the University of California, San Francisco School of Medicine, and the Stanford University School of Medicine).

Limitations to the conclusions do exist, they concede. For instance, the number of patients each physician sees was not clear – and the number and strength of pills prescribed was also not in the data. Also, the medical school category only specifies where each doctors completed the initial medical training.

But Schnell and Currie point to factors indicating that it was the training itself that was driving the increased opioid prescribing. For instance, the disparity between doctors working at the same institutions (but who were school at different places) showed distinct differences. The doctors who received the most training in pain management and related topics after medical school prescribed fewer opioids. And the difference between highest- and lowest- ranked schools was less pronounced in the most recent classes of graduates, they add.

The opioid crisis has been subject to intense scrutiny, since data shows it continues to escalate without true signs of slowing down. But researchers looking to understand how it started have found that the pharmaceutical companies continue to push their drugs – sometimes through payments taking various forms.