The opioid addiction crisis has led medical groups to recommend tightening up painkiller prescribing, but getting physicians to comply has been a challenge.

A University of Pennsylvania study has found a simple way to nudge emergency department doctors in the right direction: set the electronic medical records system to default to the recommended 10 opioid pills, or about a three-day supply, to treat acute pain.

When Penn did that in 2014, the proportion of emergency room prescriptions written for 10 tablets more than doubled, from 21 percent to 43 percent, at the Hospital of the University of Pennsylvania and Penn Presbyterian Medical Center. The proportion of scripts for 11 to 19 pills and for 20 pills decreased almost as dramatically, according to the study, published this week in the Journal of General Internal Medicine.

M. Kit Delgado, the emergency physician who led the study, said over-prescribing is a big problem because some patients become dependent on the drugs, and unused pills that sit in medicine cabinets can be diverted for abuse. "Making it easier to prescribe quantities consistent with current guidelines is an important part of addressing the opioid crisis," he said.

The national death toll from opioid overdoses continues to rise. In Pennsylvania, where Gov. Wolf last week declared the heroin and opioid addiction crisis a statewide emergency, about 3,900 overdose deaths involved opioids in 2016, according to an analysis by the federal Drug Enforcement Administration and the University of Pittsburgh School of Pharmacy.

Pennsylvania's Prescription Drug Monitoring Program, an online database, reported an 82 percent increase in emergency room visits for opioid overdoses from 2016 to 2017.

The Penn experiment with the electronic medical record system compared weekly prescribing patterns for the 22 weeks before and after the default — a preselected computer option — was set to 10 pills of a standard dose of oxycodone with acetaminophen (Tylenol). To choose a different quantity, doctors had to opt out of the default and click the 20-tablet option or manually enter another number.

One unintended outcome, Delgado said, was a decrease in the proportion of doctors prescribing fewer than 10 pills, which fell from 20 percent to 15 percent. And despite more conservative prescribing in both emergency departments, the total number of pills given out did not fall significantly — an outcome the researchers attributed to relatively judicious practices before the study began.

"We were kind of unique," Delgado said. "We were prescribing an average of 12 pills to begin with. There wasn't a lot of room to move the needle."

But on a larger scale, the opportunities are huge. The default-setting approach is now being expanded to 50 hospital emergency departments as part of an opioids study called Reduce. That three-year project is also giving individual prescribing reports to physicians so they can assess their own patterns. "If they realize they're not meeting a performance target, it's very motivating to change behavior," Delgado said.

Beyond opioids, Penn researchers have been testing "nudges" to improve medical care. For example, they have used prompts in electronic medical records and secure text messaging to boost prescribing of lower-cost generic drugs, and increase how many cardiac patients are referred for rehabilitation.

The idea that better decisions can be made simply by guiding decision-makers is the principle behind the Penn Medicine Nudge Unit, created in 2016.

"Nudges can be designed to remind, guide, or motivate behavior," physician Mitesh S. Patel, director of the innovative unit, wrote in an opinion piece in this week's New England Journal of Medicine. "As more health-care decisions are made within digital environments, nudging opportunities expand."