Emergency room study upends traditional thinking about doctors versus nurse practitioners

In emergency rooms, doctors outperform nurse practitioners on average—but that gap tells only part of the story. New research finds that difference narrows for simpler cases, and because performance varies so widely within each profession, a meaningful share of NPs outperform some doctors.  

The findings point toward a concrete strategy for hospitals struggling to cope with a national physician shortage and suggest the debate over whether NPs should be able to practice independently may be focused on the wrong issue.

“The question isn’t whether nurse practitioners should practice independently. It’s which patients they should see,” said Dr. David Chan, co-author of the study and professor of economic analysis and policy at UC Berkeley Haas. “The data show that how you deploy your workforce matters just as much as who they are.”

Published in the journal the American Economic Review, the study arrives as states across the country debate expanded “scope of practice” laws that govern what nurse practitioners can do without physician oversight. And, beyond the debate over nurse practitioner autonomy, it raises questions about whether professional credentials are the best way to judge who should do what in a hospital—or any setting where performance can be measured.

“Professions help organize work, but they’re a blunt instrument,” said Chan, the Mark and Stephanie Robinson Chancellor’s Chair at UC Berkeley Haas. “If organizations could better measure and act on individual performance, the efficiency gains could be substantial.”

An ideal research setting to establish causality

In addition to his role as a professor and faculty director of the Robinson Life Science, Business, and Entrepreneurship Program at Haas, Dr. Chan is an investigator at the Department of Veterans Affairs and co-director of the VA Center for Policy Evaluation. He also practices medicine, seeing patients as a hospitalist at the Palo Alto VA Medical Center six times a year for one-week periods.

For the study, co-authored by Yiqun Chen of the University of Illinois Chicago, the researchers analyzed 1.1 million emergency department (ED) visits at the Veterans Health Administration, which has allowed NPs to practice independently in EDs since 2016. Dr. Chan and Chen took advantage of the VA’s scheduling system: Providers’ schedules are set months in advance, but patients arrive unpredictably. That randomness allowed the researchers to attribute differences in outcomes to the type of provider rather than differences in the patients themselves.

With a design that established causality rather than just correlation, as well as a large sample size, the research represents one of the most rigorous attempts to measure whether the nation’s growing reliance on nurse practitioners comes at a cost to patients or the healthcare system. Critically, the researchers found no statistically significant difference in 30-day mortality rates between patients treated by nurse practitioners and those treated by physicians.

What the data revealed

At first glance, patients treated by NPs seemed to incur lower spending. But those comparisons masked a key fact: NPs are typically assigned healthier patients. Once researchers accounted for this non-random assignment, the apparent cost advantage not only disappeared—it reversed.

On average, patients treated by NPs in the study experienced 11% longer emergency department stays and a 20% increase in preventable hospitalizations within 30 days compared to those seen by physicians. NPs ordered more diagnostic tests and specialist consultations—patterns consistent with responding to greater diagnostic uncertainty—contributing to longer stays and resulting in 7% higher spending.

Compared with physicians, NPs were significantly less likely to prescribe opioids—drugs with high risks if overused—but more likely to prescribe antibiotics, where undertreatment can be dangerous. This pattern suggests that NPs, who may have less developed diagnostic skill, are calibrating their decisions to avoid the most dangerous errors.

NPs did not increase overall hospital admission rates on average, but for the most severe cases such as sepsis, stroke, or heart failure, they were substantially more likely to admit patients.

“It would be wrong to say it’s more costly to hire a doctor on average, despite the fact that they have almost double the salaries of NPs, because doctors have expertise and make decisions that save the system a lot of money,” said Chan.

Complexity and experience change the picture

The averages, however, obscured crucial variations. When the researchers examined results by patient complexity, the NP-physician gap was relatively small for straightforward cases. For patients with the least complex cases, the additional costs associated with NP care dropped by roughly 80% compared to average cases. The performance gap also narrowed as NPs accumulated experience, both in general and with specific medical conditions—suggesting that training differences, not innate ability, explain much of the disparity.

A team approach

The study cuts against the grain of how American medicine has traditionally thought about the boundary between the two professions. Physicians undergo four years of medical school plus three-to-seven years of residency, while NPs complete a nursing degree and a one-to-four-year graduate program with no residency requirement. The boundary between the two professions translates directly into pay: Doctors earn about twice what NPs do.

But Chan and Chen found the variation in performance among individual NPs and among individual physicians was several times larger than the average gap between the two professions. And the researchers found substantial overlap in performance: A randomly selected nurse practitioner is more productive than a randomly selected physician in about 38% of pairwise comparisons. This suggests that, despite the stark differences in training between the two professional classes, there exist a substantial share of NPs who perform at or above the level of some physicians, at least for the patients who could have been treated by either class.

One of the striking findings is that the difference in spending between a high- and low-performing physician is several times larger than the average difference between physicians and NPs. Yet hospitals appear to match tasks and pay almost entirely along professional lines rather than individual performance.

“Professional title turns out to be a very coarse signal of productivity,” Chan said.

The pressure to expand autonomy for nurse practitioners stems from arithmetic: demand for healthcare has outstripped physician supply for decades and the NP workforce has grown to more than one-third the size of the physician workforce. About 13% of emergency department visits nationwide are now handled by NPs.

Chan said rather than debate whether nurse practitioners should be granted full autonomy, or ask if medical schools might increase enrollment, health systems ought to focus on building teams that take advantage of each profession’s strengths.

“I think we’re moving from this model where we think of doctors as autonomous professionals who provide all the care,” Chan said. “When you have nurse practitioners in the equation, we need to think about how to best use them, rather than think of them as a replacement for doctors.”

AI could help close the gap

That shift requires investment. Chan said AI could help hospitals and health organizations close the gap between physicians and NPs through intuitive systems that help both types of providers make better decisions. AI could also enable administrators to identify the types of patients best suited to both classes of providers.

Policymakers could accelerate that change, he added, by paying healthcare systems for improving patient outcomes rather than just billing for procedures.

“Such a change may shift the emphasis from turf battles between physicians and NPs on who can bill procedures and turn attention to how they can work together to provide the best care,” Chan said. “I think that would automatically lead the healthcare systems to start innovating in terms of how to design their teams.”

Read the full paper:

The Productivity of Professions: Evidence from the Emergency Department
By David C. Chan Jr. and Yiqun Chen
American Economic Review, August 2026