Search the Site

Episode Transcript

Back in the 1980s, not long after she graduated college, Perri Morgan went to go visit her 2-year-old nephew, who was sick in the hospital.

MORGAN: The rest of the family was dying to get out of there. And I found myself actually kind of not wanting to leave. There was a piece of me that said, “I can do this. I have the, personality type to do this and this is what I want to do.”

Perri had grown up on a farm in rural North Carolina. She was part of the first generation of her family to even go to college. The idea of working in healthcare was appealing, and also — overwhelming.

MORGAN: I was having a hard time getting my head around four years of medical school and a long residency. And somebody told me about the P.A. profession.

“P.A.” stands for physician assistant, although as Perri will explain later, it won’t for long. Anyway, odds are at some point, you or a loved one has been, or will be, seen by a P.A. in a medical setting. In 2021, there were more than 158,000 physician assistants practicing in the U.S., an increase of more than 6.5 percent from just the year before. But back when Perri was starting her career, things were different.

MORGAN: I’m a little embarrassed for my students who work for years now to get into P.A. school to say that I learned about the profession in February and I enrolled in August, of the same year. 

The P.A. profession was established in the 1960s, around the same time as its sort of sibling profession, the nurse practitioner, or N.P. Like physician assistants, nurse practitioners have become a fixture in most healthcare settings: clinics, hospitals, nursing homes. And their numbers are rising even faster than physician assistants: by 9 percent from 2021 to 2022.

CHAN: The U.S. by far is, the leader in terms of number of nurse practitioners. It’s almost like an order of magnitude larger than the next highest countries, Netherlands and Canada. 

That’s economist and physician Dr. David Chan from Stanford University. Like a lot of people who work in and study healthcare, Dave has wondered how this explosion of P.A.s and N.P.s could be affecting not only the bottom line, but also patient care. In general, he says:

CHAN: We surprisingly know very little.

From the Freakonomics Radio Network, this is Freakonomics, M.D. I’m Bapu Jena. Today on the show, we’re going to talk about Dave’s brand-new study on the costs and quality of care provided by nurse practitioners, compared to doctors, in the emergency department.

CHAN: These implications from that single provider can kind of echo throughout the next month and beyond. 

Perri Morgan’s research has also compared the work of nurse practitioners, as well as physician assistants, to that of doctors. But she focused on a different type of patient, in a different setting. And reached a different conclusion.

MORGAN: It’s a huge public health problem. We wanted to compare apples to apples.

*      *      *

MORGAN: The P.A. profession was created at Duke back in the 60s by Dr. Eugene Stead, who was chairman of the Department of Medicine at that time. And he saw a need for people who could sort of extend the services of physicians. There’s a little bit of competition. The University of Washington created a similar profession about the same time. I think most people agree it’s Duke. I don’t bother to argue about that.

Perri Morgan is a professor and researcher at Duke University. She’s also a physician assistant, though not for long.

MORGAN: We are changing our names to physician associate, but fortunately we’re still P.A.s, so that makes it a little bit easier. 

The name change was approved in the spring of 2021 by what is now known as the American Academy of Physician Associates.

MORGAN: The physician assistant, the term assistant sort of suggests to people, I think, a lower level of skill. And so I think many of us have always wished that they had started off by naming a physician associate to begin with.

It might still be a while longer until people like Perri officially call themselves physician associates, instead of assistants. Changing the profession’s title is a long-term, complex process that involves regulators, employers, and state and federal governments.

Regardless of what the letters stand for, P.A.s have become ubiquitous throughout medicine. They work in every state in the U.S., and wear all sorts of hats depending on where they work: they take medical histories, diagnose and treat illnesses, prescribe medication, and even perform some procedures. This variation can present a challenge.

MORGAN: It’s one of the things that makes us very hard to study as a researcher because not only do we work in a lot of specialties, but even within specialties we do different things. So, for example, if you think about primary care, we might be pretty much just like a physician and have our own panel of patients. But some primary-care practices use P.A.s to just see the same day walk-ins. And some use P.A.s or N.P.s to do patient education or follow-ups on chronic diseases like diabetes. And so, it is hard to study because if you lump all those people together, you don’t get the truth of any one of those roles.

The physician assistant profession was created in the 1960s in response to a shortage of primary-care physicians. We’re now facing a similar shortage, according to the American Association of Medical Colleges. Can physician assistants fill that gap? Should they? What about nurse practitioners?

Not long ago, Perri and her colleagues started to think about those questions, and also this one: How do patients do when they’re cared for by a physician assistant or nurse practitioner, compared to a doctor? They used data from the Department of Veterans Affairs because:

MORGAN: The V.A. in primary care has a patient panel system where patients are assigned to the panel of either a nurse practitioner, a P.A., or a physician. Each of those primary care providers has a nurse, and other people that support them. So we thought the role was pretty similar. In order to make it even more similar, we limited the study to patients with diabetes in primary care. And so, we followed those patients for two years. The first year was just to establish that the primary care provider was actually their primary care provider, who they saw most of the time. And the next year we looked at outcomes and compared intermediate outcomes of diabetes care, which are how well the blood sugar’s controlled by hemoglobin A1C, how well the blood pressure is controlled, and how well the cholesterol is controlled in those patients.

JENA: Why did you focus on diabetes?

MORGAN: It’s a huge public-health problem. So, it’s one of the most important diseases. We wanted to compare apples to apples. We wanted to compare a similar role for the provider, the P.A., the N.P., or the physician. And we wanted the patients to at least have some similarities. Another nice thing about diabetes, if you’re a researcher is that it does have outcomes that can be measured in the relatively short term, we don’t have to wait and see if people have long-term complications. And we found no meaningful difference among the three groups.

JENA: Did you find anything in terms of resource use or costs?

MORGAN: Yes. One of the other reasons we were happy to use V.A. data is that you could actually calculate total cost per year per patient, which is very difficult to do in other settings. And to our surprise, we found that — well first I need to pause a minute and say, I think that P.A.s and N.P.s might be expected to see patients who are less complex. So it’s very important if you’re dealing with a patient who’s less sick, they’re gonna end up less sick at the end of the study, right? We were able to balance for all kinds of things, number of chronic diseases, health factors, but also they have data on socioeconomic factors. So after balancing for those factors, we found that the patients of P.A.s and N.P.s actually cost less per year than the patients of physicians. We were surprised. We expected that to be the same.

JENA: So similar outcomes, but lower costs.

MORGAN: Yes, and the difference in cost showed up in hospitalizations and emergency department visits. The patients of P.A.s and N.P.s were less likely to have both of those, so they ended up with lower cost.

JENA: What is your thought as to why you saw lower rates of hospitalizations and E.D. visits among P.A.s who were taking care of patients with diabetes? 

MORGAN: Yeah, we don’t know. And the study wasn’t designed for that. I’m a P.A., we had a nurse practitioner on the team, and we had a physician on the team. The physician said —”You know, when you go through long medical training like physicians do,” he said, “we’re very comfortable with hospitals. In a way, if we’re worried about a patient, we’d like to see them in the hospital.” That may just have been his opinion. We asked around quite a lot when we presented this work, “Does anybody have an idea why this is true?” It could be residual confounding, meaning that the patients of physicians were still sicker than the patients of P.A.s and N.P.s in some way that we couldn’t measure and account for. 

JENA: Ultimately what really matters is, all right, is that hospitalization for a patient good or bad? Because if P.A.s are not sending patients to the hospital or the E.R. and that leads to worse outcomes, then that’s a bad thing. But if it doesn’t lead to worse outcomes, well maybe those hospitalizations or E.R. visits weren’t required in the first place? 

MORGAN: Right. We didn’t find a difference among provider types in reason for admission to the hospital or emergency room departments. The other thing that I thought might happen is that physicians often have — and some P.A.s do —but by and large, physicians have more outside responsibilities. So in addition to taking care of patients, they’re teaching, they’re running a residency, they’re doing some other administrative task, and they might be harder to reach than those P.A.s and N.P.s who are just sitting there, five days a week seeing patients all day long. And so, if you can reach your primary care provider and say, “This is going on. What should I do? Should I go to the emergency department?” And that provider, is available to you and says, “You know, why don’t you try this first?” you may avoid some admissions.

JENA: Can you imagine a scenario where the quality of care provided by P.A.s and M.D.s would perhaps differ depending on the medical scenario? Because, you know, when I think of diabetes care part of the treatment is very protocolized. It’s identifying lifestyle factors, exercise, nutrition, other consultants a patient should see, what medication should be used, close monitoring, figuring out when to change those medications. That is a sort of different set of tasks than seeing a patient who comes to you with a fever and you have no idea why they have a fever and you’ve gotta work through that diagnosis.

MORGAN: That’s certainly true. The other thing is I will take a little bit of issue with your characterization of the care of diabetes patients. [BJ^Please, please.] There’s also good solid research that says that the relationship with the patient is crucial to long term management of diabetes. And so that’s gonna vary provider to provider. I think the particular strengths that are needed for patients with diabetes is the ability to establish rapport and build a working relationship with the patient. And that might be a doc, it might be a P.A., it might be an N.P.

JENA: I think that’s a really important point because there is a tendency to sort of lump things together. Okay? What do doctors do? What do P.A.s do? What do N.P.s do? But at the end of the day, a patient has a provider standing in front of them. What do they think about their diagnostic skills, treatment skills? If there’s a problem, do they seek help?

MORGAN: This leads into another piece of research we did that you might be interested in about patient preferences. You wanna hear a little bit, about that one?

JENA: I’d love to

MORGAN: The Association of American Medical Colleges put out a paper and the question asked to the survey participants was, “You need a new primary care provider.” And you call the clinic and you’re told that you can choose either a physician or a P.A. or N.P. And to the surprise of the folks who wrote that article, and to my surprise too actually, 50 percent of patients preferred a physician. Twenty-five percent preferred a P.A. or N.P. And 25 percent did not care. We were very intrigued by that. And so we said, “Well, next time you do that survey, can you add a question after that one saying who they would choose, the one word question, why?

JENA: Yeah, I wanna know that. Yeah.

MORGAN: So, they did. They asked, “Why did you make that choice?” and the main takeaway was that the patients who chose physicians were more often to mention higher training specialization, things like that. The patients who chose a P.A. or N.P. were much more likely to mention, bedside manner and time spent. They felt like they would have a better relationship with the P.A. or N.P.

There are some scenarios, though, where a patient doesn’t get to choose between a physician assistant, a nurse practitioner, or a doctor.

CHAN: What happens to a patient when that patient is as good as randomly assigned to a nurse practitioner versus a physician?

After the break, what can the emergency department tell us about patient care and costs, based on who provides it? I’m Bapu Jena, and this is Freakonomics, M.D.

*      *      *

CHAN: My name is David Chan. I’m an associate professor of Health Policy at Stanford. My background is that I’m an internist, as well as an economist.

JENA: So, you have a Ph.D. in economics? They told me there could be only one. I guess there’s at least 16.

In addition to his day job as a physician-economist like yours truly, Dr. David Chan also sees patients at the V.A. hospital in Palo Alto, California, near San Francisco. Not long ago, something happened at the V.A. that caught Dave’s attention.

CHAN: The V.A. had a directive a few years ago that allowed nurse practitioners to practice independently throughout their whole system.

Nurse practitioners, or N.P.s, are similar to physician assistants, but with some distinct differences. First of all, they’re all registered nurses who choose to go on and receive an advanced degree in a particular area of nursing. Nurse practitioners can practice independently in about half of U.S. states. Like physician assistants, nurse practitioners can order labs and diagnostic tests, prescribe medications, and help patients manage chronic conditions.

Over the last few decades, we’ve come to rely heavily on nurse practitioners, and that reliance is forecasted to increase. According to the Bureau of Labor Statistics, nurse practitioner is the fastest growing profession in the U.S., and the number of people in that role is expected to grow by 46 percent over the next 10 years.

CHAN: There’s a number of different trends that have happened in the U.S. We have increasing numbers of patients that physicians need to deal with in this age of efficiency. The costs of healthcare have grown, such that payers are looking for ways to reduce healthcare spending and rely on providers other than physicians. And we’ve got physicians increasingly going to specialties as opposed to primary care. Nurse practitioners have stepped in to fill that role.

Despite the rapid and sustained growth of nurse practitioners in healthcare, as Dave told us earlier, we know surprisingly little about how effective they are. Do patients who see a nurse practitioner do better or worse than those who see a doctor? It turns out that’s a hard question to study.

CHAN: If we were to compare the outcomes of patients treated by nurse practitioners versus the outcomes of patients treated by physicians, these patients might not be comparable. The patients could have, different underlying health. And in fact, in the data, we do find that nurse practitioners tend to treat patients that are younger. They have fewer comorbidities. They have a lower prediction of adverse outcomes like mortality just based on their characteristics alone.

To accurately measure the effects of being seen by one provider or another, Dave needed a randomizing device — a medical setting where someone may not get to choose the kind of provider they see and where the provider may not get to choose the type of patient.

CHAN: We are studying the emergency department. So for our setting, there’s some heterogeneity in whether a given emergency department in the V.A. employs nurse practitioners.

JENA: In your study, do you compare outcomes of patients treated by N.P.s versus doctors, or do you do something different?

CHAN: We don’t just compare the outcomes. In this paper, we are making use of a quasi-experiment that uses the patient arrival to the E.D. as a randomizing device. We’re using the availability of an N.P. as opposed to whether the patient is treated by an N.P. And you can kind of think of this as, you’re assigned to an arm of the trial. 

JENA: And what you came up with, is this idea that within a given E.D., schedules just vary. There may be some Mondays in August where there’s more N.P.s working in a given E.D. and some Mondays in August where there are fewer N.P.s who are working just for random reasons. And patients don’t know that when they decide to go to the E.D. And so, what you’re doing is relying on, this quasi-random variation in the availability of N.P.s in any given day, at any given hour. And so you can use that to say, what is the causal effect of being treated by an N.P.?

CHAN: The other thing that I think is very important is that doing a randomized trial might not be feasible or even desirable. What we’re interested in is how the patients are actually assigned. A big part of the paper is asking what’s the causal effect of being treated by an N.P. versus a physician, we’re actually interested in, a bigger question, perhaps more important question of how best to use N.P.s in a healthcare system.

JENA: I think that’s a really important point. What, are the outcomes that you studied?

CHAN: So we are, focusing on, several outcomes, and you can think of these as decisions that a provider might make in the emergency department. So the main outcomes there are length of stay and the cost of care. Then we look at the decision to admit that patient or not. Then we look at whether a patient has a 30-day preventable hospitalization, and we also look at mortality. And then, when we want to kind of unpack various mechanisms, we also look at specific types of orders that an N.P. versus a physician might order.

JENA: So, what do you find in terms of cost of care, resource use and outcomes?

CHAN: We find that on average, N.P.s use more resources in the emergency department settings. They keep patients longer and they spend more resources measured in dollars. Then when we look at outcomes after this specific stay in the emergency department, look at admissions, preventable hospitalizations and mortality, we generally find worse outcomes. So for preventable hospitalizations, N.P.s have more of those on average, and when we look at mortality and admissions, we don’t find any differences. But when we look at the sickest patients, we do find some kind of signal in the direction of worse mortality outcomes for some very sick patients, such as those with sepsis and a higher propensity to admit for the sicker patients.

JENA: And do you find that there are differences depending on the complexity of the patient? So, is it the case that there’s a small gap between N.P.s and doctors for patients who are not that medically complicated, but a larger gap for the more complex patients?

CHAN: Yes. And that’s a very important fact to highlight, in addition to the fact that we are in this emergency department setting where patients will be more complicated than a primary care setting. These gaps in performance tend to be bigger for patients that are more complex, patients with more comorbidities, and patients that are more severe, patients that have a higher predicted mortality.

JENA: What does your study say about cost? Because you mentioned that N.P.s have higher resource costs, they have higher length-of-stay, the patients are more likely to be hospitalized for preventable reasons after the E.D. visit.

CHAN: If you were just to compare the costs of patients treated by N.P.s versus the costs of patients treated by physicians, there are differences in underlying health and complexity such that you would have lower patient care costs for patients that are treated by N.P.s, but what we do differently in this study is that we have patients that actually do get seen by both NPs and physicians, depending on the availability of N.P.s. So, we find that there are higher costs for those patients. Then the second thing that’s very striking is that there are large differences in salary between N.P.s and physicians, and that might have been, a reason why it’s cheaper to provide care with an N.P. versus than with a physician. The salary of an N.P. is about half that of the salary of the physician. But when we look at our quasi-experiment in the emergency department setting, we find that on net, it’s more expensive for the hospital to hire an N.P. than for the hospital to hire a physician to treat that patient

JENA: When you study the impact of being seen by an N.P. on outcomes or costs of care that happen once the patient leaves the emergency department, how does that occur? The N.P. is not continuing to provide care afterwards, it’s a different team. Let’s say if the patient is hospitalized, like how are those costs coming into play?

CHAN: I think the other striking takeaway is that, you know, that single episode of your visit to the emergency department can lead to downstream implications in terms of health outcomes and costs and that these can be quite large even though it’s a single provider seeing you at that one point in time, these implications from that single provider can kind of echo throughout the next month and beyond. 

JENA: Is it possible that N.P.s are better suited for certain types of care? For example, better suited for chronic disease management rather than the kind of acute patient care that the E.D. demands, just on average?

CHAN: Absolutely. Our paper is specifically in the emergency department setting. We can’t really say, what it would look like in the primary care setting, but we could say that even within this emergency department setting, we see that there are differences, as I mentioned earlier, patients that are more complex, that have a greater severity, there is a bigger gap in this emergency department setting. If you were to extrapolate this into other settings the difference between nurse practitioners and physicians might be much smaller or they might even go in the opposite direction, where nurse practitioners might have better outcomes than physicians when the job is less medically complex and when the outcome of interest isn’t something like preventable hospitalization, it could be something like patient satisfaction, for example, or adherence to guidelines.

JENA: What are the questions that this research raises that should be answered but haven’t been yet?

CHAN: I think the most important questions are how should healthcare systems use this additional labor like nurse practitioners and physician assistants given that we have rising demand for healthcare and we don’t have the physician supply to meet all of this demand efficiently. And rather than asking the question of should this patient be treated by a nurse practitioner or a doctor, I think a very natural question is, how best can they both collaborate on the same team for a given patient? Or how should they divide patients such that it’s an efficient division of types of tasks to worker skill? I think those are all very large and important questions in healthcare delivery that remain unanswered.

The American Association of Medical Colleges insists that by 2034, we won’t have enough doctors to treat all the patients in the U.S. If physician assistants and nurse practitioners can help fill this gap, which is projected to be as high as 124,000 physicians across primary and specialty care, it seems like a good time to understand how, where, and when to use them. Here’s Perri Morgan again:

MORGAN: I think P.A.s especially, but also N.P.s, are more nimble in changing to emerging health workforce needs because we don’t specialize. All of our training is a basic training. So a while back, occupational medicine, it was kind of a new field and a lot of P.A.s went into that field and they could do so quickly.  A lot of P.A.s went into H.I.V. care. The team was headed by an infectious disease doc, but he needed a lot of people, and who could move over and learn that one specific thing to do? You know, P.A.s and N.P.s fit that bill.

That’s it for today’s show. I’d like to thank Perri Morgan and David Chan for their time and their dedication to understanding how to best apply the skills and knowledge of nurse practitioners and physician assistants — or “physician associates,” as we’ll soon be calling them, to make healthcare better. And thanks to you, of course, for listening!

Here’s an idea I had based on my conversation with Perri and Dave. When patients are hospitalized at a teaching hospital, like the one where I work, they’re often treated by residents, physicians in training. You might wonder if it’s better to be cared for by a trainee, or by a physician who’s completed their training. Trainees have fewer years of experience, which might concern you, but during residency, seeing patients in the hospital is basically all they do — so they become very good at it. It’s a hard question to study because patients typically aren’t randomized to either group. There are times, though, when fewer residents might be available in the hospital and more care has to be provided by attending physicians. Maybe a lot of trainees are taking an exam that day, or they’re on vacation, or called in sick. Could the availability of trainees in a hospital on any given day tell us something about their effect on patient care?

Think about it and in the meantime let us know what you thought about today’s show. Have you seen a nurse practitioner or a physician assistant instead of a doctor? Why’d you choose one over another? Did you get to choose? Send us an email at That’s B-A-P-U at

Coming up next week: Gastroenterology researchers have long wished for a randomized controlled trial to test colonoscopy as a screening tool for colorectal cancer. A few weeks ago, they finally got one.

SHAUKAT: It’s one of those, be careful what you ask for because you may or may not want to know what it says.

What does it say? And what does it mean for the future of colorectal cancer screening? That’s coming up next week on Freakonomics, M.D. Thanks again for listening.

*      *      *

Freakonomics, M.D. is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and People I (Mostly) Admire. All our shows are produced by Stitcher and Renbud Radio. You can find us on Twitter at @drbapupod. This episode was produced by Julie Kanfer and mixed by Jasmin Klinger. We had help this week from Katherine Moncure. Our staff also includes Neal Carruth, Gabriel Roth, Greg Rippin, Lyric Bowditch, Rebecca Lee Douglas, Morgan Levey, Zack Lapinski, Ryan Kelley, Eleanor Osborne, Jeremy Johnston, Daria Klenert, Emma Tyrrell, Alina Kulman, and Stephen Dubner. Original music composed by Luis Guerra. If you like this show, or any other show in the Freakonomics Radio Network, please recommend it to your family and friends. That’s the best way to support the podcasts you love. As always, thanks for listening.

JENA: Our budget keeps on going up and up and up. Pretty soon you’re gonna be in like the radio music tower, what’s, what’s that called in New York? Radio Hall City— Radio City Music song? Yeah. I got — I got most of the letters, right? Most of the words, right? 

Read full Transcript


  • David Chan, professor of health policy at Stanford University.
  • Perri Morgan, professor of Family Medicine and Community Health at Duke University.



Episode Video