top of page
  • Writer's pictureRebekah Bernard

Transcript - Cochrane's 18 Tall Tales

Updated: Jun 20, 2021

Drs. Rebekah Bernard, Niran Al-Agba, and Phil Shaffer break down the 2018 Cochrane Review "Nurses as Substitutes for Physicians in Primary Care," pointing out that of 9,000 studies reviewed over the last 50 years, just 18 were of adequate quality to include in a review of the subject. Of these 18 studies, just THREE were published in the United States, most contained high degrees of bias, had small sample sizes, were of short duration, and ALWAYS included physician supervision or nurses following physician-created protocols. Bottom line: there is no evidence that unsupervised nurse practitioners can provide the same quality of care for patients.

Rebekah Bernard MD 0:07

Hello, I'm Dr. Rebekah Bernard. And this is "Patients at Risk," a discussion of the dangers that patients face when physicians are replaced by non-physician practitioners. Tonight I'm joined by my co-host, Dr. Niran Al-Agba. She's also the co-author of our book, "Patients at risk: the rise of the nurse practitioner and physician assistant in health care."

Niran Al-Agba MD 0:29

Hi, good evening,

Rebekah Bernard MD 0:30

And I also have a special guest again, Dr. Phil Shaffer, who is a radiologist and has extensive experience in reviewing research studies.

Phil Shaffer MD 0:41

Thank you. Good to be back again.

Rebekah Bernard MD 0:43

We often hear it said that 50 years of data show that nurse practitioners can provide the same care as physicians. In fact, some advocates even say that nurse practitioners provide better care than physicians. But are these claims true? Tonight, we are going to review the best data ever published on the care provided by nurse practitioners: the Cochrane report. The subject was called: "nurses as substitutes for physicians in primary care." Let's start with this, Niran. Can you explain what Cochrane is and why it's important?

Niran Al-Agba MD 1:16

Sure. The Cochrane Collaborative is a UK-based charitable organization that started in 1993. And it uses 13,000 volunteers throughout the world to organize medical research findings and make evidence-based treatment recommendations for healthcare professionals all over the entire world. And really, I'm used to calling it the Cochrane Collaboration because that's the time when I finished my education. And that's what was previously known as, but it's now this Cochrane organization essentially, and its mission is to promote evidence-based evidence-informed health decision making. The group is so well respected that it currently collaborates with the World Health Organization and Wikipedia to provide medical information around the world. And it has been called the single best resource for methodologic research and developing the science of meta epidemiology due to a standardization and regulation process.

Rebekah Bernard MD 2:07

So, Cochrane is extremely well respected in the scientific community and when they publish a review article, the medical world and policymakers take note. So in 2018, Cochrane decided to study the available literature on care provided by nurse practitioners in a report that they called 'nurses as substitutes for doctors in primary care.' Phil, can you tell us about some of the methodologies that Cochrane used in this review?

Phil Shaffer MD 2:31

Certainly, the 2018 paper was an update, they did one in 2005. For that, they had reviewed about 4200 studies, and for one in 2018, they reviewed another 4600 studies. And what they do is say they go through these studies and look at their methods and decide if they're appropriate. And they can do this through their computers and throw out studies that are too small, didn't go long enough, or have other design flaws. And they went through these nearly 9000 studies, and only 18 of them were analyzable.

Rebekah Bernard MD 3:10

So what you're saying, Phil, is that the Cochrane looked back for 50 years of research, basically, from when nurse practitioners were first created in 1965. They found 9000 studies, but of all those 9000 they were only able to pull out 18 that were good enough to be analyzed to answer the question of whether nurse practitioners could provide adequate levels of care.

Phil Shaffer MD 3:32

Precisely, according to their criteria. But as we'll get into, even those a team of very serious problems.

Rebekah Bernard MD 3:41

Before we get into the problems, I want to review what Cochrane summarized their findings as saying - what they released as their final statement was, "the substitution of nurses for physicians is one alternative strategy that may improve access, efficiency, and quality of care. The delivery of primary care by nurses instead of doctors probably leads to similar patient health and higher patient satisfaction. Using nurses instead of doctors makes a small amount of difference in the number of prescriptions written or tests ordered but the difference may not matter clinically. Nurses probably have longer consultations with patients, and regarding the impact of information provided to patients, the extent to which clinical guidelines were followed, and the costs of health care provided by nurses instead of doctors, the results were uncertain - conclusions about these issues could not be drawn." But overall it sounds like pretty positive findings. Niran, do you think that Cochrane that, the summary is is accurate when you actually look at what the study is reviewed?

Niran Al-Agba MD 4:45

Well, it's interesting that you asked me this question because I think the way Cochrane was looking at this is not whether nurses it actually was nurses, not nurse practitioners. A lot of the studies involved clinical nurses or specialist nurses that have one extra year of training. And that's why some of these older studies, it's worthwhile discussing the details. And what's interesting is what they were looking for is, are there aspects of medicine or clinical practice where nurses could be substituted for physicians? And we'll talk about this like the phone triage study, for example, which is really very valuable. It's a little bit like we talked about in one of the earlier podcasts, the Burlington trial where it allowed physicians to have more time to see patients face to face.

So to me when I'm reading it with that, I guess, vantage point, I understand what they're saying. They're saying, Wow, yes, nurses have so much value. And we've talked about this before. Nurses are a huge part of care for patients, but I don't want that part lost, that substituting certain areas, even follow up for straightforward already diagnosed patients. There are roles where I think we need help, and we can expand our access to the underserved population. This study or Cochrane really wasn't asking the question, can nurse practitioners do everything physicians can do and substitute for physicians providing care? So I just want to be clear, I think that that's misleading. But I don't think it was intended that way. I really believe Cochrane was asking, Where can we deploy people to do work that physicians have traditionally done without harming patients?

Rebekah Bernard MD 6:22

So it sounds like what people saw were the headlines, but didn't necessarily delve into the details of what exactly these studies were looking at. And that's what we're going to get into tonight. Phil, can you tell us a little bit about - you've noted that of the 18 studies that Cochran found only three of them were performed in the United States. Can you talk a little bit about that?

Phil Shaffer MD 6:43

Yeah, certainly. one of them was written by Mary Mundinger, we covered that in great detail on the last podcast I was on. So I'm going to sort of gloss over that one and refer people back to that podcast. But suffice to say that we found that what she said she found, she didn't really find. Then there were two others, one written by Lewis in 1967, which as we've discussed briefly before, was so long ago that CT scanning didn't even exist. It used nurse practitioners that had been trained in an older system, and may not at all be relevant to what's going on right now where we have a situation where there's been a dramatic increase in nurse practitioners, nurse practitioner schools, and which don't really have the quality they used to have. And that's not by my evaluation, informed nurse practitioners will tell you that as well.

Rebekah Bernard MD 7:43

You know, Phil, what I think is interesting about the Lewis study was that there were only 66 patients in this study, they were divided into two groups of 33 each, they were mostly women, over 50 years old, and they had one of five different chronic health conditions. The nurse practitioners in the study were 100% supervised, they followed protocols. The charts were reviewed every day by physicians, the goal of the study had only one goal, would patients accept seeing a clinical nurse? And the answer was, yes. So this study actually did not even analyze the care provided by nurse practitioners.

Phil Shaffer MD 8:23

Right. And we can't emphasize your point enough that the nurse practitioners were not seeing these patients on their own and evaluating them on their own, they were helping the physicians very much. And the studies have been conflated to be something they aren't when they're taken to mean that the nurses can take care of patients without any supervision. One other thing that is very interesting is that a lot of these studies in the Cochrane Review, had patient satisfaction as an endpoint. And I kind of get upset when I read this because this is really a measure of the quality of medical care.

For one thing, patients or laypeople, they don't really know how to evaluate how complete their care was a focus on things that aren't really affecting the outcome, such as how pleasant the person was. And a lot in some of these studies, the nurses had twice as much time to work with each patient as the physicians did, and that's obviously going to improve the interpersonal relationship. The other thing is, in March of 2012, a paper published in the Archives of Internal Medicine pointed out that patients who were most satisfied with their care had a 25% greater chance of dying so that if you're very pleased with your care, it didn't mean you were getting good care at all. In fact, quite the opposite.

Rebekah Bernard MD 9:56

Isn't that amazing? And that's something that you would not expect so patient satisfaction is definitely not always correlated with improved outcomes.

Phil Shaffer MD 10:05

No, not at all. It's a marketing measure.

Rebekah Bernard MD 10:09

Niran, can you tell us a little bit about the very best scientific study in the united states that the Cochrane Review included, which was the study by Hemani in 1999?

Niran Al-Agba MD 10:20

Sure, you know, I agree with you, I think that that was probably the best study. And that was more of a question, you know, where, where nurse practitioners can be utilized best. And that study was done in 1999. so at least within the last 25 years, and it was done involving 450 patients, but what's interesting is they used a couple of groups; nurse practitioners, medical residents, and then, what we would call attending level physicians, meaning we've completed our training. And so medical residents tend to finish medical school and then have a few years of work under their belt, usually one to three years. For the attending level physicians, most of them had only seven years of practical practice experience. So they were relatively new to providing primary care. And most of the nurse practitioners had 13, I mean, on average, 13 years of experience. So that five years of education, the physicians had 11 years of education. So I think that's what they were trying to do is have 18 years of education and experience for both groups to try to standardize it a little bit.

And what's interesting is, in this study, as with all the studies that we have ever come across, physicians were required to review and sign all charts of the nurse practitioners and medical residents. And so it was 100% supervised. And if there was a problem with the quality of work or something being done, those mistakes could be corrected. And you know, that's really an important part of medical education. I know when we all went through residency, I'm sure we got called out on the carpet for mistakes or learning opportunities, things like that. And that's a really important part of becoming an expert. We talk about in our book, how feedback about your performance is so so important to improving your skills.

So nevertheless, everyone was being supervised, the nurse practitioners can make their own referrals or tests and develop treatment plans. And what the study ended up finding was, nurse practitioners utilized more healthcare resources than the medical residents in 14 of the 17 outcomes. So you know, when I looked at it, this is something that has held since the Lewis study.

And actually there was another great study in 1974, we talked about in the book where it was very clear, to get the same outcomes, it costs, sometimes twice as much, but clearly, more for the nonphysicians to provide care and have the same outcomes, this position. And it's so interesting that in a world right now, where we're trying to cut medical expenses and health care costs, that we're actually driving up those costs, I think, pretty clearly. And it's a question of whether or not the nonphysician practitioners are a part of that issue.

So compared to the medical residents, they utilize more resources. And they also utilize more resources than the physicians in 10, out of the 17 outcomes, or more lab tests. And then they ordered I believe, Phil was, was just saying, two and a half more times ultrasound CT scans. So imaging studies as a group, which was what we call a statistically significant finding, and what that term means for people who aren't physicians or statisticians is, that means that the study showed enough of a difference that random chance was less likely, or showing that difference. And so a lot of times, you know, physicians are skeptics. And so we want to know, that's a finding was what's called quote, statistically significant know that the probability that it was a finding by random chance is unlikely. And so in that case, they clearly ordered more imaging studies.

Phil Shaffer MD 13:39

I just want to point out that it's one thing if they order more CBCs, urinalyses, those are $20- $50 tests, but the CT scans, MR scans, you know, $3500, you multiply that by two and a half times, and you're talking some serious money here. So is that that was a very significant finding. And it was statistically significant. It's interesting to me that Cochrane when they reviewed this downplayed it very, very much. They did not highlight that. And when I read the paper, it that just shone out to me like a blinking light.

Niran Al-Agba MD 14:19

Well, I'll tell you what stood out to me just because I'm a primary care doc, and a lot of the push locally in this state, Washington State has been to reduce unnecessary hospital admissions. And so to me, I think the most striking finding was that the hospitalization rates were much higher for the patients in the nurse practitioner group. And they did not have patients who were considered higher risk. So those who had the highest risk were actually the patients of the attending physician. So the fully trained, full-fledged board certified physicians, and the ones managed by the nurse practitioners had a 41% higher risk of being hospitalized, which translates into 13 more hospital admissions per 100 patients per year. And to me, that's really significant parts. And every time you hospitalized a patient, you take risks of secondary infections, mistakes. I mean, it's not as simple as it sounds that only spend a few days in the hospital. So I think that's just something that's really important not to overlook.

Rebekah Bernard MD 15:14

So basically, the single best US study included in the Cochrane analysis was the study by Hemani in 1999. With basically the finding that nurse practitioners ordered more tests and had higher hospitalization rates than the physician group. Let's take a minute and talk about the Burlington trial. You mentioned it earlier. Can you tell us a little bit about it? Because this was such a sentinel report. And actually one of the very first studies ever done on nurse practitioner care, and it was performed in Canada.

Niran Al-Agba MD 15:45

Yeah, it's actually one of my favorite studies to talk about, because I feel like sometimes when we get an answer, we need to accept that answer as a great thing. And really build on that. I think sometimes we go searching for a way to tweak it or a way to somehow profit and it's a shame. It was two family practice physicians, and they were running a clinic, and they had become full, so they could not accept any more patients. And I think, Rebekah, you and I are in primary care. I'm full. I'm so full, I'm brimming at the seams. And so I love the fact that they approached the McGill University epidemiology department and said, Hey, we have these amazing nurses that have been with us for years, we really trust and have a high opinion of their skills. And we put some extra education for them, and then figure out if they can do chronic care follow up for patients. And it's so cool, just if you really do sit down and read it, I think it's one of the most frankly, just induces goosebumps just because it's such a great idea that these two docs had, and they had their hearts in the right place, which is about patient care.

And so anyway, they sent these two nurses went to school, and they started doing follow up. And what they found is roughly 55 to 67% of the time, these two nurses could see patients independently. And the patients did very well. And the patients loved it because they knew the staff at his clinic. So, it was like a medical home. I mean, which is a concept that maybe is controversial still, but to me is kind of still important. And so they increase their capacity to accept new patients in the province by 22%. During this time, there are stories about the team, really well known throughout the province for the care that they gave these patients

Rebekah Bernard MD 17:28

Niran, you said that they were independent, but they were these nurse practitioners were always working directly with physicians. Right?

Niran Al-Agba MD 17:35

Right. I'm sorry, what I should say is they would see the patients, follow up with very specific algorithms, and they didn't necessarily need to have the physician come in the room and actually see the patient with them. So I thank you for that. to correct it. They wanted to know, could they increase the capacity, see more patients per day. And so when the nurses had questions, which was about 33 to 45% of the time seeing patients, they would go and get the physician.

Rebekah Bernard MD 18:02

So they were working in very close collaboration together. And I think this is one of the sentinel studies that show that when physicians and nurse practitioners work together, patients get fantastic care. But that clearly it takes that collaboration for that to happen.

Niran Al-Agba MD 18:18

Absolutely. I think it's such a great study.

Rebekah Bernard MD 18:21

And you know, it's so interesting, because the co-author of the Burlington trial was Dr. William Spitzer, and he was actually one of the forefathers of evidence-based medicine. And like, 20 years later, he wrote an editorial for the New England Journal of Medicine, and it was titled "The nurse practitioner revisited - slow death of a good idea." And he expressed concern in that there was not good quality research on nurse practitioners. And he said that we have to really be careful not to take shortcuts when we're doing randomized controlled trials. And I know, Phil, you've pointed out that that has been a major problem, and a lot of the studies that we're seeing now

Phil Shaffer MD 18:59

Oh, yeah, the quality of the research that's coming out is, is just not good. It. To be honest, it looks as though there's a directed outcome often.

Niran Al-Agba MD 19:11

I think it's interesting that he predicted that science or medicine would potentially take shortcuts, and he warned against that. And I think it's a really important piece of advice.

Rebekah Bernard MD 19:22

We have reviewed the United States studies, we've looked at some of the Canadian Studies, we've talked a little bit about the United Kingdom studies, actually, of the 18 studies that Cochrane reviewed, seven of them were performed in the United Kingdom. I will note that nurse practitioners in the United Kingdom at this time, were always practicing under physician supervision.

Some of the studies that were included in the United Kingdom were telephone triage studies. But there was a study that was done was a randomized controlled trial in which nurses were actually clinical nurses, not nurse practitioners, and they were asked to evaluate patients with minor medical complaints, the researchers really said that there wasn't enough evidence in the study to even come up with any conclusions because the study group was not large enough to detect any rare outcomes. And this is one of the problems with some of these studies is when you're looking at low-risk medical problems, you have to have an enormous sample size to actually see, you know, people aren't going to die if they come in probably with a cold, or, you know, a little low back pain, right?

Niran Al-Agba MD 20:30

I think you're talking about the Shum study in 2000. Yes, and that was five different practices, which is kind of interesting, in a semi-rural, suburban, and urban setting. So they kind of went across all settings that clinics were in. And again, a lot of these studies talk about triage for minor medical complaints, or not undifferentiated patients. So again, having a diagnosed condition. And these clinical nurses actually completed a three-month per time educational course, to learn to manage these minor complaints. And this is very similar to what we saw and talked about in the mundane or study. So they specifically went to get an education to do this for the study, and it was 19 general practice doctors. And the study evaluated only low-risk patients, the high-risk that were excluded children under one year, pregnant women, anyone with severe chest pain, abdominal pain, difficulty breathing, vomiting, blood or fainting, or patients with psychiatric complaints. That's a lot of what walks through, I would say, an average primary care physician's door.

And again, this is another situation where they call it nurses acting semi-independently. And I think that's kind of what I was trying to refer to in the Burlington trial. They would perform examinations take histories, offer treatment, and issue prescriptions. But again, with 100% kind of oversight or ability to get feedback and get consultations from physicians when they needed it. Yeah.

And you know, the fact that they excluded these high-risk conditions is a hallmark of pretty much everything in the Cochrane trial, all the studies that were included, almost always excluded high-risk patients. And unfortunately, that's not how real life looks. And that's not how some nurse practitioners are treating patients, they are dealing with everything, not just low-risk conditions. And even in low-risk conditions. In a nutshell, nearly one in three low-risk patients was it was seen independently by the clinical practice nurses required physician collaboration. And again, we did address this, I think in one of the other podcasts in that when you're looking at a low-risk group, the risk of death as an outcome isn't really a high kind of thing. So if you're taking young people, the chances they're going to die of, say a heart attack or stroke or something unexpected, really isn't a common enough occurrence to judge whether or not these outcomes are avoided. And that's really what the study said they couldn't draw conclusions about the safety and efficacy of this nurse-led care because the study wasn't large enough to find these rare, unusual outcomes.

Rebekah Bernard MD 22:57

One of the things that I think is interesting about Cochrane is that they titled this primary care, you know, replacement of physicians in primary care, but yet, a lot of the studies were actually in specialty care. So some of the other studies that were looked at in the Cochrane in the United Kingdom was a study of patients with rheumatoid arthritis. And again, nurse practitioners were directly supervised and these were following protocols for patients with stable rheumatological disease. They also did a study in the United Kingdom on gastroenterology patients. These were patients that had dyspepsia, and patients had an endoscopy done, and then they were randomized to follow up with either a physician or nurse practitioner. And of course, high-risk patients were excluded. And they found that nurse practitioners could do adequate follow-up of these patients. And you know, these are the sorts of studies that we see there were some Netherlands studies follow up of diabetes, excluding patients that required insulin. There was one study in South Africa in which nurse practitioners were just following up on patients with HIV infection to monitor their antiretroviral treatment. So a lot of these Cochrane studies were actually very specific evaluations and not undifferentiated general care.

Niran Al-Agba MD 24:14

Yeah, it felt to me very much like they were looking for a section of medical practice, where this question was being asked, how can you substitute care and where would you choose to substitute physicians? And again, that's where I think people drawing conclusions and media headlines have taken it a little farther than the Cochrane Library.

Rebekah Bernard MD 24:35

So Phil, what's your takeaway from a lot of this data that you've seen?

Phil Shaffer MD 24:41

Well, in summary, if you're trying to apply it, the way people are trying to apply it now, you just can't do it because the methodology of the papers even in the past they're pretty stringent review just isn't good enough to do that. And one other thing I want to point out about the US studies, in particular, is there is a crossover problem, in that all these studies had a situation where patients were assigned to a group that's NP, and a group that was physician and they were comparing them, but the patients move between them. So you wound up mixing these patients, and who knows what effect the physician had. And frankly, if I were reviewing these, I would throw those out just for that reason. And you're left with no US studies to review. And I don't know if that was a viable result for them. So I think they may have left some in.

But I think that is especially now in 20 years hence when there's such a push for legislators to approve nurse practitioners at basically acting as doctors that these studies are taking total or taken totally out of context. You cannot take a study, such as even the Mundinger study and expand it to mean that people can have full practice authority, and can even declare themselves to be neurologists, dermatologists, or endocrinologists and start their own practice, which actually is what is happening out there.

Rebekah Bernard MD 26:18

Exactly. Well, the Cochrane Review summarizes the best available data on the use of nonphysicians over the last five decades, and again, out of over 9000 studies, they found 18 three of them were in the United States, all of them had variable degrees of bias. And so really, the answer is it's unclear from this data, whether or not nurse practitioners can actually substitute for physicians and that data remains to be seen. It is clear that patients with minor medical complaints and low-risk conditions can definitely be treated by nonphysicians following protocols and with physician supervision. But we definitely need more data before we should be replacing physicians just carte blanc with nonphysician practitioners in order to make sure that patients are treated safely.

For more information about this topic, please check out our book, "patients at risk the rise of the nurse practitioner and physician assistant in healthcare," it's available on Amazon. And also if you're a physician, we urge you to join us at physicians for patient protection. Our website is Thanks and we'll see you on our next podcast.

Transcribed by


bottom of page