Working with midlevels: a new note of caution

Working with midlevels: a new note of caution

A hospitalist explains why midlevels aren’t always the solution

March 2010

Published in the March 2010 issue of Today’s Hospitalist

AS THE DEMAND FOR HOSPITALISTS CONTINUES to outstrip supply, one route to enhance the hospitalist workforce continues to gain currency: hiring midlevels. As either physician extenders or as niche providers, the thinking goes, physician assistants and nurse practitioners will go a long way toward helping hospitalists treat the swelling ranks of inpatients.

But a new editorial comes with this telling title: “Nonphysician providers in hospital medicine: Not so fast.” In the piece, which was published in the February issue of the Journal of Hospital Medicine, lead author Vikas I. Parekh, MD, admits that after four years of trying to incorporate midlevels, the hospitalists in his group decided last year to phase out the use of midlevels in the general medicine service.

One big problem, writes Dr. Parekh, who is associate director of the hospitalist program for the University of Michigan Health System in Ann Arbor, is that training programs for midlevel providers don’t include enough of a focus on inpatient internal medicine. As a result, his group found, midlevels don’t achieve the level of productivity or autonomy they need to be cost-effective.

Further, the editorial pointed out, hospitalist groups rushing to incorporate midlevels are flying blind. That’s because the literature that exists on such partnerships are one-site studies with no meaningful data on midlevel productivity. Studies also fail to address the amount of time physicians need to spend on oversight and supervision.

In his market, says Dr. Parekh, who is also associate program director of the internal medicine residency program at University of Michigan, it makes better sense to just hire more hospitalists. Dr. Parekh spoke to Today’s Hospitalist about the editorial “and what his group’s experience means for hospitalist programs.

During the four years your program incorporated midlevels, they worked in what capacity? Did they serve as rounders?
We tried all sorts of things, from having them independently manage a subset of patients with attendings seeing them all to a team-based model where the attending and the PA tried to see all patients. Our last iteration was a split service model where PAs basically had responsibility for about half the patients on the service, with attending oversight.

Most iterations were primarily daytime functions, providing daily care with few admissions. We concluded that given the time and effort it took in physician supervision, it didn’t add enough value in a very academic general internal medicine setting.

Did you try using midlevels in niche services like ortho comanagement or doing discharges?
We did not. We do ortho comanagement, but there was a funding issue and not a lot of work. And in terms of finding a niche or focusing on a more non-clinical area like discharges, we floated those ideas to the PAs, but they weren’t favorably received. It wasn’t what the PAs wanted or envisioned for the job.

How many of these problems stem from working at an academic center with high-acuity patients?
It wasn’t just acuity, because we have PAs working successfully with bone marrow transplants. The bigger issue was high diagnostic complexity where there is no protocol to tell you what to do.

It was that complexity combined with acuity that made the difference. The PAs required a fair amount of physician oversight for that diagnostic component, and when we factored in our time, it didn’t work out.

You mentioned that PAs there work with bone marrow transplants. Any other successful partnerships?
Midlevels here do well in subspecialty niches such as hematology/ oncology. They have a pretty narrow subset of patients, primarily leukemia and lymphoma patients coming in for chemotherapy or protocol-driven clinical trials. There’s not much that is out of the box, once you understand the niche.

We also use PAs in the heart failure impatient service. Again, it’s easier to understand the subset of disease and patients’ primary complaints when you’re not dealing with 20 different diagnostic possibilities.

Were you hiring midlevels right out of training?
We purposefully didn’t hire very many new trainees, but we experimented with that and had some success. The advantage of people right out of training was that we could train them ourselves, but they obviously had much less clinical experience, so it took more effort. Most of the people we hired were experienced in other settings.

The article mentions the lack of data on midlevel productivity.
The literature is very limited, and studies usually look at one model in one setting, so you rarely come away from a paper with an idea of how to translate that model to your environment. Studies don’t look at how much midlevels actually do, or what happens at night and on the weekends.

But what’s really missing is the amount of time physicians spend on oversight. Even in areas where we use midlevels “hem/onc and bone marrow transplants, for instance “if you ask the physicians, they’ll say they spend a lot of time in supervision.

Your article suggested that some academic programs may want to partner with midlevel training programs to “train your own.” Is that something you’re pursuing?
We’re not. We have nurse practitioner training here, but we don’t have a PA school. But I believe some academic centers are creating such schools, and that their hospitalist programs will play a big role in them.

Any plans to resurrect the program?
In our market, I can find good hospitalists at a price point where it’s cost effective to go that route. In other markets, I’d have to think about my options.

Any advice for hospitalist programs thinking of hiring midlevels?
The big thing is knowing what to expect. If you’re not hiring PAs with hospital medicine experience, you can’t expect them to hit the ground running. Be prepared to spend a fairly long time training and supervising them.

Also, be prepared to experiment. Start with one model, but realize that it may not work in your setting.

Phyllis Maguire is Executive Editor of Today’s Hospitalist.