Practice Management

Effective hospitalist roles for NPs, PAs

Single-site study offers success story, isn’t one-size-fits-all solution.


I’m often asked about effective roles for nurse practitioners (NPs) and physician assistants (PAs), collectively known as advanced practice clinicians (APCs). My first response is always the same: They have much to contribute and can be effective members of hospitalist groups. Most hospital medicine groups (HMGs) should think about having them in their staffing mix if they don’t already.

Dr. John Nelson
Dr. John Nelson

Yet despite all that NPs/PAs can offer, my experience is that many (even most) hospitalist groups fail to develop roles that optimize their APCs’ skills.

An October 2016 study in the Journal of Clinical Outcomes Management adds additional data to help think about this issue. You may have seen the study mentioned in several news articles and blogs. Most summarized the study along the lines of “using high levels of PA staffing results in lower hospital costs per case.” Framing it this way is awfully misleading, so I’ll go a little deeper here.

Study context

The study, “A Comparison of Conventional and Expanded Physician Assistant Hospitalist Staffing Models at a Community Hospital,” is a retrospective analysis of performance measures from two hospitalist groups at Anne Arundel Medical Center (AAMC) in Annapolis, Md.1 One HMG is employed by the hospital. The other, called MDICS, is a private company that contracts with AAMC as well as approximately 13 other hospitals and 40 rehabilitation facilities. Tim Capstack, MD, is the AAMC medical director for MDICS and lead author of the study (representing a potential conflict of interest acknowledged in the article). Barry Meisenberg, MD, is a coauthor, a hospitalist in the AAMC-employed group, and chair for quality improvement and health care systems research at AAMC.

Tim told me by phone that both groups have practiced at AAMC for more than 10 years and enjoy a collegial relationship. Both groups employ PAs and pair them with a single physician in a dyad arrangement each day. Tim’s MDICS group, the “expanded PA” group, staffs each day shift with three physicians and three PAs, compared with the nine physicians and two PAs in the hospital-employed “conventional” group. The MDICS PAs are responsible for more patients each day than their conventional-group counterparts and, during the January 2012 to July 2013 study period, averaged 14.2 patients versus 8.3, respectively.

Over the course of the study, PAs in the expanded PA group saw and billed 36% of patient visits independently, compared with 5.9% for the conventional group.

Notable study findings

I think the main value of this study is in showing that the expanded PA group had rates of readmission, inpatient mortality, length of stay, and consultant use that weren’t statistically different from the conventional group.

The workloads and years of experience of doctors and PAs in each group were similar. And while there were some differences in the patients each group cared for, they seem unlikely to have a significant influence on outcomes. Clearly, there are many unmeasured variables (e.g., culture, morale, and leadership) in each group that could have influenced the outcomes, so this one study at one hospital doesn’t provide a definitive answer about appropriate APC staffing levels. However, it didn’t uncover big differences in the measured outcomes.

And this study did show that higher levels of PA staffing were associated with lower hospital charges per case. Although the difference was a modest 3%, it was statistically significant (P less than .001). I’m skeptical there is causation here; this more likely is just correlation.

It would be great to see a larger study of this.

Information applications

So does this study support the idea that HMGs can or should increase APC staffing and workload significantly to realize lower hospital cost per case and not harm patient outcomes? Not so fast!

This study only compared two hospitalist groups at one hospital. It’s probably not very generalizable.

And as described in the paper, and stressed by Tim talking with me by phone, the outcomes of their expanded PA model likely have a lot to do with their very careful recruiting and screening of experienced PAs before hiring them, not to mention a lengthy and deliberate on-boarding process (summarized in the article) to support their ability to perform well. Groups that are not as thoughtful and deliberate in how they hire and position APCs to contribute to the practice may not perform as well.


Why study only PAs? What about NPs? Tim told me that his group is agnostic regarding the training background of the APCs they hire; he suspects an identical study with NPs rather than PAs in each hospitalist group would probably yield very similar results. I see this the same way. Although there are differences in background and training between NPs and PAs, I think personal traits like years of experience in various health care settings and the ability to work efficiently are more important than training background.

A practical approach

Any group who thinks this study is evidence that adding more APCs and having them manage a higher number of patients relatively independently will go well in any setting is mistaken. But it does offer a story of one place where, with careful planning and execution, it went OK.

In my view, the real take-home message is to think carefully to ensure any APCs in your group have professionally satisfying roles that position them to contribute effectively. While common, I think configuring APCs and physicians as rounding dyads often ends up underperforming and not working out well because of inefficiency. When well executed, as is apparently the case in this study, it can be fine. But my experience is that positioning APCs to assume primary responsibility for some clinical activities, such as covering the observation unit or serving as an evening admitter/cross-cover provider (all with appropriate physician collaboration and backup), more reliably turns out well.


Capstack TM, Seguija C, Vollono LM, Moser JD, Meisenberg BR, Michtalik HJ. A comparison of conventional and expanded physician assistant hospitalist staffing models at a community hospital. J Clin Outcomes Manag. 2016;23(10):455-61.

Dr. Nelson has been a practicing hospitalist since 1988. He is co-founder and past president of SHM and principal in Nelson Flores Hospital Medicine Consultants. He is co-director for SHM’s “Best Practices in Managing a Hospital Medicine Program” course. Write to him at