At sold-out HM09 in Chicago in May, I had the pleasure of moderating a panel discussion titled “Who Says 15 Patients a Day is the Right Number?” As you might guess, each panelist (including me) said, in effect, “No one says 15 patients a day is the right number.”
Despite being a very important issue to SHM, the society doesn’t have an official position on the “right” or optimal daily patient volume or workload for a hospitalist. SHM generates and disseminates a lot of information to help each practice make decisions about workload, including SHM’s 2007-2008 “Bi-annual Survey on the State of the Hospital Medicine Movement,” articles available at www.hospitalmedicine.org, and articles in The Hospitalist. But all practices, and individual hospitalists, will have to decide what level of patient volume enables safe care, a sustainable and satisfying workload, and reasonable economic performance.
A problem that makes any workload discussion difficult is that many terms are sometimes used to mean different things. For example, “daily census” is used when comparing workloads between practices, but “daily encounters” is nearly always a more informative metric. Remember, daily encounters for a practice will always be higher (though on rare occasion, the same) than daily census.
The only definition of encounters that can be reliably compared between practices is billable encounters. Confusion arises when one person is reporting billable encounters, and another person is counting as one encounter each time a hospitalist interacted with a patient (e.g., went into the patient’s room or made a chart entry) and reports a higher encounter volume despite having the same workload and patient volume. Of course, billable encounters fail to perfectly describe our workloads, but until someone comes up with a better metric that is universally understood and applicable across all settings, billable encounters is the best metric we have.
Both billable encounters and census can be tricky. We might be convinced that because she averages 17 billable encounters per day, Dr. Krause has a higher workload than Dr. Palmer, who averages 15. But it turns out that Dr. Palmer works 210 shifts annually, generating 3,150 annual encounters; Dr. Krause’s 181 annual shifts generate 3,077 encounters. So while Dr. Krause does indeed work harder on the average day, she has lower annual productivity. My experience is that by failing to compare workloads over long periods, such as a year, many attempts to compare workloads yield misleading conclusions.
Comparing encounter volume from one practice to the next fails to capture other ways workloads differ. Dr. Krause may be the principal caregiver for a number of ICU patients; Dr. Plant might turn such patients over to intensivists. For this reason, work relative value units (wRVUs), which attempt to capture the complexity of each encounter, usually are a more meaningful—though still imperfect—metric.
Apples vs. Apples
Any truly valid method of comparing workloads should sum the annual workload for the entire practice and divide by the total provider full-time equivalents (FTEs). Yet problems arise because night shifts usually are less productive than day shifts. Consider a practice that has a distinct night shift worked by a doctor who does no day-shift work the day before or after. There is a tendency to leave this night shift out of the analysis of average workload per FTE, which makes the practice appear more productive than it really is.
For example, the practice I am part of has dedicated nocturnists who don’t work day shifts. So when thinking about how hard we’re working, we tend to sum each “day” doctor’s wRVUs for the year and divide by the number of day doctors. Calculated this way, our day doctors appear to be more productive than SHM data might indicate, but when including our nocturnists’ production and FTEs in the analysis, the overall workload per FTE in the practice is similar to the data.
Productivity Per FTE
While I think productivity per FTE per year is the best metric to use when comparing a practice to external survey data or comparing one practice to another, it is confounded by two sticky problems: inconsistent definitions of what constitutes an FTE, and the lack of an agreed-upon method of accounting for the contribution of nonphysician providers.
The most common definitions of what constitutes an FTE are based on the number of hours or shifts worked. One practice might define an FTE as 2,000 hours of work annually; another might use 180 annual shifts. Unless each shift has a clearly defined duration, it will be very difficult to reach conclusions. Although each practice might make sound decisions about how they define an FTE, they’re ultimately making arbitrary choices that aren’t consistent from one practice to the next. Controlling for this issue is very difficult.
Even trickier is how to compare the contributions of physician assistants (PAs) and nurse practitioners (NPs) from one practice to the next. In the above example, Dr. Plant’s practice has eight physician FTEs and four NPPs, and Dr. Krause’s group has eight MDs and no NPPs. How much more productive should we expect Dr. Plant’s practice to be? (NPPs make many valuable contributions, in addition to increasing productivity, but these are outside the scope of this article.)
Though survey data offer some clues, there is no established standard for increased productivity expected of PAs and NPs. One approach I use is to compare an NPP’s total compensation (salary and benefits) with that of the average physician FTE. If the cost of each NPP is 60% of an MD, then one could say that each NPP represents 0.6 “physician FTE equivalents” and could be expected to increase the productivity of the practice proportionately. So Dr. Plant’s group might be expected to have the productivity of 10.2 physician FTEs (4 NPPs X 0.6 = 2.2 “physician FTE equivalents,” added to the eight physician FTEs). (Note: While I think converting NPP FTEs into “physician equivalents” is useful in analyzing the effect of workloads on budgets, there are clearly many other very important issues when trying to quantify the contribution of NPPs to a practice.)
There are dozens of additional issues, such as the effect on productivity resulting from inefficient hospital systems (e.g., a clunky electronic health record, typed admit notes aren’t available for a few days compared with records available within two to four hours of dictation), activities such as attending Rapid Response Team activations, and differences in the social complexity of the patient population. None of these would show up in wRVU reports, so they are missing from the analysis I’ve described above.
I think it is impossible to control for all of the differences between practices that influence the definition of appropriate workload. This is the main reason SHM probably will never have a firm position on exactly what is the right or optimal number.
Although 15 patients a day is a reasonable starting point (I prefer to see fewer patients daily, but work more days/shifts annually), things get complicated in a hurry, and there will always be significant variation. TH
Dr. Nelson has been a practicing hospitalist since 1988 and is co-founder and past president of SHM. He is a principal in Nelson/Flores Associates, a national hospitalist practice management consulting firm. He also is faculty for SHM’s “Best Practices in Managing a Hospital Medicine Program” course. This column represents his views and is not intended to reflect an official position of SHM.