How hard do hospitalists work and how much are they paid? There are several sources of data to answer this question, and each has its strengths and weaknesses. Because these data influence contract negotiations and compliance with federal regulations, it is worth taking the time to understand the differences in each data set.
I’ll focus on the two most common sources of data: the biannual SHM survey of hospitalist productivity and compensation (officially titled “The Authoritative Source on the State of Hospital Medicine”), and the Medical Group Management Association’s (MGMA) annual “Physician Compensation and Production Survey.” There are many other surveys that report hospitalist data such as those by the American Medical Group Association (AMGA), Sullivan & Cotter, Hay Group, and others. Each July, Modern Healthcare magazine publishes the average compensation (but no other data) reported for hospitalists and other specialties by each of these organizations and several others (but not the SHM data). It should be easy to find a copy of Modern Healthcare in your hospital administration or library, or on the Internet.
The SHM and MGMA surveys are the most widely used sources of data for hospitalists, and some of their attributes are described in Table 1 (see below).
I should acknowledge my potential conflict of interest and potential for bias in comparing these surveys. This column is in an SHM publication. I’m very active in SHM, and I’m a past chairman and ongoing member of the Benchmarks Committee, which oversees the design and analysis of the SHM survey. And while I am familiar with and regularly review the MGMA survey, I have no other connection to that organization.
Much of the difference between the surveys is a result of the SHM survey being designed specifically for hospitalists in any type of practice setting (e.g., small hospitalist-only groups, as well as hospitalists with huge organizations like a university faculty group practice). In contrast, the MGMA survey is designed for all physician specialties, so a hospitalist answers the same questions as a traditional primary care doctor, plastic surgeon, and obstetrician.
MGMA data can be adversely affected by the inclusion of primary care office-based encounters. One of the principal ways the two surveys differ is how they address ambulatory visits. The MGMA survey reports inpatient and ambulatory visits separately, but “ambulatory” visits include any for a patient who is not a hospital inpatient. By this definition, hospitalists make ambulatory visits, most commonly to hospitalized patients who are on observation status, and also patients seen in an ED, skilled nursing facility, or pre-op clinic. Thus the MGMA survey doesn’t distinguish between ambulatory encounters a hospitalist would generate in the course of serving as a hospitalist, and those generated while that doctor might be serving in a non-hospitalist role such as office-based primary care or urgent care.
The SHM survey doesn’t include—and isn’t contaminated by—office-based primary care or urgent-care visits.
Of the 3,376 total encounters reported in the MGMA survey, 40% (1,351) are ambulatory encounters. Although the SHM survey does not distinguish between hospital and ambulatory encounters, my experience suggests few, if any, hospitalist practices make 40% of their total encounters with patients on observation status, or in an ED, SNF, or pre-op clinic. Thus, many of the ambulatory encounters reported by MGMA might have been office visits, not hospital-related visits.
Additionally, the median internal medicine hospitalist encounters (ambulatory and hospital combined) in the MGMA survey (3,376) is 42% higher than the median total encounters reported for internal medicine hospitalists in SHM’s survey (2,378). Yet the wRVUs reported in the MGMA survey (3,514) are only 8% higher than those reported in the SHM survey (3,256). Thus the calculated average wRVUs per encounter for the MGMA data is only 1.04, compared with 1.37 for the SHM data. An average of 1.04 wRVUs per encounter is very low for hospitalists, when almost all current procedural terminology (CPT) codes a hospitalist uses have a value of one or more wRVUs. Again, this suggests the MGMA data may be significantly influenced by the inclusion of office-based encounters, some of which have wRVUs of less than one. SHM has approached MGMA to discuss this data definition issue in their survey.