Survey data is often much more complicated to interpret than it first appears

John Nelson writes…

Compensation and production surveys generally set the standard on physician compensation and production.  The longer I spend looking at survey methodologies and results, the more I’m convinced that there are a lot of devils in the details that may misled many people who rely on these surveys.  As an example, I’ve adapted below something originally written by, Leslie Flores, who is the director of SHM’s Practice Management Institute and my consulting colleague.  What follows was written in response to a question about the differences in the SHM and MGMA surveys of hospitalist productivity and compensation, especially related to their results in the South.  It serves an example of the complexity of interpreting the surveys.

Until the 2007-8 survey the MGMA “hospitalist” data contained a significant proportion of doctors who worked part of their time in the hospital and part of their time in an office practice.  We know this because about 40% of the encounters reported by “hospitalists” in the 2006-7 MGMA survey were ambulatory encounters – far more than could be accounted for by hospitalists’ observation and ED visits.

In the 2007-8 survey MGMA changed their hospitalist definitions and their data improved dramatically – at least in terms of the proportion of ambulatory encounters being reported for doctors in the hospitalist categories.  This year (data released summer ’09) the proportion of ambulatory encounters reported by hospitalists is only about 5%.

So MGMA’s hospitalist data is getting better.  But there are still big differences between MGMA’s data set and SHM’s data set, including:

1)    SHM’s survey instrument allows respondents to count an individual as more than 1.0 FTE (full-time equivalent) if they worked “extra” shifts; MGMA never counts an individual as more than 1.0 FTE.  This means that the additional RVUs generated by hospitalists who are working extra shifts get counted as part of their base 1.0 FTE, which I believe increases the average production.

2)    SHM’s survey results have a higher proportion of academic practices than MGMA, which tend to have lower per-FTE production; in addition, in SHM’s last survey the large management companies (like IPC, Cogent, etc.) declined to provide individual hospitalist data – only averages for all of their hospitalists combined.  So the median data reported in the SHM survey doesn’t represent a “typical” distribution of hospitalist practices.  (There was an attempted to correct for this in the “adjusted” data columns in the SHM report.)

3)    MGMA, on the other hand, has a separate survey for academic practices so they are not represented in the Physician Compensation and Production Survey report.  In addition, of the internal medicine hospitalists in the newest MGMA data set, 953 reported working for multi-specialty medical groups and only 57 for single specialty practices.  Even more interesting, 810 reported working for groups with 75 or more total FTEs (all specialties), while only 228 reported working for groups with 25 or fewer FTEs.  Although a little over half were associated with hospital owned practices (compared to about 64% in SHM’s survey who work for a hospital or academic institution) there are clearly some different demographic represented in the two cohorts.

Despite those differences, the national median reported by MGMA for internal medicine hospitalists is 3,810 wRVUs while SHM reported a national median of 3,715 – not a huge difference.  In the South, though, SHM reported 3,719 wRVUs while MGMA is reporting 5,664 – which IS a huge difference.  I think it’s worth noting that in previous SHM surveys the South region has always had significantly higher production (and compensation) than other regions of the country; the fact that median production similar in the most recent SHM survey seems to be an anomaly.  (For the sake of an additional point of reference, the American Medical Group Association, which represents large multi-specialty groups and integrated delivery systems, reports a national median of 3,265 wRVUs and a Southern median of 3,987.)

Finally, I need to note that the wRVU schedule changed significantly in 2007 and I suspect many people reported 2006-schedule production data as if it was 2007-schedule data, which makes 2007 results from all the surveys more questionable.

Starting later this year (2009) SHM and MGMA will combine their talents and resources to survey hospitalist practices, and the results should become increasingly meaningful and reliable.

Leave a Comment