|
P E R S P E C T I V E :
P G P S T A F F I N G
W E B E X C L U S I V E
4 February 2004
Do We Need More Physicians?
The answer is to be found in a
reexamination of physician productivity.
By David C. Goodman
ABSTRACT:
A reexamination of physician productivity can resolve the current workforce
debate. If the improvement of patients’ health and well-being is the goal
of medicine, then physician productivity should use this as the output instead
of physician work effort (visits, procedures, or relative value units). Jonathan
Weiner’s study of prepaid group practices (PGPs) and regional data from
the Dartmouth Atlas of Health Care demonstrate that the combination of
low physician labor inputs with favorable outcomes is achievable in varied payer
environments. Better medical care requires not more physicians, but rather improving
physician productivity to the levels already present in many locales.
To begin, let’s agree that forecasting physician workforce requirements
is a nasty business and has been since the Graduate Medical Education National
Advisory Committee (GMENAC) sounded the alarm in 1980 about the growing physician
supply. A few years ago I steered a doctoral student away from an interest in
workforce research because the field was stagnant, politically polarized, and
lacking in funding for all but the most mundane analyses. How did we reach an
impasse in a discourse that is increasingly reminiscent, in form if not substance,
of a debate between medieval Christian theologians? What prevents us from advancing
our understanding of how many physicians are needed, and how much in the way
of societal resources should be committed to their training and to their services?
By some measures, workforce requirement models have become increasingly sophisticated
with a fuller set of parameters, sensitivity testing of default assumptions,
and the use of more accurate baseline data. New models have been developed,
such as trend analysis, in a effort to find a touchstone that will guide future
physician requirements.1 Yet amid the growing chorus
for increased physician production, there are two discordant notes that cannot
be reconciled with recent demand-based or trend forecasts. Jonathan Weiner’s
paper reporting physician-staffing levels in prepaid group practices (PGPs)
is one challenge to recent predictions of an impending physician shortage. Studies
revealing the marked regional variation with physician per capita supply are
another such challenge.2 Together they force us
to reexamine what is meant by “physician productivity” and, in turn,
how many physicians are required.
Very early in the efforts to understand physician labor effort, the notion developed
that the output of physicians was medical services, such as the number of office
visits or procedures per week.3 Knowing the number
of weeks worked per year, annual labor efforts were derived for individual or
groups of physicians. The primary advantages of using these as measures of productivity
are the modest data requirements and the ease of calculation across time and
organizations. The measurements are also consistent with the intuitive notion,
which I (a physician) share, that it is good to keep busy seeing patients. Using
medical services as the output of physicians also follows common practice in
factories where one measure of labor productivity is the number of times per
day a worker repeats a task on the production line. As a sole measure of productivity,
it is incomplete—these tasks are not an end in themselves, but a means
to produce a particular good.
It is useful to account for the relative effort of physicians, but forecasting
methods are trapped in the view that the busyness of providers (number of patient
visits, procedures, or relative value units, RVUs, per year) is the salient
measure of physician output. It is not.
In the case of health care, the rightful output is the improvement of health
and well-being. Physician productivity is the labor input for a given increment
of improved population well-being. Measuring physician labor effort (the number
of visits and procedures per unit of physician time) remains useful to compare
different types of physicians and as an adjustment in workforce supply models.
Estimating physician productivity with health outputs is ambitious because it
requires physicians linked to a well-defined patient population, where both
physician labor and patient outcomes can be measured. Other inputs need to be
considered as well, but even without this information, it is valuable to know
the attainable levels of productivity. Weiner’s paper is an exceptional
effort in the documentation of clinician inputs in tightly organized capitated
physician groups. If there were any remaining doubts about levels of physician
supply that are attainable when fully capitated payments are used to plan care,
Weiner’s paper settles the issue. PGPs can deliver care that successfully
competes in the insurance marketplace, yet with clinician (physician and nonphysician
providers) labor inputs that are 64–76 percent of overall U.S. levels.
We do not know all of the factors that permit these low staffing levels. It
does not appear to be a simple substitution of nonphysician providers or patients’
using nonplan physicians. While the adjustments for age and sex account only
partly for differences from the general population, the direction and magnitude
of the bias is debatable. Are the outputs (improvement of health and well-being)
different? Twenty years of studies have not detected systematic differences
either in the processes of care known to be effective or in health outcomes.4
Enrollees in these plans rate them highly.5 Taking
all this into account, PGP physicians are highly productive.
Comparing PGP staffing levels with the workforce across the Dartmouth Atlas
Hospital Referral Regions (HRRs; N = 306) shows similar levels of productivity
in many locales. For primary care, Kaiser Permanente staffing is within in the
lowest quintile of physicians per capita in HRRs, and Group Health Cooperative
is in the second-lowest quintile. HealthPartners’ primary care physician
staffing is in the second-highest quintile but still is lower than many regions—46
percent of the workforce in Royal Oaks, Michigan; 62 percent of Philadelphia’s,
and 64 percent of San Francisco’s. All of the PGPs specialist care staffing
levels are within or close to the lowest HRR quintile, and near the per capita
supply found in Sun City, Arizona; Dayton, Ohio; and Wichita, Kansas. It turns
out that PGP levels are exceptional when compared with the overall U.S. supply
but commonplace when compared with regional workforce levels.
The evidence for the absence of benefit associated with higher physician workforce
levels, particularly specialists, is now well established. Very low levels may,
in some instances, be associated with poorer outcomes, but when physician supply
exceeds low levels, there is little further patient benefit.6
Is there a detriment to having a greater supply of specialists? The nearly twofold
differences in Medicare spending across U.S. regions are largely attributable
to the higher-intensity practice patterns found in regions with a higher per
capita supply of medical specialists; and higher-intensity practice is not associated
with better quality or access to care, nor with slower declines in functional
status or lower mortality.7
Although further empirical work is necessary, current evidence is consistent
that by any measure meaningful to patients, higher physician supply levels are
associated with lower physician productivity. In simple terms, the extra labor
provided in regions with larger physician supply, well-intentioned though it
may be, provides no additional benefits to patients. Studies of PGPs and regional
physician supply in varied insurance markets have remarkably consistent findings.
One recent analysis predicts a 20 percent shortfall in specialists by 2020 if
training is not increased from today’s levels.8
This is approximately the difference in supply between Chicago and Hackensack,
New Jersey, or between Kaiser Permanente and Group Health Cooperative. The data
from PGPs and regions show that a 20 percent difference in supply has no consequence
for patient satisfaction or health outcomes.
When workforce forecasters promulgate their requirement models, check to see
where the production of health and well-being is included. Before we legislate
more public dollars to train physicians and support their practices, we should
expect some evidence that adding physicians to the aggregate supply will provide
something of value to patients: increased access, higher quality, improved satisfaction,
or better health outcomes. Studies of variation in health care across populations
show that these desirable outputs are attainable, but they bear little relation
to physician supply.
The author thanks Doug Staiger, Department of Economics, Dartmouth College,
for his review and suggestions.
NOTES
1. R.A. Cooper et al., “Economic Expansion Is a Major Determinant
of Physician Supply and Utilization,” Health Services Research 38,
no. 2 (2003): 675–696.
2. J. Wennberg and M. Cooper, eds., The Dartmouth Atlas of
Health Care—1998, 2d ed. (Chicago: American Hospital Association, 1997).
3. Uwe Reinhardt makes the reasonable argument that while measuring
the production of health would be “preferred,” it has also been “impossible,
at the empirical level, to control successfully for the vast array of inputs.”
U. Reinhardt, Physician Productivity and the Demand for Health Manpower
(Cambridge, Mass.: Ballinger Publishing Co., 1975), 64. See also K. Bloor and
A. Maynard, “Workforce Productivity and Incentive Structures in the U.K.
National Health Service,” Journal of Health Services Research and Policy
6, no. 2 (2001): 105–113.
4. R.H. Miller and H.S. Luft, “Does Managed Care Lead to
Better or Worse Quality of Care?” Health Affairs (Sep/Oct 1997):
7–25; and R.H. Miller and H.S. Luft, “HMO Plan Performance Update:
An Analysis of the Literature, 1997–2001,” Health Affairs (July/Aug
2002): 63–86.
5. “HMO or PPO: Picking a Managed-Care Plan,” Consumer
Reports (October 2003): 35–39.
6. Wennberg and Cooper, eds., The Dartmouth Atlas of Health
Care—1998; D. Goodman et al., “The Relation between the Availability
of Neonatal Intensive Care and Neonatal Mortality,” New England Journal
of Medicine 346, no. 20 (2002): 1538– 1544; and H. Krakauer et al.,
“Physician Impact on Hospital Admission and on Mortality Rates in the Medicare
Population,” Health Services Research 31, no. 2 (1996): 191–211.
7. E.S. Fisher et al., “The Implications of Regional Variations
in Medicare Spending, Part 1: The Content, Quality, and Accessibility of Care,”
Annals of Internal Medicine 138, no. 4 (2003): 273–287; and “Part
2: Health Outcomes and Satisfaction with Care,” Annals of Internal Medicine
138, no. 4 (2003): 288–298.
8. R.A. Cooper et al., “Economic and Demographic Trends
Signal an Impending Physician Shortage,” Health Affairs (Jan/Feb
2002): 140–154.
David Goodman (david.goodman{at}dartmouth.edu)
is an associate professor of pediatrics and of community and family medicine
at the Center for the Evaluative Clinical Sciences, Dartmouth Medical School,
in Hanover, New Hampshire.
Please click on the author's
names to read related papers by Jonathan
Weiner, Francis
Crosson, W.
Bruce Fye, Fitzhugh
Mullan, Edward
Salsberg and Gaetano Forte, and Stephen
C. Schoenbaum.
DOI: 10.1377/hlthaff.W4.67
©2004 Project HOPE–The People-to-People Health Foundation, Inc.
|
