|
Weiner Web Exclusive
W O R K F O R C E P O L I C Y :
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
Prepaid Group Practice Staffing
And U.S. Physician Supply:
Lessons For Workforce Policy
What can we learn from examining
the staffing levels of some
of the country’s largest organized delivery systems?
By Jonathan
P. Weiner
ABSTRACT:
This paper describes staffing at eight large prepaid group practices (PGPs)
serving more than eight million enrollees at Kaiser Permanente and two other
health maintenance organizations (HMOs). Even after characteristics of the patient
populations and outside referrals are accounted for, these PGPs have a physician-to-population
ratio that is 22–37 percent below the national rate. Two decades of historical
data at Kaiser Permanente indicate that its rate of specialist growth was far
higher than that of primary care. The study suggests that efficient systems
of care can readily meet the demands of patient populations with workforce staffing
ratios below current U.S. levels.
Prepaid group practices (PGPs) are highly structured, multispecialty medical
groups that are reimbursed by capitation to serve the enrollees of a health
maintenance organization (HMO). When PGPs were founded in the mid-twentieth
century, the strategies they adopted were unique. They emphasized primary, preventive,
and ambulatory care and were among the first to welcome nonphysician providers
(NPPs) such as nurse practitioners (NPs) and physician assistants (PAs). PGPs
were also among the first to develop coordinated approaches to care delivery.
Many of these innovations have been disseminated widely within the U.S. health
care system. Other staffing and organizational approaches used by PGPs still
set them apart from the less structured fee-for-service (FFS)–oriented
U.S. health care system.
Even though PGPs are a relatively small segment of U.S. practice, managers and
policymakers have looked to them since their inception as an example of how
best to “staff up” to provide care for defined populations. Within
the disaggregated U.S. health care system, there are few other instances where
a “numerator” of providers and a “denominator” of consumers/patients
are so clearly demarcated.
Over the past four decades, policymakers, labor planners, and clinic managers
have used staffing ratios derived from PGPs to help determine whether the supply
of providers in a particular geographic area or within a particular care delivery
organization is “adequate.”1 In addition,
national and regional workforce planners have assessed the potential impact
of PGPs as they, and other organizations patterned to some degree after them
(such as independent practice association, or IPA/network-model HMOs), have
become more widespread. Specifically, in the late 1980s and early 1990s, when
national health reform based on competitive integrated delivery systems was
being seriously considered, there was intense interest in staff/group-model
HMOs as one prototype.2
Although such government-led reformation did not come to pass, other changes
did occur: Corporate-controlled “managed care” became ubiquitous.
But after a decade of mainly for-profit, loosely organized plans in the mainstream,
a widespread “backlash” ensued, leading to the current environment
in which policymakers, consumers, and clinicians are seeking alternatives to
managed care. PGPs, with their unique roots in both the social welfare and cost
containment movements, represent one such alternative.
In this paper I provide a detailed case study describing how eight large PGPs
have structured their workforces to deliver care to consumers enrolled in Kaiser
Permanente (KP) and two other HMOs: the Group Health Cooperative (GHC) of Puget
Sound in Seattle, and HealthPartners (HP) in the Minnesota Twin Cities. This
study documents staffing for six distinct regions of the Permanente Medical
Group. The focus of this paper is allopathic (MD) and osteopathic (DO) physicians.
Also included are PAs, NPs, and other nurse specialists (such as midwives and
nurse anesthetists) who can be viewed to some extent as physician substitutes.
The paper also documents the PGPs’ employment of certain non-MD/DO doctors
(such as optometrists, psychologists, and podiatrists) and independent mental
health therapists.
One objective is to identify the unique approaches that each PGP has chosen
to meet the needs/demands of its patient population. To allow readers to assess
how the PGPs compare with current U.S. patterns of practice, throughout the
paper comparisons are made with current U.S. medical workforce supply ratios.
The paper greatly expands upon previous work in this area. In addition to collating
recent data, it provides detailed information for ambulatory and hospital-based
physician specialties, out-of-group contracted care, and NPPs.
Study Sites And Methods
The PGP providers documented in this study practice at more than 350 clinic
sites and 33 hospitals owned and staffed by HMOs. The PGPs serve a population
of more than eight million consumers residing in nine states and the District
of Columbia. The KP enrollees represent about 93 percent of this total; the
two California regions (Northern and Southern California), about 75 percent.
Combined, the California PGPs serve more than six million enrollees who obtain
care from more than 140 ambulatory sites and 28 fully staffed inpatient facilities.
No other private medical group (prepaid or otherwise) approaches this size and
level of comprehensiveness. The GHC and HP organizations also own and staff
their own hospitals; however, they make greater use of nonemployed contracting
physicians to serve the medical needs of their members than the KP sites do.
All PGPs in this study provide most mental health services on site and do not
contract with a “carve-out” behavioral health plan.
Two of KP’s eight regions (Ohio and Georgia), accounting for about 8 percent
of KP’s national enrollment, were excluded from this study. A sizable proportion
of the services there were provided by physicians who were not part of a Permanente
Medical Group, and the data describing out-of-plan contracting were less complete
than desired for this study. For similar reasons, only the western Washington
region of GHC was included, excluding eastern Washington and Idaho.
The patient populations included in this analysis represented only those enrollees
who were registered with primary care sites staffed by the PGP. Enrollees served
by the plans’ independent physician networks were not included. For KP
and GHC, the percentage of enrollees thus excluded was fairly modest (less than
10 percent); however, about half of all HP enrollees are served by contracting
group practices across the region. Plans provided the information used in this
analysis during mid-2002 for a reporting period of late 2001 or early 2002.
Relevant characteristics of the practices, by study site, are shown in Exhibit
1, which also presents estimates of the percentage of all hospitalizations
provided in HMO-owned hospitals and the proportion of covered physician services
that are delivered by providers not employed by the health plan’s collaborating
PGP.
Study Results
Exhibit
2 presents a comprehensive specialty-specific description of the PGP-employed
physicians in more than forty specialties.3 The
rates are expressed in terms of employed full-time-equivalent (FTE) patient
care MDs and DOs on the staff per 100,000 people enrolled at each site.
To enable a comparison with the current overall U.S. medical workforce, the
most recent data on the availability and characteristics of physicians were
obtained from American Medical Association (AMA) and American Osteopathic Association
(AOA) “masterfile” sources. For comparability, all “U.S. supply”
rates presented in this paper include only nonfederal, nontrainee providers
who are actively involved in patient care. Thus, about 18 percent of the current
active (non-retired) U.S. physician supply—which equals approximately 280
MDs/DOs per 100,000—is excluded from the comparison. Although this approach
is standard among workforce analyses, it conservatively underestimates the true
availability of clinically active providers in the United States.
Physicians employed by the group provide the majority of care at each PGP. However,
for sites other than the two extremely large California KP sites, the staffing
rates for the tertiary and hospital-based specialties documented in Exhibit
2 (and in an expanded
Exhibit 2 online; see Note 3) are likely to be incomplete.
At the other sites, nonemployed contracting physicians provide many specialized
services; I return to this point later.
At some PGPs, medical directors and other administrative physicians who are
actively involved in managing care are reported within their respective medical
specialties. Others include these physicians in a separate “medical director”
category. The proportion of employed physician time spent on nonclinical activities
(such as research) was excluded from the reported staffing rates at all sites.
Exhibit
3 presents a breakdown of NPs/PAs and other NPPs who deliver services that
in other settings are typically delivered by physicians. Some of the sites (notably
GHC and several KP sites) made considerable use of nurses (generally registered
nurses without NP training) to staff round-the-clock triage “call centers.”
Although this is an innovative use of clinicians, these nurses, along with other
ambulatory care nurses such as clinic nurses or case managers, and other inpatient
nurses were excluded from this physician services–oriented analysis.
Exhibit
4 provides a site-specific summary of the rate of overall employed providers,
including both MDs/DOs and NPs/PAs before any adjustments are applied. The physician
figures are broken down by primary care and specialty care.
Before we attempt to use these PGP staffing rates as a benchmark for other populations,
a series of adjustments must be made.4 These include
adjustments for difference between the demographics of HMO enrollees and the
U.S. population; the extent to which nonemployed physicians provide covered
care; and the proportion of providers’ time spent on patients who are not
enrolled with the PGP.
Exhibit
5 offers estimates of these three sets of adjustment factors for each PGP.
(Non-California KP sites are combined as “other.”) All of the factors
noted in this exhibit are multiplicative, which means that they should be multiplied
by the employed provider rates in previous exhibits to account for each factor
noted.
The first row of Exhibit
5 presents an approximate demographic adjustment. This is based on a comparison
of the demographic characteristics of each PGP’s enrollees with the characteristics
of the U.S. population as of 2000. An indirect standardization method was applied
using age- and sex-specific ambulatory physician contacts derived from a national
HMO survey.5 All three PGPs have a significantly
higher proportion of elderly enrollees than the U.S. HMO average and thus have
age distributions similar to that of the U.S. population. Therefore, little
adjustment of the reported staffing rates is needed to account for demographic
mix.
Even though the PGPs in this study are among the largest in the country, community-based
physicians not employed by the plan deliver some members’ care. Outside
providers are reimbursed primarily by FFS. Two of the sites (GHC and HP) could
share specialty-specific information of this type. Using a method developed
at GHC, national group practice annual billing averages by specialty were used
to estimate the FTE rates (in terms of physicians per 100,000 members) of the
outside services provided by contracting physicians.6
KP, which makes less use of such physicians, provided estimates based on historical
experience.
Exhibit
5 presents “external contract provider” adjustment factors by
major specialty group. These factors, like the others in the exhibit, are multiplicative.
For example, overall at GHC, 87 percent of the care is provided by the PGP’s
employed providers and the remaining 13 percent by community doctors. Therefore,
an adjustment factor of 1.15 should be multiplied by the employed staff rates
presented previously, to approximate the total number of FTEs serving the HMO
enrollees.
All of the PGPs in this study primarily serve the members of their affiliated
HMOs. They are paid a capitated fee for these services. But in certain circumstances,
the PGP providers care for patients from outside the enrollee denominators described
in Exhibit
1. This might include emergency care to nonmembers; “reciprocal”
coverage for members of another HMO; services to members of the PGP’s partner
HMO enrolled with an external IPA; or care for members insured under a separate
policy (such as workers’ compensation). The “nonenrollee care”
adjustment factors, based on actual recent history at each site, are presented
in the last row of Exhibit
5. They indicate the approximate downward adjustment of employed FTEs that
is necessary before they can serve as external benchmarks.
Exhibit
6 provides an overview of “adjusted” provider supply at the PGPs
after applying the three sets of factors described in Exhibit
5. U.S. rates are also presented as a point of reference. Exhibit
6 provides estimates of physician supply per 100,000 enrollees after the
“outflow” (that is, enrollees leaving the group for care) and “inflow”
(that is, nonenrollees getting care from group providers) are taken into account.
These figures represent this paper’s “bottom line” in terms of
how the overall provider staffing in PGPs compares with the current U.S. provider
supply, once all adjustments are taken into account.
An analysis similar to this one was published in 1989, using 1983 KP data. Using
these data, with augmentations from other sources for NPPs, it was possible
to do an eighteen-year trend analysis at KP. These 1983–2001 trends are
then compared with the underlying U.S. supply trends over approximately the
same time period (1980–2000). Exhibit
7 presents these two sets of figures by specialty class and separately for
physicians, NPPs, and both combined. The KP data in this exhibit have not been
adjusted; they reflect employed FTEs. To account for the slightly different
number of years included in the two data sets, annualized trends are presented
in the rightmost column.
Discussion And Policy Implications
Summary of findings.
Across the PGPs, overall adjusted physician staffing ranges from about 144 to
176 per 100,000 enrollees, compared with a national ratio of about 229 per 100,000.7
When both physicians and NPPs are combined, the overall adjusted supply ranges
from about 174 to 202, compared with about 270 per 100,000 nationally. This
is equivalent to about 1 provider for every 490 enrolled people at GHC, 1 per
495 at HP, and 1 per 575 at KP. The nation’s overall (nontrainee, nonfederal)
provider-to-patient ratio is about 1 per 370.
Staffing levels at the three PGPs (and the six KP regions) fall within a fairly
tight range. But beyond this aggregate level, a number of interesting variations
are noteworthy and show the alternative approaches these organizations have
taken to meet the needs of their populations.
The proportion of primary care physicians at each site ranges from about 40
percent to 46 percent after adjustments are made for contracting providers.
This compares with about 41 percent of primary care MDs/DOs nationally. Percentages
may be misleading, however; the U.S. primary care physician supply is about
93 per 100,000, whereas the adjusted primary care supply in the PGPs ranges
from 58 to 80 per 100,000. At HP, 10 percent of primary care providers are NPs
or PAs; at KP the non-MD/DO proportion is 17 percent, and at GHC it is 25 percent.
Nationally, about 14 percent of primary care providers are nonphysicians.
NP/PA staffing, in total, ranges across the PGPs from about 26 to 38 per 100,000
(compared with approximately 41 per 100,000 nationally). It appears that the
practices of PAs and NP/APNs (which for this analysis include midwives and nurse
anesthetists) are concentrated in the specialty care areas at HP and KP (65
percent and 60 percent, respectively). In contrast, the NPP focus at GHC is
predominantly (60 percent) primary care. When other types of NPPs (beyond NPs
and PAs) are taken into consideration, the deployment across the three PGPs
is fairly similar, in the range of 48–53 providers per 100,000 enrollees.
No comparable national rates are readily available for this expanded NPP definition.
The eighteen-year KP trend analysis suggests that its rate of annual growth
in the overall physician-to-population ratio (1.4 percent) is not too dissimilar
from the national increase of 1.7 percent during roughly the same period. However,
this trend obscures a noteworthy finding. Most of KP’s growth has been
among non– primary care physicians; there was a 2.4 percent annual increase
in specialists at KP and only a 0.3 percent annual increase in primary care
physicians. This compares with 1.2 percent for specialists and 2.4 percent for
primary care nationally.
For nonphysicians at KP, the specialist-oriented trend was even more pronounced;
specialty care NPs/PAs grew 6.9 percent annually, but there was zero growth
for primary care NPs/PAs. This lopsided situation mirrors the national picture:
Specialty care NPPs grew 13.6 percent annually, compared with a 1.8 percent
annual increase in primary care NPPs.
Limitations and generalizability.
The PGP workforce statistics presented in this paper are arguably the most comprehensive
and accurate of this type ever compiled, and certainly the most up-to-date available
in the public domain. However, a number of limitations should be noted, particularly
as they relate to the generalizability of these workforce levels beyond the
PGP setting. When collating data from the multiple practice sites, I used common
definitions and data collection frameworks whenever feasible. But, of necessity,
there was considerable reliance on local definitions and existing management
databases at the PGPs.
When one is considering the implications of the staffing patterns reported in
this paper, it is important to acknowledge the context from which they were
drawn. That is, the physician and NPP rates must be understood from within the
framework of the overall organization, its administrative support staff, and
its philosophies. While many aspects of the PGPs’ practices—including
approaches to medical staffing—can be emulated in non-PGP settings, in
certain instances it may be difficult to adopt just part of the “package.”
Although the PGPs studied here provide care to a broad cross-section of members,
their enrolled populations are not representative of all Americans. In addition
to the empirical adjustments described previously, a series of other comparability
issues has been recommended for consideration before PGP staffing ratios such
as those presented here are extrapolated to other organizations or locales.
One issue that must be considered for national planning purposes is the degree
to which people enrolled in HMOs choose to get care on their own outside of
the plan (from neither PGP-employed nor contracting providers). Because services
at these PGPs are so comprehensive, this external use rate is not expected to
be very high (estimates have suggested that it is less than 5 percent).8
Another interesting issue, and one that is difficult to measure, is the degree
to which PGPs experience “adverse” or “positive” selection
bias. That is, are their enrollees sicker or healthier than non-PGP patients,
and could they be expected to require more or less care? The national studies
on this issue are mixed, but those emphasizing PGPs (rather than HMOs more generically)
have suggested that PGP-model HMO populations are not healthier than community-based
patients.9 Adding to this national observation was
a local analysis at HP suggesting that the morbidity burden (based on the ambulatory
care group, or ACG, case-mix measure) among the PGP patient cohort was approximately
10 percent greater than for people in the HMO’s IPA network.
Another issue of comparability relates to potential need for “socioeconomic”
adjustments. These PGPs have Medicaid memberships that range from 0.7 percent
to 13 percent of their populations, and they have no uninsured patients. This
compares with 11 percent and 13 percent, respectively, for these special-need
population cohorts within the larger U.S. population. The evidence is mixed
on whether an upward or downward adjustment would be needed if the PGP patient
populations included a proportional representation of these special-need patients.10
The issue of geographic distribution also deserves comment. The PGPs in this
study serve consumers in nine states and the District of Columbia. However,
they are based primarily in the West and Midwest and do not have much presence
in rural areas. Practice patterns in the study locales could be different than
in the United States overall, although there is no evidence that the medical
needs of the populations in the underrepresented geographic areas would be higher
or lower than those of consumers in the study PGPs.
As much a conceptual issue as one of PGP/non-PGP comparability is the complicated
and controversial issue of how to define provider “productivity” and
“full-time equivalency”: Does one FTE physician at the PGP equal one
FTE outside of the PGP? Further adding to the controversial nature of this discussion
is its relationship to the provider’s sex.
AMA surveys and other sources have documented differences in terms of number
of patients seen per year by “private practice” physicians compared
with PGP-employed physicians. FFS doctors, on average, see 15–20 percent
more patients annually than those in more structured settings such as PGPs.11
Therefore, some have suggested that it takes fewer FFS physicians to provide
comparable amounts of care. These surveys do not attempt to assess comparability
or efficiency associated with patient-physician interactions, simply the number
of hours per year spent seeing patients in different settings.
Another related issue is the number of people it takes to constitute one FTE.
Because of the structure of PGPs, it is believed that a sizable proportion of
physicians work part time in such settings. For example, KP Northern California
reports that 20 percent of all permanent physicians on staff work less than
90 percent of what the organization considers to be a full-time schedule. Although
comparable national data are not available, based on a 2000 national survey,
the AMA reports that only about 10 percent of physicians worked fewer than twenty
hours per week or described themselves as not fully active.
There has been a documented tendency for female physicians to work fewer hours
per year than their male counterparts. This has been cited by some analysts
as being an important factor in understanding national supply and requirement
patterns as more women enter the medical profession.12
Numbers of male and female physicians at KP who practice part time are not available,
but medical managers at the plan believe that more women than men work part
time.
PGPs appear to be appealing to female physicians, likely because of the supportive
and flexible nature of their practice environment. Although only 22 percent
of U.S. physicians (post-training) are female, 31–47 percent of physicians
across the six KP sites in this study were female. This rate is quite similar
to where the United States is expected to be in a decade or two.
The net result of part-time status issues is that each FTE reported here represents
more than a single person: It can be estimated that every 1.00 reported FTEs
reflects 1.10 to 1.20 employed practitioners. Nationally, the comparable rate
falls in an estimated range of 1.05–1.10. The full-time equivalency and
PGP versus non-PGP productivity issues should be the subject of further inquiry.
Implications for workforce
policy. This paper
presents workforce supply ratios for large populations served by several very
large “closed” integrated delivery systems (IDSs). These staffing
ratios are likely to have direct implications for human resource planners in
other large IDSs as they seek benchmarks for “staffing up” to meet
the needs of their consumer/patients. These other structured systems include
not only PGPs but also well-integrated physician-hospital organizations, large
non-PGPs with geographically defined patient bases, government providers such
as the military, and the structured delivery systems found in other countries.13
What are the implications of the provider staffing rates described here, beyond
other structured practice settings? Are PGPs’ workforce levels appropriate
benchmarks for the majority of Americans who do not receive care from PGPs or
other organized delivery systems?
Determining whether a given provider-to-patient ratio is too low, too high,
or just about right is both technically and conceptually difficult. Over the
years, alternative methodological approaches have been used to set medical workforce
“requirement” benchmarks. The approaches have generally embraced one
of three theoretical frameworks: economic demand, clinical need, or HMO staffing.
The latter approach has been a relatively common method for setting workforce
adequacy reference points because of its intuitive appeal and its relatively
modest data requirements.
I did not set out to review the advantages and disadvantages of the alternative
methodologies used to set workforce availability standards. Nor did I set out
to apply the PGP staffing data in this paper to evaluate the adequacy of the
current or projected U.S. physician supply. Rather, in terms of policy impact,
my intent was to offer fresh information to support the formal workforce planning
activities now under way within both the public and private sectors (for example,
the federally sponsored Council on Graduate Medical Education, or COGME, and
the efforts of the Association of American Medical Colleges, or AAMC).
This study provides evidence that organized PGPs in urban and suburban areas
provide high-quality, cost-effective care to a diverse insured population with
considerably fewer physicians than are now available in the nation at large.
After adjustments are made to take differences in U.S.-to-PGP enrollee demographics
and use of providers not employed by the PGP into consideration, the physician-to-population
ratios at the three PGPs is approximately 22–37 percent lower than the
overall U.S. ratio. When NPs and PAs are added to the mix, the PGPs’ total
provider supply rate is about 24–36 percent lower than the national rate.
When the U.S./PGP provider supply differences are assessed separately for primary
and specialty care, an interesting situation comes to light. Two PGPs’
primary care provider staffing levels are closer to the national average than
their specialty care rates are. At GHC and HP, the primary care provider rate
is 15–18 percent lower than the national rate; their specialty rate is
about 30 percent lower. At KP, both the primary and specialty care supplies
are about 35 percent lower than the national supply. One issue that may affect
the comparability of PGP and U.S. practice is worth noting here. At the well-organized
PGPs, it is likely that internal medicine subspecialists (and certain other
non–primary care physicians) are serving as true referral specialists,
while in FFS practice, a mix of specialty and general care is not uncommon for
these providers.
Starting from a base of about half of all providers in the primary care specialties,
the trend analysis at KP suggests that this organization (and potentially other
PGPs) has been playing catch-up with regard to national specialist staffing.
Ironically, this PGP specialist growth seems to have occurred at roughly the
same time the national supply of generalists has approached the relatively high
proportion seen within PGPs a few decades earlier. While the causes and consequences
of these obverse trends are not entirely clear, they do lend support to the
premise that the nation must carefully reassess the appropriate balance between
primary and specialty care.
Over the past few years, interest in assessing the “adequacy” of the
U.S. medical workforce has increased dramatically. Key parties are calling for
a reappraisal of our current and future national physician and NPP workforce.
A recent high-profile paper suggested that because of aging of the population
and expansion of economic and social expectations, consumers’ demand for
physicians, particularly for specialists, is likely to grow.14
Accordingly, among some camps there is a growing call for an expansion in medical
training programs. Other analysts believe that the future situation will involve
not so much a shortage, but an inappropriate distribution of what by many yardsticks
will be an adequate or even abundant supply of providers.15
Although the PGP staffing levels reported here might not translate directly
into benchmarks for the United States as a whole, the findings of this paper
indicate that U.S. policymakers should deliberate carefully before concluding
that expansion of medical training programs is warranted, especially given the
huge taxpayer subsidy associated with supporting the training of each new medical
professional.
Another potential implication for medical workforce planners is the finding
that a sizable proportion of NPPs at the PGPs are practicing outside of primary
care. This suggests that the commonly accepted notion that PGPs rely on NPPs
mainly for primary care is not necessarily accurate. It is also noteworthy that
the total supply of NPs/PAs (26–38 per 100,000) at the PGPs was at or below
the current U.S. overall supply ratios (41 per 100,000). This suggests that
increased use of NPPs at the PGPs was not the main reason for the lower physician
staffing rates. This analysis suggests the need for increased attention to the
role of NPPs and their impact on both patients and physicians.
On the surface, workforce
planning and forecasting appears to be largely a statistical undertaking—fueled
by data, shrouded by minutiae, and confounded by countervailing assumptions.
But at its heart, the process is not a technical enterprise. Rather, determining
what a nation’s workforce should, could, or would look like ten to twenty
years hence is fraught with conceptual, political, and even moral challenges
and choices. But this is not to say that these difficult decisions should be
made without using any evidence. Analyses such as this one must provide policymakers
with information that supports rational decision making. When policymakers are
choosing from among the many alternative options, this is the only way that
benefits to society can be maximized, whatever the desired (or feasible) level
of resource commitment. PGPs have devoted considerable energy to the pursuit
of the delicate balance between benefits and costs. Therefore, it seems only
fitting that we look to them as a source of guidance as we chart our course
toward a more optimal medical workforce.
This project would not have been possible without the full cooperation of
the three prepaid group practices that freely “opened their books”
for the purpose of this analysis. The author thanks the leadership of both the
medical groups and health plans at Kaiser Permanente, Group Health Cooperative,
and HealthPartners for this unprecedented support. He also thanks the knowledgeable
and cooperative staffs at these organizations that provided administrative data
from many parts of their organizations, including Glen Hentges at Kaiser Permanente,
Philip Mealand at Group Health Cooperative, and Maureen Peterson and Tammie
Lindquist at HealthPartners. Jennifer Neisner of the Kaiser Permanente Institute
for Health Policy and Cheryl Kaplowitz at the Johns Hopkins University also
played important roles in facilitating this paper at many levels. The editorial
assistance of Tracy Lieberman at Johns Hopkins is also gratefully acknowledged.
This work was supported in part by a grant from the Kaiser Foundation Health
Plan. This work is an independent effort of the author and does not reflect
the position of that organization. The paper was derived in part from a chapter
by the author in A. Enthoven and L. Tollen, eds., Improving Health Care:
The Contributions and Promise of Prepaid Group Practice (Jossey-Bass, forthcoming).
NOTES
1. U.S. Congress, Office of Technology Assessment, Forecasts
of Physician Supply and Requirement, Pub. no. USGPO-052-003-00746-1 (Washington:
U.S. Government Printing Office, 1980); D. Steinwachs et al., “A Comparison
of the Requirements for Primary Care Physicians in HMOs with Projections Made
by the GMENAC,” New England Journal of Medicine 314, no. 4 (1986):
217–222; T. Dial et al., “Clinical Staffing in Staff- and Group-Model
HMOs,” Health Affairs (Summer 1995): 169–180; D. Goodman et
al., “Benchmarking the U.S. Physician Workforce: An Alternative to Needs-Based
or Demand-Based Planning,” Journal of the American Medical Association
276, no. 22 (1996): 1811–1844; and H.R. Mason, “Manpower Needs by
Specialty,” Journal of the American Medical Association 219, no.
12 (1972): 1621–1626.
2. R. Kronick et al., “The Marketplace in Health Care Reform:
The Demographic Limitations of Managed Competition,” New England Journal
of Medicine 328, no. 12 (1993): 148–152; J. Weiner, “Forecasting
the Effects of Health Reform on U.S. Physician Workforce Requirement: Evidence
from HMO Staffing Patterns,” Journal of the American Medical Association
272, no. 3 (1994): 222–230; and A. Tarlov, “HMO Growth and Physicians:
The Third Compartment,” Health Affairs (Spring 1986): 23–35.
3. An expanded Exhibit 2, with the specialty-specific physician
staffing rates for each of the six Kaiser Permanente regions, is available online
at content.healthaffairs.org/cgi/content/full/hlthaff.w4.43v1/DC2.
4. Weiner, “Forecasting the Effects”; J. Weiner, C.
McLaughlin, and S. Gamliel, “Extrapolating HMO Staffing to the Population
at Large,” in The U.S. Health Workforce: Power, Politics and Policy,
ed. M. Osterweis et al. (Washington: Association of Academic Health Centers,
1996), 311–326; and L.G. Hart et al., “Physician Staffing Ratios in
Staff-Model HMOs: A Cautionary Tale,” Health Affairs (Jan/Feb 1997):
55–70.
5. Weiner et al., “Extrapolating HMO Staffing to the Population
at Large.” This methodology helped to account for potential differences
across the two populations within the 65–74 and 75 and older age bands.
It did not account for any potential PGP/U.S. population differences in older
age groups.
6. Hart et al., “Physician Staffing Ratios.”
7. In the states where the three study sites are located, the
rates of physicians per 100,000 population are approximately as follows: Minnesota,
232; Washington, 217; and California, 218. In the metropolitan areas where most
of the plans’ enrollees reside, the supply is considerably higher than
these state-level rates.
8. Weiner, “Forecasting the Effects.”
9. E. Schaefer and J. Reschovsky, “Are HMO Enrollees Healthier
than Others? Results from the Community Tracking Study,” Health Affairs
(May/June 2002): 249–259; and F. Hellinger and H. Wong, “Selection
Bias in HMOs: A Review of the Evidence,” Medical Care Research and Review
57, no. 4 (2000): 405–439.
10. J. Weiner, “The Demand for Physicians in a Changing
Health Care System: A Synthesis,” Medical Care Review 50, no. 4
(1993): 411–449.
11. American Medical Association, Socioeconomic Patterns
of Medical Practice in the U.S., 2002/2003 (Chicago: AMA, 2002).
12. P.R. Kletke et al., “The Growing Proportion of Female
Physicians: Implications for U.S. Physician Supply,” American Journal
of Public Health 80, no. 3 (1990): 300–303.
13. R. Feachem, N. Sekhri, and K. White, “Getting More
for Their Dollar: A Comparison of the NHS with California’s Kaiser Permanente,”
British Medical Journal 324, no. 19 (2002): 135–143.
14. R. Cooper et al., “Economic and Demographic Trends
Signal an Impending Physician Shortage,” Health Affairs (Jan/Feb
2002): 140–154.
15. K. Grumbach, “Fighting Hand to Hand over Physician
Workforce Policy,” Health Affairs (Sep/Oct 2002): 13–27; and
J. Weiner, “A Shortage of Physicians or a Surplus of Assumptions?”
Health Affairs (Jan/Feb 2002): 160–162.
Jonathan Weiner (jweiner{at}jhsph.edu)
is a professor and deputy director of the Health Services Research and Development
Center, Johns Hopkins Bloomberg School of Public Health, in Baltimore, Maryland.
Please click on the author's
names to read related papers by Francis
Crosson, W.
Bruce Fye, David
C. Goodman,
Fitzhugh Mullan, Edward
Salsberg and Gaetano Forte, and Stephen
C. Schoenbaum.
DOI: 10.1377/hlthaff.W4.43
©2004 Project HOPE–The People-to-People Health Foundation, Inc.
|
