Medical Education: Curriculum and Financing Strategies Need to Encourage
Primary Care Training (Letter Report, 10/21/94, GAO/HEHS-95-9).

The proportion of active doctors who are primary care physicians--family
and general practice physicians, general internists, and
pediatricians--has dropped from 53 to 35 since 1960. The Department of
Health and Human Services predicts a shortage of 35,000 primary care
physicians by the year 2000. GAO analyzed student characteristics
associated with choosing primary care and surveyed medical schools and
residency programs. GAO concludes that reversing this trend would
require changes in medical school curricula to expose students to more
primary care medicine, as well as changes in the residency financing,
which now discourages teaching hospitals from sponsoring primary care
training.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  HEHS-95-9
     TITLE:  Medical Education: Curriculum and Financing Strategies Need 
             to Encourage Primary Care Training
      DATE:  10/21/94
   SUBJECT:  Hospital care services
             Students
             Physicians
             Medical education
             Education or training costs
             Financial aid programs
             Medical schools
             Hospitals
             Employment or training programs
             Education program evaluation
IDENTIFIER:  Medicare Program
             National Resident Matching Program
             
**************************************************************************
* This file contains an ASCII representation of the text of a GAO        *
* report.  Delineations within the text indicating chapter titles,       *
* headings, and bullets are preserved.  Major divisions and subdivisions *
* of the text, such as Chapters, Sections, and Appendixes, are           *
* identified by double and single lines.  The numbers on the right end   *
* of these lines indicate the position of each of the subsections in the *
* document outline.  These numbers do NOT correspond with the page       *
* numbers of the printed product.                                        *
*                                                                        *
* No attempt has been made to display graphic images, although figure    *
* captions are reproduced. Tables are included, but may not resemble     *
* those in the printed version.                                          *
*                                                                        *
* A printed copy of this report may be obtained from the GAO Document    *
* Distribution Facility by calling (202) 512-6000, by faxing your        *
* request to (301) 258-4066, or by writing to P.O. Box 6015,             *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time.                                    *
**************************************************************************


Cover
================================================================ COVER


Report to Congressional Requesters

October 1994

MEDICAL EDUCATION - CURRICULUM AND
FINANCING STRATEGIES NEED TO
ENCOURAGE PRIMARY CARE TRAINING

GAO/HEHS-95-9

Primary Care Training


Abbreviations
=============================================================== ABBREV

  AAMC - Association of American Medical Colleges
  COGME - Council on Graduate Medical Education
  HCFA - Health Care Financing Administration
  HHS - Department of Health and Human Services
  HMO - health maintenance organization
  MCAT - Medical College Admission Test
  SAIMS - Student and Applicant Information Management System

Letter
=============================================================== LETTER


B-249760

October 21, 1994

The Honorable David H.  Pryor
Chairman
The Honorable William S.  Cohen
Ranking Minority Member
Special Committee on Aging
United States Senate

The Honorable John Conyers, Jr.
Chairman
Committee on Government Operations
House of Representatives

Since 1960, the proportion of primary care physicians--family and
general practice physicians, general internists, and general
pediatricians--has dropped from about 53 percent to about 35 percent
of the nation's active physicians.  Taking into account current
health care trends emphasizing a growing need for primary care
medicine, the Department of Health and Human Services (HHS) projects
for the year 2000 a shortage of 35,000 generalist or primary care
physicians.\1 The decline in the proportion of active primary care
physicians is related to career decisions that students make during
their medical school years and later in residency training.  While no
single factor can explain why students pursue primary care or
nonprimary care specialties, various studies indicate that the
characteristics of students entering medicine and the educational
process they experience may influence career decisions. 

The federal government contributes to the financing of medical
education and training in several ways.  In 1992, the Medicare
program provided a total of about $5.2 billion in support of graduate
medical education.\2 Through its funding of biomedical research, the
National Institutes of Health helps support medical education at the
undergraduate and graduate levels.  The federal government also
supports medical education and training activities through various
programs authorized under the Public Health Service Act. 

Concerned about the declining ratio of primary care physicians to
nonprimary care physicians, you asked us to assess the role of
medical education in physician specialty choice.  You also asked us
to assess how federal financing of medical education may influence
career choices. 

More specifically, we focused on the

  characteristics associated with students who are more likely to
     choose generalist or primary care specialties in medical school,

  curriculum requirements that expose medical students and residents
     to primary care training, and

  the role federal financing plays in setting the focus of medical
     education. 


--------------------
\1 This projection assumes no changes in the current system of
medical training and a health care system dominated by managed care
arrangements.  More specifically, it is assumed that two-thirds of
the U.S.  population will be enrolled in some type of managed care
arrangement with strong utilization controls, whether a staff model
health maintenance organization, independent practice association, or
a network.  (See Council on Graduate Medical Education, Fourth
Report:  Recommendations to Improve Access to Health Care Through
Physician Workforce Reform, U.S.  Department of Health and Human
Services, Jan.  1994.)

\2 The federal government also contributes to the financing of
graduate medical education through programs administered by the
Department of Veterans Affairs, the Department of Defense, and
through federal sharing in states' costs of the Medicaid program. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

Choice of career paths in medicine is associated with the
characteristics of students admitted to medical schools and with the
curriculum and training opportunities they receive during their
medical education. 

Although the strongest predictor of whether students would choose
primary care careers was their stated preference for primary care
before they entered medical school,\3 we found that some features of
medical schools were associated with an increased likelihood that
students would go into primary care.  Foremost among these was
whether the medical school had a family practice department--students
who attended schools with family practice departments were more
likely to pursue primary care than students who attended schools
without such departments.  Other factors moderately favoring primary
care were whether a family practice clerkship was required before
career decisions were made in the fourth year and how well family
practice departments were funded. 

Similarly, once students graduate from medical school and enter
residency programs, we found a significant association between
program opportunities to expose residents to primary care faculty and
the proportion of residents choosing generalist practice. 

Although curriculum and training opportunities appear to influence
student choice of primary care, we found that many medical schools
did not require primary care experiences before students chose
residency programs.  Moreover, once in a residency program, residents
spent only a small portion of training time on providing primary care
or working in a primary care setting. 

The way residency training is financed contributes to a specialist
orientation for the clinical education of medical students.  Only
hospitals or hospital-affiliated providers are eligible to receive
Medicare payments for direct training costs.  Current financing
mechanisms for residency training can give greater incentives to
training in nonprimary care-oriented settings than in primary
care-oriented settings.  This can place primary care residency
programs at a disadvantage because of their use of nonhospital
settings for training.  In our view, the Medicare payment methodology
may need to be modified to provide incentives for training in
nonhospital settings. 


--------------------
\3 We did a logistic regression that is discussed on pp.  7-8, and
more fully in app.  I, pp.  30-32. 


   BACKGROUND
------------------------------------------------------------ Letter :2

Medical career decisions are usually made at three specific points
during the educational continuum:  (1) at the end of college when
students typically apply to medical school, (2) during the fourth
year of medical school when students elect the area of medicine to
pursue and enter residency training, and (3) at the end of residency
training when residents decide to enter practice or to train further
for a subspecialty. 

College students who choose to pursue a career in medicine take the
Medical College Admission Test (MCAT) and apply to medical schools. 
At that early point in the educational continuum, some students
express a preference for a primary care or nonprimary care
specialty.\4

During the 4 years of medical school, students are instructed in the
basic sciences and learn about the major medical disciplines through
clinical training.\5 Clinical training usually includes
preceptorships, in which medical students observe a physician in
practice, and clerkships, in which third- and fourth-year medical
students participate with medical residents and faculty in the
diagnosis and treatment of patients.  During the fourth year of
medical school, students formally select the specialty area they are
interested in by applying directly to residency training programs. 
This process is facilitated by the National Resident Matching
Program, through which student choices for residency training
programs are matched with available positions and the preferences of
program directors for candidates.\6 Several experts suggest that
educational experiences and the role models encountered during the
first 3 years of medical school are among the factors that influence
the career choices of medical students. 

After students graduate from medical school, they enter residency
programs that prepare them for independent practice in the chosen
specialty areas largely through on-the-job training.  This training
traditionally takes place in teaching hospitals.\7 Residency training
generally comprises required and elective rotations and can include
continuity of care assignments.\8 Through these training experiences,
residents treat patients under the supervision of teaching
physicians.  Residency training continues for 3 to 7 years depending
on the discipline.  In some disciplines, residents who complete
general residency training may enter practice or may further
subspecialize.  For example, a resident who completes training in
general internal medicine may decide to enter practice or to pursue
further training in cardiology, a subspecialty of internal medicine
that focuses on the heart. 


--------------------
\4 Individuals registering for the MCAT also complete a short
survey--the Premedical Student Questionnaire--detailing demographic
characteristics, their personal background, reasons for wanting to
study medicine, and specialty preference. 

\5 The basic sciences include the subjects of anatomy, biochemistry,
physiology, microbiology, pharmacology, and pathology.  The medical
disciplines include disciplines such as internal medicine,
pediatrics, family medicine, psychiatry, and obstetrics/gynecology. 

\6 Residency programs may be categorized as programs that, in
practice, typically lead to primary care careers and those that
typically lead to nonprimary care careers. 

\7 Teaching hospitals are hospitals with one or more graduate medical
education programs approved by the Accreditation Council for Graduate
Medical Education or the American Osteopathic Association. 

\8 For example, residents in family practice are required to
participate in internal medicine rotations.  However, these residents
may elect to participate in subspecialty rotations.  In addition,
family practice, as well as internal medicine and pediatrics, has
specific requirements for continuity of care training.  Continuity of
care assignments afford residents the opportunity to provide first
contact and ongoing care to a group of patients over time. 


      PRIMARY CARE MEDICINE
      INVOLVES A GENERAL BODY OF
      KNOWLEDGE
---------------------------------------------------------- Letter :2.1

Primary care is delivered by family and general practice physicians,
general internists, and general pediatricians.\9 These physicians are
broadly trained to evaluate a spectrum of undifferentiated health
problems, manage acute and chronic conditions, and address disease
prevention and health promotion.  The focus of primary care is not
organ-specific, as is the focus of such specialties as cardiology or
nephrology.  Primary care is also characterized by care that is
comprehensive and continuous, requires broad diagnostic skills, and
is usually practiced in ambulatory settings such as physician offices
or clinics.  Coordinating a patient's overall care, which entails
consulting with other physicians and referring for necessary
specialized services, is another important aspect of primary care. 


--------------------
\9 Some studies include obstetrics/gynecology in their definition of
primary care.  Several studies have shown that other specialists
provide some primary care to their patients.  In addition, nurse
practitioners, physician assistants, and others make significant
contributions to the provision of primary care. 


      INTEREST IN PRIMARY CARE
      DECLINING
---------------------------------------------------------- Letter :2.2

Interest in primary care careers among medical students has decreased
during the past decade.  The proportion of senior medical students
planning careers in primary care specialties decreased from about 32
percent in 1984 to about 19 percent in 1993.\10

Interest in primary care careers has also decreased among residents. 
Although about 40 percent of first-year residency positions are in
internal medicine, pediatrics, and family practice programs,\11 many
residents in such programs pursue additional training and enter
practice as subspecialists rather than as primary care physicians. 
For example, although first-year residents in internal medicine
constitute more than half of the pool of potential primary care
physicians, it has been estimated that 55 to 68 percent of internal
medicine residents elect to subspecialize.  In the discipline of
pediatrics, it has been estimated that between 18 and 40 percent will
subspecialize.\12 \,\13

Various studies indicate that many factors can influence the career
choices of medical students and residents.  Medical school graduates,
for example, report factors such as physician role models and
clerkships as strong influences on their specialty decisions.\14 For
residents, economic factors such as income potential and job
opportunities in a specialty area and the residency training
experience are among the factors influencing their decisions to
further subspecialize. 

This report focuses on identifying personal and educational factors
that influence student career decisions during the fourth year of
medical school and describes characteristics of the educational
process that may orient students and residents to pursue primary care
careers.  The report also explores the role of federal financing in
orienting the focus of medical education. 


--------------------
\10 The proportion of senior medical students planning careers in
primary care specialties reached a low for the 1984-93 period of 14.6
percent in 1992.  Although it increased, the proportion of senior
medical students with primary care career plans in 1993 remained
below the 1984 level.  This increase is thought to be associated with
a growing emphasis on primary care by medical schools as well as
special initiatives to foster generalist specialties. 

\11 Based on first-year positions available for 1993. 

\12 Family physicians who pursue additional training do not
necessarily become subspecialists.  For example, family physicians
may pursue additional training for added qualifications in
geriatrics, which is not a subspecialty but rather provides
additional expertise in the discipline. 

\13 These ranges are based on various subspecialization estimates
cited in the literature. 

\14 Among factors influencing specialty decisions, 1993 medical
school graduates also reported perceived fit of personality, skills,
and ability with the selected field as major influences.  Prestige
and authority factors, lifestyle variables, economic influences, and
income prospects overall were given relatively low ratings.  See D.G. 
Kassebaum, M.D., and P.L.  Szenas, M.A., "Factors Influencing the
Specialty Choices of 1993 Medical School Graduates," Academic
Medicine, 69 (1994), pp.  164-70. 


   SCOPE AND METHODOLOGY
------------------------------------------------------------ Letter :3

To identify the characteristics associated with students who are more
likely to choose primary care or nonprimary care specialties, we
combined information from 3 databases on nearly 9,000 medical school
graduates in the class of 1989.\15 We then applied a statistical
technique, logistic regression, to identify the significant
characteristics of students more likely to choose primary care
specialties.  In logistic regression, we can assess the association
of each characteristic, or factor, to the likelihood of choosing
primary care specialties while controlling for the effects of the
other characteristics in the model. 

We also surveyed all 126 allopathic and 15 osteopathic medical
schools in the United States.\16 About 89 percent of allopathic
medical schools and 100 percent of osteopathic medical schools
responded to our survey.  The questionnaire sought to determine the
extent to which schools required observation or training in primary
care as part of their curricula. 

To examine the extent to which residents are exposed to primary care
medicine, we surveyed a sample of 534 residency program directors
from allopathic and osteopathic programs in the disciplines of
internal medicine, pediatrics, family practice, and osteopathic
general practice.  About 82 percent of program directors responded to
our survey.  In the allopathic disciplines, we distinguished between
traditional track programs and primary care track programs.  We
defined primary care track programs as (1) programs listed in the
National Resident Matching Program as having a primary care focus or
(2) programs receiving funds from the Health Resources and Services
Administration to support a primary care curricular focus within the
program. 

Finally, we drew from our previous work on the federal financing of
graduate medical education;\17 reviewed numerous studies and official
documents related to medical specialty choice and medical education
funding; and interviewed officials of government agencies that fund
medical education, medical school deans, hospital administrators,
directors of residency programs, academic medical centers, and
representatives of professional medical associations.  (See apps.  I
and III for details of our scope and methodology.)

We conducted our work from May 1992 to September 1994 in accordance
with generally accepted government auditing standards. 


--------------------
\15 We aggregated data from three national surveys conducted by the
Association of American Medical Colleges (AAMC) to construct a
database of information on medical students graduating in 1989.  The
three surveys are the (1) Premedical Student Questionnaire, (2)
Matriculating Student Questionnaire, and (3) Graduation
Questionnaire. 

\16 Allopathic medicine is the most common form of medical practice. 
Graduates of allopathic medical schools receive M.D.s.  Osteopathic
medicine is a form of medical practice similar to allopathic medicine
that also incorporates manual manipulation of the body as a therapy. 
Graduates of osteopathic medical schools receive D.O.s. 

\17 Medicare:  Graduate Medical Education Payment Policy Needs to Be
Reexamined (GAO/HEHS-94-33, May 5, 1994). 


   LIKELIHOOD OF CHOOSING PRIMARY
   CARE SPECIALTY INFLUENCED BY
   VARIOUS FACTORS
------------------------------------------------------------ Letter :4

Our analyses showed that a set of factors related to medical
education is associated with an increased likelihood of students
choosing to pursue primary care careers.  These factors include
characteristics of the students that medical schools admit and
training that students receive in medical school and later in
residency programs.  Our review of the literature found that research
is generally consistent with our findings.\18


--------------------
\18 In our analyses we used two models.  Model 1 included student
data from all schools.  Model 2 included student data from schools
with departments of family practice.  About four-fifths of the
students in our original database are included in Model 2. 


      SOME PERSONAL
      CHARACTERISTICS ARE
      ASSOCIATED WITH PRIMARY CARE
      CHOICE
---------------------------------------------------------- Letter :4.1

Using logistic regression, we examined the relationship of the
characteristics of medical students and the schools they attend to
the likelihood of students choosing a primary care career in the
fourth year of medical school.  In our analyses, the strongest
predictor of choosing primary care was a student's intention to
pursue a primary care specialty as stated at what is typically the
first decision point--the last year of college before attending
medical school.  Students who stated such a preference were about
twice as likely to pursue primary care as students who expressed a
preference for a nonprimary care specialty and those who did not have
a preference. 

In addition, the following sociodemographic characteristics were
associated with the greater likelihood of students choosing to pursue
primary care careers:\19

  Married students were 40 percent more likely than unmarried
     students to pursue primary care. 

  Female students were 54 percent more likely than males to pursue
     primary care. 

  Mexican-American students were 66 percent more likely than white
     Americans to pursue primary care. 

  Students who spent most of their high school years in rural areas
     (fewer than 10,000 inhabitants) were 60 percent more likely to
     pursue primary care than students from nonrural areas.\20


--------------------
\19 Results are from Model 1 only. 

\20 We did not include students' income prospects in our models
because this variable, in other AAMC surveys, has consistently been
reported to have limited influence on specialty choice.  See "Factors
Influencing the Specialty Choice of 1993 Medical School Graduates."


      MEDICAL SCHOOL EXPERIENCES
      AFFECT STUDENTS' CAREER
      CHOICE
---------------------------------------------------------- Letter :4.2

We also analyzed characteristics of the schools that students
attended.  We were specifically interested in assessing the
association of characteristics thought to support primary care
experiences during medical school.  These characteristics include the
existence of a family practice department, a required family practice
clerkship during the third year,\21 the funding level of family
practice departments (defined as the ratio of family practice
department funding to the number of enrolled students), and whether
the school was public or nonpublic.  The research intensity of the
schools students attended was also included in the model to assess
whether medical schools that receive large amounts of research
funding orient students away from primary care medicine. 

The following characteristics of the medical schools were associated
with a greater likelihood of students choosing to pursue primary care
careers:\22

  Students who attended schools with family practice departments were
     57 percent more likely to pursue primary care than those
     attending schools without family practice departments. 

  Students who attended schools requiring a third-year family
     practice clerkship were 18 percent more likely to pursue primary
     care than students attending schools without this requirement. 

  Students attending medical schools with more highly funded family
     practice departments were 18 percent more likely to pursue
     primary care than those attending schools with lower funding.\23

  Students attending public medical schools were 38 percent more
     likely to pursue primary care careers than students attending
     nonpublic medical schools. 

The hypothesis that students attending medical schools that receive
large amounts of research funding are oriented away from primary care
was not supported in our model.  Career choices of students, for
either primary care or nonprimary care, were not associated with the
research intensity of the schools they attended.  \24 \25 Appendix I
provides the details of our analyses. 


--------------------
\21 We focused on family practice departments because this discipline
does not generally include experiences in subspecialty training. 
Internal medicine and pediatric departments provide experiences in
both the primary care and subspecialty areas in these disciplines. 

\22 The first result in this list is from Model 1; the others are
from Model 2. 

\23 More highly funded family medicine departments are those with the
highest ratio of total departmental revenues to number of students. 
In 1989, the highest ratio was $6,570 or more (the highest third)
compared with less than $3,157 (the lowest third). 

\24 The research intensity of a school was expressed as a ratio of
total federal research dollars to the number of students enrolled in
the school. 

\25 These associations were found to be statistically significant
while controlling for the effects of the other variables in the
model.  For example, the finding that students who attended schools
with family practice departments were 57 percent more likely to
pursue primary care than students attending medical schools without a
family practice department was significant, controlling for the
research intensity of the school, the students' initial preferences
to pursue primary care or specialty medicine, and the students' age
or sex. 


   MEDICAL EDUCATION AND TRAINING
   DO NOT STRESS PRIMARY CARE
   MEDICINE
------------------------------------------------------------ Letter :5

The results of our logistic regression models suggest that in
addition to characteristics that predispose students towards primary
care or nonprimary care medicine, characteristics of medical schools
also exert an influence on the career choices of medical students. 
Studies of residency programs found in the literature also suggest
that the training environment--clinical experiences and faculty role
models--can influence residents' decisions to enter practice or to
continue training and subspecialize.  Through our survey of medical
schools and residency programs, however, we found that many of the
medical schools did not have required curricula that expose students
to primary care experiences before the fourth year of medical school
when students formally select the area of medicine to pursue.\26
Similarly, only a small portion of resident training time was spent
in settings that give residents experience with continuous and
comprehensive care or undifferentiated health problems.  Finally,
instructors and faculty, who could serve as students' and residents'
role models, were infrequently primary care physicians.  This
orientation toward specialty medicine within education and training
may contribute to a smaller proportion of students and residents
choosing primary care careers. 


--------------------
\26 Since 1992, the Liaison Committee on Medical Education's
standards for accreditation of medical schools have required schools
to offer a core curriculum in primary care. 


      MEDICAL STUDENTS RECEIVE
      LIMITED EXPOSURE TO PRIMARY
      CARE MEDICINE
---------------------------------------------------------- Letter :5.1

Medical students' early contact with primary care can occur during
the first or second year of medical school through preceptorships in
which students observe practicing physicians in the community.  About
35 percent of schools responding to our survey (45 of 127 schools\27
) indicated that they required preceptorships for first- or
second-year medical students.  Of these schools, about 65 percent did
not require primary care preceptorships. 

Clerkships, which for the most part constitute the medical school
curriculum for the third and fourth years, provide students with
clinical training in which they learn by observing and actively
participating in the care of patients.  Third-year clerkships, in
particular, are an important influence on student decisions regarding
the types of residency training programs they will pursue.  Our
survey results indicate that of schools offering clerkships in
internal medicine, pediatrics, and family practice, about 53 percent
required clerkships in all three.  Moreover, about one-third of
schools with clerkship requirements for all three disciplines\28
required rotations with a primary care focus by the end of the third
year.  More specifically, about

  60 percent of schools did not require a primary care rotation as
     part of required third-year clerkships in internal medicine,\29

  38 percent of schools did not require a primary care rotation as
     part of required third-year clerkships in pediatrics,\30

  42 percent of allopathic schools did not require a family practice
     clerkship, and

  13 percent of osteopathic schools did not require a general
     practice clerkship. 

In addition, on average about 58 percent of total clinical
faculty--individuals who could serve as role models--were in
nonprimary care specialties and subspecialties.  More specifically,
about 68 percent of tenured faculty, about 62 percent of nontenured
faculty, and about 55 percent of active volunteer faculty were in
nonprimary care specialties and subspecialties. 


--------------------
\27 Overall, 127 of 141 allopathic and osteopathic medical schools
responded to our survey. 

\28 Based on allopathic schools only. 

\29 All allopathic and osteopathic schools responding to our survey
required an internal medicine clerkship for third-year students. 

\30 About 98 percent of allopathic and osteopathic schools responding
to our survey required a pediatrics clerkship for third-year
students. 


      RESIDENTS MOSTLY TRAINED IN
      NONPRIMARY CARE SETTINGS
---------------------------------------------------------- Letter :5.2

Residency programs whose residents constitute the pool of potential
primary care physicians--programs in internal medicine, pediatrics,
family practice, osteopathic general practice--provided limited
primary care training.\31 Our survey of such programs indicated that
there was little exposure to primary care medicine in ambulatory
settings.  The survey also showed that even in family practice
programs, training in ambulatory care settings was limited. 

As part of residency training, residents must complete required
rotations through which they are acquainted with various areas of
medical knowledge.  These rotations are typically comprised of
1-month "blocks." Our survey indicated that on average between
one-third and one-half of required block time was typically spent in
primary generalist rotations.  (See fig.  1.)

   Figure 1:  Proportion of Total
   Block Rotation Time Spent in
   Generalist Rotations

   (See figure in printed
   edition.)

Note:  Except for osteopathic internal medicine and primary care
pediatrics, for which all programs were sampled, sampling errors at
the 95-percent confidence level for estimates in the figure ranged
from 2 percentage points for family practice (generalist) to 6
percentage points for osteopathic general practice (generalist)
programs. 

Our survey data also indicated that even for generalist rotations,
residents spent, on average, most of their time in hospital inpatient
settings.\32 The proportion of time spent in hospital outpatient or
ambulatory settings during generalist rotations ranged from an
average of 8 percent for traditional track internal medicine programs
to 27 percent for osteopathic general practice programs.  Moreover,
the proportion of time spent in community-based outpatient
settings--which most closely resemble primary care practice
settings--ranged from an average of 2 percent for traditional track
internal medicine programs to 31 percent for osteopathic general
practice programs.  (See fig.  2.)

   Figure 2:  Average Proportion
   of Generalist Rotation Time
   Spent in Outpatient Settings

   (See figure in printed
   edition.)

Note:  Except for osteopathic internal medicine and primary care
pediatrics, for which all programs were sampled, sampling errors at
the 95-percent confidence level for estimates in the figure ranged
from 1 percentage point for traditional track internal medicine
(community-based) to 11 percentage points for osteopathic general
practice (community-based) programs. 

A portion of residency training also consists of elective rotations,
where residents can augment their training based on perceived needs
or interests.  Elective rotations constitute about one-third of total
residency training time.  Our survey indicated that
specialty-oriented rotations in hospital settings were the most
frequently chosen elective rotations--in most programs about one-half
or more of residents elected such rotations.\33 In contrast, the
average proportion of residents electing primary care rotations in
community-based ambulatory settings tended to be lower--the
proportions varied across programs from 43 percent for family
practice residents to 9 percent for osteopathic general internal
medicine residents.  (See fig.  3.)

   Figure 3:  Average Proportion
   of Residents Electing Rotations
   in Primary Care Community-Based
   Ambulatory Settings

   (See figure in printed
   edition.)

Note:  Except for osteopathic internal medicine and primary care
pediatrics, for which all programs were sampled, sampling errors at
the 95-percent confidence level for estimates in the figure ranged
from 4 percentage points for traditional track pediatrics to 12
percentage points for osteopathic general practice programs. 

A relatively small portion of residency training consists of
continuity of care assignments.  Ideally, in these assignments,
residents are assigned specific patients who are seeking care for the
first time for a new condition or routine care.  Residents are
expected to follow these patients over time, provide continuous care,
and learn to recognize and manage illnesses.  The amount of time
spent in continuity of care assignments is usually accumulated in
terms of half-days. 

Our survey found that the amount of time that programs required
residents to spend in continuity of care assignments varied across
programs from an average of 383 half-days for family practice
programs to an average of 142 half-days for traditional track general
internal medicine programs.  (See fig.  4.)

   Figure 4:  Average Number of
   Half-Days Spent in Continuity
   of Care Assignments

   (See figure in printed
   edition.)

Notes:  Data reflect half-days spent in continuity of care
assignments during postgraduate training years one through three.

Except for osteopathic internal medicine and primary care pediatrics,
for which all programs were sampled, sampling errors at the
95-percent confidence level for estimates in the figure ranged from
12 half-days for traditional track pediatrics to 25 half-days for
osteopathic general practice programs. 

Continuity of care assignments should offer experiences that closely
approximate what a generalist physician will do in primary care
practice; thus, such assignments are considered a fundamental
training component for primary care physicians.  Our survey
indicated, however, that on average a majority of residents in almost
all programs fulfilled their continuity of care assignments in
hospital-based general medicine clinics.\34 Moreover, substantial
proportions of patients in continuity of care assignments within most
programs were not assigned to residents for first contact care for a
new condition or routine care.  In our survey, the proportion of
patients assigned for follow-up care, rather than for first contact
care, varied across programs from an average of 16 percent for family
practice programs to an average of 54 percent for osteopathic general
internal medicine programs.  (See fig.  5.)

   Figure 5:  Average Proportion
   of Patients Assigned to
   Residents in Continuity Clinics
   for First Contact Care and for
   Follow-Up Care

   (See figure in printed
   edition.)

Notes:  Totals do not add to 100 percent because of an "other"
category that is not shown here.

Except for osteopathic internal medicine and primary care pediatrics,
for which all programs were sampled, sampling errors at the
95-percent confidence level for estimates in the figure ranged from 4
percentage points for family practice (first contact) and traditional
track internal medicine to 16 percentage points for osteopathic
general practice (first contact) programs. 


--------------------
\31 More specifically, the categories of programs surveyed were
traditional track internal medicine, primary care track internal
medicine, osteopathic internal medicine, traditional track
pediatrics, primary care track pediatrics, family practice, and
osteopathic general practice.  The "averages" presented hereafter
represent the average for each program category. 

\32 The exception was osteopathic general practice programs in which
residents spent, on average, about 38 percent of their time during
generalist block rotations in hospital inpatient settings. 

\33 The exception was residents in osteopathic general practice
programs.  About 38 percent of osteopathic general practice residents
elected specialty-oriented rotations in hospital settings. 

\34 The exceptions were osteopathic general internal medicine and
general practice programs.  For osteopathic general internal medicine
programs, about 47 percent of residents on average fulfilled
continuity of care assignments in other primary care settings.  For
general practice programs, about 45 percent of residents on average
fulfilled their continuity of care assignments in other nonprimary
care settings. 


      RESIDENTS MAINLY ENCOUNTER
      NONPRIMARY CARE ROLE MODELS
---------------------------------------------------------- Letter :5.3

Residents in almost all programs were trained predominately by
nonprimary care faculty.\35 For internal medicine and for pediatrics
programs in our survey, an average of 40 percent or less of the
faculty were generalist physicians.  Thus, in these two disciplines,
programs had few generalist role models training residents in primary
care medicine.  (See fig.  6.)

   Figure 6:  Average Proportion
   of Total Faculty Who Were
   Generalist Physicians

   (See figure in printed
   edition.)

Note:  Except for osteopathic internal medicine and primary care
pediatrics, for which all programs were sampled, sampling errors at
the 95-percent confidence level for estimates in the figure ranged
from 3 percentage points for traditional track internal medicine and
pediatrics to 7 percentage points for family practice and osteopathic
general practice programs. 


--------------------
\35 The exceptions were osteopathic general practice and allopathic
family practice programs.  For general practice programs, on average
about 55 percent of faculty were generalists; for family practice
programs, on average about 68 percent of faculty were generalists. 


      INTERNAL MEDICINE PROGRAMS
      THAT EMPHASIZED PRIMARY CARE
      PRODUCED MORE GRADUATES WHO
      ENTERED PRIMARY CARE
      PRACTICE
---------------------------------------------------------- Letter :5.4

While residents in internal medicine programs constitute the largest
portion of primary care residents, it has been estimated that more
than half of such residents pursue additional training and enter
practice as subspecialists.  Our survey data suggest that internal
medicine residency programs provided limited exposure to primary care
faculty and primary care medicine in ambulatory settings.  To further
explore the relationship between program opportunities to acquaint
residents with primary care faculty and medicine and the proportion
of residents choosing to enter generalist practice, we conducted a
separate analysis of both categories of internal medicine
programs.\36

Our analysis showed a significant association between program
opportunities to expose residents to primary care faculty and the
proportion of residents choosing generalist practice.\37 On average,
a greater proportion of residents entering generalist practice was
found to be associated with internal medicine programs with a higher
than average proportion of

  full-time primary care faculty,

  hospital rounds taught by primary care faculty, and

  required inpatient block rotations taught by primary care faculty. 

Our analysis also showed a significant association between certain
characteristics of a program's continuity of care assignments and a
program's output of generalist physicians.  A greater proportion of
residents entering generalist practice was associated with programs
that provided for

  continuity of care clinic experiences in which more than 80 percent
     of training time was spent in a primary care setting and

  exposure to a higher than average percentage of continuity of care
     clinic patients seeking care for the first time for a new
     condition. 

These results suggest that training programs and their curricula may
influence residents' decisions to practice general medicine or
specialize. 


--------------------
\36 In this analysis, primary care and traditional track programs
were combined. 

\37 Program characteristics reported to be statistically significant
reflect achieving a chi square test result with a p-value at the
95-percent confidence level or greater (p<.05); that is, if there was
no difference in the universe, it is unlikely (less than a 5-percent
chance) our sample results would show a difference of this magnitude. 


      LIMITED REIMBURSEMENT FOR
      NONHOSPITAL-BASED TRAINING
      CITED AS A BARRIER
---------------------------------------------------------- Letter :5.5

The lack of reimbursement for training residents in settings other
than hospitals was reported to be a major barrier to the
establishment or maintenance of community-based ambulatory training. 
With regard to barriers to training residents outside the hospital
setting, the following three were most commonly cited by most
residency programs:\38

  insufficient government reimbursement for training residents in
     community-based ambulatory settings,

  insufficient private payer reimbursement for training in
     community-based ambulatory settings, and

  insufficient government reimbursement for services provided in
     community-based ambulatory settings. 


--------------------
\38 The exception was primary care internal medicine programs.  The
three most commonly cited barriers for those programs were
insufficient government reimbursement for training residents in
community-based ambulatory settings, insufficient private payer
reimbursement for training in community-based ambulatory settings,
and hospital service or staffing needs (instead of insufficient
government reimbursement for services provided in community-based
ambulatory settings). 


   HOSPITALS PLAY DOMINANT ROLE IN
   MEDICAL SCHOOL AND RESIDENCY
   TRAINING
------------------------------------------------------------ Letter :6

Residency programs are primarily sponsored by and based in teaching
hospitals.  Medical schools rely on teaching hospitals for the
clinical training aspects of the medical school curriculum.  Through
such affiliations, teaching physicians in the hospital supervise
residents and, assisted by residents, instruct third- and fourth-year
medical students.  Our surveys indicated that through such teaching
arrangements, the clinical training of residents and, in turn, of
medical students consisted mainly of experiences with specialist role
models and hospital patients.  For the most part, hospital patients
do not require those diagnostic or clinical practice skills
characteristic of primary care medicine:  evaluation of
undifferentiated health problems and comprehensive and routine care. 
As a result, many residents and medical students have little
opportunity to have experiences that most resemble primary care
practice. 


   FINANCING MECHANISMS CONTRIBUTE
   TO FOCUS ON SPECIALTY MEDICINE
------------------------------------------------------------ Letter :7

The way residency training is financed contributes to a specialist
orientation within medical education and training.  In general, there
are financial disincentives for teaching hospitals to sponsor primary
care training; current financing mechanisms for residency training
are more supportive of training in specialist-oriented settings than
in generalist-oriented settings.  Because the clinical experiences of
medical students are linked to the training of residents, residency
program financing can shape the types of role models and training
experiences medical students have.\39

The chief means of support for residency programs are teaching
hospital revenues from patient care.\40 Hospital- based services
usually generate more revenue for medical service plans and the
hospital itself.  Primary care, for the most part, is an ambulatory
practice; that is, it is largely conducted in nonhospital settings,
such as doctors' offices and clinics.  Because inpatient care
services and specialty education generate more revenues, there is a
disincentive for educators to increase the time that residents spend
in outpatient or ambulatory care settings.\41 With such differences
in revenues, sponsoring primary care training programs may be
financially disadvantageous to teaching hospitals. 

Medicare, unlike private third party payers, makes separate payments
to hospitals for its portion of the "direct" and "indirect" costs of
graduate medical education.\42,43

Historically, the Congress viewed Medicare support for residency
training programs as necessary to help meet community needs for
trained health personnel.\44 Absent federal guidance on the number
and types of residents to be trained, in effect Medicare relies
primarily on hospitals to determine the specialty distribution of
physicians to be trained.  During the 1989-91 period, hospitals used
Medicare direct medical education funds to support the training of 75
percent specialists and 25 percent generalists. 

In addition, Medicare's payment methodology also creates barriers to
primary care training by limiting payment for training in
nonhospital-based settings.  Under current Health Care Financing
Administration (HCFA) rules, only hospitals and hospital-based
providers are eligible to receive Medicare payments for training
costs in nonprovider settings.  That is, when residents do train in
outpatient or ambulatory settings, Medicare only reimburses the
direct costs of such training when the ambulatory care provider has a
teaching agreement with a hospital.  This is because Medicare limits
such reimbursement for training in ambulatory settings to those
programs for which hospitals incur almost all or substantially all of
the training costs. 


--------------------
\39 See app.  II for information on certain federal programs that
support residency training and medical school education. 

\40 Patient-care revenues are also a revenue source for medical
schools.  Under a financing arrangement known as medical service
plans, a portion of the revenue generated by clinical faculty patient
services helps fund medical school departments.  During the 1990-91
period, medical service plan revenues comprised about 31 percent of
total revenues for U.S.  medical schools.  According to Eli Ginzberg
and others, these funds are used to cover the salaries of most of the
expanded clinical staff and also help pay for some departmental and
general medical school operations (The Economics of Medical
Education, Josiah Macy Foundation (New York, 1993), p.  34). 

\41 See Primary Care Physicians:  Financing Their GME in Ambulatory
Settings, Institute of Medicine (1989). 

\42 Hospital charges are generally set at levels high enough to cover
a portion of the facility's training costs; private payers contribute
toward such costs in this way.  However, in the current marketplace,
many large-scale purchasers make no distinction between the price
they are willing to pay to a teaching hospital versus a nonteaching
hospital, despite teaching hospitals' higher costs.  Furthermore,
many purchasers try to encourage their beneficiaries to use less
costly providers.

\43 Direct costs include teachers' and residents' salaries as well as
facility and equipment expenses.  Indirect costs include those higher
patient care costs thought to be due to such factors as increased
diagnostic testing, increased number of procedures performed, and
higher staffing ratios.  In 1992, Medicare provided about $1.46
billion for direct costs of resident training and $3.56 billion for
indirect costs.  See GAO/HEHS-94-33. 

\44 Committee reports indicated that these educational activities
enhance the quality of care in an institution and that Medicare
should recognize these costs for reimbursement purposes until
communities undertake to bear such costs in another manner. 


   CONCLUSIONS
------------------------------------------------------------ Letter :8

Our analyses of student characteristics associated with choosing
primary care and the results of our surveys of medical schools and
residency programs suggest that training institutions may be able to
do more to increase the number of practicing primary care physicians. 

Medical schools, for example, could evaluate their recruitment and
admissions policies to assess how much importance is placed on
recruiting and admitting students who are interested even before
entering medical school in pursuing primary care specialties.  Our
model results indicate that such students were twice as likely to
pursue training in primary care.  Medical schools could also assess
their success in recruiting and admitting students from diverse
sociodemographic backgrounds.  Our model results suggest that schools
with diverse student bodies are more likely to have a larger pool of
students interested in pursuing primary care careers. 

The results of our survey also indicate that primary care training
did not have a prominent place in most medical schools.  Schools
could help cultivate or maintain an interest in primary care by
providing students with greater exposure to primary care curricula
and role models before the fourth year when medical career decisions
are made.  Our survey results of residency programs that train the
pool of potential primary care physicians indicate that these
programs provided relatively few role models and clinical experiences
that closely resemble primary care practice.  This situation was due,
in part, to current financing mechanisms that provide more support
for residency training in hospital settings and for
specialty-oriented clinical faculty than for primary care training
and faculty.  In particular, Medicare's payment methodology for the
direct costs of residency training tends to reinforce a specialty
orientation within physician training.  Because medical care
continues to move outside the boundaries of the hospital, we believe
that in addition to supporting hospital-based training, the federal
government may want to encourage greater training in nonhospital
settings. 


   MATTER FOR CONGRESSIONAL
   CONSIDERATION
------------------------------------------------------------ Letter :9

To support the training of primary care physicians, the Congress may
want to consider modifying Medicare's payment methodology for the
direct costs of graduate medical education to provide incentives for
training in nonhospital settings. 


---------------------------------------------------------- Letter :9.1

Officials at the Bureau of Health Professions, Health Resources and
Services Administration reviewed a draft of this report.  They
generally agreed with the information presented.  We have
incorporated their comments where appropriate. 

Previously, HHS officials commented on a draft of our report,
GAO/HEHS-94-33, on Medicare's payment methodology for the direct
costs of graduate medical education.  In those remarks, HHS officials
stated that the Council on Graduate Medical Education (COGME), which
is adminstered by the Public Health Service and reports to the
Secretary and the Congress on matters related to graduate medical
education, has stated many of the same concerns regarding barriers to
primary care training contained in that report.  HHS officials
further stated that COGME is concerned that this payment methodology
provides an incentive to add residency positions based on hospital
service needs rather than societal and educational needs.  This
incentive is inconsistent with the view that there should be more
educational experiences at nonhospital, community-based sites. 

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days
after its issue date.  At that time, we will send copies to others on
request.  If you have any questions about this report, please call me
at (202) 512-7119.  GAO contacts and staff acknowledgments are listed
in appendix IV. 

Sarah F.  Jaggar
Director, Health Financing
 and Policy Issues


PERSONAL AND INSTITUTIONAL
CHARACTERISTICS ASSOCIATED WITH
CHOICE OF PRIMARY CARE RESIDENCY
PROGRAM
=========================================================== Appendix I

Using statistical models, we performed two analyses to identify
selected characteristics associated with the likelihood that a
graduating medical student would seek certification in a primary care
specialty.  On the basis of literature reviewed and interviews with
medical education experts, we identified the major factors claimed to
influence student specialty choice and incorporated into our analyses
those factors for which there were data.  Specifically, we sought to
(1) identify the characteristics of students who were more likely to
choose generalist or primary care specialties, (2) estimate the
effect of a medical school's emphasis on primary care training, and
(3) estimate the effect of a medical school's research funding. 

In our analyses, we looked at the relationship of student and school
characteristics to the likelihood of choosing generalist residencies. 
We used a statistical technique, logistic regression, to assess these
relationships.  Our analyses included 8,939 students graduating in
1989 from allopathic medical schools in the United States; these data
were obtained from the Association of American Medical Colleges
(AAMC).  Characteristics assessed for their relationship to student
specialty choice are identified in tables I.1 and I.2. 


   RESULTS OF OUR ANALYSES
--------------------------------------------------------- Appendix I:1

Particular demographic characteristics were associated with the
likelihood that medical students would choose generalist (or primary
care) specialties when they graduated.  Our logistic models showed
that students who were female, married, Mexican-American, or from a
rural area were more likely to indicate an interest in pursuing
primary care specialties than students who were male, single, white,
or from a nonrural area.  In measuring the consistency of students'
specialty choices before they entered medical school and upon their
graduation, we found that a student's indication of early interest in
primary care was a strong predictor of the generalist residency
choice. 

Our analyses also showed a statistically significant association
between the medical school's commitment to primary care education and
the likelihood that a graduating student would choose a generalist
specialty.  In our analyses, students attending schools that were
public, had a highly funded family practice department,\45 or
required a family practice clerkship in the third year were more
likely to pursue a generalist specialty.  Our analyses did not show,
when controlling for other factors, a statistically significant
association between the amount of research funding a school received
and student specialty choice.  This finding does not support other
studies that suggest that the research intensity of a school
encourages students to develop an interest in specialty medicine and
ultimately to choose a subspecialty.\46


--------------------
\45 Highly funded family practice departments are those with the
highest ratio of total departmental revenues to number of student
enrollees.  In 1989, the highest ratio was $6,570 or more (the
highest third) compared with less than $3,157 (the lowest third). 

\46 In our analyses, the level of a school's research funding was a
proxy for its research intensity. 


      SOME DEMOGRAPHIC GROUPS MORE
      LIKELY TO SELECT PRIMARY
      CARE
------------------------------------------------------- Appendix I:1.1

Our analyses showed that several demographic variables were
statistically significant predictors of specialty preference. 
Mexican-American students, students from rural areas, female
students, and married students were more likely than their
counterparts to indicate a preference for primary care.  These
differences in likelihood ranged from 66 percent for Mexican-American
students (compared with white students) to 40 percent for married
students (compared with unmarried students).  While some studies have
shown age to be significantly related to specialty selection, our
analysis, which controlled for several demographic variables, showed
no significant difference between older (age 30 and over) and younger
(less than age 30) graduates in their likelihood of selecting primary
care. 


      STUDENTS WITH MOST AND LEAST
      EDUCATIONAL DEBTS LESS
      LIKELY TO SELECT PRIMARY
      CARE
------------------------------------------------------- Appendix I:1.2

Amount of medical education debt was also a statistically significant
predictor of student specialty choice.  Students with education debts
in the middle quartiles of more than $10,000 to $50,000 were about 23
to 26 percent more likely to select primary care than students with
debt exceeding $50,000.  It has been hypothesized that the higher the
level of debt, the more inclined a student would be to pursue a
specialty with a high earning potential. 


      PREMEDICAL SCHOOL PREFERENCE
      IS STRONG PREDICTOR IN
      MODELS
------------------------------------------------------- Appendix I:1.3

Some students graduating from medical school may have been
predisposed, even before they entered medical school, to a primary
care career.  In our models, students' intentions for medical
careers, as stated typically in their last year of undergraduate
college before attending medical school, were the strongest predictor
of student specialty choice.  Our models estimate that students who
indicated a primary care preference in college were about twice as
likely to indicate a preference for primary care in their final year
of medical school as students who indicated a preference for a
nonprimary care specialty or indicated no preference in their senior
year of undergraduate college. 


      VARIABLES REFLECTING MEDICAL
      SCHOOL ENVIRONMENT
------------------------------------------------------- Appendix I:1.4

We also used the models to assess the relationship between selected
medical school characteristics and students' likelihood of selecting
primary care.  The estimates from our models reflect the net effects
of the medical school characteristics, controlling for the influence
of students' preferences before entering medical school and their
demographic characteristics. 


      EXISTENCE OF FAMILY PRACTICE
      DEPARTMENT IS SIGNIFICANT
      PREDICTOR
------------------------------------------------------- Appendix I:1.5

Because we considered the existence of a family practice department
to be an indicator of a school's emphasis on or commitment to primary
care, we incorporated a variable in one model that compared choices
of students who attended schools with family practice departments
with choices of those students who attended schools without such a
department.  The results indicated that students who attended medical
schools with family practice departments were 57 percent more likely
to select primary care than students from schools without these
departments. 


      OTHER SCHOOL FACTORS SHOW
      ASSOCIATIONS
------------------------------------------------------- Appendix I:1.6

For students who attended schools with family practice departments,
we also examined whether the level of departmental funding, the
requirement of a third-year family practice clerkship, and type of
school ownership were related to the likelihood of choosing primary
care.  We found that funding and clerkship requirements were related
to the choice of a primary care career; the likelihood of selecting
primary care was slightly higher (by about 18 percent) for students
attending schools with the most highly funded departments than for
those students at schools with the least funded departments.\47 We
estimated that the requirement of a third-year clerkship also had
about an 18-percent effect.  School ownership, however, also had a
moderate effect; students attending schools that were publicly owned
were 38 percent more likely to select a primary care specialty than
their counterparts at nonpublic schools.  One explanation is that
since public medical schools depend more on public funds than
nonpublic medical schools do, public schools may be more pressured to
graduate more primary care physicians.\48


--------------------
\47 Highest level of funding was categorized at $6,570 or more per
student; whereas, the lowest level of funding was $3,157 or less per
student. 

\48 See D.  Campos-Outcalt and J.H.  Senf, "Characteristics of
Medical Schools Related to the Choice of Family Medicine as a
Specialty," Academic Medicine, 64 (1989), pp.  610-15. 


      RESEARCH INTENSITY NOT A
      SIGNIFICANT PREDICTOR
------------------------------------------------------- Appendix I:1.7

Some studies show an inverse relationship between the amount of
research funding a medical school receives and the proportion of
primary care graduates selecting primary care, these studies did not
control for the effects of other variables on specialty selection. 
Specifically, the studies showed bivariately that the more research
funding a medical school receives, the smaller the proportion of
primary care generalists it graduates.  Our bivariate data also
showed that schools with the highest level of federal research
funding per student (top quartile) had a slightly smaller proportion
of students who selected primary care (15 percent compared with 18 to
20 percent of students in the other quartiles).\49 Our models, which
control for the effects of other variables, did not show a
statistically significant association between research funding and
the likelihood of a student selecting primary care. 


--------------------
\49 In our analyses, federal research dollars was a proxy for a
school's research intensity or "milieu."


   METHODOLOGY
--------------------------------------------------------- Appendix I:2

To identify characteristics that could be associated with the
likelihood of students selecting a career in primary care medicine,
we constructed a database from information on students who graduated
in 1989 from 125 allopathic medical schools in the United States and
in Puerto Rico.\50,51 The AAMC Graduation Questionnaire was the
source for our outcome variable of interest, which captures the
students' specialty intentions in their senior year of medical school
before they enter the National Resident Matching Program.\52 Of the
15,573 students who graduated in that year, we excluded 4,445
students who did not respond to the AAMC questionnaire.  Of the
remaining 11,128 respondents, our models excluded an additional 2,189
individuals for whom we did not have complete data.  Thus, our final
models were based on 8,939 (57 percent) of the graduates.  Because
our models excluded a substantial number of 1989 graduates, we used
available information to make comparisons between the cases who were
included and those who were excluded (see data limitations section
below).  These analyses did not show any substantial differences
between the groups. 


--------------------
\50 The University of Minnesota at Duluth, School of Medicine, was
excluded from our analysis because its students actually graduate
from the University of Minnesota at Minneapolis.

\51 Because similar databases were unavailable for the osteopathic
medical schools, we were unable to include osteopathic students or
schools in our analyses. 

\52 A.M.  Singer, "The Class of 83:  A Follow-up Study of 1983
Medical School Graduates Through the First Six Postgraduate Years,"
Contract # 240-87-00067 (Washington, D.C.:  Health Resources and
Services Administration, 1990).  This report showed that student
specialty indications in the Graduation Questionnaire were a good
measure of the medical careers students ultimately entered. 


      DATA SOURCES
------------------------------------------------------- Appendix I:2.1

To develop our analytic database, we combined information on medical
school graduates with information about the institutions they
attended.  Data on individual graduates were obtained from the AAMC's
Student and Applicant Information Management System (SAIMS).\53 From
SAIMS, we obtained demographic and financial information on students
as well as information about their career intentions both before
entering medical school and in their senior year.  We compiled
institutional data from several sources.  We used 1987-88
AAMC-published directories to determine whether schools had family
practice departments and whether they required third-year family
practice clerkships.  Data on funding and ownership were obtained
from AAMC's Institutional Profile System. 


--------------------
\53 Data obtained from SAIMS came from the following surveys:  (1)
Premedical Student Questionnaire, (2) Matriculating Student
Questionnaire, and (3) Graduation Questionnaire. 


      MULTIVARIATE ANALYSIS
------------------------------------------------------- Appendix I:2.2

We used two multivariate logit models to quantify the statistical
impact of selected factors on the likelihood that a student intended
to seek certification in a primary care specialty.  These models were
used to produce estimates of the effect of each factor, while holding
constant the other factors that could influence the decision.  The
dependent variable of these models was the preferences of students
for residency programs, as stated in January of their senior year. 
The variable was coded as 1 if they indicated that they planned to
seek certification in family practice, general internal medicine, or
general pediatrics; otherwise the variable was coded as 0.  The
independent variables included in the models reflected both
individual characteristics of the graduates as well as
characteristics of the institutions they attended.  Table I.1 shows
the number of students and percentage selecting primary care for each
category of the individual student variables.  Table I.2 shows the
number of students and percentage selecting primary care for each
category of institutional variables.  These variables are described
below as they were defined in the models. 


         INDIVIDUAL STUDENT
         VARIABLES
----------------------------------------------------- Appendix I:2.2.1

  Initial preference--Students indicating a preference for family
     practice, general internal medicine, or general pediatrics
     before entering medical school were classified as having an
     initial preference for primary care; all other student
     preferences were classified as other/unknown (data were obtained
     from the AAMC Premedical Student Questionnaire).  This variable
     was used to control for student inclinations prior to entering
     the medical school environment. 

  Student's hometown size--Students who spent the major portion of
     their high school years in nonsuburban towns of less than 10,000
     persons were classified as being from rural areas. 

  Marital status--Students who indicated on the Graduation
     Questionnaire that they were married or separated were
     classified as married. 

  Age--We classified students into two groups according to age at
     graduation:  under 30 years old and 30 and older.\54

  Race--Students were classified into the following categories
     according to how they described themselves in the Graduation
     Questionnaire:  (1) black, not of Hispanic origin, (2) Asian or
     Pacific Islander, (3) Mexican-American, including other Hispanic
     or Chicano, (4) white, not of Hispanic origin, and (5) other. 

  Sex--Students' gender was obtained from the Graduation
     Questionnaire. 

  Debt--We used education indebtedness (premedical education plus
     medical school debts) as reported in the Graduation
     Questionnaire as an indicator of financial status of students at
     graduation.  We classified students into the following
     categories of debt:  (1) less than $10,000, (2) $10,000 to
     $29,999, (3) $30,000 to $49,999, and (4) $50,000 or more. 


--------------------
\54 See S.S.  Allen and others, "Effect of Early Exposure to Family
Medicine on Students' Attitudes Toward the Specialty," Journal of
Medical Education, 62 (Nov.  1987), pp.  911-17.  The results of this
study indicated that students aged 31 and older, at entry into
medical school, were more likely to choose family medicine. 


         INSTITUTIONAL VARIABLES
----------------------------------------------------- Appendix I:2.2.2

  Family practice department--We categorized schools as having a
     department (100 schools) or not (25 schools) based on the
     1987-88 AAMC Directory of American Medical Education.  Students
     were assigned to a category on the basis of the school they
     attended. 

  Research funding--We categorized the schools into the following
     groups on the basis of the ratio of total federal research
     support dollars to number of students enrolled in 1988-89:  (1)
     $13,560 or less (31 schools), (2) $13,561 to $25,414 (31
     schools), (3) $25,415 to $71,800 (31 schools), and (4) $71,801
     or more (32 schools).  Students were assigned to a category on
     the basis of the school they attended. 

  Family practice department funding--We categorized the schools with
     family practice departments into the following three groups on
     the basis of the ratio of total departmental revenues to number
     of students enrolled in 1989:  (1) less than $3,157 (32
     schools), (2) $3,157 to $6,569 (33 schools), and (3) $6,570 or
     more (33 schools).  Students were assigned to a category on the
     basis of the school they attended.\55

  Required third-year family practice clerkship--We categorized the
     schools with family practice departments according to whether
     their students were required to take a family practice clerkship
     in their third year (35 schools) or not (65 schools).  This
     classification was based on the 1987-88 AAMC Curriculum
     Directory.  Students were assigned to a category on the basis of
     the school they attended. 

  Ownership--We categorized the schools with family practice
     departments according to whether they were a public institution
     (71 schools) or a private institution (29 schools).  This
     classification was based on information about school ownership
     obtained from the AAMC Institutional Profile System.  Students
     were assigned to a category on the basis of the school they
     attended. 


--------------------
\55 Data on departmental revenues were not available for two schools. 


      MODELING APPROACH
------------------------------------------------------- Appendix I:2.3

We used a two-stage modeling approach to assess the effect of the
institutional variables systematically.  The first model, which
captured data on students from all institutions, was used to test
whether the existence of a department of family practice had an
effect, controlling for the other variables in the model.  The second
model, which captured data on students who graduated from schools
with a department of family practice, was used to examine the effects
of departmental funding, a third-year clerkship requirement, and
school ownership.\56 Both models included the individual student
variables as well as the institutional research funding variable. 


--------------------
\56 These variables were not included in Model 1 because they only
apply to schools with departments of family practice.  That is, there
was no departmental funding if a department did not exist, none of
the schools without a department had a third-year clerkship
requirement, and nearly all of the schools without a department were
private.  About one-fifth of the graduates were excluded from Model 2
because they attended schools without family practice departments. 


      PRESENTATION OF RESULTS AS
      ODDS RATIOS
------------------------------------------------------- Appendix I:2.4

The logistic regression results for both models are presented in
tables I.3 and I.4 as adjusted odds ratios.  The odds ratio is a
measure of association that compares the likelihood of an event
occurring (for example, selection of primary care) in one group with
a defined reference group.  The reported odds ratio indicates the
effect of a particular factor (for example, initial preference),
controlling for the effects of the other variables in the model.  The
estimate of the effect, reflected in the odds ratio, is a net effect
for a particular variable.  The greater the odds ratio differs from
1, in either direction, the larger the effect it represents. 


   DATA AND MODEL LIMITATIONS
--------------------------------------------------------- Appendix I:3


      COMPARISON OF RESPONDENTS TO
      NONRESPONDENTS
------------------------------------------------------- Appendix I:3.1

Our intent was to base the analyses on all 1989 graduates of U.S. 
allopathic medical schools.  For many graduates, however, data were
not available for the dependent variable in our models.  In
particular, 29 percent of the graduating class (4,445 individuals)
did not respond to the AAMC Graduation Questionnaire, the source of
information for our dependent variable.  When we compared the
nonrespondents with the respondents on selected independent variables
in our models, however, we found their distributions to be similar
(tables I.5 and I.6.) The similarities suggested that the
nonrespondents were not dramatically different from the
respondents.\57


--------------------
\57 In addition to excluding the 4,445 nonrespondents, an additional
37 respondents to the AAMC Graduation Questionnaire were excluded
because they did not respond to the question on specialty intentions. 


      EFFECT OF MISSING DATA FOR
      INDEPENDENT VARIABLES
------------------------------------------------------- Appendix I:3.2

Our multivariate analyses also excluded cases with missing data on
one or more independent variables.  Of the 11,091 graduates with
information on the dependent variable, we did not have complete
information on the independent variables for 19 percent of the cases. 
Missing data on one variable only, size of student's hometown,
accounted for a large proportion of these cases (14 of the 19
percent).  Thus, if we had not included size of student's hometown as
a variable in the models, we would have lost only 5 percent of the
cases. 

Because we were concerned about the potential impact of excluding 19
percent of the cases with information on our dependent variable, we
compared our results with those we would have obtained without the
hometown variable in the models.  The effects of the independent
variables were similar for both models and suggested that our results
were not dramatically biased because of the exclusion of individuals
without information on hometown size. 


      OTHER VARIABLES
------------------------------------------------------- Appendix I:3.3

Another limitation of our models was the lack of information on
certain variables that may be significant factors affecting career
intentions of medical school graduates.  For example, we were not
able to directly control for such factors as the importance that
students place on the prestige, intellectual stimulation, and
earnings potential of various specialty choices.  However, our
initial preference variable--specifying student intentions before
attending medical school--may indirectly reflect some of these
factors. 



                         Table I.1
          
            Individual Variables Included in the
              Models by Number of Students and
             Percentage Selecting Primary Care


                                                Percentage
                                  Percentage     selecting
Variable                 Number     in group  primary care
------------------  -----------  -----------  ------------
Initial preference
Primary care              2,852           26          27.1
Other/unknown             8,239           74          14.4
Marital status
Married                   3,908           36          22.0
Not married               7,056           64          15.4
Education debts
Less than $10,000         2,932           28          15.8
$10,000-29,999            2,153           20          19.3
$30,00-49,999             2,508           24          20.6
$50,000 or more           3,036           29          16.2
Age
30 or older               2,167           20          19.7
Under 30                  8,917           80          17.2
Sex
Female                    3,737           34          21.9
Male                      7,354           66          15.5
Race/ethnicity
White                     9,086           82          18.0
Black                       526            5          16.3
Asian                       856            8          13.3
Mexican-American            152            1          26.3
Other                       459            4          18.1
Hometown
Rural                     1,652           18          26.6
Nonrural                  7,675           82          16.4
----------------------------------------------------------


                          Table I.2
           
           Institutional Variables Included in the
               Models by Number of Students and
              Percentage Selecting Primary Care

              (Total students in group: 11,091)


                                                      Percen
                                                        tage
                              Number          Percen  select
                                  of            tage     ing
                              school              in  primar
Variable                           s  Number   group  y care
----------------------------  ------  ------  ------  ------
Family practice department
------------------------------------------------------------
Yes                              100   8,753      79    19.2
No                                25   2,338      21    11.8

Research dollars per student
------------------------------------------------------------
Quartile 1 (less than             31   2,161      20    18.5
 $13,560)
Quartile 2 ($13,561-              31   3,171      29    20.0
 $25,414)
Quartile 3 ($25,415-              31   2,816      25    17.8
 $71,801)
Quartile 4 (above $71,801)        32   2,943      27    14.5

Departmental funding\a
------------------------------------------------------------
Highest third (above $6,569)      33   2,521      29    21.8
Lowest third (less than           32   3,244      37    17.0
 $3,157)
Middle third ($3,157-             33   2,886      33    19.2
 $6,569)

Required third-year clerkship\a
------------------------------------------------------------
Yes                               35   2,559      29    21.4
No                                65   6,194      71    18.4
Ownership\a
Public                            71   6,207      71    20.8
Nonpublic                         29   2,546      29    15.3
------------------------------------------------------------
\a Includes only students who attended schools with departments of
family practice. 



                          Table I.3
           
           Model Results--Adjusted Odds Ratios and
              Confidence Interval for Individual
                          Variables


                                         95%             95%
                              Adjust  confid  Adjust  confid
                                  ed    ence      ed    ence
                                odds  interv    odds  interv
Variable                       ratio      al   ratio      al
----------------------------  ------  ------  ------  ------
Initial preference
------------------------------------------------------------
Primary care                  2.04\b   1.83-  1.95\b   1.72-
                                        2.29            2.20
Other/unknown                 1.00\c          1.00\c

Marital status
------------------------------------------------------------
Married                       1.40\b   1.25-  1.36\b   1.20-
                                        1.57            1.54
Not married                   1.00\c          1.00\c

Education debts
------------------------------------------------------------
Less than $10,000               1.02   0.88-    0.98   0.82-
                                        1.20            1.17
$10,000-29,999                1.23\b   1.05-    1.11   0.93-
                                        1.44            1.33
$30,000-49,999                1.26\b   1.08-    1.17   0.98-
                                        1.47            1.39
$50,000 or more               1.00\c          1.00\c

Age
------------------------------------------------------------
30 or older                     1.03   0.90-    0.98   0.84-
                                        1.19            1.14
Under 30                      1.00\c          1.00\c

Sex
------------------------------------------------------------
Female                        1.54\b   1.37-  1.48\b   1.31-
                                        1.72            1.68
Male                          1.00\c          1.00\c

Race
------------------------------------------------------------
Black                           0.89   0.68-    0.81   0.60-
                                        1.17            1.10
Asian                         0.75\b   0.59-    0.87   0.67-
                                        0.95            1.13
Mexican-American              1.66\b   1.12-  1.58\b   1.03-
                                        2.46            2.43
Other                           1.06   0.80-    1.18   0.87-
                                        1.40            1.60
White                         1.00\c          1.00\c

Hometown
------------------------------------------------------------
Rural                         1.60\b   1.41-  1.63\b   1.42-
                                        1.83            1.88
Nonrural                      1.00\c          1.00\c
------------------------------------------------------------
\a Based only on students who attended schools with departments of
family practice. 

\b Significantly different from 1.00 (reference category) at the
95-percent confidence level. 

\c Reference category. 



                          Table I.4
           
           Model Results--Adjusted Odds Ratios and
            Confidence Interval for Institutional
                          Variables


                                         95%             95%
                              Adjust  confid  Adjust  confid
                                  ed    ence      ed    ence
                                odds  interv    odds  interv
Variable                       ratio      al   ratio      al
----------------------------  ------  ------  ------  ------
Family practice department
------------------------------------------------------------
Yes                           1.57\b   1.32-
                                        1.86
No                            1.00\c

Research dollars per student
------------------------------------------------------------
Quartile 1 (less than           1.09   0.91-    1.08   0.88-
 $13,560)                               1.31            1.34
Quartile 2 ($13,561-            1.15   0.98-    1.12   0.91-
 $25,414)                               1.34            1.38
Quartile 3 ($25,415-            1.01   0.85-    1.02   0.83-
 $71,801)                               1.19            1.24
Quartile 4 (above $71,801)    1.00\c          1.00\c

Departmental funding\a
------------------------------------------------------------
Highest third (above $6,569)                  1.18\b   1.00-
                                                        1.40
Middle third ($3,158-                           1.09   0.93-
 $6,569)                                                1.28
Lowest third (less than                       1.00\c
 $3,157)

Required third-year clerkship\a
------------------------------------------------------------
Yes                                           1.18\b   1.02-
                                                        1.35
No                                            1.00\c

Ownership\a
------------------------------------------------------------
Public                                        1.38\b   1.18-
                                                        1.61
Nonpublic                                     1.00\c
------------------------------------------------------------
\a Based only on students who attended schools with departments of
family practice. 

\b Significantly different from 1.00 (reference category) at the
95-percent confidence level. 

\c Reference category. 



                         Table I.5
          
           Comparison of Graduation Questionnaire
             Respondents to Nonrespondents for
               Selected Individual Variables

                        (Percentage)

                           Respondents      Nonrespondents
Variable                      (11,128)             (4,445)
------------------  ------------------  ------------------
Sex
----------------------------------------------------------
Female                              34                  33
Male                                66                  67

Race/ethnicity
----------------------------------------------------------
White                               82                  78
Black                                5                   6
Asian                                8                   8
Mexican-American                     1                   2
Other                                4                   5
----------------------------------------------------------


                         Table I.6
          
           Comparison of Graduation Questionnaire
             Respondents to Nonrespondents for
              Selected Institutional Variables

                           Respondents      Nonrespondents
Variable                      (11,128)             (4,445)
------------------  ------------------  ------------------
Department/clerkship
----------------------------------------------------------
No department or                    21                  22
 clerkship
Department but no                   56                  50
 clerkship
Department and                      23                  28
 clerkship

Departmental funding\a
----------------------------------------------------------
Highest third                       29                  24
 (above $6,570)
Middle third                        33                  39
 ($3,157-$6,569)
Lowest third (less                  37                  36
 than $3,157)

Research funding
----------------------------------------------------------
Quartile 1 (less                    20                  21
 than $13,560)
Quartile 2                          29                  28
 ($13,561-
 $25,414)
Quartile 3                          25                  29
 ($25,415-
 $71,801)
Quartile 4 (above                   27                  23
 $71,801)

Public institution
----------------------------------------------------------
Public                              58                  62
Nonpublic                           42                  38
----------------------------------------------------------
\a Based on students who attended schools with departments of family
practice.  About 1 percent of both respondents and nonrespondents
attended schools for which we lack information on funding. 


PUBLIC HEALTH SERVICE FUNDING OF
MEDICAL EDUCATION
========================================================== Appendix II

In addition to funding medical education and training through the
Medicare program, the federal government also provides funds through
programs authorized under the Public Health Service Act.\58 Under
title VII of the act, the Department of Health and Human Services
provides two types of assistance for medical education and
training:\59 (1) institutional support to medical schools through
grants and contracts for special training programs and (2) student
assistance through loans, loan guarantees, and scholarships. 

Several programs authorized under title VII focus on promoting
primary care education and training.  By funding family practice,
general internal medicine, and general pediatrics residency programs
and family practice departments in medical schools, title VII has
provided modest but crucial support for primary care training.  (See
table II.1.)



                                        Table II.1
                         
                          Title VII Programs Supporting Medical
                                Education and Primary Care
                                      Specialization

                                    (Fiscal year 1993)

                                                                              Outlays (in
Program                 Description                                            thousands)
----------------------  ---------------------------------------------------  ------------
Health Professions      Supports analytical and descriptive studies of the           $640
 Analytical Program      health professions, including evaluations and
                         projections of the supply of health professionals
                         by specialty and geographic location
Health Professions      Awards grants and contracts to conduct research on          1,112
 Educational Research    health professions issues, including the extent to
                         which educational debt affects medical student
                         specialty choice, and factors affecting selection
                         of careers in primary care
Establishment of        Awards grants to allopathic or osteopathic medical         11,494
 Departments of Family   schools to establish, maintain, or improve family
 Medicine                medicine programs, including pilot testing of
                         model curricula\a
Graduate Training in    Supports residency programs for training physicians        15,711
 Family Medicine         who will enter family medicine, including support
                         for such trainees\a
Predoctoral Training    Supports programs, trainees, curriculum                     1,797
 in Family Medicine      development, clerkships, and preceptorships\a
Faculty Development in  Supports family medicine programs and trainees in           6,403
 Family Medicine         such programs who plan to teach in a family
                         medicine training program\a
Graduate Training in    Supports new residency positions or conversion of          11,587
 General Internal        "traditional" programs to those that emphasize
 Medicine and General    longitudinal, preventive, and comprehensive care
 Pediatrics              (unlike programs in internal medicine and
                         pediatrics from which many physicians enter
                         subspecialty training, supported programs
                         emphasize continuity, ambulatory, and preventive
                         medicine)\a
Faculty Development in  Helps meet the cost of programs for training                4,892
 General Internal        physicians who plan to teach in general internal
 Medicine and General    medicine and general pediatrics, and for trainees
 Pediatrics              in such programs\a
Predoctoral Training    Supports programs, trainees, curriculum                       580
 for General Internal    development, clerkships, and preceptorships\a
 Medicine and General
 Pediatrics
Primary Care Loan       Provides low-interest, need-based loans for
 Program (formerly       students who specialize in primary care and
 Health Professions      practice primary care throughout the life of the
 Student Loans, as       loan (borrower's failure to honor agreement
 applied to schools of   results in interest rate of 12% instead of 5%, and
 allopathic and          3-year repayment deadline)\b,\c
 osteopathic medicine)
Exceptional Financial   Funds awards of need-based, tuition scholarships to        10,331
 Need Scholarship        medical students who must complete residency
                         training in primary care and practice in primary
                         care for 5 years
Financial Assistance    Same as program above for students who must come            6,181
 for Disadvantaged       from disadvantaged backgrounds
 Health Professions
 Students (FADHPS)
-----------------------------------------------------------------------------------------
\a Preference will be given to institutional applicants that
demonstrate a commitment to train primary care clinicians and
underrepresented minority students, and that have a high rate of
students who go into practice in medically underserved areas. 

\b Participating schools must meet specified standards for output of
primary care physicians or return a portion of the loan funds made
available to the school.  Schools that fail to meet certain
conditions will be required to repay a percentage of Primary Care
Loan Program funds received during the 1-year period in which the
school did not comply. 

\c The Primary Care Loan and Health Professions Student Loans
programs are supported by revolving funds.  In fiscal year 1993, $1.9
million dollars were available for redistribution. 

Source:  U.S.  Department of Health and Human Services, Public Health
Service, Health Resources and Services Administration. 

Although not focused on primary care, several other programs
authorized under title VII complement efforts to promote primary care
education and training.  These programs include efforts to increase
the numbers of health care providers from minority or disadvantaged
backgrounds and to promote educational strategies to recruit and
retain health care providers for underserved populations.  Special
loan and scholarship programs for disadvantaged and minority students
reflect the perception that a disproportionate number of such
students enter primary care and practice in underserved areas. 

Title III of the Public Health Service Act also provides support for
improving access to care in "Health Professional Shortage Areas"
through the National Health Service Corps.\60 These shortage areas
can be designated based in part on a lack of primary care physicians. 
The Corps funds salary and benefit costs of program physicians, a
variety of clinical and professional support activities, and
scholarship and loan repayment programs.  In fiscal year 1993, the
federal government provided a total of about $116 million (about $43
million for field operations and about $73 million for recruiting and
associated activities) to support the Corps and its programs. 


--------------------
\58 The federal government contributes to the financing of graduate
medical education also through programs administered by the
Department of Veterans Affairs, the Department of Defense, and
through federal sharing of states' costs of the Medicaid program.  By
funding biomedical research at the undergraduate and graduate medical
education levels, the National Institutes of Health indirectly
contributes to the financing of medical education and training. 

\59 In addition to medicine, title VII provides federal support for
health professions education in osteopathy, dentistry, veterinary
medicine, optometry, podiatry, pharmacy, public health, and graduate
programs in health administration. 

\60 In 1981, authority for the National Health Service Corps
Scholarship Program was transferred to title III by Public Law 97-35. 


SURVEY METHODOLOGY FOR MEDICAL
SCHOOL AND RESIDENCY SURVEYS
========================================================= Appendix III

Using survey methodology, we sought to measure the extent to which
medical schools and residency programs maintain requirements and
provide opportunities for students and residents to gain experience
in primary care medicine. 


   SURVEY OF ALLOPATHIC AND
   OSTEOPATHIC MEDICAL SCHOOLS
------------------------------------------------------- Appendix III:1

To determine how much primary care experience medical schools provide
students, we mailed questionnaires to all 126 allopathic and 15
osteopathic medical schools in the United States.  Specifically, we
sought to determine the extent to which schools required observation
or training in primary care medicine as part of their curricula. 

We developed two self-administered questionnaires (for allopathic
schools and for osteopathic schools) based on a review of relevant
research and interviews with medical school officials.  We pretested
the questionnaires with participants from three medical schools and
submitted copies to AAMC for review.  Based on the pretest results
and discussions with the reviewers, we modified and finalized the
questionnaires and mailed them to the dean of each allopathic and
osteopathic medical school.  To obtain a higher response rate, we
mailed a second questionnaire to nonrespondents.  About 89 percent
(112) of allopathic medical schools responded and 100 percent of
osteopathic medical schools responded. 


   SURVEY OF RESIDENCY PROGRAMS
------------------------------------------------------- Appendix III:2

To determine the extent to which residents are exposed to primary
care medicine, we mailed questionnaires to 534 directors of
allopathic and osteopathic residency programs.  The survey sought to
determine residents' contact with primary care medicine by reviewing
aspects of three components of residency training:  required
rotations, elective rotations, and continuity of care assignments. 

We developed items for the survey with the input of directors and
department chairs from several residency programs.  Based on these
discussions, we developed eight self-administered questionnaires. 
While the questionnaires were similar in content, the response
choices were tailored to fit five allopathic and three osteopathic
residency programs.  The allopathic programs included\61

  internal medicine, traditional track,

  internal medicine, primary care track,

  pediatrics, traditional track,

  pediatrics, primary care track, and

  family practice. 

The osteopathic programs included

  internal medicine,

  pediatrics, and

  general practice. 

We pretested the questionnaires with representatives from nine
residency programs (eight allopathic and one osteopathic).  Pretest
participants included hospital department chairs, residency program
directors, and faculty members.  We also submitted the questionnaires
to several experts for review.  On the basis of the pretest results
and expert discussion, we modified and finalized the questionnaires. 

We identified the universe of civilian residency programs for each of
the five allopathic and three osteopathic program categories through
listings in The 1991-1992 Directory of Graduate Medical Education
Programs and The 1991-1992 Directory of Osteopathic Postdoctoral
Education Programs.  The residency programs selected to participate
in the survey were identified through simple random sampling of the
following residency programs:  allopathic internal medicine,
traditional track; internal medicine, primary care track; pediatrics,
traditional track; family practice; and osteopathic general
practice.\62 The entire population of osteopathic internal medicine
and pediatrics programs and allopathic pediatrics primary care track
programs was included in the study because of their relatively small
numbers nationally. 

About 82 percent (482) of the 534 program directors surveyed
responded to the questionnaires.  Response rates for each are listed
in table III.1. 



                         Table III.1
           
               Percentage of Residency Programs
            Responding to Survey by Discipline and
                            Track

                                                      Percen
                                       Total            tage
                                      number              of
                                          of  Number  sample
Residency program discipline and      progra  survey  respon
track                                     ms      ed     ded
------------------------------------  ------  ------  ------
Allopathic
------------------------------------------------------------
Internal medicine, traditional track     247     151      75
Internal medicine, primary care          152     108      82
 track
Pediatrics, traditional track            163     113      85
Pediatrics, primary care track            41      41      83
Family practice                          367      76      84

Osteopathic
------------------------------------------------------------
Internal medicine                         43      43      86
General practice                          97      48      83
------------------------------------------------------------

--------------------
\61 A program was considered to be a "primary care track" program if
it either advertised in the National Resident Matching Program as a
primary care program or received funds from the federal Health
Resources and Services Administration to support a primary care
curricular focus within its residency program.  We included as
"traditional track" those programs that were listed in the National
Resident Matching Program as categorical programs. 

\62 Because we surveyed a statistical sample of these residency
programs, our estimates have a measurable precision or sampling
error.  In this analysis, the sampling errors are stated at a
95-percent confidence level. 


GAO CONTACTS AND STAFF
ACKNOWLEDGMENTS
========================================================== Appendix IV

GAO CONTACTS

Rose Marie Martinez, Assistant Director, (202) 512-7103
Andrew Bhak, Senior Health Policy Analyst, (202) 512-7134

ACKNOWLEDGMENTS

Prior to leaving GAO Carolyn Cocotas and Nancy Kim served as Project
Manager and Senior Health Policy Analyst.  Other staff members who
participated in field work activities include Patricia Padilla,
Sheila Nicholson, and Paul Wright.  Design and data analysis support
was provided by Robert DeRoy, Steve Machlin, Ed Murphy, Linda
Stinson, and Ed Tuchman.  Hannah Fein contributed to the writing of
the report, and Peter Amory, Lester Baskin, and Jessica Weisz
provided varied assistance during their summer internships. 


BIBLIOGRAPHY
============================================================ Chapter 0

Allen, S.S., and others.  "Effects of Early Exposure to Family
Medicine on Students' Attitudes Toward the Specialty." Journal of
Medical Education, Vol.  62, (1987), pp.  911-17. 

Association of American Medical Colleges.  Financing Graduate Medical
Education:  Final Report of The AAMC Committee on Financing Graduate
Medical Education.  Washington, D.C.:  1986. 

Babbott, D., and others.  "The Stability of Early Specialty
Preferences Among U.S.  Medical School Graduates in 1983." Journal of
the American Medical Association, Vol.  259 (13), (1988), pp. 
1970-975. 

Babbott, D., and others.  "Racial-Ethnic Background and Specialty
Choice:  A Study of U.S.  and Medical School Graduates in 1987."
Academic Medicine, Vol.  64 (10), (1989), pp.  595-99. 

Babbott, D., and others.  "Medical Student Attitudes About Internal
Medicine:  A Study of U.S.  Medical School Seniors in 1988." Annals
of Internal Medicine, Vol.  114, (1991), pp.  16-22. 

Campos-Outcalt, D., and J.M.  Senf.  "Characteristics of Medical
Schools Related to the Choice of Family Medicine as a Specialty."
Academic Medicine, Vol.  64, (1989), pp.  610-15. 

Colwill, J.M.  "Where Have All The Primary Care Applicants Gone?" New
England Journal of Medicine, Vol.  326 (6), (1992), pp.  387-93. 

Council on Graduate Medical Education.  Fourth Report: 
Recommendations to Improve Access to Health Care Through Physician
Workforce Reform.  U.S.  Department of Health and Human Services,
Jan.  1994. 

Duerson, M.C., and others.  "Impact of Required Family Medicine
Clerkship on Medical Students' Attitudes about Primary Care."
Academic Medicine, Vol.  64, (1989), pp.  546-48. 

Eisenberg, J.M.  "How Can We Pay for Graduate Medical Education in
Ambulatory Care?" New England Journal of Medicine, Vol.  320 (23),
1989, pp.  1525-531. 

Feltovich, J., and others.  "Teaching Medical Students in Ambulatory
Settings in Departments of Internal Medicine." Academic Medicine,
Vol.  64, (1989), pp.  36-41. 

Ganem, J.L., and others.  "Review of U.S.  Medical School Finances,
1992-1993." Journal of the American Medical Association, Vol.  272
(9), (1994), pp.  705-11. 

Ginzberg, E., and others.  The Economics of Medical Education.  New
York:  Josiah Macy Foundation, 1993. 

Godkin, M., and M.  Quirk.  "Why Students Choose Family Medicine: 
State Schools Graduating the Most Family Physicians." Family
Medicine, Vol.  23, (1991), pp.  521-26. 

Goldsmith, G.  "Factors Influencing Family Practice Residency
Selection:  A National Survey." The Journal of Family Practice, Vol. 
15 (1), (1982), pp.  121-24. 

Goodson, J.D., and others.  "The Training of Physicians Outside the
Hospital." Archives of Internal Medicine, Vol.  146, (1986), pp. 
1805-809. 

Greer, D., and others.  "Comments on the AAMC Policy Statement
Recommending Strategies for Increasing the Production of Generalist
Physicians." Academic Medicine, Vol.  69 (4), (1994), pp.  245-60. 

Huntington, C., and others.  "The Shortage of Generalist Physicians."
American Family Physician, Vol.  45 (4), (1992), pp.  1573-576. 

Institute of Medicine.  Primary Care Physicians:  Financing Their GME
in Ambulatory Settings.  Washington, D.C.:  National Academy Press,
1989. 

Kassebaum, D.G., and R.A.  Haynes.  "Relationship Between Third-year
Clerkships in Family Medicine and Graduating Students' Choices of
Family Practice Careers." Academic Medicine, Vol.  67 (3), (1992),
pp.  217-19. 

Kassebaum, D.G., and P.L.  Szenas.  "Rural Sources of Medical
Students, and Graduates' Choice of Rural Practice." Academic
Medicine, Vol.  68 (3), (1993), pp.  232-36. 

Kassebaum, D.G., and P.L.  Szenas.  "The Declining Interest Of
Medical School Graduates in Generalist Specialties:  Students'
Abandonment of Earlier Inclinations." Academic Medicine, Vol.  68
(4), (1993), pp.  278-80. 

Kassebaum, D.G., and P.L.  Szenas.  "Relationship Between
Indebtedness and the Specialty Choices of Graduating Medical
Students:  1993 Update." Academic Medicine, Vol.  68, (1993), pp. 
934-37. 

Kassebaum, D.G., and P.L.  Szenas.  "Factors Influencing the
Specialty Choices of 1993 Medical School Graduates." Academic
Medicine, Vol.  69, (1994), pp.  164-70. 

Kassler, W.J., and others.  "Why Medical Students Choose Primary Care
Careers." Academic Medicine, Vol.  66 (1), (1991), pp.  41-43. 

Lieu, T.A., and others.  "Specialty Choices at One Medical School: 
Recent Trends and Analysis of Predictive Factors." Academic Medicine,
Vol.  64, (1989), pp.  622-29. 

Markert, R.J.  "Why Medical Students Change to and from Primary Care
as a Career Choice." Family Medicine, Vol.  23 (5), (1991), pp. 
347-50. 

Martini, C.J.M., and others.  "Medical School and Student
Characteristics That Influence Choosing a Generalist Career." Journal
of the American Medical Association, Vol.  272 (9), (1994), pp. 
661-68. 

McKay, N.L.  "The Economic Determinants of Specialty Choice by
Medical Residents." Journal of Health Economics, Vol.  9, (1990), pp. 
335-57. 

McLaughlin, M.A., and others.  "The Impact of Medical School Debt on
Postgraduate Career and Lifestyle." Academic Medicine, Vol.  66 (9
Supplement), (1991), pp.  S43-S45. 

Nobel, J., and others.  "Career Differences Between Primary Care and
Traditional Trainees in Internal Medicine and Pediatrics." Annals of
Internal Medicine, Vol.  116 (6), (1992), pp.  482-87. 

Politzer, R.M., and others.  "Primary Care Physician Supply and the
Medically Underserved:  A Status Report and Recommendations." Journal
of the American Medical Association, Vol.  266 (1), (1991), pp. 
104-09. 

Rabinowitz, H.  "The Relationship Between Medical Student Career
Choice and A Required Third-year Family Practice Clerkship." Family
Medicine, Vol.  20, (1988), pp.  118-21. 

Rivo, M.L., and others.  "Defining the Generalist Physician's
Training." Journal of the American Medical Association, Vol.  271
(19), (1994), pp.  1499-504. 

Rivo, M.L., and D.  Satcher.  "Improving Access to Health Care
Through Physician Workforce Reform." Journal of the American Medical
Association, Vol.  270, (1993), pp.  1074-078. 

Rosenblatt, R.A., and others.  "Which Medical Schools Produce Rural
Physicians?" Journal of the American Medical Association, Vol.  268
(12), (1992), pp.  1559-565. 

Rosenthal, M.P., and others.  "Influence of Income, Hours Worked, and
Loan Repayment on Medical Students' Decision to Pursue a Primary Care
Career." Journal of the American Medical Association, Vol.  271 (12),
(1994), pp.  914-17. 

Schroeder, S.A.  "The Making of a Medical Generalist." Health
Affairs, Vol.  4 (2), (Summer 1985), pp.  22-46. 

Schwartz, M.D., and others.  "Medical Student Interest in Internal
Medicine.  Initial Report of the Society of General Internal Medicine
(SGIM) Interest Group Survey on Factors Influencing Career Choice in
Internal Medicine." Annals of Internal Medicine, Vol.  115, (1991),
pp.  6-15. 

Singer, A.M.  The Class of 83:  A Follow-up Study of 1983 Medical
School Graduates Through the First Six Postgraduate Years. 
Washington, D.C.:  Health Resources and Services Administration,
1990. 

Starfield, B.  Primary Care:  Concept, Evaluation, and Policy.  New
York:  Oxford University Press, 1992. 

Stimmel, B.  "The Crisis in Primary Care and the Role of Medical
Schools." Journal of the American Medical Association, Vol.  268
(15), (1992), pp.  2060-065. 

U.S.  General Accounting Office.  Medicare:  Graduate Medical
Education Payment Policy Needs to Be Reexamined.  GAO/HEHS-94-33, May
5, 1994. 

Weiner, J.  "Forecasting the Effects of Health Reform on U.S. 
Physician Workforce Requirement:  Evidence from HMO Staffing
Patterns." Journal of the American Medical Association, Vol.  272
(3), (1994), pp.  222-30. 

Whitcomb, M.E., and others.  "Comparing the Characteristics of
Schools that Produce High Percentages and Low Percentages of Primary
Care Physicians." Academic Medicine, Vol.  67, (1992), pp.  587-91. 

