Robert C. Bowman, M.D.
The new education must be less concerned with sophistication than compassion…it must teach man the most difficult lesson of all - to look at someone anywhere in the world and be able to see the image of himself. Norman Cousins, Anatomy of an Illness via John West MD Nebraska Academy of Family Physicians president
Family medicine, as measured by the choices of US medical students, has had a consistent 30 year decline. Only a brief period of popularity stimulated by managed care efforts interrupted this decline from 1994 – 1998, with continued decline since. The consistent nature of this decline does not suggest a recent problem. Nor does it indicate the swing of a pendulum. Without significant numbers of physicians imported from international medical schools, the situation would be even worse. The managed care panic years and the contributions of foreign medical schools have hidden the magnitude of the changes in US medical schools.
Methods
Hypothesis:
Medical schools admitting more students born in rural areas (or fewer from urban origin), schools admitting more instate students, medical schools that have a rural mission or person, schools with a family medicine department, schools with a lower MCAT average scores, those located further west, and medical schools located in states with better education opportunity are likely to graduate more family and general practice physicians.
Dependent Variable
The dependent variable was the percentage of family medicine and general practice physicians graduating from each medical school that had a final year of residency training between 1997 and 2003. This included 21000 graduates who completed a family medicine residency program as noted in the AAFP database and 4000 additional physicians from the AMA Masterfile who designated their specialty as FP and had at least one year of primary care graduate training.
Independent Variables
Medical school type: public, private, osteopathic, and allopathic.
Establishment of a family medicine department by 1992 20 (Kahn)
Existence of a rural mission 21 (Rosenblatt) or rural medical education person at the school (surveys by the STFM Group on Rural Health survey).
Percentage of students born in rural areas using the birth city and state from the AMA Masterfile. Over 98% of US cities were coded rural or urban (Hart reference).
Percentage of instate medical students for each school
Average MCAT for each school in 2000
Grade Point Average for students at the school
Average of all Achievement Test scores for the state’s 8th graders
Product of high school graduation rate and college continuation rate for the state. 24 (from Education Weekly).
State data was compiled from the US Census regarding population. 23 The percentage of non-metro population was used to weight the regression as several of the variables were sensitive to population distribution.
Footnote: Commercial sources and internet sources that compile MCAT scores for schools had to be utilized as AAMC would not release individual MCAT scores for schools.22
These scores were verified by accessing medical school admissions web sites ( of 130 accessed). Use of 5 different sets of MCAT scores from 2000 - 2003 revealed no differences in the regressions.
Some medical schools were excluded due to significant variation in mission, ethnicity, and comparability of state education data. These atypical schools included the 2 year, branch, or regional campuses (University of Minnesota Duluth, Charles Drew); the 6 year medical school (U of Missouri Kansas City); those with large variation in ethnicity and mission (Howard, Meharry, Charles Drew, Hawaii, Puerto Rico schools); and those with too few graduates (Arizona College of Osteopathic Medicine, Florida State, Touro, Edward Via, and Pikeville).
The data was analyzed with SPSS v 11.5 using linear regression weighted for the percentage of the population of the state in non-metropolitan areas. 25
Results
|
|
Mean |
Std. Deviation |
|
TRUFPGPR% of FP and GP graduates from a school |
18.80 |
36.22 |
|
MCATALL |
9.30 |
3.75 |
|
LONGITUDE - negative with western direction |
-87.89 |
52.51 |
|
FPDEPT92- presence of FP Department by 1992 |
0.87 |
1.65 |
|
OVER30 - % admitted who are over 30 yrs at graduation |
22.84 |
38.64 |
|
HSXCOLC High school grad rate x college continuation rate |
39.28 |
32.42 |
|
RBRN - % admitted who were born in rural locations |
18.53 |
54.17 |
|
INSTAVGY % admitted who were born in state |
72.05 |
132.93 |
|
RMSNPER presence of rural mission or rural person |
0.12 |
1.62 |
Pearson Correlations
|
|
TRU |
MCAT |
LONG |
FPDE |
OVR30 |
HSXCO |
RBRN |
INST |
RMS |
|
FPGP |
1.00 |
-0.61 |
-0.41 |
0.58 |
0.52 |
0.33 |
0.69 |
0.53 |
0.46 |
|
MCAT |
-0.61 |
1.00 |
-0.03 |
-0.48 |
-0.41 |
-0.05 |
-0.58 |
-0.49 |
-0.24 |
|
LONG |
-0.41 |
-0.03 |
1.00 |
-0.12 |
-0.23 |
0.05 |
-0.25 |
-0.27 |
-0.17 |
|
FPDE |
0.58 |
-0.48 |
-0.12 |
1.00 |
0.39 |
0.04 |
0.39 |
0.54 |
0.15 |
|
OVR30 |
0.52 |
-0.41 |
-0.23 |
0.39 |
1.00 |
-0.01 |
0.23 |
0.19 |
0.22 |
|
HSXC |
0.33 |
-0.05 |
0.05 |
0.04 |
-0.01 |
1.00 |
0.32 |
-0.22 |
0.16 |
|
RBRN |
0.69 |
-0.58 |
-0.25 |
0.39 |
0.23 |
0.32 |
1.00 |
0.43 |
0.39 |
|
INSTA |
0.53 |
-0.49 |
-0.27 |
0.54 |
0.19 |
-0.22 |
0.43 |
1.00 |
0.20 |
The weighted regression explained 80 % of the variance.
|
|
B |
Std. Error |
Beta |
t |
Sig. |
|
(Constant) |
-3.10 |
6.766 |
|
-0.46 |
0.6482 |
|
MCATALL |
-2.11 |
0.620 |
-0.218 |
-3.40 |
0.0010 |
|
LONGIT |
-0.17 |
0.034 |
-0.250 |
-5.02 |
0.0000 |
|
FPDEPT92 |
3.78 |
1.237 |
0.172 |
3.05 |
0.0029 |
|
OVER30 |
0.19 |
0.049 |
0.203 |
3.91 |
0.0002 |
|
HSXCOLC |
0.32 |
0.056 |
0.288 |
5.73 |
0.0000 |
|
ALLRBRPR |
0.11 |
0.043 |
0.160 |
2.50 |
0.0138 |
|
INSTAVGY |
0.05 |
0.017 |
0.186 |
3.01 |
0.0033 |
|
RMSNPER |
3.24 |
1.056 |
0.145 |
3.07 |
0.0028 |
Interpretation
Each percentage point increase in fp choice by medical students represents an additional 1.31 students choosing family practice in a typical graduating class.
· Medical school locations 300 miles further west have a 1 percentage point increase in students choosing family medicine (Represents 1.31 more FP docs per graduating class).
· Each 3 percentage point increase in higher education opportunity results in a 1 point increase in choice of family medicine (schools in states with 3 points higher high school grad rate times college continuation graduated an additional FP doctor each year). Education is the Key
· Medical schools with an average MCAT 1 unit higher graduate 2.1 percentage points fewer family physicians or about 3 - 5 fewer FP doctors per graduating class. MCAT Correlations
· Medical schools with a rural mission or person have a 3 point higher choice of FP (4 more fp docs).
· Medical schools with a family medicine department have 3.8 point higher choice of FP (8 more fp docs).
· Each 9 point increase in rural birth percentage results in a 1 point increase in FP choice (8 more rural origin students results in 1 additional FP doc). When this rural group is replaced by a highly urban group with a lower probability of FP choice (the reality of the past 20 years), the impact is likely greater. This is most recent grads after decline of rural born.
· Each 5 point increase in older medical students admitted increases FP by 1 point.
Admissions characteristics contribute directly and indirectly to most variables. The infrastructure/mission variables have the most magnitude of impact, but there are few remaining schools without FP departments. Only a few schools have a rural mission or person.
Several of the variables interacted in a significant way.
§ MCAT scores correlated highly with NIH funding, and both of these could not be included in the regression together. Either variable inserted alone resulted in similar regression results. Medical schools with higher MCAT scores and also those with more dollars from the National Institutes of Health grants graduated fewer family physicians. MCAT Correlations
§ Schools with more rural born students, schools with a rural mission/person, and those with an FP department also had a lower MCAT. Removing any the former variables increased the magnitude of the MCAT component. Schools with more instate admissions, osteopathic schools, and public schools all graduated more FP doctors, but these variables interacted with the rural and FP department variables.
§ Schools with a rural mission/person had more rural born medical students and inclusion of both variables tended to weaken the contribution of either.
§ Older interacts with rural and MCAT also. Older and rural born have lower ratios of applicants to matriculants.Age and FP Physicians
See predictions of FP "match" based on background and ethnicity (social status)
Family Physicians Are Different In almost every characteristic that matters to admissions, those choosing FP are different. It takes different to be different in career choice, location, and integration.
Before Admissions - Physician workforce for the US can be predicted by characteristics of US medical schools. If the US truly desires more to serve underserved areas, it better start with education and work through admissions first and foremost.
Admissions and Social Status Could it be that rural, older, instate are all proxies for social status
See longer version at
Choice of Family Medicine: Past, Present, Future
Choice of FP Update Progress beyond the Arizona Study
Medical Schools and the Family Medicine Match
Rural Choices by Medical School Origin
Rural and Urban Comparison of all FP graduates as of 2003
