Robert C. Bowman, M.D.
Distribution of Physicians other links
The following table compares medical schools by their choice of family medicine regarding students graduating 1987 - 2000. The latest match data is used for an update, compared to MCAT average scores for schools in the group for 2000, and choice of rural FP by the students in these schools averaged as a group.
Divisions MCATavg FPMatch04 % RurFP 94-00
Allo Priv Least FP (11) 10.94 1.9% 0.5%
Allo Priv Next Least (11) 10.52 5.6% 1.0%
Allo Priv Next Most (11) 9.55 8.0% 1.7%
Allo Priv Most FP (11) 9.23 12.9% 3.4%
Allo Pub Least FP (20) 9.84 7.7% 2.0%
Allo Pub Next Least (21) 9.54 11.0% 4.0%
Allo Pub Next Most (20) 9.64 13.2% 5.6%
Allo Pub Most FP (20) 9.26 18.9% 8.3%
Osteo Least FP (9) 8.36 13.0% 4.9%
Osteo Most FP (8) 8.25 18.2% 10.1%
Note that the rural fpgp % was impacted by managed care by +25%, particularly in the schools least likely to distribute well. The FP match data above for 2004 is much lower than during managed care and even lower than 1987 - 1993. The impact on FP choice was +50% during managed care. Osteopathic data from Masterfile sources is missing about 11% of specialty choice, only 1 % data on specialty missing for allopathic. Location data is zip code by RUCA of Large, Medium or Isolated rural locations, all that were not Urban or Urban-focused, as per Hart and WWAMI site.
The MCAT is a key point to consider. MCAT scores do not tend to vary when comparing each other or past MCAT scores for 4 years. There is a .85 or higher correlation with each other.
MCATALL 2000 average Pearson Correlation Value, all highly significant
NIH Grant $ to school +0.659
RESEARCH Grads from school +0.65 - .75
Core Urban % +0.70
FMGP PER -0.607
FM % -0.607
PRICARE% -0.604
Rural GRAD % -0.594
% Rural Born Admitted -0.537
% over age 30 -0.504
IntMed Office Based % -0.482
Rural FP % -0.478
NON MET % of state 2000 -0.473
BACH or More % +0.449
GPA 2003 +0.444
Rural Mission or Person -0.440
Median Income of State +0.438
Ortho % +0.432
Internal Med % (all IM) +0.420
HIGHER Ed spending 93 +0.411
Rural Mission at School -0.401
% White Male -0.383
Cardio doc % +0.379
ASIA % admits + 0.375
These relationships involve comparisons of allopathic schools, excluding the atypical schools such as military, traditional black, and FP focused such as Duluth.
The correlations of MCAT are striking in impact on family medicine, primary care, and rural FP via the admissions emphasis that we give MCAT. Basically everything that would help distribute physicians is defeated by reliance on the highest MCAT scores. The factors of age, less urban origins, US born, and income levels all correlate with lower MCAT scores and better distribution. Schools with the highest scores distribute almost nothing yet receive the most NIH and GME dollars. Students with the highest MCAT do not choose FP or rural. The effect of MCAT using controls and other variables such as longitude and education is such that each 1 point increase in MCAT is about a 4 percentage point decrease in the FP choice of the students at a school, even when the higher and lower MCAT schools are taken out.
Accreditation, legal pressures, emphasis on physician researchers, US News and World Report ratings, USMLE 1 emphasis, all push the US to less and less distribution through admissions, aided by poor orientation to distributional careers during school (likely decreasing now), less than stellar and decreasing intensity of training (impact of liability, audits, and loss of teaching emphasis in medical schools), centralization of medical school and residency training in the most urban areas (only 7 outside most urban areas), loss of accountability for physician career choices (loss of managed care and moving to "market forces"), and poor support and finance of those who do choose distribution, particularly in the liability category.
The MCAT is no different than the SAT and ACT regarding income, ethnicity, and urban bias and the cumulative impact of ACT/SAT and MCAT is likely. The problem is not standardized testing, but over-reliance on standardized testing. The standardized tests are also the least valid on the students most likely to distribute well. Studies on those choosing primary care note a clear discrimination bias in the MCAT, likely a result of income differences. Studies in lower income and black and non-traditional students all demonstrate that the correlation between MCAT and USMLE 1 is not the same 0.6 or higher as in the "usual" urban high income medical school admission.
No physician would use any medical test with the specificity and sensitivity of the MCAT. Studies in Alabama note that MCAT based determinations can identify 90% of those at academic risk, but 80% will still graduate. Multi-school studies note that MCAT subscores of 8, a total of 25, and a GPA of 3.0 give maximal opportunity for diversity without increasing academic failure to a great degree (Albanese). The MCAT appears to be a good predictor of high academic performance, but does not discriminate well of those with average or above average levels.
There is also even more to be concerned about. The increase of the MCAT of 1 point in bioscience alone in the past decade may indicate that students and colleges are gaming" the system and teaching to the test, without increasing academic preparation. These games may be the exclusive venue of those of higher income or schools with more higher income students who have already been scoring higher. This, coupled with the rapid increases in admissions of those of higher income (see yearly data past decades from AAMC), also indicates that medical students are increasingly unlikely to choose family medicine, rural practice, and service to the underserved. The MCAT use may also be responsible for declines in the probability of rural born students being admitted, as well as lower numbers and ratios of rural, older, and low income students over time. The declines in some minorities choosing family medicine also suggests that higher income minorities such as urbanized Native Americans are gaining access to medical school while lower income minorities are not.
The legal controversies in admissions are also a concern and have basis in science. White and Asian students have a very narrow margin between those accepted and not accepted of about 1 point in the MCAT. Many if not most of the students not admitted have scores better than the 8 - 25 and 3.0 noted above. Other ethnicities have a 1.6 - 2 point MCAT differential between those admitted and not (AAMC Minorities in Medicine 2001). There are also differences within the designated ethnicities. Vietnamese parents have lower income origins and Vietnamese medical students choose FP at 28.5%. Mexican Americans have more rural and lower income origins and choose FP at 20%. Indian-Pakistani parents have the highest income levels and their children choose FP at 2.2%. Other Hispanics have higher income and lower FP choice (Bowman studies FP grads 2001-2003 compared to 1998-2001 US allopathic med school grads by same ethnicity).. The more rural the county type for birth origins of the student, the lower the income, and the higher choice of family medicine. Education resources are also related to income levels through property taxes. Some states clearly just do not have the resources to prepare the students who will become family physicians to the level needed to gain entry to medical school.
There are also concerns regarding use of the MCAT. Asian students peak out with MCAT and do not do as well on USMLE 1 as predicted by past MCAT and GPA levels. Hispanics also share some of these characteristics. White, primary care, and lower income students do not gain admission (declines in past decade in AAMC data) and improve in USMLE 1 and 2 and in residency (Jefferson Longitudinal studies: Xu, Veloski, Callahan, Nash, Erdman, Gonnella, Laine, Hojat, Cooter, Barzanski, others). This is not likely an impact of ethnicity so much as income. There can be nothing wrong with a group that dedicates itself to education. Asian students are a pure group of higher income, most urban, highest educated and professional parents. About half are born in the US and half overseas. This group represents the direction of US medical school admissions with increase from 400 to over 3000 in the past 15 years. Again this is not so much Asian as a group with a concentration of characteristics also seen in urban whites and other ethnicities highly urban and higher income. Students with privileged origins and every parent advantage are likely all that they can be before medical school. Students of more humble origins have greater potential for improvement if given better opportunity. Older students in particular seem to be able to take advantage of the opportunities in medical school and choose careers involving direct patient care in much greater percentages. The changes seen with age also tell us what we must do to improve all US physicians, we must require those who are academically prepared to be maturationally prepared. We need to raise the bar in people skills as high as we have in academic skills and not cave in and admit physicians who cannot meet the tests in both categories.
Just did a new study involving Teachability and Education efficiency, indexes involving student preparedness for school and barriers to education. Again any gifted student can become a physician, but it takes real education investment, distribution, and efficiency to graduate those who will become family physicians. States can be ranked in adjusted school efficiency with states such as Utah Montana and Idaho on top and DC, NJ, and CT on the bottom. Note from Greene's publication "The Adjusted School Efficiency Index gives students' academic achievement as a percentage of the achievement level predicted by their teachability and state education spending adjusted for cost of living." (Jay P Greene, The Teachability Index: Can Disadvantaged Students Learn? Education Working Paper, Manhattan Institute).
http://www.manhattan-institute.org/pdf/ewp_06.pdf
I compared admissions of those who will become family physicians with the final state index in the pdf file. There was a raw unadjusted 0.38 correlation with this scale that would likely be higher when detailed analysis was done. When I compared states that had increasing admissions of students born there compared to other states, the states that are making gains in admissions compared to other states have higher index scores. In some ways measuring family medicine admissions is a great measure of education outcomes, given the tendency of those with speeded intellect to choose other specialties. Again, it takes the efforts of an entire state to graduate family physicians, as those in Iowa and West Virginia and others have found.
Admissions Ratios and US Medical Students
Robert C. Bowman, M.D.