Regressions in Family Medicine and Rural Medical Education
Variables in the Medical School Database
Abstract: Regression variables can interact in some important and some confusing ways. They are often used and abused in policy analyses. It is important to understand some of these interactions to be able to interpret such studies. It is also important to explore new variables. Some of these may be more difficult to obtain, but they may do a better job of explaining why physicians choose rural practice.
Introduction
Regression studies are important to rural medical education. Rabinowitz studied Jefferson Medical School students to demonstrate that 78% of the decision for rural practice could be predicted from just two factors, rural background and family practice interest at medical school matriculation. Rosenblatt used regressions in a national study of allopathic medical schools to demonstrate that those schools graduating rural physicians tended to have a rural mission, less dependence on National Institutes of Health research funding, and a more rural population. Bowman did much the same with rural graduation rates of FP residency programs, adding rural rotations to the list of factors associated with this outcome.
One of the problems with doing regression analyses is that many of the variables interact in ways that interfere with interpretation. The greater the number of variables, the more difficult it is to interpret the findings. The challenge of a researcher in this arena is to choose the appropriate variables to avoid this confusion.
Some variables are so closely related that it is difficult to separate them. It is often difficult to avoid the associations between race, rurality, and poverty. Rural areas tend to be predominantly white and less wealthy. Not surprisingly with less wealth comes less educational expenditures.
The following are some of the interactions between education expenditures, education outcomes, income and poverty measures, race, and rurality.
Correlations inserted here
Primary Care and Rural Physician Associations
Also in rural areas certain associations are inevitable. There is a close association between primary care graduation rates and rural graduation rates. A major part of this reason is that rural doctors are by and large, primary care doctors, and in the medium to smaller rural locations, they are nearly all family practice doctors. Although some might use this association to point out that support of Family Medicine is important, there are alternative explanations that may make more sense. Studies of senior medical students note that those interested in rural practice are far more likely to know that this is their final career pathway than other students (30% vs 15%). Declines in the numbers of rural background students have tended to result in decreases in rural graduation rates and in the numbers choosing family medicine. Also boosting the numbers of family medicine residents has not resulted in more rural physicians graduating from FP residency programs. It seems more likely that Family Medicine is dependent on those who are committed early to rural practice and that changes in admissions (or in the preparation of rural background students for admissions consideration), is responsible for the declines in both rural and primary care numbers.
Minority and Rural
Minority physicians tend to originate from and return to inner city areas. Underrepresented minorities do tend to choose underserved areas and underserved populations at a rate 4 times the usual medical school graduate (40% vs 10%), but the practice locations chosen are not likely to be rural underserved areas. Xu demonstrated that black physicians did consider rural practice at rates approximating whites, but it was a rare Hispanic or Asian physician that did so.
Gender and Race
Minority Family Medicine graduates tend to be predominately female. Because of this and also because few minority graduates choose rural areas, gender and rural practice studies can be more challenging to interpret.
Perspectives To Consider: State or Medical School level
Another difficulty involves the selection of the population to be studied. One can select state comparisons, or compare medical schools. States vary greatly in size, composition, and proximity to other states. Medical schools vary by their focus, size, age, mission, and sources of revenue. Some have a more national presence as noted by higher numbers of students coming from outside of the state and a larger dependence upon National Institutes of Health grant funding.
Here are some of the variables that are useful when studying the relationship between education, higher education, professional education, and rural location.
Mismatch of data and outcomes
Sometimes the data involving a cohort that is being studied must correlate with a certain year. Interventions in admissions involve a delay of 4 years medical school and at least 3 years in residency.
Outcome measurements
There are a variety of outcomes to consider.
Total number of rural doctors graduating from allopathic medical schools divided by state population
Total number of rural doctors graduating from allopathic medical schools plus estimate of 15% of osteopathic graduates of the state in 1991 (average osteopathic rural graduation rate) divided by population of the state
Total number of rural doctors graduating from FP residency programs
Rural doctors can be defined by
· Location in a town of less than 10,000 by AAMC and AAFP data.
· Location in a town of less than 25,000 not adjacent to a metro area by AAFP data.
· Location in a town of less than 50,000 by AAMC data.
· Location in an area with a certain continuum code
Other variables to consider
Population of the state in 2000
Population of the state in 1990
Percentage of the state population in rural areas
State population that is in counties of less than 10,000
Hypothesis: States that spend more on education or that distribute educational resources more equitably graduate more rural or minority physicians
Education Weekly Data
State higher education expenditures per capita
State education expenditures K-12 per capita
Determination of educational grade according to resources
Determination of educational grade according to educational equity
Determination of educational grade
Hypothesis: The number of rural or minority doctors is related to the political power as measured by the voting population
Percentage of the voting population that is black
Percentage of the voting population that is white
Percentage of the voting population that is hispanic
Percentage of the voting population that is native
Underrepresented minority graduation rate
Per capit
New Variables to Explore
Service orientation