The Physician Distribution by Concentration Coding System
Robert C. Bowman, M.D.
Physicians per 100,000 Population (300 average) |
Total % of US Physicians By Type of Practice Location |
% of Total US Population and % of Total FM Physicians By Type of Practice Location |
Physician to Population Ratio |
Higher or Lower Probability of Medical School Admission compared to 100% For National Average | |
Super Center 200 or more physicians and < 1% land area |
1100 phys per 100,000 about 40 � 80 FM per 100,000 |
41 � 46%, 50% if residents counted, 55 � 70% of some subspecialties |
12% of pop 20% of FM physicians |
4 to 1 |
200% to 900% |
Major Center 75 to 199 physicians and 3% land area |
400 phys per 100,000 about 50 � 70 FM |
20 � 25% of physicians |
22% of pop 27% of FM physicians |
1 to 1 |
80% to 150% |
Marginal Urban higher income and lowest poverty |
150 phys per 100,000 about 40 FM per 100,000 |
20% of physicians |
35% of pop 25% of FM |
1 to 2 |
80% to 120% |
Urban Underserved lower income and higher poverty |
80 phys per 100,000 about 30 FM per 100,000 |
4.2% of physicians |
13% of pop 7% of FM |
1 to 3 |
20% to 40% |
Marginal Rural average income and average poverty |
130 phys per 100,000 about 51 FM per 100,000 |
4% of physicians |
8 - 9% 10% of FM |
1 to 2 |
50% to 70% |
Rural Underserved lower income and higher poverty |
105 phys per 100,000 about 39 per 100,000 |
3% of physicians |
8 - 9% 9% of FM |
1 to 3 |
20 � 50% |
Existing zip code practice locations were also smoothed while considering adjacent zip codes. The zip code and adjacent zip codes are a proxy for the typical catchment area of a primary care practice.
The national admission average for 1994 - 2000 was 1 medical student admitted for 200 of age 18 - 24 in the US. The 1990s admission ratio compared to 1970 birth county population was 7 admitted per 100,000 county population or 8 per 100,000 per class year when including the 14% foreign born raised in the US and admitted to US schools.
As probability of admission increases, probabilty of exclusive school and exclusive career choice increases along with top probability of super center location. Physician origins associated with lower probability of admssion have fewer choices of types of medical schools, are more likely to be admitted to more normal and less exclusive medical schools, have greater probability of choice of family medicine to multiply rural or underserved choice, and have the top probability of most needed health access career choices. Logistic regressions on complete populations of physicians confirm these findings for individual physicians tracked from birth to admission to career choice to practice location.
Concentrations shape admission, training, career choice, and practice location with regard to physicians.
Only physicians born outside of concentrations, family physicians, physicians older at graduation, and physicians trained in about 60% of medical schools are associated with higher probability of practice location in zip codes with 65% of the US population outside of concentrations in rural and urban, marginal and underserved locations.
Physicians with exclusive origins, younger or normal age at admission, career choices other than family medicine, and training in the top 40 US medical schools ranked by MCAT scores all have lower probability of most needed health access career choices (rural location, underserved location, family medicine). The combination of all concentration factors results in over 83% found in Super Center and Major Center concentrations with absolute lowest rural, underserved, primary care, and family medicine careers. For any type of physician, any type of career, any type of medical school, and any type of health policy, when greater concentrations are found in super centers, this results in fewer found in the most needed health access careers.
With 75% to 92% of the highest paid specialists, 72% of total physicians, 70% of internal medicine and pediatrics, top concetrations of hospital facilities, and over 90% of research, medical school, and graduate medical education distributions going to 3400 zip codes in 4% in the land area in top concentrations of physicians and income and people, health care is more challenging and less likely for more than a majority of the United States population.
Abstract:
Purpose: Geographic coding systems typically cost millions of dollars to develop but fail to consider the most important determinant of physician practice location � concentrations of physicians. A direct coding system based on physician concentrations makes more sense compared to coding physicians by concentrations of people (rural vs urban) or income. A coding system that considers local zip codes as well as adjacent zip codes is also appropriate for primary care for populations in most need of health care. Elderly populations with limited mobility, lower and middle income populations, and populations with reduced transportation depend upon local care. Geographic coding systems also do little to indicate potential solutions as geography is not amenable to change. A coding system that considers concentrations of physicians is an aid to understanding the forces that concentrate physicians as well as the forces shaping shortages of physicians. Nations can change concentrations of health funding and concentrations of physicans to result in better distributions of physicians, health resources, and the economics related to physicians. But nations must first translate existing concentrations into text and graphics that can help the nation to visualize the problems and the solutions.
Methods: Physicians were divided into concentrations inside of zip codes with 75 or more physicians and distributions outside of concentrations in practice zip codes with less than 75 physicians. Zip codes with 19% or more of the population in poverty were coded as underserved along with zip codes designated as a Community Health Center, National Health Service Corps, or whole county primary care shortage area site. Zip codes with less than 75 physicians that did not have high poverty or a designation were coded as marginal. The underserved and marginal zip codes were also categorized into rural and urban locations using RUCA 2.0 coding. The final coding included 8 categories - Super Centers with 200 or more physicians at a zip code, Major Centers with 75 � 199 physicians, Marginal Urban locations, Marginal Rural locations, Urban Underserved locations, Rural Underserved locations, Military locations, and International locations. The RUCA coding was also used to consider isolated rural locations. Comparisons included concentrations of physicians, population, primary care physicians, family physicians, and populations in poverty,
Results: Over 72% of physicians (even higher levels when considering residents in training) were found practicing inside of top concentrations in super center and major center locations with 35% of the population in less than 4% of the land area. Family physicians were evenly distributed inside and outside of concentrations, internal medicine and pediatric forms of primary care had 70% levels of concentration inside, and specialists had 75 � 92% of graduates located inside. Non-family physicians were three times more likely to be found in concentrations. Family physicians were twice as likely to be found in urban locations in need of physicians outside of concentrations and were 3 � 4 times more likely to be found in rural locations outside of concentrations of physicians.
Areas outside of concentrations were marginal locations with half of the national average physician concentration of 300 physicians per 100,000 and underserved locations with 80 to 100. Isolated rural locations and urban underserved locations had the lowest physician concentrations and primary care physician concentrations with about 60 physicians per 100,000 or one-fourth of the nation average of 300. With decreasing concentrations of physicians at a location, the proportion of family physicians increased relative to total physicians or total office based physicians. The concentrations of family physicians were steady. Concentrations of non-family physicians declined with decreasing concentrations of physicians, income, and people.
Rural locations with concentrations of physicians contained about 3% of physicians or about one-third of rural physicians. Rural locations inside of concentrations had the same lower family medicine and primary care percentages found in urban super center and major center locations that also had top concentrations of physicians. Rural locations inside of concentrations were very different in physician workforce compared to marginal or underserved rural locations that had 40 � 100% of local physicians found in family medicine. This is likely to present a problem as large systems based in rural concentrations of physicians clearly have a different type of physician and a different type of health care in mind as they increasingly dominate nearby smaller rural locations.
Conclusions: The Physician Distribution by Concentrations coding system helps to establish a dependent variable suitable for studies of physician practice location. While the coding system does capture the most underserved locations, it also reveals a most important element in American health care with regard to practice locations with top saturations of physicians. The current health care design concentrates physicians and health resources with 72% of physicians (and now likely more) found in 3400 zip codes in 4% of the land area. As a consequence this design results in 65% of the population found outside of concentrations. Matters are complicated as primary care saturations are also found only for the 4% of the land area inside of concentrations while zip codes with 65% of the population are left behind. Given the same problem in the finance of health care with health care coverage greatest and most beneficial for those of top concentrations and lowest for lower and middle income peoples and given the same problems with overutilization and increased utilization in populations associated with top access and lower or insufficient utilization in populations with lesser access, the US health care design appears to be consistently flawed in form and function as well as in infrastructure, finance, and outcomes.
The studies based on the Physician Distribution by Concentration coding are consistent with the medical literature regarding physician practice location. Consistently the physicians ordered by their life experiences related to concentrations are consistent in career and location choice. Using logistic regression odds ratio probability of practice location (super center, underserved, rural, inside of concentration, outside of concentration) can be generated. The physicians that are found serving populations left outside by the American health care design include physicians with origins outside of concentrations, physicians older at medical school graduation, family physicians, and physicians trained in more normal and less exclusive medical schools. Most urban and highest income physician origins or birth associated with concentrations of physicians; younger or normal age at graduation (early admit or no delay); subspecialty, specialty, or hospital support career choice; or training in allopathic public or a top ranking MCAT school is associated with greater probability of practice location inside of concentrations.
The physicians tracked by PDC can also be linked to their birth origins using the Masterfile. This can be used to generate the probability of admission for physicians in each type of practice location, specialty, or medical school. Those most likely to gain admission are most likely to be found in exclusive medical schools and are least likely to choose family medicine (and are most likely to choose an exclusive specialty) and are least likely to be found serving in most needed health access careers. Those with origins associated with lower probability of admission are more likely to be found in more normal medical schools and are more likely to choose family medicine and are more likely to be found in rural, underserved, primary care, and family medicine careers.
As the US moves to more exclusive in admission, training, and health policy distributions of resources, basic health access is more and more difficult to address. Physicians and non-physicians respond to admission, training, and health policy changes. Physician assistant and nurse practitioner movements away from primary care and away from the family practice broad generalist mode in practice also involve movements toward hospital, specialty, and subspecialty careers is also associated with declines in basic health access contributions.
The Physician Distribution By Concentration coding reveals the impact of life experiences or experiential place involving concentrations (income, people, physicians, social organization, higher standardized test scores, higher probability of admission) with regard to career and practice location choice as well as the influences of less concentration in life experiences (more normal in income origins, population density origins, physician concentrations, more normal in scores, generalist versus specialist lifestyle, lesser social organization). Studies illustrate that physicians that are born, raised, educated, and trained for the first 30 years in concentrations of income, people, professionals, and physicians are least likely to address the health care needs of the 65% of Americans left outside of concentrations. Those more normal and less exclusive in origins, training, and career choice have greater probability of meeting the nation�s most important health access needs.
The only factors known to influence location outside of concentrations are physician origins outside of concentrations, family practice broad generalist career choice, older age at medical school graduation, and medical education focused on less than the most exclusive admissions, career choices, and training locations. Distributions of physicians are also most important for the elderly that are 70% outside of concentration that quadruple in primary care needs from age 40 to 80. Even with close to universal financial access, the elderly found outside of concentrations of physicians with highest costs of health care and living may not be able to find a physician to care for them.
Given highest reimbursements for the most subspecialized physicians as well as their hospital locations, given multiple lines of revenue and the highest lines of revenue in each line going to zip codes inside of concentrations (especially medical schools), and given fewer physicians and the lowest paid physicians and facilities outside of concentrations, a reasonable estimate is 85 � 90% of health care revenues associated with physicians going to 3386 zip codes in 4% of the land area in top concentrations. Once again the US health system design makes health access most difficult with only 10 � 15% of the health resources attributable to physicians flowing to zip codes with 65% of the population that are found in marginal and underserved zip codes. The challenges are magnified since marginal and underserved populations include 70% of the elderly and others most likely to have health literacy issues, chronic illness, poverty, deficient education, unemployment, and lack of health care coverage.
When the impact of physician concentrations is understood, problems with many health care situations could be resolved. These include government grants that end up sending more funding to zip codes that already have top concentrations of physicians. Another consideration is the impact of concentrations of physicians (Manhattan, large metro concentrations) to suppress nearby physician location (surrounding boroughs, rural areas near to physician concentrations). Also rural children raised in concentrations of income (professional parents, higher income parents) or in other concentrations (major college or research facility in the county) share the same highest probability of admission with lower choice of health access careers as those raised in the 33 counties with top concentrations of people (over 2500 people) income, physicians, and medical education. The same issue applies to underrepresented minority students with parent influences involving concentrations that may not share much in common with lower and middle income Americans left behind in the health care design. Attempting to send physicians with origins and training in top concentrations to locations with more normal or lower concentrations has not been a good strategy to distribute physicians equitably.
Only greater balance in physician origins, more life and health experience prior to admission, more normal and less exclusive in training, and family practice choice are associated with more equitable distributions and the potential of resolving basic health access problems in America.
Radical changes shifting substantial resources away from concentrations such that basic health access can be improved seems indicated; however experience with medicine and medical education indicates that such changes are followed by other types of problems and the rebound of medicine against such changes. But a steady movement of resources toward the 65% of Americans left behind is indicated. Of course American leaders can risk not addressing these changes but this would not be advisable given 70% of the elderly left behind that will double in the next two decades or the distrust in government that is a constant fixture when most Americans have limited access to health care.
Most importantly the PDC coding system reveals the great difference in perspective between physicians and American leaders as compared to 65% of the population. Leaders of the nation, physician leaders, and concentrations of physicians reside 75 � 90% inside of concentrations in just a few percentage points of the land area. These are locations, populations, and situations that are different from the Americans outside of concentrations that are far more numerous but less socially organized. This is most relevant in basic infrastructure areas such as child development, basic education, and basic health access. It is difficult for those immersed in concentrations to understand the situations facing most Americans, their children, and grandchildren. For health care in particular, if there is not consideration of populations, physicians, and medical students from outside of concentrations, there will be less and less health care delivered outside of concentrations. The economic divisions alone arising from this great divide involving 2 trillion dollars a year in expenditure and economic impact are enough to move the United States from an advanced developed nation to a very different category of nation.
Introduction
The Physician Distribution by Concentration coding tool has been a consistent effort by the author over the past decade. The author has been immersed in basic health access delivery, medical education, and research for 26 years. Not surprisingly the PDC coding system is most relevant for basic health access.
The first challenge regarding the development of a new coding system for physicians is a data. The American Medical Association Masterfile was used to code physician origins, career choices, medical schools, age at graduation, and practice locations. In order to work with data, a researcher must learn to speak the language of Masterfile as well as Masterfile limitations. One example is the need to avoid use of the most recent class years of graduates (2001 to 2005) as the data contained in the Masterfile is least accurate for the most recent class years. For the purposes of coding, the year 2005 version captures a representative distribution of physicians by career and location choices. Coding requires thorough immersion in the data to begin to understand distributions.
Another barrier for a new coding system is funding. There was no specific grant or funding. As with the delivery of basic health access, the author worked after hours and weekends to develop the database. There was cooperation and consultation from family practice and rural associations.
Acceptance of any coding system is most difficult without development by the government or by medical associations that control the reporting and publication. Most importantly those reading reports based on PDC coding must understand the basics of such coding. Hopefully the utility of a coding system that more closely captures the nature of physician distribution will be realized as important in areas such as health access. With greater acceptance, more will understand the PDC system. Actually the concepts are not difficult. As with other areas involving major progress, much must be unlearned so that progress can result. This was true in the case of the author and hopefully his work will allow readers to accelerate their own unlearning so that they can learn about important areas such as basic health access.
The PDC system may not be as important for the physicians found in top concentrations. For basic health access, local zip code or adjacent zip code practice locations are important for all and especially for the elderly that steadily lose mobility and for all with limited transportation. A major role of health care is to promote efficient and effective societal function. People facing barriers of access for basic health care are not as efficient or as effective. The convenience care more population has the potential for introducing quality and cost problems as well as decreasing basic continuity primary care access as primary care forms are consumed for convenience care. If physicians and non-physicians are tracked moving toward concentrations class year to class year and with each year after graduation, the nation could anticipate worsening problem with basic health access. By knowing the types of physicians and non-physicians that are associated with basic health access, the nation would know what is required to maintain basic health access in the United States.
Common variations remain a problem for physician coding. Physician locations have been characterized by concentrations of income, poverty, economics, and people. While these coding systems remain true to distributions of income, economics, or people, there are inconsistencies with regard to physicians.
Family physicians also elude economic and geographic coding. Unlike other physicians that follow top concentrations of income and economics, family physicians remain at 30 � 40 family physicians per 100,000 across a wide range of US populations. The same is not true for pediatric and internal medicine primary care forms that concentrate in concentrations of physicians, income, and people. Studies of cost, quality, access, and distribution that lump primary care forms together are significantly flawed. Family physicians are also tracked to birth origins with the same 30 � 40 family physicians arising per 100,000 people. More normal in origin, in training, and in practice location is a consistent family medicine characteristic. The same is found for nurse practitioner and physician assistant graduates who remain in the family practice broad generalist mode. Family practice modes sharing race, ethnicity, or geographic origins also have top distribution to most needed health access locations.1-6 When physician coding by concentrations is considered, the family practice forms distribute with 50 � 60% of graduates found in locations with 65% of the population with only 40 � 50% found inside of concentrations. Other specialties in physicians and non-physicians are found with only 10 � 30% outside of concentrations. Non-family practice specialties are associated with concentrations with 70 � 90% of graduates found in 4% of the land area in top concentrations of physicians. This makes sense when health resources are understood as concentrated 90% in such locations. Also movements toward these locations are not a surprise.
Rural locations generally have lower and middle concentrations of income, people, and physicians but a common mistake made by workforce experts is to consider rural areas and populations to be alike. For example using the PDC coding, one-third of rural physicians are found in locations with top concentrations involving hundreds of physicians. Physicians can be found concentrated in small and even isolated rural zip codes at the highest levels despite lower concentrations of people and income.
Shortage areas are also inconsistent without consideration of physician concentrations in local or adjacent zip codes. Shortage area definitions could be adjusted yearly or even monthly, but these efforts can be counterproductive and can even inhibit most needed health access locations. Abuses of all federal programs to distribute physicians are common. Even the best efforts have required revisions and regulations year after year to reduce abuse, fraud, and misuse to a tolerable level. Even with such efforts, the increased regulation forces hiring of consultants, placing even more barriers in the way of populations in most need of health access. Federal dollars are commonly used to support practice locations that have the top concentrations of physicians and health resources in America. Social organization plays a role in physician concentration that cannot be measured by studies of population density or income density. Physician concentration does capture physician concentration directly, however.
One theme is apparent in all of these variations as well as in normal distributions. Physicians locate practices according to concentrations of physicians.
Until the nation understands concentrations, it will fail to understand distributions. In the health care dimension this is often about primary care and health access.
Under the current system, health care coverage is linked to jobs and economics. Workforce experts often consider poor physician distribution to be the result of poor economics in rural or underserved areas. If admission, training, and health policy choices result in greater concentrations of the health care economics closely associated with physicians, poor economics in rural and underserved areas can be the result of national policies and practices that concentrate physicians.
Extreme concentrations of physicians can represent a major problem for health care access. Balanced distributions of physicians and populations facilitate health care access. Concentrations of physicians also mean that significant United States populations are left behind. Also physicians in concentrations of physicians fail to have much awareness of populations outside of concentrations. Since physicians are involved in health system design, this lack of awareness can be programmed into the system with the development of medical education, graduate medical education, and reimbursement policies that fail to understand or support populations and physicians who are located outside of concentration
Methods
The Physician Distribution by Concentration methodology was based on a single zip code practice location within the context of adjacent zip codes. The coding system was also established at the same time as the author was coding the birth origins of physicians in the United States. This also involved a comprehensive study of geography and history to identify origin points that no longer exist (ghost towns, Japanese Detention Camps), those with name changes, and past and present military bases. Both efforts required 5 years to capture the basic framework of physician distribution in a process of immersion, translation, categorization, and application. The Framework of Experiential Place captures these concepts best with the dominant theme being physician origins, training, or practice locations inside or outside of concentrations.
The process began by compiling the number of total active physicians at each zip code using the 2005 American Medical Association Masterfile. If there was no practice zip code listed, the alternative practice zip codes were used, then the home zip code, then any zip code listed for the physician. Physicians had to be listed as alive and active to be included in the zip code counts. The physician zip code distributions were grafted into an existing zip code database with year 2000 census data.
Categorization began with a division of inside and outside although at the time the choice was an arbitrary dividing point. The initial coding methods considered levels of 50, 75, or 100 physicians at a zip code as the defining point for concentrations of physicians. After a review of the types of physicians at each of the locations, the facilities at such locations, and the demographics of the locations, the 75 physician level was selected as the most consistent definition for physician concentration.
Consideration was given for a separate academic or medical school location category as compared to typical physician concentrations. Medical school zip codes did have greater physician concentrations often due to graduate medical education positions, but physician practice zip codes associated with concentrations were similar regardless of isolated rural, urban, military, or medical school location.
The efforts to consider medical schools did reveal that there were differences in the levels of concentration with separations between those with 200 or more physicians. Locations with 200 or more physicians had different types of physicians and different demographic characteristics. Major Center zip codes were defined as locations with 75 � 199 physicians. Locations with 200 or more physicians became Super Center concentrations. Super Centers and Major Centers as used in the categorization are not representations of existing corporate entities. These terms represent concentrations of physicians.
For simplicity all military locations were kept in the military category. International practice locations were a small fraction and the data was considered poor quality in this area outside of the United States. There are few mechanisms to update data in this area and many are still listed in their final residency or practice locations prior to departure, creating inaccuracies particularly in international medical graduates who return to home nations or move to other nations.
Since the focus was on physician distribution, the Masterfile was constantly used to verify the categorizations in an iterative process using all possible data fields. Concentrations of physicians with medical teaching, research, and resident primary practice activity fields were used to identify locations likely to be medical school locations. The physician specialties with 85% or greater levels found in concentrations were used to identify other zip code locations likely to represent concentrations. Type of practice fields were used to identify government or other facilities likely to represent concentrations of physicians.
The Underserved Categories: Half Served to One-Fourth Served
The next step was a consideration of locations with the lowest concentrations of physicians. This began with a zip code plot of ratios of primary care physicians to population. With poverty approaching 19% and beyond, levels of primary care physicians declined. Locations with high poverty have difficulty maintaining primary care without mechanisms of health care support. Primary care levels increased in Community Health Center and residency training zip code locations. A convenience listing of family medicine residency training zip codes was used for this latter comparison.7
Census data for 2000 was used to divide zip codes into locations with at least 19% of the population in poverty to compare this method of underserved coding to existing categorization systems. Eventually levels of 0 � 14%, 14 � 19%, and 19% and over were established for use in detailed coding. Levels of 14% and below were considered at or below average poverty levels of 12.7% for the United States.
Zip code listings of Community and Migrant Health Center (CHC) sites, National Health Service Corps (NHSC) sites, and whole county primary care (PC) shortage areas were obtained from federal web sites from 2001 � 2003. Newer federal designations since this time were not included as these did not impact the 2005 Masterfile distributions. There were other federal designations considered (partial county, township, Medicaid, prison designation), but examinations did not reveal the same consistency. The underserved category could be divided into rural, urban, isolated rural, poverty, or various designated subgroups. Many of the zip codes had two or more of the factors. While there were small differences such as 21% poverty levels for rural underserved compared to 24% for urban, there was consistency across the subgroups.
In the process of setting the standards, there was not a predetermined level of underserved physicians based on standard deviations or a specific level of 5% or 10% of total physicians. The consistency was set by the coding process involving poverty levels or federal designations. Even some zip codes with federal designations had to be excluded from the underserved category due to high concentrations of physicians (over 75 at a zip) or poverty levels below 14%, levels not much different than the 12.7% national average for poverty. Because other factors such as health care outcomes are used in shortage determinations, the federal sites with 14 � 19% poverty level were maintained as underserved. The current federal shortage designation inclusion of zip codes with normal poverty surrounded by other zip codes also without significant poverty appeared to be inconsistent.
The unique zip codes presented a challenge. These zip codes often had no population or population levels insufficient to support a single physician. Adjacent zip codes were used to determine patient populations, physician concentrations, and poverty concentrations. The 4 zip code grouping represents the catchment area of a typical primary care physician location. The use of adjacency for unique zip codes was also applied to all zip codes for consistency, grouping up to 4 adjacent zip codes together for comparison. The author reviewed over 43,000 zip codes across geographic categories, latitude, longitude, poverty, income levels, and physician concentrations to assure consistency in coding across adjacent codes.
Consistent coding of underserved zip codes was also a goal. Adjacent zip code comparisons using zip code poverty levels smoothed the underserved location designation. Zip codes with less than 2000 people were reviewed in detail. Zip code total populations and poverty populations were considered for adjacent zip codes. In rural counties with few zip codes, county income levels were used to facilitate this process. Zip codes in rural counties with income below $37,000 in median family income in 1999 were also included as underserved. These counties represented the bottom 7% of the population in income. Few of these zip codes were added as most were already included as underserved. This method used in rural areas could not be used with urban counties. Urban counties had multiple zip codes and a variety of location types within the same county. One advantage of the PDC system is that the locations within urban counties can be coded by concentrations of physicians.
The coding was maintained for 36 different types of locations with internal consistency for each type across concentrations of people, poverty, income, and physicians as well as shortage designations.
Integrating Rural and Urban Coding
Physician practice locations are entered as zip codes. Zip codes based on concentrations can be integrated with geographic zip coding methods. RUCA 2.0 urban zip codes typically are codes of 1 � 3 with rural codes 4 � 10. Rural zip codes with 30% of the population commuting for work to adjacent urban areas were considered urban focused (4.1, 5.1, 7.1, 8.1, and 10.1). The �.1� or urban focused codes join the urban codes (1 � 3) for the urban totals. RUCA Categorization A divides rural locations into large rural (4 � 6), small rural (7 � 9), and isolated rural (10 � 10.6) locations, again with the exception of urban focused codes.8 This also allows coding for isolated rural locations.
Geographic coding was initially used to divide concentrations of physicians into rural and urban. The rural component only contained about 3% of total physicians. The rural and urban locations were also different only in rural versus urban location. The division into Super Centers with over 200 physicians and Major Centers with 75 � 199 represented a different type of practice location with different physicians, physician concentrations, and marked differences in primary care physicians.
There are other issues important to understand regarding the new coding system.
Half Served, One-Third Served, and One-Fourth Served Locations
More than a few National Health Service Corps, Community Health Center, and whole county primary care shortage zip codes had over 75 physicians. These locations often had top concentrations of physicians. For this reason all locations with over 75 physicians were coded as major centers or super centers regardless of federal designation or poverty level. These are locations where access to care is not likely to be about the physician access factor. Access involves financial, health care coverage, transportation, and numerous other dimensions. Adding a few physicians is unlikely to make a difference with hundreds already present and top concentrations of primary care in the nation at 100 to 350 primary care physicians per 100,000, levels more than sufficient for health care access. These are locations that also hire the lowest percentages of family practice and primary care physicians and there is nothing to stop them from accepting government funding for primary care positions while deciding to support fewer.
With 65% of the population in urban or rural locations with one-fourth to one-half of the national average physician concentration of 300, allowing special funding to locations with physician concentrations remains questionable. Leaving populations behind in health access is also about re-routing physicians where they are most needed.
The remaining locations that were not major center or underserved locations were coded initially as �served� locations. The initial assumption was that there was a middle ground. This was incorrect. The �served� locations were not much different than underserved locations. The national average of concentration was 300 per 100,000. Physician concentration levels in major centers began at 400 and increased to 1100 per 100,000 for super centers. The so-called served locations had 120 (rural) to 150 (urban) physicians per 100,000. This indicated a marginal concentration of physicians at a level only half of the national average. The term Half Served captured the overall physician concentration and the fact that Half Served locations also had half of the recommended concentration for primary care. The terms Marginal or Half-Served are used interchangeably in this study. Half served urban locations had the lowest poverty levels and higher income levels. Half served rural locations had slightly above average levels of poverty and lower income levels although not as low as underserved locations.
The descriptive nomenclature relative to physician concentration was continued in urban underserved locations that had One-Fourth Served concentrations and rural underserved locations with One-Third Served concentrations. As a group underserved locations had the lowest concentrations of physicians, but urban areas had the lowest physician concentrations of all. This is a possible result of proximity to top concentrations of physicians in nearby urban areas suppressing urban underserved physicians to the lowest levels. Higher costs of delivering health care may also squeeze physician support levels. Poverty levels were twice as high in underserved locations compared to marginal locations with slightly lower differentials for rural marginal compared to rural underserved.
The error rate in the coding system appeared to be greater in the urban marginal or half-served locations. When physician practice locations only included zip codes marked as practice zip codes, the levels of physicians in urban marginal locations decreased with more coded in concentrations or in underserved locations. Urban marginal locations have higher levels of unique zip codes. Zip codes without population can often be health care facilities and may well represent branches of super center and major center institutions. Physicians that list home or alternate zip codes may be coded in urban marginal locations but may be working in nearby super centers and major centers. With 75% of physicians in such locations, this is likely. The urban marginal locations with lowest poverty are likely to represent residential locations although distributions of primary care physicians to such locations are common and complicate further analysis.
A comprehensive coding system with 36 different categories (super, major, rural, urban, isolated rural, poverty over 19%, poverty 14 � 19%, Community Health Center zip code, National Health Service Corps site, whole county primary care shortage area, military, international, medical school) was maintained but the major focus was concentrations and distributions.
Definitions of Specialties
Specialty designations in the Masterfile are self-designated by physicians but there is input from graduate medical education sources. Comparisons with family practice data reveal very few differences other than in the most recent osteopathic graduates. Delays in capturing recent osteopathic data, data on Puerto Rican graduates, minority graduates, and international graduates are common. These are also populations that are less likely to respond to surveys from the allopathic data sources. Over time the data does move toward consistency. Including only physicians graduating before 2001 (active graduates 1940 � 2000) using 2005 Masterfile data can reduce the delay factor to a minimal level.
The various primary care specialties of family medicine, general practice, internal medicine, pediatrics, and medicine pediatrics were combined together as the physician form of primary care. There is a problem including all generalists together. Inactive primary care is still generalist primary care. Not classified primary care physicians have unknown locations and many are actually not in the United States (international graduates). Not classified and osteopathic generally means poor data. Other in primary practice activity is another unknown but this is a group with major center and super center location at higher levels, not generally an indicator of direct patient primary care. Researcher, resident, and medical teaching categories are also poorly reflective of primary care delivery. Administrative and hospital-based careers also indicate lesser primary care delivery. With increasing hospitalists, generalist primary care levels are likely to reflect actual primary care and health access poorly. For these reasons, the author includes as primary care only the physicians who designate a primary care specialty and also the office-based primary practice activity.
An example regarding international medical graduate studies can help. Studies attempt to include data on the most recent graduates. About 20 � 40% of international medical graduates are listed in residency. Another 20 � 40% are listed as not classified. Departures to other nations at graduation are common as are returns to residency training after a few years of service. Researchers and associations should not use secondary databases when attempting to study recent graduates that are known to have higher levels listed in residency or not classified. Direct practice location capture is required. Cross section studies should include appropriate time for the delays in data entry and time for physicians to complete training to assure final career selection, representative practice location, and verified United States location.
The office based response in the primary practice activity field combined with a primary care self-designation in the specialty field is consistently the best indicator of actual direct primary care contributions. The level of office based primary care percentage increases with medical schools that graduate more family physicians (also have lower MCAT), with birth origins (lower and middle income and rural birth, older graduates from medical school), with practice locations associated with higher levels of primary care (outside of concentrations, rural), and for graduates of the 1970s and 1990s, time periods with better primary care policy when graduates were more likely to choose and remain in primary care. For example the internal medicine residency graduates during the 1980s decreased to 44% but then returned to a majority of 54% found in office based primary care. The office based proportion, including only those who remain in internal medicine specialties (no transitional careers), has recently declined to less than 30%. Peak retention in office based internal medicine for the residents graduating in the early 1990s coincided with return to primary care in physician assistants (by class year and annual surveys of all graduates) and peak family practice choice for the medical school graduates of 1995 � 1998. Internal medicine primary care choice involves the final years of residency and family practice choice involves the final years before the third year medical student match, although the factors involve birth to admission, training, and the health policies and practices present at the time of final decision. Class year analysis reveals these relationships.
In these categorization studies, the family medicine and general practice specialties are combined for a number of reasons. The first reason involves historical comparisons. The major reason for the FPGP combination is coding differences in allopathic, osteopathic, and international graduates. Regardless of allopathic, osteopathic, or international graduates, the family practice or general practice physicians share the highest levels of distribution. For these reasons the two specialties were combined into the FPGP combination. The division is becoming a moot point. By and large the general practice numbers have substantially declined in physicians in recent decades due to credentialing requirements. For the most recent 1987 � 2000 US MD Grads 24,888 of 25,207 or 98.7% of the FPGP physicians were family physicians and only 1.3% were general practice physicians. These were also concentrated in the osteopathic schools that had more retained in osteopathic family practice residency programs.
Family medicine remained separated from all other types of physicians by a consistent 20 percentage points regardless of definitions of physician concentration involving 50, 75, or 100 physicians. With the 75 physician zip code level, family physicians from different types of medical schools, different geographic origins, and different birth county levels also remained grouped in the 35 � 52% range found in zip codes with 75 or more physicians.
Four sets of MCAT scores for each medical school were collected for the 2000 � 2003 class years from medical school web sites. In each class year set of MCAT scores, about 10% of the scores were missing and were interpolated from the surrounding class year values. The 4 year set of score average was ranked and used to categorized allopathic
The development of the birth origins Masterfile, the zip code data base, the county and state databases, the medical school database, specialty databases, and the family medicine databases with ethnicity, race, and gender involved comprehensive reviews of the literature relevant to sociology, economics, education, geography, census data, and history as related to physician workforce and distributions. The studies were greatly facilitated by continued advances in internet search capabilities.
Results
Physician Distribution by Concentrations (PDC) categories can be compared to demographic variables.
Table I. Physician Distribution by Concentrations
|
Concentrations |
Marginal |
Underserved |
|
| |||
|
Super Centers (200+) |
Major Centers (75-199) |
Urban Half Served |
Rural Half Served |
Urban Fourth Served |
Rural Third Served |
Military |
Total |
Practice Zip Codes |
1,117 |
2,231 |
16,020 |
9,312 |
3,955 |
9,391 |
1,816 |
43,842 |
% of Land Area using Total US Land Area |
0.51% |
2.52% |
11.55% |
22.03% |
4.24% |
37.51% |
0.47% |
78.3% |
% Land Area Using Zips with physicians or pop |
0.65% |
3.19% |
14.65% |
27.94% |
5.38% |
47.58% |
0.60% |
100.0% |
US Pop 2000 (millions) |
32.021 |
61.434 |
96.230 |
23.887 |
34.556 |
23.705 |
1.949 |
273.8 |
% By Location |
11.7% |
22.4% |
35.1% |
8.7% |
12.6% |
8.7% |
0.7% |
1 |
Per Square Mile |
1769.83 |
689.56 |
239.20 |
30.87 |
232.77 |
18.31 |
122.99 |
99.8 |
US Pop Poverty (millions) |
3.895 |
6.508 |
7.738 |
2.485 |
8.425 |
4.636 |
0.195 |
33.88 |
% By Location |
11.5% |
19.2% |
22.8% |
7.3% |
24.9% |
13.7% |
0.6% |
100% |
Per Square Mile |
215.28 |
73.05 |
19.23 |
3.21 |
56.75 |
3.58 |
12.301 |
12.36 |
Poverty Percentage |
12.2% |
10.6% |
8.0% |
10.4% |
24.4% |
19.6% |
10.0% |
12.4% |
Poverty to Pop Index |
0.983 |
0.856 |
0.650 |
0.841 |
1.970 |
1.581 |
0.808 |
1 |
Physician to Pop Index |
3.643 |
1.295 |
0.489 |
0.428 |
0.290 |
0.347 |
1.029 |
1 |
Physicians Per Sq Mile |
19.145 |
2.651 |
0.348 |
0.039 |
0.200 |
0.019 |
0.376 |
0.297 |
Active Physicians Per 100,000 Population |
1081.76 |
384.45 |
145.28 |
127.07 |
86.12 |
103.19 |
305.50 |
296.97 |
All Active Physicians (Total Minus Retired) |
346,389 |
236,186 |
139,807 |
30,354 |
29,760 |
24,460 |
5,954 |
813,099 |
42.6% |
29.0% |
17.2% |
3.7% |
3.7% |
3.0% |
0.7% |
100.00% | |
1987 � 2005 Physicians Including Residents |
49.6% |
25.2% |
13.4% |
3.0% |
4.1% |
2.7% |
1.6% |
417,110 |
All Grads of 1987-2000 |
46.1% |
26.5% |
14.5% |
3.6% |
4.2% |
3.3% |
1.7% |
316,511 |
Recent Grads 1987-2000, Classified, Not Residents |
41.4% |
28.6% |
15.5% |
4.3% |
4.5% |
3.7% |
1.9% |
246,573 |
% FPGP of Total Physicians at Location |
7.7% |
15.2% |
25.9% |
38.6% |
23.8% |
36.9% |
21.6% |
16.1% |
FPGP Across Locations |
19.8% |
27.1% |
25.0% |
10.2% |
6.6% |
8.6% |
2.6% |
100.0% |
Not FPGP Across |
45.6% |
28.9% |
13.7% |
3.1% |
4.1% |
2.8% |
1.8% |
100.0% |
FPGP / Not FPGP Ratio |
0.43 |
0.94 |
1.83 |
3.29 |
1.63 |
3.05 |
1.44 |
1.00 |
Super centers had more physicians per square mile (19) than the rural underserved areas had people per square mile (18).
The entire major medical center location grouping included concentrations of 75 or more physicians by combining super and Major center locations. This grouping involved 71% of total
Super Centers and Major Centers represent the top concentrations of physicians and people. About 41% of all graduates and 46% of recent graduate physicians were found in super center zip code locations with 200 or more physicians. This increases to half of total physicians when including residents in training. The super center zip code locations with less than 1% of the land area of the United States had a concentration of 1100 physicians per 100,000 or nearly 4 times the national average of 300.
In the rural underserved (One-Third Served) and urban underserved (One-Fourth Served) areas, physician concentrations were one third the national average and one fourth the national average in physicians per 100,000. Underserved areas had concentrations of poverty above 20% and lower income levels. Physician concentrations doubled from urban underserved (one fourth served) to marginal urban (one half served) locations even though the poverty level was cut from 24% in poverty to 8% and even though income levels increased from far below average to above average.
With urban underserved physician concentrations suppressed to even lower levels than rural underserved and with urban underserved areas in close proximity to super centers and major centers, this introduces a new theme. Physician concentrations to the extreme may play a role in suppressing nearby physician concentrations to even lower levels.
Rural physician concentrations did not increase much from underserved to marginal or Half Served locations. These were locations dominated by family physicians and primary care. Rural areas also had extremes of physician concentration in rural super centers and rural major centers. These rural physician concentrations shared the low percentages of family physicians and primary care physicians found in urban concentrations.
The nation�s rural physicians are about 10% of the total with about one-third found in rural concentrations, rural Half Served, and rural One-Third Served locations. Even isolated locations had rural concentrations such as Cooperstown NY with hundreds of physicians as well as marginal and underserved locations. The defects of coding systems based on county types or Rural Urban Commuting Area codes can be seen in places such as Cooperstown and other concentrations that are better termed Super Centers rather than isolated rural. Also recruitment and retention efforts are different in concentrated physician locations that can attract a wide number of physicians with various origins and backgrounds. The nation�s current deficit primary care policies compromise health care to marginal or underserved locations that are most dependent upon family physicians.
In the super center locations population concentrations were 1700 or 17 times greater than the 100 people per square mile average but super center physician concentrations (19.1) were 64 times greater than the national average for physicians (0.297) in physicians per square mile. The total difference between super center and rural underserved physician concentrations in physicians per square mile was 1000 times greater.
The national standards for underserved location are best seen in recent graduates of 1987 � 2000 without including residents. About 4.5% of physicians were in Urban One-Fourth Served locations. About 3.7% were in Rural One-Third Served locations, again aptly named because these are areas with about one-third of rural physicians and one-third of the land area of the United States. Total national underserved physician levels were also 8.2% or one-third of the 21% of the US population found in zip codes that were coded as underserved. Physicians from lower income rural locations also have one-third the national probability of medical school admission and are three times as likely to be found practicing in rural locations outside of concentrations.
The
Specialty physicians, general internal medicine, and general pediatric physicians concentrate 70% or more into major medical centers. Over 50% of family physicians were found outside of major medical centers, the major factor in enhanced family medicine distribution. This increased to 60% outside for osteopathic family physicians, a good match to the 65% of the population outside. Super center locations had significantly lower concentrations of family physicians at only 6%. Since 50% of physicians are found in super centers, a specialty that escapes this concentration is most able to contribute to distribution. Family physicians reached national averages at major centers with 75 � 199 physicians or levels of 13 � 14%. For all other locations family physicians increased past 20 to 50% or more of total physicians. Percentages of family physicians consistently improved with decreasing levels of income, population, education, physicians, health facilities, and professionals. The concentrations of family physicians remained steady. Concentrations of other physicians increased with concentrations of physicians, income, and population.
Table II. Physician Concentrations and Distributions from the 2005 Masterfile
All Sources |
As a Percentage of Total Physicians |
Distribution Across the Nation |
Active Physicians per 100,000 Population |
| |||||
Locations |
FM |
Office PC |
Non FM GME Slots |
FM GME Slots |
FM |
Office Primary Care |
All Active |
Active with GME |
All |
Concentrations |
|
|
|
|
|
|
|
|
|
Super Center |
6.2% |
28.6% |
64.6% |
42.8% |
49.6 |
230.3 |
804.1 |
1103 |
1191.3 |
Major Center |
11.0% |
35.2% |
20.8% |
29.5% |
35.9 |
114.9 |
326.5 |
409.7 |
434.6 |
Half Served |
|
|
|
|
|
|
|
|
|
Marginal Urban |
17.0% |
40.7% |
10.3% |
18.4% |
22.6 |
54.1 |
132.7 |
168.9 |
181.8 |
Marginal Rural |
28.0% |
46.9% |
0.6% |
1.9% |
37.1 |
62.1 |
132.4 |
162 |
168.3 |
Marginal Isolated |
39.5% |
56.5% |
|
|
32.4 |
46.5 |
82.2 |
112.4 |
116.5 |
Underserved |
|
|
|
|
|
|
|
|
|
Urban Underserved |
16.2% |
42.7% |
2.8% |
4.8% |
11.9 |
31.3 |
73.3 |
88.4 |
95.7 |
Rural Underserved |
26.5% |
50.4% |
0.4% |
1.6% |
30.2 |
57.4 |
114 |
115.3 |
121 |
Isolated Underserved |
37.7% |
62.6% |
|
|
23.3 |
38.7 |
61.8 |
55.3 |
58.7 |
Only Super Center and Major Center locations exceed the 300 physicians per 100,000 that is average for the nation and the Super Center and Major Center locations are the only ones with greater than the recommended 100 primary care physicians per 100,000. All physicians include retired and inactive and other physicians who not uncommonly are found in marginal or underserved locations with lower cost of living, a problem for workforce studies if physicians with active primary practice activities are not specifically selected for analysis.
Graduate medical education positions were compiled using the practice zip codes noted by residents in training in the Masterfile. These were also divided into family practice graduate medical education or other types of training. As with physician practice locations, residency locations in family practice are distributed outside at higher levels although all graduate medical education positions are 86% our more inside of concentrated physician locations.
Does the PDC coding represent actual concentrations of physicians?
In the following, the physician concentrations per 100,000 were compared. Zip codes with greater than 4 standard deviations of difference were combined into the extremes of -4 or +4 S.D.
Table III. Standard Deviations of Physician Concentration
Divisions by Physician Concentration: Standard Deviations for Physicians Per 100,000 at a zip code |
Total Zip Codes |
|
|
Urban Half Served |
Urban Fourth Served |
Rural Third Served |
Rural Half Served |
All Under-Served |
-4 Standard Deviations |
3165 |
0.0% |
0.0% |
40.8% |
46.4% |
7.7% |
4.7% |
54.2% |
-3 Standard Deviations |
3749 |
0.0% |
2.0% |
53.5% |
23.0% |
11.3% |
8.9% |
34.4% |
-2 Standard Deviations |
5533 |
0.0% |
0.8% |
50.0% |
23.6% |
12.9% |
10.8% |
36.5% |
-1 Standard Deviations |
8897 |
0.0% |
6.6% |
52.3% |
13.8% |
13.7% |
12.4% |
27.5% |
0 Standard Deviations |
12492 |
0.6% |
14.8% |
43.6% |
11.0% |
12.9% |
15.3% |
23.8% |
1 Standard Deviations |
19075 |
2.3% |
33.7% |
34.4% |
6.8% |
9.3% |
11.8% |
16.1% |
2 Standard Deviations |
29244 |
6.8% |
59.5% |
18.2% |
3.3% |
4.6% |
6.8% |
7.9% |
3 Standard Deviations |
44638 |
28.3% |
58.3% |
7.4% |
1.5% |
1.2% |
1.9% |
2.7% |
4 or more Std Dev |
88800 |
78.8% |
18.1% |
1.4% |
0.5% |
0.1% |
0.3% |
0.5% |
Unique zip codes |
30980 |
55.0% |
6.7% |
18.5% |
4.7% |
4.2% |
3.3% |
8.9% |
|
246573 |
41.4% |
28.6% |
15.5% |
4.5% |
3.7% |
4.3% |
8.2% |
The super center and major center codes do represent the top concentrations of physicians as measured in physicians per 100,000 population. The underserved locations coded by the PDC do involve the lowest physician density locations. The Unique Zip Codes that had little or no population (created for business or government purposes) required the capture of adjacent zip code populations for determinations of concentrations. Even using adjacent populations, the concentrations of physicians were significantly higher.
Underserved concentrations of physicians were the lowest. This was expected in isolated rural underserved locations, but the same low level of physician concentration in urban underserved locations was a surprise. It is entirely possible that urban underserved concentrations are suppressed by nearby physician concentrations. Counties and zip codes adjacent to urban concentrations of physicians have long been known to have fewer physicians.
Rural health access is complicated by geographic distances and a variety of factors related to lower concentrations of physicians, including specialists who are geographically distant. This is another consequence of concentration of specialists and specialist training in a few major medical center locations. Normally about 75% of rural physicians have urban origins. This is due to the fact that 90% are urban in origin. About 10% of physicians have rural origins and this group becomes 25% of rural physicians. In locations with the top rural physician concentrations, about 50% of the physicians were rural origin and 50% were urban origin. Even in locations with physician concentrations, the nation is dependent upon maintaining rural origin medical school admission. This has been a problem as rural origin medical students have declined from over 25% to less than 8%. Rural males were once 27% of medical school admissions and are now less than 4%. Even rural females have twice the level of medical school admission. This is the same situation facing African American medical school admissions. Both share potential medical student populations that have greater levels of lower and middle income. Decreases in both in admission make it difficult to distribute physicians where they are most needed in urban and rural areas beyond concentrations. Indeed the medical schools that admit the most black males and the most rural males lead the nation in physician distribution outside of concentrations. The principles of physician distribution have been successfully practice for over 100 years by Historically Black schools and osteopathic schools. The same principles pushed 1970s creation allopathic public and osteopathic public medical schools to top levels of distribution. Of course there are many that deny that physicians can be distributed, often pointing to the fact that most black or rural origin physicians do not distribute. They are of course correct when examining the percentages where 80% are found in concentrations, but they are incorrect in that odds ratios of distribution are 2 � 3 times greater. Family practice also doubles or triples distribution. Of course in the top ranking MCAT schools with a tiny fraction of lower and middle income origin physicians, little training outside of concentrations, and the lowest percentages of family physicians and primary care physicians, it is indeed possible to reduce distribution from 28% levels to 16% levels or lower. Increasing parent income levels, increasing MCAT scores, and increasing parents who are professionals and physicians in admitted medical students all indicate decreasing potential for distribution. Record low levels of family practice choice (and in all physician and non-physician forms of primary care) greatly limit distribution. Only those that fail to see beyond concentrations can fail to see the consequences for the 65% of the nation that is outside of physician concentrations.
Table IV. Birth Origin Factors of Older Age at Graduation, Career Choice, School Type, and Location Distribution
|
Marginal |
Underserved |
Concentrations | |||||||
Older Grad |
FP |
School |
Total |
Urban |
Rural |
Urban |
Rural |
|
|
Both |
Y |
Y |
Osteopathic 49-57% Older Grads |
1135 |
25.7% |
8.6% |
7.0% |
15.9% |
27.0% |
13.0% |
40.1% |
N |
Y |
931 |
27.2% |
7.5% |
5.5% |
10.8% |
32.1% |
14.5% |
46.6% | |
Y |
N |
1308 |
21.3% |
6.3% |
5.2% |
8.2% |
32.6% |
24.2% |
56.7% | |
N |
N |
1556 |
17.2% |
5.0% |
5.3% |
6.6% |
34.1% |
29.6% |
63.8% | |
Y |
Y |
Osteopathic 35 � 42% Older Grads |
1160 |
29.2% |
11.9% |
5.9% |
13.2% |
24.1% |
13.5% |
37.7% |
N |
Y |
1683 |
29.1% |
11.9% |
5.4% |
8.7% |
27.6% |
15.6% |
43.2% | |
Y |
N |
1665 |
19.6% |
9.2% |
4.8% |
5.9% |
31.1% |
27.3% |
58.4% | |
N |
N |
3422 |
21.3% |
6.7% |
3.6% |
3.5% |
34.4% |
28.6% |
63.0% | |
Y |
Y |
Osteopathic 24 � 33% Older Grads |
656 |
29.3% |
7.8% |
5.6% |
6.1% |
29.0% |
20.0% |
48.9% |
N |
Y |
1642 |
32.6% |
8.5% |
3.5% |
3.0% |
31.4% |
19.0% |
50.4% | |
Y |
N |
1228 |
19.1% |
5.9% |
5.0% |
3.3% |
33.1% |
31.4% |
64.4% | |
N |
N |
3892 |
20.6% |
4.5% |
2.8% |
1.7% |
35.0% |
33.2% |
68.2% | |
Y |
Y |
Puerto Rico Schools |
39 |
10.3% |
2.6% |
15.4% |
2.6% |
43.6% |
23.1% |
66.7% |
N |
Y |
187 |
11.2% |
1.6% |
18.2% |
11.2% |
32.6% |
23.0% |
55.6% | |
Y |
N |
179 |
9.5% |
1.1% |
14.5% |
3.4% |
31.3% |
36.3% |
67.6% | |
N |
N |
2373 |
8.4% |
1.1% |
15.0% |
4.7% |
32.8% |
37.0% |
69.8% | |
Y |
Y |
Meharry Morehouse Howard |
121 |
23.1% |
5.0% |
14.9% |
5.0% |
28.9% |
18.2% |
47.1% |
N |
Y |
303 |
22.8% |
5.3% |
15.5% |
7.9% |
26.4% |
19.1% |
45.5% | |
Y |
N |
382 |
20.7% |
3.1% |
7.9% |
4.2% |
26.7% |
34.6% |
61.3% | |
N |
N |
1425 |
14.5% |
1.8% |
8.6% |
2.9% |
29.5% |
39.8% |
69.3% | |
Y |
Y |
West Coast Distri-butional |
397 |
20.4% |
7.1% |
7.6% |
7.1% |
31.2% |
25.4% |
56.7% |
N |
Y |
834 |
20.1% |
4.3% |
8.3% |
5.6% |
33.8% |
26.9% |
60.7% | |
Y |
N |
1165 |
11.6% |
3.5% |
5.3% |
2.1% |
29.8% |
46.8% |
76.6% | |
N |
N |
3696 |
10.2% |
1.9% |
4.1% |
1.4% |
30.5% |
51.1% |
81.6% | |
Y |
Y |
Allopathic MCAT 8.5-9.25 |
1060 |
20.9% |
13.3% |
7.7% |
17.5% |
24.5% |
14.5% |
39.1% |
N |
Y |
2559 |
24.7% |
13.9% |
6.5% |
12.3% |
25.9% |
15.0% |
40.9% | |
Y |
N |
3380 |
13.7% |
5.0% |
5.9% |
6.1% |
31.1% |
36.7% |
67.8% | |
N |
N |
11080 |
12.6% |
4.1% |
5.1% |
4.6% |
32.5% |
39.7% |
72.2% | |
Y |
Y |
AllopathicMCAT 9.25-9.5 |
1081 |
24.2% |
11.7% |
8.5% |
10.9% |
23.2% |
19.7% |
42.9% |
N |
Y |
3050 |
23.3% |
8.6% |
6.2% |
11.5% |
27.8% |
20.0% |
47.7% | |
Y |
N |
4470 |
14.4% |
4.5% |
4.8% |
3.9% |
27.0% |
42.7% |
69.7% | |
N |
N |
16277 |
12.9% |
2.9% |
4.0% |
3.2% |
29.4% |
45.8% |
75.2% | |
Y |
Y |
AllopathicMCAT 9.5-10 |
1825 |
23.3% |
12.3% |
6.4% |
8.9% |
24.2% |
22.3% |
46.5% |
N |
Y |
5432 |
26.5% |
10.8% |
5.4% |
6.5% |
27.6% |
20.6% |
48.3% | |
Y |
N |
8841 |
14.6% |
4.0% |
4.7% |
3.1% |
29.4% |
42.2% |
71.6% | |
N |
N |
38523 |
13.8% |
2.7% |
3.2% |
1.6% |
29.7% |
47.1% |
76.8% | |
Y |
Y |
AllopathicMCAT 10-10.5 |
1256 |
24.5% |
13.6% |
6.1% |
7.1% |
25.5% |
21.3% |
46.7% |
N |
Y |
3780 |
26.7% |
14.0% |
4.9% |
4.5% |
26.6% |
20.8% |
47.4% | |
Y |
N |
6241 |
13.9% |
3.9% |
3.1% |
1.8% |
28.7% |
46.9% |
75.5% | |
N |
N |
27768 |
12.3% |
2.6% |
2.9% |
1.2% |
28.9% |
50.2% |
79.1% | |
Y |
Y |
AllopathicMCAT 10.5-12 |
547 |
20.8% |
7.9% |
6.8% |
6.2% |
29.3% |
26.9% |
56.1% |
N |
Y |
1829 |
21.1% |
7.8% |
7.9% |
5.4% |
28.4% |
26.9% |
55.3% | |
Y |
N |
4896 |
10.8% |
2.0% |
3.1% |
1.7% |
22.3% |
59.0% |
81.3% | |
N |
N |
28286 |
9.5% |
1.6% |
2.6% |
1.0% |
24.2% |
59.6% |
83.8% | |
Y |
Y |
Allo Early Admit |
31 |
25.8% |
16.1% |
6.5% |
9.7% |
32.3% |
6.5% |
38.7% |
N |
Y |
261 |
31.8% |
6.9% |
5.0% |
8.8% |
27.6% |
18.0% |
45.6% | |
Y |
N |
151 |
19.2% |
6.6% |
4.6% |
3.3% |
30.5% |
33.8% |
64.2% | |
N |
N |
1754 |
14.7% |
3.0% |
3.2% |
1.4% |
30.4% |
46.3% |
76.7% |
The older graduates and those choosing family practice are consistently more likely to distribute outside of major medical center concentrations. The same characteristics are closely associated with distributional medical schools and top levels of primary care contributions. Narrow admissions in age and MCAT scores, narrow training focus, and the nation�s current health policies act to further concentrate physicians. MCAT scores and younger age admission are not necessarily the focus of medical schools, but higher scores and younger ages are an indication of narrowing in origins, narrowing in distribution, and declining primary care contributions.
Medical schools vary across a continuum that can be ranked by scores, rural origins, birth in a city or county with a medical school (proxy for major medical center origin), birth county income (1969 per capita), and career choice. Urban and rural served locations are not shown.
Table V.
Also Medical School Type and Career Choice and Most Needed Health Access
Most Recent Graduates 1987 � 2000 with a Classification, Not Residents in Training
|
|
|
Underserved Practice Location |
Concentrations in Practice |
Birth Origins | ||||
Type or Location of School and Percent Found in Office Primary Care |
Career Choice and FPGP % |
Total |
Urban Under-served |
Rural Under-served |
Major Center |
|
Bottom Quart Birth County Income |
Rural Birth |
Birth in Med School City/ County |
Puerto Rican 30.9% |
Not FPGP |
2,121 |
14.2% |
5.0% |
33.2% |
35.5% |
0.9% |
0.3% |
75.4% |
FPGP 13.6% |
335 |
22.4% |
10.1% |
33.1% |
18.2% |
1.2% |
0.8% |
72.2% | |
Historically Black 44.2% |
Not FPGP |
1,761 |
8.0% |
3.4% |
29.1% |
39.9% |
11.8% |
6.6% |
73.1% |
FPGP 19.7% |
433 |
15.9% |
7.9% |
28.2% |
17.8% |
19.0% |
11.4% |
68.6% | |
Early Admission 32.6% |
Not FPGP |
1,957 |
3.4% |
1.5% |
30.0% |
45.7% |
21.6% |
13.1% |
65.2% |
FPGP 14.7% |
336 |
4.5% |
8.3% |
28.9% |
17.6% |
20.5% |
20.5% |
56.0% | |
West Coast Distributional 42.1% |
Not FPGP |
4,887 |
4.5% |
1.6% |
30.6% |
49.7% |
5.6% |
7.5% |
73.4% |
FPGP 21.6% |
1,343 |
8.0% |
6.0% |
33.1% |
26.8% |
7.9% |
11.1% |
64.8% | |
MCAT 10.5-12 27.4% |
Not FPGP |
32,153 |
2.8% |
1.1% |
24.4% |
58.8% |
8.2% |
6.7% |
77.2% |
FPGP 7.6% |
2,626 |
7.8% |
5.6% |
28.0% |
27.0% |
12.6% |
11.7% |
68.1% | |
MCAT 10-10.5 33.9% |
Not FPGP |
34,001 |
2.9% |
1.3% |
29.2% |
49.1% |
9.1% |
9.5% |
71.3% |
FPGP 14.2% |
5,622 |
5.5% |
5.4% |
26.8% |
20.9% |
14.4% |
18.7% |
59.7% | |
MCAT 9.5-10 34.6% |
Not FPGP |
47,512 |
3.5% |
2.0% |
29.8% |
45.8% |
9.8% |
9.3% |
71.4% |
FPGP 14.5% |
8,045 |
5.6% |
7.4% |
26.5% |
20.9% |
15.0% |
15.9% |
59.8% | |
MCAT 9.25-9.5 37.3% |
Not FPGP |
20,577 |
4.0% |
3.5% |
29.3% |
44.4% |
19.0% |
14.0% |
64.4% |
FPGP 17.8% |
4,471 |
7.0% |
11.6% |
26.0% |
20.0% |
24.5% |
19.0% |
56.1% | |
MCAT 8.5-9.25 41.0% |
Not FPGP |
14,723 |
5.3% |
5.1% |
32.5% |
38.5% |
28.8% |
21.7% |
53.6% |
FPGP 21.9% |
4,139 |
6.8% |
14.3% |
25.4% |
15.0% |
36.2% |
32.1% |
44.6% | |
Uniformed Services 14.4% |
Not FPGP |
1,515 |
3.2% |
2.1% |
18.3% |
18.9% |
12.7% |
11.4% |
64.0% |
FPGP 21.2% |
408 |
2.7% |
6.4% |
15.2% |
9.6% |
16.7% |
14.7% |
58.2% | |
Osteopathic Low MCAT 48.2% |
Not FPGP |
5,278 |
4.1% |
4.8% |
34.9% |
28.3% |
13.1% |
12.5% |
65.9% |
FPGP 37.1% |
3,111 |
5.7% |
10.3% |
29.7% |
16.4% |
17.5% |
17.9% |
60.9% | |
Osteopathic High MCAT 46.0% |
Not FPGP |
8,454 |
3.8% |
3.3% |
32.6% |
32.0% |
12.4% |
12.6% |
67.1% |
FPGP 35.0% |
4,556 |
5.4% |
8.3% |
26.8% |
16.9% |
16.7% |
19.3% |
58.1% | |
Canadian 26.7% |
Not FPGP |
2,337 |
2.1% |
3.2% |
15.2% |
39.3% |
1.0% |
1.2% |
75.8% |
FPGP 22.9% |
693 |
3.6% |
7.6% |
22.9% |
16.0% |
1.1% |
1.6% |
66.0% | |
Central American 53.2% |
Not FPGP |
2,167 |
11.2% |
4.2% |
30.5% |
36.2% |
4.1% |
1.6% |
76.6% |
FPGP 22.6% |
631 |
20.1% |
11.4% |
26.3% |
19.5% |
9.3% |
5.5% |
64.6% | |
42.1% |
Not FPGP |
480 |
6.5% |
1.0% |
29.0% |
46.5% |
0.0% |
0.0% |
86.4% |
FPGP 7.3% |
38 |
21.1% |
0.0% |
34.2% |
26.3% |
0.0% |
0.0% |
93.3% | |
53.3% |
Not FPGP |
7,211 |
5.6% |
5.2% |
31.0% |
38.2% |
0.1% |
0.1% |
75.7% |
FPGP 6.8% |
522 |
5.7% |
7.3% |
27.8% |
23.9% |
0.0% |
0.0% |
73.6% | |
Distant International 40.0% |
Not FPGP |
11,283 |
5.7% |
4.7% |
25.9% |
47.7% |
0.4% |
0.5% |
75.8% |
FPGP 7.2% |
874 |
9.2% |
7.2% |
29.5% |
27.9% |
0.5% |
0.9% |
71.6% | |
60.3% |
Not FPGP |
685 |
10.8% |
5.3% |
27.4% |
34.3% |
0.0% |
0.0% |
72.3% |
FPGP 9.3% |
70 |
18.6% |
11.4% |
22.9% |
24.3% |
0.0% |
0.0% |
60.6% | |
The |
Not FPGP |
2,254 |
8.4% |
11.4% |
30.9% |
26.3% |
0.7% |
0.3% |
80.1% |
FPGP 12.7% |
329 |
8.2% |
8.5% |
25.8% |
17.3% |
1.3% |
0.6% |
79.1% | |
44.6% |
Not FPGP |
2,711 |
6.3% |
11.2% |
29.5% |
32.0% |
0.0% |
0.1% |
89.6% |
FPGP 5.2% |
150 |
7.3% |
8.0% |
30.7% |
18.0% |
0.0% |
0.0% |
92.5% | |
59.5% |
Not FPGP |
2,924 |
4.1% |
3.5% |
32.6% |
37.0% |
7.1% |
7.0% |
74.2% |
FPGP 22.5% |
850 |
5.4% |
8.4% |
29.9% |
19.6% |
12.6% |
11.8% |
64.0% | |
All 37.2% |
Not FPGP |
206991 |
4.1% |
2.8% |
28.9% |
45.6% |
11.3% |
9.7% |
70.4% |
FPGP 16.1% |
39,582 |
6.6% |
8.6% |
27.1% |
19.8% |
17.9% |
17.7% |
58.9% |
About 70% of osteopathic and 50% of international graduates have birth origins in the Masterfile. Birth origins were calculated as the percentage of a known origin compared to those with all known origins. Uniformed Services office based levels are low compared to hospital based (military hospital) location.
Family physicians are consistently found in rural underserved, rural, and urban underserved locations at greater levels, typically double or triple the percentages of other types of graduates from the same types of medical schools. This level of distribution is consistent dating back across 30 years of graduates. The very few who manage to choose family practice in the highest ranking MCAT medical schools have much greater distribution than graduates not choosing family practice. In these schools, family physicians appear to be the only hope for distribution.
Family physicians are consistently distant from physician concentrations in origins, training, and practice locations.
Medical schools that admit more students with origins involving concentrations such as cities or counties with medical schools concentrate physicians at the highest levels, distribute at the lowest levels, and graduate the fewest family physicians, primary care physicians. Concentrations of Medical College Admission scores also result in the same outcomes.
The cities and counties with medical schools dominant admission with nearly 70% of those born in the United States or in other nations. Even the segment with missing birth origins has characteristics that indicate origins involving concentrations rather than a mix or a origins outside. Those with missing birth data are also likely to have concentrated origins.
Birth in a medical school county indicates lower distribution and also lower choice of family practice (12% versus 19%). Ratios of medical school admission were higher for those born inside and were lower for those born outside. Those most likely to gain admission are least likely to distribute.
Those least likely to become family physicians are being admitted at higher levels, replacing those most likely to become family physicians, primary care physicians, rural physicians, underserved physicians, women�s health physicians, and physicians who care for the elderly. This is because 65% of the population is outside of concentrations and even higher levels of older Americans move away from the highest cost (living, health care, more) locations to more reasonable cost areas of the nation, typically lower and middle income. Higher cost areas demand higher income populations to support these costs. Movements of the middle and lower income population away from high cost of living areas have been dramatic in some counties.
Medical schools that admit the most exclusive students by score rankings concentrate those with medical school city or county origins, most urban origins, and highest income origins. Allopathic private schools supply on average about 33% of total physicians but only admit 20% of the total born in the lowest income counties and admit 50% of the total admitted from the highest income and most urban counties. As origins increase in concentrations of income and population density, elite allopathic private schools take an increasing share of medical students. These are admission locations associated with decreasing levels of family practice, primary care, rural, and underserved locations.
Elite medical schools are more likely to have younger medical students, more born in higher income counties, more born in the most urban counties, and fewer born in rural or lower income counties. Medical schools selecting the most exclusive students concentrate the most into major medical center locations and graduate the fewest for rural, underserved, family medicine, and primary care careers.
Medical schools admitting a wider distribution of medical student types distribute more physicians. They also admit older medical students which is another indicator of a broad admissions process. Even the average medical schools distribute physicians above national averages because the upper crust medical schools concentrate physicians at such high levels. Outstanding distribution beyond concentrations can be found in the Historically Black, osteopathic, and lower scoring allopathic medical schools. These are school that focus on the characteristics of the students, not their scores. This involves special mission, detailed admission considerations, and differences in training. There is also a self selection process of medical students. Those with higher MCAT scores can and do self select higher ranking MCAT medical schools in the hopes of a better subspecialty placement (also with higher income). This tends to concentrate practice locations in the elite schools, leaving distributional schools with graduates more likely to distribute.
Historically Black, osteopathic, and lower scoring allopathic medical schools admit higher levels of physicians with origins �outside� of concentrations. These include rural origin, lower income origin, and older medical students. Older medical students are most often those who were delayed in admission by barriers of income and education.
Distributional medical schools distribute physicians outside of major medical centers at the highest levels by admitting differently, training differently, graduating more family physicians, graduating more into primary care who remain in primary care, and distribute the most to underserved locations.
Birth origins fields use the city and state location. Birth origins represent proxy socioeconomic and geographic origin markers. Unfortunately in urban locations the medical students cannot easily be divided into those born and raised in concentrations and those born �outside.� Because Hispanic and African American levels of concentration (income, education, admission probability) are lower, use of race and ethnicity can add a marker of �outside� origins to examine distributions within counties. The Historically Black medical schools have predominantly black medical students and provide useful information regarding career and location choice. In addition birth origins can be mapped to counties with a majority of African Americans and these studies have verified the findings. For example in the past decade the decline in family practice choice in Historically Black medical schools as well as predominantly black counties has been 70%, a level greater than the 50% average decline. The doubling impact of family practice on distribution to marginal and underserved locations will be greatly missed by these practice locations. The smaller percentage of the population in concentrations will benefit while the 65% of the population outside and the even higher levels of Hispanic and African American and elderly and underserved and marginal and rural populations �outside� will not benefit. The combination of fewer African American and Hispanic medical students admitted, delays in admission as graduates move to fewer medical schools who will admit them at higher levels, exchanges of more African American and Hispanic graduates of higher income origins replacing those of lower and middle income origins, fewer family practice graduates, fewer remaining in office based primary care, declines in Medicaid support, and cuts in Medicare support will be devastating to populations outside in the most need of health care and physicians who match up in race, ethnicity, origins, socioeconomic levels, and understanding.
Physician Concentrations By State
A test of the PDC coding can be performed at the state level. These studies include physicians found in administration, hospital, medical teaching, not classified, office based, other, research, or locum tenens primary practice activities. Residents were listed separately. Inactive or retired physicians were excluded.
Table VI. State Distributions of Physicians
Physicians Per 100,000 |
Concentrations |
Marginal |
Underserved | |||||
Residents in 2005 |
All Active Physicians |
Super Center |
Major Center |
Urban |
Rural |
Urban |
Rural | |
200 to 600 Super Center and Major Center Physicians per 100,000 Population in the State or District |
||||||||
DC |
210.6 |
764.8 |
1472.2 |
426.4 |
140.0 |
|
344.9 |
|
MD |
59.2 |
443.7 |
1108.8 |
358.9 |
155.8 |
133.5 |
106.2 |
122.0 |
MA |
87.8 |
461.3 |
1903.9 |
505.5 |
174.9 |
192.7 |
80.4 |
212.7 |
NY |
83.2 |
408.5 |
991.6 |
363.2 |
163.0 |
123.5 |
72.4 |
143.1 |
RI |
67.5 |
436.9 |
1059.2 |
366.3 |
175.9 |
|
92.4 |
|
CT |
56.2 |
380.0 |
1014.4 |
420.9 |
200.7 |
196.7 |
80.1 |
278.9 |
HI |
30.3 |
328.9 |
651.8 |
317.5 |
124.7 |
203.0 |
55.3 |
43.3 |
PA |
60.4 |
345.5 |
1212.7 |
390.5 |
154.2 |
107.2 |
101.3 |
128.7 |
NJ |
35.3 |
347.6 |
879.0 |
410.8 |
202.4 |
187.1 |
75.1 |
137.8 |
IL |
55.9 |
294.1 |
909.2 |
350.9 |
118.3 |
101.2 |
59.3 |
98.5 |
DE |
33.7 |
297.0 |
1099.1 |
418.6 |
146.5 |
139.6 |
145.7 |
|
150 � 200 Super Center and Major Center Physicians per 100,000 Population in the State |
||||||||
CA |
29.2 |
283.3 |
954.7 |
322.0 |
137.1 |
168.3 |
70.0 |
89.8 |
MI |
48.4 |
291.3 |
1117.8 |
349.8 |
125.4 |
135.5 |
62.0 |
91.8 |
OH |
51.1 |
290.4 |
1133.5 |
381.8 |
133.8 |
100.8 |
87.7 |
83.2 |
VA |
38.0 |
296.7 |
873.8 |
386.2 |
152.0 |
141.0 |
114.7 |
143.3 |
VT |
59.0 |
372.6 |
1549.7 |
641.2 |
327.1 |
196.0 |
130.1 | |
MN |
46.5 |
295.2 |
1656.1 |
404.4 |
144.8 |
119.2 |
107.4 |
94.5 |
OR |
24.5 |
288.8 |
1015.0 |
417.0 |
129.3 |
131.7 |
97.3 |
147.4 |
MO |
45.2 |
280.8 |
1188.9 |
379.4 |
110.2 |
122.2 |
88.1 |
85.0 |
CO |
26.9 |
289.1 |
1234.2 |
355.6 |
147.8 |
193.0 |
104.0 |
128.5 |
TN |
35.2 |
274.9 |
1139.4 |
408.8 |
138.8 |
99.6 |
63.8 |
88.7 |
LA |
44.7 |
272.9 |
1170.8 |
403.2 |
136.2 |
0.0 |
96.7 |
92.4 |
WA |
24.6 |
284.4 |
1207.8 |
411.5 |
140.0 |
149.4 |
98.0 |
118.9 |
FL |
18.4 |
287.9 |
997.5 |
419.3 |
163.2 |
187.9 |
87.5 |
118.9 |
NC |
36.7 |
269.7 |
1136.0 |
403.8 |
124.2 |
128.1 |
60.0 |
104.5 |
AZ |
23.1 |
256.1 |
816.1 |
312.3 |
142.2 |
167.2 |
58.2 |
118.1 |
WI |
32.2 |
270.9 |
1125.3 |
358.0 |
155.0 |
130.5 |
76.9 |
67.1 |
NM |
31.2 |
253.7 |
667.8 |
386.0 |
291.7 |
259.0 |
66.0 |
113.7 |
GA |
26.3 |
245.8 |
1137.0 |
342.1 |
117.7 |
116.9 |
85.7 |
101.2 |
TX |
32.7 |
240.8 |
1278.8 |
357.7 |
124.7 |
111.1 |
70.7 |
85.1 |
WV |
33.7 |
259.1 |
1164.6 |
416.7 |
147.7 |
105.5 |
139.0 |
117.9 |
|
Less than 150 Super and Major Center Physicians per 100,000 population in the State |
|||||||
SC |
31.7 |
246.7 |
847.1 |
493.5 |
116.4 |
160.6 |
112.2 |
89.9 |
ME |
22.7 |
300.8 |
1447.1 |
483.8 |
183.5 |
183.2 |
198.1 |
203.6 |
UT |
33.5 |
232.9 |
853.3 |
378.5 |
102.4 |
121.5 |
98.5 |
88.7 |
NE |
41.1 |
255.5 |
2561.6 |
435.5 |
190.6 |
119.5 |
50.6 |
67.8 |
IN |
22.9 |
235.0 |
1063.6 |
378.2 |
137.1 |
91.0 |
107.0 |
67.3 |
KY |
28.9 |
240.5 |
1352.4 |
426.8 |
131.5 |
119.5 |
94.6 |
97.7 |
NV |
10.5 |
225.4 |
565.7 |
500.9 |
121.8 |
153.3 |
56.5 |
61.3 |
NH |
23.8 |
282.8 |
4249.4 |
541.3 |
138.8 |