Why Physician Workforce Research Needs New Tools
Robert C. Bowman, M.D.
Research is an important component of problem solving.
Reasons for New Tools
Primary Care Workforce Research
Tools are needed that
The Standard Primary Care Measuring Tool
Flexible Primary Care Versus Permanent Primary Care
The Physician Distribution by Concentration Coding Tool
The Current Status of Health Access and Health Workforce Studies
Paradoxically much of what contributes to health access is known. Of course research can be scientifically correct and involve entire populations, and still not be implemented. Unlike brain tumor children or HIV patients, there is not the same attitude regarding “breaking the research code” in a double blinded study with an immediate implementation of a successful intervention. This is especially true when the intervention that works involves changes in career choices and practice locations from the status quo.
Important and critical need to know information is not being presented to those who make the final decisions regarding health care for the nation. The defects involve
The nation has had a decidedly limited analysis of the consequences of admitting more and more children of concentration - or more correctly more and more children with origins associated with combinations of concentration. While the most urban children and the highest income children and children of professionals all have advantages with top scores, there are grave limitations in these students when it comes to health access careers.
The most exclusive children that have rarely been exposed to “normal” or less than top concentrations and that have been trained in top concentrations are not likely to embrace health access careers, careers serving lower and middle income people, or career serving rural or Native people. Most exclusive and less normal is not a good plan for health access.
Regardless of specialty, physicians must have superior people skills, people orientation, and the ability to work with the widest variety of peoples in the widest variety of roles, from colleagues to employees to employers to leaders at all levels to those who provide the basic infrastructure for health care facilities.
Many if not most of the difficulties above are directly or indirectly related to a focus in American health care on major medical center forms of health care and poor understanding of the needs of the 65% of the population found outside of major medical centers, largely those in lower and middle income America.
American Medical Association (AMA) and Association of American Medical Colleges (AAMC) databases can track medical students from birth throughout training and all the way to retirement. The quantity and range of data is not a problem. The various data sources have not been sufficiently focused on the major areas of physician workforce: costs, quality, and access. When reviewing the current status of physician workforce research, a number of themes emerge.
Convenience Samples and Convenient Explanations
Most workforce research focuses on the convenience sample provided by medical students and residents who are located in major medical centers (medical schools or zip codes with 75 or more physicians).1 With 70% of physicians and nearly 100% of medical students and residents trained in major medical centers it is difficult not to view physician workforce through the major medical center lens.
Medical students represent a narrow subset of society, particularly in the elite medical schools where most research originates. There is little variation across the narrow range of students in this group. Medical school variations are often even less within medical schools. Each medical school is consistent from year to year in student ages, origins, scores, and other characteristics. When considering medical students across the wide range of all medical schools and over a number of years, there is much more total variation to consider. This also includes sufficient numbers of medical students in important subgroups (minority, rural, lower income, socially or otherwise different) that are very different in characteristics and outcomes. Research of this nature takes a coordinated effort across many types of medical schools. It is often easier to do a national sample than to gather multiple schools together, but national data access is limited.
Without considering influences before and after medical training, it is also far too easy for medical educators to claim credit for the positive outcomes and deflect blame for the problem areas. It may be in vogue to consider the more challenging career choices as victims of “controllable lifestyle” but alternative hypotheses should be considered even if they challenge medical education. The best evidence for this is provided by consensus efforts involving medical education regarding the quality of medical education.2 Medical education faces major problems with decisions made that will only make matters worse.
Without researchers and associations willing to explore all possibilities and perspectives, the research will be limited.
The major medical center perspective may also limit the ability of medical leaders to grasp different points of view. Understanding rural, lower income, diverse, or even middle income origin patients or medical students may be difficult for medical education leaders that hail from a small and exclusive subset of 146 top American colleges where 74% came from the top income quartile,3 from the elite set of American medical schools training most researchers and leaders, and from an even more exclusive subset of American parents. All too commonly medical education leaders have attempted to deflect blame or limit physician distribution to a limited small subset4, 5 even when their own colleagues attempt to encourage them into a “Season of Accountability and Social Responsibility.”6 Distributional medical education has remained stagnant since this 1991 call to action and has been forced into retreat in the past decade for both rural and underserved areas.
Medical education has yet to carefully explore the admissions changes that have been forced upon them by influential parents and a few individuals who have used their expertise in government, legal areas, and mass communication to change admissions. Massive changes in admissions in the last ten and twenty year periods have gone largely unreported and the changes have all involved lower probability of distribution.7
Another problem involves the numbers and concentrations of students in elite medical schools. When considering rural, low income, older, and diverse medical students; the lowest levels are found in the elite medical schools. It takes many years of graduates of elite medical schools to get a suitable comparison group. Even the rural, low income origin, and inner city origin medical students at elite medical schools have higher scores and slightly lower levels of distribution when compared to other medical schools. When medical students are representative of the
Studies also tend to blame poor inner city and rural physician distribution on lifestyle issues, security issues, poor economics in rural or underserved areas, or the paucity of cultural outlets. The lack of a substantial commitment to quality of care for rural and underserved peoples when temporary rookie physicians are seen as “better than none.” In the support of National Health Service Corps and more support for medical school tuition, statements are made such as “We need to de-emphasize the necessity for permanent physician settlers in areas unattractive to most physicians. I believe, along with many others, that we can solve the geographic maldistribution problem by placing fully trained
Rural and underserved areas have no less lifestyle, culture, or recreational opportunity compared to urban areas. Many of the cultural, social, and recreational opportunities available in rural and underserved areas are flatly missing in major urban areas. Lifestyle differences are merely different, not superior or inferior. There are also professionals who prefer a less organized location where individuals make more of a difference, a place with more social interaction, and a place that feels more like “home.” These professionals are not likely to be found in east coast major metropolitan areas. It is difficult for those who have not experienced different peoples, different colleagues, any of the different rural lifestyles, or training outside of major medical centers to comprehend this.
There is an alternative explanation for the relationship of poor economics to lack of physicians. Health care is a significant contributor to the economies of rural and underserved locations. Failure to admit medical students that will distribute to rural or to underserved areas may actually be a cause of the poor economics blamed for poor physician distribution. In rural and underserved areas the impacts of the health care and education sectors can involve nearly half of some economies.
Attempts to explain career choice without carefully considering the 30 years before training, the impacts of health policy during and after training, and the student perceptions of a planned future career are likely to be flawed. Lack of understanding regarding the influences from birth to medical school has made it far too easy to attribute admissions, career, and location decisions to medical school and residency factors such as curricula, residency location, and medical school mission. With a single dimension perspective, much can be missed.
There are different perspectives to consider. These include the perspectives of rural or underserved peoples, the perspectives of child development and education, the perspective of primary care, perspectives involving time dimensions and health policy effect, and the impact of the first decades of life on career choice. Studies fail to consider the perspectives of those geographically and socially different. A simple way to describe this involves a contrast between major medical center populations, medical students, forms of training, and health policy and the contrasts between those who are least associated with major medical centers by origin, location, income, status, geography, culture, language, and other factors. Major medical centers are seen as the solution even when better distribution of health care is about distributing resources, physicians, and opportunity away from major medical centers.
The limited perspectives can even involve primary care. For example, primary care generalists are considered a factor in lower health care costs and higher quality in studies comparing states and nations.8 This is a limited perspective claiming credit for cost and quality correlations, but numerous other factors share the same correlations. The fact is that some states and nations invest in children and distribute child development, education, and opportunity. Other states concentrate income, opportunity, education, and other resources. The states that concentrate resources deprive lower and middle income populations. Children raised in these environments have an impaired start in life and make poorer decisions in education, jobs, and life. They also make poor decisions in health care. All of these are more costly for states. States that distribute child development, education, and opportunity have fewer gaps and citizens that make better decisions, and a more efficient society with alternatives more likely to but the same states and nations with advantages in health outcomes also do better in areas such as child development and education.9 By visualizing workforce, health policy, and medical education from a single perspective, the most important solutions for health care access, cost, and quality fail to gain even simple consideration.
The influences of parents may be outstanding as in the devotion of some lower income parent(s) who clearly change the life course of their child. Or they may be less than flattering when a pattern of admission suggests too much influence regarding the children of professionals, physicians, alumni, or wealthy parents. The influences of medical schools and programs can also be outstanding or less than flattering. Without appropriate and objective studies, neither of these areas will gain attention.
There are also those that perform better than predicted despite lower MCAT scores10 (veloski) and those who distribute at levels higher that predicted given the most urban origins and younger ages at medical school entry. The fact that this is one of the few studies to involve more than one medical school is testament to the lack of workforce research coordination and the potential of having greater numbers and more variation in students.
In an article in the Chronicles of Higher Education, Anthony Carnevale, a former vice-president of ETS, claimed that the testing service terminated a promising line of research regarding “strivers” or those with better performance than predicted based on scores and other characteristics. Those in charge of the testing deny these charges. Although there is no clear evidence that such research is truly promising or that it has been squelched at the medical education level, research has not been innovative either. Research relating higher MCAT scores to poorer distribution has been considered not worthy of exploration by the Vice President of AAMC in charge of the MCAT. Dr. Ellen Julian sees her task as preparing a test that best reflects performance in the first two years of medical school and meets the needs of medical schools. Beyond valid performance predictions, the next priority is consideration of the colleges and the students themselves. There is no question that maintaining the validity of a test in the age of instant information is a major challenge. However the focus of the test is narrow and getting ever more narrow.11 The usual performance measurements involve correlations with USMLE 1, another standardized test.12, 13 Those who do well on one standardized test are likely to do well on others. However there are those who manage to do less well than expected on the USMLE 1 compared to MCAT, and these are the Asian students who have the highest probability of admission.7, 10 It is entirely possible that those who do best on the MCAT with a narrow range of characteristics focused on academics and sciences may have limitations as medical school demands a more mature and clinically oriented graduate. Dr. Julian has also stated that an element of speededness is needed to make the MCAT valid for predicting medical school performance. What this means is that the MCAT depends upon differences in student ability to rapidly process multiple choice test questions. Those most likely to move more slowly through tests are students who are the most different in background, culture, income, social distance, and geographic origin. These are students that did not have every advantage of income and education. When given more time regarding the MCAT test, student performance levels out. Admission to medical school may represent the first level playing field many disadvantaged students have ever had.
Physician workforce studies also fail to encompass a very relevant and ignored area of all professional education, the concepts of eliteness or exclusiveness. Rarely do studies consider the impact of standardized test scores, elite college status, AOA status, and elite medical school status that exert great influence on eventual career choice, especially regarding areas such as future physician quality (high or low) and physician distribution. Those found to have the least satisfaction levels as physicians14 appear not to be a consistent pattern except the specialties with the least satisfaction are also those that tend to have AOA status, the highest board scores, the most exclusive selections.15 They are also more likely to come from schools with the highest MCAT scores and there is little doubt that they themselves had the highest MCAT scores.
Seen through the major medical center lens, those choosing the people oriented careers may be interpreted as having less ability, especially from the perspective of those who place far too much emphasis on scores. Another explanation for career choice and lower scores is different origins (associated with lower scores), a different focus on serving patients directly, and anticipation of a different career and location compared to most medical students. Students can be seen as moving two ways. Both groups pursue opportunities but the pursuit is different. One group pursues opportunity and the other group pursues opportunity for service.
The highest MCAT scores and physician distribution seem to be incompatible. The schools with the highest scoring students have the lowest levels of distribution, primary care, and family medicine. The students with the highest MCAT scores have the highest board scores, they are most likely to have AOA status, and they choose careers that are least likely to be found outside of major medical centers. Students with lower MCAT scores are the most different, the least connected to major medical centers, the least likely to be admitted, and the most likely to distribute. Those with the highest scores, the highest income and most urban origins (birth county), and the youngest ages are the most likely to concentrate in major medical centers.
Important studies in these and other areas will not proceed with the current limited perspective. Other perspectives should be considered beyond physicians, types or physicians, the major medical center perspective, or even the limited perspective of those immersed in health care alone. The 3000 by 2000 efforts ended in frustration with the major concern being the lack of qualified minority candidates.16 Earlier efforts have been successful at the high school level17 or at the high school and college level. These efforts have improved admissions and decreased attrition rates.18 However the lack of earlier considerations in middle school, elementary school, and child development hamper the research and the potential to change admissions and health care in the nation. When the perspective is the highest standardized test scores, the exclusive major medical center types of students can be tracked back through college testing, advanced placement, and the earliest elementary testing. Child development is a huge advantage for those of the top socioeconomic level.19 Those left behind are left out, as are the populations in most need of professionals.
There are perspectives that can address a number of areas related to physician distribution across the wide range of physician characteristics and all outcomes. Birth origins are available for well over 90% of US MD Grads. These origins can be linked to birth county characteristics such as income, education, or geography. Birth origin studies can compare across a national population of student or focus on a selected subgroup. Birth origin studies proceed in the natural progression of choices made by medical students and shaped by parents.
Birth origin studies encompass elements related to parent factors, education, concentrations of professionals, social organization, and geographic distance. Role modeling may be reflected in the types of locations where students are born and raised or it may be a function of social class. The lifestyle preferences expressed by students may not be controlled or uncontrolled, but may represent student choices to maintain the major medical center lifestyles that most have had since birth. Other students may prefer the lifestyles found beyond major medical centers and make the career choices that facilitate this location. The apparent success of medical schools with missions for rural health or for the underserved may be due to the strong influences of admissions, location factors, and prevailing career choices. Public medical schools in Midwestern and western states have admissions, environmental influences, and workforce factors that shape future rural physicians and family physicians. Underserved origin medical students concentrated in certain schools may allow these schools to effectively address specific missions.
The broader perspective of birth origin studies can also help answer the riddle of increased male rural distribution patterns. With better understanding regarding distributions of education, admissions probabilities, and gender differences it is possible to explain why some medical schools graduate more females into rural family practice compared to males even among a national group where most schools maintain a male advantage. In some states, females are the major group that overcome the barriers of education and income found in rural areas and those returning to rural locations are female. Distribution back to Native reservations would be improved, except that females are the only ones contributing. Native males have been less able to navigate the difficult path to college, medical school, and family medicine residency. Regarding distribution, gender may be far less important than birth origins, education, income, and career opportunity. These are reflected in standardized test scores and age at medical school admission. Actual performance as a physician may not be related to scores in those most different, but poor distribution is related to higher scores.
One thing is clear, without reasonable controls on the trainees regarding the contributions of specific characteristics involving origins and age and other factors, important outcomes such as physician distribution and physician quality cannot be determined.
The increasingly diverse varieties of student types, specialty types, and practice environments present a daunting array of factors to study and comprehend. There are also new considerations.
Unified Theories Should Be Questioned
Current workforce evaluations are a jumble. Attempts to form a unified theory such as “controllable lifestyle” or to consider all primary care types alike or to attribute career choice decisions to medical school influences, role modeling, or Title VII investments made years earlier only confuse matters. Each of the specialties has some variation. Some can be traced to new specialties or increased demand. Others may benefit from specific health policy changes such as including dialysis patients. Career choice may suffer when supply is considered greater than demand.20
Another problem involves the lack of national standards for distribution. When there is no baseline for rural workforce, underserved workforce, or an understanding of the baseline influences attributable to age, geographic origins, and socioeconomics, it is difficult to discern distribution from the standpoint of program interventions or the impact of student characteristics. Divisions of physicians into major medical center locations, underserved locations, and served locations can allow standards to be set. Rural workforce levels in the United States are 11% of the total physician workforce and underserved workforce levels are 7% when the 1987 – 1999 graduates from all medical school sources are considered.1 The standards for family physicians are a consistent 24% rural distribution and 12 – 14% underserved distribution for national populations of family physicians for 1987 – 1999 class years.
Lack of national standards also allows selected subsets of practitioners to promote themselves as a solution for physician workforce problems. Any health policy piece can find a suitable subgroup to make a point to politicians regarding funding needs. However the real needs are national studies that demonstrate consistent and sustained outcomes such as distribution to rural or to underserved areas or retention in primary care. Physician assistants ten years ago used to have 25% rural distribution when 40% were working with FPGP physicians, but now have 17% rural distribution with only 28.5% in FPGP offices.22 Lack of consideration of time and health policy means failure to understand retention in primary care and the attractive salaries connected with major medical centers.
Closer examination often reveals that the outcomes are a result of selected student characteristics or failure to study the outcomes over a long enough period after graduation. Failure to consider time, health policy, geographic origins, socioeconomics, scores, ages, and relevance to national standards for distribution should be grounds for rejection regarding workforce publication. Without considering various outlier student and physician groups, it is difficult to interpret results.
Limitations in perspectives, data access, numbers of researchers, and varieties of researchers have clearly limited the range and quality and utility of workforce research.
Focusing on the Obvious
Physician workforce studies do not focus on the obvious: concentrations of physicians. Failure to understand that 70% or more of physicians have connections to major medical center types of settings before admission means failure in understanding physician concentrations inside major medical centers and lack of physician distribution outside.1 There are clearly factors such as health policy, facilities, and social organization that concentrate physicians.
A focus on selected “reasons” for poor distribution such as limited economics or the “paucity of culture”4 does not excuse medical education from the responsibility of physician distribution. All of the excuses of medical education leaders melted away when the
Studies never demonstrate what happens to those who fail to gain admission, but their practice locations are not in doubt. Those most like them with slightly lower scores and lower probability of medical school admission are students that are most like the lower and middle income 70% of the nation that is most in need of health care. Those with the lowest probability of admission clearly have the highest probability of distribution. Those with the highest probability of admission have the most connections to medical schools before admission, are the youngest, and are most likely to be admitted early. They are also the least likely to distribute outside of major medical centers. They are also replacing those most likely to distribute. Focusing on the elite students is a process that excludes the most distributional. Rather than a focus on excluding students to minimize attrition levels or maximize board scores and prestige levels, studies could focus on admissions of students above a sufficient academic threshold.23, 24 It would seem to be common sense that admissions of students with service orientation, empathy, diligence, awareness of the needs of a wide variety of patient types, those most likely to be satisfied as physicians, and other indicators of future physician quality would share top priority along with academics.
A narrow admissions pool means narrow career choices and narrow distribution and may mean narrow physician and health care quality regarding the care of lower and middle income patients. In every other study of physician quality or the perception of physician quality, those most like the physicians in gender, race, and ethnicity have been evaluated to be higher quality. Why should socioeconomic differences be different? Empathy and service orientation also appear to be lacking in the privileged/higher scoring. Studies of awareness clearly demonstrate that there are problem areas regarding lack of awareness of the needs of lower income peoples in those most commonly admitted25 and this can be measured by income levels, social distance, and probably geographic distance.
What is perhaps most concerning regarding health access and physician distribution is the failure to see that those most different are the students who distribute. Understanding the importance of differences appears to be the foundation of studies involving health care costs, quality and the perception of quality, and access to physicians. These three also include dimensions of time as in short and long term costs, quality, and access. There are also dimensions of class differences as in costs, quality, and access for the highest status peoples with too much and the lowest status peoples with too little or inappropriate health care.
The most difficult part regarding admissions for distribution is attempting to understand how the limitations inherent in standardized testing make it difficult to determine whether lower scores mean lower performance or whether lower scores are the result of different origins in a non-standard student. Studies demonstrate that the relationships between MCAT scores and the usual outcomes are linear and consistent, but are not significant in magnitude or relevance. Research must not only be significant but it must be relevant. When visually portraying attrition, delay, or distinction curves compared to MCAT scores, there are dramatic differences at the extremes such as the lowest MCAT scores. However the changes are flat across the usual range of student MCAT scores. Selecting a student with a slightly higher MCAT does not provide any real advantage in short or long term performance measures.12 However it is increasingly clear that students with higher MCAT scores are the least likely to leave major medical centers, to choose primary care and family medicine, and are rarely found in rural or underserved areas.
Again, workforce research is framed by perspective. Physicians seek out the familiar. Physicians return to familiar practice locations and areas with similar levels of social organization. Those born in rural areas return rural. Underserved physicians return to underserved locations. Family physicians born in military bases return to military careers at 22%. Those most familiar with major medical centers remain in such locations, even when a career change is dictated to maximize the potential of staying local. When students have maximal choice of all locations as in family medicine, the return to the most familiar is most possible. Limitations in opportunity and career choice restrict admissions to a certain type of physician and limit physician location to major medical centers. Policies regarding child development, education, higher education, medical education support, and health policy all influence the location of physicians
Without careful studies regarding the wide range of physicians all along the pathway to a medical career, without standards for comparison, and without studies that consider concentrations of physicians, it is difficult to understand what drives career choice and practice location. Studies then fail to identify the medical students and the health policies that will aid in distribution.
Bias in Data Access
Bias or the perception of bias is a major limitation of current physician workforce studies and the use of such studies for health policy purposes. Limitations in data access and research support contribute to this problem. Associations own the testing, surveys, accreditation, and data collections from student origins to career choices to retirement. Those not connected to associations are forced into extensive, expensive, and time limited data collection. These small local studies may fail to demonstrate differences in a single medical school with a consistent medical student admission. Individual studies of small subgroups may inject more bias or may have different outcomes that confuse leaders and the public. The public is constantly confused by studies that drugs with side effects not seen in smaller populations or when studies involve different populations with conflicting results, as in studies of estrogens in older or younger populations. Association research efforts are only possible for those that have established years of a relationship with the association. These are researchers with an established track record and the results are just as traditional. Data access is limited except for special purposes.
Innovative research, critical research, probing research, new themes, and new perspectives are not likely in such settings. Despite calls for more basic science research, associations are not likely to support “basic science” probing forms of workforce research. Most studies are driven by lobbying needs designed to shape health policy in ways favorable to the association or group sponsoring the study, making the entire process subject to bias and uncertainty.
Associations rarely publish research with inconclusive results or studies that are not flattering to the association or to physicians. As a result the nation still supports medicine pediatrics, nurse practice, and physician assistants at high levels even though there is clear evidence that they have moved away from primary care and underserved areas to become major medical center providers. The nation needs to know that specific forms of providers designed for primary care in rural and underserved areas have been distorted by market forces, major medical center demand, and health policies. Until the nation’s leaders understand this, they will not know the true magnitude of how bad health policies have become for those in lower and middle income areas of the nation or why the nation needs a true foundation as primary care, rather than market forces.
There are likely reasons why the current workforce situation does not change. Again this has to do with perspective. Those who aspire to leadership positions are often very different than those who choose to practice. The leaders are often those who have had the most elite and atypical lives from birth throughout private school to top college to elite medical schools and top fellowships. Some have even skipped practice and moved directly into leadership. Even a brief review of physician workforce gatherings reveals that the researchers are currently at elite schools or else they have graduated from elite schools. From the time of Flexner a very few schools have had an enormous influence upon American medicine.26 Throughout much of the last century, these reforms were needed and the influence of the elite schools was an important part of the process. However the urbanizing, centralizing, academizing process may have gone too far. The most elite may have difficulty believing research results that are different than their perceptions based on their life experiences and their observations of colleagues, who are just as exclusive in origins and life experiences. There is little understanding of family medicine because they experience few family physicians and the few family physicians that they experience are academics like themselves.
Few can deny that the top 146 colleges train top future leaders or that these top colleges capture 74% of the top income quartile origin students while only 3% come from the bottom income quartile.3 Medical schools are also concentrations of the elite with 60% coming from the top 20% and 80% coming from the top 40% in income.27 Elite professional schools are concentrations of the elite with 90% from the top levels, compounded by concentrations of the children of professionals, alumni, and those with elite scores that have been treated exclusively since the day these scores were discovered. It may be difficult for physicians raised exclusively to consider the needs of a distant 70% of the nation’s population without significant shared immersion experiences. This is something that the nation tends to avoid across life experiences, education, admissions, college, professional school, and associations.
The message of Flexner was that physicians and medical schools needed higher academic standards. Academic standards are just fine now and some would consider them excessive. This generation of physicians must figure out how to relate better to patients, staff, management, leadership, and people at all levels. A focus on ever more exclusive physicians appears to be a step in the wrong direction.
Limitations in Definitions, Standards, and the Grant Process
The lack of objective data and few established tools and benchmarks has been a huge problem. Entire meetings can break down over what constitutes primary care or what a rural definition involves. Numerous coding systems have been established and researchers can often choose the coding that improves the desired results.
Sometimes no coding system exists. There is no coding system for major medical centers. This seems odd since this is the major physician location.
Underserved coding is difficult on a national scale. Federal and state designations are constantly changing. The designations are subject to bias in the process involved. Designation efforts require some level of organization on the part of populations, states, or medical centers. Those with low levels of organization may have the same level of need, but they can fail at multiple points. They can fail to recognize the opportunity to gain resources. They can fail to organize a sufficient effort or they can fail to “polish” the effort to maximize the probability of selection. The smaller or more “minority” the population and the more distant and different, then the process is more difficult. Working with a larger coalition may help, but this can distract from the original mission or force a sharing of resources with the same potential for maldistribtion.
The same failures ring true for underserved families, underserved populations, and potential underserved origin medical students. Recognizing potential and opportunity, coordinating the effort, and gaining the “polish” to look enough like those that succeed can be huge challenges. In the case of Native Americans, such an effort can result in being considered outside the tribe although efforts by tribes and medical schools are improving this situation. Those who appear to be polished have a huge advantage. A simple definition of polish is easy. Polish is “someone like me” or “something done the way I would do it.” Unless evaluations of people, programs, and proposals evaluate potential to a greater level than “polish,” the nation will fail evaluating those most different with impacts upon physician quality and in physician distribution. Those most polished will rarely be found serving any populations other than patients within major medical centers.
The current distributions of information and influence mean that those with the highest levels of organization may obtain resources even though they may have less need and even when they actually get the highest levels of resources in other ways by current health policy. Abuses of each new system of payment and each new system of support for underserved areas has become the rule rather than the exception. The list is long and comprehensive dating back to cost-plus funding and extending forward to maximizing reimbursement from prospective payment, abuses in rural health clinics, the growth and use of graduate medical education funding, bonus payments for underserved practice used by visiting subspecialists, and J-1 Visa programs now used to help recruit medical school faculty. Often programs and schools compete for grants not based on true need or demonstrated ability or creativity or potential for contribution, but based on how well organized they are and how well they manipulate data. This is a much easier process when there are no objective sources to verify the information submitted.
Searching for Objective Approaches
New organizations of data can categorize physician concentrations and dilutions. Birth origins tools can test past assumptions and help unlock the complex interactions between physician characteristics, admissions, training, and career decisions. Objective standards and comparisons can be established for entire national populations instead of small selected subsets of practitioners, states, or programs. New methods can critique current policies involving shortage designation decisions, reimbursement levels, and distributions of revenue. The practice locations can be compared to physician origins with the use of birth origins data in the Masterfile. It is relatively easy to identify and categorize major medical centers using three levels of physician concentration and two levels of geographic coding.1
The dimension of time has often been neglected in workforce studies. The effect of health policy can be tested by comparing class years of graduates across the nation, by types of students, or by types of medical schools. Deterioration of primary care is also seen over time beyond graduation for every primary care type except family medicine. The dimension of time can be neglected and can even be abused with the choice of particular class years, resulting in more favorable results. Choosing a health policy period with poor choice of family medicine and then an ending point with a maximal family medicine choice will improve outcomes in primary care, family medicine, and physician distribution. Choosing a peak period such as 1995 – 1997 graduates and then comparing the data to more recent career decisions during a major change in health policy is also subject to major error. Without careful controls or better understanding regarding career choice in all specialties, the studies may erroneously conclude the effect of economics, specific government programs, or lifestyle considerations.
Accountability in the Evaluation and Publication of Research
Another problem involves limitations in the scope of the research presented. It is possible to cherry-pick a selected subset with the best outcomes or characteristics or researchers can minimize the less flattering areas. An example would be comparisons of physician assistants, nurse practitioners, or family physicians.
The nation has missed some of the most important indications of collapsing primary care health policy. In addition to declines in family medicine choice for US MD Grads and the lowest levels of primary care retention in history for internal medicine residency graduates, there have been massive movements of nurse practitioners and physician assistants away from primary care. Just as medical schools tend to hide primary care declines by using first specialty choices and women’s health practitioners, NPs and PAs have used the same tactics. The losses of primary care have been hidden by allowing geriatrics, women’s health, and other hospital activities to count as primary care. Emergency care is the biggest NP and PA attraction. The growth of urgent and convenience and corporate care clinics also dilutes primary care workforce. This is not a surprise as it pays the most and saves major medical centers the most compared to emergency room physicians.
Studies have failed to address some areas that should be obvious to medical schools and nursing schools. The most obvious areas involve shortages. These include shortages of physician researchers and medical school faculty. With increasing subspecialty salaries and more and more regulations placed on teaching hospitals, medical schools are unable to keep up with competitive faculty salaries. Nursing faculty shortages may also be a result of the increased demand for nurse practitioners to substitute for emergency, urgent care, hospitalist, and subspecialty physicians.
Finally there is the ultimate bias of socioeconomics and the huge influence of income, wealth, power, position, and professional parents. As noted in recent exposés involving Duke and Brown, college admissions decisions are influenced by family wealth. New questions have been raised about standardized test scores involving errors and the widespread and immediate sharing of questions and answers. Standardized tests are not the problem. However dependence upon standardized testing can be a problem when used in ranking and rating medical students, schools, and national efforts. No standardized test has relevance when considering much beyond short term performance.
A focus on who will make the best physician in 5, 10, 15, or 25 years should be the primary task of medicine. Despite many efforts over many decades, there is no single test that can determine the highest quality physician for the future. A single test that selects mainly the best basic science students should not be the foundation of admissions efforts. Recently deans have come to the defense of basic science preparation and focus. When considering the physician specialties that depend upon higher level sciences and math, this is a group that involves less than 33% of US MD Grads. For nearly all physicians people skills and management abilities are much more important and rarely emphasized in admissions, preparation, or training.
Even though the Association of American Medical Colleges derives most of its income from testing, the proceeds should be devoted to a sincere and dedicated effort exploring all avenues and approaches to finding and admitting those who will be the best physicians for decades to come. Computerized testing efforts also are not a problem, however when staff was directed to come up with testing for communication skills and current areas considered more challenging, the choice to pursue computerized testing represents the easier path taken and a distraction. Should someone else develop a better tool for evaluating medical students, the AAMC would lose a major source of revenue and would not retain the ability to influence medical schools.
Each individual specialty should also examine whether top board scores and AOA status is best for the image of the discipline or the profession, whether it is best for long term quality, and whether scores and status are best for the types of patients served by the discipline.
The Importance of a Different Perspective for Considering Physician Distribution
When reviewing the entire pipeline from parents to birth to education to admissions, the nation’s physicians divide into two major groups.
State and national outcomes in education, health, and medical education can be tracked to the relative balance or imbalance regarding the two competing groups.
The elite group dominates medical school admissions. Elite students experience few obstacles of education or income or cost. Not surprisingly they are the youngest at medical school admission. They naturally have the best standardized test scores. They are most like past and present medical students. They naturally have “polish.” The concentration of elite students into major medical centers is a given.
The humble origin group represents 70% of the population of the
The pipeline to medical school for those most likely to distribute is paved with a wide range of underfunded, inconsistent, intermittent, and dependent grant programs. Child development is a major player. Rich parents spend tens of thousands on their children in the early years. Poor parents, poverty locations, and low income area schools cannot afford these expenditures. Headstart spends $7000 per child but the efforts are focused on personnel that are less expensive, but less likely to be effective than in Belgium where preschool teachers have college degrees and are called pedagogues.9 The various agencies run local Headstart efforts but spend huge amounts of energy just obtaining funding rather than a primary focus on meeting specific needs. Much the same is true in education where applying for grants distracts from delivering services. No Child Left Behind is fully capable of focusing on earlier years, but often the funding is distracted to higher levels. Studies also reveal the inequities in state education funding across states and within states. Those most likely to distribute as physicians fail to obtain the same opportunities as those in the highest income levels that live in areas with higher property values. Grants at the highest levels as in Title VII have much the same problem.
A more subtle form of abuse is present in need based grant programs. Some states and cities make attempts to keep income, education, and health resources distributed. This results in lower “scores” and less probability of federal or state resources. Other states have plenty of resources, but fail to distribute them well and then are “rescued” by government support programs. These abuses can only be identified with objective considerations of the current distributions and how these impact distributions of important resources such as professionals. By sending funds into the most impossible locations where multiple problems stem from societal failure, it is clearly possible to demonstrate that interventions do not work. The sad fact is that anything done after age 8 is often too little and too late. Children with a stellar beginning retain the highest potential for education, employment, and professional school. They have the lowest probability of generating major costs for society through prison, social programs, unemployment, extra education costs, and additional public safety expenditures. Until many of the need based programs gain full and sufficient funding on a regular annual basis, they will continue to spend most of their energies applying and seeking grants and grant renewals instead of renewing the
A focus on rural and underserved and primary care should be infrastructure, just like child development and education for low income children and for increasing amounts of middle income children. A primary example of the problems ingrained in the fabric of the nation involves preschool versus high school. The current emphasis in the nation is science, math, and verbal in the high school level. This predominantly impacts the college bound group that may represent the top 30% of students in the nation, the current 30% and those moving up into the 30%. Proposals such as universal preschool are not likely to impact the top 30% as they self fund these programs, however universal preschool would have a significant impact on the 70% in middle or lower income populations. The national emphasis at the higher grade levels is a reflection of who votes, who is socially organized, who has influence on governments, and general ignorance regarding the primary importance of a superior first six years of life for all children.
1. Bowman RC. Major Medical Centers. http://www.unmc.edu/Community/ruralmeded/major_medical_centers.htm. Accessed October 26, 2006.
2. Carl J Shapiro Institute for Education and Research at Harvard Medical School. Executive Summary, Recommendations and Conclusions from the Millennium Conference on the Clinical Education of Medical Students. Boston April 2001.
3. Carnevale AaR, SJ. Left Behind: Unequal Opportunity in Higher Education, Reality Check Series. In: Kahlenburg R, ed. New York: The Century Foundation Press; 2004:p. 106.
4. Cohen JJ. Why doctors don't always go where they're needed. Acad Med. Dec 1998;73(12):1277.
5. Kassebaum DG, Szenas PL. Rural sources of medical students, and graduates' choice of rural practice. Acad Med. Mar 1993;68(3):232-236.
6. Butler WT. Academic medicine's season of accountability and social responsibility. Acad Med. Feb 1992;67(2):68-73.
7. Bowman RC. Admissions Ratio By Birth Origins in U.S. Medical Students. http://www.unmc.edu/Community/ruralmeded/underserved/admissions_ratio_by_birth_origin.htm.
8. Starfield B. Primary Care: Concept, Evaluation, and Policy. 1992.
9. Organization for Economic Cooperation and Development. Education at a Glance 2005. http://www.oecd.org/dataoecd/41/12/35341224.pdf. Accessed August 4, 2006.
10. Veloski JJ, Callahan CA, Xu G, Hojat M, Nash D. Prediction of students' performances on licensing examinations using age, race, sex, undergraduate GPAs, and MCAT scores. Acad Med. Oct 2000 2004;75 (10 Suppl):S28-30.
11. McGaghie WC. Assessing readiness for medical education: evolution of the medical college admission test. Jama. Sep 4 2002;288(9):1085-1090.
12. Julian ER. Validity of the Medical College Admission Test for predicting medical school performance. Acad Med. Oct 2005;80(10):910-917.
13. Julian ER, Ingersoll DJ, Etienne PM, Hilger AE. The impact of testing accommodations on MCAT scores: descriptive results. Acad Med. Apr 2004;79(4):360-364.
14. Leigh JP, Kravitz RL, Schembri M, Samuels SJ, Mobley S. Physician career satisfaction across specialties. Arch Intern Med. Jul 22 2002;162(14):1577-1584.
15. Jolly P. Charting Outcomes in the Match. https://services.aamc.org/Publications/showfile.cfm?file=version68.pdf&prd_id=159&prv_id=189&pdf_id=68. Accessed November 10, 2006.
16. Nickens HW, Ready TP, Petersdorf RG. Project 3000 by 2000. Racial and ethnic diversity in U.S. medical schools. N Engl J Med. Aug 18 1994;331(7):472-476.
17. Butler WT, Thomson WA, Morrissey CT, Miller LM, Smith QW. Baylor's program to attract minority students and others to science and medicine. Acad Med. Jun 1991;66(6):305-311.
18. Crump R, Byrne M, Joshua M. The University of Louisville Medical School's comprehensive programs to increase its percentage of underrepresented-minority students. Acad Med. Apr 1999;74(4):315-317.
19. Hart B, and Risley, T. Meaningful Differences in the Everyday Experience of Young Children. Baltimore: Paul H. Brookes; 1995.
20. Bowman R. Changes in Specialty Choice 1987 - 1999; 2006.
21. Bowman RC. Five Periods of Health Policy and Physician Career Choice; 2006.
22. American Academy of Physician Assistants. Data and Statistics. http://www.aapa.org/research/index.html. Accessed October 26, 2006, 2006.
23. Albanese MA, Farrell P, Dottl S. Statistical criteria for setting thresholds in medical school admissions. Adv Health Sci Educ Theory Pract. 2005;10(2):89-103.
24. Albanese MA, Farrell P, Dottl SL. A comparison of statistical criteria for setting optimally discriminating MCAT and GPA thresholds in medical school admissions. Teach Learn Med. Spring 2005;17(2):149-158.
25. Association of American Medical Colleges. Minority Students in Medical Education: Facts and Figures XI Available at https://services.aamc.org/Publications/showfile.cfm?file=version12.pdf&prd_id=89&prvid=87 Accessed April, 2003. Washington DC 1998.
26. Barry J. The Great Influenza; 2005.
27. Association of American Medical Colleges. Medical Student Education Costs and Student Debt Available at https://services.aamc.org/Publications/showfile.cfm?file=version35.pdf&prd_id=121&prv_id=137&pdf_id=35 Accessed July 2006. Washington DC 2005.