Joshua Freeman, MD
We need more primary care physicians. I have written about this often, and cited extensive references that support this contention, most recently in The role of Primary Care in improving health: In the US and aroundthe world, October 13, 2013. Yet, although most studies from the US and around the world suggest that the optimum percent of primary care doctors should be 40-60%, the ratio in the US is under 30% and falling. A clear reason for this is that relative lack of interest of US medical students in entering primary care at the rates needed to maintain, not to mention increase, our current primary care ratio. In addition, the ratio of primary care to other specialty residency positions is too low. Here we confront the fact that the large majority of medical students completing Internal Medicine residencies enter subspecialty fellowships rather than practicing General Internal Medicine. At the Graduate Medical Education level, a simple way of estimating the future production of primary care doctors would be to add the number of residency positions in Internal Medicine (IM), Pediatrics (PD), Family Medicine (FM), and combined Internal Medicine-Pediatrics (IMPD) and subtract the number of fellowship positions they might enter. This still overestimates the number of general internists, however, since it does not account for doctors who practice as “hospitalists” after completing their residency because such a role does not currently require a fellowship (as does, say cardiology). Estimates are now that 50% or more of IM graduates who do not pursue fellowship training become hospitalists.
Thus, we welcome the research report from the Association of American Medical Colleges (AAMC) “The role of in medical school culture inprimary care career choice” (1), by Erikson et al. that appears in the December 2013 issue of AAMC’s journal Academic Medicine. The authors surveyed all 4th-year medical students from a random sample of 20 medical schools to assess both student and school level characteristics that were associated with greater likelihood of entering primary care. The first, and arguably most important finding, was that only 13% of these final-year medical students were planning on primary care careers. This is despite the fact that 40% were planning to enter the “primary care” residencies of IM, PD, FM, and IMPD, with most of the fall-off in internal medicine and least in family medicine. This finding strongly supports my assertions above, and makes clear that the historically AAMC-encouraged practice of medical schools reporting “primary care” rates by entry into residencies in those fields is not valid. It also, even more important, shows the extent of our problem – a 13% production rate will not get us from 30% to 40% or 50% primary care no matter how long we wait; obviously it will take us in the other direction.
The primary outcome variable of the study was entry into primary care, and it specifically looked at two school level (but perceived by students, as reported in the survey) characteristics: badmouthing primary care (faculty, residents or other students saying it is a fall back or something that is a “waste of a mind”) and having greater than the average number of positive primary care experiences. It turns out that both were associated with primary care choice (in the case of badmouthing, students from schools with higher than average reported rates were less likely to be planning primary care careers, while students who were planning such careers reported higher rates of badmouthing), but, after controlling for individual student and school characteristics, accounted for only 8% of the difference in primary care choice. Characteristics of the student (demographics such as sex, minority status or rural origin, academic performance defined as the score on Step 1 of USMLE, as well as expectation of income and feeling of a personal “fit” with primary care) and of the school (research emphasis, private vs. public, selectivity) accounted for the rest. Interestingly, debt was not a significant factor in this study.
I would argue that many of these individual and school characteristics are highly correlated. A school that prides itself on being selective (taking students with high scores) and producing subspecialists and research scientists does not have to badmouth primary care; the institutional culture intrinsically marginalizes it. On the other side, the students selected at those schools are more likely to have those characteristics (particularly high socioeconomic status and urban or suburban origin) not associated with primary care choice. It is worth noting that the measure of academic performance in this study was USMLE Step 1, usually taken after the first 2 years and focusing more on the basic science material covered in those years, rather than USMLE Step 2, which covers more clinical material (perhaps because not all 4th-year students studied have taken Step 2 yet). This biases the assessment of academic qualification; many studies have demonstrated high levels of association of pre-medical grades and scores on the Medical College Admissions Test (MCAT) with pre-clinical medical school course grades and USMLE Step 1 scores, but not with performance in any clinical activity, not to mention primary care. Perhaps most students improve their scores from Step 1 to Step 2, but it is particularly true for FM and primary care. A quick look at our KU students applying to our family medicine program shows an average increase of nearly 30 points in these scores.
So the problem is in the overall culture of medical schools, in their self-perception of their role (creating research scientists vs. clinicians, creating subspecialists vs. primary care doctors) and in their belief that taking students with the highest grades is equivalent to taking the best students. This culture, simply put, is bad, defined as “it has undesirable outcomes for the production of the doctors America needs”, and must change. Erikson and colleagues acknowledge that schools could do a better job of taking rural students, offer more opportunities to engage in public health and community outreach activities, and have more experiences in primary care, all of which were somewhat associated with primary care career choice. These are tepid, but coming from the AAMC, a reasonably significant set of recommendations. I say we need an immediate change in every single medical school to recruit at least half of every class with students whose demographic and personal characteristics are strongly associated with primary care choice, present a curriculum that has much less emphasis on “basic science” and more on clinical, especially public health, community health, and primary care. One of the primary bases for assessing the quality of a medical school should be its rate of primary care production, and this is going to require a major qualitative shift in their practices and the beliefs of many of their faculty and leaders.
I am NOT saying is that we don’t need subspecialists or research scientists. We do. I AM saying that the emphasis on production of these doctors compared to primary care doctors is out of whack, not just a little but tremendously so, and can only be addressed by a major sea change in attitudes and practices in all of our medical schools. I do not expect that all schools should produce the same percent of primary care physicians. Some might be at 70%, while others are “only” at 30%, but ALL need a huge increase, by whatever means it takes. Even if we produce 50% primary care physicians on average from all schools it will be a generation before we get to their being 50% of the workforce. At less than that it will take longer, and at less than 30% we will not even maintain where we are.
13% is not just “insufficient”, it is a scandalous abrogation of the responsibility of medical schools to provide for the health care of the American people. They should be ashamed, should be shamed, and must change.
(1) Erikson CE, Danish S, Jones KC, Sandberg SF, Carle AC, “The role of in medical school culture in primary care career choice”, Acad Med December2013;88(12) published online before print.