Bob Phillips MD, Executive Director of the Robert Graham Center, the American Academy of Family Physicians’ (AAFP) DC-based policy center, gave one of the plenary speeches at the recent annual meeting of the Society of Teachers of Family Medicine (STFM). His talk, available at the Family Medicine Digital Resource Library (www.fmdrl.org) presented a great deal of data regarding the current, and anticipated future, state of the US primary care workforce. There are currently 222,308 primary care doctors (FPs, GPs, general internist, general pediatricians and geriatricians), or one for every 1358 people, and an additional 86,000+ NPs and PAs in primary care, for a total of about 308,000, or about 1 primary care provider for every 1000. This is not a bad ratio, looking at the nation as a whole, but geographic distribution is another matter, with the ratio of primary care providers (PCPs) to person ranging from 500:1 to 5,000:1 in various primary care service areas (PCSAs); the nearly 5000 PCSAs with a shortage have a deficit of over 54,000 PCPs, which is equal to the “surplus” in the 1,600 or so PCSAs who have higher ratios. And, of course, those underserved PCSAs comprise the vast majority of the US geographic area (see figure).
Phillips and colleagues estimate that we will need an additional 50,000-60,000 primary care physicians by 2025. The largest cause of the growth is, of course, the projected growth in the US population, but it is augmented by the aging of that population (older people require more health care services) and, significantly, the increase in the number of insured-and-thus-likely-to-access-health-care created by the new coverage provisions in the Affordable Care Act (ACA). This portion of the increase is smaller, but unlike the gradual growth resulting from the other two causes, will “hit” all at once in 2014, and our primary care workforce is in no way prepared to meet it. This lesson was emphasized by the experience that Massachusetts had when it introduced near-universal coverage; in addition to adding more people to the “insured pool”, the group that is added has pent-up health care needs. Moreover, the currently uninsured are largely clustered in areas with the lowest current PCP:population ratios, so that what will happen when they gain insurance is that they will end up seeking primary care in the emergency room. While the Graham Center estimates a need for an additional 8,500-10,000 primary care doctors, given what Dr. Phillips calls our “voodoo” workforce policy -- which not only does not incent students to enter primary care or locate in rural and underserved areas, but in fact does quite the opposite -- the actual increase in the number of providers would need to be many times higher. Much recent work has demonstrated the decrease in student interest in primary care, and in particular the phenomenal decrease in the percent of internal medicine graduates entering primary care (as opposed to entering subspecialties or hospitalist practice), as documented by Garibaldi and Hauer among others (see A Quality Health System Needs More Primary Care Physicians, Dec 11, 2008). More recent data collected by the American College of Physicians (ACP – the internists’ group) from residents taking required examinations in 2009 indicate that only 21% of these residents are planning careers in general internal medicine (65% subspecialties, 10% hospitalist) and the actual results tend to show that these plans tend to skew even more to the latter two groups when decisions are actually made.
I have pointed to money, specifically the anticipated income related to student debt, as a major determinant of specialty choice, and Dr. Phillips makes this quite graphic by comparing the ratio of primary care and family physician incomes to more highly paid specialties over time. Using Diagnostic Radiology and Orthopedic Surgery as comparators, primary care incomes, which were about 60% in 1979, dropped to barely 35% in 2003, a trend that has not decreased since. Other graphs show that the % of graduate training (residency) positions filled by US medical graduates tracks linearly with specialty income, and that the growth in new residency positions has been almost entirely in those high-income specialties with drops in primary care positions. (This is not only because of student interest; it is also because many of these new positions are funded by hospitals. The same specialists – radiologists, cardiologists, orthopedists, anesthesiologists – that make big incomes for themselves also make big profits for the hospitals, so that hospitals are more interested in increasing their capacity to do these functions by having more trainees, residents and fellows, in these specialties.)
This creates a problem. The Graham Center data support much other research that has been cited in this blog by Starfield and others indicating that a health system that is based on primary care, with 40-50% PCPs, creates the greatest benefit in health and lowers cost. We have currently about 32% PCPs. With an interest in primary care among medical students now at about 22%, the problem is going to grow, not shrink. And, as I have often written, if we are interested in increasing primary care specialty choice, we are largely taking the wrong students into medical schools (e.g., Primary care specialty choice: student characteristics, July 12, 2010). Given that these characteristics are in large part negatively associated with family income, the changes in funding for medical schools are also troubling. Phillips cites an interview with the founding dean of the University of Missouri-Kansas City (UMKC) medical school, E. Grey Dimond MD, in the Kansas City Star (April 25, 2011). Dr. Dimond is asked how UMKC, as the “public” medical school with the least state funding of any in the US, survives, and he answers that they have increased tuition to become the highest tuition school in the country. This, of course, does not bode well for low-income students, urban or rural, becoming physicians: “Farm kids in Missouri from little towns that need doctors can’t pay what we have to have.” And, for those low (and middle, and even upper-middle) income students who graduate with debts often exceeding $250,000, those income differences among specialties loom very large – and this does not bode well for primary care.
Phillips provides evidence that Medicare costs and avoidable hospitalizations and hospital readmissions drop dramatically when there are higher primary care ratios (ratios of 1 FP+NP+PA:1500-2000 people, or 1:1000 if all PCPs considered). He cites a large number of studies demonstrating essentially the same thing.
Is there a bright side? Are there solutions? Well, the contributions of primary care are now being widely acknowledged, and there are lots of calls for increasing primary care physicians. The ACA bill provides some increased funding for primary care (about a 10% increase under Medicare) and major funding increases for the National Health Service Corps (NHSC), which pays for medical education by (some) scholarships or (mostly) loan repayment for physicians who enter primary care (and sometimes general surgery) and practice in an underserved area for a period of years. Unfortunately, these are not sufficient; a 10% increase sounds like a lot, but if it brings the primary care doctor’s income from 33% to 37% of that of a specialist (and this would be if the whole practice were Medicare), it is not going to do the trick. The loan repayment from NHSC is good, but it rarely covers the whole bill.
What would work? Medicare, taking the lead among all payers, needs to increase primary care physicians’ income dramatically. The Council on Graduate Medical Education (COGME) estimates in its very impressive 20th Report, Advancing Primary Care, that a family physician must be able to anticipate earning 70% of what a subspecialist makes if the goal of having a 40% primary care workforce in 20 years is to happen., the level at which income expectations tend to wash out of the decision on specialty choice. The federal and state governments should learn from successful models and repay all of the loans of medical school graduates who enter primary care over 8-10 years (enough time to ensure they are actually practicing primary care) and do it twice as fast for those who practice in an underserved setting.
This is what it will take to bend the curve of specialty choice, and, as a result, to bend the cost curve of providing health care.
 Garibaldi, RA, Popkave C, Bylsma W, “Career plans for trainees in internal medicine residency programs”, Acad Med 2005 May;80(5):507-12
 Hauer KE, Durning SJ, Kernan WN et al., “Factors associated with medical students’ career choices regarding internal medicine”. JAMA 2008;300(10):1154-64
 The benefit on cost of hospitalization, avoidable hospitalizations, and readmissions is more difficult to assess for general internists than for family physicians; this would be unsurprising given that their training in almost all in the hospital rather than in the ambulatory setting.