The Balanced Budget Act of 1997 fixed the number of graduate medical education (GME) positions funded by Medicare, which tightly governed the growth of the physician workforce in the United States for about five years.1,2 Then, between 2002 and 2007, training institutions largely self-financed a nearly 8% expansion in residency training, in which the majority of positions were for non-primary-care specialties and tightly aligned with teaching hospitals' financial interests.3,4 In 2007, the U.S. Council on Graduate Medical Education (COGME) called for further expansion of GME positions,1 but in May 2009 it expressed concern to the U.S. Secretary of Health and Human Services that recent expansions had effectively reduced the production of primary care physicians and reiterated the importance of “aligning GME with future health care needs.”5
The Resident Physician Shortage Reduction Act of 2009,6 introduced in both the U.S. House and Senate during debate on health care reform, proposed increasing residency slots by expanding federal funding to qualified GME teaching institutions. The bill included a 15% increase in Medicare-funded GME positions, taking the total to about 115,000. To be eligible for the additional slots, a hospital already beyond its government-funded cap would need to have (1) 10 or more resident positions above its cap and (2) at least 25% of its full-time equivalent residents in primary care and general surgery. Going forward, the hospital would not be allowed to go below this 25% threshold for 10 years.
This language created some concern because it committed nearly half of the proposed expansion to supporting the GME growth that hospitals had already financed on their own—growth that had effectively reduced primary care and general surgery production. In addition, it set the 25% threshold for eligibility at entry into residency, meaning that residents who subsequently subspecialized would be counted. A recent COGME report7 recommends that setting the threshold at a point after completion of the first residency would be a better measure of production because the majority of internal medicine residents and an increasing number of pediatric and surgery residents choose to subspecialize.
Although the bill was not passed, it is likely that a version of it will be reintroduced. To inform that debate, we tested the 25% threshold for each U.S. training institution at residents' completion of initial training and again two to four years out to demonstrate how the difference in the timing of measurement affects hospitals' eligibility. The bill set a relatively low bar given that nearly 38% of the physician workforce consists of primary care providers or general surgeons, so we also investigated how thresholds higher than 25% would affect institutional eligibility for expansion.
We assessed institutional production of primary care physicians and general surgeons in the United States between 2005 and 2007 using retrospective analysis of data from the 2009 American Medical Association Physician Masterfile.8 These files track residency and fellowship training for individual physicians for up to six experiences. We initially classified a resident as primary care if the specialty of his or her residency was family medicine, general internal medicine, or pediatrics.
For each institution providing GME to residents, we first identified all physicians who completed primary care or general surgery residencies during the study period. Then, for each institution, we calculated the production rate after initial training as the sum of primary care and general surgery graduates divided by the total number of graduates. Second, to calculate production rates that take into account further training, we excluded from the numerator those primary care or general surgery graduates who went on to complete another residency within the next two to four years (based on length of time from graduation date to 2009) and analyzed the data again. The rates we calculated, therefore, reflect production after initial training and for graduates two years (graduated in 2007) to four years (graduated in 2005) out from their completion of a primary care or general surgery residency training program. We then calculated the numbers of institutions with production rates above and below the 25% and 35% thresholds of primary care and general surgery graduates who did not pursue training beyond their first residency.
From 2005 to 2007, 683 institutions in the United States provided GME training to 116,004 residents. Among physicians completing their first residency, 54,245 (46.8%) were trained in primary care or general surgery, and 586 (85.8%) of the training institutions met the proposed 25% threshold for expansion eligibility if it were assessed at completion of initial residency. The 97 institutions that fell below the 25% threshold trained 2,307 (2%) residents.
At two to four years out, only 29,963 (25.8%) of the physicians had not pursued training beyond an initial primary care or general surgery residency, and an additional 135 institutions training 73,466 (63.3%) of the residents fell below the 25% threshold. Moving the threshold to 35%—a rate still below that needed to sustain the current primary care and general surgery workforces—at two to four years after initial training further reduced eligibility, leaving just 369 (54.0%) of the institutions (Figure 1). This excluded 314 institutions collectively training 93,774 (80.8%) of the residents (Figure 2). Raising the threshold to 40%, a level proposed by COGME in recent draft recommendations,9 left only 323 (47.3%) institutions when applied two to four years out from completion of initial residency. This higher threshold excluded 360 institutions training 99,269 (85.6%) of the residents.
We found that thresholds for GME expansion eligibility are sensitive to the point in the physician training pipeline at which institutional production is measured. Although the Resident Physician Shortage Reduction Act of 2009 contained much language that supported purposeful expansion of primary care and general surgery, the thresholds it set and the point at which they were applied may not have produced the desired effect. As written, the bill would apply its 25% threshold at the time of entry into first residency and perpetuate the hospital-funded expansion of the past decade, which strongly favored training other than and led to reduced production of primary care and general surgery physicians. However, were the same 25% threshold to be applied two to four years after completion of first (and perhaps only) residency, 135 institutions (with nearly two-thirds of current residency positions) would be eliminated from expansion eligibility. Applying higher thresholds, which are more appropriate to maintaining the current primary care and general surgery workforces, would remove from contention the majority of institutions, which train 80% to 85% of young physicians. Our findings demonstrate that setting such higher thresholds and applying them at least two years after completion of initial training would reward institutions that train generalist physicians and be more consistent with the intent of the bill's sponsors.
Since 2000, primary care GME production has dropped to levels too low to sustain the current primary care physician workforce.3,10 More than 1,250 resident positions in adult primary care training programs have been lost because of closures, absorption into other specialty training programs, and increased opportunity for subspecialty training.4 In fact, fellowship positions for internal medicine increased from 7,774 in 1999 to 10,062 in 2008, which may reflect an intent to increase the number of trainees who specialize and move out of primary care.11 COGME, the Medicare Payment Advisory Commission, the Josiah Macy Jr. Foundation, and others have made recent recommendations about improving primary care training and recommitting GME to the health care needs of the United States.1,9,12 Setting the expansion eligibility threshold at 25%—at time of entry or after completion of first residency—will not accomplish either goal. Production of at least 35% to 40% measured two to four years out from initial residency completion is required to sustain the current primary care and general surgery workforce, and, as noted above, COGME is considering a recommendation of at least 40%.13
The Patient Protection and Affordable Care Act of 2010 (ACA)14 is projected to exacerbate anticipated primary care shortages by helping 30 million or more people gain access to health insurance. ACA may help mitigate this situation through the preferential redistribution of currently unfilled residency positions to primary care and general surgery programs. It also supports community-based training by providing funding for new residency positions in teaching health centers and by lifting administrative restrictions on GME funding for training in outpatient settings—both changes that will favor primary care. ACA also expands debt relief for physicians who serve in the National Health Service Corps, which also favors primary care and general surgery, specifically in underserved areas. Further, the Secretary of Health and Human Services recently announced the Primary Care Residency Expansion Program, which will provide $168 million to fund temporary expansion of primary care residencies with an emphasis on community-based training.15 These aspects of ACA offer important opportunities to improve access to generalist services. Any further efforts to expand GME training will need similar tailoring to prevent continuation of the trends seen over the last decade and supply the workforce necessary to meet the country's health care needs.
Our study has potential limitations. By applying our proximal threshold measurement at graduation from first residency, we are underestimating likely institutional eligibility of using the entry criteria proposed in the GME expansion bill because a small proportion of residents start primary care or general surgery residencies but leave before completion of training or to enter other training. Similarly, because we only looked at data two to four years out from first residency graduation, we may be overestimating institutions' actual generalist production. Finally, we did not account for the estimated 20% of internal medicine graduates who work as hospitalists, or in other nonclinical roles in the decade after they finish their initial training, which would effectively reduce generalist production even further.16
The GME expansion legislation proposed in the House and Senate in 2009 would not have met legislative expectations using the proposed threshold and timing of measurement. Rather, it may have solidified the GME expansion that has reduced production of primary care physicians and general surgeons over the last decade to the point that it cannot sustain the current workforce. Applying thresholds at least two years out from completion of initial residency is a better indicator of final output, but production will not be sustained or increased unless a threshold higher than 25% is set. The workforce research community should help policy makers marry intent with evidence to produce the outcomes that many congressional advisory groups are advocating.
The authors wish to thank Winston Liaw, MD, Nicholas Weida, and Kim Epperson for their assistance in thinking about this report.
The opinions expressed in this report are solely those of the authors and the Robert Graham Center and do not necessarily reflect those of the American Academy of Family Physicians.
2Goodman DC. Improving accountability for the public investment in health profession education: It's time to try health workforce planning. JAMA. 2008;300:1205–1207.
3Salsberg E, Rockey PH, Rivers KL, Brotherton SE, Jackson GR. U.S. residency training before and after the 1997 Balanced Budget Act. JAMA. 2008;300:1174–1180.
4Weida NA, Phillips RL, Bazemore AW. Does graduate medical education also follow green? Arch Intern Med. 2010;170:389–390.
5Council on Graduate Medical Education. Letter to Kathleen Sebelius, Secretary of Health and Human Services. May 5, 2009. http://www.cogme.gov/cogmeletter.htm
. Accessed January 21, 2011.
15U.S. Department of Health and Human Services, Health Resources and Services Administration, Bureau of Health Professions, Division of Medicine and Dentistry. Affordable Care Act: Primary Care Residency Expansion (PCRE) Program. Frequently Asked Questions. http://www.hrsa.gov/grants/healthprofessions/pcrefaqs.pdf
. Accessed January 21, 2011.
© 2011 Association of American Medical Colleges
16Baron RJ. The chasm between intention and achievement in primary care. JAMA. 2009;301:1922–1924.