In 2006, when the IOM process that led to the publication of Retooling for an Aging America: Building the Health Care Workforce kicked off, I had the opportunity to speak to the committee. As the then–Atlantic Philanthropies’ representative, I thought very hard about what I would say to this august group that would be memorable and worth their time. The advice I gave (and pretty much all I can remember) was that they should not just count heads but weigh them. My point was that the workforce for an aging America is not just about the numbers of health care workers available, but just as much about what they know how to do. This is the crux of our work and our grantmaking: doing more of the same old care is not nearly as useful as providing more skilled, better organized care.
I was reminded of this point when I read "Primary Care Physician Workforce and Medicare Beneficiaries’ Health Outcomes" in the May 25 issue of JAMA. This paper is a masterpiece of health services research by Drs. Chang, Stukel, Flood and Goodman of the Dartmouth health services research empire (Dartmouth Institute for Health Policy and Clinical Practice). It very creatively uses various data sources to look at the relationship of local availability of primary care physician workforce to individual Medicare beneficiary outcomes like hospitalization, cost, and mortality.
Looking only at Medicare fee-for-service beneficiaries, the authors constructed measures of family physician and general internist workforce and related them to Medicare spending and two patient outcomes: death from all causes and hospitalizations for chronic conditions like diabetes and congestive heart failure, termed “ambulatory care sensitive condition hospitalizations” (e.g., diabetes, congestive heart failure, etc.). The authors adjusted for a raft of patient level information (race, age, income, health, etc.) to be sure that patient outcomes were due to availability of primary care service.
Given their gargantuan sample of over 5 million beneficiaries and all the services they received in 2007, the authors report statistically significant, but very small, effect size estimates for the impact of workforce on these outcomes. In the strongest effect, relative risk of hospitalization for an ambulatory care sensitive condition (those conditions that primary care is supposed to be able to take care of) was reduced by .09 between the lowest and highest quintiles of primary care full-time equivalents (FTEs). However, the raw risk of hospitalization for one of these conditions in any quintile was quite low--between 7 and 8 percent of beneficiaries annually, so the difference between high and low is less than 1 percent.
I had several reactions to this paper. First, I was terrifically impressed that the authors were able to link Medicare claims information from physicians, Medicare claims related to individual beneficiaries, census information, and AMA survey data in ways that had never been intended. This process enables them to control for many alternative explanations for relationships between the primary care workforce and health outcomes of older Americans. Finding even weak "signals" of relationships in what are very "noisy" and imperfect data is very impressive.
However, I think there are a few issues in how the work was conceived worth reassessing and that might even clarify the relationships between workforce and outcomes. For example, I really wonder about the exclusion of geriatricians and nurse practitioners/physician assistants from the primary care workforce. While their numbers may be small, the share of effort given to primary care services for Medicare beneficiaries may be disproportionately high. For example, 65% of geriatrician revenue comes from outpatient evaluation and management billings, the highest of any medical specialty, and it’s all from Medicare.
But most importantly, upon reflection on my own advice to the IOM committee, I wonder why anyone would have expected a relationship between primary care generalist workforce and the health outcomes of Medicare beneficiaries. While the Foundation has been firmly committed to the principle that primary care generalists and non-geriatrician specialist are critical to the well-being of older adults, we equally firmly believe that they need to learn more about caring for their older patients to be effective. Increasing the number of primary care workers only helps if they are good at caring for older adults. We do not feel that generalists or non-geriatrician specialists routinely get the training they need to be competent to care for older adults, particularly complex and frail older adults.
How do I know?
We know from the ACOVE studies that vulnerable elders have many physician encounters (their doctors bill for lots of evaluation and management services) and still only receive 30% of indicated care for geriatric conditions like falls and dementia. And we know that the more indicated care they receive, the better their outcomes, even for mortality.
We know from the Macy/ABIM “scariest ever” study that the quality of chronic care and care for geriatric conditions is still incredibly poor in internal medicine and family medicine training programs, despite years of effort that has slightly increased resident exposure to these skills.
We also know that even if physician providers had all the medical skills needed to provide competent geriatric care, routine primary care is not organized to provide that care. There still isn’t enough time in the day to do all that is indicated, and the physician’s medical skills are only a portion of what is needed to make a difference in the outcomes of complex older adults--it takes a team, and most current practices do not include effective teams.
We know that even in those capitated, primary care–centered, and highly organized systems of care such as Group Health Cooperative and Kaiser-Permanente, that our innovative delivery models such as IMPACT or Care Transitions lead to far superior outcomes than usual care. While these systems may lead the nation in care for younger people with single chronic conditions like diabetes, their efforts to really manage the care of complex older adults are only just beginning and their usual geriatric care is not particularly well developed. So, our expectations for the benefits to geriatric patients of primary care in disorganized fee-for-service should be much more modest.
A test of my hypothesis that generalists don’t have the skills to influence the outcomes of complex older adults is possible in these data. In my argument, the outcomes of frail and complexly ill older adults should not be much influenced by the availability of the generalist primary care workforce. Given the existing workforce’s lack of geriatric skill, what they know how to do and do well, is of relatively little value. However, for less complex older adults, primary care physician availability might still be valuable, and therefore be a stronger predictor of patient outcomes. I would love to see the authors do their analysis separately for more or less complex older people.
At the Foundation, we think that a better trained workforce is at least as valuable as a larger workforce. For example, we would like to see all medical schools incorporate the AAMC/Hartford 26 physician competencies into their programs and rigorously assess physicians in training on their attainment. Other health professions should continue to move in this direction as well. The first step in achieving higher quality of care is not throwing more providers at the problem—it’s making sure that the professionals that are out there have their heads stuffed full of the knowledge they need to take better care of older adults.