Analyzing cause and effect in the U.S. physician workforce.

In this expression, the product atctSt denotes the projected supply in year t of fulltime-equivalent (FTE) physicians in patient care. Component St denotes the number of physicians projected to be alive in year t, component at is the fraction of those physicians who are projected to be professionally active in year t, and component ct is the fraction of the professionally active number of FTE physicians in year t who are projected to be active in patient care. The product (Dt/Qt)Nt represents the projected “demand” or “requirements” of physicians in year t. Variable Dt denotes the projected average number of physician services per capita thought to be needed or demanded in year t. Qt is the projected average number of physician services projected to be produced per FTE physician in year t (physician productivity). Nt is the size of the population to be served. The ratio Dt/Qt is the familiar physician-to-population ratio, although here expressed as the number of FTE physicians per capita rather than per 100,000 population.1 Finally, variable Xt represents the projected physician surplus (if positive) or shortage (if negative) in year t. Even this highly compact forecasting model shows just how daunting an enterprise it is to estimate the physician surplus or shortage one or two decades into the future. Any of the variables in the equation can change over time, sometimes in unforeseen ways. Forecasters have tried to cope with this uncertainty on the demand side of this equation simply by positing normative physicianpopulation ratios (that is, normative values of the ratio Dt/Qt), either for all physicians or for subgroups of physicians. In his well-known projection of physician workforce requirements, for example, Jonathan Weiner used for that purpose the relatively low physician staffing ratios now observed in fully integrated managed care networks, including the Kaiser Permanente plans.2 On that basis, and on the assumption that 40–65 percent of Americans would receive their care from fully integrated managed care plans in 2000, Weiner projected a huge physician surplus for that year and beyond. As it turned out, Americans have not flocked into managed care organizations of this sort, and the projected surplus never materialized. In their paper in this volume, Richard Cooper and colleagues proceed at an even higher level of aggregation on the demand side. They observe that the actual physicianto-population ratio (Dt/Qt) tends to be highly positively correlated with per capita gross domestic product (GDP), within nations over time, across regions within nations at a point