Summary

[D]

Models to project physician supply and demand are often sensitive to assumptions regarding the characteristics of the future healthcare system and whether current trends will persist. Replete with examples of projected trends that failed to fully materialize and the emergence of trends that were never anticipated, the history of the United States healthcare system shows a system that is continually evolving. As Uwe Reinhardt (2002, p. 196) states: it is a “daunting enterprise . . . 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.” This fact is especially true when projecting demand for physician services, where much uncertainty exists regarding the characteristics of the future healthcare system.

Factors leading to potential underestimates of physician requirements include: (1) underestimates by the United States Census Bureau of actual population growth, (2) overestimates of the proportion of population insured through plans with aggressive managed care practices, (3) overestimates of proportion of care provided by NPCs, (4) underestimates of increased per capita use of physician services over time, and (5) overestimates of increases in physician productivity. Although we are unable to predict with certainty whether current trends in the healthcare operating environment will persist and what new trends will emerge, efforts to model physician supply and demand require educated predictions of major trends that affect the physician workforce. These uncertainties, combined with an ever changing healthcare system, highlight the need to frequently reassess supply and requirements projections.

In addition to the uncertainties mentioned above that affect the accuracy of projections, the PSM and PRM, like all projections models, have their limitations. For example, both models are static in that they do not model how physicians, patients, and insurers will react to changing conditions. As an example, physician earnings tend to increase as demand exceeds supply, resulting in financial incentives for physicians to enter specialties with a shortage of physicians rather than entering specialties with a surplus of physicians. Similarly, the scope of practice in particular specialties is changing over time. An expanded scope of practice could result in greater physician requirements for that specialty, with the possibility that requirements might fall for a specialty with an overlapping scope of practice that competes for many of the same patients.

A limitation of a utilization-based approach to model physician requirements is that, by definition, the approach assumes that the physician labor market is in balance in the base year. Inefficiencies in the delivery of care are extrapolated into future years’ projections.

Another limitation is that the PSM and PRM are national models. Although they can be adapted to project supply and demand for smaller geographic regions such as States, the models do little to inform the debate regarding the future adequacy of physician supply in currently underserved areas. Past government policies to improve physician supply in underserved areas have relied in part on the assumption that physician surpluses (especially surpluses of primary care physicians) will create financial motivations for physicians to gravitate to underserved areas. The baseline projections suggest that the supply of primary care physicians will grow at about the same rate as demand through 2020, which will create little financial pressure for primary care physicians to disperse to traditionally underserved areas.

Additional research that might improve the supply projections include the following:

· Estimating more exact retirement patterns. As discussed, the PSM uses historical data to estimate separation rates that we think reflect long-term trends, rather than short-term fluctuations reflecting current market conditions. Preliminary results from the PWS are consistent with the concern that AMA Masterfile data underestimate the number of retired physicians, which could lead to overestimates of physician supply.

· Modeling specialty choice. The PSM uses historical data to estimate the distribution of new physicians into various specialties. This reliance on historical data might understate the importance of new trends in specialty choice—especially as it pertains to relatively new specialties such as critical care.

· Estimating long-term trends in physician productivity. With the exception of modeling trends in average hours worked as women and older physicians constitute a growing portion of the physician workforce, the PSM does not explicitly model changes in physician productivity. We calculated the productivity scenario presented in this paper outside the model and assumed a 1 percent annual increase in physician productivity. Improved training, technological advances, and increased use of NPCs and other health workers could lead to increased productivity, and additional research could inform how such productivity increases should be incorporated into the physician supply projections.

Additional research that might improve the demand projections includes the following:

· Estimating the impact of economic growth on physician requirements. Economic growth could change patient expectations and the ability of patients, employers, and the government to purchase additional physician services. The recent work by Cooper et al. has opened the debate on this topic, but the research community is far from reaching a consensus on the implications of economic growth for the future adequacy of physician supply.

· Estimating the impact of growing NPC supply on physician requirements. Expansion of the clinical or business autonomy of NPCs could increase competition between NPCs and certain physician specialties, resulting in slower growth in physician requirements. Similarly, collaboration between NPCs and supervising physicians can increase physician productivity, which in turn reduces physician requirements.

· Estimating the impact of new technologies on short-term and long-term requirements for physicians. New technologies could allow physicians to provide new services, and they could reduce mortality, increasing long-term requirements. Likewise, new tests, procedures, pharmaceuticals, or equipment could provide a substitute for some physician services thus slowing the growth in physician requirements.

Despite the limitations of projection models like the PSM and PRM, and despite the uncertainties of how the healthcare system will look in the future, these two models are powerful tools for understanding the implications of changing demographics, changing government policies, and other trends on the future adequacy of physician supply.