A New Open-Source, Dynamic Model for Projecting Physician Supply and Demand

Model Overview
With a grant from the Physicians Foundation, the Cecil G. Sheps Center for Health Services Research (Sheps Center) is developing a dynamic, open-source, web-based tool to project physician supply and demand and estimates of shortages in the United States. The model will be intuitive, easy to use, customizable, and freely available to all who wish to use it. The tool will allow policymakers, physician practice and human resource managers, physician associations, medical societies, and other stakeholders to identify imbalances across geographical areas in the distribution of physicians of various specialties.

The model is not intended to generate a single “right” answer. Instead it will allow users, operating in a rapidly changing system, to simulate the effects on physician supply and demand of various proposed health care workforce policies, anticipated changes in physician characteristics and behaviors, health system changes and trends in population demographics and disease burden. Because the model is open-source and transparent, everyone can critically evaluate the model’s assumptions and suggest improvements.

The model is being developed with input from a clinical advisory board comprised of physicians from a wide variety of specialties. Clinical input into the model is essential to identify real-world practice issues affecting the workforce and this input is being facilitated through a collaboration with The North Carolina Medical Society Foundation (NCMSF). The NCMSF provides a “boots on the ground” perspective from practicing physicians that will bolster the validity of the model and the physician community’s confidence in model forecasts. The Foundation’s involvement in the project is critical to engaging the wider physician community in identifying and implementing interventions to address shortages for given specialties and specific geographic areas.

The Model in 3 Parts
The model has three main elements (Figure 1):

1. A demand-side component that accounts for changes in population demand for services in 18 major health areas (e.g.,complications of pregnancy and childbirth; diseases of the digestive system; endocrine and metabolic disorders, etc.) that are provided in hospital-based (e.g., inpatient and outpatient), office-based and emergency department settings;
2. A supply side component that simulates individual physicians’ location and other work choices throughout their careers from training to retirement. Individual physicians’ choices are modeled by estimating probabilities that are dependent on their age, specialty and geographic location. The model includes a geographic migration component to account for the probabilities of movement of physicians from their training locations into initial practice locations and also for migration during their careers; and
3. An empirically-based “plasticity” factor that reconciles supply with demand across 35 clinical service areas (Figure 2).

The plasticity factor will compare the demand for physician services to the supply of those services to determine whether supply can meet local demand. The key contribution of the plasticity factor is that it accounts for the heterogeneity of services provided by physicians. For example, analyses of National Ambulatory Medical Care Survey (NAMCS) data suggest that dermatologists provide a much more uniform mix of services than family physicians. Family physicians may practice more pediatrics and be more likely to provide obstetrical delivers in areas where there are fewer pediatricians and obstetricians, but dermatologists will remain focused. This translates into a series of different potential combinations of specialties that can respond to a need for care in a specific place. The model will offer variable “solutions” to local or regional need estimation, including scenarios regarding how changes in NP and PA substitution rates and productivity might address physician supply shortages.

To make the model tractable and useful to the many different physician specialties as well as human resource managers planning for the workforce needed to meet future demand for health services, the model merges physician specialties into clinical service areas that can be used in both the demand- and supply-side models. Determining the optimal number of clinical service areas to model required finding a balance between utility and complexity. The clinical services areas were chosen based on the frequency of visits for conditions and resulted in the classification of physicians shown in Figure 2.

Progress in Building Model
The model-building process is currently underway. In the coming months, the project team will produce prototypes to test functionality and calibrate the model. The team will develop healthcare policy and environmental change scenarios to test the model’s face and internal validity with input from clinical collaborators and clinical advisory group. The team will then finalize user inputs, formats for model outputs, and design the model’s website. Once the model’s website has been launched, the team will work to ensure that it reaches the broadest possible policymaking audience through fact sheets, a user guide, a final report, peer-reviewed publications, and conference presentations. In the interim, a detailed document will be available on the project’s website in July 2012 that summarizes the current state of workforce modeling in the United States, compares the project’s model to previous workforce models, and describes the model approach, innovations and methods.