A DYNAMIC STRUCTURAL EQUATION MODEL OF PHYSICIAN DISTRIBUTION

DAVID EDWARD BARTKUS, Purdue University

Abstract

The geographic and specialty maldistribution of physicians in the forty-eight states was documented and the paucity of studies of distribution dynamics cited. A dynamic structural equation model was developed to study the factors affecting the change over time of physician-population ratios. Three physician groups were studied: general and family practitioners, primary care specialists, and high level specialists. Within the model these three group variables were considered jointly dependent with five other variables: hospital facilities, medical centers, barriers to entry and mobility, per capita Medicare expenditures, and per capita Blue Cross and commercial insurance expenditures. Other explanatory variables included demographic factors, socioeconomic factors, and medical environmental factors. Five of the eight structural equations (three physician equations, hospital equation, medical centers equation) were hypothesized to be dynamic: by inclusion of a lagged dependent variable in these equations the coefficients of the explanatory variables took on dimensions of time--amount of change per time period. Time series data for the years 1963-1973 were gathered for all factors from published sources. A lag structure was empirically derived. A model was developed for each of the nine census divisions, a covariance analysis having dictated that different structures were involved. The two-stage least squares method of structural parameter estimation was used. Structural results indicated that, contrary to past literature, change in rates of general and family practitioners was much less affected by demographic and socioeconomic factors as by medical environmental factors, especially the presence of specialists in an area. With regard to specialists, while their previously cited dependence upon medical environmental factors (especially hospital facilities and medical centers) was confirmed, they actually showed a stronger relationship to sociodemographic factors than did general and family practitioners. Change in hospital facilities showed a stronger relationship to demographic and socioeconomic factors than to numbers of specialists, a somewhat unexpected result. Not unexpected was that both public and private health expenditures were found to be greater where physicians and hospital facilities were more prevalent. The structural model was used as the basis of a simulator to determine effects over time of various policies implemented in the health care arena. Implementation of a national health insurance plan would have little effect upon distribution of physicians, though expenditures would rise. A policy limiting growth of hospital facilities would as expected result in huge decreases in expenditures as well as slow the decrease of general and family practitioners. However, the concomitant decrease in primary care specialists produced a net decrease of primary care practitioners, posing an interesting tradeoff between availability of primary care and control of health expenditures. The last policy, which placed an arbitrary lid upon numbers of high level specialists, seemed to accomplish its objectives without any damaging side effects to the system.

Degree

Ph.D.

Subject Area

Sociology

COinS