Volume 40 | Number 5p1 | October 2005

Abstract List

John C. Fortney Ph.D., Diane E. Steffick, James F. Burgess Ph.D., Matt L. Maciejewski, Laura A. Petersen M.D., M.P.H.


To determine whether strategies designed to increase members' use of primary care services result in decreases (substitution) or increases (complementation) in the use and cost of other types of health services.

Study Setting

Encounter and cost data were extracted from the Department of Veterans Affairs (VA) administrative data sources for the period 1995–1999. This timeframe captures the VA's natural experiment of increasing geographic access to primary care by establishing new satellite primary care clinics, known as Community‐Based Outpatient Clinics (CBOCs).

Study Design

We exploited this natural experiment to estimate the substitutability of primary care for other health services and its impact on cost. Hypotheses were tested using ordinary least squares (OLS) regression, which was potentially subject to endogeneity bias. Endogeneity bias was assessed using a Hausman test. Endogeneity bias was accounted for by using instrumental variables analysis, which capitalized on the establishment of CBOCs to provide an exogenous identifier (change in travel distance to primary care).

Data Collection

Demographic, encounter, and cost data were collected for all veterans using VA health services who resided in the catchment areas of new CBOCs and for a matched group of veterans residing outside CBOC catchment areas.

Principal Findings

Change in distance to primary care was a significant and substantial predictor of change in primary care visits. OLS analyses indicated that an increase in primary care service use was associated with increases in the use of all specialty outpatient services and inpatient services, as well as increases in inpatient and outpatient costs. Hausman tests confirmed that OLS results for specialty mental health encounters and mental health admissions were unbiased, but that results for specialty medical encounters, physical health admissions, and outpatient costs were biased. Instrumental variables analyses indicated that an increase in primary care encounters was associated with a decrease in specialty medical encounters and was not associated with an increase in physical health admissions, or outpatient costs.


Results provide evidence that health systems can implement strategies to encourage their members to use more primary care services without driving up physical health costs.