2019 HSR&D/QUERI National Conference

4052 — Predicting Unmet Primary Care Need using Economic Disequilibrium Models

Lead/Presenter: Edwin Wong,  COIN - Seattle/Denver
All Authors: Wong ES (Center of Innovation for Veteran-Centered and Value-Driven Care, VA Puget Sound Health Care System), Wilson CR (Primary Care Analytics Team, Veterans Health Administration), Nelson KM (Center for Veteran-Centered and Value-Driven Care, VA Puget Sound Health Care System)

Ensuring timely access to care is a key element of VA's Patient Aligned Care Team model. Inadequate internal capacity has been identified as a key access barrier. There is currently limited evidence using rigorous methods measuring how much additional capacity is needed to meet Veterans' healthcare needs. The objective of this study was to develop a new methodological application to quantitatively estimate unmet primary care (PC) demand.

We examined a random sample of 51 VA medical centers using VA administrative databases, linked with geographically aggregated data from the Area Health Resource File and the Centers for Disease Control and Prevention from 2014 to 2018. The outcome was PC visits, defined using selected stop code groupings. We modeled PC supply as a function facility-level PC staffing, local area characteristics and seasonality. We modeled PC demand as a function of Veteran enrollment, demographics and risk scores, flu severity and seasonality. We applied econometric disequilibrium models to jointly estimate parameters of the supply and demand equations. These models addressed the fact that observed number of visits does not represent quantity demanded when unmet demand exists. Joint supply and demand models were estimated separately for each facility using a maximum likelihood estimator.

The average number of visits per month provided by VA facilities decreased from 3,104 (SD = 1,750) to 2,654 (SD = 1,557) between 2014 and 2018. For 34 facilities, we did not reject the null hypothesis that the estimated number of visits demanded equaled visits supplied. For 8 facilities, visits demanded exceeded visits supplied by an average of 32%. For the remaining 9 facilities, visits demanded was less than visits supplied by an average of 22%.

In this novel application of econometric disequilibrium models, the estimated quantities of PC supplied and demanded were balanced at most facilities. However, model estimates suggest the level of PC needed by Veterans exceeded resources at some facilities, while at other facilities, staffing likely exceeded Veterans' PC needs.

Model estimates can be used to identify facilities where unmet PC demand exists and inform the level of staffing needed to meet expected PC demand of Veterans.