Geographic Variance in Maryland’s Potentially Preventable Emergency Visits: Comparison of Explanatory Models

Abstract

The use of emergency departments (EDs) for potentially preventable visits is costly and inefficient. In Maryland, about 20%–30% of such visits are ambulatory care sensitive and thus potentially preventable. The uninsured are often perceived to account for a disproportionate share of such visits. This analysis aimed to (a) compare and explain the geographic variance in Maryland’s potentially preventable ED visit (PPV) rates for the total and uninsured populations and (b) test the predictive value of regression models developed. Geographic hot spots of increased PPV rates were highly correlated for uninsured and total populations, but uninsured rates were more clustered in urban areas. Poisson and geographically weighted regression (GWR) models best fit the data and predicted 40%–52% and 46% of the variance in 2009 total and uninsured rates, respectively. Significant predictors of increased PPV rates were social determinants of health: lower per capita income and education levels, and higher percentage of female-headed households.