|| Checking for direct PDF access through Ovid
In seeking to identify modifiable, system-level factors affecting emergency department (ED) use, we used a statewide Medicaid database to study community variation in ED use and ascertain community characteristics associated with higher use.This historical cohort study used administrative data from July 1, 2003 to December 31, 2004. Residence ZIP codes were used to assign all 555,219 Medicaid enrollees to 130 primary care service areas (PCSAs). PCSA characteristics studied included rural/urban status, presence of hospital(s), driving time to hospital, and several measures of primary care capacity. Statistical analyses used a 2-stage model. In the first stage (enrollee level), ED utilization rates adjusted for enrollee demographics and medical conditions were calculated for each PCSA. In the second stage (community level), a mixed effects linear model was used to determine the association between PCSA characteristics and ED use.ED utilization rates varied more than 20-fold among the PCSAs. Compared with PCSAs with primary care capacity less than need, PCSAs with capacity 1 to 2 times the need had 0.12 (95% CI: −0.044, −0.20) fewer ED visits/person/yr. Compared with PCSAs with the nearest hospital accessible within 10 minutes, PCSAs with the nearest hospital >30 minutes’ drive had 0.26 (95% CI: −0.38, −0.13) fewer ED visits/person/yr.Within this Medicaid population, ED utilization was determined not only by patient characteristics but by community characteristics. Better understanding of system-level factors affecting ED use can enable communities to improve their health care delivery systems–augmenting access to care and reducing reliance on EDs.