357 Utilizing a Health Information Exchange to Improve Care Transitions After an Emergency Department Visit

Abstract

Improving care transitions for ambulatory sensitive conditions such as diabetes between the ED and primary care is important to patient safety. Health information exchanges (HIEs) may be useful tools in this process, although data demonstrating this impact is lacking. The primary objective of this study was to assess the feasibility of implementing a tailored care plan in the emergency department Information Exchange (EDIE), a statewide emergency department HIE and care coordination platform endorsed by the American College of Emergency Physicians. We hypothesize that patients who receive this intervention will have decreased ED visits and increased contacts with primary care providers after an ED visit compared to patients who did not receive the intervention. We performed a randomized parallel experimental study with a pre-post design. A total of 260 patients who were attributed to one of four primary care clinics within the University of Washington system, had a diagnosis of diabetes and had at least one ED visit were randomly assigned to the CARE (Coordination and Referral from Emergency) intervention or a control arm. CARE included a tailored care plan in the EDIE which is a care coordination HIE platform that is used in all EDs in Washington and across 22 states. The care plan included primary care clinic information, current diabetes medications, and HgA1c values and was pushed to emergency providers at the time a patient checked in to any ED that utilizes EDIE. Care plans were updated weekly. Measured outcomes included statewide ED visits and primary care clinic contacts within 10-days of an ED visit. These outcomes were compared using “differences in differences” models 12 months pre- and post-intervention. Overall, the number of ED visits decreased significantly in the post-intervention period (mean ratio: 0.75, p=0.002). The decrease was statistically significant in the intervention group (0.66, p=0.003) while the decrease in the control group was non-significant (0.83, p=0.10) over the same period. Similarly, the rate of primary care provider contacts after an ED visit decreased significantly overall (odds ratio: 0.76, p=0.048). In this case the decrease was statistically significantly in the control group (0.68, p=0.018) but not in intervention group (0.86, p=0.45). However, the difference in differences was not significant for either ED visits (p=0.11) or clinic contacts (p=0.37). Results were similar when adjusting for site, sex, age, and English as primary spoken language. Overall implementation of tailored care plans within a HIE care coordination platform is feasible. Patients with diabetes and previous ED visits with a tailored care plan in a statewide care coordination HIE had fewer ED visits after implementation. However, differences in post-period ED visits and clinic contacts were not statistically significant between experimental and control groups.