Abstracts

Rationale: Medical information fragmentation occurs when patients seek care at multiple hospitals that do not have common medical record systems, and places patients at risk for medical errors, duplicate testing, and increased costs. Among patients with epilepsy, the degree of medical information fragmentation and the affected visit types are not well understood. We used a novel modification of a medical information fragmentation measure (Bourgeois et al 2010) to assess the degree of information fragmentation for three visit types (outpatient, emergency department, and inpatient). Methods: We extracted a de-identified data set from NYCLIX (New York Clinical Information Exchange), a regional health information organization (RHIO) that draws clinical data from several New York City hospitals. We defined a cohort of 8660 people with epilepsy as patients with an ICD9 code of 345.x during the one-year period April 2009 - March 2010. We extracted the year and location of three visit types (outpatient, emergency department, and inpatient) over a three-year period (April 2009 - March 2012). We created a visit type specific fragmentation index to answer the following question: among patients who had a specific visit type at least twice, what percentage made that visit type at multiple hospitals? For each visit type, we calculated the fragmentation index as the number of patients with that visit type at multiple hospitals divided by the number of patients with at least two of that visit type. To assess how many patients may have benefited from electronic health information exchange, we reviewed the NYCLIX usage logs. NYCLIX was only accessed by a pilot group of emergency department users during the study period. Results: The fragmentation index was 36% for outpatient visits (i.e. 1615 patients made outpatient visits at multiple hospitals out of 4440 patients with at least two outpatient visits), 20% for emergency department visits (593 of 2993 patients), and 13% for inpatient admissions (397 of 3152 patients). The fragmentation index for outpatient visits was significantly higher than for emergency department visits and inpatient admissions (test of proportions; p < 10^-15 for both comparisons). Clinical information was accessed via NYCLIX for 0.5% of the cohort (44 of 8660 patients) during 72 of their emergency department visits. Conclusions: We used a novel information fragmentation index to show that people with epilepsy frequently seek care at multiple sites for outpatient, emergency, and inpatient care. We found that information fragmentation for outpatient visits is significantly worse than for emergency and inpatient visits, both in terms of the number of patients affected and the degree of fragmentation. These data suggest that to improve care continuity for patients with epilepsy, health information exchange networks should prioritize efforts to share clinical information in the outpatient setting.