Abstract T P305: Fragmented Care Among Stroke Patients at 4 Chicago Hospitals

Abstract

Background/Objective: Stroke patients may be unable to provide medical history. Given the urgency of decision-making in acute stroke and importance of medication compliance post-stroke, prompt access to prior health records could reduce delays and errors. We hypothesized that stroke patients often have fragmented care in the Chicago Health Atlas. Methods: The Chicago Health Atlas encompasses clinical data on over 1 million unique patients within the 606xx ZIP codes of Chicago. A unique ID is created for each patient through the use of a secure one-way hash function. The creation of a unique ID enables longitudinal tracking of a patient’s care across participating sites. We identified all acute stroke admissions to any of 4 Chicago Health Atlas hospitals using ICD-9 codes (430-436) and searched for prior inpatient or outpatient encounters at any of those sites. Fragmented care was defined as having prior encounters at a site different from the site of stroke hospitalization. In the same cohort, we assessed post-hospitalization outpatient encounters within the Health Atlas. Results: There were 5,980 stroke patients with emergency department or inpatient visits between 2006 and 2010. Of these, 3,732 (62.4%, mean age 61.1 years, 47% black) had non-fragmented care prior to stroke while 2,066 (34.5%, mean age 58.1 years, 31% black) were completely new encounters (first-ever) in the Health Atlas and only 182 (3.0%, mean age 57.5 years, 59% black) had fragmented care. The percent with fragmented care increased from 1.3% in 2006 to 3.3% in 2010 (p<0.001 for trend). Rates of follow-up within following 12 months varied across the 3 groups (non-fragmented: 75% and fragmented: 64% vs. new: 51.4%; p<0.001). Conclusions: Two-thirds of stroke patients have previously accessed the health care system, with a majority having records at the same site of stroke hospitalization. However, 3% of patients have fragmented care, a rate that is increasing over time. Our estimate of fragmentation may be an underestimate since only 4 sites were included in this analysis. Having health information exchange may be valuable in reducing fragmentation of care for stroke patients.