Abstract 173: Use of Linked Datasets to Analyze Post-Stroke Outcomes in Massachusetts

Abstract

Introduction: Stroke is the 4th leading cause of death in the United States and continues to be a leading cause of severe long-term disability. In Massachusetts (MA) there are over 18,000 inpatient hospitalizations attributed to stroke each year. Since 2004, the CDC-funded Paul Coverdell Acute Stroke Registry (MA-PCNASR) has collected detailed clinical and in-hospital data on stroke patients at 58/72 hospitals. Overall goals of the registry are to improve stroke care by tracking and evaluating adherence to evidence based stroke performance measures. While data has shown a significant improvement in adherence over time, to date there has been no method of examining post discharge patient outcomes of re-admissions and mortality. We sought to examine the feasibility of linking this dataset to statewide administrative datasets to analyze post-stroke outcomes. Methods: MA-PCNASR patients from 2005-2008 (n=31,113) were matched against stroke cases in the MA Hospital Discharge Database (HDD) (n=73,577). De-identified records were probabilistically linked ( LinkPlus, CDC ) on five points: patient admission date, discharge date, age, gender and hospital. Once linked, patient records were matched within the HDD by medical record number and encrypted social security number to examine re-admission rates as well as previous admissions before the stroke event. Taking a conservative approach, records with a probabilistic linking score of 9 or above (out of a possible 17.6) were accepted as true matches. Patient characteristics and co-morbidities were compared between the matched and un-matched groups. Results: Overall, 82% (n=25,480) of PCNASR cases were matched to inpatient cases. There was no significant change in match percentage per year. It is expected that the presence of patients treated with t-PA and then transferred to another hospital within the un-linked cohort accounts for much of the difference seen between the groups (Table 1). Secondary matching will aim to identify these patients and link them to the final site of their inpatient care. Conclusions: Results indicate the feasibility of linking MA-PCNASR and HDD data. Preliminary linking has begun between the linked database and mortality data from vital records in order to assign 30 d, 90 d and 1 year mortality. The resulting dataset will be a valuable resource which will allow, for the first time in MA, stroke surveillance across the entire continuum of care - from the emergence of risk factors, to the index stroke event, and post-discharge outcomes- and targeted efforts at increasing adherence to measure4s that demonstrate an impact on health outcomes.