Abstract
BACKGROUND: Though evidence is limited in Japan, clinical controlled studies overseas have revealed that specialized care units are associated with better outcomes for acute stoke patients. This study aimed to examine the effectiveness of hospital functions for acute care of ischemic stroke on in-hospital mortality, with statistical accounting for referral bias. METHODS: We derived data from a large Japanese claim-based inpatient database linked to the Survey of Medical Care Institutions and Hospital Report data. We compared the mortality of acute ischemic stroke patients (n = 41 476) in hospitals certified for acute stroke treatment with that in non-certified institutions. To adjust for potential referral bias, we used differential distance to hospitals from the patient's residence as an instrumental variable and constructed bivariate probit models. RESULTS: With the ordinary probit regression model, in-hospital mortality in certified hospitals was not significantly different from that in non-certified institutions. Conversely, the model with the instrumental variable method showed that admission to certified hospitals reduced in-hospital mortality by 30.7% (P < 0.001). This difference remained after adjusting for hospital size, volume, staffing, and intravenous use of tissue plasminogen activator. CONCLUSIONS: Comparison accounting for referral selection found that certified hospital function for acute ischemic stroke care was associated with significantly lower in-hospital mortality. Our results indicate that organized stroke care--with certified subspecialty physicians and around-the-clock availability of personnel, imaging equipment, and emergency neurosurgical procedures in an intensive stroke care unit-is effective in improving outcomes in acute ischemic stroke care.