Introduction: Higher hospital volumes have been related to better outcomes in previous researches. However, the mechanisms of the volume-outcome relationships in stroke have not been sufficiently figured out. The aim of this study is to uncover the roles of processes in acute phase stroke care among the volume-outcome relationships.
Hypothesis: Processes of care mediate the volume-outcome relationships in stroke patients.
Methods: We analyzed 45,598 acute ischemic stroke patients in 267 certified training hospitals in Japan. Data between April 1,2011 and May 31, 2012 was obtained from Japanese Diagnosis Procedure Combination Database. We divided annual hospital case volumes into two groups: high-volume and low-volume. Main outcome was 30-day mortality. We set following process of care as mediators: early initiation of aspirin prescription (less than 24 hours) and early initiation of stroke rehabilitation (less than 3 days). Multilevel mediation models were performed adjusting for age, gender, comorbidities, Charlson comorbidity index to examine the relationships among hospital volume, processes of care and outcome.
Results: The early initiation of aspirin increased from 19.4% to 24.1%, the early initiation of rehabilitations increased from 59.6% to 65.6% and 30-day mortality decreased from 6.1% to 4.2%, for low-volume and high-volume hospitals. In the multilevel logistic regression analyses, admission to high-volume hospitals had 38% increased odds (OR = 1.38; 95% CI, 1.02-1.88) and 118% increased odds (OR = 2.18; 95% CI, 1.59-2.98) of early aspirin and rehabilitation initiation, compared with low-volume hospitals. Before adjusting for processes of care, admissions to high-volume hospitals had 30% decreased odds of death compared with admission to low-volume hospitals (OR = 0.70; 95% CI, 0.61-0.81). When processes of care were further adjusted, the volume-mortality relationship became non-significant. The early initiation of aspirin and rehabilitation accounted for 71% of the relationship (the regression coefficient was reduced from -0.355 to -0.102 when adjusted for processes of care).
Conclusions: Processes of care mediated the volume-outcome relationships in stroke patients.