Online Program


Is higher hospital system spending intensity associated with better acute care outcomes?
David Alter, Institute for Clinical Evaluative Sciences 
Rick Glazier, Institute for Clinical Evaluative Sciences 
Astrid Guttmann, Institute for Clinical Evaluative Sciences 
*Therese Stukel, Institute for Clinical Evaluative Sciences 

Keywords: health system research, efficiency of care, quality and costs, sustainability

Canada's universal healthcare system provides access to all medically necessary physician and hospital care. Moderate variations in healthcare utilization and spending exist across regions. Previous US work has shown that higher regional spending was not associated with better outcomes for acute care patients. We undertook a population-based study of 5 acute care cohorts hospitalized in Ontario, Canada, to determine whether hospital systems with higher healthcare spending achieved lower rates of mortality and readmissions, and higher rates of preventive care, shared care (visit to primary care physician and specialist within 4 weeks of discharge) and prescriptions of evidence-based medications. Using health administrative data, we enrolled patients age 20-99 with an incident admission for acute myocardial infarction, congestive heart failure, hip fracture, colorectal cancer or breast cancer between 1998-2007 and followed them for 1 year. As in the US study, we removed potential reverse causality between spending and illness by using an end-of-life expenditure index that reflected the component of spending variation that was unrelated to hospital-specific illness differences. A hospital's end-of-life expenditure index (EOL-EI) was estimated as the per capita spending on hospital, physician and ED services in the last 2 years of life among decedents who were loyal to the hospital. Each cohort member's exposure to different levels of spending was then defined as the EOL-EI in their hospital of admission. We demonstrate that the hospital EOL-EI is an instrumental variable (IV) in that it is related to mean cohort spending but unrelated to mean cohort illness severity. Study patients were similar in baseline health status, but those admitted to hospitals with a higher EOL-EI received more care. Using Cox proportional hazards models with reduced form IV analysis, we showed that increased hospital spending was associated with lower mortality and readmission rates, and higher rates of shared care and prescriptions of evidence-based medications. Higher spending hospital systems tended to be higher volume, teaching hospitals where the attending physician was more likely to be a specialist or see a higher volume of inpatients, and have better coordination between ambulatory and hospital care. A universal healthcare system founded on a primary care system with central management of hospital resources appears to be able to contain costs without sacrificing quality.