PURPOSE: Medical imaging is a large and growing component of health care expenditures. To better understand some of the determinants of imaging ordering behavior, the authors analyzed the effect of differential capacity on the imaging workup of patients with acute nonhemorrhagic stroke. METHODS: All patients at a US teaching hospital and a two-campus Canadian teaching hospital between 2001 and 2005 discharged with diagnoses of acute nonhemorrhagic stroke were identified. Billing data were linked with clinical information systems to identify all imaging studies performed, comorbidities, and patient disposition. RESULTS: Nine hundred eighteen patients at the US hospital and 1,759 patients at the Canadian hospital were included. Patients were similar in age and distribution of comorbid illnesses. The rate of MRI scans at the US hospital was more than twice that at either of the Canadian hospitals (95.75 scans per 100 patients vs 41.39 scans per 100 patients). The length of stay was significantly shorter and the inpatient mortality rate significantly lower at the US hospital compared with the Canadian hospital. A multivariate regression analysis demonstrated that only patient age and site (US vs Canada) were significant predictors of MRI use, controlling for patient gender, comorbidities, and use of anticoagulants. CONCLUSIONS: Scanning utilization varied at hospitals with differential access to scanning technologies. There was less frequent use of MRI scanning at hospitals with limited access to this modality. Patient and health system factors are important considerations when interpreting the mechanisms for this variation, its importance, and the potential relationship of imaging use with patient outcomes.
PURPOSE: Medical imaging is a large and growing component of health care expenditures. To better understand some of the determinants of imaging ordering behavior, the authors analyzed the effect of differential capacity on the imaging workup of patients with acute nonhemorrhagic stroke. METHODS: All patients at a US teaching hospital and a two-campus Canadian teaching hospital between 2001 and 2005 discharged with diagnoses of acute nonhemorrhagic stroke were identified. Billing data were linked with clinical information systems to identify all imaging studies performed, comorbidities, and patient disposition. RESULTS: Nine hundred eighteen patients at the US hospital and 1,759 patients at the Canadian hospital were included. Patients were similar in age and distribution of comorbid illnesses. The rate of MRI scans at the US hospital was more than twice that at either of the Canadian hospitals (95.75 scans per 100 patients vs 41.39 scans per 100 patients). The length of stay was significantly shorter and the inpatient mortality rate significantly lower at the US hospital compared with the Canadian hospital. A multivariate regression analysis demonstrated that only patient age and site (US vs Canada) were significant predictors of MRI use, controlling for patient gender, comorbidities, and use of anticoagulants. CONCLUSIONS: Scanning utilization varied at hospitals with differential access to scanning technologies. There was less frequent use of MRI scanning at hospitals with limited access to this modality. Patient and health system factors are important considerations when interpreting the mechanisms for this variation, its importance, and the potential relationship of imaging use with patient outcomes.