Abstract WP34: Factors Associated with Reduced use of Stroke MRI: an Analysis of Practice Patterns
Background: Guidelines support the use of Magnetic Resonance Imaging (MRI) in the evaluation of acute stroke, however the perceived value of MRI information to providers in different clinical scenarios is not known. We aimed to identify patient characteristics associated with lower MRI use as means of identifying possible clinical scenarios where providers may perceive MRI to be less valuable.
Methods: Stroke hospitalizations (principal ICD-9-CM 433.x1, 434.x1, 436) were identified from the State Inpatient Databases (SID) from 2003-2009 for 12 states. MRI utilization was identified using revenue codes and ICD-9 procedure codes. Patient characteristics were abstracted from the hospitalization record. Multi-level logistic regression with a random hospital-level intercept was used to estimate the association between patient characteristics (demographics, vascular risk factors, stroke mimics, comorbidities, regional socioeconomic status) and MRI. Effect sizes were estimated using posterior probabilities with all covariates held at their means. Sensitivity analyses accounting for the distribution of MRI contraindications were performed.
Results: 670,309 hospitalizations were included in our sample; 51% of the population received MRI. From the model (area under ROC 0.82), age and potential cardioembolic source (atrial fibrillation, congestive heart failure, myocardial infarction) were associated with a lower probability of MRI utilization (table). The predicted probability of MRI in a 50 year old was 63.7% (95% CI 61.5%-65.9%) compared to 41.6% (39.4%-43.9%) in an 80 year-old. In sensitivity analysis, the association between cardioembolic sources and lower MRI use remained significant after accounting for extreme assumptions about the distribution of MRI contraindications.
Conclusions: In routine care, MRI may have less perceived value by providers in stroke patients who are older and have a potential cardioembolic source.
- © 2012 by American Heart Association, Inc.