To determine if MRI is cost-effective in the emergency department evaluation of suspected stroke. 

Short duration or “rapid MRI” protocols have become more common for the evaluation of ischemic stroke, but little has been published on their cost-effectiveness. 

We evaluated patients who presented to our emergency department as “Brain Attack” codes, suspected of having acute stroke. We included sequential patients who had CT perfusion (CTP) as part of their acute stroke evaluation from the years before rapid MRI was available and compared them to patients who had an 8-minute rapid MRI for acute stroke evaluation. 467 patient encounters were used to assess differences between MRI (n=168) and CTP (n=299). We used propensity score matching to balance patient age, race, NIH Stroke Scale, and final diagnosis (stroke vs. mimic). The final matched cohort included 164 MRI and 115 CTP patients. We evaluated the differences in log-scale cost of patient care to our healthcare system and the length of stay (LOS) with Average Treatment Effect for the Treated (ATET), a robust methodology to account for possible confounders.

Propensity score Kernel matching showed a significant decrease in total cost [ATET, -0.550; 95% CI (-0.952, -0.149)], pharmacy [ATET, -1.085; 95% CI (-2.010, -0.160)], and supplies [ATET, -0.196; 95% CI (-0.909, 0.518)] healthcare costs in the MRI group. There was also a marginally significant decrease in LOS days [ATET, -0.712; 95% CI (-1.49, 0.064)] in the MRI group.  

Rapid MRI in the emergency department workup of stroke results in significantly lower healthcare cost. Because our cohort was matched on potential confounders, such as stroke severity and final diagnosis, these results provide compelling evidence that emergency department utilization of rapid MRI is cost-effective. Prior to the rapid MRI, the majority of suspected strokes were admitted to the hospital, which can potentially be avoided in MRI-negative patients.