Background: Treatment in mobile stroke units (MSUs) is associated with faster time to thrombolysis in acute ischemic stroke than treatment in conventional (non-MSU) ambulances. The optimal approach to automating the comparison of both treatment modalities is unclear.
Methods: Using structured query language (SQL) queries on our enterprise data warehouse (EDW), we extracted key treatment metrics from all patients who were transported on the MSU and conventional ambulances at New York Presbyterian Hospital (NYP) between October 2016 and November 2018. We identified the conventional care group by ingesting an external vendor feed into the EDW and using SQL to join pre- and in-hospital records to select patients that were transported by ambulance for suspected stroke and were evaluated by the stroke team in the hospital. Both query sets were automated to run at regular intervals, and we used a web-based analytics dashboard to visualize both query outputs.
Results: The dashboard went live in March 2018, after 8 months of development. Over the study period, we identified 83 patients treated by MSU and 123 patients treated by conventional ambulance. The lack of a common identifier between pre-hospital and in-hospital patient records posed a significant challenge and required development of a patient matching algorithm to join both data sets. Additionally, migration to a MSU telemedicine-based care model in late November 2017 introduced new clinical documentation that required modification of the MSU group database query. Multiple rounds of extensive query optimization were necessary prior to go-live to ensure acceptable refresh rates that could support use of the dashboard .
Conclusion: We automated the identification of a comparison group for the evaluation of a MSU program at a large urban health system. Our main challenge was integrating pre- and in-hospital data.
Describe the new knowledge and additional skills the participant will gain after attending your presentation.: Describe the benefits and evidence supporting treatment with mobile stroke unit for acute ischemic stroke
Describe an approach to utilizing automated database queries to evaluate MSU treatment
Describe technical difficulties involved in integrating pre and in-hospital data sources
Benjamin Kummer (Presenter)
Mount Sinai Health System
Michael Zelenetz, NewYork-Presbyterian Hospital
Michael Lerario, NewYork-Presbyterian Queens
Joshua Willey, Columbia University College of Physicians & Surgeons
David Vawdrey, NewYork-Presbyterian Hospital
Matthew Fink, Weill Cornell Medicine