Utilization of Automated Keyword Search to Identify E-Scooter Injuries in the Emergency Department.

Background and objective Accurate identification and categorization of injuries from medical records can be challenging, yet it is important for injury epidemiology and prevention efforts. Coding systems such as the International Classification of Diseases (ICD) have well-known limitations. Utilizing computer-based techniques such as natural language processing (NLP) can help augment the identification and categorization of diseases in electronic health records. We used a Python program to search the text to identify cases of scooter injuries that presented to our emergency department (ED). Materials and methods This retrospective chart review was conducted between March 2017 and June 2019 in a single, urban academic ED with approximately 80,000 annual visits. The physician documentation was stored as combined PDF files by date. A Python program was developed to search the text from 186,987 encounters to find the string "scoot" and to extract the 100 characters before and after the phrase to facilitate a manual review of this subset of charts. Results A total of 890 charts were identified using the Python program, of which 235 (26.4%) were confirmed as e-scooter cases. Patients had an average age of 36 years and 53% were male. In 81.7% of cases, the patients reported a fall from the scooter and only 1.7% reported wearing a helmet during the event. The most commonly injured body areas were the upper extremity (57.9%), head (42.1%), and lower extremity (36.2%). The most frequently consulted specialists were orthopedic and trauma surgeons with 28% of cases requiring a consult. In our population, 9.4% of patients required admission to the hospital. Conclusions The number of results and data returned by the Python program was easy to manage and made it easier to identify charts for abstraction. The charts obtained allowed us to understand the nature and demographics of e-scooter injuries in our ED. E-scooters continue to be a popular mode of transportation, and understanding injury patterns related to them may inform and guide opportunities for policy and prevention.