Comparing Strategies for Identifying Falls in Older Adult Emergency Department Visits Using EHR Data

The study compared different strategies for identifying falls in older adults during emergency department (ED) visits using electronic health record (EHR) data, with manual chart abstraction as the gold standard. It involved 500 ED visits from patients aged 65 and older. The research found that natural language processing (NLP) was the most accurate method for identifying falls, outperforming strategies based on International Classification of Diseases (ICD) codes and clinical classification (CC) data. While ICD and CC strategies were less accurate, combining them improved performance but still fell short of NLP's accuracy. The study highlighted the potential of NLP to enhance fall identification and suggested that, in its absence, code-based definitions with CC data could be used, albeit with limitations. The research was supported by grants from the Agency for Healthcare Research and Quality and the National Institutes of Health.