Document resource
Documentation practices for regional anesthesia vary widely due to limited guidance, which can result in missing and inaccurate data capture. This study developed a natural language processing (NLP) algorithm to identify regional anesthesia in unstructured clinical notes, and compared results to the current referent for regional anesthesia research data (Corporate Data Warehouse structured data). 1 The analysis included postoperative notes from elective non-cardiac surgeries at one of six Veterans Health Administration hospitals in California between January 1, 2017 and December 31, 2022. Results showed that the algorithm identified 96.6% of the regional anesthesia cases recorded in the referent, with a low false negative rate of 0.8% and an accuracy of 82.5%. Notably, the NLP algorithm found more than twice as many regional anesthesia cases as the referent, indicating that there might be issues with the accuracy and completeness of documentation in research databases. This suggests that NLP is a promising tool for improving the completeness and accuracy of clinical documentation.