Grant: Travel Awards - Institute Community Support

Surgical site infections (SSIs) are one of the most common complications following surgery and represent a major threat to patient safety. Patients with SSIs stay in the hospital an average of 11 days longer, and the cost of treatment is about five times higher. Studying SSIs could dramatically reduce healthcare costs, but one necessary first step is to accurately identify them. Traditional surveillance methods rely on hospital discharge codes and manual chart reviews by infection control experts. However, it is well known that chart reviews often require extensive human resources and are time-consuming, and the ICD codes for SSIs based on patient discharge databases are often under-reported and inaccurate, which makes population-level monitoring impossible. Electronic medical records (EMRs) contain detailed information regarding various aspects of patient care and have been widely implemented across Canada. EMRs include doctors' and nurses' notes, which are a valuable resource for timely detection of SSIs. However, these free-text notes have not been leveraged for SSI detection because traditional analytical tools cannot analyze unstructured text data. To bridge this gap, this study developed and tested computer algorithms using artificial intelligence and machine learning to automatically identify infections after hip and knee replacement surgery by analyzing both coded hospital data and free-text clinical notes. When compared to expert review, the algorithms achieved 96% accuracy and could reduce chart review workload by up to 95%. These results show that computer algorithms can support infection detection and enable population-level SSI surveillance, ultimately improving patient outcomes and helping hospitals use resources more efficiently.