Grant: Project Grant

Drug injection related-bacterial infections (skin/soft tissue infection, bone/joint, heart valves; hereafter IRBI) are one of the main causes of emergency room (ER) consultations for people who inject drugs (PWID), and appear to be increasing in incidence in the context of the overdose crisis. The clinical outcomes of IRBI consultations are often unfavorable in PWID. The recent establishment of a data lake (storage repository of raw data) at the CHUM, a university hospital serving a large proportion of PWID in Montreal, opens up opportunities to use big data to understand and respond to drug-related harm. However, because information on drug use is not collected in a standardized manner in hospital charts, there is a need for foundational work to better identify PWID in the data. The goals of this project are to: use the data lake to develop a cohort of patients who consulted the ER for a bacterial infection between 2012 and 2021; develop and validate an algorithm to identify PWID in this patient cohort; and compare trends in consultations between PWID and non-PWID. Among PWID, we will identify predictors of poor system-related outcomes, defined as intensive care unit admission, leave against medical advice, re-consultation and/or re-hospitalization at 7, 30 and 90 days. The algorithm developed for this project is critrical for the development of future big data research projects on drug-related harm. Evaluating changes in IRBI over time will enable exploration of the impact of changes in the drug use landscape in recent years, and will help to improve public health interventions and community services. Identification of predictors of poor system-related outcomes will inform the development of interventions in the ER to improve the intra- and post-hospital trajectory of patients.