Clostridioides difficile infection (CDI) is an important healthcare-associated infection and a significant driver of poor patient outcomes and preventable cost. An existing bundle of antimicrobial stewardship (AMS) methods of CDI prevention are limited by difficulties identifying high-risk patients most likely to benefit from interventions. Recently we have developed a machine learning classification tool capable of accurately identify risk of developing CDI in hospitalized patients. Here we propose adding and rigorously testing this highly innovative precision medicine approach to CDI prevention within a robust, real-world AMS infrastructure Objective: To decrease CDI incidence by implementing an electronic health record-integrated CDI-risk classification tool within a framework of precision-medicine AMS.
Specific Aim : Measure the effect of implementing a real-time CDI-risk classification tool to guide AMS efforts for CDI prevention among high-risk patients. The primary outcome will be hospital-associated CDI incidence. We will measure outcomes in patients identified as high-risk by the risk classification tool, comparing a 24- month pre-implementation period with a 12-month post-implementation period using interrupted time series (ITS) segmented regression. Secondary outcomes will include antimicrobial utilization rates, CDI test ordering, hospital length of stay, total cost, tool use, and AMS satisfaction Hypothesis: We hypothesize that tool implementation will result in a 35% relative reduction in CDI incidence, which aligns well with the lower end of estimates of bundle effect in prior studies. Significance: While evidence supports the efficacy of AMS interventions for CDI-prevention in general, our proposal is the first to our knowledge to employ a precision-medicine approach to CDI prevention. This high- impact, precision medicine proposal, based on extensive, sound preliminary data has a high probability of success.
Clostridioides difficile infection (CDI) is an epidemic in hospitalized patients, and current prevention strategies are limited by difficulties identifying patients most likely to benefit from interventions. We developed a machine learning classification tool capable of accurately estimating risk of CDI. This study will use the tool to target antimicrobial stewardship CDI-prevention bundle recommendations toward patients at highest risk who are missed by current case finding strategies and investigate whether this will decrease rates of CDI.
