Data-Driven Methods for the Characterization of the Implementation of Evidence-Based Medical Practices


The goal of the work reported here was to answer two important questions with regards to LTVV use for patients with ARDS: 1) How do we measure adoption? and 2) What are the drivers of provider adoption? To this end, I have demonstrated the influence of patient height, hypoxemia severity, and ARDS documentation on tidal volume selection for ARDS patients. I have shown evidence that the association of patient height with standardized tidal volume is not an ARDS-specific phenomenon, but instead is an effect of mechanical ventilation for hypoxemia. This finding suggests the clinician use of a simple height-based heuristic for tidal volume selection. Further, I have validated these associations in an international cohort, implying that my results are generalizable to the patient population worldwide. Then, I provide methods to measure ARDS recognition at both the population and individual clinician level that account for these effectors. Using this metric, I show that local team-based culture is a stronger driver of ARDS recognition than specific position within an interaction network, which raises questions about the previously utilized opinion leader targeting approach for adoption interventions. Finally, I demonstrate that different local cultures report different barriers to implementation and that engagement with the studied innovation should be considered when evaluating the importance of specific reported barriers. In summary, this work provides methods for the characterization of the adoption process as well as specific insights for the design of future implementation interventions.

Alternate Identifier
Date created
Resource type
Rights statement