A multi-site cognitive task analysis for biomedical query mediation

• We present a validated task model to characterize biomedical query mediation.
• We found the biomedical query mediation process to contain similar elements to the Librarian reference interview.
• Our study participants believed that BQM tasks cannot be automated.

ObjectiveTo apply cognitive task analyses of the Biomedical query mediation (BQM) processes for EHR data retrieval at multiple sites towards the development of a generic BQM process model.Materials and methodsWe conducted semi-structured interviews with eleven data analysts from five academic institutions and one government agency, and performed cognitive task analyses on their BQM processes. A coding schema was developed through iterative refinement and used to annotate the interview transcripts. The annotated dataset was used to reconstruct and verify each BQM process and to develop a harmonized BQM process model. A survey was conducted to evaluate the face and content validity of this harmonized model.ResultsThe harmonized process model is hierarchical, encompassing tasks, activities, and steps. The face validity evaluation concluded the model to be representative of the BQM process. In the content validity evaluation, out of the 27 tasks for BQM, 19 meet the threshold for semi-valid, including 3 fully valid: “Identify potential index phenotype,” “If needed, request EHR database access rights,” and “Perform query and present output to medical researcher”, and 8 are invalid.DiscussionWe aligned the goals of the tasks within the BQM model with the five components of the reference interview. The similarity between the process of BQM and the reference interview is promising and suggests the BQM tasks are powerful for eliciting implicit information needs.ConclusionsWe contribute a BQM process model based on a multi-site study. This model promises to inform the standardization of the BQM process towards improved communication efficiency and accuracy.