Pharmaceutical analysis of inpatient prescriptions: systematic observations of hospital pharmacists' practices in the early user-centered design phase (Preprint)

BACKGROUND The healthcare sector's digital transformation has accelerated, yet adverse drug events (ADEs) continue to rise, posing significant clinical and economic challenges. Clinical Decision Support Systems (CDSS), particularly those related to medication, are crucial for improving patient care, identifying Drug-Related Problems (DRPs) and reducing ADEs. Hospital pharmacists play a key role in utilizing CDSS for patient management and safety. Human Factors and Ergonomics (HFE) methods are essential for designing effective, human-centered CDSS. HFE involves three phases: exploration, design, and evaluation, with exploration being critical yet often overlooked in literature. For medication-related CDSS, understanding hospital pharmacists' tasks and challenges is vital for creating user-centered solutions. OBJECTIVE The aim of this study is to explore the actual practices and identify the needs of hospital pharmacists analyzing electronic prescriptions. This study focuses on the preliminary stage of user-centered design of a pharmacist-centered CDSS. METHODS The study involved observing 16 pharmacists across five hospitals in mainland France, including a university hospital, two large general hospitals, a smaller general hospital, and a specialized clinic. Pharmacists were selected regardless of expertise. The observation method used systematic in situ observation with shadowing posture, following pharmacists as they analyzed prescriptions. Researchers recorded activities, tools used, verbalizations, behaviors, and interruptions, using an observation grid. Data analysis focused on modeling pharmacists' cognitive work, categorizing activities by action type, specificity, and information source. Sequential time data analysis and distance matrices were employed to generate hierarchical clustering and identify similarity groups among the pharmacists' analysis. Each group is described using its typical sequences of analysis and related covariates. RESULTS 16 pharmacists analyzed and validated electronic prescriptions for 140 patients, averaging 5.48 minutes per patient. They spend 91% of their time searching for information rather than transmitting it. Most information comes from the list of prescriptions but it's the time spent in the Electronic Medical Record (EMR) that dominates at the heart of the analysis. Pharmaceutical interventions are most frequently transmitted in the last third of the sequence. The pharmaceutical analysis were grouped into four clusters: A (22%): Interventionist clinical analysis with extensive crossing of various sources of information and almost systematic pharmaceutical interventions. B (52%): Most common clinical analysis focusing on EMR and biology results. C (13%): Logistical analysis, focusing on the pharmacy workflow and the medication circuit. D (13%): Quick, trivial analyses based exclusively on the list of prescriptions. CONCLUSIONS The pharmaceutical analysis process is complex and multifaceted. Pharmacists are detectives, accessing a wealth of information in order to discriminate DRPs and respond accordingly. They also carry out different types of analysis, which lead to different needs and require different solutions from CDSS. This exploratory study is an essential prerequisite for meeting the challenge of designing tools to support pharmaceutical analysis and pharmacists.