Effective use of clinical decision support in critical care: using risk assessment framework for evaluation of a computerized weaning protocol

Background: Clinical decision support aids such as computerized weaning protocols (CWPs) aim to reduce medical errors and improve patient safety. However, the dynamic nature of critical care environments demands context-specific and complexity - inclusive assessment of these support tools for optimal results. Objective: To apply and validate the use of a risk assessment method called Functional Resonance Accident Method (FRAM), which is originally proposed for adverse event analysis in the aviation industry, to evaluate effective use of a CWP in a medical intensive care unit. Study Design and Methods: Multiple data collection methods including (1) ethnographic observations, (2) semi-structured interviews, and (3) review of hospital documents related to workflow, procedures, and training were used to simulate a FRAM based model of the CWP and identify factors affecting its use. Subsequently, we validated our findings by shadowing clinicians during 65 weaning attempts (120 h of in vivo data). Results: The factors posing risk to effective use of CWP included misinterpretation of CWP's sedation assessment scale, communication and collaboration breakdowns, problems with on-time support delivery, and negative perception of the protocol among clinicians. During the in-situ validation, we found that 45 of the 65 attempts were favorable, 16 fell under near-miss category, while the remaining four were unfavorable. Conclusions: Non-linear risk assessment method based on resilience engineering concepts is an effective approach for identification of factors for safe use of decision support aids in the real-world health care environment.