Medical education should teach heuristics rather than train them away.

CONTEXT: Ample research suggests that most decisions are based on heuristics-simple rules of thumb-that violate prescriptions of logic and probability theory and should therefore be avoided. Yet findings on decision making in everyday work contexts support the idea that heuristics are in fact the very basis of good decision making if adapted to the challenges and performance criteria of the specific work domain. Because traditional pedagogies aim at circumventing heuristics in (clinical) decision making, ways in which to improve the use of heuristics via (medical) education have rarely been explored. OBJECTIVE: To describe the rationale for teaching and learning proper use of heuristics, rather than stigmatising them, and to identify principles and potential implications for the design and improvement of pedagogies for training in clinical decision making. RESULTS: Based on theory and evidence concerning heuristic decision making in everyday work domains, we suggest that heuristics should not be avoided as irrational or a mere source of errors, in particular in domains where errors are unavoidable. Instead, we should teach and learn how to use heuristics to make fewer and 'smarter' mistakes rather than 'dumb' ones. Based on concepts borrowed from signal detection and control theory, we demonstrate that, to improve heuristic decision making, teaching should focus on differential diagnoses and learning from feedback and mistakes, in teams and in contextually rich settings where the frequencies, costs and trade-offs between different types of errors (misses versus false alarms) can be experienced. We discuss three possible teaching formats and how to best implement them based on our findings. CONCLUSIONS: The most promising way to train (future) physicians and other health professionals in clinical decision making is not to circumvent heuristics or correct deviations from logic and probability theory but to enhance the use of heuristics by improving perspicacity, that is, by tuning the (recognition) processes that underlie the domain-specific adaptive selection of heuristics and management of ensuing errors.

The heuristic version of Cognitive Work Analysis: A first application to medical emergency situations.

Cognitive Work Analysis is an original method that seeks to describe work systems made up of nested sets of constraints, from ecological constraints imposed by the work domain to cognitive constraints. This top-down approach starts with a work domain model in order to analyze and specify contexts of activity. To complement this method, we propose a bottom-up version of Cognitive Work Analysis focusing on contexts of activity and depicting how operators adapt to the ecological constraints. Based on Rasmussen's Dynamic Safety Model, the ecological constraints involved are those bounding the workspace in which operators dynamically navigate with control loops, strategies, work organization, and competencies. This analysis relies on the simulation of specific contexts of activity. A first illustrative application of this framework to a simulated medical emergency situation with a team of nurses and nursing aids shows that this framework can help identify design issues.

Decisionmaking in practice: The dynamics of muddling through.

An alternative to conventional models that treat decisions as open-loop independent choices is presented. The alterative model is based on observations of work situations such as healthcare, where decisionmaking is more typically a closed-loop, dynamic, problem-solving process. The article suggests five important distinctions between the processes assumed by conventional models and the reality of decisionmaking in practice. It is suggested that the logic of abduction in the form of an adaptive, muddling through process is more consistent with the realities of practice in domains such as healthcare. The practical implication is that the design goal should not be to improve consistency with normative models of rationality, but to tune the representations guiding the muddling process to increase functional perspicacity.

Emergent features and perceptual objects: re-examining fundamental principles in analogical display design.

Two sets of design principles for analogical visual displays, based on the concepts of emergent features and perceptual objects, are described. An interpretation of previous empirical findings for three displays (bar graph, polar graphic, alphanumeric) is provided from both perspectives. A fourth display (configural coordinate) was designed using principles of ecological interface design (i.e. direct perception). An experiment was conducted to evaluate performance (accuracy and latency of state identification) with these four displays. Numerous significant effects were obtained and a clear rank ordering of performance emerged (from best to worst): configural coordinate, bar graph, alphanumeric and polar graphic. These findings are consistent with principles of design based on emergent features; they are inconsistent with principles based on perceptual objects. Some limitations of the configural coordinate display are discussed and a redesign is provided. Practitioner Summary: Principles of ecological interface design, which emphasise the quality of very specific mappings between domain, display and observer constraints, are described; these principles are applicable to the design of all analogical graphical displays.

Striving for safety: communicating and deciding in sociotechnical systems.

How do communications and decisions impact the safety of sociotechnical systems? This paper frames this question in the context of a dynamic system of nested sub-systems. Communications are related to the construct of observability (i.e. how components integrate information to assess the state with respect to local and global constraints). Decisions are related to the construct of controllability (i.e. how component sub-systems act to meet local and global safety goals). The safety dynamics of sociotechnical systems are evaluated as a function of the coupling between observability and controllability across multiple closed-loop components. Two very different domains (nuclear power and the limited service food industry) provide examples to illustrate how this framework might be applied. While the dynamical systems framework does not offer simple prescriptions for achieving safety, it does provide guides for exploring specific systems to consider the potential fit between organisational structures and work demands, and for generalising across different systems regarding how safety can be managed. PRACTITIONER SUMMARY: While offering no simple prescriptions about how to achieve safety in sociotechnical systems, this paper develops a theoretical framework based on dynamical systems theory as a practical guide for generalising from basic research to work domains and for generalising across alternative work domains to better understand how patterns of communication and decision-making impact system safety.

Management of test results in family medicine offices.

PURPOSE: We wanted to explore test results management systems in family medicine offices and to delineate the components of quality in results management. METHODS: Using a multimethod protocol, we intensively studied 4 purposefully chosen family medicine offices using observations, interviews, and surveys. Data analysis consisted of iterative qualitative analysis, descriptive frequencies, and individual case studies, followed by a comparative case analysis. We assessed the quality of results management at each practice by both the presence of and adherence to systemwide practices for each results management step, as well as outcomes from chart reviews, patient surveys, and interview and observation notes. RESULTS: We found variability between offices in how they performed the tasks for each of the specific steps of results management. No office consistently had or adhered to office-wide results management practices, and only 2 offices had written protocols or procedures for any results management steps. Whereas most patients surveyed acknowledged receiving their test results (87% to 100%), a far smaller proportion of patient charts documented patient notification (58% to 85%), clinician response to the result (47% to 84%), and follow-up for abnormal results (28% to 55%). We found 2 themes that emerged as factors of importance in assessing test results management quality: safety awareness-a leadership focus and communication that occurs around quality and safety, teamwork in the office, and the presence of appropriate policies and procedures; and technological adoption-the presence of an electronic health record, digital connections between the office and testing facilities, use of technology to facilitate patient communication, and the presence of forcing functions (built-in safeguards and requirements). CONCLUSION: Understanding the components of safety awareness and technological adoption can assist family medicine offices in evaluating their own results management processes and help them design systems that can lead to higher quality care.