Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Improving patient safety during handovers has become a public health priority.1 Over the past decade, a number of best practices have emerged, which, taken together, represent the first generation of handover interventions. Largely adapted from industries (such as aviation and railroad) in which transition errors have high consequences,2 these first-generation best practices aim to reduce information loss and distortion via structured communication protocols such as face-to-face and written sign-out that use mnemonics and standardised templates, interactive questioning and distraction-free environments.1
These efforts have been fruitful. Interventions that bundle these practices have yielded improvements in educational and clinical outcomes.3 Yet, while these protocols improve safety, handovers still remain an important source of medical error and potential harm to patients. Accordingly, we must now choose how best to identify strategies that improve upon these first-generation interventions. In our view, since handovers are a complex cognitive task, these efforts will require deeper appreciation of human cognitive abilities. The sender and receiver must simultaneously apply and integrate multiple sets of (clinical, communication and systems) knowledge, skills and attitudes into one, time-limited and highly constrained activity.4 As a result, the task demands can easily exceed the information-processing capacity of the clinicians, resulting in impaired learning and performance, errors and harm to the patient.5
To date, handover research and practice has not explicitly applied cognitive theories of learning and information processing to address these cognitive limitations. One such theory, cognitive load theory (CLT), has received increased attention in the medical education literature,5 ,6 and, in a recent study, has been used to unpack the complexity of handovers.7 By highlighting the constraints of human beings’ working memory, we believe that CLT identifies specific cognitive limitations highly relevant to handovers, and can help guide a second generation of handover research and new best practices.
Prevailing models of human memory, as initially developed by Atkinson and Shiffrin,8 include three subsystems of memory: sensory, working and long-term memory. CLT focuses on one of these subsystems, namely, working memory. While sensory and long-term memories are relatively capacious, human beings’ working memory can only hold a limited number of independent information elements at a time (4–7, ±2),9 and can actively process not more than 2–4 elements at any given moment.10 As a result, working memory is viewed as the ‘bottleneck’ for information processing. CLT identifies three types of cognitive loads that consume limited working-memory resources:11 (1) intrinsic load (task related; eg, the complexity of the patient(s) being signed out), (2) extraneous load (task unrelated; eg, interruptions or background noise) and (3) germane load (learning related; eg, asking a clarifying question). Some CLT researchers prefer a two-load model in which germane load is understood as a subset of intrinsic load.
With these three types of cognitive load in play, CLT focuses on how the total load of a task (sum of the three types of load) compares with the clinician's working-memory capacity. When the sum of intrinsic and extraneous load surpasses a clinician's working-memory capacity, his/her performance suffers, and the individual has no remaining working-memory resources to allocate to germane load, and thus, learning suffers as well. In order to regulate cognitive load and, thereby, enhance performance and facilitate learning (schema construction and automation),5 CLT recommends the use of three primary strategies:
Titration of intrinsic load to the developmental stage of the clinician. For example, simplifying the task can be accomplished by reducing complexity, increasing time allotted or, when necessary, decomposing (often via simulation) the tasks.
Reduction of extraneous load. For example, minimising distractions such as interruptions.
Optimisation of germane load. Encourage the use of strategies that facilitate construction of an accurate mental model of the patient, such as activating prior knowledge, compare/contrast and monitoring understanding and performance.
Cognitive load in handovers: implications for second-generation strategies
The goal of a clinical handover is to transfer the sender's mental model of the patient and his/her clinical responsibility to the receiver.12 Communication may occur through a number of methods, including visual (electronic data, notes) and auditory (telephonic or face-to-face direct communication).13 This is a complex cognitive task, and a CLT approach yields important insights and recommendations for managing such complexity.
Intrinsic cognitive load in handovers
The intrinsic load is imposed by the essential features of the task (table 1). Intrinsic load arises from the mental activities of selecting, organising and integrating the relevant words and images in order to perform the task.6 Intrinsic load depends on:
Number of information elements. Signing out more patients or assigning more tasks increases load.
Amount of time available. Signing out in the context of a life-threatening process such as severe trauma or an acute coronary syndrome demands faster processing, which consumes more working-memory resources.
Extent to which the information elements interact with each other. Cognitive load increases during handovers when there are major uncertainties (diagnostic, therapeutic and/or informational contingencies), disease/therapeutic interactions (eg, disease–disease, drug–drug and disease–drug interactions) or a limited evidence base for the disease.
Participant's knowledge and expertise. The intrinsic load of a handover increases when the sender or receiver has less experience with a disease, patient type or problem. A more knowledgeable clinician already possesses and immediately activates the relevant illness script that vastly simplifies the task (eg, many symptoms and signs are reduced to a single information unit, such as diabetes poorly controlled in the context of poor adherence). A novice, lacking such illness scripts, experiences higher intrinsic load.
The first-generation best practices do not focus on intrinsic load. Second-generation strategies should identify the drivers of intrinsic load that affect handover errors and performance the most, including the relative impact of number of patients, complexity of patients (eg, comorbidities), uncertainties in diagnosis or treatment and time compression (table 2). If research were to identify critical thresholds for a given level of trainee or attending (eg, number of patients with significant uncertainties in diagnosis or treatment or number of patients designated complex), beyond which, handover performance significantly deteriorates, electronic algorithms could be developed to identify when a panel of patients to be handed off has crossed the threshold, and is at elevated risk for handover errors. In these situations, more time could be allocated to the handover, more experienced clinicians activated and/or some of the patients reallocated to a different team/clinician. One early study demonstrated a technique to balance the mental workload associated with outpatient panels at the point of academic year-end transfer, and reported 50%–61% reduction in intercaseload variation for mental workload.14
Similarly, we know that knowledge and expertise is an important determinant of intrinsic load. Second-generation interventions should explore what type of experience is most important in modulating intrinsic load and reducing handover errors, whether it is general experience (such as years in practice) or experience with the specific type of diseases or other contextual factors. If proxies for general experience are inadequate then future handover protocols may need to certify clinicians for specific contexts or illnesses, and handover trainings may need to become more context-specific or disease-specific.
Extraneous cognitive load in handovers
Extraneous load results from any task-unrelated elements such as features of the task design or of the environment (table 1). These include:
Single-channel versus dual-channel communication. Extraneous load increases when communication occurs only across one channel (auditory or visual). Because working memory has partially independent channels for auditory and visual information, human beings can process more elements when they are distributed between the two channels.6
Information search. Extraneous load also rises when information is separated in space and/or time (important clinical data must be acquired from multiple different sources), and working-memory resources are then dedicated to the search for that information.
Distractions. Extraneous load also results from environmental sources such as interruptions15 or background noise or internal sources such as preoccupation with a personal concern.
From the perspective of CLT, the key components of the first generation of handover best practices focus heavily on reducing extraneous load (table 2). For example, standardising the communication process with mnemonics for verbal communication and templates for written communication decreases extraneous load by distributing information across the auditory and visual channel, presenting information in a predictable sequence and reducing the amount of handover-irrelevant information. Similarly, the use of electronic technologies to integrate all relevant information into a single space decreases information search, and the recommendations for the design of schedules and the environments minimise fatigue (eg, duty hours), interruptions and noise. In other words, the efficacy of current best practices may be due to decreasing the amount of working-memory resources diverted to non-essential aspects of the handover, such as locating the relevant information or negotiating how to structure the communication and the transfer of responsibility.
Second-generation handover practices should focus on additional strategies that expand the focus to internal sources of distraction and the clinician's ability to manage distractions (table 2). Are internal distractions such as preoccupation with a personal matter or how one is being perceived important drivers of extraneous load and errors? We should test techniques, such as mindfulness, deep breathing or attention management that may improve the ability of a clinician to screen out distractions (eg, internal anxiety or external noise), and thereby reduce extraneous load and improve performance. This represents a novel approach to managing extraneous load; rather than focusing on controlling external factors, these interventions focus on clinician's skills that can mitigate the impact of distractions, including internal ones. These may effectively augment first-generation strategies since distractions, despite efforts to minimise, will always exist in the clinical environment.
Germane cognitive load in handovers
Germane load involves clinician's deliberate concentration, including the use of cognitive strategies to understand, remember and generalise from experience; in other words, to learn (table 1). Unlike intrinsic or extraneous load, germane load is regulated by the individual—the clinician chooses how much effort to give to his or her performance and learning. Self-regulated learning occurs at multiple levels, including (1) effort to understand the language, diseases, concepts, numbers and acronyms; (2) effort to remember the cases and (3) effort to draw generalisable lessons from the handover. Strategies include self-monitoring to identify when inadequate understanding exists, asking clarifying questions, holding several different representations in working memory in order to compare and contrast and activating prior knowledge about the patient or the disease. If intrinsic and extraneous loads are too high, there will be insufficient working memory available for germane load, and no learning will occur.
Some of the first-generation best practices address germane load (table 2). For example, the opportunity to ask clarifying questions or the step in I-PASS,3 which prompts the receiving clinician to synthesise what he/she has heard. These actions help improve understanding. Second-generation interventions should test other types of interventions that increase germane load and thereby improve performance and learning. Promising avenues of investigation include to what extent do self-regulation or metacognitive skills, such as self-monitoring of understanding and slowing down when you should,16 or active listening17 improve the ability of a given clinician to process, understand and retain the information during a handover, and recognise and address potential gaps in understanding or meaningful differences of opinion. If the use of these germane load optimising behaviours improves handover quality then clinician training and protocols may be enhanced to promote these behaviours.
The emphasis on improving the quality of handovers by standardising the communication process is welcome, and has been associated with improved performance.3 However, these efforts, which primarily focus on reducing extraneous load, do not fully account for other significant drivers of cognitive load and handover errors. By focusing attention on the limits of working memory for both the sender and receiver, CLT identifies dimensions of all three types of cognitive load not addressed by first-generation protocols.
The second generation of handover research, strategies and best practices should embrace CLT and strive to optimise all three types of load. The resulting interventions, when added to the current best practices, could yield a much more potent handover bundle. A better understanding of the drivers of cognitive load would open the ‘black box’ of clinician's thinking and performance in handovers. This should enable us to maximise cognitive strengths while augmenting and bolstering cognitive limitations in handovers.
Contributors JQY, RMW, PSO, OtC and DMI all contributed to the planning, conduct and reporting of this article.
Competing interests None declared.
Provenance and peer review Not commissioned; internally peer reviewed.
- Original research
- Original research