Article Text
PDF
Abstract
Background Interruptions are a part of many hospital settings. During medication administration, interruptions have been shown to lead to medication errors. Understanding interruption management strategies during medical management could lead to the design of interventions to reduce and mitigate related errors.
Methods Semi-structured interviews with paediatric nurses in an in-patient setting were used to identify types of interruptions, strategies for safe medication administration and interruption management, as well as factors influencing the interruption management strategy choice. Nurses also worked through use cases and provided verbal protocols about their strategies. To confirm and refine a framework for interruption handling, on-the-job observations were also conducted.
Results Four case studies of medication administration highlight four interruption handling strategies. Three allow the interruption: 1) the primary task is suspended so that the higher priority secondary task may be engaged immediately; 2) multi-task by dividing attention between the primary and secondary tasks; and 3) mediating the interruption with an action that supports resumption of the primary task. The fourth blocks the interruption, keeping attention on the primary task (blocking). Interviews and on-the-job observation suggest that nurses dynamically assess the primary and (interrupting) secondary tasks. They prioritise task execution based on both risk and workflow efficiency assessments. Specific interruption handling depends on both task and experience related factors.
Conclusions Paediatric nurses have developed sophisticated strategies to manage interruptions and maintain patient safety and work efficiency during medication administration. To support a more resilient healthcare system, interruption management strategies should be supported through process, task support tools and education.
- Human factors
- medication safety
- nurses
- paediatrics
- quality improvement
- hand-off
Statistics from Altmetric.com
Introduction
There are high numbers of interruptions in many hospital settings and such interruptions may lead to human error.1–16 Interruptions have cognitive costs in diverted attention and forgetfulness leading to omissions.17–20 Models of interruption have typically focused on task-switching when the primary task is suspended to attend to a secondary (interrupting) task (see figure 1). Between the secondary task alert (ie, the interruption) and commencement of the secondary task, there is an interruption lag as the interruptee prepares to and then stops the primary task execution. After completing the secondary task, the interruptee experiences a resumption lag which includes a process of reorientation to the primary task and recognising the state in which the primary task was left.21 ,22
After Trafton et al 21 the interruption and resumption process, involving a primary (interrupted) and a secondary (interrupting) task.
The research summarised in table 1 indicates that not all interruptions result in a suspension of the primary task.8 ,11 ,23–25 The interruption lag presents an opportunity for choice of interruption handling by the interruptee. After an interruption alert, the interruptee can select an interruption-handling strategy that addresses the task and clinical context and enhances workflow.
Interruption handling strategies proposed or observed
A four level taxonomy includes three strategies that allow the interruption to occur and a fourth strategy blocks it:
-
Engaging: high priority secondary task. The primary task is suspended so that the higher priority secondary task may be engaged immediately. The primary task may be resumed after completion of the secondary task.
-
Multi-tasking: similar priority of primary and secondary tasks. The interruptee multi-tasks by dividing attention between the primary and secondary tasks; both tasks are performed synchronously.
-
Mediation: high priority task generated before suspension of primary task. The interruptee mediates the interruption with an action that supports resumption of the primary task. Mediation measures usually support prospective memory (the memory that we need to make actions in the future), mark the state of task or complete a subtask of the primary task before switching tasks. Sometimes, the interruptee mediates the interruption by deflecting the secondary task to another worker. In this case, the delegation is the high priority task that is generated.
-
Blocking: high priority primary task. Primary task takes priority over the secondary task and the secondary task is blocked.
Westbrook et al 12 have shown an association of interruptions with an increased risk and severity of medication errors. Thus we demonstrate this four part taxonomy using descriptions of interruption handling strategies during medication administration. There are many types of interruptions in a hospital ward (monitor alarms, overhead announcements, etc); we focused on external interruptions initiated by colleagues or patients during medication administration. We illustrate the interruption strategies with four medication administration case studies on an acute care paediatric ward. We summarise nurses' descriptions of task prioritisation depending on the specific context of the task or on the nurses' experience. We show insight into the dynamics of interruption management and suggest possible interventions to support nurses who are frequently interrupted.
Methods
Ethical approval was granted from the institution and interview participants gave informed consent.
Six in-depth semi-structured interviews were held with individual nurses regarding medication administration errors, types of interruptions and strategies for safe medication administration in light of frequent interruptions. Critical incident technique was used to elucidate steps that were prone to error if interrupted.26 ,27 Use cases and a hierarchical task analysis were developed.28 ,29 The use cases provided the medication administration context and the hierarchical task analysis identified distinct subtasks. Understanding the subtasks supported meaningful discussion of interruptee options when an interruption alert occurred.
Six different nurses followed the use cases to simulate medication administration. As they performed medication administration tasks, they were stopped at various points and gave verbal protocols (explanations of what they were thinking) about their strategy for managing the interruption.30 Eight additional nurses were involved in interviews focusing on their personal strategies for interruption handling. In addition, a single researcher (LC) observed nurses administering medications during the research period in order to develop the evolving themes. Notes were iteratively reviewed and discussed with a second researcher (EJB). Based on these notes and reflexive journals, a modified analytic induction analysis was used to develop a framework for interruption handling.31
Results
Empirical evidence for interruption management strategies
Based on observations, four case studies of medication administration are presented that highlight the four interruption-handling strategies (see table 2). In these cases, the primary task is medication administration and the secondary task is presented through an interruption. Cases 1–3 illustrate strategies for allowing an introduction: engaging, multi-tasking and mediating. Case 4 illustrates the alternative: blocking an interruption.
Case studies of interruption handling strategies
Nurses' descriptions of factors influencing interruption handling
In interviews, nurses suggested that both risk and efficiency assessments determine task priority. The risk assessment is based on the likelihood and severity of a medication error. The efficiency assessment considers the ease of medication administration task resumption and the cost of delaying the interrupting secondary task. This determination of task priority is a dynamic subjective process based on task context and the experience of the nurse. The following summary of task and experience factors indicates the rich context of the medication administration environment. Quotes from nurses interviewed are included to add to the context.
Task-related factors
-
1. Urgency of task: Most drug administration occurs during routine non-urgent conditions. Medications that address an urgent need, like pain relief, are prioritised. Similarly, alarms that indicate urgent patient conditions are attended to immediately and primary tasks are suspended. “I always want to give pain medication as soon as possible” (participant nurse 4).
-
2. Dynamics of task: Some interruptions introduce tasks that can wait while others can only be addressed for a short time-period. For example, consulting healthcare staff come and go on the paediatric ward because they are caring for patients elsewhere. Consequently, the consultant's interruption may present the only chance to speak face-to-face with the clinician. Asynchronous communication is not well supported and there is an efficiency cost if this interruption is blocked and the nurse has to find the clinician later. Thus, interruptions from itinerant staff take high priority and often lead to suspension of medication tasks. “Dr. X never returns his pages, so I have to catch him when I can” (participant nurse 2).
-
3. Medication-specific factors: All medications are not created equal. Some drugs, like acetaminophen are very familiar to nurses and they are very unlikely to lead to a dosing error. Other drugs, (eg, chemotherapy) have greater potential to cause harm and their administration is prioritised. In case 4, the nurse blocks an interruption to continue checking her chemotherapy.
Sometimes medications are prescribed outside of routine use. These dosages are often triple-checked by a nurse; the time and extra cognitive load required often leads to prioritisation of these tasks. “Seizure meds make me nervous; they are often at high dosages so I always call the pharmacy and the doctor to check the amount” (participant nurse 3).
-
4. Patient-specific factors: The ‘family-centered’ focus of paediatric wards generally prioritises any interruption originating from a patient or parent. Additionally, medication administration may be specifically prioritised for certain patients. For example, sicker patients tend to have more complicated drug regimens and a higher risk of an adverse event. Sometimes families are difficult and the cost of an error is particularly high. “This family is really tough and if anything goes wrong, it will take hours to explain and I'll never hear the end of it” (participant nurse 2).
-
5. Task-specific factors: Some tasks have subtasks that can be suspended with reasonable support for task resumption. These tasks are easier to suspend and resume than others. For example, the nurse in case 3 performs tasks like marking a list and placing a cup to provide visual cues to remind him where he was in the medication administration process. These cues support prospective memory. Thus, as he is about to task-switch, he generates a high priority task that place-holds the primary task.
Cognitive tasks are not easy to resume after interruption. Although a note to oneself may support prospective memory, there is no way to place-hold a cognitive task. For example, checking an unfamiliar medicine puts a greater cognitive load than checking a familiar medication, like ibuprofen. Nurses performing higher cognitive load tasks are more likely to block interruptions. When they must engage with an interruption (task-switch), the cognitive task is often resumed from the beginning, which is inefficient.
Experience-related factors
-
Previous medical errors: Negative experiences with medical errors (‘near misses’ or ‘adverse events’) can affect future behaviour. “I once gave a patient the wrong antibiotic because there was another drug with a similar name in the drawer. Now, whenever I have a sound-alike drug, I am really nervous and triple check it” (participant nurse 1).
-
Habit: Routines are sometimes employed to make tasks easier and performance safer. A rigid routine supports resumption because the order of subtasks is clearly established and familiar. “I always lay out the meds on the counter this way so I don't lose track of where I am” (participant nurse 6).
Discussion
This paper introduces an integrated taxonomy for interruption management and provides illustrative examples based on empirical observation and interviews. We highlight the complexity of interruptions and nurses' ability to manage them.
Strategies
We propose a taxonomy of four strategies for the handling of an interruption; three strategies allow the interruption and one strategy blocks it. Our framework suggests that nurses use sophisticated strategies to manage workflow in a dynamic environment. This is similar to Ebright, Patterson and Saleem's seven-level prioritisation hierarchy of nursing activities.32 ,33 Their work discusses prioritisation across high-level activities; our work discusses prioritisation of tasks and subtasks within the context of medication administration.
Context
Prioritisation of tasks depends on assessment of clinical and situational workload factors. All tasks are not created equal; some interruptions present a high priority task that must be addressed immediately and other interruptions present tasks that can wait. Paediatric nurses indicate that task-specific factors and personal experience affect their choices of interruption handling.
Interventions
Interruptions foster conditions that are likely to result in memory lapses.34 Recent interventions, such as ‘no interruption zone’ signage or artifacts35–38 assume that interruptions are bad and aim to reduce or eliminate all interruptions. These interventions treat all medication tasks as equal; our findings suggest these tasks are not equal. These barriers also assume all nurses are equal and do not allow for the variation in interruption-handling skill that comes with experience and supports safe medication administration. Indeed, barriers to interruptions may actually interfere with nurses' ability to select and engage important interruptions. This may lead to inefficiencies and care delivery that is out-of-date. How can we create a system where good interruptions are allowed and bad interruptions are blocked?
First, education and a culture change are needed. Coiera and Tombs noted a bias towards interruptive communication.1 Biron described nurses' propensity to accept all interruptions.24 We also noted nurses' willingness to engage interruptions. Recognising that prioritisation of competing tasks is a constant challenge makes the reality of high and low priority tasks explicit. Education about the anatomy of interruptions and management strategies could increase awareness and improved management of interruptions. Empowering interruptees to block or mediate interruptions may improve their ability to filter distracting low priority secondary tasks. Training interrupters to consider timing their interruptions at low priority tasks or at moments when the interruptee's prospective memory is best supported, could also help. The availability of asynchronous methods of communication (email, text) to interrupters who could identify low priority interrruptions could safely decrease interruption load. Experienced nurses could teach task prioritisation to novice nurses to support acquisition of interruption management skills.
Next, like other safety-critical industries, healthcare workflows could be designed to accommodate interruptions.39–42 Boehm-Davis and Remington present a framework that considers the cognitive work at each stage of an interruption, and propose solutions for interruption-disruption reduction.22 Redesigns support nurses of all experience levels and address task factors specifically. Tasks can be designed for suspension and resumption; cues can be provided that support prospective memory or subtasks can be created such that they can be completed before task-switching.
Healthcare interruptions are here to stay. Acceptance, fostered through education, and workflow redesign inclusive of interruptions will contribute to the resiliency of our care delivery system.
References
Footnotes
-
Funding This research is supported by the National Library of Medicine T15LM009462 Training Grant: A systems engineering focus on Medical Informatics. The content is the sole responsibility of the authors and does not necessarily represent the official views of the National Library of Medicine or the National Institutes of Health.
-
Competing interests None.
-
Ethics approval Ethics approval was provided by University of Virginia Institutional Review Board.
-
Provenance and peer review Not commissioned; externally peer reviewed.
-
Data sharing statement We are able to share anonymised field notes by email request.