Article Text
PDF
Abstract
Objectives To map interruptions encountered by a senior physician performing a variety of everyday tasks on an emergency department (ED) ‘shop floor’ in the UK in order to identify tasks most likely to be interrupted, modes of interruption and those interruptions most likely to result in breaks as suspension of the original task.
Methods A self-observational audit study of interruptions was undertaken by a consultant emergency physician in a medium-sized ED over 25 separate shifts totalling 119 h. The main outcome measures were type and occurrence of interruption in relation to mode of original task. ‘Success’ of interruptions and number of outstanding tasks were also recorded.
Results 718 interruptions were recorded, with an average of 6 per hour. A mean number of 2.44 outstanding tasks were recorded on each occasion of interruption. Verbal advice, telephone calls and interpretations of x-rays were the most common forms of interruption. 498 interruptions (69%) were successful, defined as interruptions that resulted in a task break (over-riding and suspension of the original task). The most successful interruptions were calls to the resuscitation room (95%). Interruptions from electronic telecommunications systems were extensive (33% of total) with success dependent on the type of communication system. Telephone conversations were rarely interrupted (16% compared with a mean of 69%).
Conclusions Overt electronic communication systems may have a disproportionate impact in determining the likelihood for successful interruptions. Formal consideration of how to prioritise and manage interruptions from various channels could be usefully added to emergency medicine education and training.
- Interruption
- communication
- emergency medicine
- safety
- teamwork
- mental illness
- methods
- risk management
- forensiclegal medicine
- hypothermia
- interpersonal
- environmental medicine
- violence
- communications
Statistics from Altmetric.com
- Interruption
- communication
- emergency medicine
- safety
- teamwork
- mental illness
- methods
- risk management
- forensiclegal medicine
- hypothermia
- interpersonal
- environmental medicine
- violence
- communications
Background
Improvements in healthcare incident reporting systems show serious patient error to be more prevalent than previously indicated with as many as 3000 deaths reported annually in the UK.1 The fluid, unpredictable and multiprofessional components of emergency medicine (EM) make it particularly susceptible to risk with as many as 18 minor errors occurring for every 100 patient admissions.2 Government targets focused on patient throughput add additional organisation pressure in an environment where there is limited capacity for review of rapid decisions afforded to other specialities.3
Good communication is a prerequisite for safe, accurate and timely decisions required in high-pressured hazardous multiprofessional working environments such as EM.4 EM practitioners are required to negotiate particularly high communication loads5 due to complex communication patterns involving extensive multiprofessional, multispecialty and multidepartmental interaction. Time constraints, rapid turnover and complex unpredictable tasks heighten the need for effective communication.6
Error in EM is more preventable than in other hospital departments. Estimates suggest that 27–51% of patient-related incidents across healthcare environments are avoidable, but in EM this is as high as 53–82%.2 Improvements in teamwork behaviours based on effective multiprofessional communication could prevent as many as 43% of incidents.7 The study reported in this paper focuses on one aspect of communication in the emergency department (ED)—namely, the role of interruptions (defined in the context of this paper as an unplanned intervention that significantly disrupts the flow of planned work either negatively or positively) during live multiprofessional practice. Interruptions lead to increased patient risk8 and are likely to reduce workflow and efficiency in the ED.5
Fluid multiprofessional medical and nursing practice is ‘interrupt-driven’.3 This is particularly evident in unpredictable environments such as the ED where the delivery of patient care necessitates continuous, mostly synchronous (face-to-face), interruption.3 ,8 Interruptions are widely accepted and little criticised in healthcare, being viewed as unavoidable or having limited detrimental effect.9
There are clear contexts where interruptions in the ED are appropriate rather than detrimental, including interruption following sudden or unanticipated complications such as a patient who develops ventricular dysrhythmia. ED physicians are also highly skilled in judging whether interruptions require their immediate attention and the suspension of their current task. However, the wider medical culture where interruptions are predominantly welcomed by senior clinicians5 is at odds with other high-risk industries where the deleterious effects of interruptions are well documented.10 ,11
Analysis of patient safety errors in healthcare environments such as EM is predominantly focused on investigating the root cause of incidents12 with limited scope for analysing active error contributors, such as interruptions during live activity. In other industries where errors can be analysed in depth through simulation exercise or in real time, the detrimental effect of interruption is more evident.
Although the flight deck and ED are unique working environments with separate associated risks requiring different skill sets, the high intense team-based nature of activity suggests that potential lessons may be learnt. NASA researchers studying flight deck safety suggest interruptions predispose to error by distracting, diverting and disrupting the original task, with short and intense interruptions the most dangerous.10 Nearly half of aviation crew errors are attributed to lapses of attention associated with interruptions, distractions or preoccupation with one task to the exclusion of another.13 Interruptions themselves account for 7% of all aviation risk reports in the USA11 and 15% of nuclear reactor shutdowns.14
The unpredictable nature of EM care, dictated by unscheduled patient visits each requiring an extensive number of communications,6 adds to potential for unavoidable interruption. Emergency physicians regularly have to shift their attention to prioritise the unannounced arrival of critically ill or injured patients, and subsequently spend a third of their time managing three or more patients concurrently.3 As shifts get busier, risk magnifies as the number of multiple tasks accumulate and interruptions increase.15
Research in North America indicates that EM physicians are interrupted roughly 10 times per hour in the USA3 and 4.4 times per hour in Canada.15 In Australia a recent study reported an interruption rate of 6.6 times per hour.16 In this paper we report on interruptions encountered by a consultant physician in a UK ED working on the ‘shop floor’. This key role involves, among other responsibilities, the delivery of optimal and safe patient care, provision of support for junior medical colleagues and collaboration with key nursing and administrative staff responsible for patient coordination and throughput. Communication with external EDs and key hospital departments is consistent and crucial. ED staff involved in such a coordinating role are particularly vulnerable to interruption and are far more likely to have tasks suspended than those with an allocated patient load.17
This study is an audit of the frequency and type of interruptions. The necessity for appropriate interruption in EM is not questioned. Data include original (background) tasks at the time of interruption and the number of outstanding tasks. The findings will allow us to ascertain which interruptions are likely to be successful and which background tasks are the most likely to be interrupted.
Methods
A prospective self-observational case study was carried out in a UK ED seeing approximately 60 000 patients per annum. The ED is situated in a large acute teaching hospital serving a mainly rural population. Clinical areas within the ED include a three bed resuscitation room, a 12–15 bed major injuries area and a 12–15 bed minor injuries area. Adjacent to the major injuries area is a four bed observation ward and ‘see and stream/treat’ facilities. The study focused on a single consultant with extensive clinical experience. Data were collected over 25 shifts, each ranging from 3 to 10 h in length (total 119 h) and on a variation of weekdays to ensure the data remained representative of the ED. The consultant acted as his own observer, conforming with action research tradition of practitioners reporting their own practices, covertly recording interruptions and background tasks using predetermined abbreviations in a pocket book as he was interrupted.
The typology of background tasks, interruptions and abbreviations was developed and revised by the consultant prior to the commencement of the study and codes were identified for each interruption type. Successful or unsuccessful interruptions were coded with ‘<’ or ‘>’. Thus, if the consultant stopped viewing an x-ray to provide verbal advice ‘X < VA’ would be coded. Additional interruptions would be added as they occurred with a full stop indicating a task was complete before a new task commenced. This method allowed the number of incomplete tasks to be scored. In order to avoid alteration in practitioner behaviour, ED staff were unaware the audit was being undertaken. All data collected were anonymised, with no possibility for retrospective identification of staff members involved in interruptions.
Results
Typology of interruptions and background events
The typology of interruptions and background events, along with data collected from the audit, are shown in tables 1 and 2. A ‘successful interruption’ is defined as an interruption that results in a task break (the interruption taking precedence through the over-riding and suspension of the original task). The data from tables 1 and 2 are illustrated graphically in figures 1 and 2.
Interruptions (ranked by most successful interruptions)
Original tasks (ranked by percentage of background events successfully interrupted)
Interruptions occurring in the ED.
Original task at the point of interruption.
Number and type of interruptions
A total of 718 interruptions were recorded. A mean number of six interruptions occurred per hour (range 0–14). Request for verbal advice was the most common interruption (n=218, 30.5%), followed by telephone calls (n=119, 16.5%) and requests for x-ray interpretation (n=12.5%). 65.8% of the interruptions were face-to-face, with 63.3% from fellow multiprofessionals and only 2.5% from patients or relatives. 33% of the interruptions were from electronic communication systems (119 telephone calls, 61 minor injuries pager calls and 58 responses to the ED tannoy/overhead page system). Only a minority of interruptions were face-to-face requests involving an immediate clear and identifiable clinical necessity related to a patient consultation. This totalled 64 interruptions (8.7% of all interruptions) and consisted of calls to see a patient in majors (n=30, 4.1%), the resuscitation room (n=20, 2.8%) and minors (n=13, 1.8%).
Successful interruptions
Four hundred and ninety-eight (69%) of interruptions were successful, with background tasks suspended remaining incomplete. The most successful interruptions were calls to see patients in the resuscitation room (95% success with only one interruption deferred) or when the consultant was informed of a situation (94%). Other synchronous interruptions from fellow clinical staff were less successful (verbal advice 61%; help to interpret ECG 47%). Direct interruptions from patients or relatives were successful on 89% of occasions.
Successful interruption from electronic communication systems was dependent on the form of communication. Telephone and tannoy calls were almost always successful (93% in each instance), but the minor injuries pager was the second most unlikely of all interruptions to succeed (16% success with 84% of interruptions deferred).
Original tasks
On each occasion that the consultant was interrupted, the original (background and incomplete) task was noted. By far the most common original task involved writing patient notes, either in majors (n=119), minors (n=172) or occasionally in the resuscitation room (n=13). The study hospital has a paper rather than electronic notes system and, in total, documentation was being completed on almost half of the occasions when an interruption occurred (n=304, 42.4% of all background tasks). When interrupted, the consultant was also often providing verbal advice (n=73, 10.2%), speaking on the telephone (n=56, 7.8%) or engaged in patient consultation (n=135, 18.8%), either in majors (n=34), minors (n=52) or the resuscitation room (n=49).
Suspension of original task
Of the 718 background tasks interrupted, 498 (69%) were successful with the interrupting task gaining precedence. Teaching was the most likely background event to be suspended (98%), followed by documentation (92%). The consultant was also likely to be interrupted successfully during formal lunch and coffee breaks (90%). On more than half of the occasions the consultant was successfully interrupted while engaged in patient consultation. This was more likely in minors (62% success) than in majors (56%) or the resuscitation room (51%).
Interruptions attempted during an ongoing telephone conversation, which accounted for 56 events, were the least likely to succeed (16%), with the consultant completing the conversation (and task relating to it) before beginning a new or suspended task.
Number of outstanding tasks following interruption
The number of incomplete outstanding tasks at the time of each interruption is shown in table 3.
Number and frequency of outstanding tasks
On each occasion of attempted interruption, at least two tasks were considered outstanding (the original and interrupting task). On average, 2.44 concurrent tasks were outstanding following each interruption attempt. It was unlikely therefore that the consultant had more than three outstanding tasks at any one time. On only one in 10 occasions (11.4%) when interrupted were there four or more concurrent outstanding tasks. When reviewing these data it is important to note that the data collection method dictated that, on 20 of 718 occasions, the consultant was either having a coffee or lunch break rather than undertaking a specific ED-based task.
Discussion
Study limitations
This study is idiosyncratic. It was undertaken in a single unique UK ED and focused on one consultant's self-reports, limiting both the validity and generalisability of the findings. Self-reports are often seen as invalid data owing to bias and, despite extensive experience as an emergency doctor, it cannot be discounted that certain interruptions may have been missed. Aware of the problems of data collection without triangulation following the audit described here, an independent medical education researcher undertook an observation of five different consultant physicians (including the original auditor) using the same typology of interruptions and original tasks. One hundred and sixty-two interruptions were recorded over 31 h of practice (5.2 per hour). The mode of interruption and original tasks were comparable to themes from the original audit, suggesting that the findings reported in this paper replicate the experience of consultant physicians in the study hospital ED. The data were not, however, sufficient for statistical testing. Statistical testing of correlations between specific interruptions and background events or analysis of other ED staff members was also beyond the scope of this study.
Where the study focused on a doctor's work, it necessarily excluded other ED practitioners. We know that, in the last 15 years, nursing and administrative staff members have played an increasing role in both the provision of patient care and management of patient throughput, adding to the complexity and frequency of communication. Errors are common across all work groups in the ED.2 ,11 Observation of nursing staff illustrates that they are exposed to higher levels of interrupting events.17 ,18 This may have particular implications for breach management in UK departments, where a recent audit undertaken in a busy UK ED illustrated that nursing staff are involved in an average of 100.9 communication events per hour (or 1.68 per minute), with a mean number of 42 interruptions per hour.19
Despite its limitations and idiosyncracies, the results of this study compare with observational research conducted in EDs in the USA and Canada. In the studies performed in the USA a higher rate of interruptions was recorded3 ,14 while, in Canada, interruptions were less frequent.15 In both North American studies a variety of emergency physicians were observed. Variation in role and experience impact on recorded interruption. A small-scale study in Australia found that registrars, with a role that included an element of coordination, experienced 23.5 interruptions per hour compared with 8.3 per hour for junior doctors with an allocated patient load.17 Our study focused exclusively on a consultant physician.
It was evident that the consultant observer used a task management strategy when interrupted, with 31% of interruptions deferred (to be dealt with following the completion of the original preceding task). Interruptions relating to clinical necessity (such as resuscitation calls) were rarely deferred, whereas the lower priority minor injuries pager was the least successful interruption. This pattern suggests that the consultant did not defer purposeful or clinically necessary interruptions that had potential benefits for patient care. The management of interruptions reflects a personal hierarchy of attributed importance given to both interrupting and interrupted events. Although clinical necessity was likely to be the over-riding prioritisation, the consultant was also frequently involved in the management of patient throughput as the ED team attempted to see and treat 98% of patients within the UK government 4 h target.20
Prioritisation of clinical necessity did not appear to change when the consultant was reviewing patients, with a high proportion of consultations successfully interrupted. Consultations interrupted in the resuscitation room, an area reserved for actual and potentially life-threatening emergencies, may appear high (51%), but these data are likely to reflect skilled prioritisation rather than poor task management. Unlike aviation where a ‘sterile cockpit’ rule is used during critical manoeuvres,21 this is not possible in the resuscitation room where a constant flow of multiprofessional and multispeciality personnel is required to deliver care. Practitioners must judge when interruptions are appropriate and which interruptions will result in the suspension of the current task. No relevant training is provided for staff. It could be argued therefore that the management of interruptions represents an opportunity for the emergency physician to demonstrate professional artistry based on expertise.
Despite the consultant not carrying a mobile telephone, findings illustrate an apparent disproportionate significance of telecommunication systems compared with synchronous communication, indicating a more problematic area for the management of interruptions. Telephone and tannoy calls were almost as likely to result in a task break as a call to the resuscitation room. Such calls are usually answered blindly and may be harder to defer once one is engaged in conversation. Indeed, the study data illustrate that, once the consultant becomes engaged in a telephone-based task, interruptions were almost always deferred.
These findings are consistent with previous research where telephone conversations were found to be three times less likely to be interrupted than professional face-to-face discussions, and patient consultations were twice as likely to be interrupted by telephone calls than by other providers.22 The popularity of communication systems in the ED results from preference for individual contact as opposed to printed or electronic material, as well as staff bias to effective interruptive communication methods.8
As wireless communication devices increasingly permeate the healthcare setting, the potential for an increased number and complexity of interruptions is likely to occur.23 Information management systems related to patient throughput in the UK also increase the potential for interruption where government 4 h breaching targets and the time patients have been in the department are displayed electronically.
No direct evidence links interruptions in the ED to error rates, but their adverse effects in other high-risk industries is well documented where multiprofessional teams work in unpredictable and fluid environments. As indicated in this paper, certain interruptions in the ED are necessary and should be encouraged. However, a large percentage are deferred and may be unnecessary. Additional research identifying how such interruptions may be reduced or avoided may be beneficial. A USA study illustrated how the use of an electronic whiteboard as a focal point for interpersonal communication has allowed practitioners to better prioritise their tasks and avoid interruption.8 Systems reducing interruption rates should be welcomed, but there is also a need for additional research focusing on the number of interruptions occurring across staff groups in UK EDs, which are likely to vary considerably,17 ,18 as well as the detrimental effect of interruptions in healthcare settings such as the ED.
Staff members in EDs make limited efforts to account for the conflicting priorities they inflict through interruption or the potential sequelae of interrupting a colleague.8 This trend is likely to remain in a medical culture where there is limited scope for reflection on the necessity of interruptions and little explicit education for reflection in work practice. Formal consideration of how to prioritise and manage interruptions from various channels could be usefully added to EM education and would undoubtedly help in raising further awareness of communication error and of the potential detrimental effect of interrupting colleagues at inappropriate times. This would, in turn, help to inform guidance for medical or nursing staff regarding the inappropriateness of interrupting colleagues.
The consultant observer in this study noted that traditional hierarchy and location play a key role in whether or not a doctor is interrupted. Certain interruptions that are not tolerated in the doctor's office may be treated differently at the nurses' station. This has additional implications for reflection on the appropriateness and education surrounding interruptions.
References
Footnotes
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Primary survey