Clinical risk and collective competence in the hospital emergency department in the UK

https://doi.org/10.1016/S0277-9536(99)00441-4Get rights and content

Abstract

The hospital emergency department (ED) is a risky environment, often subject to litigation for negligence. Risk is defined as an avoidable increase in the probability of an adverse outcome for a patient. With the aim of identifying the sources of risk, this study carried out participant observation and collected critical incidents in two EDs in the UK for a period of 30 months. Active failures included delay in beginning investigations or treatment, failure to obtain diagnostic information, misinterpretation of diagnostic information and the administration of inappropriate treatment. Three latent conditions underlay these failures: patients' unrestricted access to the ED, cognitive errors by individual members of staff and a strict horizontal and vertical division of labour. An analysis of the incidents resulting from the third latent condition identified a contradiction between the division of labour and working conditions in the ED. The paradigm circumstances under which this contradiction can result in active failures are described. The management of risks arising in this way could be improved by developing a workplace culture in which ‘sapiential authority’ — authority derived from experience, special access to information or being at hand in an emergency — is recognised in addition to authority derived from a formal status. However, as long the contradictions between the division of labour and working conditions remain, accidents should be considered normal events.

Introduction

In the UK the hospital emergency department (ED) provides medical care for members of the public who are suddenly taken ill, or believe they are ill. A wide spectrum of medical and surgical conditions is encountered, ranging from life-threatening illnesses to minor injuries. When the patient reports to the reception desk, a patient record (the ‘cas-card’) is created and the patient is triaged — examined by a nurse and given a clinical priority as an emergency (red), very urgent (orange), urgent (yellow), standard (green) or non-urgent (blue) (Mackway-Jones, 1997). The cas-card is colour-coded accordingly and this determines the patient's initial progress in the department. Emergency, very urgent and urgent cases are taken at once to treatment areas where life support facilities are available, while standard and non-urgent ones go to a waiting room to be called in order of priority.

EDs are the subject of a considerable amount of litigation for negligence, originating 8.5 per cent of all claims in England and Wales (Bevan Ashford Solicitors, 1996). In response to the rising tide of litigation in the health service generally, the British government has called for the implementation of risk management procedures (NHS Executive, 1996). In the present paper, we report an empirical study of two EDs, with the objectives of clarifying the nature of risk and its management in this context and contributing to the theory of competence and error in complex, dynamic environments. The need for the research arises from the limited number of previous studies of the fundamental causes of risk in EDs. Previous research has analyzed patients' complaints (Kadzombe and Coals, 1992) and their reasons for beginning law suits (Trautlein et al., 1994, Rusnak et al., 1989). However, because a variety of factors are included in the decision to complain or sue, these studies do not give a full picture of the root causes of risk in the ED.

Three bodies of social science research provide the theoretical background for the present enquiry: studies of the way work is organised in EDs, studies of the relations between formal and informal organisational structures in EDs and the theory of organisational accidents.

In a recent study of the way work is organised in EDs, Dodier and Camus (1998) draw attention to a fundamental contradiction. Access is unrestricted, resulting in large numbers of attenders including many patients who are not ill or who would be more appropriately treated by their GPs. However, the departments are responsible for dealing with life-threatening conditions, which require immediate attention, so it is necessary to place patients in order of priority. The official procedure for this is nurse triage, described above. However, several sociological studies have found that staff not formally responsible for prioritizing patients vary their own “mobilisation” according to their perception of the patient's needs. For example, Hughes (1980) describes how ambulance crew judge the urgency of each case using information obtained while transporting the patients to hospital. Depending on these judgements, they drive faster or slower and either rush the patient to the resuscitation room on arrival or take time to complete the registration formalities first. Other staff involved in the pre-processing of attenders, such as receptionists, also influence the prioritization process (Hughes, 1989). Several studies have examined the category systems which staff use to make their unofficial judgements about priority. Jeffery (1979) found that they construed patients on a continuum ranging from ‘good or interesting’ to ‘bad or rubbish’, the good patients being those who provided the medical and nursing staff with opportunities to practise their specialist skills, while the bad patients were responsible for their own illnesses or refused to co-operate. These judgements affected the treatment as well as the priority that was given. Taken overall, these findings demonstrate that what happens to patients in the ED is not determined solely by rational biomedical decisions by those with formal authority. It is also affected by social practices through which all grades of staff express their moral judgements about the patients.

Sociologists have also investigated the relationships between the formal and informal structures within the ED. The formal division of labour assigns responsibility for making diagnoses and deciding on treatment to the doctor and (unless special arrangements have been made) restricts the nurse's role to ‘nursing’ duties. However, EDs are staffed by a mixture of experienced nurses (many with specialised qualifications in Accident and Emergency Nursing) and relatively inexperienced junior doctors (most of whom are appointed within one year of registration, do not intend to make emergency medicine their specialism and will leave after six months to take up other posts). Many of the nurses acquire skill in medical tasks such as interpreting X-rays and become familiar with organisational routines such as the protocols for treating different conditions. This contradicts the official division of labour and several researchers have investigated how the contradiction is resolved. Drawing on his own experience as a doctor, Stein (1967) describes the ‘doctor–nurse game’, in which experienced nurses use non-verbal communication and cryptic remarks to suggest actions to inexperienced doctors without explicitly challenging their formal responsibility. An ethnographic study by Hughes (1988) revealed that diagnosis and treatment decisions were sometimes predetermined by nurses, using their influence over the movement of patients through the department. Thus directing a patient to one room rather than another could express the nurse's judgement about the nature of the illness. Pre-processing work, such as carrying out routine observations, obtaining information about the patient's condition from a carer and preparing the patient for the doctor's arrival, offered similar opportunities. For example, Hughes (1988) observed that nurses would often prepare equipment for a particular treatment before the doctor had directed which treatment should be given. While not challenging the doctor's formal responsibility, this provided cues for inexperienced doctors about how (in the nurse's view) the case should be managed. Another example of the doctor–nurse game reported by Hughes (1988) was when a nurse asked a patient key questions while the doctor was carrying out a physical examination. This generated diagnostic information, which the doctor might not have elicited otherwise. Sometimes, nurses exerted influence on doctors more directly: Hughes (1988) reports cases of nurses offering advice in an open and straightforward way, or even intervening bluntly to point out shortcomings of the work of certain junior doctors.

Recent years have seen important developments in the theory of organisational accidents. While early investigations concentrated on errors by individual employees, today the focus has shifted to failures at the level of the system. As Moray (1994) puts it, “the fundamental claim is that the systems of which humans are a part call forth error from humans, not the other way round” (p. 37). An influential work in this field is Turner (1978), which analyses accidents in complex organisations and concludes that disasters are often the end point of a long chain of events. Early indicators of the coming disaster are often overlooked, which Turner (1978) attributes to organisational norms about the riskiness of the operations. Reason (1997) introduced an important distinction between active failures and latent conditions. Active failures include slips by individuals (such as picking up the wrong ampoule from a shelf) and mistakes (such as lack of knowledge of a protocol). Latent conditions, on the other hand, are characteristics of the organisation which do not lead directly to accidents, but may do so indirectly. Reason compares latent conditions to “resident pathogens” in the human body. They lie dormant for long periods until they combine with other factors to penetrate the defensive measures designed to prevent accidents. Building on Reason's theories, Vincent et al. (1998) produced a model of accident causation in health care, in which the important latent conditions are institutional context, organisation and management, working environment, team factors, staff factors, task factors and patient characteristics.

Within the systems approach to organisational failure, a division exists between ‘normal accident’ theories, which state that accidents are inevitable in large, complex enterprises and ‘high reliability organisation’ theories, which assert that accidents can be prevented by organisational development. Normal accident theory was developed by Perrow (1984), who argues that complex systems generate novel and unexpected sequences of events that are hidden or difficult to comprehend immediately. He distinguishes between loosely-coupled and tightly-coupled systems. In the former, there is some opportunity to recover a critical situation before it becomes an accident, but in the latter, disasters are inevitable. A case study that strengthens the normal accidents hypothesis is Vaughan's (1996) analysis of NASA's decision to launch the spaceship Challenger, despite engineers' concerns that a vital component (the O-rings) could fail in low temperatures. The widely accepted explanation of the disaster is that middle managers concealed the technical problem from top administrators. This focuses the blame on individual staff. In her alternative interpretation, Vaughan (1996) attributes the disaster to a collective process which she calls the normalisation of deviance (p. 394). Each time a launch recommendation was required, diverse opinions had to be pulled together into a collective public position within a bureaucratic production culture that required a clear-cut decision to launch or not launch. In such a context, the deviant (i.e. abnormal) performance of the O-rings in test situations was progressively re-interpreted as normal. One contributory factor was that while technically qualified individuals continued to make accurate assessments of the risk, the launch decision was collective and their views were muted by their position in NASA's social structure. For example, a resource analyst noticed that the anomalies in the O-ring data were increasingly being interpreted as normal variations and wrote a memorandum to warn that the technical problem was compromising flight safety. However, because he was not a member of the engineering group and had only recently joined NASA, this made no impact on the bureaucratic production culture that continued to move inexorably towards the fatal launch. Vaughan's (1996) analysis is important for several reasons. One is to emphasise the collective nature of decision-making and error generation in enterprises. Another is to demonstrate that, while an individual might possess information crucial for a decision, his or her position in the bureaucratic structure might prevent it influencing the collective process.

A more optimistic view is taken by high reliability organisation theorists. Their fundamental assumption is that organisational accidents can be prevented by appropriate organisational design, once the parameters of the reliable organisation have been identified. Writers in this field include Weick, 1987, Roberts, 1989, who identify the main attributes of the high reliability organisation as a safety culture, ability to learn from failures, procedures for recovering critical situations before they become disasters and the effective communication of doubt through the organisation.

Section snippets

Method

Two EDs in Britain provided the data for the study. Both are located in university teaching hospitals and have more than 60,000 new attendances annually. Following a study of working practices in these departments by the first author, the second author carried out fieldwork under the guidance of the third author (a consultant emergency physician). The fieldwork began with attendance at the induction days for the senior house officers (SHOs) who make up most of the medical workforce in EDs. The

Active failures

Twenty-five critical incidents were collected and classified into four basic types of active failure (Table 1).

  • Type 1: Delay in beginning the initial nurse assessment, the medical investigations or the treatment (N=14).

  • Type 2: Failure to obtain essential diagnostic information (N=6).

  • Type 3: Misinterpretation of diagnostic information (N=3).

  • Type 4: Administration of inappropriate treatment (N=2).

Latent conditions

Three categories of latent condition were found. First, some incidents were due to large number of

Discussion

All seven incidents reveal a tension between the need to work flexibly in the ED and the rigidity of the division of labour. The vertical and horizontal division of labour is not specific to the ED, but evolved in other departments of the hospital where the work process is different. In those wards, patients are treated by different teams of specialists and diagnoses and treatment decisions are made by consultants and passed down to junior members of the medical staff. The expectations this has

References (40)

  • R.A. Rusnak et al.

    Litigation against the emergency physician: common features in cases of missed myocardial infarction

    Annals of Emergency Medicine

    (1989)
  • R.F. Belli et al.

    Autobiographical misremembering: John Dean is not alone

    Applied Cognitive Psychology

    (1997)
  • Healthline

    (1996)
  • D.E. Broadbent

    Word frequency effect and response bias

    Psychological Review

    (1967)
  • B. Dankbaar

    New production concepts, management strategies and the quality of work

    Work, Employment and Society

    (1988)
  • Withdrawal of Guidance on the Extended Role of the Nurse

    (1992)
  • The extending role of the clinical nurse: legal implications and training requirements

    (1977)
  • N. Dodier et al.

    Openness and specialisation: dealing with patients in a hospital emergency service

    Sociology of Health and Illness

    (1998)
  • S. Dowling et al.

    Nurses taking on junior doctors' work: a confusion of accountability

    British Medical Journal

    (1996)
  • Y. Engestrom

    Developmental studies of work as a testbench of activity theory

  • J.C. Flanagan

    The critical incident technique

    Psychological Bulletin

    (1954)
  • R.P. Fisher et al.

    Field test of the cognitive interview: enhancing the recollection of actual victims and witnesses of crime

    Journal of Applied Psychology

    (1989)
  • R.E. Geiselman

    Improving eyewitness memory through mental reinstatement of context

  • D. Hughes

    The ambulance journey as an information generating process

    Sociology of Health and Illness

    (1980)
  • D. Hughes

    When nurse knows best: some aspects of nurse/doctor interaction in a casualty department

    Sociology of Health and Illness

    (1988)
  • D. Hughes

    Paper and people: the work of the casualty reception clerk

    Sociology of Health and Illness

    (1989)
  • R. Jeffery

    Normal rubbish: deviant patients in casualty departments

    Sociology of Health and Illness

    (1979)
  • E.A. Kadzombe et al.

    Complaints against doctors in an accident and emergency department: a 10 year analysis

    Archives of Emergency Medicine

    (1992)
  • C.C. Lundberg

    Managing in a culture that values learning

  • Cited by (44)

    • Between demarcation and discretion: The medical-administrative boundary as a locus of safety in high-volume organisational routines

      2018, Social Science and Medicine
      Citation Excerpt :

      Fieldnotes, Practice 5, back office, 16.07.13) Receptionists also frequently made moral judgements about patients by grading the risk of different medication requests (Boreham et al., 2000), with receptionists commonly stating: ‘it depends on who it is [i.e. which patient] and what it is they want [i.e. which medication/result]’ (Practice 1, Receptionist 5). Thus, while the practice IT systems and formal protocols guided everyday action, they did not constitute scripts for action (Swinglehurst and Greenhalgh, 2015).

    • Attitudes of emergency care staff towards young people who self-harm: A scoping review

      2014, International Emergency Nursing
      Citation Excerpt :

      The findings from the review indicate that the nature of the setting or service influence attitudes, however the studies reviewed do not consider whether professional culture and ideology have a bearing on the formation of attitudes, which might explain for example why those working in CAMH services would seem to have more positive attitudes. There is evidence from the wider literature that practitioners’ expectations of working in A&E are counterintuitive to the reality of A&E work, (Dodier and Camus, 1998; Boreham et al., 2000). Doctors working in acute hospital care have been found to prioritise curative functions in order to ‘fix’ the patient (Jeffery, 1979; Cassell, 2004; Hadfield et al., 2009).

    • Identifying nontechnical skills associated with safety in the emergency department: A scoping review of the literature

      2012, Annals of Emergency Medicine
      Citation Excerpt :

      During stage 1, a series of databases was searched, including MEDLINE, EMBASE, and PsychINFO (covering behavioral sciences in medicine, law, and other fields). This was supplemented by searching Health Information Resources (a service open to National Health Service staff to enable simultaneous access to a number of databases) and reviewing reference lists of seminal articles5,14,15 and relevant systematic reviews.9-11 Key words and Medical Subject Headings terms were generated around 3 main concepts as dictated by the research question, namely, emergency medicine, nontechnical skills, and error.

    • Accidents in "normal" operation - Can you see them coming?

      2010, Journal of Loss Prevention in the Process Industries
      Citation Excerpt :

      The quality of information flows and resources, mentioned above, form the latent conditions that may lead to a possible ineffective control process, i.e. Dekker's so-called “unfolding situation” (Dekker, 2002). In literature various numbers of latent conditions are presented, e.g. Hudson, Wagenaar, Reason, Groeneweg, Meeren van der, and Visser (1991) identified 11 types of latent conditions, Jacobs and Haber (1994) identified 20 different types of specifically organization related latent conditions, Weil and Apostolakis (1999) identified 6 types of these, Boreham, Shea, and Mackway-Jones (2000) types, while Busby and Hughes (in press) identified 8 types of latent conditions. Most of these types of latent conditions need clarification because they are not specific enough and too encompassing for an organization to benefit.

    • New approaches to researching patient safety

      2009, Social Science and Medicine
    View all citing articles on Scopus
    View full text