Maxine Offredy *,Martin Rhodes MA and Yvonne Doyle
Reader in Primary Health Care, University of Hertfordshire, Hatfield, UK
Medical Director, Harrow Primary Care Trust, Harrow, UK
Acting Interim Head of Department, Harrow Primary Care Trust, Harrow, UK
Received date: 12 October 2009; Accepted date: 29 October 2009
This paper provides a structured chronology of an investigation into a significant untoward incident in an elderly care ward. Using Reason’s Swiss Cheese Model, which has become one of the dominant paradigms for analysing clinical and patient safety incidents, it charts the interplay of national and local policies resulting in unsafe practice.A qualitative approach was used in this multidimensional investigation. This approach aimed to discover what actually happened in the specific and related incidents and the underlying causes. Thus, the anatomy of the incident refers to the structure of staffing, the physiology includes the process in place at the time of the incident and the pathogenesis alludes to the development of the incident.The findings report on the patients involved in the incident. The investigation also explores how strategic financial directions from the Department of Health impact on staffing levels and training. These are contextualised using the concepts of the Swiss Cheese Model to assist understanding of how and why the incident occurred.Key points emanating from a learning event are captured to aid understanding and the importance of being cognisant of the ever present risks in clinical practice. The impact of the investigation on staff and the primary care trust are also presented.
protection of vulnerable adults, risk, significant untoward incident
At five o’clock one morning in November 2007, a member of staff entered a primary care trust (PCT) ward for the elderly and found a patient with advanced dementia tied to a chair with a bed sheet. The folded sheet was tied around both patient and chair with the knot behind the chair. The matter was reported to the PCT managers. As there was uncertainty surrounding the identification of the individual who tied the patient to the chair, all fourmembers of staff who were on duty at the timewere identified and suspended. Following the procedures laid down in the Protection of Vulnerable Adults (POVA) guidelines, the PCT informed the police. The police conducted an investigation over the following three months and referred their findings to the Crown Prosecution Service, who declined to mount a prosecution after reviewing the information presented to them. In February 2008, the matter was referred back to the PCT which began its own investigation into the incident and the surrounding circumstances.
This paper describes the investigationa and the findings; the findings are analysed using Reason’s Swiss Cheese Model of system accidents. The chronology of events, the learning from the investigation and the impact for both staff and the PCT are also highlighted.
The multidimensional investigation comprised interviews, review of ward documentation including patients’ records, oral and written information, site visits and information from the police investigation. This approach aimed to discover what had actually happened both in the specific incident described and in any related incidents, and also the underlying causes. Thus the anatomy of the incident refers to the structure of staffing, the physiology includes the process in place at the time of the incident and the pathogenesis alludes to the development of the incident. Box 1 summarises the timeline.
Structured interviews of staff were mostly taperecorded (with consent from interviewees) and undertaken with 33 employees. Contemporaneous notes were made and the transcripts were made available to the interviewees. Interviews of patients and relatives were conducted by PCT staff, trained in sensitive interviewing techniques, who were not front-line clinicians. The investigator re-interviewed some staff to clarify issues identified in the transcripts.
Documentation was used to identify issues and triangulate findings with other information obtained. Guidelines on good practice concerning the protection of vulnerable adults and the use of restraint were used as references in the investigation.[1,3–6] Restraint is defined by the Mental Capacity Act 20057 as action that uses, or threatens to use, force to secure the doing of an act which the client resists, or restricts the client’s liberty of movement, whether or not the client resists.
The project group consisted of the investigating team (managers, directors, the chairman of the PCT, a representative from the Hospital Trust and an outside expert) who met frequently throughout the investigation and beyond. The team directed the investigation, made decisions about staff suspensions and ensured sensitive release of information to patients, relatives and staff.
Following the compilation of a report and disciplinary proceedings having been taken against staff, a learning event was held by the PCT and included all stakeholders: representatives of the Healthcare Commission, a union representative, non-executive directors of the PCT, police personnel and members of the POVA strategy group. The event was conducted under the Chatham House Rule8 and consisted of a presentation, outlining the incident, investigation and examination of predisposing factors. Small groups discussed what had gone well and what might be improved in relation to the investigation, staff management and relationships with colleagues. The aim of the exercise was to capture lessons from the incident and to disseminate good practice.
The first patient
The first patient
Patient C was an elderly patient with advanced dementia who was found by a member of staff tied to a chair. The four staff members on the night shift denied tying Patient C to the chair or that the patient was tied to a chair. They further maintained that a sheet was placed around Patient C for warmth as there were no blankets available.
The investigation identified a further four patients who had been treated similarly, two of whom were deceased by the time of the investigation. They shared a history of confusion and disruptive behaviour. Multidisciplinary team discussions about one of these patients had concluded that, for a test period of two weeks, the patient should be allowed to sit in a chair with a loosely-tied restraining seat belt; after the test period this stratagem would be reviewed by the consultant at regular intervals. Staff interviewed had observed this restraint but, apart from the ward manager, none had recognised that this was part of a specific and agreed strategy that was under regular review as there was not a care plan concerning restraint in the records.
Relatives of the two deceased patients reported that their relatives had told them that they were tied to their beds. One relative witnessed the restraint of their family member whilst other relatives observed patients being restrained by being tied to furniture. The case notes of one patient recorded that a patient’s relative had asked that the patient be restrained in a chair but a nurse had refused to do so on the grounds that thiswas unethical. The investigations established that the effect of inappropriate restraint on the patients could not be demonstrably ascertained because of the cognitive impairment of the patients. Nonetheless, the possible effects of such behaviour, including ethical issues and the impact this behaviour might have had on the patients, relatives and friends, were of concern to the investigating team.
Some members of staff confirmed that these patients had been restrained with either bed sheets or men’s braces while others deny witnessing, or knowing of, patients who had been restrained. Those nurses who had witnessed restraint of patients did not report the practice. One healthcare assistant thought that restricting movement by tightly applying a sheet and placing a table close to the patient was acceptable practice.
The analysis presents the predisposing factors of the incident and their fit with the Swiss Cheese Model (Figure 1).2 Box 2 highlights the main points of the learning event.
The basic premise of the Swiss CheeseModel implies that hazards or errors are prevented from occurring by a series of barriers and safeguards. In an ideal organisation the safeguards and barriers would be intact. However, Reason postulates that most organisations are akin to slices of a Swiss cheese, with many holes. Unlike a Swiss cheese, these holes continually shift their position, opening and closing in different areas. The existence of holes in a ‘slice’ does not normally result in an accident. Accidents occur when, by chance, the holes in many ‘slices’ align to allow a route for the hazard to reach the patient, causing harm.
Each barrier has unintended weaknesses or holes (Figure 1). Holes in the defence occur because of two reasons: active failures and latent conditions. Active failures refer to unsafe acts committed by individuals at the ‘sharp end’ of the system (pilots, air traffic controllers, maintenance workers, doctors, nurses) whose actions can result in immediate adverse consequences. Latent conditions are strategic misjudgements or poor decisions usually taken at the senior management level of the organisation or within society, away from the front line and at the ‘blunt end’ of the system. Latent conditions have a further three layers: organisational influences, unsafe supervision and preconditions for unsafe acts. The following sections contextualise these concepts within the current investigation.
The latent conditions identified in this investigation include strategic direction from the Department of Health for NHS Trusts to achieve financial balance or to break even in the financial year. This background drive to achieve financial balance affected staffing levels and training. The PCT in this investigation had significant financial problems and was required to produce a robust recovery plan to achieve financial balance over the shortest possible period.
The Trust Board approved the recovery plan at its meeting in June 2006 when it was described as being ‘non-negotiable’. It was subsequently approved by the Strategic Health Authority. The plan resulted in a sudden cut of more than 20% in the nursing budgets for the Care of the Elderly wards.Notes made by a PCT staff member in May 2006 suggested that no risk assessment had been made of this action. The recovery plan had small sections on risk and its mitigation; this had been evaluated by three PCT directors alone. The PCT’s Risk Register does not contain information to suggest that the risk to patient safety was monitored following the implementation of the recovery plan. Enquiries by the investigator to appropriate personnel in the PCT did not uncover any information to suggest that any such evaluation had occurred. Consequently strategic decisions such as the reduction in nursing budgets, the lack of identification of risk and its management and the lack of wider discussions about risk and its evaluation are latent failures and predisposing factors which impacted on the patients and staff 15 months later.
A Health Impact Assessment day which involved 19 people, including practicing clinicians, was held in June 2006, after the recovery plan had been signed off. The issue of not filling vacant posts was judged to be by far the greatest risk of the whole recovery plan. The risk was to be managed by a service redesign and one of the four Care of the Elderly wards that were managed by the PCT was closed.
Between June 2006 and November 2007 all five Executive Directors moved on from the PCT. The investigator judged that the ‘organisational memory’ was affected by this, including the monitoring of the consequences of the recovery plan.
The staffing levels for the 26-bed ward at the time of the incident were meant to be two trained nurses and four healthcare assistants during the day and evening shifts, and two trained nurses and two healthcare assistants on the night shift. The investigator was told that there are no national guidelines regarding staffing levels on Care of the Elderly wards, but local comparisons suggest that this complement of staffing was low. Differing opinions exist as to how often the ward was understaffed. It is undisputed that there was heavy reliance on ‘bank’ staffb to cover periods when there was a staff shortage and when patients required one to one care – as in the case of the first patient.
Reason’s model takes into account predisposing factors which might lead to an incident. In this case, the reliance on bank staff is viewed as being both a latent and active predisposing factor. The latent failure has been discussed above in relation to financial instability. However, some responsibility for active failure rests with the individual because of the need for their physical and emotional stamina to work long hours. Errors in their delivery of care are more likely to occur; it is at such times that shortcuts might well take place. The data showed that some staff drew the attention of senior managers to the high use of bank staff and poor staffing levels. In terms of the model, the individuals to whom the staff had voiced their concerns inadvertently allowed unsafe supervision of both staff and patients, thus giving the impression of condoning poor staffing levels.
The data showed that only twostaff had received training on restraint or protection of vulnerable adults. Of the two, one staff member had received training in a previous employment and the other undertook training in her own time. Interviews with some staff showed that lack of training was a predisposing factor to their substandard practice as some had poor understanding and knowledge of what constituted restraint.
An interview with a member of staff highlighted that reduced staffing levels resulted in suboptimal quality of care for patients. Examples given were lack of attention to dietary, continence and mobility needs. Poor team communication was highlighted by staff although they were encouraged to discuss issues at monthly meetings. Minutes of meetings were held on computer but could not be accessed by the investigator because of suspension of staff.
Staff appraisals had been undertaken, although not all within a 12-month period. Some were judged by the investigator to be superficial; only one interviewee identified any learning or professional development plan that emerged from the appraisal. Lack of support for training was also an issue of concern for some staff.
The Swiss Cheese Model demonstrates that the events outlined in this paper do not occur in isolation but as a result of certain preconditions being breached. Some of the issues to emerge from this investigation include non-technical skills, for example, social or interactive skills and situational awareness skills. These skills relate to the individual per se and are required in team and individual situations. They should be part of an induction programme so that the concepts of patient safety, respect, dignity and management are grounded in practice.
Managerial procedures for recruiting staff were curtailed because of the financial position of the PCT, which in turn was responding to government policy. These latent conditions and predisposing factors lay dormant for many months before interacting with other failures to produce unacceptable nursing care. Analysis of error in any complex organisation requires investigation into both the active failures that happen to the individual on the front line and the latent failures of policy, procedures and culture. Reason’s Swiss Cheese Model allows organisations to find holes in their defences and to develop mechanisms to address the underlying structural weaknesses.
The PCT’s written procedures viewed suspension as a neutral act. However, it was not perceived as such by the staff. Suspension together with the involvement of the police in the investigation was not conducive to a climate of openness. This made investigation into the circumstances of the first patient and the surrounding issues difficult. In particular, although there is clear evidence of patients being restrained, no member of staff admitted to doing this or knew anyone who had done this. In the event, the investigation resulted in disciplinary procedures being applied to four members of staff and a further member was reported to their professional body.
The team interviewing the patients and their relatives found that the majority said they were satisfied with the care received on the ward. Some patients and relatives were highly complimentary about the services provided. However, there were some who said they were not satisfied with the care delivered but would not complain because their relative could be readmitted onto the ward. This mixture of reviews would suggest that the quality of care on thewardwas variable. Further, little consideration appears to have been given to the impact of such actions as restraint on people with dementia. There appears to be an underlying issue among staff and management that behavioural disturbances in people with dementia are indicators of unmet needs; this would indicate a failure to address provision of training in best practice in dementia care. If staff have not been well trained in communicating with people with dementia, they are unable to understand what needs the person with dementia is communicating through behavioural symptoms and are unable to respond appropriately to these needs. This makes poor practice not only likely, but inevitable. Poor practice will perpetuate itself.
A pertinent issue for nurses (and other healthcare professionals) to bear in mind is that their practice decisions have an ethical component. The component is made up of a number of principles and includes: avoiding harm, assessing the consequences of action, autonomy and rights and values and beliefs. Tension arises in the decision about applying restraint because to do otherwise could also cause harm. Thus the consequences of using or withholding restraint need to be carefully considered by determining the potential benefit and harm. Two of the principles laid down by the Mental Capacity Act 2005 state that:
An act done or decision made, under this Act for or on behalf of a person who lacks capacity must be done, or made, in their best interests. Before the act is done, or the decision is made, regard must be had to whether the purpose for which it is needed can be as effectively achieved in a way that is less restrictive of the person’s rights and freedom of action.
The investigation team found that the actions of some of the nurses breached these values when treating elderly patients with dementia. Implicit in this behaviour is the lack of person-centred care, a lack of understanding about the impact of dementia on patients and a lack of skills in effective communication by staff when working with people with dementia.
Incidents such as those described in this paper do not occur in isolation of the context in which practitioners work, including the culture and climate of the organisation. Recent thinking suggests they occur due to a combination of issues in individuals and systems. Thus the allegations were investigated in their wider context. There is little controversy among clinicians about the importance of good clinical practice, but the consensus of good clinical practice dissipates when errors occur, making investigation of incidents difficult. It is imperative to know, understand and learn from the ever-present risks in clinical practice.
There is very clear evidence that the first patient was tied to a chair with a sheet in November 2007 and possibly on previous nights and this was performed in a way contrary to national guidelines on restraint. There is no evidence that points to which particular individual or individuals were responsible for participating in restraining patients. There is evidence that some staff knew about this unacceptable practice but failed to manage or report this behaviour. Knowledge of policy and practice on restraint was poor; some senior staff did not recognise that the use of a seat belt was restraint. This was partly due to the lack of training on this subject, but was also about lack of training on the condition of dementia and its impact and the communication skills of staff. There are lessons that were learnt from the latent preconditions relating to the incident, resulting in sensitivity to the needs of patients and their relatives throughout the investigation (apologies were given and feedback was given to them before the official report), special training for interviewers and the timely response to the incident. The events described in this paper are both of national and international significance. Extensive recommendations (www.harrowpct.nhs.uk) were provided for individual nurses, ward staff, medical consultants and the Trust.
Department of Health. www.publications.parliament.uk/ pa/cm200607/cmhansrd/cm070522/debtext/70522–0001. htm 2007.
Department of Health. No Secrets: guidance on developing and implementing multi-agency policies and procedures to protect vulnerable adults from abuse. London: HMSO, 2000
Commissioned; not externally Peer Reviewed.
a It should be noted that major national television and radio stations reported the incident when it occurred;further reports were published in local newspapers.
b Bank staff were nurses engaged by the PCT to work shifts on an ‘as needed’ basis. Bank staff may have held substantive contracts of employment with the PCT, or may have been engaged only under the terms of a bank staff letter of registration.