Investigation report: Wrong site surgery - wrong tooth extraction

A number of national reports were in progress when the COVID-19 pandemic significantly affected the UK in 2020. Much of the work associated with developing the investigation reports necessarily ceased as HSIB’s response was redirected.

For this national report, while the learning described has not changed due to COVID-19, the scope of the investigation and processes HSIB used to engage with staff had to be adapted. These changes are acknowledged and described further in section 3 of this report.

We would like to thank the Patient whose experience is documented in this report, and her family. We would also like to thank the healthcare staff who engaged with the investigation for their openness and willingness to support improvements in this area of care.

This report is intended for healthcare organisations, policymakers and the public. Its aim is to help improve patient safety in relation to understanding the importance of systems intended to ensure that measures put in place to minimise the risk of patient safety incidents are effective. The report explains how other safety-critical industries have adopted, and often regulate for, such systems. It uses a case of wrong tooth extraction as a case in point.

This investigation examined the most common form of wrong site surgery as reported by NHS England and NHS Improvement (NHSE/I) over the past five years – wrong tooth extraction. Wrong site surgeries – procedures carried out on the wrong patient or the wrong part of a patient’s body – are classed as Never Events. Never Events are defined by NHSE/I as ‘patient safety incidents that are wholly preventable where guidance or safety recommendations that provide strong systemic protective barriers are available at a national level and have been implemented by healthcare providers’.

The investigation uses a real patient case, referred to as ‘the reference event’, to examine the issues relating to wrong tooth extraction.

Background

Following the introduction of National Safety Standards for Invasive Procedures (NatSSIPs) in 2015, all NHS trusts in England were required to identify relevant clinical procedures in their organisation, including those undertaken outside hospital settings, and produce Local Safety Standards for Invasive Procedures (LocSSIPs). In terms of the numbers of interventions carried out, dental surgery is one of the largest areas of NHS surgical activity in the NHS; the majority of these interventions are carried out away from the operating theatre environment, on outpatients under local anaesthesia. It was intended that the introduction of LocSSIPs should take teamwork and training, human factors and cultural aspects into consideration when implemented. In 2018, the Royal College of Surgeons Faculty of Dental Surgery published a LocSSIPs toolkit for wrong site extraction in dentistry to assist trusts in the development of local procedures.

Identifying teeth is challenging due to the need to capture each tooth’s position with reference to upper and lower jaw and left and right side of the mouth. A young person has up to 20 deciduous teeth (commonly known as milk teeth) which are replaced by up to 32 permanent ‘adult’ teeth – at some stage there will be a mixture of both (referred to as mixed dentition). Each tooth is named (for example upper right first premolar) and also identified using a system of notation. In the UK, there are several systems of dental notation in use including Palmer notation, an adaptation of Palmer notation that uses letters and numbers, and Federation Dentaire International (FDI).

The reference event

The investigation examined the case of a girl aged seven who was referred by her general dental practitioner to a local Trust that provided more complex dental services in both community health centres and hospitals. After a six-month wait, the girl was seen by a senior dental officer (SDO) who undertook an examination and referred to an orthodontic consultant (a specialist who deals with irregularities or abnormalities in the teeth or jaw) for an opinion. Following that referral, the SDO agreed a treatment plan with the girl’s parents. The SDO extracted an infected deciduous tooth and prescribed a course of treatment with a dental therapist colleague. The treatment was planned over two appointments; the plan was to extract a deciduous tooth under local anaesthetic during the second appointment. Following the extraction, it was noted that a neighbouring permanent tooth had been extracted instead of the intended deciduous tooth. The permanent tooth was successfully reimplanted within approximately seven minutes. The incident was correctly reported and investigated as a Never Event.

The national investigation

Following analysis of the reference event, the investigation focused on wrong tooth extraction and the requirement for there to be ‘strong systemic protective barriers’ in place at a national level in order for a serious incident to be deemed a Never Event. In this context, the term ‘barriers’ refers to measures put in place to prevent the occurrence of patient safety incidents. In other safety-critical industries, the term ‘barrier’ is well defined and has a specific meaning. Barrier analysis and management methods – ways to ensure barriers are, and remain, effective – are routinely used, and sometimes mandated, in other safety-critical industries. The investigation sought to identify what barriers existed in the care pathway for tooth extraction.

Findings

• Wrong tooth extraction is a serious occurrence and is a locally reportable incident with the associated investigation requirements that are within the Serious Incident Framework. Incidents should continue to be notified and investigated to understand where the controls that have been put in place to prevent it happening have failed so they may be strengthened.
• The investigation saw many good examples of LocSSIPs, designed at trust level, with the intention of introducing controls and a standardised process to tooth extraction specifically in the outpatient setting.
• An understanding of how well barriers will perform when put to the test is important when understanding and proactively managing risk. While most safetycritical industries have invested heavily in systems, mandated by regulatory bodies, for identifying, analysing and assuring barriers, the NHS has not.
• In safety-critical industries the term ‘barrier’ is reserved for a safety control with specific properties.
• The description of what constitutes a ‘barrier’ is not clearly defined in the NHS Never Events policy and framework and is inconsistent with other literature.
• The investigation considers the NHS Never Events policy and framework document offers minimal guidance on the categorisation of barriers to help organisations understand their function. This makes implementation of national policy open to variation.
• The exclusion of deciduous teeth from the previous list of wrong site surgery Never Events suggested a focus on the potential level of harm to patients (the outcome) posed by this type of incident, rather than the development of reliable systems that aim to mitigate all wrong tooth extraction.
• All of the controls identified by the HSIB investigation to prevent wrong tooth extraction relied on human performance or were intended to protect against loss of human reliability.
• The investigation found no controls at a national level that represented strong, systemic barriers to tooth extraction.

HSIB makes the following safety recommendation

HSIB notes the following safety action

Introduction

1.1 This investigation followed the care of a child who attended a community dental service and, while undergoing a course of planned treatment, had a permanent tooth extracted in error. NHS England and NHS Improvement (NHSE/I) classes wrong tooth extraction as wrong site surgery (surgery performed on the wrong patient or the wrong part of a patient’s body) which, in turn, is defined as a Never Event (see 1.3).

1.2 HSIB has conducted a number of investigations into Never Events and recently published a national learning report summarising these investigations (Healthcare Safety Investigation Branch, 2021). The report recommended that NHSE/I remove Never Events that do not have strong and systemic (system-wide) safety barriers. In this context, the term ‘barriers’ refers to measures put in place to prevent the occurrence of patient safety incidents. The HSIB report also recommended that NHSE/I commission programmes of work to find strong and systemic safety barriers for specific incidents where barriers are felt to be possible but are not currently available. By analysing a Never Event using tools routinely employed in other safety-critical industries but unfamiliar to the NHS, this report introduces principles of how to improve safety by gaining an understanding of how reliable barriers are at preventing unintended outcomes.

Never Events

1.3 The NHS Never Events policy and framework (NHS Improvement, 2018a) defines a Never Event as a ‘Serious Incident that is wholly preventable because guidance or safety recommendations that provide strong systemic protective barriers are available at a national level and should have been implemented by all healthcare providers’. Incidents that are classified as Never Events are described in the Never Events List. This list is subject to periodic review and was last published in 2018 and revised in February 2021 (NHS Improvement, 2018b).

1.4 Never Events must be investigated under the NHS’s Serious Incident framework and reported to both the Strategic Executive Information System (StEIS) and the National Reporting and Learning System (NRLS) (two national databases of information about serious patient safety incidents).

Wrong site surgery

1.5 Wrong site surgery was one of the core Never Events when the programme was introduced in 2009, and remains one of 16 main classifications of Never Event. Wrong site surgery is described as:

The definition explains that interventions that are considered to be ‘surgical but may be done outside a surgical environment’ should be included when considering wrong site surgery Never Events. This applies to tooth extraction, which often takes place in an outpatient settings, in hospitals, community settings and general dental practices.

As shown in table 1, the most common category of wrong site surgery between 2015 and 2020 was wrong tooth extraction. In 2018 to 2019 there were 82,149 outpatient hospital appointments in England for tooth extraction (NHS Digital, 2019). There were almost 25,000 operations to extract teeth in children aged five to nine years old. Most of these operations would have been to extract multiple teeth. This was in addition to the approximately 179,000 teeth extracted from children up to the age of nine in general dental practices. Interventions previously excluded from the definition of wrong site surgery included the removal of wrong deciduous teeth (unless done under a general anaesthetic) and local anaesthetic blocks for dental procedures.

Introduction of National Safety Standards for Invasive Procedures (NatSSIPs)

1.6 Along with the definition of surgical Never Events, the use of the World Health Organization (WHO) surgical safety checklist was made mandatory for all surgical procedures, with the aim of reducing the number of surgical Never Events. When the anticipated reduction did not occur, a Surgical Never Events Taskforce was formed; it reported in 2014 (NHS England, 2014). In response, NHS England published National Safety Standards for Invasive Procedures (NatSSIPs) in 2015 (NHS England, 2015a). These procedures were intended to standardise key elements of surgical care and ensure consistency of application, and included procedures undertaken outside hospital environments, such as surgical procedures undertaken by dentists. NatSSIPs were designed to provide the basis for the production of Local Safety Standards for Invasive Procedures (LocSSIPs) and their implementation was required by a patient safety alert published by NHS England in 2015 (NHS England, 2015c). To assist with developing local guidance, the Royal College of Surgeons Faculty of Dental Surgery published a LocSSIPs toolkit for wrong site extraction in dentistry in early 2018 (Royal College of Surgeons Faculty of Dental Surgery, 2018).

Tooth identification

1.7 Identifying teeth is complicated by defining their position on the upper (maxillary) or lower (mandibular) jaw and on the left or right side of the mouth. Adults have up to 32 permanent teeth which, in childhood, are preceded by up to 20 deciduous teeth (commonly known as milk teeth). As a permanent tooth grows, the root structure of the deciduous tooth is eroded and teeth are shed naturally. As a child matures, they will have a period of mixed dentition – a combination of deciduous and permanent teeth. The age at which this happens will vary.

1.8 Teeth are identified by various methods including shape, colour, size, root structure and their position in the mouth in relation to other teeth. They are also referred to by type, such as molar, pre-molar, canine and incisor, and are mapped in the mouth using a system of notation.

1.9 There are several systems of dental notation in general use in the UK (Pemberton and Ashley, 2017). Common examples are the Palmer notation and Federation Dentaire International (FDI) system. Both divide the mouth into quadrants; upper right, upper left, lower right and lower left (see figure 1).

1.10 In the adult mouth, each quadrant contains a maximum of eight teeth. A child’s mouth has up to five teeth per quadrant. Teeth are numbered from the front centre to the back of the jaw. The unambiguous reference to tooth location has increasingly had to contend with the ability to be easily transcribed to computerised dental record systems and referrals. Whereas FDI uses the numbers 1 to 4 to describe the quadrants, the alphanumeric adaptation of the Palmer system describes them using upper left (UL), upper right (UR), lower left (LL) and lower right (LR) together with a tooth number in accordance with Palmer notation. For example, the second premolar on the bottom left side of the mouth would be referred to as lower left 5 and would be typed as LL5.

1.11 Palmer annotates deciduous teeth using the letters A to E and the same deciduous tooth position would be typed LLE. As well as noting teeth, the condition of the gum (gingival tissue) is recorded separately. For this, the mouth is divided into six segments, three upper and three lower. This difference is intended to group teeth together by area where gum conditions tend to be similar.

Provision of planned dental care in England

1.12 Under commissioning arrangements, the NHS and private primary care providers (that is, general dental practitioners) deliver the majority of dental care in England. Dentistry is undertaken in a variety of settings including primary care, community settings and secondary care (hospitals). The complexity of care required should generally dictate in which setting the service is delivered (NHS England, 2015b).

1.13 Dental professionals are registered by the General Dental Council (GDC). The GDC sets out the standards of conduct as well as the scope of practice which applies to all members of the dental team (General Dental Council, 2013). Care is provided by a range of dental healthcare professionals with varying levels of qualification, including:

• Dentists – who are qualified to treat diseases and other conditions that affect the teeth and gums, especially the repair and extraction of teeth and the insertion of artificial ones. Dentists may register specialist skills on 13 specialist lists regulated by the GDC which include orthodontics (the specialism dealing with irregularities or abnormalities in the teeth or jaw) and paediatric (children’s) dentistry.
• Dental therapists (DTs) – who have a scope of practice that is greater than that of a dental nurse (DN) (see below), but more limited than that of a dentist. For example, a DT may undertake dental examinations and charting, take radiographs (X-ray images), give dental block analgesia, carry out fillings on deciduous and permanent teeth, and extract deciduous teeth under local anaesthetic.
• Dental hygienists – who carry out treatment which helps patients maintain their oral health by preventing and treating periodontal disease and promoting good oral health practice. Many dental hygienists also have an additional qualification as a dental therapist; this was true of the dental therapist in the reference event.
• Dental nurses (DNs) – who provide clinical and other support to the team, including dentists and dental therapists.

1.14 Although the GDC permits DTs to ‘carry out their full scope of practice without prescription and without the patient having to see a dentist first’ – referred to as direct access – DTs are sometimes limited by their employers to only operate under prescription from a dentist. This was the case in the Trust where the reference event took place, although some DTs at the Trust had other employment which allowed greater scope of practice.

This investigation used the following patient safety incident, referred to as ‘the reference event’, to gain a greater understanding of the reliability of controls intended to prevent wrong tooth extraction, and whether wrong tooth extraction meets the criteria for a Never Event.

2.1 A general dental practitioner (GDP) referred the Patient, a girl aged seven, to a trust (referred to in this report as ‘the Trust’) that provided dental clinics in eight local community health centres and had an agreement with two acute trusts for the use of operating theatres for dental procedures under general anaesthetic. In 2019 to 2020, the Trust dental service extracted more than 11,000 teeth and had no previous reports of wrong tooth extraction.

2.2 The GDP had previously attempted to treat the Patient, but she did not tolerate the use of suction. They noted on the referral that the Patient was not in pain but required additional support to ensure she was in a relaxed and prepared state to receive treatment. The referral included restorations, or fillings, to the upper 6 teeth, sealant to the lower 6 teeth and extraction under local anaesthetic of a grossly decayed deciduous tooth, LLE.

2.3 Six months later, the Patient attended a new patient appointment at a local community health centre, accompanied by her Father. She was seen by a senior dental officer (SDO) with an interpreter present; on subsequent appointments, an interpreter was not considered necessary. Having noted the referral from the GDP, the SDO examined the Patient’s mouth. Assisted by a dental nurse (DN), the teeth were then charted onto a specialist computer system. A radiograph was taken of the Patient’s teeth (see figure 3).

2.4 The SDO recorded in the clinical notes:

• the reasons for the referral and that the decayed tooth originally intended for extraction (LLE) was now missing and had been replaced by a partially erupted LL5
• LRE had gross decay with clinical signs of infection • all the permanent upper and lower 6 teeth were present and showing signs of decay
• LLD was decayed and loose. The SDO made a treatment plan to extract LRE and temporarily fill the upper first molars on the right and left (UR6 and UL6). Because of the condition of all the permanent molars, the SDO made a referral for a specialist opinion from an orthodontic consultant in another trust. Four weeks later, while awaiting an appointment with the specialist, the Patient returned to the clinic with her Father for the extraction of LRE. This was completed under local anaesthetic and the Patient’s co-operation was noted as ‘excellent’.

2.5 After 10 weeks, the Patient saw an orthodontic specialist in clinic. The appointment was followed up with a letter to the SDO, recommending extraction of UR6 and UL6 but to leave the lower first molars in place. The Patient was then invited for an appointment with the SDO where the treatment plan was discussed with her Mother and Father. The various options for extraction of the permanent teeth were discussed along with a plan to undertake the following treatment:

1. extraction of UR6 and UL6
2. filling LL6 under local anaesthetic
3. extraction of LLD under local anaesthetic
4. application of a protective coating to LL6 and LL5.

2.6 The SDO wrote a prescription for procedures 2 to 4 to be undertaken by a dental therapist (DT). The Trust had five DTs who had a broad scope of practice and could, on prescription, undertake fillings and extractions of deciduous teeth under local anaesthetic. Owing to the limitation of extracting only deciduous teeth, their practice was mainly in children’s dentistry and they had significant experience of treating anxious patients and those with additional needs.

2.7 Two weeks later the Patient again attended the community health centre with her Father for a 40-minute appointment with the DT. The DT had printed out the SDO’s prescription and decided to undertake treatments 2 and 4 during the first appointment. The clinical records noted that the DT gained verbal consent and undertook the filling of LL6 and applied fissure sealant to LL5 and LL6 without incident. The treatment was recorded on the Patient’s electronic record and marked as complete on the DT’s printed prescription. The Patient’s co-operation was again recorded as ‘excellent’.

2.8 The final appointment with the DT for the extraction of LLD was scheduled for three weeks later. On that day, the DT arrived 15 minutes early for work to prepare for the day ahead. She printed a list of patient appointments (the ‘day list’) and made notes from the electronic record system. The DN arrived and prepared the surgery for the first and subsequent appointments. This included checks on equipment such as the suction machine, ensuring availability of local anaesthetic and preparing instrument trays with sterile equipment laid out in bags to be opened prior to use. Having made their preparations, the DT and DN attended a 5- to 10-minute ‘safety huddle’ in a communal area, at which general departmental information was passed on to all staff.

2.9 The Patient, accompanied by her Father, arrived in good time for the planned extraction. Verbal consent for the extraction was confirmed and a patient medical questionnaire was completed by the DT at 08:53 hours. With the local anaesthetic prepared by the DN, the DT identified the area for the anaesthetic injection and administered articane, a local anaesthetic, to numb the gum area. After a short wait, the DT confirmed that the anaesthesia had been effective by prodding the area with a sharp point. Having broken the ligament attaching the tooth to the bone, the DT used a set of posterior deciduous forceps (an instrument designed for extracting teeth from the back of the mouth) to grip the tooth for extraction. The DT stated that the forceps appeared “quite loose” on the selected tooth and she therefore exchanged them with the DN for another set.

2.10 The DT stated that the tooth came out easily and she passed it to the DN to be cleaned and put in a ‘tooth fairy’ envelope. The DT placed a pack on the wound and then completed her notes on the electronic record system. When the DN checked on the packing, she suspected that the wrong tooth may have been extracted. On re-examination the DT realised the error and, after a brief explanation to the Father, she sought immediate help from an SDO colleague.

2.11 The SDO confirmed that the extracted tooth was LL5 not LLD and sought advice from a specialist in paediatric dentistry about re-implanting the tooth into the gum socket from which it had been extracted. The tooth was placed into saline solution and, after approximately seven minutes out of the gum, it was re-implanted by the SDO. Following further advice, LLD was also extracted owing to the infection in the gum and the potential for it to compromise the successful re-implantation of LL5.

2.12 During regular appointments since the incident the re-implanted tooth was noted to have ‘no discolouration, normal mobility … continued root growth indicating tooth vitality’. The Patient complained of some pain, but it was uncertain whether this was associated with the extraction or the decay in UL6 and UR6. After discussion with the Patient’s parents, it was agreed to continue with the extraction plan for the permanent teeth using a combination of local anaesthetic and inhalation sedation (a mixture of gases breathed through a nosepiece to help the child feel relaxed – also known as ‘happy air’).

This section outlines how HSIB identified the patient safety issue. It also describes the criteria HSIB used to decide whether to go ahead with the investigation, and the methods and evidence used in the investigation process.

Identification of the reference event

3.1 On 7 October 2019 HSIB conducted a search of the Strategic Executive Information System (an NHS database of serious incidents in healthcare). The search was for incidents that occurred between 1 January and 7 October 2019, using the search term ‘never event’. This identified 410 results in England and Wales. These results were then filtered by ‘speciality (dental)’ and ‘never event (wrong site surgery)’ to identify a reference event for the national investigation.

Decision to investigate

3.2 Following a preliminary investigation, HSIB’s Chief Investigator authorised a national investigation based on the following patient safety risk criteria:

Outcome impact – what was, or is, the impact of the safety issue on people and services across the healthcare system?

3.3 While there was limited literature to aid understanding of the potential for harm as a result of wrong tooth extraction, litigation data provided evidence of the harm associated with wrong tooth extraction. Tooth loss can result in pain, shifting of other teeth, cosmetic impairment which may in turn create psychological harm, and interference with day-to-day life, such as eating and speaking. Incorrectly extracted teeth can be re-implanted, but this is not always successful and places an additional burden on already stretched specialist dental service resources.

Systemic risk – how widespread and how common a safety issue is this across the healthcare system?

3.4 There are three Never Events that are particularly relevant to dentistry: wrong site surgery, wrong implant/prothesis and retained foreign object post-procedure (foreign objects being left behind in the body after a procedure). In 2018 to 2019 there were around 2.2 million outpatient hospital appointments for dentistry (NHS Digital, 2019). More than 180,000 people were admitted to hospital for dental procedures, either involving an overnight stay or as day cases. Tooth extraction is a widely performed invasive procedure and wrong tooth extraction is the most common wrong site surgery Never Event. Teeth are extracted in error across both primary and secondary care even when apparently robust local procedures are in place in an attempt to prevent such errors. Learning potential – what is the potential for an HSIB investigation to lead to positive changes and improvements to patient safety across the healthcare system?

3.5 Identifying and extracting the correct tooth involves a number of specific challenges as a result of anatomy and the difficulties of undertaking uncomfortable procedures on anxious, awake patients. Dentistry is an area that has not previously been explored by HSIB. The investigation represented an opportunity to understand the structures and challenges faced in this sector and how effective organisations are at managing controls to prevent a wrong tooth extraction Never Event.

Scope of the investigation

3.6 Following an analysis of the reference event, HSIB planned to broaden the scope of the investigation. The intention was to review the wider contributory factors identified in relation to the settings in which tooth extraction, and other minor invasive procedures, were conducted. Crucially, by observing care being delivered in the relevant clinical settings, the investigation intended to explore how local safety standards were implemented in practice. Shortly after the plan was made, the COVID-19 pandemic was declared, curtailing activity on the investigation.

3.7 In view of the COVID-19 pandemic and the impact on traditional ways of working, a decision was made not to undertake observations in clinical areas. The reasons for this decision included a desire not to impose an additional, avoidable burden on NHS dental services at a time of extreme pressure, and a recognition of the fact that many trusts ceased providing all but centralised, emergency dental care during the early stages of the COVID-19 pandemic.

Investigation methodology

3.8 Having gained a valuable insight into an incidence of wrong tooth extraction, and given the importance placed in the NHS Never Events policy and framework on strong systemic barriers when defining a Never Event, the investigation focused on wrong tooth extraction as the most frequently reported wrong site surgery over the past five years (NHS Improvement, n.d.). HSIB saw this investigation as an opportunity to explain barrier management concepts (see section 4), in particular, how an understanding of the strengths and weaknesses of controls may be applied to reducing incidences of wrong tooth extraction. By studying the reference event, the investigation sought to understand what it revealed about the reliability of controls intended to prevent wrong tooth extraction, and whether this type of incident met the criteria for a Never Event.

3.9 The investigation used the following methods to gather evidence and verify findings:

• review of the Patient’s dental records from August 2018 to August 2019
• review of Trust policies, procedures and practice regarding tooth extraction
• interview with the Patient’s parents
• interviews with eight clinical staff at the community trust where the patient was treated from February to July 2019
• interviews with the senior dental officer, dental nurse and review of the Trust Serious Incident statement from the dental therapist involved in the incident (who was unavailable for interview)
• review of the Trust’s internal Serious Incident investigation report
• search of the Strategic Executive Information System and the National Reporting and Learning System for reported incidents relating to dental extractions
• overview of literature relevant to dental extractions and Never Events
• conversations with relevant national organisations and professional bodies including the National Advisory Board for Human Factors in Dentistry
• discussions with a subject matter advisor.

4.1 The NHS Never Events policy and framework (NHS Improvement, 2018a) implies there are ‘strong systemic protective barriers’ in place to prevent wrong tooth extraction. The guidance also states that systemic protective barriers need to be ‘successful, reliable and comprehensive safeguards or remedies’. The HSIB investigation sought to understand the definition of a strong barrier and how barriers can be managed, to gauge whether they meet the criteria set out in the Never Events policy and framework.

4.2 The concept of protective barriers is well understood in the field of hazard and risk management. The literature on this topic uses a variety of descriptions to convey the idea of the ‘lines of defences’ or ‘layers of protection’ inherent in a barrier management system; Reason’s ‘Swiss cheese’ model of accident causation is a wellknown example (Reason, 1990).

4.3 Barrier management is the process of ensuring that the safety controls an organisation relies on are robust enough to be successful in protecting against serious adverse events and their consequences. Important in many aspects of risk mitigation in healthcare is the reliance on humans to perform, support or protect the control functions.

4.4 The hierarchy of controls describes the properties of a safety control. It is based on the concept that not all controls are equal, with those at the top of the hierarchy considered to be more effective and sustainable solutions to safety issues than those at the bottom (National Institute for Occupational Safety and Health, 2020) (see figure 4). The most effective control is to eliminate the threat. An HSIB national learning report which provided an analysis of Never Events (Healthcare Safety Investigation Branch, 2021) reported that common controls in Never Events, such as checklists, were administrative and relied on human behaviour to be effective. Administrative controls are not strong and systemic, and their effectiveness is low on the hierarchy of controls.

4.5 The reference event demonstrated that a task such as tooth extraction is an activity that happens within a complex system. In such a system, the social, cultural and technical context within which the task is performed influences and limits the way the system operates and how effective controls will be.

4.6 While most safety-critical industries place emphasis on the importance of barriers to guard against unintended consequences, some, such as oil and gas, aviation, rail and nuclear, have invested heavily in systems for identifying, analysing and assuring barriers. In some industries regulators have mandated safety management systems with inherent barrier management. For example, the UK civil aviation regulator has embedded bowtie analysis (BTA) in its Performance Based Oversight and Regulatory Safety Management System. It is a model used by operators to help understand their safety risks (Civil Aviation Authority, 2020).

Bowtie analysis

4.7 Developed in the early 1990s and used in the North Sea oil and gas industry, BTA is an example of a barrier management system in widespread use in high-hazard industries (Chartered Institute of Ergonomic and Human Factors, 2016; De Ruijter, 2016).

4.8 BTA is undertaken for a specific hazard (in the case of this investigation, dental surgery for tooth extraction in an outpatient setting). A bowtie diagram has at its centre a knot, or top event (in this case, forceps placed on the wrong tooth), which may result in an undesired outcome (in this case, extracting the wrong tooth). There can be multiple BTA diagrams depicting different top events (another example might be ‘retained object following dental surgery’).

4.9 Barriers depicted on the lefthand side of the ‘knot’ reduce the likelihood of a known threat leading to the top event – sometimes referred to as prevention controls (see figure 5). Those on the right-hand side reduce the likelihood of a consequence owing to the top event or mitigate the severity of the consequence – otherwise called recovery controls.

4.10 BTA takes into account the cultural and technical (socio-technical) context within which a hazard sits at various levels in a system, such as a hospital, a hospital ward or a side-room. By introducing the concept of degraded but not eliminated defences (see 4.13), it allows a pro-active understanding of how controls decay and can fail. The UK Health and Safety Executive recognises BTA as a model for a barrier-based approach (Health and Safety Executive, 2008) and in a review of literature published prior to 2017, McLeod and Bowie (2018) found it has been used to a limited extent in healthcare.

4.11 An advantage of BTA is that it provides a means to ‘communicate not calculate’. By representing risk diagrammatically, the intention is that organisations can communicate and understand risk more effectively and prioritise activity correctly (De Ruijter, 2016; Health and Safety Executive, 2008). The civil aviation regulator provides a library of generic bowtie template diagrams on its website for organisations to adapt; one, for example, addresses human performance (Civil Aviation Authority, 2014).

Defining ‘barrier’

4.12 The term ‘barrier’ is often used in everyday language to denote any type of mitigation to a threat or an obstacle to a positive outcome. Within safety-critical industries, the term ‘barrier’ has a distinct meaning, although the description is inconsistent in literature (Sklet, 2006) and poorly defined in NHS policy.

4.13 The International Association of Oil and Gas Producers (2016) explains a barrier as a ‘risk control that seeks to prevent unintended events from occurring or prevent escalation of events into incidents with harmful consequences’. BTA also recognises factors that can degrade (weaken) barriers, making them less effective. These degradation factors can, in turn, have controls in place to mitigate their impact (see figure 6).

4.14 There is currently no agreed standard for BTA (Civil Aviation Authority, 2020). Recognising the ‘lack of standardisation or establish best practice’, the Chartered Institute of Ergonomics and Human Factors (CIEHF) produced a white paper which included the following definitions:

• '‘controls’ comprise all of the measures expected to be in place to prevent incidents. Controls comprise barriers and safeguards.
• ‘barriers’ are controls that are assessed as being sufficiently robust and reliable that they are relied on as primary control measures against incidents.
• ‘safeguards’ are controls that support and underpin the availability and performance of barriers but that cannot meet the standards of robustness or reliability to be relied on as primary measure (i.e. as a barrier).’ (Chartered Institute of Ergonomics and Human Factors, 2016)

The introduction of safeguards was intended to recognise that, in some industries (such as healthcare), there are often controls that are not sufficiently robust to meet the quality criteria required of a barrier, but do provide a level of risk reduction and may be the only control present. The white paper suggested their role ‘should be recognised in any comprehensive approach to barrier analysis as ineffective safeguards can create the conditions for barrier failure or degradation’ (Chartered Institute of Ergonomics and Human Factors, 2016).

Barrier categorisation

4.15 The Never Events policy and framework offers examples of what it refers to as strong systemic barriers:

• ‘physical barriers (eg equipment that makes it impossible to connect medications via the wrong route)
• time and place barriers (eg withdrawal of concentrated medications from settings to prevent them being accidentally selected) or
• systems of double or triple checking where these are supported by visual or computerised warnings, standardised procedures or memory/communication aids.’ (NHS Improvement, 2018a)

4.16 The investigation found various definitions for how barriers are categorised in the literature. Classification was in keeping with the hierarchy of controls, with a consensus that barriers fell into two categories; active or passive (Chartered Institute of Ergonomics and Human Factors, 2016; De Ruijter, 2016; Sklet, 2006).

• Active barriers are characterised by a requirement to be able to; detect the existence of a threat; to decide what action needs to be taken; and, to act to block the threat (McLeod and Bowie, 2018). Active barriers can be fully technical (a defibrillator that notices a failing heart rhythm and delivers a shock automatically), fully human (a person observes cardiac arrest and commences resuscitation) or a combination of both (alerted by a heart rate monitor alarm a clinician commences resuscitation).
• Passive barriers are usually physical features or structures that are capable of blocking the progress of a threat simply by their existence (a locked medicine cabinet, for example).

Human barrier elements

4.17 There are certain controls that rely on a person to achieve their function (a handheld fire extinguisher or setting up a syringe driver) and/or maintain its availability or readiness to function (putting a phone on charge or ensuring the crash trolley is stocked and in place). CIEHF, in its white paper, states:

When designing safety features, Porto (2009) concluded that ‘getting it right the first time and noticing when it is wrong require the vigilance of the individual, a skill at which humans are generally unreliable’.

4.18 Human barriers fall into two categories (Chartered Institute of Ergonomics and Human Factors, 2016); organisational and operational.

• Organisational controls are where ‘there is intended to be little room for autonomy or discretion in what is done’. This can be achieved by written rules or procedures that an individual is expected to follow when making decisions or taking action.
• Operational controls are less prescriptive and ‘rely on individuals’ skill and experience, capabilities in problem solving, decision making, and imagination, as well as team working skills including coordination and communication’.

4.19 The CIEHF makes 33 recommendations to improve those elements of barrier management that rely on human performance or are intended to protect against loss of human reliability.

Effectiveness of controls

4.20 In safety-critical industries the term ‘barrier’ is reserved for a safety control with specific properties. The UK civil aviation regulator explains these properties on its website (Civil Aviation Authority, n.d.) as does the CIEHF white paper (Chartered Institute of Ergonomics and Human Factors, 2016):

• Ownership – clear responsibility for maintaining and assuring effectiveness.
• Traceability – the ability to track a barrier to a specific requirement or process.
• Specificity – a statement of staff performance required to prevent the hazard identified.
• Independence – each barrier is independent of another.
• Effectiveness – every barrier alone should be able to prevent the threat.
• Assurance – the effectiveness of the barrier, in all situations, can be assured and there is an understanding of the critically of the barrier and the consequence of absence or malfunction.

4.21 The NHS Never Events policy and framework acknowledges that the human in the loop is fallible and the importance of independence – ‘processes that rely solely on one staff member checking the actions of another or referring to written policies are not strong barriers’ (NHS Improvement, 2018a).

4.22 The significance of dependence when considering human reliability was highlighted in a study in the nuclear industry (Swain and Guttmann, 1983) and independence is seen as critical to barrier effectiveness (Chartered Institute of Ergonomics and Human Factors, 2016). Where one individual is responsible for a number of barriers in a system or process, if a single event could degrade all those barriers (for example if the operator was fatigued) they are not independent and should be treated as one single control or barrier element. Also, where a worker is responsible for checking another operator the process may be affected by the operator being less careful because they know their work will be checked. Conversely, the worker checking may not be as diligent as required because they know and trust the operator and have confidence in their ability (Chartered Institute of Ergonomics and Human Factors, 2016). The white paper quotes a report into an explosion at a fuel storage site:

4.23 All of the controls identified by the HSIB investigation to prevent wrong tooth extraction involved a human element. In many cases, the same person was responsible for several barrier elements in a series of measures or there was a heavy reliance on one team member checking the other in the dental surgery.

4.24 In conclusion, the value of barrier management is summarised by McLeod and Bowie (2018), ‘… when it is done properly, BTA can provide a rich understanding of the controls that are expected to be in place to protect against incidents, how they can fail, and how they need to be implemented, supported and managed. And it can do so without having to make any assumptions about the mechanisms or nature of accident causation.’

5.1 With an understanding of barrier management systems, the investigation analysed the process involved in extracting and reimplanting a tooth in the reference event. Within that process the investigation sought to identify controls that were intended to ensure that the correct tooth was identified and extracted without error. The Trust undertakes approximately 11,000 tooth extractions per year and made the investigation aware that this was the only reported incidence of wrong tooth extraction in over 10 years. Where controls were identified that had not been a factor in the reference event, the investigation considered what may degrade their effectiveness in other circumstances.

Trust Local Safety Standards for Invasive Procedures (LocSSIPs)

5.2 NHS trusts in England are required to use a surgical safety checklist in operating theatres (NHS National Patient Safety Agency, 2009) to prevent incorrect site surgery and ensure safe management of patients. However, for the majority of patients, dental extractions (and other oral surgical procedures) are undertaken in an outpatient setting. The risk of a wrong site procedure, such as wrong tooth extraction, also exists in the outpatient setting.

5.3 Given the emphasis on local process in the national guidance, LocSSIPs were intended to be created by multi-professional clinical teams, involving service users in the design, and implemented against a background of education in human factors and teamworking. To assist dental teams to implement LocSSIPs, the Royal College of Surgeons Faculty of Dental Surgery, together with the Faculty of General Dental Practice and other bodies, published a ‘Toolkit for LocSSIPs for wrong site extraction in dentistry’ on the Royal College of Surgeons website (2018). The toolkit recognised that wrong site surgery in dentistry might not cause ‘significant physical harm when compared to the loss of a limb’ but was ‘potentially symptomatic of problems in the clinical systems and processes of the environment in which it occurs’.

5.4 The toolkit included a process flow diagram with accompanying explanatory notes. Nationally, the investigation saw many good examples of LocSSIPs, designed at trust level, which incorporated the principles from the toolkit with the intention of introducing a standardised process for tooth extraction specifically in the outpatient setting (see appendix 1). LocSSIPs were intended to cover the part of the patient pathway that related specifically to the performance of an invasive procedure; starting at the point at which a patient was admitted to the procedure area and ending at the point at which the patient was discharged from the procedure area. However, the national guidance went on to recognise the importance of contextual factors beyond this procedure area and encouraged organisations to consider the invasive procedure patient pathway as a whole. The HSIB investigation found that some controls intended to prevent wrong tooth extraction sat outside the period covered by admission and discharge.

5.5 At the time of the incident, the Trust where the reference event took place had a LocSSIP in place dated August 2016. The origin of the LocSSIP coincided with a September 2016 deadline set by an NHS England patient safety alert (NHS England, 2015c) for implementation of a LocSSIP based on the National Safety Standards for Invasive Procedures (NatSSIPs). The intent of the patient safety alert was for trusts to ‘Identify all procedures undertaken across clinical settings in your organisation that the NatSSIPs are applicable to’. The Trust undertook invasive procedures in a variety of settings, but the LocSSIP referred to, and used language associated with, the operating theatre. The Trust explained to the investigation that dental procedures requiring general anaesthetic were conducted in surgical theatres in two hospitals operated by another trust. The Trust was required to have a LocSSIP for theatre procedures in the acute trust and the investigation considered the language used in the LocSSIP reflected the specific purpose for which it was intended.

5.6 Most of the staff interviewed by the investigation conducted their work largely in dental surgeries; the limits to scope of practice for a dental therapist (DT) meant they did not deliver care in the operating theatre environment. During interviews the investigation found that awareness of the LocSSIP was poor. Interviewees were then asked about awareness of safety procedures for tooth extraction and the answers did not reveal a significantly different response to the understanding of process.

5.7 Dentists and dental nurses (DNs) who did provide treatment in operating theatres explained they had LocSSIPs for that environment. These clinicians worked in mixed teams with operating theatre staff and anaesthetists in an environment where the World Health Organization surgical safety checklist, required by the LocSSIP, was accepted practice. The investigation saw no evidence that the controls in place in the operating theatre environment were carried across and routinely used in the dental surgery setting.

5.8 The Trust Serious Incident investigation noted that the LocSSIP was held on the Trust intranet and ‘available to all staff’. It went on to note that familiarisation with the LocSSIP was not included in the induction education programme, or preceptorship, for DTs at the time. The reason given by the Serious Incident investigation was that a DT’s remit only extended to the extraction of deciduous teeth under local anaesthetic ‘and these are not currently classed as a Never Event if the wrong tooth is removed’. The LocSSIP stated that it would be best practice for it to apply to both permanent and deciduous teeth.

5.9 The intent of NatSSIPs was that a LocSSIP be in place to ensure that any procedure within the given definition of invasive surgery was carried out safely. It should therefore have been routinely applied to any tooth extraction in all environments. As it was not part of the induction programme at the time of the reference event, the investigation believed staff may have been unaware of the LocSSIP or considered it not relevant to the environment in which they worked. Consequently, there were some important controls, including the team briefing and certain checks, that were absent in the dental surgery but probably present in the operating theatre. The investigation concluded that the nature of the local environment was significant when considering the effectiveness of controls. It cannot be assumed that controls that are present and successful in one environment will automatically be adopted and successful in another.

5.10 The investigation noted that a month after the reference event, the Trust LocSSIP was revised and republished. It drew on the Royal College of Surgeons toolkit and included separate sections for procedures in the surgery and operating theatre. Although not available at the time of the incident, it reinforced certain controls for each environment, such as counting from the mid-line for tooth confirmation. The Trust also included LocSSIPs on the induction for DNs, DTs and dentists.

Procedure identification (pre-surgery)

5.11 The investigation identified several potential error points in the pathway when information was transcribed or relayed prior to the start of treatment. The Trust’s LocSSIP highlighted ‘common failures which could lead to the extraction of a wrong tooth’ which included, ‘Information transfer errors may occur during transfer of planned treatment onto paperwork such as charting, consent forms, theatres lists and white board’. Any such error could result in a clinician referring to inaccurate documentation and subsequently inadvertently planning to undertake the incorrect procedure on a correctly identified patient (see figure 8, threat 1).

5.12 Following referral, the Patient’s details were correctly entered into a specialist dental electronic records system. Along with identifying information such as name, date of birth and NHS number, the system displayed icon alerts to clinicians highlighting details such as communication requirements, medical allergies, special learning needs or mobility considerations which may impact the way treatment was undertaken. For example, the general dental practitioner (GDP) considered the Patient would require an interpreter to assist understanding. This was noted and acted upon. A translator was present for the new patient appointment but was not considered necessary for subsequent visits and the alert was removed.

5.13 The initial handwritten GDP referral contained references to tooth type in longhand (first molars) and position using the alphanumeric adaptation of Palmer notation (see 1.9 to 1.11). The writing was clear and the information was interpreted and correctly transcribed into the dental records by the senior dental officer (SDO). The SDO conducted an examination of the mouth, which included charting of the Patient’s teeth.

Dental charting

5.14 Dental charting is a process during which a dental healthcare professional lists and describes the health of the teeth and gums. Charting usually involves the relaying of information verbally from one dental healthcare professional who is examining the mouth to a second, usually a dental nurse, who is at a computer. The information is entered onto the patient’s electronic record using the keyboard and mouse to form a diagrammatic representation showing all the surfaces of the teeth. The technical vocabulary to describe the three-dimensional mouth for a two-dimensional dental chart is standardised to identify the correct tooth and surface of the tooth with accepted notations for clinical observations such as cavities, fillings or fractures. The investigation heard that standardised terminology is taught during professional training, but commercial electronic recording systems varied in functionality. The investigation also heard that dental clinicians trained in other countries may be more familiar with other terminology and notation systems.

5.15 The investigation was told the Trust required the use of Palmer notation; however, the investigation noted the way charting was conducted was not standardised. Differences included the starting point in the mouth and subsequent system for examining the quadrants of the upper and lower jaw. Other variables in communication style which may affect information transmission and reception within the team included speed of delivery, accent, volume and environmental factors, such as background noise. Physical barriers to communication included wearing facemasks and the lack of visual contact because one team member is focused on the patient’s mouth while the other has to attend to a computer display screen – the positioning of computers in the dental surgeries visited meant that the staff member receiving information had their back to the dental chair. The description of charting given to the investigation did not contain any closed-loop feedback, a method whereby the person receiving the information repeats it back to confirm that it has been correctly received and understood.

5.16 A senior dental nurse told the investigation that the Trust’s electronic record system was part of the induction training for new staff and familiarity improved with use of the system. It was explained that one of the challenges to accurate charting was when the delivery of information was too fast. If this occurred, the DN was taught to request a pause and a slower delivery. The investigation heard during interview that the DT and DN, as well as other clinicians, often worked together on a regular basis so “got used to working together” and became accustomed to individual styles, and a request to “slow down” may be easier in such a well-bonded team. The investigation also heard that staff absence and the requirement to provide services in different geographic locations resulted in variation in team constitution.

5.17 The investigation considered that to work well, dental charting required a high degree of competence, an effective human-machine interface, and good teamwork and communication. The socio-technical controls identified to ensure the correct transfer of information during charting relied predominantly on human performance.

5.18 In the reference event, charting took approximately four minutes during the new patient examination, which lasted a total of 40 minutes. The charting was accurate and the SDO correctly noted during the examination that the clinical situation had changed since the referral six months previously. The deciduous tooth intended for extraction (LLE) was no longer present and the lower left second permanent premolar (LL5) was present in its place and partially erupted. Since the referral, LLD was showing signs of decay and was mobile as was LRE. Both teeth required extraction.

The treatment plan

5.19 Having examined and charted the mouth, the SDO was in a position to make a treatment plan. They noted the plan in the electronic dental record and, at the end of the day, transcribed the plan. It was represented on a second dental chart as ‘planned treatment’. This chart used colour coding to highlight work that was yet to be completed.

5.20 The investigation heard how records were generally updated following treatment. The time available for this in a standard 40-minute appointment was variable. When appointments overran, records would be completed from contemporaneous notes or memory, sometimes hours later and following several other appointments. The Trust had recognised this pressure and made the appointment slot before the lunch break available to clinicians for updating records.

5.21 The investigation found that notes made under pressure of time or not made immediately following the appointment were at greater risk of factual inaccuracy, with workload, stress and time elapsed influencing the likelihood of error.

5.22 Two days after the new patient examination, the SDO submitted an electronic referral requesting a specialist orthodontic opinion on the optimal timing for the extraction of the decayed upper 6 teeth and whether extraction of the lower 6 teeth should be considered. The electronic referral was unsuccessful and, 10 days later, a handwritten referral was completed and submitted by the SDO. Questions posed in the referral were answered in a letter from the specialist two months later. The letter was uploaded onto the system and the advice to extract the two upper 6 teeth was correctly transcribed into the dental records and treatment plan.

5.23 Having already extracted the tooth of most concern (LRE), on receipt of the specialist advice the SDO discussed and agreed a treatment plan with the Patient’s parents during a separate appointment. They noted in the dental record: ‘

PLAN – THERAPY - in comms, thank you
(1) EXT LLD [extract tooth LLD]
(2) F/S LL5 [apply fissure sealant, a temporary plastic coating that is painted on to the grooves of the back teeth to form a protective layer to prevent future decay, to LL5]
(3) LA PRR LL6 [fill and apply a more permanent preventative resin restoration under local anaesthetic]
SDO –
(1) EXT UR6 UL6 if wishes to proceed under LA [extract UR6 and UL6]
(2) REF BACK TO GDP’ [refer back to general dental practitioner]

The SDO then wrote an electronic referral and prescription for the treatment to be completed by the DT. This correctly identified the teeth using alphanumeric Palmer notation with the treatment to be undertaken and the maximum dosage of local anaesthetic for both the filling and extraction.

5.24 The investigation noted that the information presented on the referral to the DT was a true representation of the intended treatment plan and was not a factor in the wrong tooth extraction. This had required either the transcription of handwritten documents into electronic records, electronic records into handwritten documents, verbal information into electronic records or manual transfer of information from one electronic record to another no less than seven times and had involved the input of at least four different healthcare professionals. Due to the protracted period between initial referral and treatment, the Patient’s dentition had changed rendering some aspects of the initial referral obsolete.

5.25 The investigation considered that to be effective controls intended to prevent information transfer errors were reliant on human performance. The controls consisted mostly of a requirement for an individual to crosscheck whether information was correct against other documents available on the record system. It was explained that the review of records is time consuming, relies on the availability of the electronic record system and a thorough review is not always possible in the scheduled time available. It also requires an impetus to query that information might be incorrect; an individual must acknowledge that they may have made an error during transcription or a co-worker must be willing to routinely check if a trusted colleague has made a mistake in transcription. The perception of time pressure and workload can influence individuals to adopt timesaving strategies when applying process in order to achieve their intended goal with an accepted level of performance (Hancock and Warm, 1989). The Trust had recognised the implications of time pressure created by back-to-back appointments with a break for lunch and had blocked the last morning appointment before the scheduled lunchbreak for completion of medical notes and administration.

5.26 The controls intended to prevent a plan to extract the wrong tooth as a result of documentation errors are shown in figure 8. The diagram also highlights some of the degradation factors and controls.

Patient input into procedure identification

5.27 NatSSIPs (NHS England, 2015a) and the Royal College of Surgeons toolkit (2018) encourage the involvement of the patient in identifying the correct procedure where feasible. The 2019 Trust LocSSIP required that ‘When the patient is present, patient identity checks must be undertaken and verification of the teeth to be extracted with the patient/parent/carer’. In the reference event, the treatment plan was originally explained to both of the Patient’s parents by the SDO and agreed with them five weeks prior to the incident. Care then transferred to the DT who reported confirming with the Father on the morning of the appointment that there had been no changes to the Patient’s medical history and gaining verbal consent for the extraction of LLD. During interview the Father referred to “the tooth” but the investigation is unclear to what extent they would have been able to assist with confirming the correct tooth for extraction. The investigation heard from staff that while some patients attending for treatment were fully aware of their procedure, others were often anxious and confused or unaware of the exact source of pain or which tooth was to be extracted. Staff also explained that language and understanding was often a factor and sometimes they had to converse through an interpreter, when one was available. The investigation concluded that the involvement of the patient in procedure identification was a weak control to prevent wrong tooth extraction.

Tooth extraction

5.28 With all the arrangements for the treatment in place, the investigation then studied the pathway from the point at which the Patient arrived for her tooth extraction (see figure 9, threat 2).

Site marking

5.29 NatSSIPs are clear about procedural verification of site marking (marking the part of the body on which the procedure is to be undertaken) and state that ‘surgical site marking is mandatory for all procedures for which it is possible’. They go on to highlight exceptions:

The Trust LocSSIP did not refer to site marking and the investigation was unaware of this as a recognised technique anywhere in general dental practice. In lieu of site marking, the LocSSIP did require that ‘All documentation to be checked by the pre-assessing or operating clinician prior to procedure to confirm that the [electronic] charting, the written notes, hospital cover sheet, consent form and theatre lists are all consistent with respect to the teeth to be extracted’.

Documentation check

5.30 Checking all documentation, in lieu of site marking, is a control intended to confirm the correct procedure and does not assist with the physical identification of the tooth. The investigation noted that there were various layers of administrative controls intended to make this a more reliable process.

5.31 In the reference event, the DT and DN were familiar with the Patient, having treated her in the same surgery three weeks prior to the incident. On that occasion the DT accessed the patient record and printed a copy of the SDO’s prescription. It is not known how thoroughly the dental notes or radiograph were reviewed at this point. The DT did consider the notes sufficiently to inform her decision on a sequence for the prescribed treatment. She elected to conduct the oral hygiene treatment, less invasive application of fissure sealant to LL5 and 6 and the filling to LL6 during the first appointment in the hope of building rapport with the Patient before undertaking the potentially more stressful extraction of LLD on the second appointment. The subsequent treatment on that day was completed in accordance with the prescription and the dental record was updated with comprehensive notes, with a final note for a review of oral hygiene and extraction of LLD on the next visit. The DT also ticked the elements as complete on the printed prescription sheet, which she retained with other patient paperwork in a plastic wallet. The investigation found that the documentation had been checked to some extent prior to the first appointment.

5.32 Three weeks elapsed between the first and second appointment. The Patient’s appointment was the first of the day and was the only extraction. The investigation was advised that the Patient’s treatment was considered the most complex of the day and had been allocated a double appointment of 80 minutes. The DT and DN were scheduled to be working together as a team for the whole day and it was normal practice for the DT to make handwritten notes on the day list to assist the DN in preparing for the surgery. The DN recalled the day list being annotated with ‘Extraction – LLD’. The DN did not recall having a discussion or briefing about the day’s list. Checking the notes required access to the computer and was reliant on human performance and availability of the IT system. The practice of using the annotated printout of the prescription as a reference for treatment on subsequent appointments degraded the control. It was an adaptation that made it less likely that notes would be comprehensively checked on subsequent visits and handwritten notes on the day list introduced another opportunity for transcription error.

5.33 The investigation concluded that the DT had annotated the day list with the correct information at the start of the day, most likely using the pre-printed prescription from three weeks earlier as the reference. The DT and DN then went for a short daily whole-team meeting which took them away from the dental surgery. The investigation considered that the DT had probably reviewed the dental records prior to the first appointment. The DT had developed an individual practice that meant the electronic records were not thoroughly reviewed before the second appointment, but the information available to the team was correct. Checking documentation for consistent information was reliant on socio-technical and human controls such as thoroughly cross-referencing electronic medical records against referral letters and prescriptions to identify error. Factors which degraded the controls included lack of access to functioning IT, lack of familiarity with the system and a lack of independence.

Use of the white board

5.34 As a reinforcement of the procedure during preparation and as a visual reminder throughout the treatment, the Trust LocSSIP stated: ‘Operating clinician responsible for writing the planned treatment on to the white board in theatres where one is available, otherwise hospital cover sheets used.’ The investigation was told that when marked up on a whiteboard, the cross symbol (+) used in Palmer notation to denote the four quadrants of the mouth gives an intuitive, map-like visual representation of the tooth, or teeth, for extraction, especially when orientated behind the patient. At the time of the investigation, the Trust’s dental surgeries were equipped with laminated A4 paper as a substitute for a white board. This had been introduced as part of a recent amendment to the LocSSIP. It is unclear what was in place at the time of the reference event. The Trust Serious Incident investigation was told that, although it would be their normal practice, neither the DT nor the DN annotated the ‘whiteboard’ with a D in the lower left quadrant, to indicate the tooth for extraction.

5.35 Following interviews with a number of staff, the investigation found that the use of a whiteboard did not feature prominently in any description of the extraction process. When asked, staff were unable to demonstrate how the laminated charting sheet would be marked and where it would be placed in the dental surgery. One clinician explained they chose not to annotate it as they considered this another opportunity for information transfer errors.

5.36 The investigation identified that completion of the whiteboard was a control intended to improve team situational awareness and ensure that both the DT and DN had the same mental model for the intended procedure (that is, a shared understanding of what was to happen). From the wording of the LocSSIP and after talking to staff, it was apparent that this control had originated in operating theatre-based practice for extractions under general anaesthetic. While dentists and some DNs were familiar with this environment, DTs and other DNs were not. The investigation found that use of the whiteboard was not embedded in Trust staff practice in the outpatient setting. Although the control had clear ownership, it lacked the traceability, effectiveness, assurance and independence of a reliable control.

Reference to radiograph images

5.37 The NatSSIPs recommend that ‘Reference to radiological imaging may be useful’. The Trust LocSSIP made no specific reference to reviewing images as part of the document check. The subject matter advisor told the investigation that reviewing images improved staff’s understanding of patients’ dentition and allowed for a certain amount of treatment planning. The Royal College of Surgeons toolkit (2018) and Trust LocSSIP both recommend that it is good practice to have the radiograph displayed during treatment. A practical demonstration during interview showed that to view the image on a standard computer monitor in a brightly lit surgery the operator would have to leave the patient. The investigation also heard that computer screens were liable to ‘time out’ to screen saver, the electronic record system could freeze, run slowly or be offline (in one surgery the IT socket was behind the backrest of the office chair and held in with sticky-tape to avoid disconnection). Staff were required to use portable laptops to access records in the acute hospital operating theatres where internet connection was sometimes an issue.

5.38 The Patient’s dental records show an entry at 08:53 hours: ‘medical questionnaire taken’. This indicates that the dental records for the patient were open on the morning prior to treatment, but the DT stated she did not review or display the radiograph on the morning of the incident.

5.39 The investigation was aware of incidents in other trusts where images had been displayed back to front, causing confusion between left and right and resulting in wrong tooth extraction. A clinician shared with the investigation their potential for disorientation between left and right when viewing images displayed in any position other than behind the patient. The investigation considered that the potential for error degraded the radiograph check as an effective barrier.

Tooth identification

5.40 The Trust LocSSIP required confirmation of selection of the correct tooth before forceps were placed on the tooth. The DT recalled identifying the tooth and the area for a local anaesthetic injection, which she then administered. After a short wait, she confirmed anaesthesia by prodding the gum with a sharp point. She identified the tooth again without external confirmation and, having broken the ligament attaching the tooth to the bone, used a set of posterior deciduous forceps to grip the tooth for extraction. The DT stated that the forceps appeared “quite loose” on the selected tooth and she therefore exchanged them with the DN for another set. She then re-applied the forceps and extracted the tooth.

5.41 Based on interviews and the DT’s statement, the investigation undertook a task analysis and concluded there were five occasions when tooth LLD needed to be identified during the procedure (see figure 8).

5.42 The local anaesthetic was effective for the extraction of LL5 which suggests it was administered in the correct lower left area of the mouth. The actions of initially identifying the tooth to administer the local anaesthetic and placing of forceps for extraction will always be punctuated by a pause for the local anaesthetic to take effect. During this time, it would be usual for the DT to talk to the patient to keep them relaxed and distracted from the impending extraction. In the reference event, without the benefit of site marking, the DT then re-examined the mouth, identified the tooth and broke the ligaments. She then reidentified the tooth and placed the first set of forceps, which slipped. The exchange of forceps required communication with the DN to explain the requirement. After a finite period, the DT then reidentified the tooth and extracted it. It is unclear whether the obviously decayed and mobile LLD was correctly identified initially and mis-identified on second application of the forceps, or the wrong tooth was selected initially when the ligaments were broken. The investigation considered it would have needed a positive external check at all five stages of the process to ensure the correct course of action.

5.43 The LocSSIP did require that ‘If there is any uncertainty the procedure should be stopped and the notes/treatment plan reviewed’. The DT reported an issue with the first set of forceps which would have introduced a distraction, but there was no indication of doubt about the procedure or that this caused them to stop and review, or question, their intended actions.

5.44 Staff explained to the investigation the challenge of identifying a tooth, often in a conscious, anxious patient. Determining tooth position requires description of left and right plus upper and lower, with possible consideration of a mixture of permanent and deciduous teeth. Identification is influenced by a variety of factors including position, colour, shape, condition or root structure. In turn these can be influenced by decay, partially erupted teeth resembling other teeth, lone standing teeth or missing teeth.

5.45 The DT stated that the Patient’s LLD looked like a C and that the LL5, to which they had applied fissure sealant in the previous appointment, was small. The SDO who assisted after the extraction also noted that the LL5 had “unusual crown morphology”, meaning the top surface of the tooth did not have a typical structure. On the basis of the statements of the DT and the SDO, it is considered likely that the unusual tooth morphology was a factor in the wrong tooth being extracted.

5.46 The Trust LocSSIP stated: ‘Before forceps are placed on to tooth, the teeth should be counted with a second clinician or dental nurse who understands the annotation to confirm that the correct tooth is being removed.’ The toolkit for LocSSIPs for wrong site extraction in dentistry available at the time (Royal College of Surgeons Faculty of Dental Surgery, 2018) recommended ‘verbal counting of the dentition from midline, whilst pointing with an instrument to each tooth to confirm surgical site (tooth to be extracted) with confirmation from your assistant (of side and countdown), will provide clarification and verification’.

5.47 When asked in interview to explain how they would identify a tooth for extraction, staff all stated, “by counting”. When asked to describe how they would count, there was no consensus for technique. All but one of those questioned described counting starting from the back of the mouth moving forward, not from the centre line towards the back of the mouth. The investigation observed an extraction at the Trust where counting took place from the mid-line to the back of the mouth to identify the tooth prior to placing the forceps. This observation supports the investigation’s finding that there was no consensus on tooth counting methods when identifying teeth.

5.48 With regards to the involvement of the DN in tooth confirmation, one DN explained that they watched and did a visual check, while another explained they did not have the training to identify teeth but could confirm the count. The investigation noted that while DNs have a competence of recording assessments on dental charts (National Examining Board for Dental Nurses, 2020), identification of teeth is outside the General Dental Council core Scope of Practice for DNs (General Dental Council, 2013).

5.49 In interview, DNs expressed varied opinions about the extent to which they would input or challenge the confirmation and decision process for extraction. Although some explained they had received additional training in communicating with patients with specific needs, none of those questioned remembered receiving non-technical skills training in team communication. None of the DTs or DNs questioned by the investigation had received human factors training. The Royal College of Surgeons LocSSIPs toolkit highlights:

5.50 In a wrong tooth extraction incident investigated and reported by another trust, the investigation found:

5.51 The HSIB investigation noted that the DT had not used the DN to confirm the tooth for extraction and the DN had not challenged the DT to confirm the correct tooth to be extracted. The DN was subsequently the first to notice that the wrong tooth had been extracted so had appropriate knowledge to assist.

5.52 The investigation found that the lack of consensus among staff regarding tooth identification using counting had the potential to create confusion. This potential for confusion was further aggravated by the absence of role definition for the DN. The control of confirming tooth identity was degraded by the lack of standardisation in process, definition of individual responsibilities and competencies. The investigation found that for external confirmation to be effective as a control it had to be practised in a culture which encouraged challenge.

5.53 The investigation found that the layers of controls in place to ensure the correct tooth was identified, selected and extracted were absent or degraded to varying degrees. Although they represented safeguards, there was a lack of a strong, systemic barrier to wrong tooth extraction.

Recognising and recovering from the wrong tooth extraction

5.54 During the reference event, once the permanent tooth had been extracted, it was passed to the DN to be cleaned and placed in a small envelope for the Patient to take home. The DT then went to the computer and at 09:07 hours entered ‘completed – extraction LLD’. This was 14 minutes after the previous entry that recorded the medical questionnaire had been taken; this indicated to the HSIB investigation that the DT was under the impression the correct procedure had been undertaken.

5.55 The investigation was told that the DT would have examined the tooth to ensure that it had been removed without leaving any of the root retained in the gum. A dental subject matter advisor explained that there is a noticeable difference between the root structure of a first deciduous molar (LLD) and a permanent second premolar. Neither the DT nor the DN noticed at this point that LL5 had been extracted in error. The investigation considers that the entry in the dental records indicated that the DT considered the extraction had gone to plan. Equally, the DN washed the tooth and placed it in the envelope without noticing that it was LL5 not LLD that had been extracted. During the extraction there was no obvious reminder of the intended plan displayed; controls such as marking on the whiteboard and a verbal confirmation were absent.

5.56 The error was first noticed when the DN inspected the wound and identified that LL5 was missing. The team acted swiftly to seek advice from a senior colleague, who in turn sought advice from a consultant by telephone. The tooth was placed in saline solution and had only been out of the gum for approximately seven minutes when it was reimplanted.

5.57 There was no standard operating procedure in place for emergency reimplantation and therefore no controls in place. The investigation found that the measures that did recover the situation relied on human performance, teamwork, experience and knowledge to be successful. 5.58 The recovery controls and degradation factors are shown in figure 11. Figure 12 shows the high level Bowtie diagram which gives a visual representation of the controls in place and the nature of those controls.

Patient identification

5.58 Identifying the correct patient for the correct care is critical to avoid a mismatch between a patient and the intended intervention and avoid a wrong site surgery. This is the subject of a separate HSIB investigation, ‘Wrong site surgery – wrong patient’, the results of which will be published in due course. The Trust LocSSIP highlighted the potential for ‘time pressures both before treatment and during treatment’ and ‘changes to list order’ to result in error.

5.59 The Trust could receive up to 500 referrals per month which impacted the referral to treatment time and introduced a pressure to reduce waiting times. The investigation was told there was often reactive planning to fill cancelled appointments at short notice, resulting in changes to the planned appointment lists.

5.60 In the reference event, the controls were effective and patient identification did not influence the wrong tooth extraction in this event. The investigation found that the outpatient referral system and patient identification relied on human controls to be effective and lacked the robustness required of a barrier.

6.1 Using the analysis of the reference event, the investigation also explored how a more formal approach to barrier management could help identify where systemic weaknesses are present and drive improvements in patient safety.

6.2 The importance of understanding control effectiveness was demonstrated when a trust introduced a correct site surgery checklist (for outpatient tooth extractions) in 2009. Despite the new checklist, the trust went on to experience five incidents of wrong tooth extraction over the next three years. A subsequent article (Saksena et al, 2014) detailed that ‘while introduction of an outpatient correct site surgery checklist was an important step in the right direction, there was clearly more that was required’. By a system of investigating when things went wrong, process and risk mapping, plus a monthly/annual audit programme, the trust found:

• teams were unclear about their roles in undertaking checks and there was a non-standardised approach
• there was confusion about who should undertake checks and write on the whiteboard
• often only the operating team member verified certain checks before the patient was treated
• the verbal ‘time-out’ rarely took place.

6.3 The trust then undertook a process of clarifying ownership, traceability, specificity and assurance of the various factors within the mapped process to improve effectiveness of those controls. The outcome was no further wrong tooth extractions in the 30 months following the improvement work. The article noted various improvements that had been instrumental in this success, including:

• a verbal ‘time-out’ pause and confirmation observed in 100% of cases
• staff worked as a team with more confidence about their roles and responsibilities
• 98% of staff stated if used correctly the checklist and policy prevented wrong tooth extraction
• staff valued use of the whiteboard as a communication and verification tool.

The report concluded:

6.4 This work, which included observational audit, demonstrated the concept that often frontline staff act to avoid adverse events despite weaknesses in, or the absence of, suitable barriers. The audit process reflected that controls cannot be considered to be static, as staff adjust their actions to influence the effectiveness of the control or compensate for weaknesses that may develop. Ultimately, an accumulation of events or factors in the working environment may erode the effectiveness with which staff are able to compensate for absent or ineffective barriers and prevent unintended consequences (Reason, 2008).

6.5 The investigation considers that this example of barrier management was effective at identifying unreliable or inadequate controls. By applying some of the properties required of an effective barrier management system, the trust was able to substantially mitigate the risk of wrong tooth extraction. Teamwork and checking improved with understanding of roles and responsibilities, and there was greater assurance and auditability. The HSIB investigation considered that there remained a lack of independence. Controls still relied on a single staff member completing a series of checks or ‘one staff member checking the actions of another or referring to written policies’ which, by definition in the Never Events policy and framework, cannot constitute ‘a strong barrier’.

Exclusions to the list of Never Events

6.6 The list of Never Events (NHS Improvement, 2018b) contains several exclusions. First on the list is ‘removal of wrong primary (milk) teeth unless done under a general anaesthetic’. The investigation understands that this exclusion relates primarily to the level of potential harm. The list specifically states that ‘teeth extracted in error that are immediately reimplanted’ should be included as Never Events. The investigation considered that both events would be indicative of a similar compromise in controls intended to prevent wrong tooth extraction. If considering systemic barriers to this type of incident, the exclusion of deciduous teeth suggests a focus on level of harm rather than on reliable systems that are in place to mitigate any wrong tooth extraction.

6.7 The investigation is clear that wrong tooth extraction is an avoidable, serious incident and should be investigated at an appropriate level. However, the involvement of exceptions based on outcome and not process introduces an element of unnecessary confusion.

Summary

6.8 This report builds on the HSIB national learning report which provided a themed analysis of Never Events (Healthcare Safety Investigation Branch, 2021) and found that controls were often administrative and ineffective. The report went on to make several safety recommendations including:

6.9 This investigation into wrong tooth extraction found a consensus in the literature review that a barrier should be reliable and adequate as a primary defence to stop a threat developing into an unwanted event and outcome. By using a system of barrier management to investigate and analyse the process of tooth extraction in the outpatient setting, the investigation found that the controls in LocSSIPs intended to prevent wrong tooth extraction were reliant on human elements to make them effective. They often had none of the systemic assurances and ownership expected of a barrier in other safety-critical industries. They lacked independence and used layers of safeguards in an attempt to achieve a level of reliability. When considering the effectiveness of controls, an external review of Never Events at a trust noted:

While there were many safeguards to prevent wrong tooth extraction, the investigation did not identify any ‘strong, systemic barriers at a national level’ which made this form of serious incident ‘wholly preventable’.

6.10 Following inclusive discussions between the NHS England and NHS Improvement safety team, key stakeholders representing dental professionals and HSIB in November 2020, it was agreed that wrong tooth extraction did not meet the criteria for a wrong site surgery Never Event. NHS England and NHS Improvement announced in February 2021 that wrong tooth extraction would be removed from the list of Never Events, effective as of April 2021.

HSIB notes the following safety action

6.11 Although wrong tooth extraction has been removed from the list of Never Events as it did not meet the criteria, the requirement to investigate when something goes wrong remains an inherent part of any effective safety and barrier management system. Wrong tooth extraction remains a notifiable incident to the Care Quality Commission (2013) and understanding how controls should have responded and how they were compromised resulting in an unintended outcome, is fundamental to a safety culture.

6.12 The investigation found that the description of what constitutes a ‘control’ or ‘barrier’ is not clearly defined in the NHS Never Events policy and framework and also inconsistent with other literature. The investigation also notes the NHS Never Events policy and framework document offers minimal guidance on the categorisation of barriers to help organisations understand their purpose. This makes a clear understanding how controls should work difficult and implementation in local policy open to variation. By having staff qualified to proactively identify where controls are robust and where they are weak using a tool such as bowtie analysis, the reliability of systems can be built to a point where risk is as low as practical.

HSIB makes the following safety recommendation

Appendix 1: Example dental extraction LocSSIP from the toolkit for LocSSIPs for wrong site extraction in dentistry (RCS, 2018).

Appendix 2: Illustrative bowtie diagrams created during the investigation.

Care Quality Commission. (2013) Notifications required by the Health and Social care Act 2008 Guidance for English NHS providers [Online]. Available at https://www.cqc.org.uk/sites/default/files/documents/statutory_notifications_for_nhs_bodies_-_provider_guidance_v6.pdf (Accessed 10 November 2020).

Chartered Institute of Ergonomics and Human Factors. (2016) Human factors in barrier management. A white paper by the Chartered Institute of Ergonomics and human factors.

Civil Aviation Authority. (n.d.) Control effectiveness [Online]. Available at https://www.caa.co.uk/Safety-initiatives-and-resources/Working-with-industry/Bowtie/Safety-risk-priorities/Control-effectiveness/ (Accessed 11 November 2020).

Civil Aviation Authority. (2014) Bowtie document library [Online]. Available at https://www.caa.co.uk/uploadedFiles/CAA/Content/Standard_Content/Safety_initiatives_and_resources/Bowtie/Templates/Document_library/Umbrella%208.1%20Human%20Performance.pdf (Accessed 25 February 2021).

Civil Aviation Authority. (2020) CAA strategy for bowtie risk models. An updated briefing from the Civil Aviation Authority [Online]. Available at https://publicapps.caa.co.uk/docs/33/CAP1329%20Bowtie%20Strategy%20MAY%202020.pdf (Accessed 12 November 2020).

De Ruijter, F. G. (2016) The bowtie method: a review. Safety Science, 88, 211-218. General Dental Council. (2013) The scope of your practice [Online]. Available at https://www.gdc-uk.org/information-standards-guidance/standards-and-guidance/scope-of-practice (Accessed 17 January 2021).

Hancock, P. & Warm, J., 1989. A dynamic model of stress and sustained attention. Human Factors, Volume 31, pp. 519-537.

Health and Safety Executive. (2008) RR637 Optimising hazard management by workforce engagement and supervision. Health and Safety Executive. (2011) Leadership and worker involvement toolkit. Management of risk when planning work: the right priorities [Online]. Available at https://www.hse.gov.uk/construction/lwit/assets/downloads/hierarchy-risk-controls.pdf (Accessed 29 December 2020).

Health and Safety Executive. (2019) Introduction to risk management [Online]. Available at https://www.hse.ie/eng/about/qavd/riskmanagement/ (Accessed 19 March 2021).

Healthcare Safety Investigation Branch. (2021) National learning report. Never Events: analysis of HSIB’s national investigations.

International Association of Oil and Gas Producers. (2016) Standardization of barrier definitions.

McLeod, R. and Bowie, P. (2018) Bowtie analysis as a prospective risk assessment technique in primary healthcare. Policy and Practice in Health and Safety, 16 (2), 177-193.

National Examining Board for Dental Nurses. (2020) National Examining Board for Dental Nurses national diploma syllabus [Online]. Available at https://www.nebdn.org/app/uploads/2020/03/NEBDN-National-Diploma-Syllabus-2020-Website-Version-V2.pdf (Accessed 18 March 2021).

National Institute for Occupational Safety and Health. (2020) Hierarchy of controls [Online]. Available at https://www.cdc.gov/niosh/topics/hierarchy/default.html (Accessed 11 March 2021).

NHS Digital. (2016) Hospital admitted patient care activity, 2015-2016 [Online]. Available at https://webarchive.nationalarchives.gov.uk/20180328130140/http://digital.nhs.uk/catalogue/PUB22378 (Accessed August 2020).

NHS Digital. (2019) Hospital outpatient activity 2018-19 [Online]. Available at https://digital.nhs.uk/data-and-information/publications/statistical/hospital-outpatient-activity/2018-19 (Accessed 5 December 2020).

NHS England. (2014) NHS England Surgical Never Events Task Force report. NHS England. (2015a) National Safety Standards for Invasive Procedures (NatSSIPs) [Online]. Available at https://improvement.nhs.uk/documents/5405/NatSSIPs_Final_updated_June_2019.pdf (Accessed March 2020).

NHS England. (2015b) Introductory guide for commissioning dental specialties [Online]. Available at https://www.england.nhs.uk/commissioning/wp-content/uploads/sites/12/2015/09/intro-guide-comms-dent-specl.pdf (Accessed 17 January 2021).

NHS England. (2015c) Patient safety alert. Supporting the introduction of the National Safety Standards for Invasive Procedures.

NHS England and NHS Improvement. (n.d.) Never Events data [Online]. Available at https://improvement.nhs.uk/resources/never-events-data/ (Accessed 19 November 2020).

NHS Improvement. (2018a) Never Events policy and framework. Revised February 2012 [Online]. Available at https://improvement.nhs.uk/documents/2265/Revised_Never_Events_policy_and_framework_FINAL.pdf (Accessed March 2021).

NHS Improvement. (2018b) Never Events list 2018 [Online]. Available at https://improvement.nhs.uk/documents/2899/Never_Events_list_2018_FINAL_v7.pdf (Accessed 19 November 2020).

NHS Improvement. (2018) NatSSIPs implementation survey: full results.

NHS National Patient Safety Agency. (2009) Patient safety alert. WHO Surgical Safety Checklist.

Oxford University Hospitals. (2016) External review of ‘Never Events’ that occurred at Oxford University Hospitals NHS Trust during the period 13 September 2013 to 26 March 2015.

Pemberton, M. and Ashley, M. (2017) The use and understanding of dental notation systems in UK and Irish dental hospitals. British Dental Journal, 223 (6), 429-434.

Porto, G. (2009) Safety by design: ten lessons from human factors research. Journal of Healthcare Risk Management, 21 (4), 45-52.

Reason, J. (1990) Human Error. Cambridge: Cambridge University Press.

Reason, J. (2008) The Human Contribution, Unsafe Acts, Accidents and Heroic Recoveries. Surrey: Ashgate.

Royal College of Surgeons Faculty of Dental Surgery. (2018) Tool kit for ‘Local Safety Standards for Invasive Surgical Procedures’ (LocSSIPs) for wrong site extraction in dentistry [Online]. Available at https://www.rcseng.ac.uk/dental-faculties/fds/publications-guidelines/locssips-toolkit-dental-extraction/ (Accessed 4 December 2019).

Saksena, A., Pemberton M. N., Shaw, A., Dickson, S. and Ashley, M. P. (2014) Preventing wrong tooth extraction: experience in development and implementation of an outpatient safety checklist. British Dental Journal, 217 (7), 357-362.

Sklet, S. (2006) Safety barriers: definition, classification, and performance. Journal of Loss Prevention in the Process Industries, 5 (19), 494-506.

Swain, A. and Guttmann, H, for the US Nuclear Regulatory Commission. (1983) Handbook of human reliability analysis with emphasis on nuclear power plant applications.

Trbojevic, V. M, for the Health and Safety Executive. (2008) Optimising hazard management by workforce engagement and supervision.

Wright, S., Ucer, T. and Speechley, S. (2017) Errors and adverse events in dentistry: a review. Journal of Primary Care Dentistry, 9 (1), 142-146.

We welcome feedback on our investigation reports. The best way to share your views and comments is to complete our online feedback form.

This document, or parts of it, can be copied without specific permission providing that the source is duly acknowledged, the material is reproduced accurately, and it is not used in a derogatory manner or in a misleading context.