Investigation report: The use of an appropriate flush fluid with arterial lines

This investigation aimed to understand the risks for patients associated with blood sampling from arterial line systems used in adult critical care. An arterial line is a system used to continuously monitor a patient’s blood pressure and intermittently monitor blood glucose levels by taking a blood sample from the arterial line. A thin tube called a cannula is inserted into an artery, usually in the person’s forearm, and tubing is used to attach a device called a transducer. This is connected to a bag of fluid (the flush fluid) and the pressure change of the fluid within the connecting plastic tubing transmits to an electrical monitoring device. This device displays the blood pressure within the artery as a continuous wave line on the monitoring screen. Arterial lines are used in other areas of healthcare and the findings of this investigation may also be relevant to these areas.

The selection and attachment of the incorrect flush fluid is a recognised risk in the use of arterial lines. When a flush fluid contains glucose rather than saline (0.9% sodium chloride) and a blood sample from an arterial line is contaminated, a false high blood glucose is recorded. This will mislead the clinician who may start the patient on insulin, which controls blood sugar levels. This may lead to unrecognised and dangerous levels of hypoglycaemia (low blood glucose levels), which can lead to the patient going into a coma. Careful monitoring of blood glucose levels has become established practice for critical care patients. It provides the information clinicians need to understand whether treatment is appropriate to correct an imbalance in blood glucose levels. Blood glucose monitoring using an arterial blood sample is recommended in critical care environments.

As an example, which is referred to as ‘the reference event’, the investigation reviewed the care of a patient named Keith. Keith received treatment to drain his gallbladder but then became very unwell. He was admitted to a critical care unit with significantly low blood pressure and sepsis (a reaction to infection that causes a person’s body to damage its own tissues and organs). An arterial line was inserted, and the incorrect flush fluid was connected to the transducer. The use of the incorrect flush fluid led to the contamination of the blood samples taken from the site of the arterial line, which consequently misled clinicians to give Keith an unnecessary and potentially harmful treatment.

This investigation’s findings, safety recommendations and safety observations aim to demonstrate the risk associated with the set-up of the arterial monitoring line and use of the correct flush fluid and to improve care for patients across the NHS.

Keith was 66 years old and was recently retired. After feeling unwell for 25 days and attending medical appointments to investigate his gallbladder, he was admitted to hospital. Following an investigation and the drainage of his gallbladder his condition deteriorated. He was admitted to the critical care unit late in the year of 2020 with sepsis and very low blood pressure. Keith was medically unstable and required numerous medications and devices to be attached to him to enable accurate monitoring of his condition.

The insertion of the arterial line was completed around 16:20 hours. The doctor inserted a cannula into an artery in Keith’s arm, while a nurse went to look for the correct equipment, an arterial line (marked red) and a 500ml bag of saline (sodium chloride 0.9%) flush fluid. Saline is the recommended flush fluid for an arterial line and other fluids should not be used for this purpose. The nurse looked first within the ‘lines’ drawers at the bedside (intended to include all necessary equipment for the insertion of an arterial line). When unable to find the correct arterial line transducer set, the nurse looked in two other areas of the critical care unit. They also collected the bag of saline flush fluid from the drug cupboard. The nurse was aware that the cannula had been inserted and the blood would clot and block the line without the arterial line and saline flush fluid attached. They returned to the bedspace with a blue line central venous transducer set. The line and bag of flush fluid were checked and attached. The risk of attaching a blue transducer line was recognised by staff, as a blue transducer line is intended to be connected via a vein and can indicate it is safe to administer medication, which would not be safe through an arterial transducer line.

The need to change the line was prioritised by staff but the nature of Keith’s condition created several competing tasks. Keith’s blood glucose levels were found to be low. A different nurse went to retrieve a bag of glucose so that a glucose infusion (where glucose in a liquid solution is delivered via a patient’s vein) could be commenced to increase Keith’s blood glucose levels. Unable to locate the required strength of glucose, the nurse returned with two alternative strengths for the doctor to select for treatment. While collecting the glucose the nurse also found and returned with the correct red arterial transducer line. The red arterial transducer line and remaining bag of glucose were left together by the bed. The other nurse interrupted their current task and collected both items to replace the blue transducer line that was in position with the correct red arterial transducer line. They replaced the central venous transducer set (blue line) and the correct saline flush fluid with the arterial transducer set (red line) and the incorrect bag of fluid containing glucose as the flush fluid.

Subsequent blood samples taken from Keith’s arterial line were contaminated with the flush fluid containing glucose. This led clinicians to conclude that Keith was suffering from high blood glucose levels and at risk of further harm. An infusion of insulin, the required treatment for high blood glucose, was administered for approximately 8 hours. This treatment reduced Keith’s blood glucose levels to below the recommended limit. The incorrect fluid was identified the next day, during morning safety checks completed by the nurse taking over Keith’s care. The treatment of insulin was stopped, and glucose administered to correct the blood glucose levels.

Keith had a brain scan on the same day, which concluded at that time there was no neurological damage associated with the abnormal blood glucose levels. He was discharged from the critical care unit 15 days after his admission following care for his underlying condition. Sadly, Keith died of COVID-19 later during his hospital stay.

HSIB was contacted by the Department of Health and Social Care in response to receiving a prevention of future deaths notice from a coroner’s investigation into the death of a woman aged 57 years. The investigation was completed on the 7 September 2020 and concluded that the use of a solution containing glucose instead of saline, the recommended fluid to flush an arterial line, contributed to the patient’s death.

HSIB identified Keith’s case as a similar patient safety incident. Although the event did not result in the death of a patient, it involved a similar sequence of events and the same error was identified.

The key findings from the investigation include:

• The physical layout and design of the clinical and storage areas will influence how reliably staff are able to select and collect similar-looking equipment and medication.
• The labelling of bags of fluids, similar looking medications and manufacturers’ packaging reduce the reliability of selecting the correct flush fluid in the context of a critical care unit with time pressures and high workloads.
• The procurement and design of arterial transducer line equipment, the pressure infusion bags and transducer, do not assist in the identification of the incorrect flush fluid or prevent contamination from the flush fluid of a blood sample taken from the arterial line. Alternative equipment, for example transparent pressure infusion bags and closed arterial transducer lines, are currently available to the NHS. These may reduce the risk but are not routinely in use.
• Challenges in the provision of a consistent suitable workforce and high workloads have a detrimental effect on the safety controls currently relied upon to avoid or identify the risk of using the wrong flush fluid. Safety checks and training lack resilience to organisational pressures regularly experienced within critical care units.
• There can be a delay in identifying the contamination with glucose of an arterial line blood sample due to a normalisation and acceptance that critically ill patients may have altered blood glucose levels and require insulin treatment, and a perceived low risk associated with the use of a flush fluid.
• The design of systems to record and monitor information relevant to the arterial transducer line system and blood glucose levels do not easily alert staff to the potential use of the wrong flush fluid.
• Recommendations issued over the last 14 years by national safety bodies and professional healthcare organisations to address the safety of blood sampling associated with arterial lines have not been effectively implemented.

The intention of this safety recommendation is to provide design guidance for manufacturers to manage the risk associated with fluid selection. All aspects of label design should be considered this recommendation is broader than the judicious use of colour as the approach to increasing label safety.

The intention of this safety recommendation is to ensure fluid labels can be consistently read from all directions at all times when the pressure infusion bag is inflated.

The intention of this safety recommendation is to increase awareness of and action on known risks related to the design of the medical devices.

The intention of this safety recommendation is to assure appropriate action is taken to manage the known risks related to the design of the medical devices.

The intention of this safety recommendation is for the Association of Anaesthetists to revise existing national guidance in collaboration with all relevant healthcare professionals including the following clinical areas: critical care, theatres and emergency departments.

The intention of this safety recommendation is to provide assurance that NHS providers have implemented the future national guidance.

1.1.1 Critical care refers to a specialist area of healthcare that cares for seriously ill patients with complex medical needs. Critical care patients may have single or multiple organ failure and require close observation and medical interventions.

1.1.2 Critical care service specifications indicate that one nurse may care for two patients, except for patients that may require advanced respiratory (breathing) support and additional equipment to manage multiple organ failure. For these patients a ratio of one patient to one nurse is required (NHS England, 2019; The Faculty of Intensive Care Medicine and Intensive Care Society, 2019).

1.1.3 The guidance for the provision of critical care services highlights the infrastructure and design of the physical environment as essential to support safe and effective care (The Faculty of Intensive Care Medicine and Intensive Care Society, 2019). This guidance reflects common challenges for existing critical care units which include good visibility of patients, adequate storage, provision of space and natural light.

1.2.1 The amount of sugar in the blood is known as the blood glucose level. The blood glucose levels of a healthy person should be between 4 and 8 millimoles per litre (mmol/litre) (Diabetes.co.uk, 2022). Blood glucose levels may fluctuate during a severe illness (Kerr et al, 2017). It is recognised that control of blood glucose levels is associated with better patient outcomes (Mesotten et al, 2015).

1.2.2 Hypoglycaemia is when the blood glucose level is less than 4mmol/litre (National Institute for Health and Care Excellence, n.d.). This will occur in approximately 2 to 3% of critical care patients (Badawi et al, 2012). Hypoglycaemia can inhibit the nervous system and prolonged periods of hypoglycaemia may cause brain damage or death (Association of Anaesthetists of Great Britain and Ireland, 2014).

1.2.3 Mild hyperglycaemia (blood glucose levels between 10.0 and 12.2mmol/litre) is common in critically ill patients and is a stress response to the severity of an illness (Marik and Egi, 2014). The clinical implications of hyperglycaemia are poor clinical outcomes, higher death rate and longer stays in critical care units (Stapleton and Heyland, 2022).

1.2.4 Careful monitoring of blood glucose levels has become established practice for critical care patients. It provides the information clinicians need to understand whether treatment is appropriate to correct an imbalance in a patient’s blood glucose levels. Blood glucose monitoring using an arterial blood sample is recommended to avoid inaccuracies associated with more commonly used blood glucose tests obtained by pricking the person’s finger (known as a capillary blood test) (Mesotten et al, 2015).

1.3.1 An arterial line is a system used to continuously monitor a patient’s blood pressure and to monitor blood glucose levels. It has five separate components (see figure 1):

• a cannula (a plastic tube inserted over a sharp needle)
• arterial transducer set (short and stiff plastic tubing)
• 500ml bag of saline flush fluid (0.9% sodium chloride)
• a pressure bag to inflate over the bag of saline flush fluid
• a transducer (which transmits the blood pressure recording).

1.3.2 The 500ml bag of saline (0.9% sodium chloride) is often referred to as the ‘flush’ fluid as its function is to maintain the patency (openness) of the tubes, prevent blood from clotting and flush the arterial tubing. Clinical staff will also draw the flush fluid through the system to prevent clotting of residual blood when taking an arterial line blood sample.

1.3.3 The transducer records the pressure change of the saline flush fluid within the plastic tubing. This change in pressure is transmitted to an electrical monitoring device, which reflects the blood pressure within the artery as a continuous wave line on the monitor. The use of an arterial line is considered the ‘gold standard’ for the monitoring of a patient’s blood pressure. This enables regular blood tests to obtain an accurate and real-time continuous monitoring of a patient’s vital signs (Plowright and Sumnall, 2022).

1.3.4 There are several steps to setting up an arterial line (Plowright and Sumnall, 2022). Depending on local policies, these steps may be completed by one or two members of staff, including a doctor and a nurse. The process for setting up an arterial line includes (but may not occur in this order):

• insertion of the arterial cannula into the patient’s artery in the forearm
• collection of transducer and tubing, 500ml bag of saline (flush fluid) and pressure bag
• connecting a 500ml bag of saline to the tubing extending from the transducer
• hanging the 500ml bag of saline covered by a pressure bag, which is inflated to 300 millimetres of mercury (mmHg)
• attaching the transducer cable to the transducer tubing and the bedside monitor
• attachment of tubing and transducer to arterial cannula site and labelling to identify as arterial line
• checking the measurement of arterial blood pressure via the bedside monitor.

1.3.5 The process for taking a blood sample from an arterial line will vary depending upon the type of transducer system used. There are two main types of transducer systems: ‘open’ and ‘closed’. The difference between these is the length of tubing between the cannula and the access point, known as a ‘port’, where a blood sample can be taken. This distance is referred to as the ‘dead space’ and described in more detail below.

1.3.6 An ‘open’ system (see figure 2) has a ‘3-way tap’ as a port close to the arterial catheter site. Staff attach a syringe to the port to remove residual flush fluid and will then take a blood sample from the same port using a blood sampling syringe. Guidance recommends that an amount of blood equivalent to three times the volume of the dead space be removed to reduce the contamination of a blood sample with the flush fluid (Medicines and Healthcare products Regulatory Agency, 2014a). Other guidance highlights that even taking five times the dead space, contamination from the flush fluid is still possible when using an ‘open’ system (Association of Anaesthetists of Great Britain and Ireland, 2014).

1.3.7 A ‘closed’ system (see figure 2) has a port beyond a separate sampling site (from which blood can be taken). This enables staff to remove the residual flush within the dead space, while conserving the patient’s blood. A closed system is reported to reduce the chance of contamination of a blood sample (Association of Anaesthetists of Great Britain and Ireland, 2014).

1.4.1 The selection and attachment of the wrong flush fluid is a recognised risk in the use of arterial lines. When a blood sample from an arterial line is contaminated with flush fluid that contains glucose, a false high blood glucose is recorded. This will mislead the clinician who may initiate treatment in the form of insulin. The inappropriate administration of insulin to a patient may lead to unrecognised and dangerous levels of hypoglycaemia. Arterial lines are used in other areas of healthcare and the findings of this investigation may also be relevant to these areas.

1.4.2 Public awareness of this safety concern has been raised by the reporting of the death of Susan Warby (who was mistakenly given glucose instead of saline through an arterial line) and the subsequent prevention of future deaths report issued to the Department of Health and Social Care. This report identified systemic issues (that is, issues inherent to the healthcare system as a whole rather than isolated problems) that warranted change to protect the public (Courts and Tribunals Judiciary, 2020).

1.4.3 There is a large body of further evidence of the multiple systemic factors believed to contribute to the risks associated with the wrong flush fluid used in conjunction with arterial transducer line systems. These include:

• look-alike labelling and packaging of intravenous fluids (National Patient Safety Agency, 2008a)
• storage, clarity and ease of selection of flush fluids from working environment and storerooms (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013)
• the physical environment including lighting and ability to see text within labels (Gupta and Cook, 2013)
• inadequacy in checks of flush fluids (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013; National Patient Safety Agency, 2008a)
• staffing and caseload may influence adherence to local procedures (Gupta and Cook, 2013)
• obscured labelling of intravenous fluid because of the need to cover with a pressure bag (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013; National Patient Safety Agency, 2008a)
• blood sampling contaminated by inadequate flushing (Gupta and Cook, 2013; National Patient Safety Agency, 2008a)
• design of the arterial line may increase the likelihood of the contamination of a blood sample with glucose-based flush fluid (Gupta and Cook, 2013; National Patient Safety Agency, 2008a)
• presentation of data and glucose levels within medical records may inhibit the detection of the wrong flush fluid as the cause of altered blood glucose levels (Gupta and Cook, 2013; National Patient Safety Agency, 2008a).

Existing recommendations to address the known risks

1.4.4 There have been three key national attempts to address the risk of the wrong flush fluid used with arterial lines. These stem from the National Patient Safety Agency (NPSA) in 2008 (National Patient Safety Agency, 2008a), the Medicines and Healthcare products Regulatory Agency (MHRA) in 2014 (Medicines and Healthcare products Regulatory Agency, 2014a) and the Association of Anaesthetists of Great Britain and Ireland in 2014 (Association of Anaesthetists of Great Britain and Ireland, 2014).

1.4.5 In 2008 the NPSA issued a rapid response report to highlight the risk associated with the use of the wrong flush fluid with arterial lines. This document states ‘sampling from arterial lines is risky’. This report asked healthcare providers for an immediate response which included:

• increased awareness of the risk
• clearly labelled arterial lines
• assurance of the prescription and checking of infusion (a volume of liquid given to a patient through a vein) fluids
• only saline to be used as an arterial line flush fluid
• visibility and clarity in the labelling of fluid bags irrespective of the use of a pressure bag.

A national review of previously issued safety alerts, published in 2022, confirmed that this NPSA rapid response report was an enduring standard which remained valid, is unlikely to change in the immediate future and ‘should already be embedded systematically across NHS provider organisations’ (NHS England and NHS Improvement, n.d.).

1.4.6 The MHRA issued a drug safety update in 2014 stressing the risk associated with the use of glucose solutions compared to saline in arterial lines (Medicines and Healthcare products Regulatory Agency, 2014a). It recommended the use of saline as the solution of choice and that staff ‘remain vigilant’ when selecting solutions as similarities between glucose and saline bags of fluid means that ‘confusion may occur’. When drawing a blood sample, a minimum volume of three times the dead space of the cannula (tubing) should be discarded first to avoid contamination in the event of the wrong flush fluid being used.

1.4.7 The Association of Anaesthetists developed a working party to address the incorrect use of arterial flush fluids (Association of Anaesthetists of Great Britain and Ireland, 2014). It recognised that the NPSA guidance and good sampling practices were insufficient to prevent patient harm or death due to the consequence of sample contamination. The working party completed a robust systems analysis and issued a comprehensive set of system-wide recommendations which included:

• guidance to store saline separately from other infusion fluids
• double checking of prescribed saline fluids on setting up an arterial line, and checking at least once during a shift
• pressure bags to have a fully transparent front panel, and use of ‘closed’ arterial transducer line systems
• vigilance to glucose thresholds – an unexpectedly high glucose level should trigger a medical and equipment review
• starting or increasing an insulin infusion based on samples taken from an arterial line must require a medical review
• trends in blood glucose levels over time are more easily identified from a graphical representation.

1.4.8 The NPSA and the Association of Anaesthetists both conclude that manufacturers should develop a universal safer system to address this problem. This would require consideration of potential engineered solutions (that is, solutions built into the design of the equipment) to prevent or minimise the risk.

3.1.1 The Department of Health and Social Care contacted HSIB following a prevention of future deaths notice from a coroner’s investigation into the death of Susan Warby (Courts and Tribunals Judiciary, 2020). The coroner’s investigation was completed on the 7 September 2020 and concluded that the use of a solution containing glucose instead of saline contributed to the patient’s death.

3.1.2 HSIB identified Keith as a man aged 66 years who experienced a similar patient safety incident while being treated in an NHS critical care unit. Although the event did not result in Keith’s death, a similar sequence of events occurred and the same error was identified.

3.2.1 HSIB conducted an initial scoping investigation which determined that the patient safety concern met the criteria for investigation (see below). HSIB’s Chief Investigator authorised a national investigation.

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

Arterial transducer line systems are routinely used for patients admitted into the critical care environment. They are essential to ensure accurate and continuous monitoring of a patient’s vital signs, which informs clinical decision making and the management of the patient’s condition.

An arterial transducer line requires the use of a saline flush fluid to ensure the connecting tubes remain open. If an incorrect flush fluid containing glucose is used, the contamination of any blood sample taken from the arterial transducer line may falsely indicate that the patient has high levels of blood glucose. Subsequently, clinicians may judge that the patient is suffering from hyperglycaemia (high blood glucose) and deliver treatment in the form of an insulin infusion. This will reduce the patient’s blood glucose levels causing hypoglycaemia (low blood glucose), which can cause fatal neuroglycopenic brain injury (Association of Anaesthetists of Great Britain and Ireland, 2014).

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

A review was performed of the National Reporting and Learning System (NRLS) (a database of patient safety incidents reported by the patients, the public and healthcare professionals in England and Wales) (see appendix 1). A search was undertaken for historical incident data reported on and between 1 September 2016 until and including 31 August 2021, using the search term ‘arterial’ which returned 39,543 results. The search was filtered to focus on adults by removing those where the patient’s age was reported to be aged 17 years or less. This returned 32,132 results. A total of 447 relevant reports involving the wrong flush fluid were identified in this 5-year period. This supports the view that this remains a systemic problem and glucose is not the only incorrect flush fluid that might be used; reports also included the use of water for irrigation, mannitol (used to manage tissue swelling) and potassium chloride infusion (used to replace potassium in the blood).

Patel et al (2020) suggest that despite multiple recommendations and guidelines, cases of incorrect treatment of falsely elevated blood glucose levels from arterial line blood sampling continues to occur. Patel et al reports data obtained between 2005 and 2015 from the NRLS, which recorded 299 incidents of this type of error. The investigation completed a similar analysis of the data between 1 September 2020 and 31 August 2021 and concluded on average there was an increase in the number of incidents reported per year between 2005 and 2021. This may be due to an improved culture of reporting these incidents. However, of note there were more episodes of hypoglycaemia reported in the detailed analysis between 1 September 2020 and 31 August 2021 than reported between 2005 and 2015 (18 compared with 6 respectively) (see appendix 1).

A survey of UK adult critical care units in 2013 highlighted 30% reported errors involving the use of glucose solution attached to an arterial line (Gupta and Cook, 2013). This suggests a greater number of incidents may be occurring than formally reported.

Learning potential – what is the potential for an HSIB investigation to lead to positive changes and improvements to patient safety across the healthcare system?

Despite safety alerts (National Patient Safety Agency, 2008a), academic reviews (Patel et al, 2020; Gupta and Cook, 2013) and national guidance (Association of Anaesthetists of Great Britain and Ireland, 2014; Medicines and Healthcare products Regulatory Agency, 2014a), the risk of harm and fatal brain injury caused by low blood sugar levels, associated with the use of the wrong flush fluid with arterial lines, continues to exist.

The existing body of evidence concludes that engineered solutions and national changes need to be considered to address this safety risk. There have been many attempts for improvement which include training, procedures and modification to the task of blood sampling. These have been unable to avoid ongoing harm and in some cases contributed to patient death (Association of Anaesthetists of Great Britain and Ireland, 2014) (Gupta and Cook, 2013), (Patel, et al, 2020).

HSIB has explored the reference event to consider the potential for existing safety measures to manage the risk associated to the wrong flush fluid being used with arterial transducer line systems. HSIB has identified that a national investigation would be beneficial to understand and learn how, despite a focus on the improvement of this issue, patient harm continues to occur.

3.3.1 The investigation was completed between March 2021 and March 2022.

3.3.2 The investigation was delayed in starting as it took account of the ongoing demand and effect of the COVID-19 pandemic on NHS critical care resources.

3.3.3 The HSIB investigation team included the following expertise:

• healthcare leadership
• healthcare investigation
• human factors and systems engineering
• clinical subject matter advisors.

3.3.4 The investigation collated and reviewed evidence from Keith’s family, clinical staff from the reference event Trust, and staff supporting the storage and procurement of supplies for the critical care unit. Information sources included:

• patient medical records
• interviews with Keith’s Wife and three daughters
• interviews with staff who cared for Keith at the time of the incident
• interviews with operational staff supporting the critical care unit
• interviews with staff in governance roles and involved in the Trust’s internal investigation into the incident.

3.3.5 The adult critical care unit was observed to understand the layout and physical characteristics of the environment in which Keith was cared for. This provided the opportunity to observe the storage of arterial transducer line equipment and fluids, and Trust staff demonstrated how an arterial transducer line would be set up, checked and recorded in the medical records.

3.3.6 Stakeholders with national influence on the safe management of arterial transducer lines were identified. These stakeholders informed the investigation team on the current level of safety and potential terms of reference for a national investigation.

3.3.7 A period of engagement with relevant stakeholders included one-to-one conversations with representatives of pharmacy and clinical professional bodies along with commercial and national organisations. These conversations ensured representation of the professional colleges and regulatory bodies with the greatest influence and knowledge of arterial transducer line systems in the UK.

3.3.8 Three workshops were held to facilitate discussions on recognised safety issues and the implementation and reliability of safety controls. These considered where future controls should be focused. Appendix 2 summarises the organisations present at the workshops. The final workshop focused on where HSIB safety recommendations could be directed to increase future safety of arterial transducer line systems.

HSIB adopts a no-blame approach to all investigations. The healthcare system is considered in its entirety, including the equipment, physical space, tasks, human capabilities, organisational culture, to understand the factors most likely to have contributed to the outcome. The analysis of the data collated considers principles and evidence relevant to the science of design, engineering and psychology and relies heavily on the safety science of human factors. This section describes in more detail how the investigation was carried out.

3.4.1 Before completing interviews or a site visit, the investigation team familiarised themselves with the equipment and task of attaching an arterial line and the necessary flush fluid. Relevant HSIB clinical staff were interviewed to walk through the task and training videos were consulted. This informed the development of a hierarchical task analysis to assist the investigation (Kirwan and Ainsworth, 1993). The investigation undertook interviews and observations at the Trust to understand how each task was completed, the equipment required and factors which may contribute to how staff complete the tasks.

3.4.2 The investigation team visited and interviewed Keith’s family to understand their recollection of events and perceptions of the incident.

3.4.3 The combination of family and staff interviews informed the development of a timeline for the reference event. The timeline was verified and added to following a review of electronic documentation.

3.4.4 The investigation process was iterative; as further information was gained, additional interviews or data sources were identified.

3.4.5 Interview and observation data were either digitally recorded or notes taken, which were then thematically coded and analysed, based on the Systems Engineering Initiative for Patient Safety (Carayon et al, 2006). This informed the investigation of the factors interacting and contributing to the outcome from the wrong flush fluid being used.

3.4.6 Several approaches were applied to the analysis of the evidence to consider a top-down perspective of the wider system. Finally, analysis was completed to understand how the context of the adult intensive care unit influenced the reliability and accuracy of tasks required to set up and take a blood sample from the arterial transducer line. This analysis was based on the Systematic Human Error Reduction and Prediction Approach (SHERPA) (Embrey, 2014), completed with a clinical subject matter expert. The findings were visualised for sharing using a Bowtie analysis (Chartered Institute For Ergonomics and Human Factors, 2016). This described the threats to the safe use of arterial lines and the sufficiency of existing safety controls. This enabled a shared understanding across multiple stakeholders within the healthcare system to address the threats and develop safety recommendations.

Organisational context relative to COVID-19 pandemic

4.1.1 The Trust provides adult critical care at two sites. Nursing staff are rostered to work at either site and are required to call on the day of their shift to find out which site they will work on. The adult intensive care unit (AICU) is the larger unit compared to the critical care unit (CCU). The incident occurred within the AICU. The AICU has 16 beds, and the unit is divided into three sections. Nurse 1, who admitted Keith, explained she was more familiar with working at the CCU.

4.1.2 In November 2020, the second wave of the COVID-19 pandemic was just starting in England. On the day of Keith’s admission there were nine patients on the AICU and a further three admitted overnight. Acuity (a high number patients requiring a high level of care) was rated as high, and staff at the Trust reported that three night-time admissions were unusual.

4.1.3 The investigation heard that the COVID-19 pandemic had resulted in several months of high levels of workload for staff. A senior nurse explained there had not been enough of staff to accommodate the need for staff to isolate if they were in contact with or confirmed as having COVID-19. Subsequently, staff not affected by COVID-19 were encouraged to work additional duties. The investigation heard that generally staff were suffering a certain degree of burnout and fatigue, with breaks not always protected, and a greater reliance upon bank (temporarily contracted) staff was necessary around the time of the incident.

4.1.4 Before the pandemic, the Trust had found it a challenge to ensure nursing staff were consistently available with the correct skill sets to cover all shifts. COVID-19 had exacerbated this staffing issue further.

Staff wellbeing and fatigue

4.1.5 The investigation explored staff wellbeing and the Trust’s approach to the managing the risk of staff fatigue. Appendix 3 summarises the typical shift patterns and hours of work for some staff involved in the incident. This suggests examples of staff completing regular shifts and then being rostered on to complete additional bank shifts during scheduled days off. In the month of November 2020, this included a member of staff completing seven consecutive 12.5-hour night shifts. In some weeks there were examples of staff working consecutive weeks of 87.5, 70 or 62.5 hours. Cumulative fatigue and extended periods of being awake are both well recognised as influencing human performance including memory recall and attention to information (Dawson and McCulloch, 2005).

4.1.6 Nursing staff said that both night and day 12.5-hour shifts were often started earlier or finished later; therefore, typically these became 13.5-hour shifts. These additional hours are unpaid unless a specific request is made and felt necessary by staff to ensure the effective sharing of information between shifts. The working time directive states a worker should not exceed, including overtime, an average of 48 hours for each 7-day period. They also should not exceed 8 hours of nightwork within 24 hours. The duration of working hours is calculated as an average over 17 weeks (The Working Time Regulations, 1998).

4.1.7 The investigation spoke with the AICU wellbeing lead. They recognised the Trust’s approach to managing staff fatigue was reactive, which had become increasingly apparent during the COVID-19 pandemic. This finding was reiterated by a member of staff who suggested a reluctance to highlight fatigue to senior staff at the risk of being perceived badly. They suggested that knowledge on the impact of fatigue and the need to report on fatigue was not well understood. This reflects the wider NHS culture which is highlighted in a recent publication by the Royal College of Nursing, which quotes the Health and Safety Executive recommendation to avoid shifts longer than 8 hours and where 12-hour shifts are implemented these should be limited to three consecutive shifts (Royal College of Nursing, 2021). This report recognises the risk associated with staff fatigue to both patient safety and staff wellbeing. There is currently no regulatory requirement to prevent excessive working hours for nursing staff.

4.1.8 The rostering of shifts, in the context of the recent COVID-19 pandemic, was described as challenging by staff at the Trust. Senior staff are made aware of the availability and competence level of nursing staff through the IT rostering system. This enables senior staff to create a staff roster and to cover AICU duties that can fall outside of the working time directive.

4.1.9 The Trust and NHS Professionals (NHSP) (a service that contracts with staff who are often already employed within the organisation) agreed over the winter period to offer a financial initiative for staff; on completing five shifts with NHSP they would be paid for six. This may have had the unintended consequence of encouraging a norm where staff work a high number of hours. The IT system is limited in its ability to alert managers when staff are working excessive cumulative hours across shifts completed through the Trust and NHSP. Senior nursing staff suggested the only time they may look at an individual’s historical shift pattern was retrospectively, if staff showed signs of fatigue or reduced levels of performance. There were various factors behind staff members’ motivation to work additional shifts. Some staff suggested a concern for their peers and patients if a shift was inadequately staffed; others highlighted the high cost of living in the local area and a need to increase their earnings.

4.1.10 The investigation found there had been a normalisation and acceptance of staff working extended shifts and additional hours during the start of the COVID-19 pandemic. The concept of fatigue management is yet to become an established process in healthcare, as applied in other industries (International Civil Aviation Organisation, n.d.). Staff at the Trust reported that the current approach to managing NHS staff wellbeing was not prospective in managing the associated risk of fatigue.

Physical characteristics of the clinical environment

4.1.11 Many of the staff interviewed reported that the AICU bedspaces and the physical environment were noisy, with poor lighting and insufficient space to work. The investigation heard terms used to describe the bedspace during the time of the incident as “messy”, “cluttered” and “chaotic”. The Trust is aware that due to the unit’s age the physical design of the bedspaces does not comply with the existing recommendations (Department of Health, 2013). The investigation was shown plans for the new larger AICU currently under construction, intended to comply with these recommendations.

Noise

4.1.12 During the observation visit, the investigation noted high levels of background noise from an industrial-sized ventilation system. The ventilation system had recently been introduced in response to the risk of airborne infection recognised during the COVID-19 pandemic. This had the unintended consequence of making communication more challenging (made worse by the need for staff to wear face masks) and created a working environment that affected people’s level of comfort and stress (see figure 4).

This section considers factors that influenced how the team completed the work and activities required to manage Keith’s deteriorating condition. This section will consider how opportunities arose for the incorrect flush fluid to be used at different points of Keith’s care.

Management of Keith’s admission to the AICU

Staff perceptions

4.2.1 Staff described Keith’s admission to the AICU as an emergency as he was clinically very unstable. The description used by two clinical staff reflecting on Keith’s admission was “firefighting”, suggesting there were multiple simultaneous clinical challenges. The handover received on admission identified an urgency for staff to treat Keith’s very low blood pressure. The need to insert an arterial line was prioritised by a consultant to enable effective blood pressure monitoring.

4.2.2 During Keith’s admission there were at least five staff working around the bedspace, under time pressure to complete competing tasks to stabilise his condition. Nurse 2 recalls offering assistance and took on the role of supporting the doctor with the insertion of the arterial and central line, to enable nurse 1 to continue to set up the bedspace and administer medication as required.

4.2.3 Nurse 1, who was allocated to receive Keith on admission, had been away from work for 5 weeks on leave and usually worked at the CCU. They were conscious of constant changes to the work environment in the AICU and had regularly checked their emails while on leave to keep up to speed with the changes. This was their first shift back and they also had the responsibility of supervising a nurse undertaking specialised training. These factors were described by nurse 1 to the investigation as creating feelings of being “overwhelmed”. The lack of familiarity with staff and equipment storage systems (modified during the COVID-19 pandemic) were specifically identified as adding to their stress.

Setting up the arterial line

4.2.4 Nurse 2 was on duty as part of the NHS Professional bank service. Nurse 2 explained that setting up an arterial line was a daily task for them. They also described differences in the way equipment was stored compared with the work environments they were more familiar with; the CCU and operating theatres.

4.2.5 The storage of arterial line equipment in the CCU and operating theatre environment was suggested as being clearer and more consistent compared to the AICU. In the AICU it was described as regularly changing, with different types of stock being contained within the unit’s three clinical areas. Both nurses reflected that this had improved since the incident, with allocated and labelled drawers (see figure 5) containing the arterial line equipment, except the flush fluid.

4.2.6 The doctor completed the insertion of the cannula, to start effective monitoring of Keith’s deteriorating condition. Nurse 2 worked under time pressure to find a red arterial transducer line, which was not located where expected. The investigation heard that it was the responsibility of the care support workers (CSWs) (rostered on to a shift to support nursing duties) to ensure bedspaces were equipped with the necessary equipment. Nurse 2 explained to the investigation that the time pressure was associated with their concern about losing the patency (opening) of the cannula, which by now had been inserted by the doctor. They made the decision to connect the only available line, a blue venous transducer line, asking the CSW to continue to look for the correct equipment.

4.2.7 The investigation heard from doctors and nurses familiar with inserting arterial lines. The findings suggest there is variability in how the specific tasks of inserting, collecting and installing an arterial line is completed. Typically, there is an implicit understanding in AICU that the doctor will insert the cannula and the nurse will collect and connect the remaining equipment, including the checking and connection of the flush fluid. On some occasions or in different clinical areas the doctor may complete all the tasks.

4.2.8 The division and completion of different tasks by the different job roles placed a time pressure on the collection of equipment. This division of tasks spreads the responsibility for the reliability and safety of the whole task. The organisation of equipment and the organisation of the timing of the task across team members appeared to influence how the arterial line was set up. The process of connecting the transducer to the cannula and connecting a flush fluid had to be undertaken twice, as a consequence of the correct transducer set not being available at the time of the cannula being inserted.

Management of episode of hypoglycaemia (low blood sugar)

4.2.9 At 16:24 hours, as the arterial line was being inserted, a blood test identified that Keith’s blood glucose level was 3.4mmol/litre and he was diagnosed with hypoglycaemia. Nurse 1 spoke with the doctor and treatment with glucose was advised. The Trust policy for treatment of hypoglycaemia in adult inpatients depends upon the level of alertness of the patient. On every ward, including the AICU, there is a ‘hypo box’ which contains a 100ml vial (glass bottle) of 20% glucose and treatment would require the delivery of 75ml of the glucose as an infusion. Within critical care units, glucose is readily available to staff within the drug cupboard, hence staff explained to the investigation the ‘hypo box’ was never used. The location of the hypo box within the AICU reflected this, as it was stored at a far point of the unit obscured by equipment.

4.2.10 The nurse went to the drug cupboard to look for 20% strength glucose (see figure 6). They were only able to locate 500ml bags of glucose at 10% and 50% strength. The investigation heard from the critical care lead pharmacist at the Trust that the stock of infusion bags provided to the AICU only included 5%, 10% and 50% glucose strengths. An infusion of 20% glucose was only available in the AICU as a 100ml vial (glass bottle) (see figure 7). The investigation observed that such vials were stored on shelves opposite the bags of fluid, which nurse 1 would have had their back to when looking for glucose.

4.2.11 The investigation visited the CCU, the other unit in which nurse 1 was more familiar with working. Due to the nature of CCU patients’ healthcare needs, the investigation heard from the critical care lead pharmacist and later observed, that 500ml bags of 20% glucose were routinely stocked and stored in the CCU’s drug cupboard (see figure 8). Four staff working in both the AICU and CCU were asked by the investigation how they would treat hypoglycaemia for their patients. All staff (of different grades and job roles) described the use of a 500ml bag of 20% glucose as an infusion. None suggested they would access the ‘hypo box’ or use the 100ml vial of 20% glucose. The use of a 500ml bag of 20% glucose was also described in the Trust’s incident report as the accepted approach to managing hypoglycaemia on the AICU. A senior member of staff reflected that the current local hypoglycaemia policy does not reflect typical management of critical care patients, where intravenous access is easily achieved to deliver intravenous fluids.

4.2.12 The investigation found the accepted work practices within the critical care units were not reflected by the current local policy, evidenced by the AICU not stocking 500ml bags of 20% glucose. As nurse 1 did not find the glucose in the AICU drug cupboard in the form they expected, it appears they concluded it was out of stock, rather than looking for an alternative volume of 20% glucose. The nurse selected 10% and 50% strengths of glucose to bring to the bedspace to allow the doctor to select a suitable alternative. This suggests a mismatch between Trust staff tasked with developing safety policies and those providing the medication with those working every day in this clinical area.

Checking and recording of medication

4.2.13 At 19:30 hours the night shift nurse arrived and received a handover from nurse 1. This involved both nurses viewing Keith’s record on the IT system. At the start of a shift this verbal handover would usually be followed by a walk around the bedspace with a visual check of all equipment and medication attached to the patient. This is supported by a safety checklist, which at the time of the incident did not include a reminder to check the arterial transducer line fluid; this has since been added. These visual checks were not completed by the night nurse until 22:00 hours and were also delayed the next morning at 09:00 hours. The IT system has two interfaces to record medications: the ‘medication administration record’ and ‘flowsheet’. There were gaps and inconsistencies in how the arterial flush fluid was recorded and delays and incompleteness of the safety checks (see figure 3 in section 2).

4.2.14 Keith’s Wife arrived at about 19:00 hours. Keith’s condition was deteriorating, and she recalls there was a lot of activity by staff around the bedside. The night nurse commenced the visual checks but was interrupted on several occasions. The interruptions were necessary to respond to Keith’s deteriorating condition and the need for him to be intubated and for a nasogastric (NG) tube to be inserted; there were some challenges with putting the NG tube in the correct position. Further interruptions occurred when two new admissions were received onto the unit. The staffing levels on the night shift did not allow for a float nurse and the night nurse was required to support other nurses in receiving new patients. The night nurse described how multiple competing tasks and time pressures influenced the reliability of the checking and documentation completed. The intensity of the nurse’s workload was commented on to the investigation by Keith’s Wife and by one of the doctors arriving for the medical handover at 20:00 hours.

4.2.15 The delay of safety checks the following morning was also suggested to be a result of staff workload. The nurse on duty that morning had the additional responsibility of co-ordinating the staff and tasks within that section of the unit. This nurse did raise concerns that these additional responsibilities may not be compatible with providing the attention required by Keith due to the seriousness of his illness. They were informed that, because of a lack of additional senior nurses, there was no alternative.

4.2.16 Nursing and pharmacy staff described how the arterial flush fluid, although a relatively harmless fluid, is considered a medication and as such requires a prescription, which should be signed by clinical staff once they have completed safety checks: check of name and strength of medication on the bag, the expiry date, and the name of the patient for whom it is intended. A collection of medications regularly given to patients in the AICU have automatically pre-populated prescriptions; this includes insulin and saline for a flush fluid. This reflects how frequently these medications may be given by either a nurse or doctor in a critical care setting.

4.2.17 One member of staff told the investigation that flushes may not be consistently signed for by clinical staff. There was a consensus across staff that the arterial flush was considered a “low-risk” medication, as its function is to maintain the patency of the arterial transducer line, rather than being therapeutic for the patient. Staff also suggested to the investigation there was a belief that there is a low probability of the flush fluid being incorrect. This suggests the awareness and perceived risk associated with the administration of the flush fluid may differ between staff.

4.2.18 Checking is often added as a safety measure in healthcare, without strong evidence to reflect its effectiveness in avoiding errors (Koyama, et al, 2020). The challenge of multiple checks being used as a reaction to a safety concern or incident was suggested to the investigation: “… you find a problem, so you add it to the checklist and the next thing you know your checklist takes you an hour.” This was highlighted to explain how the duration of time required to complete long lists of checks may compromise the reliability and quality of checks. In this situation, the competition for the nurse’s attention disturbed the sequence of checks. The checks the following morning, although again delayed, were successful in identifying 10% glucose as the flush fluid.

Blood sampling and management of hyperglycaemia (high blood sugar)

4.2.19 At 17:25 hours a blood test was completed from the arterial line and returned a blood glucose reading of 19.9mmol/litre. This inferred Keith had moved from a state of hypoglycaemia to hyperglycaemia. It was documented on the ‘flowsheet’ that an insulin infusion was started at 02:53 hours; this is not recorded in the ‘medication administration record’. The initials entered in the ‘flowchart’, to indicate delivery of insulin, were not those of the night nurse and the investigation heard they were the initials of another AICU nurse, likely to be covering the night nurse’s break. The last dose of insulin is recorded at 11:12 hours on the second day.

4.2.20 The treatment of hyperglycaemia in the AICU is outlined in a local protocol. This indicates that if a patient’s blood glucose levels are above 8mmol/litre on two consecutive occasions then insulin should be started and blood glucose initially monitored hourly. Once blood glucose levels fall within the range of 4mmol/litre to 7mmol/litre then monitoring of blood glucose should be continued every 2 hours.

4.2.21 In Keith’s case, five entries of blood glucose at a level greater than 8mmol/litre were recorded between 17:25 hours and 02:00 hours before he was given insulin. The investigation heard that a nurse spoke with one of the doctors who agreed to starting the insulin. Once the insulin had been started Keith’s next blood glucose recording was at 05:17 hours and was below 8mmol/litre. The night nurse suggested to the investigation the intensity of the workload influenced the lack of blood samples taken as indicated in the protocol.

4.2.22 The first increase in blood glucose was from 3.5mmol/litre to 19.9mmol/litre between 16:24 hours and 17:25 hours. The investigation considered the significance in the size of the increase. Staff explained in hindsight this is “a big jump”. However, this could be explained in response to the recent administration of 50% glucose and the deterioration of Keith’s condition, as hyperglycaemia is a recognised response to sepsis in critical illness. Two further opinions from staff suggested that on reflection an episode of hyperglycaemia would be unusual in a patient with impaired liver function, which Keith had been diagnosed with.

4.2.23 The impact of the 10% glucose flush fluid in place was considered by the investigation. The investigation heard from staff that this can contaminate the blood sample taken from the arterial line to give an erroneously high level of glucose in blood test results. Reflecting on the significant change in Keith’s blood glucose levels, a nurse and doctor both suggested that a capillary blood test would have been the only way to confirm whether this reflected a true record of Keith’s blood glucose levels. Capillary blood tests are not usually undertaken for patients in the AICU because there is a high level of confidence placed on the reliability of arterial blood test results.

4.2.24 The combination of workload, the context of the AICU, where hyperglycaemia and the administration of insulin is common, and the result of the blood sample all influenced the decision to administer insulin to Keith. The risk of a contaminated blood sample was not suggested to the investigation as a likely consideration for staff, because of the familiarity of the pattern of events, namely a patient with severe sepsis becoming hyperglycaemic. This reflects a well-recognised approach and limitation to real-world decision making in the context of experienced people, where either information or time is limited (Kahneman and Klein, 2009). The absence of information to contradict the assumptions or ‘mental model’ held by an individual about a situation, may cause them to not consider alternative explanations for the information they are presented with.

4.2.25 An opportunity to identify the 10% glucose and its contribution to providing inaccurate blood glucose levels occurred when a blood sample was taken from another site on Keith’s body. This opportunity arose when the temporary renal dialysis line (central line for renal replacement therapy) was inserted by a doctor and, although it was not standard practice, the nurse offered to test the blood the doctor had removed during the procedure. The doctor told the investigation that they were mindful that they were at the end of their first night shift and aware of the level of fatigue they were experiencing. They explained to the investigation that moving to a nightshift can make it difficult to sleep the day before, so by 07:00 hours they estimated they may have been awake for almost 24 hours. Remaining awake beyond 17 hours has been demonstrated to reduce performance levels to those seen in a person at the legal blood alcohol limit for driving (Dawson, 1997). The doctor’s awareness of their fatigue motivated them to adopt a high level of vigilance to their tasks and they agreed to the blood tests to be sure the temporary renal dialysis line was positioned in a vein. The blood test result was read with the motivation to check the oxygen level and this reduced the attention given to the remainder of the results by either the doctor or the nurse.

4.2.26 The situation and reason for the blood test being taken, combined with how the blood results were presented was not sufficient to alert staff to the significance of blood glucose being outside the expected range. Staff fatigue may also have reduced the amount of attention given to all the available information.

Medicines and Healthcare products Regulatory Agency (MHRA)

5.1.1 The MHRA is the regulatory body responsible for the safe and effective use of both medication and medical devices in the UK. It operates the ‘Yellow Card’ reporting system which collects and monitors voluntarily reported information (by health professionals and/or patients) on suspected safety concerns with healthcare products.

5.1.2 The investigation asked the MHRA for details of any Yellow Card reports associated with the use of the wrong flush fluid used with an arterial line. The MHRA responded following a search of its database between 1 September 2020 until and including 31 August 2021. This did not identify any reports relating to the investigation.

5.1.3 The investigation identified several considerations regarding the reliability of using the National Reporting and Learning System (NRLS) (see 3.2.1) and MHRA Yellow Card reports to establish the frequency of incidents and associated level of harm. These are:

• There is variability in how information is entered into the systems and therefore how it can be searched and reported.
• It is mandatory to report serious incidents, but voluntary for other patient safety incidents (NHS England and NHS Improvement, 2021a).
• Yellow Card reporting by the public and healthcare professional is promoted but reporting is not compulsory.
• Medical device manufacturers are obliged to report certain types of incidents directly to the MHRA; this is influenced by the robustness of their post-market surveillance processes (the monitoring manufacturers are required to undertake to evaluate the safety and performance of their products).
• The number of reports submitted reflects the reporting culture, not how often incidents happen (NHS England and NHS Improvement, 2021a).
• Reporting rates were lower during the COVID-19 pandemic (NHS England and NHS Improvement, 2021a).

5.1.4 The absence of reports on medical devices or medicines through the Yellow Card reporting system has been highlighted as an issue in previous HSIB investigation reports (Healthcare Safety Investigation Branch, 2018) (Healthcare Safety Investigation Branch, 2019). Issues relating to staff awareness and effective communication with manufacturers have been highlighted in a recent study (Tase et al, 2022). This study indicates that evaluation of the risks related to medical devices does not consider the actual environment and context in which they are used, and recommends improvements in medical device reporting and the communication between end users and manufacturers.

Investigation survey

5.1.5 The investigation was aware that the reporting system may not reflect the likelihood of the risks associated with blood sampling from arterial lines. Furthermore, in the context of the COVID-19 pandemic the safety concerns may have changed. Two professional bodies, the British Association of Critical Care Nurses and College of Operating Department Practitioners, offered to issue a brief anonymous electronic survey to their membership in December 2021. The survey included questions on whether staff had experienced, or were aware of, the selection and use of the wrong flush fluid in an arterial line, the likelihood of this happening and the consequences of taking a blood sample from the arterial line when an incorrect flush solution had been used.

5.1.6 Replies were received from 138 members, 101 (73%) of whom had experienced or heard about incidents with the wrong flush fluid being used within an arterial line. Several respondents commented that this “occurs only infrequently”. It was recognised that the consequences of taking a blood sample from the arterial line when the wrong flush fluid was used may lead to false results and inappropriate treatment decisions. Some commented that the sampling method should minimise contamination from the flush fluid.

5.1.7 Where a solution containing glucose had been inadvertently used, respondents reported the actual impact on the patient ranged from none, as the wrong fluid was found before it was used, through to insulin treatment initiated for hyperglycaemia and the death of patients.

5.1.8 There was a wide range of responses about the likelihood of selecting and using a wrong flush solution in the arterial line. These included:

• “Quite likely, the bags are stored together, look similar and packaging is similar.”
• “Should be unlikely but it does happen.”
• “This is rare, but an ongoing potential hazard on a daily basis.”

5.1.9 Staff highlighted the importance of taking a systemic view of the risk, especially the impact of contextual factors such as the COVID-19 pandemic. Staffing numbers and the impact on the ratio of qualified nurses available on a unit were thought to influence the reliability of checks.

5.2.1 The investigation explored the known risks associated with the tasks required to set up an arterial line and take a blood sample. The investigation also sought to analyse the robustness of existing safety controls to manage these known risks.

5.2.2 A high-level task analysis was completed based on expert descriptions and observations of the setting up and taking of a blood sample from an arterial line (see figure 13). This high-level and simplified description of tasks was used to further analyse why the task may influence patient safety. The method used for this further analysis was based on the Systemic Human Error Reduction and Prediction Approach (SHERPA) (Embrey, 2014). This approach considers how the contextual factors identified from the reference event, literature (see 1.4.3) and the survey completed by two professional bodies may influence human performance. SHERPA facilitates consideration of how systemic factors influence the reliability of each task. An example of the detail obtained from this method can be found in appendix 4.

5.2.3 The investigation also heard from stakeholders about the challenges associated with the implementation of safety controls relevant to arterial lines. The following sections summarise the evidence on the risks already known.

Setting up the arterial line system

5.2.4 It has been well recognised that medication packaging can have an impact on the selection and subsequent use of the wrong medication (National Patient Safety Agency, 2008b). The National Patient Safety Agency (NPSA) report acknowledges that injectable medication, including medication delivered in infusion bags, is particularly susceptible to error. The NPSA report made several recommendations for the presentation of written information on bags of fluid. The investigation reviewed the current design of fluid labelling and concluded these do not reflect the NPSA recommendations (National Patient Safety Agency, 2008b), which include:

• placement of key information, for example the name of the medication, at the bottom of the bag close to the attachment of tubing
• varying elements of the design to enable differentiation between similar products
• judicious use of colour for high-risk infusions
• where colour can’t be used, varying the presentation of text and using graphic components, for example the medication name within a shape
• medication storage boxes should be clearly labelled on three sides and judicious use of colour for highlighting information.

5.2.5 A more recent review by the Department of Health and Social Care into reducing medication-related harm still highlights the need to work with industry and the MHRA to ‘develop solutions’ to reduce the risk associated with ‘look-alike sound-alike drugs’ (Department of Health and Social Care, 2018). Furthermore, the coroner’s report into the death of Susan Warby concluded that future deaths ‘could occur unless action is taken’ on the lack of distinctive labelling of fluids intended for use with arterial lines (Courts and Tribunals Judiciary, 2020). A response by the MHRA to the coroner’s report explains why colour cannot be used to support correct identification in fluid bags. However, the response does not address any other potential design solution and indicates risk mitigation must therefore ‘be employed locally within clinical areas’ (Medicines and Healthcare products Regulatory Agency, 2020b). Guidelines and a recent study both present alternative designs, which do not rely on colour, to increase the visibility of labelling with the intention of reducing selection errors (Lusk et al, 2022; National Patient Safety Agency, 2008b).

5.2.6 The storage of bags of flush fluids was identified as contributing to the outcome in the reference event (see 4.1.16 to 4.1.20). The risk that storage plays in the ease of selection of intravenous fluids from working environments and storerooms had been previously recognised by others (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013; National Patient Safety Agency, 2008a). The investigation heard that often little attention is given to the adequacy of storage for large volumes or to the inherent risks in infusion products in critical care units. ‘Health building notes’ provide best practice guidance on the design of healthcare spaces. These were considered for the storage of infusion products in critical care areas, but provide limited information on how healthcare storage environments can be designed to support safe selection of these products (NHS England and NHS Improvement, 2021b; Department of Health, 2013). Guidelines for the provision of intensive care services highlight that issues remain relating to adequate storage space, despite new critical care units being built (The Faculty of Intensive Care Medicine and Intensive Care Society, 2019).

5.2.7 The availability of pressure infusion bags that obscure the labelling on the flush fluid bag, as seen in the reference event, reduces the opportunity to identify an incorrect flush fluid. The use of non-transparent pressure infusion bags has been previously raised as a safety issue (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013; National Patient Safety Agency, 2008a). However, a recent small simulation study suggested that even with a transparent pressure infusion bag the label may still not draw the clinician’s attention and lead to the identification of an incorrect bag (Patel et al, 2020).

5.2.8 Staffing and caseload have been suggested as influencing adherence to local procedures that recommend checks (Gupta and Cook, 2013). The investigation found from the reference event and subsequent SHERPA analysis that the workload, staff availability and levels of fatigue within the team influenced the likelihood of safety checks being completed. These findings reflect the current concern within healthcare, where checking forms the main safety control relied upon to manage known risks (Koyama et al, 2020). The robustness of checking to ensure the safety of arterial flush fluids is a well-established problem (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013; National Patient Safety Agency, 2008a).

Prescribing and recording

5.2.9 In 2008, the NPSA highlighted the need for assurances of the prescription and checking of infusion fluids (National Patient Safety Agency, 2008a). The investigation of the reference event and evidence from pharmacy and clinical professions suggested that when patients are admitted to critical care units, prescriptions may be pre-populated in electronic prescribing systems. The investigation heard from stakeholders that Trust audit data suggests these prescriptions are not consistently signed when a flush fluid is administered.

5.2.10 To prescribe medication requires a qualified member of staff to document the type, strength, route of delivery and frequency of a medication. Currently the NHS has both paper and electronic systems and signing of a prescription on administration of a medication is a legal requirement. The SHERPA analysis highlighted factors influencing the reliability of the checks or signing of the prescription. These included a perception of the low risk of the flush fluid to the patient and a conflict in tasks due to time pressures associated with administering treatment to a critically ill patient.

5.2.11 Saline is used as an ‘off-label’ medication for use as a flush fluid. This means that it is used in a way that differs from the use for which it was authorised (Medicines Healthcare products Regulatory Agency, 2014b). This potentially increases the risk since it means that the MHRA has not examined the risks or benefits of using the medication in that way. Off-label prescribing remains acceptable if there is no suitable alternative and use is in accordance with the body of respected medical opinion. Manufacturers cannot advertise or recommend that medication is used in any way other than that specified in its licence. The MHRA monitors the safety of medication including off-label use via the Yellow Card reporting system.

5.2.12 The current off-label use of saline for an arterial flush fluid suggests that manufacturers do not have responsibility for the risk incurred for this use. This transfers the risk to the healthcare system; however, when this becomes an accepted practice healthcare providers may not know of the risk and subsequently may not mitigate for it.

Taking a blood sample

5.2.13 The design of transducer systems – that is, whether they are open or closed – influences the likelihood of blood sample contamination from the flush fluid (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013). A recent small-scale simulation study indicated that despite training and procedures to inform on the recommended sampling strategy contamination of a blood sample occurred where a flush fluid containing glucose was in place. This was the case for both open and closed transducer systems; however, the contamination was significantly reduced with the closed system (Patel et al, 2020). The conclusion from the existing evidence is that despite following the recommendations on how to take a sample, contamination of a blood sample will still occur, although there is less of a risk with a closed system (Patel et al, 2020; Gupta and Cook, 2013; National Patient Safety Agency, 2008a). These findings contradict the evidence heard by the coroner in issuing the prevention of future deaths notice for Susan Warby’s case. The evidence heard suggested that ‘a good technique used by staff would prevent false readings being obtained’ (Courts and Tribunals Judiciary, 2020).

Interpreting blood test results

5.2.14 The investigation heard that in the context of critically ill patients, blood glucose levels outside of the expected range and the use of insulin occur frequently and can become normalised by staff. The process for recording blood glucose levels and insulin may vary between paper and electronic systems. Both systems can make it challenging for staff to effectively recognise abnormal patterns, sufficient to direct attention to the flush fluid bag as a potential source of the altered blood glucose levels (Association of Anaesthetists of Great Britain and Ireland, 2014; Gupta and Cook, 2013).

5.2.15 The justification heard by the investigation for interpreting blood glucose levels from an arterial line were convenience and accepted practice. This does not allow for an independent blood sample (separate from the flush fluid) to routinely be used to compare blood glucose levels.

The investigation has established that taking a blood sample via an arterial line has persistent safety issues, which may not be reflected accurately in the formal reporting systems. Hence, it is impossible to capture a true picture of the prevalence and level of associated harm. The investigation found concerns within the healthcare system that recent events, namely the COVID-19 pandemic, have heightened this risk due to variability in staff resources as organisations respond to increasing demands.

The investigation sought to understand the effectiveness of, and challenges to, the implementation of previously recommended safety controls. Key stakeholders were engaged and remote interviews completed with each. This provided further evidence of the safety issues summarised in section 5.2. Three workshops were subsequently held to explore this evidence and consider suitable recommendations to address the persistent safety issues with these stakeholders.

5.3.1 The investigation identified 22 stakeholder organisations (see appendix 2). These included the professional bodies representing healthcare staff who use arterial lines, healthcare regulators and safety bodies and representatives from the commercial sectors of pharmaceutical and medical devices. The workshops were held between December 2021 and March 2022. The first workshop, held in December, was attended by 23 participants representing the pharmaceutical community, and the second, held in January, was attended by 22 representatives of the clinical community. The third workshop, held in March, included representatives from the first two workshops and stakeholders from national and regulatory organisations.

5.3.2 The aim of the workshops was to gather further insight into current guidance, procedures and existing safety controls and challenges for the implementation of stronger safety controls.

5.3.3 The first two workshops facilitated discussions on the safety issues relevant to the key tasks of setting up and taking a blood sample from an arterial line system (see figure 13 in section 5.2). Each discussion considered the potential risks associated with each of the tasks. Information was recorded on how these risks are currently managed or where there are opportunities to identify and rectify the wrong flush fluid.

5.3.4 The discussions progressed to identifying potential stronger controls available for managing the risks. This was done with reference to the hierarchy of controls model (The National Institute for Occupational Safety and Health, n.d.) (see figure 14).

5.3.5 When developing safety controls it can be beneficial to aim at the highest level possible of the hierarchy of controls model, as mitigations decrease in effectiveness the further down they are in the hierarchy. Therefore, the workshop participants were encouraged to start with a focus on how these risks could be eliminated, substituted or designed out, rather than solely relying on administrative controls, for example checks and training.

5.3.6 The evidence from the first two workshops and the investigation of the reference event were integrated and a visualisation developed to enable a shared understanding for the final workshop. This approach facilitated a multidisciplinary engagement with those who are able to influence the implementation and the co-design of recommendations.

Stakeholder workshop findings

5.3.7 The visualisation produced was based on a Bowtie approach (see figure 15). Bowtie analysis allows the visualisation of threats to safety (McLeod and Bowie, 2018; Chartered Institute for Ergonomics and Human Factors, 2016). Bowtie analysis considers the cultural and technical (sociotechnical) 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 weakened but not eliminated defences, it allows an understanding of how safety controls can fail. An advantage of bowtie analysis is that it provides a means to communicate by representing risk diagrammatically to a diverse audience. The intention is that organisations can prioritise activity correctly (De Ruijter and Guldenmund, 2016).

5.3.8 Previous HSIB reports provide a detailed description of bowtie analysis (Healthcare Safety Investigation Branch, 2021a, Healthcare Safety Investigation Branch, 2021b).

5.3.9 This approach will be used to describe the findings from the workshops. Figure 16 illustrates the two main events in the centre of the image, which both have the potential to cause patient harm: glucose within the flush fluid and the subsequent inaccurate diagnosis of hyperglycaemia. These may lead to the outcome (right-hand side of figure 16) of the unnecessary delivery of insulin. Both events were systematically considered during the workshop, to understand the effectiveness of safety controls intended to prevent these events and those safety controls intended to identify and enable recovery to avoid harm if the events occurred.

5.3.10 The investigation’s evidence highlighted four clear threats (left-hand side of figure 16) which may lead to these events. The remainder of this section of the report will present the investigation findings for these four threats:

1. incorrect collection and set-up of the flush fluid
2. no or incorrect prescription of flush fluid
3. contamination of an arterial line blood sample
4. misinterpretation of the blood glucose results.

Table 2 summarises the findings relative to the strength of the main safety controls relied upon to manage all four threats.

5.3.11 The following sections will consider significant findings specific to each of these threats. A bowtie image for each threat uses colour to indicate the effectiveness of current safety controls as reflected by the hierarchy of controls model (the darker the tone, the higher in the hierarchy – see figure 14). The workshops also considered how to strengthen existing controls by considering the following questions:

Incorrect collection and set-up of the flush fluid

5.3.12 The evidence suggested several systemic factors may influence the reliability of the tasks required by staff to select and set up a flush fluid. The accuracy or variability in the way these tasks were completed was influenced by the storage, design of labels and procurement of pressure infusion bags. Figure 17 illustrates the existing safety controls relied upon to manage these risks.

Design of flush fluid label and pressure infusion bags

5.3.13 The practice of storing bags of fluids within their original boxes and the design of the manufacturer’s label on the fluid bags do not help staff to easily differentiate a bag of fluid with and without glucose. Current guidance encourages trusts to consider storage as an approach to enhance safety in the selection of similar looking and sounding medicines (Royal Pharmaceutical Society, 2018). The investigation heard from stakeholders that in reality, enhancing safety through storage is not easy to achieve:

“Storage is a labelling, packaging, space and staffing issue.”

5.3.14 The stakeholder workshops reiterated the investigation’s findings from the reference event and historical reports (see 5.2.4 to 5.2.7). There was a consensus by stakeholders that labelling presents a risk and is a key contributory factor to selection errors.

5.3.15 The MHRA provides best practice guidance on labelling and packaging (Medicines and Healthcare products Regulatory Agency, 2020a) but does not make specific reference to infusion bags and the recognised challenges associated with them (Medicines and healthcare products Regulatory Agency, 2014a; National Patient Safety Agency, 2008b).

5.3.16 Stakeholders also considered the use of non-transparent pressure infusion bags as an obvious risk in the system. There was a strong consensus that where work might be done to improve the design of the flush fluid label the use of a pressure infusion bag that obscured the label did not make sense. Two stakeholders described how they had been involved in removing non-transparent pressure infusion bags at their trust or at a regional level.

5.3.17 The investigation heard from representatives of NHS Supply Chain, the organisation that sources and supplies healthcare products for NHS care providers. They reported that the current technical specification for procurement of pressure infusion bags by trusts requests the ‘ability to see substance contents and fluid levels in fluid bags and containers’. Stakeholders described how even where best practice had been adopted in the use of transparent pressure bags, they had recently experienced problems with consistently obtaining clear pressure bags. Variability in procurement practices across the NHS were considered as highly influential to the procurement of devices and clearly did not reflect NHS Supply Chain’s specification.

5.3.18 The workshop stakeholders also concluded that reliance upon checks to manage the risk of a selection error was inadequate. A stronger design solution was considered necessary. There was an agreement that the context of very unwell patients, staff workload, workforce and time pressure all weakened the reliance on checks. The use of pressure infusion bags that obscured flush fluid labels and the lack of distinctive features to distinguish between a clear fluid with or without glucose all contributed to the likelihood that the wrong fluid may be selected. Previous safety recommendations and the evidence from the reference event and the coroner’s judgement in the Susan Warby case have been considered. HSIB makes the following safety recommendations with the intention that they will provide design guidance for manufacturers to manage the risk associated with fluid selection, and ensure fluid labels can be consistently read from all directions at all times when the pressure infusion bag is inflated.

Storage of flush fluids

5.3.19 Stakeholders reiterated well-recognised issues relating to the availability of sufficient and high-quality space for storage within NHS facilities. Stakeholders considered that a low priority is placed on storage, the responsibility for storage and the impact on safety.

5.3.20 The investigation sought an international perspective of this problem and contacted the Institute for Safe Medication Practices and International Medication Safety Network. This was not a problem they currently recognised or were focused on. The investigation also spoke with an anaesthetist based in the United States. They indicated that in critical care environments, in their experience, glucose fluid was stored remotely and a technical closed-loop medicines administration system and barcode scanning were used to ensure the correct fluid was retrieved. Similar practices in storing glucose-based fluids away from saline were suggested by stakeholders familiar with UK operating theatre environments. Stakeholders recognised this may delay retrieval time.

5.3.21 Stakeholders also shared how some trusts chose to store all of the separate components required for an arterial line as a complete set. This implies removing the flush fluid from the locked drug cupboard and storing it in an alternative way, which may vary from policies that require all medications to be stored in a designated locked facility.

Prescription of flush fluid

5.3.22 Stakeholders acknowledged that although a prescription for medication is a legal requirement, it appears to present as a weak safety control in the context of the use of saline as a flush fluid in an arterial line system. The evidence indicated that the context of work and normalised practices, pre-population of the prescription, and the lack of reliability of subsequent checks were all factors in reducing the effectiveness of the prescription as a safety control (see figure 18).

5.3.23 The perception of saline as a low-risk product and the frequency with which it is used are recognised as reasons why prescribing fails to function as an effective control to prevent the incorrect flush being used. Stakeholders told the investigation that “nurses selected sodium chloride (saline) for multiple different things, multiple times a day and it is perceived as ‘non-risky’ which will affect how much attention is given”.

5.3.24 Pharmacists play a critical role in checking the accuracy and completeness of prescriptions. In the critical care environment, the recommendations suggest a minimum of 5 days of pharmaceutical support a week. However, NHS England and NHS Improvement has advised units to increase this to 7 days a week by 2020 (NHS England, 2016), although it has been suggested that it may not be feasible to implement this for all units (The Faculty of Intensive Care Medicine and Intensive Care Society, 2019). The double checks required and signed on the prescription by nursing and medical staff were recognised as inconsistently completed. Stakeholders echoed the academic literature recognising workload pressures, perception of the risk and reliability of a double check as reasons for discrepancies (Koyama et al, 2020). Variability in the need for a prescription in different clinical contexts where arterial lines are used was also cited. Medication practices in operating theatres differed from those used in critical care units and with recent pressures caused by the COVID-19 pandemic more operating theatre staff have been required to support critical care units. Stakeholders suggested this may have increased the variability in practices.

5.3.25 Stakeholders also expressed concern about the unintended consequence created by the use of pre-populated prescriptions of ‘bundles’ of medications typically required for critical care patients. These bundles include flush fluid and insulin. It was suggested to the investigation that their use may lower the attention staff give to these medications and reduce the level of conscious decision made to administering these medications. The unintended consequence of using pre-populated prescriptions, which are intended to aid efficiency and reduce workload, is similar to a longstanding problem recognised in automation within technical systems (Bainbridge, 1983). Stakeholders suggested that technology to support closed-loop medicines administration and checking (where technology confirms the correct medication) was not widely in use.

5.3.26 In summary, although a prescription is still necessary, the variability in practice across clinical environments and the lack of a reliable closed-loop medicines administration system reduce the effectiveness of this safety control. NHS England and NHS Improvement stakeholders stated that the use of barcode medication checking systems was currently under review to consider their use within the NHS.

Contamination of arterial blood samples

5.3.27 The transducer device available to trusts for arterial lines require staff to mitigate for the transducer’s inability to prevent contamination from the flush fluid. This is only an issue if a flush fluid contains anything other than saline. Checks and training in the procedure of taking blood are all relied upon to enhance safety and manage the risk associated to the transducer’s current design (see figure 19). There are two types of transducer available to healthcare: open and closed systems (see figure 2 in section 1). It is recognised that some closed systems reduce but do not eliminate the risk of contamination of a blood sample by flush fluid. Trusts train staff on a technique to reduce the likelihood of contamination from the flush fluid. This relies on withdrawing a volume of blood three to five times the ‘dead space’ between the port and the patient (see 1.3.5). This dead space is not a fixed volume as transducer systems vary and there are no markings on the tubes to indicate the volume required. The MHRA acknowledged this risk in 2014 (Medicines and Healthcare products Regulatory Agency, 2014a), but there does not appear to have been any further modification to labelling, device instructions or design to address the risk.

Interpretation of blood glucose results from arterial line samples

5.3.30 The investigation heard that the normalisation of patients in critical care having high levels of blood glucose may reduce clinicians’ attention to blood glucose levels outside of the expected range. This normalisation reduces the likelihood that further checks would routinely direct staff to suspect the use of an incorrect flush fluid. The risk of the incorrect interpretation of the cause of high blood glucose levels is compounded by the way trends in blood glucose levels are presented on patient record systems. The existing safety controls relied upon to manage the risk of misinterpretation are illustrated in figure 20.

5.3.31 Stakeholders who attended the investigation’s workshops discussed the elimination of the risk of misinterpretation through alternative approaches to monitoring blood glucose levels. Systems currently available to continuously measure blood glucose were considered positively. However, stakeholders suggested that the lack of reporting of arterial line system incidents meant it was not possible to justify the scale and costs associated with the implementation of such systems. The investigation concluded that this should be regularly reviewed and considered as a safety control in the future.

5.3.32 Stakeholders recognised that, until changes to the design of transducers eliminated the risk of contamination, alternative safety controls were required to increase the likelihood of identifying an inaccurate blood glucose level. Such controls might involve the taking of a second blood sample from an alternative site; a capillary blood glucose test (finger prick test) is most frequently used for this purpose. Completing a second test at the time of every arterial blood sample was considered impractical in the context of critical care. This is due to the physiological condition of the patient’s circulation, and the time pressure and likelihood that it may not be reliably completed. The introduction of ‘trigger point’ capillary testing was suggested as a pragmatic approach, for example testing prior to administering insulin, on putting up or changing a flush fluid and even at the start of a shift. This may not prevent the contamination from the flush fluid but would increase the likelihood of its identification and increase safety controls associated with the delivery of insulin treatment.

5.3.33 Paper and electronic systems currently record glucose levels using numerical values rather than a trend line. Figure 21 shows the different ways of presenting the data. It is recognised that use of a trend line supports early identification of change, which can be enhanced with lines to indicate where a change may fall outside of a predefined limit (Burns and Hajdukiewicz, 2004).

5.3.34 Stakeholders considered existing guidance or practice may not reliably increase clinician’s vigilance to the identification and questioning of extreme changes in blood glucose levels, such as an erratic trend or a single outlier. This was seen in the investigation of the reference event. The presentation of information or the alerts generated by IT systems were not enough to increase clinicians’ attention.

Management and detection of the risk

5.3.35 The investigation has considered how to avoid using the wrong flush fluid or contamination of blood samples from the flush fluid. The bowtie image (to the right of the events) also presents the potential for detection once the incorrect flush fluid has been used (see figure 22). The safety controls for detection mirror those for prevention and therefore will equally vary in their reliability.

5.3.36 The investigation has heard that currently there is no device or process that can eliminate the risk of arterial line blood contamination if the wrong flush fluid is used. The investigation has highlighted a number of systemic factors likely to influence the reliability of the use of the correct flush fluid. Until these systemic factors can be addressed there is a need for healthcare systems to raise awareness of this risk and understand how procurement decisions, storage facilities and staff practices can be optimised to strengthen safety around this issue.

5.3.37 The existing safety controls rely heavily upon repetitive checks and staff attention to avoid the traps that remain in selecting similar looking products and sampling blood to avoid contamination from the flush fluid. Stakeholders were emphatic that staff workload, work pressures and conflicts in staff tasks heavily contributed to the reliability and the effectiveness of the existing safety controls. These opinions were supported by the reference event findings which appeared to impact the. effectiveness and ability for staff to complete all necessary checks.

5.3.38 A number of practical approaches to managing the residual risk were considered during this investigation and believed valuable by stakeholders. Until post-market surveillance can inform manufacturers and the design of arterial line systems, there will be a need to develop further guidance to support clinical staff to follow practices to manage this systemic risk. Trusts will also need to consider the findings of this investigation in their procurement decision making and guidance on/priority given to storage for arterial line products.

5.3.39 The investigation considers that the existing controls are low in the hierarchy of controls and that there is a continual need to mitigate for these risks. The investigation acknowledges that the risks will remain until medication and device design issues can be addressed by regulators and manufacturers. The investigation also heard in the future use of arterial lines may extend to enhanced care settings (a ward setting where an upskilled workforce with ready access to the critical care team). Professional bodies and regulators should consider the evidence from this report to develop national guidance to implement and evaluate the following controls:

• review of the development and use of ‘arterial line kits’ developed by trusts to meet local protocols and separate storage from glucose products
• storage to use principles of design to reduce known errors for example, layout, labelling, separation of common look-alike sound-alike medications
• procurement and use of only transparent pressure infusion bags for use with arterial lines
• procurement of closed system transducer sets
• detailed description of best practice in the sampling technique for an arterial line
• trigger point testing via a second route, for example before insulin is started – this may include during high or erratic blood glucose recording, changes of shifts, change of fluid bag (allowing for some time to elapse before testing)
• closed-loop medication administration and checking, for example barcoding medication administration systems
• acknowledge and review risks associated with pre-populated prescribing associated with prescribing in critical care environments.

5.3.40 The investigation was told that implementing and evaluating these measures would require support from a range of national stakeholders but would need to be co-ordinated by a central body to ensure they were effective. HSIB therefore makes the following safety recommendation with the intention that the Association of Anaesthetists will design national guidance with representation from all relevant healthcare professionals within the following clinical areas: critical care, operating theatres and emergency departments.

5.3.41 The intention of the following safety recommendation is to provide assurance that NHS providers have implemented the future national guidance.

• The physical layout and design of the clinical and storage areas will influence how reliably staff are able to select and collect similar-looking equipment and medication.
• The labelling of bags of fluids, similar looking medications and manufacturers’ packaging reduce the reliability of selecting the correct flush fluid in the context of a critical care unit with time pressures and high workloads.
• The procurement and design of arterial transducer line equipment, the pressure infusion bags and transducer, do not assist in the identification of the incorrect flush fluid or prevent contamination from the flush fluid of a blood sample taken from the arterial line. Alternative equipment, for example transparent pressure infusion bags and closed arterial transducer lines, are currently available to the NHS. These may reduce the risk but are not routinely in use.
• Challenges in the provision of a consistent suitable workforce and high workloads have a detrimental effect on the safety controls currently relied upon to avoid or identify the risk of using the wrong flush fluid. Safety checks and training lack resilience to organisational pressures regularly experienced within critical care units.
• There can be a delay in identifying the contamination with glucose of an arterial line blood sample due to a normalisation and acceptance that critically ill patients may have altered blood glucose levels and require insulin treatment, and a perceived low risk associated with the use of a flush fluid.
• The design of systems to record and monitor information relevant to the arterial transducer line system and blood glucose levels do not easily alert staff to the potential use of the wrong flush fluid.
• Recommendations issued over the last 14 years by national safety bodies and professional healthcare organisations to address the safety of blood sampling associated with arterial lines have not been effectively implemented.

The intention of this safety recommendation is to provide design guidance for manufacturers to manage the risk associated with fluid selection. All aspects of label design should be considered this recommendation is broader than the judicious use of colour as the approach to increasing label safety.

The intention of this safety recommendation is to ensure fluid labels can be consistently read from all directions at all times when the pressure infusion bag is inflated.

The intention of this safety recommendation is to increase awareness of and action on known risks related to the design of the medical devices

The intention of this safety recommendation is to assure appropriate action is taken to manage the known risks related to the design of the medical devices.

The intention of this safety recommendation is for the Association of Anaesthetists to revise existing national guidance in collaboration with all relevant healthcare professionals including the following clinical areas: critical care, theatres and emergency departments

The intention of this safety recommendation is to provide assurance that NHS providers have implemented the future national guidance.