Investigation report: Weight-based medication errors in children

Felicity was 4 years old at the time of the incident. She was admitted to a cardiology (heart) ward in mid-February. She was diagnosed with fluid on her lungs. This is a recognised and common complication of a complex heart procedure which Felicity had had 3 months previously.

During her stay in hospital, Felicity underwent a further complex heart procedure and was under the care of the paediatric cardio-respiratory team – a specialist team which looks after children with heart and lung problems. In March, she was diagnosed with a blood clot in her right leg, known as a deep vein thrombosis (DVT).

Doctors identified that there was a need to balance the risk between providing blood thinning medicine for Felicity’s heart condition and to treat the DVT, without increasing the possibility of a bleed on the brain. A multidisciplinary team agreed that Felicity should be prescribed 100 units/kg of dalteparin twice daily. This recommendation was in line with the British National Formulary for Children, which is the guide clinicians refer to when prescribing, dispensing and administering medication for children.

Felicity’s weight at the time of the prescription was 15.2kg, so the dose for administration was calculated manually by the prescriber as being 15.2 x 100 = 1,520 units twice daily, rounded down to 1,500 units twice daily. The dalteparin was then inadvertently prescribed at a dose of 15,000 units twice daily, using the Trust’s electronic prescribing and medicines administration system (ePMA), due to the prescriber not making the necessary manual alteration to the dose within the system. The ePMA system itself and the subsequent processes for approval of the prescription and the dispensing, checking and administration of the medicine did not identify the incorrect prescription. This meant Felicity received 15,000 units of dalteparin (10 times the dose intended) on 5 occasions over one weekend. A subsequent CT scan showed that Felicity had a new right-sided bleed in the brain.

The acute NHS trust where the reference event took place (referred to as ‘the Trust’) notified HSIB of the incident. The Trust also escalated the incident, reporting it on the Strategic Executive Information System (a database of serious patient safety incidents), and conducted a serious incident investigation. HSIB gathered additional information and assessed the incident against its investigation criteria. HSIB decided to progress to a national investigation.

The national investigation focused on:

• An exploration of the factors that both support and inhibit multidisciplinary co-ordination and decision-making, including communication of critical decisions in relation to weight-based medications in support of a child’s treatment.
• Understanding the factors which contribute to a reduction in the effectiveness of checking as a barrier to medication errors, including those that result in workarounds.
• The implementation of ‘off-the-shelf’ electronic prescribing systems in specific contexts, particularly paediatric prescribing.

Multidisciplinary co-ordination and decision-making

• Email discussions may be being used for discussion of critical decisions in relation to patient care, with limited dissemination to the wider team.
• There is limited standardisation of handovers, ward rounds (visits to each patient in a ward to review and discuss their care) and huddles (short, focused staff briefings), in terms of which members of the multidisciplinary team are involved, and how they are conducted for maximum effectiveness.
• There is limited national guidance on the management of ward rounds in paediatrics (as is available for adult care).
• There is variability in hospital clinical pharmacy provision. This reduces the availability of ward-based pharmacists and may result in a dispensary-based service only at weekends.
• Nursing staff perceived themselves to be the final barrier to prevent an incorrect dose prior to the administration of medication, resulting in them feeling accountable for the error.

Factors undermining the effectiveness of checking as a barrier

• Multiple cues influenced whether staff considered medication doses to be correct.
• Processes for the checking of medicines varied without evidence of what constituted the most effective process.
• The distinction between verification and checking was not explicit, that is, checking that the prepared medication correlated with the prescription versus verification that the prescription was appropriate against a standard.
• The environments within which staff prepared and checked medicines influenced their performance.
• Environmental layouts and limited resource resulted in workarounds.

Implementation of ‘off-the-shelf’ electronic prescribing systems in specific contexts

• There are no standards for what safety-critical functionality should be available in ePMA systems configured for use in paediatrics (for example, the use of weight-based dose bands, where individually calculated doses are rounded to a set of predefined doses).
• The use of free-text comment boxes in ePMA systems is not specified or standardised.
• The usability and functionality of ePMA systems need to be assessed through user-testing across a range of different settings.
• Local configuration of ePMA systems potentially introduces variability and risks if not undertaken with clear understanding of the potential hazards and their mitigations.
• Software could qualify as a medical device if it meets the definition, set out in the Medical Device Regulations 2002 (as amended).
• Local governance of ePMA systems is limited with evidence of gaps in training and an absence of safety cases (reports that provide a transparent, evidence-based argument for why a system is safe for use in a particular setting).

Safety recommendations

HSIB’s safety recommendations are directed to a specific organisation for action. They are based on information derived from the investigation or other sources, such as safety studies, and are made with the intention of preventing future, similar events.

The HSIB investigation focused on errors in the prescription of weight-based medication for children, in the context of electronic prescribing and medicines administration (ePMA) systems. The responsibility for ensuring the safety of medicines prescribing and administration is with individual trusts, with input from NHSX, NHS Digital and the Medicines and Healthcare products Regulatory Agency (MHRA). Therefore, safety recommendations made in this respect are directed towards those organisations. Different parts of the healthcare system have also been identified to address the other risks highlighted in this investigation relating to paediatric ward rounds and second-checking processes. Safety recommendations have been directed to them accordingly.

1.1.1 The prescribing and administration of medicines are among the most common healthcare interventions. However, it is estimated that 237 million medication errors occur at some point in the medication process in England each year; 66 million are potentially clinically significant (Elliot et al, 2020). In paediatrics (children’s healthcare), 13% of prescriptions written for children contain errors (Ghaleb et al, 2010).

1.1.2 Medication errors can occur at the procurement (ordering) stage in addition to the prescribing, dispensing, administration and monitoring stages of the medication process. Specific factors complicate prescribing and administration of medicines for children and these factors can contribute to errors. Some of these factors are shown in figure 1 (Conn et al, 2019).

(adapted from Conn et al, 2019) (In this figure, the term ‘off-licence prescribing’ refers to the use of medicines outside of the indications for which they are licensed by national regulatory bodies).

1.1.3 The World Health Organization has previously highlighted the challenge of medication errors and in 2017 launched its third Global Patient Safety Challenge, ‘Medication Without Harm’ (World Health Organization, 2017). This aimed to reduce the global burden of severe and avoidable medication-related harm by 50% over 4 years. In response, the Medicines Safety Improvement Programme (NHS England and NHS Improvement, 2019) was established. The aim was to address the most common causes of severe harm related to medication by influencing safety culture, safety systems and high-risk medicines in common use. It is also supporting the development and implementation of ePMA systems as part of this programme of work.

1.2.1 Because children vary in terms of body size, weight, and their ability to process medications as their bodies develop, prescriptions for children must be individualised. Calculations are typically based on a child’s weight, age, gestation (the time between conception and birth) or body surface area. A child’s organ function (how efficiently their kidneys/liver are working) and underlying or new illnesses may also affect the dose calculation, which adds to the complexity.

1.2.2 A scoping review by Conn et al (2019) found that junior doctors frequently miscalculate doses, with tenfold errors (prescribing ten times the amount intended) mostly arising from decimal point inaccuracies. Only around half of doctors double-checked calculations.

1.2.3 In organisations where ePMA systems are used, calculations are often still undertaken manually. This is because not all systems have an automated calculation function, and some systems have an automated calculation but staff do not use it (Shah and Chui, 2019).

1.3.1 An ePMA system is an electronic medication management system which replaces paper medication charts. It should support the end-to-end management of medication, from prescribing and dispensing, through to administration (this is known as ‘closed loop medicines management’). An ePMA system also supports medicines reconciliation and is linked to electronic patient records for clinical decision support functionality, for example, alerting.

1.3.2 As outlined in the report ‘Electronic prescribing and medicines administration systems and safe discharge’ (Healthcare Safety Investigation Branch, 2019), ePMA systems are able to reduce certain prescription errors. However, they have also created new types of errors (NHS Connecting for Health, 2009). The reduction in medication errors depends on optimising commercial ePMA systems so that the available functionality is switched on, appropriately used, integrated with other relevant IT systems, and aligned with clinical workflows (National Institute for Healthcare Research, 2018).

1.3.3 A study by Puaar and Franklin (2018) found that there is limited knowledge and data relating to unintended consequences of introducing ePMA systems because of the varied nature of health IT products and the lack of common criteria against which to measure the impact. The study described three components of an ePMA system that may contribute to errors. These were ePMA system functionality and design; organisational decisions around ePMA system implementation and use; and different approaches to prescribing in the context of the electronic system.

1.3.4 In relation to terminology used in the report, the UKCA (UK Conformity Assessed) marking is a new UK product marking that is used for goods being placed on the market in Great Britain (Department for Business, Energy and Industrial Strategy, 2020). It covers most goods which previously required the CE marking. It came into effect on 1 January 2021.

1.4.1 Healthcare staff use various processes to check that they are giving the right medication to the right patient. In this report, the generic term used to describe these processes is ‘second checking’. However, as outlined later in the report, there are different terms and definitions used in relation to the checking processes prior to the administration of medication.

1.4.2 The second checking of medication prior to administration has been embedded in nursing practice for decades as an intervention to safeguard against errors and associated harm (Westbrook et al, 2020). In children’s health services, second checking is widely applied given the complexity of medication management, and the vulnerable nature of patients.

1.4.3 The second-checking process involves two registered healthcare professionals checking medication prior to administration. To prevent errors in administration, nurses are encouraged to check that they have the correct patient, medication, route (for example, by mouth or injection), time, and dose. This involves a step-by-step process which should be defined at an organisational level, according to the Royal College of Nursing (2020). Because the process is defined by individual organisations, practice varies and there is no standardised approach.

1.4.4 In relation to terminology used in the report:

• ‘Double-checking’ is the more generic term used to describe some form of second check undertaken by a second individual.
• ‘Single-person double-checking’ is the term used to describe one person undertaking both checks.
• ‘Independent double-checking’ involves both the requesting and checking nurse separately performing a check without sharing information.
• ‘Primed double-checking’ involves two people either working together or influencing the checking process by suggesting what the checker should find.

The investigation was completed between August 2020 and October 2021.

3.1.1 Investigative approach

HSIB adopts a no-blame approach to all investigations. It considers the healthcare system in its entirety to identify the factors that have contributed to the patient safety incident.

This investigation used the Systems Engineering Initiative for Patient Safety (SEIPS) and the Functional Resonance Analysis Method (FRAM), (see 3.1.7).

3.1.2 Investigation team

The HSIB investigation team was multidisciplinary:

• In relation to the skills and knowledge that the investigation team applied, it had expertise associated with clinical domains and safety science.
• The team was supported by a subject matter advisor (SMA), who is a Fellow of the Chartered Institute of Ergonomics and Human Factors. Excerpts from the SMA’s human factors report are included as figures throughout the report, to support the analysis carried out by the investigation.

3.1.3 Reference event investigation

The reference event investigation required visits to the Acute NHS Trust where the patient’s care was provided.

Engagement (reference event)

Felicity’s family was contacted and interviewed to establish their perspective on the reference event. The staff directly involved in the reference event were interviewed and included:

• eight members of the nursing team from the cardiology ward
• two consultant paediatric cardiologists
• the specialist registrar (paediatric cardiology)
• the specialist trainee 2 doctor (paediatrics)
• two specialist pharmacists.

3.1.4 National investigation

HSIB launched an investigation to explore the implementation of safety improvements for reducing the number of weight-based medication errors in children. The goal of the healthcare system is to ensure that the appropriate dose of a medicine is prescribed and administered for a patient’s weight or other characteristics (see 1.2.1). The safety risk is that this does not occur and that the system does not support healthcare providers to prescribe, dispense and administer medicines safely. The safety risk is focused on paediatric care, but may be applicable to other groups of patients where medicine dose calculations are required.

Findings from the scoping investigation led to a decision to broaden the investigation and identify healthcare settings to collect further data to understand the national context.

Information was gathered from three further hospital trusts, as outlined in table 1.

Engagement (national investigation)

Stakeholders across the healthcare system were contacted and interviewed to establish their perspective on the national context. In addition, there was engagement with:

• contacts across the UK
• secondary care organisations in regions across England
• software manufacturers for electronic prescribing and medicines administration (ePMA) systems in healthcare.

Limitations to engagement or access to information (national investigation)

Observation visits to hospitals necessarily ceased during 2020/21 when the COVID-19 pandemic significantly affected the UK. For this investigation, even though not all of the planned observations and visits were completed, it was thought that the findings and safety recommendations were supported by the evidence gathered and analysed by the investigation.

3.1.5 Evidence gathering

Multiple sources of evidence were gathered and reviewed by the investigation, including:

• Felicity’s clinical records
• Trust policies, procedures, and practice
• the findings of the Trust’s internal incident investigation report
• relevant incidents reported to the two national patient safety incident databases: the Strategic Executive Information System and the National Reporting and Learning System
• national guidelines and standards
• literature relevant to the identified safety risks.

The evidence gathering process adopted an iterative approach; as further information was gained, additional data sources were identified.

The investigation gathered both interview and observational data from the healthcare settings.

• An interview plan was developed to gather information on safety risks
• observations were conducted and a thematic analysis was performed on the fieldnotes
• virtual meetings were held with a number of software manufacturers.

3.1.6 Analysis

The analysis process had the following aims:

• To generate findings through group discussions after evidence gathering, analysis days, multi-disciplinary team working, reflections on what evidence was missing or inadequate.
• To develop visualisations of the system involved in the patient’s pathway and a timeline of the events. This assisted in recognising communications, interactions and decision making.
• To develop a comprehensive understanding of the healthcare system so that safety recommendations could be identified, and their potential impact on the system could be considered.

3.1.7 Analysis methods used in the investigation

Analysis methods were used to consider local and national practices, and practices evidenced in research literature. This enabled a detailed analysis of how the healthcare system influenced the reference event and allowed potential recommendations for improvement to be considered.

To help understand the healthcare system, the investigation used two methods:

1. The Systems Engineering Initiative for Patient Safety (Holden et al, 2013; Carayon et al, 2006). SEIPS is a framework for understanding structures, processes and outcomes and the relationships between them. SEIPS is explained in more detail in appendix 1.
2. The Functional Resonance Analysis Method (FRAM) (Hollnagel, 2012). FRAM aims to reflect risks within complex systems, by describing variability relative to the functions within the system and potentially modelling what is needed for everyday performance to go right. It is explained in more detail in appendix 2.

3.1.8 Verification of findings

The findings were shared with Felicity’s family, the healthcare organisations involved in the reference event, and with key stakeholders within the healthcare system. This enabled checking for factual accuracy and overall sense-checking.

Communication of management plans

4.1.1 The multi-specialty medical team’s discussion by email concluded that dalteparin should be prescribed at the treatment dose of 100 units/kg/dose twice a day. There was evidence of joint decision making and the communication of plans regarding the intended dose of dalteparin. However, teamworking in this context did not include the specialist pharmacy and nursing teams on the cardiology ward. The medical staff told the investigation that all professional groups were invited. However, this was at variance with the perception of other staff groups who reflected that they would not be told information on the ward round or during a ‘huddle’ (short, focused staff briefings). Some staff considered that they were not part of the discussions at all: “Emails float around and [name of professional group] are the last to hear.”

4.1.2 Correspondence by email appeared to be common practice in cardiology to facilitate discussion across different medical specialties and to generate a written record. The intention was then for these emails to be filed in the medical records. This was supported by a junior doctor who told the investigation that patient plans were either agreed on the telephone or by email, and they were not always part of these discussions. However, they would be updated separately (verbally) by the consultants to ensure they were aware of the plans.

4.1.3 Consultant Paediatric Cardiologist 1 advised the investigation that the email plan had been forwarded to more junior members of the medical team, with a request for it to be added to Felicity’s medical records. A high-level, weight-based plan was made by the consultant team for dalteparin, with the expectation that the ward doctors would undertake the calculation and prescribe a specific dose. The specialist trainee 2 doctor (ST2) inadvertently prescribed dalteparin via the ePMA system at a dose of 1,000 units/kg/dose twice a day, rather than the intended 100 units/kg/dose twice a day.

4.1.4 A free-text box on the ePMA system was completed to say, ‘as per discussion with haematology’. This was intended to reflect the multidisciplinary team discussion involving the haematology. However, it instead later served to falsely reassure the dispensing pharmacist and the nursing staff administering the medicine that there was a rationale for the unusually high dose.

4.1.5 The ST2 told the investigation that while it was not mandatory to use the free-text field for annotations, he did so frequently because nurses often queried doses. At times when a doctor may not be present on the ward (at night-time for example), comments in the free-text box provided additional information about a prescription for the nursing team.

Pharmacy teamworking and workload

4.1.6 In relation to pharmacy team working and communication, checking of the dalteparin prescription was undertaken by a specialist paediatric pharmacist (Specialist Pharmacist 1 (SP1)) and subsequently dispensed by a pharmacy technician. On Friday 13 March, the SP1 was covering the paediatric cardiology ward for the day. The ward normally sits under the remit of a senior advanced specialist pharmacist (Specialist Pharmacist 2 (SP2)). SP1 was covering the ward because SP2 was on leave the previous day and SP1 had covered the ward then, so SP1 provided cover for the ward on the Friday too for continuity.

4.1.7 SP1 told the investigation about a non-work issue that took place on the morning of the dalteparin dispensing, which caused her distress and meant she was away from work for a short time. The unplanned absence from her pharmacy duties created additional pressure for her, as she attempted to make up the lost time. Additionally, the pharmacy workload was described as being generally more intense on Fridays due to weekend leave, replenishing stock, and preparing discharge medicines. On this day, a third specialist pharmacist was on leave which, together with the other factors, compounded the workload pressure for the pharmacists still further.

4.1.8 In relation to pharmacy processes, dispensing required a number of steps and different systems:

4.1.9 In the reference event, the only communication between pharmacy and the ward was the alleged phone call from pharmacy following the dispensing of the dalteparin, to reassure staff regarding the dose. The pharmacist’s professional check would normally have been undertaken in the clinical environment, particularly for newly prescribed medicines. In the case of the reference event, SP1 undertook the professional check for Felicity’s dalteparin prescription remotely, while she was in the pharmacy department. This was due to the non-work issue alluded to in 4.1.7.

4.1.10 SP1 recalled seeing the ‘15’ and ‘as per discussion with haematology’. She was unaware of the plan to commence dalteparin, as she had not been updated from the previous day when the intention had been to start Felicity on clopidogrel. Normal practice for SP1 would have been to undertake a manual calculation to check the dose was correct. On this occasion, the calculation was not made. SP1 thought that potential factors that contributed to this omission were time constraints and being aware of previous delays regarding Felicity’s medication. Pharmacist checking is undertaken to assess for clinical suitability and the safety of medicines, which is essentially verification of the prescription (see subject matter advisor’s analysis 4 on page 58). The investigation heard that this is a common task, but requires manual calculations.

4.1.11 In relation to the dalteparin, SP1 described the choice of anticoagulant as being unusual, since a different type of anticoagulant, ‘unfractionated heparin’, was more commonly prescribed on the paediatric cardiology ward than dalteparin. The investigation heard from SP1 that dalteparin could be prescribed as a treatment (therapeutic) or preventative (prophylactic) dose. SP1 noted that the dose was unusual because it was at a therapeutic (rather than prophylactic) level. The SP1 told the investigation of her initial concerns around how the nurses were going to prepare that dose for administration. The preparation would normally involve withdrawing the prescribed dose from a vial and diluting it to provide a sufficient volume to be given. SP1 then realised that 15,000 units was available via a pre-filled syringe, which avoided the need for a vial and diluting agent.

4.1.12 SP1 was in the pharmacy office when she checked the dalteparin prescription. The investigation observed the working environment to be potentially distracting because of the busy, confined space and elevated levels of noise from other members of staff, bleeps and phones.

4.1.13 The investigation was informed that a phone call was made to the ward by someone from pharmacy, informing the ward of the large dose for dispensing. This was after the pharmacy check and prior to the dalteparin reaching the ward, but further detail could not be ascertained by the investigation. A pharmacy technician informed the investigation that they would regularly contact the ward about medicines being ready for collection. This differed from a specialist pharmacist’s account, who said that a ward would not be routinely contacted in this way.

4.1.14 Following the dispensing of the dalteparin, the next interaction with a pharmacist was 3 days later. This was because the dalteparin was issued on a Friday afternoon and pharmacy provision was reduced to a dispensary-based service only at the weekend, since the Trust did not have 7-day ward pharmacist cover. During the week, the cardiology ward had a higher level of pharmacy service. This included a daily review of every patient’s medication chart, and pharmacist presence on the ward, both of which would have provided an opportunity to verify the dose. The SP2 confirmed this by reflecting to the investigation that: “If it had have happened on a Monday, I feel pretty sure that it wouldn’t have got past Tuesday morning.”

Handovers

4.1.15 Weekend handover from Consultant Paediatric Cardiologist 1 (Consultant 1) to Consultant Paediatric Cardiologist 2 (Consultant 2) occurred at 17:00 hours on Friday 13 March. Consultant 1 informed his colleague of the discussions he had earlier in the day with the neurology team about Felicity’s anticoagulation management. He informed Consultant 2 that he had copied him into the relevant email correspondence confirming the decision to start the dalteparin, along with the recommended dose of 100 units/kg twice a day.

4.1.16 The investigation heard from Consultant Paediatric Cardiologist 2 that it might have been more appropriate for the handover to occur during the Friday morning ward round to allow for increased familiarity with the patients. That said, Felicity was known to most of the medical team, as well as the nursing and pharmacy team, due to the number of times she had visited hospital and the length of her stay on both the cardiology ward and the paediatric intensive care unit (PICU).

4.1.17 The decision to start Felicity on dalteparin was made at 09:47 on Friday 13 March, which would have been after a morning handover had taken place. In relation to subsequent nursing handovers on Saturday and Sunday, the investigation heard that general aspects of care would have been considered during these handovers. This may have included the new dalteparin prescription, but specific information (such as the dose) would not have been discussed. The nursing handover would have been unlikely to detect the error since the purpose was to convey information from one nurse to another about a child’s condition during the shift.

Ward rounds and huddles

4.1.18 The investigation heard that in general terms, the ward round was more about a strategic view from the consultant’s perspective (at the reference event Trust). The consultant said: “My main role at that stage is to see what is keeping the patient in hospital, how can I think about sending her home.” From the ST2’s perspective, their role was more of an operational one, with the expectation that they prepare information about the patients for discussion during the ward round.

4.1.19 The investigation heard that the nurse in charge was supposed to join the ward round, but could only do so on approximately 10% of occasions, due to their own patient workload. Pharmacists and individual nursing staff who were looking after the patients joined the ward round infrequently, if at all.

4.1.20 In terms of the culture of weekend ward rounds, they were conducted differently than on weekdays, with different expectations (custom and practice – practice that has developed over a period of time). The investigation heard that over a weekend period, ward round reviews consisted of a summary of each patient and updates of significant events. Clinicians would usually aim to make decisions for children requiring changes in therapy or discharge decisions. For more patients with long-term, more complex health problems, the aim would be to ensure their condition was stable throughout the weekend.

4.1.21 The ward rounds would be led by a consultant or specialist registrar, depending on the seniority of the registrar. If the ward round was led by a specialist registrar, the on-call consultant, junior doctors, and the nurse in charge would usually be briefed in a post ward round huddle. The ward round would be discussed at the huddle and plans amended and made for the day ahead.

4.1.22 In the reference event, Consultant Paediatric Cardiologist 2 led the Saturday ward round. The investigation was told that he would normally look at the bedside paper prescription charts during a ward round. However, Consultant 2’s lack of familiarity with the ePMA system meant that prescription information was not accessed and Felicity’s prescription was not viewed electronically. Instead, the medications were verbally discussed in terms of the doses of medications by weight (but not the individualised doses prescribed). This was in line with the routine practice of not second checking prescriptions generally. In Felicity’s case, Consultant Paediatric Cardiologist 2 considered that, as per the plan outlined in the email trail, the recommended dose of dalteparin was clear and there was no reason to second check.

4.1.23 The investigation heard that checking of medication during a ward round tended to be undertaken by exception, if medical staff were asked to do so, or if concerns were raised about a particular medicine. The investigation heard that sometimes dose adjustments (increases or decreases) were discussed and documented on the ward round plan by the junior doctors. However, generally, a review of medication would only include starting or stopping medicines, rather than checking individual doses. Medical staff told the investigation that they assumed that a pharmacist would have checked the dose.

4.1.24 The Sunday ward round was led by the Specialist Registrar. Felicity was noted to be on dalteparin, and the medical notes recorded the dose correctly as 1,500 units. The investigation was told that the ePMA system was not checked as the consultants and specialist registrars relied on junior doctors to use the ePMA system. Consultant staff described not being confident using the electronic system, partly because they were unfamiliar with it.

4.1.25 A consultant told the investigation that now everything was electronic, he missed looking through paper charts, partly because of ease of access. The investigation heard that: “Since the introduction of electronic prescription, we don’t have time. It is easy to flip to a [paper] medicine chart and see what it is and then close it, rather than logging into … different systems.”

4.1.26 A huddle followed each ward round, on weekdays and at the weekend. It would involve a specialist registrar or a more junior doctor going through the patient’s condition and the plan for the day. The huddle would not discuss the detail of medications (including doses) unless a query had been raised, though the introduction of a new medication or a dose change would normally be highlighted.

4.1.27 Most of the time, a consultant would be present at the huddle, especially if they had not been on the ward round. The nurse in charge joined the huddle if they had not been on the ward round. They would convey the relevant information to the nurse looking after the patient; this nurse then received the information third hand.

4.1.28 The investigation heard that the role of pharmacy staff was considered to be invaluable by the medical and nursing staff. However, pharmacy staff were rarely present during ward rounds or in huddles. The variability in pharmacy attendance at huddles was because they did not take place at set times and often conflicted with other commitments. The pharmacists instead worked independently on the ward, undertaking pharmacy tasks and responding to queries, mainly from the nursing team.

4.1.29 The investigation heard perceptions that the nursing staff were not always part of multidisciplinary team discussions. This was evidenced by the nursing staff not always being involved in email communication, handovers, ward rounds and huddles. The nursing staff told the investigation that these factors may have influenced the incorrect dose of dalteparin not being detected by the clinical staff over the weekend.

Learning and accountability

4.1.30 As part of the Trust’s serious incident investigation, a review of the error was undertaken from the prescribing, dispensing and administration perspectives. This was carried out to identify ‘where things went wrong and identify areas for action/improvement’. The outcome of the review regarding the administration process was ‘confirmation bias’, in that the dose was appropriate in line with the ePMA annotation. The nursing team felt that, despite this, they were ultimately responsible for the medication error since they were responsible for checking and giving the medicines. Nursing staff told the investigation that they perceived themselves to be the final barrier, and the last opportunity to prevent an incorrect dose from reaching the patient. The investigation witnessed the distress the event had caused them personally and professionally.

4.1.31 In relation to the prescribing stage, concerns about the suitability of the ePMA system for paediatrics was noted by the reference event Trust as being a ‘root cause’. The learning from this was used by the Trust to positively influence quality improvement:

‘The doctor was offered the opportunity to work with our e-medication team, who are working with [new ePMA system] in the development of the paediatric section of the e-medication application, and he is extremely enthusiastic about this opportunity.’

4.1.32 In respect of the culture of the medical team, the investigation heard concerns that consultants made high-level plans, and expected junior doctors to implement them. These plans may not have included guidance on the actual doses of medicines to be prescribed. The investigation team saw evidence of this in practice.

4.1.33 Conversely, members of the medical team told the investigation that consultants would generally state the dose to junior staff for high-risk medications, although it would not always be written in the patient’s notes. In the reference event, the ST2 considered the communication to be clear. The investigation heard that in view of the complex and highly specialised nature of the work in children’s cardiology, junior doctors would check every decision with either the registrar or consultant. This included the doses of certain medicines. The ST2 said “I find in paediatrics it’s generally a very open culture”, adding that “I can ask a dose three times and no one will mind because they know I’m just doing it so that I can get the right dose for the right patient in the right way”.

4.1.34 In relation to nursing staff questioning decisions/medication doses, the investigation heard differing perspectives. The majority of nursing staff told the investigation that it was not uncommon to administer medicines where the prescription did not match the dosage recommended in the British National Formulary for Children (BNFC). Therefore, it may not be second checked or challenged if there appeared to be reasonable justification, which made it difficult to recognise an unintentional deviation. The investigation heard: “We all knew it [the dose] was high, but had a legitimate reason for it.”

4.1.35 A nurse told the investigation that the culture on the cardiology ward was such that she felt able to challenge her colleagues when required. For example, she felt able to ask colleagues to be quiet when she was preparing medicines. Another member of the nursing team told the investigation that when a prescription was outside of the BNFC guidance, and something did not seem right, she would question it: “I will ask if I don’t know.” The nurse said she would feel confident in challenging other staff until she received a satisfactory response.

4.1.36 The dalteparin medication error was detected by Nurse 8 after seeing the unfamiliar box of dalteparin (a purple bannered box) in the medicines cupboard while preparing dalteparin for her patient. It was during her break that she reflected on the unexpected box of dalteparin and afterwards she checked the dose, noting the 15,000 units. The nurse knew the dose of dalteparin from memory and her knowledge of normal doses combined with the unfamiliar packaging, in the context of a bloodstained vomit, triggered her to question and take action. She checked Felicity’s paper medical records for the dose (which Nurses 1 to 7 appeared not to do), and recalled seeing the email in the records summarising the multidisciplinary team decision. Following this, she checked the BNFC and the error was reported. Further doses were then withheld pending a medical review and the incident was reported.

Prescribing

4.2.1 The ePMA system was being used by the reference event Trust on adult wards on three separate hospital sites. While ePMA systems had been used in adult areas for 8 years prior to the reference event, an ePMA system had only been introduced into paediatrics 10 months before the incident. The usability and functionality of the ePMA system was raised as an issue across the multidisciplinary team in paediatrics.

4.2.2 The ePMA system used by the reference event Trust was configured to encourage prescribing for both adults and children via ‘dose sentences’ where possible. Dose sentences are inbuilt prescriptions within the ePMA system that generate pre-populated doses meaning the prescriber does not need to enter a dose. In the Trust’s ePMA system, dose sentences could be accessed via a ‘quick list’ for adults and through a ‘protocol function’ for children. The dose sentences for adults and children were intentionally separated.

4.2.3 The investigation heard about the challenges in developing some dose sentences in children because doses will vary for each medicine according to the child’s weight (newborn to 100kg or greater) and other characteristics (see 1.2.1). SP2 told the investigation that the system lacked:

‘… the human element when it comes to rounding up or rounding down critical doses, so we were unable to really sign off the dalteparin protocol because we couldn’t get the system to make that human decision point when something sits in the middle of two doses.’

4.2.4 The intention was for prescribers to use the standard ‘prescribe’ function instead, which allowed them to select the required medication in the most appropriate strength, which might come in either a pre-filled syringe or vial. SP2 advised the investigation that this was not always possible because “on the list of the options to pick from, our preferred product was always greyed out, so it wasn’t intuitive to pick”. The system had been pre-set to encourage the selection of the preferred strengths for adults.

4.2.5 The introduction of the ePMA system in paediatrics was described as being time pressured by some staff groups, compounded by additional configuration requirements reflecting the complexity of prescribing in paediatrics. The investigation was told that in order to transfer the medication prescriptions and protocols from the test system to the live system, the system had to be rebuilt and tested again. Not all of the planned and tested features were then transferred to the live system, resulting in the system being rolled out to paediatrics in an incomplete state.

4.2.6 In the reference event, the ST2 was able to select an adult dalteparin ‘quick list’ and the dose defaulted to the syringe size intended for adults. The system then relied on the doctor manually entering the correct dose, replacing the dose information that had been automatically entered by the system. The investigation was advised by a software manufacturer that if a trust was using adult protocols in a paediatric environment, the expectation would be that this would be written into a safety case. Safety cases are used to demonstrate that patient safety risks have been addressed rigorously and proactively. The risk would then be mitigated by the local organisation through appropriate controls. The investigation found no evidence of this.

4.2.7 The need for manual alteration of a default dose was at variance with prescriber expectations of the ePMA system, one of whom said:

“… you put weight, you put a dose and the system should say this doesn’t add up, but that wasn’t the case.”

The clinicians’ perceptions were that the ePMA system would support their clinical decision-making and protect them from entering wrong doses. The investigation also heard that, while the system might alleviate some errors, it could not be assumed it would prevent all. The investigation found that it was not widely recognised by the clinical staff that the ‘alert’ functionality, which flags up potential errors, depended on the Trust’s configuration of the ePMA system.

4.2.8 In relation to available functionality, the ePMA system had an integrated calculator function. The software manufacturer advised this should have been a key functionality of an ePMA system used in paediatrics, where all medication doses are prescribed by a weight-based calculation. It was a manual function, which auto-populated the patient’s weight and prompted the prescriber to enter the dose required. The recommended dose from the calculator would then be transcribed back into the system. If a weight had not been entered (as occurred in the reference event), then this functionality could not be used. The investigation was told that some staff at the trust were concerned about the validity of the calculator and training did not appear to adequately prepare the prescriber to use it. The ST2 told the investigation that he had never used the calculator function. He said:

“There is a built-in calculator that you can use just to, like, double-check your prescription, but ... I’ve never heard of anyone using it.”

The process of prescribing the dalteparin

4.2.9 Following confirmation from Consultant Paediatric Cardiologist 1, combined with consultation of the BNFC and the medical records, the ST2 calculated the dose manually, using Felicity’s weight and the dose of 100 units/kg twice daily. The ST2 then accessed the ePMA system.

4.2.10 The prescribing of the dalteparin was described by the doctor as being straightforward. It was undertaken at the ward nurses’ station, using a laptop. The doctor described the environment as quiet and he was undisturbed during the task. No personal factors were noted on the day.

4.2.11 The following steps were undertaken as part of the prescribing process:

1 The doctor typed in the name of the medicine in the ‘Name’ box in top left corner of the screen, and a set of formulations appeared on the right of the screen. Anything which is greyed out was not in the Trust’s formulary.

2 The doctor selected dalteparin sodium 15000/0.6ml injection from the drop down ‘Adult A-Z, Quick List’ (see figure 3).

3 On a second screen, 15000 was auto-populated in the dose field (see figure 4), which the doctor planned to manually alter to 1500 (but either did not do this or entered the number incorrectly).

4 On a third (final) screen, there was a requirement to ‘update’ the prescription, to finalise the process.

4.2.12 The prescription of dalteparin then underwent a pharmacy check by SP1 on the ePMA system before being transferred electronically to the pharmacy department. There were three different electronic systems in pharmacy used for the checking and dispensing of medicines (the ePMA system, a prescription tracking system and a dispensing system), with limited interconnectivity. Once the dalteparin had been dispensed, the nursing staff were required to administer it. They used tablet computers for accessing the ePMA system, which they also used for recording observations via a different application.

Administration

4.2.13 The investigation was told that the Wi-Fi at the hospital site was “temperamental”. At the medicines administration stage, this presented a risk for nursing staff to mitigate, because a drop in the Wi-Fi could result in a medication not being signed for. This potentially could prompt a second dose to be given in error. The investigation saw evidence of this where a dose of dalteparin had been given, but it had not registered on the system because the Wi-Fi had dropped out. The nursing staff then needed to sign in later when the Wi-Fi was available to retrospectively sign off the dose.

4.2.14 The ePMA system was described by the nursing, medical and pharmacy teams as being suboptimal in terms of its functionality and usability. The investigation was told that the ePMA system was rigid, and did not allow for manipulation of the functionality to support safe practice. This included limited safety features such as limited alerting functionality and functions to prevent the user from taking an action. Regarding usability, while the investigation did not undertake a specific usability assessment, staff feedback was considered in relation to Nielsen’s principles for user interface design (see table 2) (Nielsen and Molich,1990). Not all of these principles were found to be contributory to the reference event.

4.2.15 Following the reference event (but not as a direct consequence), paediatrics was withdrawn from the ePMA system implementation programme. This was pending a different digital system, which was in the process of being rolled out in the Trust at the time of the investigation’s site visit.

4.3.1 The ePMA system was introduced into the Trust in 2012 for adult patients. The investigation was told that three clinical safety officers were in post (who were professionally medical and nursing staff), to oversee the initial implementation of the system in the clinical environment. The three clinical safety officers were trained and there was a clinical safety assessment for the first deployment of the ePMA system, in accordance with the national standard (DCB0160) (NHS Digital, 2018) (see ‘Clinical risk management standards’ below). The investigation was unable to determine whether the Trust went through a further clinical safety assessment process ahead of the implementation of the ePMA system on the paediatric cardiology ward in May 2019, which was the first paediatric area to use the system. Expectations for clinical safety assessment are outlined in ‘Clinical risk management standards’ below.

4.3.2 The investigation heard that general system configuration of an ePMA is governed by the software manufacturer, which retains responsibility for any changes that a Trust may request. The local configuration specific to a trust is governed by the trust, which has responsibility and ownership. Examples of local configuration include the development of quick lists (to organise medicines into folders for selection), dispensing practice, dose range checking order sets/guidance and restrictions. Dose range checking is required to prevent overdosing of medicines and can be set up by Trusts via an alert mechanism.

4.3.3 The investigation was told by a software manufacturer that there appeared to be limitations in the way the quick list had been set up by the reference event Trust. This was because there appeared to be no rule or alert set up for dalteparin to prevent overdosing. The Trust told the investigation that this could be challenging to set up because of other decisions that needed to be made in terms of recording a patient’s weight in the ePMA system. The Trust’s expectation was that if there was not a paediatric quick list, the prescriber should have been able to select from a generic formulary item. However, only the adult quick list was visible in the reference event and the ST2 was unable to select a generic item.

4.3.4 The investigation was told by another trust that was using the same ePMA system as the reference event Trust that it was not possible to safely separate quick lists for adults from quick lists for paediatrics. They stated that quick lists could be organised into folders, so allowing for segregation when users went into the ‘quick list’ function. However, on the prescribing search screen, quick lists for a given medicine were all listed on the same screen. This increased the amount of information that paediatric prescribers had to scroll through and did not support the separation of quick lists for adults and children across the same organisation. This trust does not use the quick lists for children, but rather medication ‘protocols’ for prescribing.

4.3.5 The investigation was told by the clinical staff involved in the reference event that they had concerns about the implementation of the ePMA system. These concerns related to the adaptation of an adult system to paediatrics, training, testing, functionality, and the rationale for roll-out.

4.3.6 Staff members’ perception was that the main driver for the paediatric cardiology ward being included in the ePMA system roll-out was to determine how the system functioned in paediatrics. The investigation heard from the clinical staff that other reasons for the implementation were that the ward had fewer beds than other paediatric areas, staff were keen to pilot and the rest of the hospital site was also using the ePMA system. An alternative hospital site in the same Trust where the majority of children’s services were located, presented challenges with implementation, due to the existence of different systems with limited interoperability.

4.3.7 The Trust’s implementation team told the investigation that the decision to roll out the ePMA system to paediatrics was influenced by the paper-based prescribing errors reported by the specialty. The Trust had discussions with another trust that was using the same ePMA system in paediatrics. The second trust reported that the system offered a number of benefits over paper systems, in particular greater clarity of prescriptions. This trust’s configuration was used by the reference event Trust when it piloted the implementation on the paediatric ward. From a strategic perspective, the Clinical Commissioning Group and national NHS policy was to move to ePMA and electronic systems; however, the deployment of such systems was not a high priority for the Trust at that time.

4.3.8 Following implementation of the ePMA system, some paper prescription charts still needed to be used on the paediatric cardiology ward alongside the ePMA system, resulting in a hybrid situation. Paper prescription charts were available for medicines with variable doses, since these could not be safely managed within the ePMA system. Paper charts were also transferred with patients admitted from the emergency department and PICU, as the ePMA system had not been implemented in these areas.

4.3.9 Training of the paediatric clinical staff on the ePMA system was described as ‘ad hoc’ and took place during the working day. Staff who did the training considered that it did not prepare them to use the system and there was no competency assessment. Telephone support was available 24 hours a day for the first weeks after implementation and there was a paper log on the ward where staff could note issues for troubleshooting. The investigation heard that issue logs from the wards would have been reviewed by the project team, which at this point was led by the ‘eMeds pharmacist’. Usually, issues were discussed with the ward team if clarification was required and triaged for action, depending on whether it related to medication file configuration or required action from the Trust’s IT department or the system supplier. However, the ward team could not recall any improvements being made as a result of the log and the investigation was unable to access a copy to view. The investigation also heard that issues were not always reported on the Trust’s incident reporting system.

4.3.10 The consultants involved in the reference event could not recall receiving any training on the ePMA system, although they acknowledged that it might have occurred alongside other mandatory training. The ST2 recalled receiving training on the ePMA system at the beginning of the year in paediatrics and was also familiar with the system from previous roles in adult medicine. He found the system fairly straightforward to use adding that he got “on well with computers and technology in general”.

Loss of cues

4.4.1 In pharmacy, the worksheet used for the dispensing process used to be a handwritten transcription of the prescription. The worksheet was yellow for adult patients and pink for paediatric patients. Following the implementation of the ePMA system, all worksheets printed off the system were white. The investigation was advised that this made it difficult to identify a paediatric prescription from the automated worksheet, particularly as it did not state ‘child’ or ‘paediatric’ anywhere on the worksheet. The investigation heard how the colour of the sheet may have previously alerted the dispensing technician to it being a paediatric patient, so acting as a visual cue.

4.4.2 Nursing staff described how in hindsight cues during the medicines administration process led them to believe that it was the correct dose. This was particularly evident in relation to the ongoing administration of the medicine, despite individual concerns. The cues included:

• documentation which described ‘as per discussion with haematology’
• the phone call from pharmacy confirming the dose was high, but appropriate
• the patient’s name on the box containing the dalteparin
• the fact that patients often received doses outside of those in the BNFC
• and following the first dose, the fact it had been given before.

4.4.3 Multiple nursing staff informed the investigation that they recognised the unusually high dose at the time of preparing the dalteparin for administration, and questioned the dose. It appears that the free-text comment on the ePMA, ‘as per discussion with haematology’, may have allayed any concerns, resulting in the medication being administered. The exception was the final dose, at which point the error was detected.

4.4.4 The packaging of the dalteparin, labelled ‘15,000IU/0.6ml’ and with a purple banner (see figure 5), was unfamiliar to staff and the majority of staff did not recognise its significance. The dosing instructions on the box instructed to ‘give the full vial’, so they did. However, while unfamiliar medication packaging acted as a cue for some staff that the wrong medicine dose may have been chosen, it was not a cue for all.

Figure 5 Ward stock of dalteparin (left) and dalteparin 15,000 units

Second checking of medicines

4.4.5 The nursing staff who administered the first dose described to the investigation that they were still on the ward at 20:30 hours, and were “shattered” after a 12-hour shift, which should have ended at 20:00 hours. They gave the dalteparin during their shift to support their colleagues on the night shift.

4.4.6 There was evidence of different levels of familiarity with dalteparin among staff on the ward. Some nursing staff described it as being a commonly used medicine in paediatric cardiology and readily knew the dose. Others said it was not used every day. Dalteparin was a stock item on the ward, but not the ‘15,000IU/0.6ml’ concentration. A deep vein thrombosis (DVT) was also described as rare on the ward and this potentially led to rationalisation of the unusual dose, namely that a higher dose was required for therapeutic treatment.

4.4.7 The preparation room for medicines was located part way up a corridor on the cardiology ward, away from the central nurses’ station and bed area. Within the room there was a doorbell which was used to summon colleagues to check a medicine when required. The investigation considered this represented a workaround, which was required because of the location of the room. This contributed to an absence of a second checker from the beginning to the end of the medicine process.

4.4.8 The investigation heard how an independent second check took time and this was challenging when there was limited staffing on the ward. If there were only two registered nurses on a shift, they could not both be in the medicines room together because of the need to have sight of the patients. Medicines were therefore often prepared alone with a second check being undertaken before administration. Checking processes, particularly at the point of administration, differed from those specified in Trust policy/procedures.

4.4.9 During the first administration of dalteparin, both nurses were in the medicines room, but preparing medicines for different patients. Nurse 1 asked Nurse 2 about the dalteparin and Nurse 2 directed Nurse 1 to check the medicines guide, as Nurse 2 had never had to administer it fully from a pre-filled syringe. Nurse 2’s second checking involved checking the prescription and comparing it to medicines guide. They felt the prescription was correct after seeing the note about the haematology plan. Nurse 2 was not familiar with the detail of Felicity’s plan of care.

4.4.10 The investigation heard that the door of the preparation room might be propped open if two nurses needed to be in the room at the same time, and no other staff were on the ward. This was a particular issue out of hours. An investigation into a previous medication incident in cardiology had recommended a reduction in the number of times the door was propped open, to reduce interruptions.

4.4.11 The first dose of dalteparin was given by a nurse who was also the nurse in charge. Nursing staff described short staffing on a number of shifts during the weekend period, resulting in medicines being given late. Nurse 4 was the second checker for the second dose given at 06:35 hours on Saturday 14 March and remembered commenting on the unusually high dose. However, she was then falsely reassured by the note saying it had been discussed with haematology, the pharmacist’s check, Nurse 3’s assurances, the box of dalteparin with Felicity’s name on it and the fact that the medicine was being used to treat a DVT.

4.4.12 There was no specific mention by nurses 1 to 7 of referring back to the medical records to check the documentation of the intended prescription. The intended dose of dalteparin was not documented in the medical records before prescribing, other than in the email which was supposed to be filed in the records. There was an annotation made by the specialist registrar during the Sunday ward round which stated ‘dalteparin 100u/kg BD’. Therefore, unless the nursing staff had seen the email correspondence which included the dosing information, the medical records would not have helped them to detect the error before the Sunday evening dose.

Interventions which support multidisciplinary communication

Handovers

5.1.1 There is evidence that clinically significant information may be lost during the handover process, and presence of a consultant at handover improves outcomes for children (Royal College of Paediatrics and Child Health, 2015). Relevant to this, one of the standards for acute general paediatric services outlined in ‘Facing the future’ (Royal College of Paediatrics and Child Health, 2015) states: ‘At least two medical handovers every 24 hours are led by a consultant paediatrician.’ The investigation observed and heard that this was not always practical because of the on-call arrangements for consultants, particularly at weekends.

5.1.2 An audit by the Royal College of Paediatrics and Child Health (2013) showed there was support for handovers involving paediatric consultants and senior nursing staff to ensure the multidisciplinary team was kept up to date. It highlighted that ‘many clinical directors and ward managers felt that an inclusive, well documented handover was the glue that held their service together’.

5.1.3 In relation to nursing teams, there are no national handover standards. The investigation contacted the Royal College of Nursing (RCN) about this and was advised about work that was undertaken in 2019 around the use of human factors and the situation, background, assessment, recommendation (SBAR) communication tool in handover (Royal College of Nursing, 2019). This was intended to provide a structured handover framework for nursing teams. The investigation also noted there were local policies in place and observed these in practice during site visits to trusts.

Quality improvement initiatives and huddles

5.1.4 The Royal College of Paediatrics and Child Health (RCPCH) told the investigation that it had previously worked with more than 50 paediatric units over 4 years on the Situation Awareness for Everyone (S.A.F.E.) programme (Royal College of Paediatrics and Child Health, 2018). This was implemented by developing and trialling quality improvement techniques, to improve communication and build a safety-based culture. By using S.A.F.E principles, the RCPCH anticipated that paediatric units could, amongst other outcomes: ‘reduce avoidable error and harm to acutely sick children’ (Royal College of Paediatrics and Child Health, 2018).

5.1.5 S.A.F.E is also a key theme for Making it Safer Together (MiST). This is a paediatric patient safety collaborative of hospitals whose goal is to continually reduce healthcare-associated harm, by sharing experiences, tools, and ideas. The safety ‘huddle’ was included as part of the S.A.F.E. project to improve situational awareness. It has been described as an intervention to improve the quality of care, and has been implemented in various settings within healthcare (Royal College of Paediatrics and Child Health, 2018).

5.1.6 Following on from this, the investigation was advised that a ‘DRUG-gle’ could be a valuable intervention to reduce medication errors (Royal College of Paediatrics and Child Health, 2020). A DRUG-gle is a specific medication-focused safety ‘huddle’, involving the paediatric ward doctors, nurses and pharmacists meeting as part of the ward routine, usually after the ward round. Integral to this is a short, focused discussion when checking individual patient’s medication charts on the ward round. This is described using the acronym ‘APT’:

• ‘Is it Appropriate? – do we need to stop or change it?
• Is it the correct dose Prescribed? – can it be rationaled?
• Is the Timing OK? – is it ambulatory friendly?’

DRUG-gles were not observed during site visits, but research literature indicates they are undertaken in some organisations across the UK (Shore et al, 2017).

5.1.7 The principles of S.A.F.E. huddles/DRUG-gles in paediatrics encourage more multidisciplinary input and have been used at several sites in England. The investigation found that however huddles are not always multidisciplinary; content is varied, as is attendance; and there are no clear aims or objectives. The investigation observed that their primary purpose appeared to be the updating of staff who have not been able to attend the ward round. The investigation observed a huddle during a comparison site visit. The huddle appeared to be informal, with nursing and medical staff gathered behind a nursing station, with frequent interruptions and distractions.

5.1.8 In summary, evidence from the reference event investigation, observation undertaken at other trusts and information from the RCPCH and the RCN indicates that while there are handover standards for paediatricians, there are no similar standards for paediatric nursing staff. The SBAR communication tool is advocated by the RCN, but the investigation found no evidence that this has been widely adopted by trusts for the purpose of handovers and huddles. Additionally, there is no standard format for conducting huddles.

Ward rounds

5.1.9 In the case of the reference event, the nursing and pharmacy teams did not appear to be completely integrated into the cardiology multidisciplinary team. This was particularly evident in relation to irregular attendance and active participation in ward rounds. In addition, the nursing and pharmacy teams were not integral to the discussions about the initiation of anticoagulant therapy.

5.1.10 The investigation heard that the irregular ward round attendance by the nursing and pharmacy teams was resource and capacity related, in addition to the timing of the round, which took place when other tasks needed to be undertaken. The pharmacy team reflected on whether their attendance on ward rounds was routine practice:

“It is in an ideal world, and I have done in the past where possible. It’s not always possible to follow the ward round around because they don’t always happen at the same time of day.”

The Trust where the reference event took place reflected that for ward rounds, ‘work as done’ (that is, the way staff did them in practice) (Shorrock, 2016), did not meet the national guidance ‘Standard Ward Round NHS Improvement’. This was because ‘standards and expectations were inconsistent’.

5.1.11 The investigation observed practice at a comparison site trust and noted that the pharmacist was an integral part of the ward round, amending any changes to the children’s medicines directly on the ePMA system during the ward round. In addition, the investigation observed the nursing staff caring for the individual patients providing a direct handover to the consultant leading the ward round. The nurse in charge was not part of the round, but joined a post-round huddle to ensure they were updated in respect of key changes.

5.1.12 The Royal College of Physicians (2021) states that all professional staff should input into ward round discussions, adding that the nurse caring for the patient was essential to these discussions. The role of the ward co-ordinator is also described as a key role, particularly in respect of inputting into pre- and post-ward round huddles, and receiving updates during the round.

5.1.13 In relation to pharmacists, the ‘Modern ward rounds’ guidance published by the Royal College of Physicians (2021) states that: ‘Pharmacy input is also essential for most patients and, where resources allow, there are demonstrable benefits to the ward pharmacist being part of the ward round.’ The guidance specifically refers to medication reviews being incorporated into ward rounds. ‘Professional standards for hospital pharmacy services’ (Royal Pharmaceutical Society, 2017) similarly advocates that:

‘Pharmacy team members are integrated into multidisciplinary teams across the organisation and provide patient facing clinical services to ensure safe and appropriate medicines use for all patients …’

5.1.14 Sutherland et al (2019) found that while there are currently no paediatric interventional studies evaluating the impact of ward-based pharmacists in the UK, their study demonstrated their contribution to the detection and resolution of medicine-related issues. Their study on prescribing errors in paediatric intensive care units found that pharmacists were an important control for medication error, but were only present on an ad hoc basis (Sutherland et al, 2019).

5.1.15 In the reference event, the lack of a 7-day clinical pharmacy service meant that the specialist pharmacists were not present on the cardiology ward from Friday (when the dalteparin was dispensed) until the following Monday. A higher level of pharmacy service would have afforded a daily review of Felicity’s medication chart and pharmacist presence on the ward, both of which would have provided an opportunity to verify the dose of dalteparin.

5.1.16 The requirement for improved 7-day services has been recognised for many years, as outlined in the publication ‘Transformation of seven day clinical pharmacy services in acute hospitals’ (NHS England, 2016a). This includes potential solutions to improve workforce capability 7 days a week, and the means by which national/local organisations and individuals can support delivery. This is an existing focus for improved 7-day services.

5.1.17 Additionally, HSIB has previously made a safety recommendation to the Royal Pharmaceutical Society (RPS) relating to hospital clinical pharmacy provision. This was in connection with the national investigation ‘The role of clinical pharmacy services in helping to identify and reduce high-risk prescribing errors in hospital’ (Healthcare Safety Investigation Branch, 2020a). The aim of the safety recommendation was to reduce variability in hospital clinical pharmacy provision and is therefore relevant to this investigation. The investigation was advised that the RPS was planning to start refreshing its professional standards for hospital pharmacy services at the end of 2021. However, implementation has been delayed pending the provision of funding for the development of the guidance.

5.1.18 Evidence gained through analysis of the reference event, comparison site visits, discussions with the RCPCH and the RCN and a review of the research literature indicates that there is no standardised approach to conducting ward rounds in paediatrics.

5.1.19 Recent work has been undertaken to update the ‘Modern ward rounds’ guidance (Royal College of Physicians, 2021). A multidisciplinary approach to modern ward rounds was adopted within this guidance, with a focus on the inclusion of patients, carers and families in decision-making processes and communication. Ward rounds were described as being a:

‘… focal point for a hospital’s multi-disciplinary teams to undertake assessments and care planning with their patients. Co-ordination of assessments, plans and communication is essential for effective and efficient care.’

5.1.20 The guidance also specifies that information on the patient’s condition gathered at the shift handover should be discussed in a board round or huddle prior to the ward round, attended by all members of the multidisciplinary team (Royal College of Physicians, 2021).

5.1.21 The publication reflects that ‘most ward rounds in UK hospitals require considerable improvement and research and quality improvement is necessary to inform effective practice’. It incorporates best practice principles for multidisciplinary teams such as ‘Agree principles, standards, functions and structure for local ward teamworking’ and ‘Clarify each team member’s role’ (Royal College of Physicians, 2021).

5.1.22 The ‘Modern ward rounds’ guidance (Royal College of Physicians, 2021) is currently for clinicians treating adult patients and not children. There is no specific guidance for ward round management in paediatric settings.

5.1.23 The RCPCH told the investigation that if the guidance was to be used in paediatrics, it would require greater emphasis on the views of the parents and the voice of the child, both during ward round discussions and decision making. The RCPCH advised that there would also need to be consideration of the role of the paediatric pharmacist in supporting the ward round process in different hospital environments. This role includes prescribing during the ward round, stopping medications/reducing a dose, reviewing the length of treatment and monitoring of pros and cons of the medicine.

5.2.1 Children are estimated to be three times more likely than adults to experience medication-related harm (Sutherland et al, 2019), which is thought to be related to the increased complexity in the approach to medication for children. Administration errors are reported to occur in 20% to 25% of dose administrations (Koyama et al, 2020). In a systematic review that included paediatric and adult medicines administration, Koyama et al (2020) identified that interventions to reduce errors during this stage in the process are particularly important, since it is the final step before a patient receives a medicine.

5.2.2 Information technology is increasingly being used by hospitals to facilitate the medication administration process and associated with this is the implementation of significant workflow changes (Koyama et al, 2020). This provides an opportunity to examine work practices and to reconsider long-held practices. While the implementation of ePMA systems has reduced the risk of transcription errors, the technology may have unintended consequences. One of these consequences is that it takes longer to complete the medicines administration process, due to both nurses needing to log on to the system independently to verify the second-checking process (Koyama et al, 2020).

5.2.3 Qualitative studies also indicate that nurses are significantly less likely to second check medications when using an ePMA system. There is evidence of substantial disruption to nurses’ work in both seeking out and responding to requests for second checking, which can contribute to task errors and potentially increased safety risks (Westbrook et al, 2020).

5.2.4 The investigation was advised by a comparison site trust that many ePMA systems are configured to mandate a second check for certain medications or certain types of patients. Therefore, this step cannot be omitted since it requires the second nurse to add their ‘PIN’ (unique registration code) before they can document the administration of the medication.

Second-checking process

5.2.5 Within the research literature, definitions of second checking vary considerably, with terminology often used interchangeably. Several studies have demonstrated that organisational second-checking policies often differ in terms of how the process should be conducted, which contributes to variation in how nurses apply the policies (Koyama et al, 2020). This was also demonstrated by the investigation’s analysis using the Functional Resonance Analysis Method (FRAM) to explore variability in the application of organisational policies (see figure 6).

5.2.6 Some organisations require second checking for all medicines while others only require it for high-risk medicines. Koyama et al (2020) also reported that the majority of non-numerical qualitative studies they reviewed and cognitive theory (focused on thought), provided limited detail on the steps in the medication administration process which required second checking.

5.2.7 Second checking of medications has been implemented in hospitals based on an assumption that it will result in fewer errors (Koyama et al, 2020). However, to date there is limited evidence of the effectiveness of second-checking procedures to reduce errors in prescribing for children, despite the extensive use of the policy in paediatric hospitals worldwide (Westbrook et al, 2020). Koyama et al (2020) attribute the limited evidence to the paucity of high-quality randomised controlled trials.

5.2.8 ‘Primed double-checking’ (see section 1.4) may lead to confirmation bias (the tendency to focus on or interpret information that confirms one’s existing beliefs) (Westbrook et al, 2020). As the Institute for Safe Medication Practices (2019) states:

‘Two people working independently are less likely to make the same mistake; if they work together or suggest what the checker should find, both could follow the same path to error.’

There is limited information on how independent and primed double-checking is differentiated (Koyama et al, 2020). However, independent double-checking is preferred since if the checker is primed, an error may not be detected due to confirmation bias (Koyama et al, 2020).

5.2.9 In relation to ‘independent double-checking’ (see section 1.4), published studies have produced conflicting findings. Findings by Konwinski (2020) indicated that ‘there was no compelling evidence to adhere to the IDC [independently double-checked] process at all’. The Institute for Safe Medication Practices (2019) believes that the selective and proper use of manual independent double-checks can play an important role in medication safety, particularly among vulnerable patients (including children) and for high-risk medications. Koyama et al, (2020) concluded that independent double-checks should minimise errors that arise from one individual, providing that an independent calculation is carried out by a clinician with no prior information, to avoid influencing their decision making.

5.2.10 Several studies (see appendix 3) have demonstrated the ability of independent double-checks to detect up to 95% of errors (Institute for Safe Medication Practices, 2019). Despite positive attitudes about their use, independent double-checks have ‘long been disputed, discounted, and misused in healthcare’ (Institute for Safe Medication Practices, 2019). As outlined by Dickinson et al (2010):

‘The adoption of IDC [independent double-checks] in health care settings must have in place: policy and guidelines that clearly define the process of checking, educational support, an environment that supports peer critique and review, well-designed medication areas and accessible resources to support drug administration.’

5.2.11 Westbrook et al (2020) suggested that the current application of double-checking policy may no longer be fit for purpose in modern clinical settings. Pfeiffer et al (2020) supported this, stating that consideration should be given to reconceptualising the double-check process to include a wider range of options, such as ‘single-person double-checking’.

5.2.12 A hospital in the US used a human factors approach to minimise medication administration errors by piloting and introducing ‘single-checks’ (Konwinski et al, 2021). This resulted in a reduction in the medication error rate from 1.21% to 0.78%. In addition, less time was spent searching for another to nurse to perform the check and to document medication administration.

Cognitive processes involved in checking medicines

5.2.13 Koyama et al (2020) have explored the factors which may further influence the reliability and effectiveness of double-checking. These include the following, many of which were evident from the analysis of the reference event:

• the automatic nature of the task which may decrease staff members’ attention
• failure to look for and process additional information once initial information looks correct (‘satisficing’)
• excessive trust in the person whose work is being checked
• deference to authority when a junior nurse may not challenge an error made by a more senior colleague
• distractions and interruptions
• diffusion of responsibility (where person is less likely to take responsibility when other colleagues are present) and an overreliance on double-checking, in which staff believe someone else will catch any mistakes, leading to a false sense of safety (Koyama et al, 2020).

5.2.14 The ‘Efficiency-Thoroughness Trade Off’ concept (Hewitt et al, 2016) and ‘Social Loafing’ (observed when checking another person’s work) (Koyama et al, 2020) are said to contribute to the diffusion of responsibility and lack of concentration on the task. Koyama et al (2020) found that in the case of medicines administration tasks, where independent double-checks were required, these factors contributed to errors not being detected during the checking process.

5.2.15 There are different cognitive processes involved in checking medicines. Double-checking against a prescription requires a person’s attention for the activity, but not their critical thinking (White et al, 2010). However, critical thinking is required (using a nurse’s own knowledge) to evaluate whether a dose calculation is correct and for identifying an error in the prescription (Dickinson et al, 2010). Pfeiffer et al (2020) consider that calculations involve generating (as opposed to comparing) information, with the calculated value being new, ‘generated’ information. It was suggested that creating space and points in time in the medication process for plausibility reviews (using own knowledge) represents a way of using reflective thinking for detecting errors (Pfeiffer et al, 2020).

5.2.16 The investigation observed that what was often missing in the independent double-check process was a ‘sterile cockpit’ environment – that is, an environment void of unnecessary distractions. According to Federwisch et al (2014), this should improve patient safety if applied at key points in clinical procedures.

5.2.17 To broaden the understanding of double-check processes, the investigation explored how other, high-reliability industries carried out their checking processes. It was noted that checks that rely on humans pose a safety risk in all industries. This is exemplified in various reports by the Air Accidents Investigation Branch (2020; 2018). The investigation also had the opportunity to observe checking processes in helicopter engineering and simulated flight processes during a visit to a Royal Naval Air Station. The sterile cockpit concept was observed, where pilots and staff were required to refrain from non-essential activities during critical phases to reduce distractions to a minimum. The investigation’s learning from the visit is provided in learning case study 1.

5.2.18 In summary, evidence from the reference event investigation, observations of work during comparison site visits, learning from the Royal Navy, learning from Air Accidents Investigation Branch and a review of the research literature (Westbrook et al, 2021; Koyama et al, 2020; Castiglione and Ekmekjian, 2017; Hewitt et al, 2016; Schwappach et al, 2016) undermine confidence in the effectiveness of second checking in healthcare.

5.2.19 There is limited evidence comparing the effectiveness of independent versus synchronous (checking side by side) manual second checks on reducing the risk of medication administration error (Castiglione and Ekmekjian, 2017). The whole process of ‘checking’ appears not to be specified in sufficient detail. According to Koyama et al (2020), there remains a need for higher-quality studies to determine if, and in what context, second checking produces sufficient benefits in patient safety to warrant the considerable resources required.

5.2.20 The investigation recognises that the safety recommendations made in this report focus on checking processes that are reliant on humans. The investigation also found evidence, supported by research literature, of the important role of the working environment in supporting effective checking processes. For example, the investigation saw how the set-up and environments within which medicines were being checked resulted in distractions and interruptions.

Uptake of electronic systems and the impact on medication errors

5.3.1 Data indicates the use of health IT systems is becoming increasingly widespread and the functionality more sophisticated (NHS Digital, 2018). NHSX, the organisation tasked with driving forward digital transformation in healthcare, advised the investigation that around 50% of trusts are using ePMA systems, with a further 25% to 30% funded for go-live dates over the next 12 to 18 months. It was anticipated that the increased uptake of ePMA systems by trusts would correspond with a 30% reduction in medication errors (NHS England and NHS Improvement, 2019).

5.3.2 In paediatrics, ePMA systems were viewed as being key improvement measures to reduce medication errors (Mohsin-Shaikh, 2019; Campbell et al, 2006). In support of this finding, the investigation regularly heard from frontline staff that their perception was that the ePMA systems would prevent medication errors.

5.3.3 However, contrary to expectations, the findings of a recent systematic review and meta-analysis conducted by Gates et al (2021) concluded that ePMA systems had variable impact on medication errors and harm. Furthermore, NHS Digital (2021) stated that while ‘it must be recognised that failure, design flaws or incorrect use of such systems have the potential to cause harm to those patients that the system is intended to benefit’. This is supported by Sutherland et al (2019) who state that poor design and the implementation of ePMA systems can lead to a worsening of safety, and higher numbers of deaths.

5.3.4 According to Fox et al (2019), there are three types of ‘technology-related errors’ created by ePMA systems; these are discussed further in subsequent sections:

• not preventing the majority of harmful prescription errors
• the varying levels of decision-making support and warnings provided to prescribers
• differences in the systems being used, and between the same system on different sites.

5.3.5 Sutherland et al (2019) reflected that with the drive to improve safety in the NHS by introducing more technology, there was a need to understand these systems and how people and technology work together to ensure patient safety. This has been recognised by NHSX in its Digital Clinical Safety Strategy (NHSX, 2021), which includes ‘Ensuring that rapidly evolving digital systems do not cause or contribute to adverse events’ as one of its system-wide priorities.

Weight-based prescribing in paediatrics

5.3.6 Many medication doses in paediatrics need to be based on either the child’s body weight or body surface area to achieve the most appropriate dose (Paul et al, 2011). The investigation was advised that the weight of children can vary greatly, by 200-fold differences in some cases (for example, a 500g baby versus a 100kg teenager). Therefore, an individualised approach to treatment is required, sometimes without suitable preparations of medicines.

5.3.7 Weight-based calculation has been shown to generate risks in prescribing (80% of errors) and administration (16% of errors) (Tse and Tuthill, 2020). This study highlighted that half of all prescribing errors resulted from incorrect dose selection, and technical prescribing errors occurred almost twice as often as clinical errors. This is attributed to a lack of standardised methods for the preparation and administration of medicines for children, resulting in adult formulations often being adapted for administration (Furniss et al, 2018).

5.3.8 Sutherland et al (2019) found that while dosing errors were the second most prevalent type of error in prescribing for children, the proportion of errors involving a 10-fold dose or more was uncertain. While acknowledging that there are many factors that can cause dosing errors, the problems relating to decimal points in clinical practice when calculating and measuring doses have been highlighted. Interventions over the years have not provided a solution and there has not been a systematic approach to reduce these problems in the UK (Rashed and Tomlin, 2020).

Use of ePMA systems in paediatrics

5.3.9 The investigation was told by a software manufacturer that some ePMA systems were not yet being used in paediatrics because the manufacturer wanted to ensure optimum safety in terms of the functionality before such systems were implemented. The investigation was advised by one manufacturer that it wanted to include functionality that allowed a standardised approach, preventing mis-selection of adult dosing. This would include controls for certain patient groups, including weight, different levels of alerts and permissions.

5.3.10 The investigation observed ePMA systems designed for adult patients being modified for paediatric patients. This included paediatric-only facilities, where the age and weight ranges of patients sometimes resulted in adult equivalent doses of medicines being prescribed.

5.3.11 In the reference event, trust-level configuration resulted in access to adult doses (of dalteparin) in a paediatric environment. As outlined in this investigation’s interim bulletin, published in March 2021 (Healthcare Safety Investigation Branch, 2021), the investigation identified the following safety observation from the investigation of the reference event.

5.3.12 Evidence over a number of years indicates that local configuration of ePMA systems has resulted in different outcomes in simulation testing, including adverse consequences (Metzger et al, 2010). Local configuration introduces variability and may introduce deviations in terms of a system’s functionality. A key observation made by the investigation was that all the ePMA systems observed were systems that were initially configured for prescribing for adults which were then adapted for paediatrics.

5.3.13 The investigation explored with trusts and software manufacturers the risks caused by having adult and paediatric prescribing in the same ePMA system. It was established that the risks needed to be adequately identified and mitigated with clear documentation in safety cases. Safety cases form part of a proactive safety management approach (Sujan and Habli, 2021) and are discussed in further detail in this section, from 5.3.34.

Safety functionality in ePMA systems

Use of the free-text field in ePMA systems

5.3.14 During the investigation, the usability and safety functionality of ePMA systems in paediatrics resulted in an exploration of user-testing and the implementation in different settings. One example was the use of the free-text field. The investigation was advised that the free-text box was not consistently populated in terms of content; ePMA systems ranged from having no free-text options at all to multiple free-text options. They usually had a specific focus on aspects such as additional instructions and administration information, and some of the free-text information was thought to be valuable when used within local guidance. NHSX advised the investigation that it was not feasible to recommend that ePMA systems were amended to remove free-text boxes, and that removing them may in fact have a detrimental impact.

5.3.15 It is evident that free-text fields in ePMA systems enable staff to supplement standard data capture with important information. There are however no standards in place guiding the use of these fields, resulting in varying interpretation and inconsistent use at an organisational level. This may have unintentional consequences, as occurred in the reference event. The free-text information ‘as per discussion with haematology’ was misleading when read in context of the dose prescribed.

Clinical decision support

5.3.16 It has been suggested that the clinical decision support (for example, dose range checking (DRC) built into most e-prescribing systems is mainly derived from adult dosing, which may contribute to the reported dosing errors in children (Wald et al, 2018). The investigation heard from NHSX that DRC (to help clinicians monitor the appropriateness of medication dosing) may have prevented the adult dose being selected and prescribed at the reference event Trust. However, DRC is often not individualised for each patient and there are different levels of maturity in the various software applications used by trusts.

5.3.17 During a comparison site visit to another trust, the investigation was told about the challenges of identifying and developing DRC software for paediatrics, because of the wide range of doses used for differing medical conditions. At this trust, the ePMA system used a decision support system from an external provider. However, only 30% of medications could have DRC applied and they were not the most frequently prescribed medications, for example, paracetamol. As a workaround, the ePMA team at the trust built local DRC rules for the 10 most commonly prescribed medications.

5.3.18 The investigation also considered the configuration of clinical decision support software in the context of alerting. Evidence from the visit to the reference event Trust and other comparison sites indicated that potentially fatal medication dosages could be entered (as test cases) into ePMA systems without this triggering an alert. This was due to the way in which the software had been configured by organisations and the level of alerts built into the system. Conversely, ‘alert fatigue’ could be an issue for system users, as described by trust during a comparison site visit. The trust told the investigation that they had opted to ‘switch off’ the lowest level of medication interaction alerts on their ePMA system because of this.

5.3.19 A safety recommendation was made to the Department of Health and Social Care as part of HSIB’s investigation ‘Electronic prescribing and medicines administration systems and safe discharge’ (Healthcare Safety Investigation Branch, 2019). The safety recommendation was to:

‘… consider how to prioritise the commissioning of research on human factors and clinical decision support systems; particularly in relation to the configuration of software system alerting and alert fatigue, to establish how best to maximise clinician response to high risk medication alerts.’

A study was commissioned which focused on the importance of human factors and the human-computer interaction in the system design of ePMA systems, with a plan to commission primary research on this topic in the future.

Dose rounding

5.3.20 The investigation heard that a current limitation of ePMA systems is that they cannot convert complex specific doses that have been prescribed to doses that support ease of administration. This has also been referenced in the research literature. Wald et al (2018) found that most ePMA systems do not easily generate doses required for administration in children and ‘rounding tolerances’ (rounding the dose up or down) depend on a child’s age and medical condition.

Weight-based dose-bands

5.3.21 Standard dosing bands (NHS England, 2016b) based on body weights were introduced as a way to ensure safe prescribing for children. Implementing dose-banding limits, and guidelines to support the process of administering medicines in hospitals, were found to be important in minimising medication errors and promoting medication safety (Al-Turkait and Khan, 2015).

5.3.22 The establishment of weight-based dose bands, in conjunction with dose-rounding tolerances, could be of fundamental importance in developing paediatric medicine catalogues, to aid safe and effective e-prescribing for children (Rashed and Tomlin, 2020).

Use of the calculator function

5.3.23 As mentioned in section 4.2.8, the ePMA system used by the reference event Trust had a calculator function, which was recommended by the manufacturer but not routinely used by the Trust. The decision was instead taken (after consultation with other sites using the same system in paediatrics) to build ‘dose sentences’ for medicines, to assist dosing.

5.3.24 During a site visit to another trust, the investigation observed there was an inbuilt calculator for commonly used medicines or for those that were the subject of errors. For those medicines, a recommended dose would be provided to select from, alongside the possibility of self-populating or editing the dose. The trust had optimised safety by ensuring that generally (depending on the medication), a low dose would be auto-populated for the prescriber to increase, rather than the other way round.

5.3.25 During comparison site visits to a number of trusts, the investigation observed demonstrations of ePMA system functionality in test settings. It was apparent that the configuration of their trusts’ ePMA systems may have prevented the reference event. Mitigating factors were automatic functions being built into the system to prevent the prescription of excessive doses and mandating particular ways of prescribing in paediatrics, for example using a protocol-based system.

Governance of ePMA systems

5.3.26 Information standards from various national bodies are used to strengthen and support national healthcare initiatives. These standards ‘provide the mechanism for introducing requirements to which the NHS, those with whom it commissions services, and its IT system suppliers, must conform’ (NHS Digital, 2018).

5.3.27 The two standards relating to the clinical safety of health information technology are DCB0129 and DCB0160, as outlined in ‘Clinical risk management standards’ on page 50. These standards are intended to provide clarity for users and developers with regards to the registration of standalone software as a medical device and how to comply with the standards (NHS Digital, 2018).

5.3.28 The investigation found that software (‘a set of instructions that processes input data and creates output data’ (Medicines and Healthcare products Regulatory Agency, 2021c)) could qualify as a medical device if it meets the definition, set out in the Medical Device Regulations 2002 (as amended) (Medicines and Healthcare products Regulatory Agency, 2021a).

5.3.29 The Medicines and Healthcare products Regulatory Agency (MHRA) launched a public consultation in September 2021 on the regulation of medical devices (Medicines and Healthcare products Regulatory Agency, 2021a). A broad range of regulatory issues were considered, including the updating of regulations that apply to software as a medical device.

5.3.30 Guidance on regulating medical devices in the UK was published in 2020 and updated in January 2022 (Medicines and Healthcare products Regulatory Agency, 2022a). The guidance states that since January 2021, there have been a number of changes to how medical devices are placed on the market in Great Britain. This was introduced through secondary legislation. It contains a key requirement relevant to this investigation: ‘all medical devices, including in vitro diagnostic medical devices (IVDs), custom-made devices and systems or procedure packs, need to be registered with the MHRA before they are placed on the Great Britain market’. The investigation has identified that once introduced, the legislation applied to devices already on the UK market.

5.3.31 The MHRA has advised the investigation that given the description of the functionality of the ePMA system used by the reference event Trust, the product would most likely qualify as a Class I medical device (Medicines and Healthcare products Regulatory Agency, 2021b). This is in accordance with the Medical Devices Regulations 2002 (SI 2002 No 618, as amended) (Medical Devices Regulations 2002).

5.3.32 There are requirements arising from this legislation at manufacturer and healthcare organisation levels. The manufacturer has a responsibility to ensure that the device is compliant with the Medical Devices Regulations 2002, as outlined by the Medicines and Healthcare products Regulatory Agency, (2020):

• ‘Manufacturer shall ensure that the system is designed to safety principles, taking account of the generally acknowledged state of the art.
• The manufacturer shall provide adequate information so that the device can be used safely and properly, taking account of the training and knowledge of the potential users. This shall indicate any warnings and/or precautions to take, for example, specific instructions for paediatrics.
• The manufacturer shall review experience gained from devices in the post-production phase and apply any necessary corrective actions, for example, reviewing why users do not use specific safety features.’

5.3.33 The MHRA has produced a software flowchart for the developers and manufacturers (and users) of software as a medical device to guide them in respect of what is likely to meet the definition of a medical device (Medicines and Healthcare products Regulatory Agency, 2021b).

5.3.34 The investigation found that guidance is required to provide specific information to manufacturers of ePMA systems to ensure they are compliant with the requirements under the Medical Devices Regulations 2002.

5.3.35 Healthcare organisations are responsible for ensuring that devices are deployed and used in line with manufacturers’ instructions (Medicines and Healthcare products Regulatory Agency, 2021c):

• ‘A device management policy should help to ensure that risks associated with the use of medical devices are minimised or eliminated.
• Users need to carefully consider the content of all warnings of risks and the device should only be used and maintained in line with the manufacturer’s instructions.
• Training is a key element in device safety. Ensure adequate training programmes are in place for users. Ensure such training programmes are repeated regularly, where necessary.’

5.3.36 In line with MHRA guidance, there is also a requirement to report suspected adverse reactions to medicines in children and young people under 18 years through the Yellow Card Scheme (Medicines and Healthcare products Regulatory Agency, 2022b). Yellow Cards can be used for reporting suspected adverse drug reactions to medicines. This includes suspected adverse drug reactions associated with overdose and medication. Adverse incidents involving medical devices can also be reported to the MHRA through the Yellow Card Scheme.

Clinical safety cases and the role of clinical safety officers

5.3.37 Clinical safety officers (CSO) are people in organisations who are ‘… responsible for ensuring the safety of a Health IT System … through the application of clinical risk management’ (NHS Digital, 2018). CSOs are expected to be knowledgeable about risk management and its application to clinical domains, ensuring that expected processes are followed (NHS Digital, 2018).

5.3.38 The DCB0160 implementation guidance (NHS Digital, 2018) outlines the training requirements of a CSO. These include being qualified in risk management or having an understanding of risk and safety in relation to health IT systems. The investigation was told that training is provided by NHS Digital in partnership with other bodies if staff have not acquired skills not through academic study. Training includes the consideration of the responsibilities of CSOs and expected safety standards for their organisations.

5.3.39 The investigation was told of the importance of CSOs having appropriate knowledge, skills and support to safely manage the implementation and continuing configuration of ePMA systems in trusts. However, while the aforementioned training from NHS Digital is available, the investigation heard that uptake is limited. The NHS Digital training is not mandated and training is available from third-party suppliers.

5.3.40 In 2021, NHS Digital trained 800 people, double the number trained in 2019. The investigation heard that requests for training are increasing and are anticipated to rise further following publication of the Digital Clinical Safety Strategy (NHSX, 2021). NHSX and NHS Digital are looking to deploy a ‘train the trainer’ system over the next 12 to 18 months with the responsibility for training sitting within integrated care systems (partnerships between health and care organisations that co-ordinate and plan services for a specific geographical area). NHS Digital informed the investigation that this decision was made based on the NHS not accepting responsibility and ownership for digital safety and the requirements of the DCB standards.

5.3.41 The investigation was also advised of the importance of providing training for staff who are responsible for the day-to-day management of ePMA systems. However, it was heard that few of these staff had accessed the first level of clinical safety risk management training.

5.3.42 A core responsibility of CSOs is to approve clinical safety case reports (NHS Digital, 2018). These are defined as:

‘A report that presents the arguments and supporting evidence that provides a compelling, comprehensible and valid case that a system is safe for a given application in a given environment at a defined point in a Health IT System’s lifecycle.’

They are seen as integral to the governance of ePMA systems in healthcare.

5.3.43 Essentially, safety cases provide a transparent, evidence-based argument for why a system is safe for use in a particular environment/situation (Sujan and Habli, 2021). One of the commitments of the Digital Clinical Safety Strategy (NHSX, 2021) is to:

‘Ensure the transparent nature of documentation used to build a safety case for existing clinical safety standards. This will be led by NHSD [NHS Digital] and industry partners on an ongoing basis.’

5.3.44 The investigation was given the opportunity to review the safety case for a previous release of the ePMA system used in the reference event. This described efforts to ‘identify, assess and, where appropriate, mitigate hazards associated with this product’ (anonymous source). It also stated that if used as specified, the ‘release did not include any clinical hazards or functionality which could give risk to unacceptable clinical risk to patients’.

5.3.45 The safety case acknowledged the role of local configuration and stated an expectation for the customer to conduct comprehensive user acceptance testing and mitigation of any local hazards through configuration. For example, testing of the system needs to ensure lists of prescribable medications are presented to each local user, depending on their role (such as paediatrics).

5.3.46 The investigation heard that a manufacturer’s safety case is specific to the version of the ePMA software released, based on the scope of the release. Items in that release that may pose safety risks at local implementation are described. The investigation was advised by a manufacturer that it is an expectation that clients will review their safety cases. Further, the decision whether to test according to the advice the manufacturer provides, is at clients’ discretion.

5.3.47 During site visits, the investigation asked those involved in the implementation of ePMA systems whether they had safety cases in line with DCB160. One trust only had its original safety case for adult prescribing from the software manufacturer for the system, and not a local safety case for modifications or roll-out to paediatrics. Several trusts referred to the existence of a risk register or risk logs, by way of a response to the investigation and they questioned how widely the safety case requirements had been disseminated.

5.3.48 At a national level, when looking at a new system, a safety case/hazard log is part of the procurement/implementation process and is accessed from the manufacturer. NHS Digital advised the investigation that the impact of the statutory requirements behind the standards for health IT were that a system should not be permitted to roll out more widely until the safety case/hazard log had been reviewed.

5.3.49 The investigation found evidence of safety cases not being available or not being used in line with the expectations of DCB0160. There was also found to be limited awareness of the need for or role of safety cases. It was apparent that national work was needed to support organisations to understand the requirements and their responsibilities.

5.3.50 Safety cases are commonly used in other safety-critical industries, such as the nuclear industry (Sujan and Habli, 2021). Therefore, the investigation approached the Office of Nuclear Regulation (ONR) to understand how the nuclear industry uses safety cases (see learning case study 2). The learning from ONR aids understanding in respect of how to structure a safety case, the evidence needed, the role of testing, and the need for a comprehensive assessment of safety, where equipment has been further adapted. In healthcare, a comparison could be made with a situation where an adult ePMA system is adapted for paediatric use and the need to test, assess, and evidence the safety of the adaptations before activation.

5.3.51 The issue of safety cases has been highlighted by HSIB in its report on the ‘Procurement, usability and adoption of ‘smart’ infusion pumps’ (Healthcare Safety Investigation Branch, 2020b). Following on from this, the investigation has identified the need to further establish the role and importance of safety cases in healthcare. The aim of this is to improve the availability and quality of safety cases, and support CSOs (and other staff involved in the local configuration of ePMA systems) to undertake their roles with appropriate training and support.

5.3.52 In support of this safety recommendation, the Care Quality Commission (CQC) has agreed to review whether a provider’s assurance of their compliance with the standard specific to ePMA systems in healthcare (including the authoring of a DCB 0160), can form part of the CQC’s developing regulatory model in the future.

Te investigation found that:

Multidisciplinary co-ordination and decision-making

• Email discussions may be being used for discussion of critical decisions in relation to patient care, with limited dissemination to the wider team.
• There is limited standardisation of handovers, ward rounds (visits to each patient in a ward to review and discuss their care) and huddles (short, focused staff briefings), in terms of which members of the multidisciplinary team are involved, and how they are conducted for maximum effectiveness.
• There is limited national guidance on the management of ward rounds in paediatrics (as is available for adult care).
• There is variability in hospital clinical pharmacy provision. This reduces the availability of ward-based pharmacists and may result in a dispensary-based service only at weekends.
• Nursing staff perceived themselves to be the final barrier to prevent an incorrect dose prior to the administration of medication, resulting in them feeling accountable for the error.

Factors undermining the effectiveness of checking as a barrier

• Multiple cues influenced whether staff considered medication doses to be correct.
• Processes for the checking of medicines varied without evidence of what constituted the most effective process.
• The distinction between verification and checking was not explicit, that is, checking that the prepared medication correlated with the prescription versus verification that the prescription was appropriate against a standard.
• The environments within which staff prepared and checked medicines influenced their performance.
• Environmental layouts and limited resource resulted in workarounds.

Implementation of ‘off-the-shelf’ electronic prescribing systems in specific contexts

• There are no standards for what safety-critical functionality should be available in ePMA systems configured for use in paediatrics (for example, the use of weight-based dose bands, where individually calculated doses are rounded to a set of predefined doses).
• The use of free-text comment boxes in ePMA systems is not specified or standardised.
• The usability and functionality of ePMA systems need to be assessed through user-testing across a range of different settings.
• Local configuration of ePMA systems potentially introduces variability and risks if not undertaken with clear understanding of the potential hazards and their mitigations.
• Software could qualify as a medical device if it meets the definition, set out in the Medical Device Regulations 2002 (as amended).
• the investigation found that:stems is limited with evidence of gaps in training and an absence of safety cases (reports that provide a transparent, evidence-based argument for why a system is safe for use in a particular setting).

HSIB’s safety recommendations are directed to a specific organisation for action. They are based on information derived from the investigation or other sources, such as safety studies, and are made with the intention of preventing future, similar events.

The HSIB investigation focused on errors in the prescription of weight-based medication for children, in the context of electronic prescribing and medicines administration (ePMA) systems. The responsibility for ensuring the safety of medicines prescribing and administration is with individual trusts, with input from NHSX, NHS Digital and the Medicines and Healthcare products Regulatory Agency (MHRA). Therefore, safety recommendations made in this respect are directed towards those organisations. Different parts of the healthcare system have also been identified to address the other risks highlighted in this investigation relating to paediatric ward rounds and second-checking processes. Safety recommendations have been directed to them accordingly.

SEIPS is a systems engineering approach with human factors principles embedded within it (figure A1). SEIPS describes how components of the work system produce work processes which result in different outcomes. Work system factors are described below (Holden et al, 2013; Carayon et al, 2006):

• person(s): the people working in the particular system and the patient
• tasks: undertaken by the persons which may vary in complexity or variety
• tools and technology: used to undertake the tasks which may vary in usability and functionality
• internal environment: the physical space around the persons, for example layout, noise, and temperature
• organisation: conditions external to the persons to support the organisation of, for example, resources and activity
• external environment: factors outside of the healthcare institution that might include policy, societal or economic factors.

Processes can be physical, cognitive, or behavioural and lead to outcomes for the patients, professionals, or healthcare institutions. The interactions between the various components of the work system lead to different outcomes, both positive and negative. The framework includes feedback loops which represent the adjustments systems make over time.

FRAM aims to reflect risks within complex systems. It does this through describing variability in the functions within the system and looks to model what is needed for everyday performance to go right.

The FRAM involves exploring ‘work as done’ with frontline staff to identify the ‘functions’ that are being performed. A function is defined as ‘the activities – or set of activities – that are required to produce a certain outcome’ (Hollnagel, 2012). In doing this, FRAM develops a model of the core functions to illustrate how each links, how variability might occur, and how this may affect outcomes. To achieve this, links are created between functions by identifying six specific aspects of each function: input, output, preconditions, resources, controls, and time factors.

Identified system functions were entered into the FRAM Model Visualiser software (FMV).