Investigation report: Electronic prescribing and medicines administration systems and safe discharge

1.1.1 A definition of ePMA is: ‘The utilisation of electronic systems to facilitate and enhance the communication of a prescription or medicine order, aiding the choice, administration and supply of a medicine through knowledge and decision support and providing a robust audit trail for the entire medicines use process.’ (NHS Connecting for Health, 2009).

1.1.2 In practical terms, an ePMA system supports the safe, effective, and cost-effective use of medicines from a patient’s admission to hospital until their discharge. Figure 1 illustrates a well-defined, properly configured ePMA system, including ‘closed loop medication management’ (CLMM). A CLMM system is a fully electronic medication management process that integrates automated and intelligent systems to manage the process of prescription, dispensing and administration of medicines, and seamlessly document all the relevant information.

1.1.3 In relation to terminology used in this report, see Appendix 1 for a detailed glossary. In general terms, throughout the report, electronic prescribing is referred to in the context of being an integral part of ePMA in the United Kingdom; in the United States it is referred to as a ‘computerised provider order entry’ (CPOE) system.

1.2.1 A guideline on the safe and effective use of medicines in health and social care has been issued (National Institute for Health and Care Excellence, 2015). The guideline aims to ensure that medicines provide the greatest possible benefit to people by encouraging medicines reconciliation (see 1.2.2), medication review(4), and the use of patient decision aids.

1.2.2 Medicines reconciliation, as defined by the Institute for Healthcare Improvement, is the process of identifying an accurate list of a person’s current medicines and comparing them with the list currently in use, recognising any discrepancies and documenting any changes. This process results in a complete list of medicines that can be accurately communicated.

1.2.3 The medicines reconciliation process should be undertaken when a patient moves between care settings and will vary depending on the care setting that the person is moving from and to, for example, from primary care into hospital. In an acute setting, a patient’s medicines should be accurately listed, and medicines reconciliation should be carried out within 24 hours, or sooner if clinically necessary (National Institute for Health and Care Excellence, 2015). Medicines reconciliation may need to be carried out more than once during a patient’s hospital stay, for example, when the patient is admitted, transferred between wards or discharged.

1.2.4 In relation to primary care, the National Institute for Health and Care Excellence (2015) states there is a need to ‘ensure that systems are in place for people discharged from a care setting to have a reconciled list of their medicines in their GP record within one week of the GP practice receiving the information, and before a prescription or new supply of medicines is issued’.

1.2.5 In all care settings, organisations should ensure that a designated health professional has overall organisational responsibility for the medicines reconciliation process. The process should be determined locally and should include:

• organisational responsibilities
• responsibilities of health and social care practitioners involved in the process
• individual training and competency needs.

1.2.6 Organisations should ensure that medicines reconciliation is carried out by a trained and competent health professional, ideally a pharmacist, pharmacy technician, nurse or doctor.

They will have the necessary knowledge, skills and expertise including:

• effective communication skills
• technical knowledge of processes for managing medicines
• therapeutic knowledge of medicines use.

1.2.7 Patients and their family members or carers should be involved, whenever possible, in the medicines reconciliation process

_{(4) A structured medication review is a critical examination of a person’s medicines (https://www.nice.org.uk/guidance/QS120/chapter/Quality-statement-6-Structured-medication-review).}

1.3.1 Medicines optimisation is defined as ‘a person-centred approach to safe and effective medicines use, to ensure people obtain the best possible outcomes from their medicines’ (National Institute for Health and Care Excellence, 2016).

1.3.2 Medicines optimisation looks at the value that medicines deliver, making sure they are clinically effective and cost effective. It is about ensuring people get the right choice of medicines, at the right time, and are engaged in the process by their clinical team (NHS England, 2019).

1.3.3 The Royal Pharmaceutical Society (2013) refers to four principles of medicines optimisation (Figure 2).

1.4.1 Anticoagulants, commonly referred to as blood thinners, are medicines that prevent or reduce coagulation of blood, prolonging the time it takes for blood to clot. There are three main groups of anticoagulants: low molecular weight heparin (LMWH), warfarin and direct oral anticoagulants (DOAC).

1.4.2 Two of the medications discussed within this investigation are apixaban, which is a DOAC in tablet form and dalteparin, which is a LMWH and given by subcutaneous injection(5). There is currently no reliable way of reversing the effects of apixaban once it has been administered. For dalteparin, protamine sulphate can be given if there is evidence of haemorrhage (bleeding) or if very large amounts of the medication have been injected, but anticoagulation will not be completely reversed (National Poisons Information Service, 2019). This is relevant for patients who are over-anticoagulated, as occurred in the reference event.

1.4.3 In 2014, the National Institute for Health and Care Excellence issued a guideline on the management of atrial fibrillation (National Institute for Health and Care Excellence, 2014). This guideline advocated the use of anticoagulant medication for the treatment of atrial fibrillation (AF) to prevent embolic stroke(6).

_{(5) An injection into the layer of fat under the skin.}

_{(6) A stroke that occurs when a blood clot that has formed elsewhere in the body travels to the brain via the bloodstream and blocks the flow of blood to the brain}

1.5.1 There were several drug interactions noted in the reference event, the key ones being:

• The administration of the oral anticoagulant, apixaban, which the patient administered herself in addition to dalteparin. Taken together, these drugs can increase the risk of bleeding; this is graded as ‘moderate’ (amber) in the British National Formulary (BNF)(7) .
• The prescribing of amiodarone with crizotinib (anti-cancer drug). Amiodarone is an antiarrhythmic medication, used to treat and prevent a number of types of irregular heartbeat. The administration of both crizotinib and amiodarone increases the risk of bradycardia (low heart rate), with the severity of interaction graded as ‘severe’ (red) in the BNF(8).
• The prescribing of an increased dosage of bisoprolol (to treat high blood pressure) with crizotinib. The combined medicines increase the risk of bradycardia(9), with the severity of interaction graded as ‘moderate’ (amber) in the BNF.

_{(7) ‘Moderate—the result could cause considerable distress or partially incapacitate a patient; they are unlikely to be life-threatening or result in long-term effects’. Accessed via: https://bnf.nice.org.uk/interaction/dalteparin.html#bnf_i1567199228225}

_{(8) ‘Severe—the result may be a life-threatening event or have a permanent detrimental effect’. Accessed via: https://bnf.nice.org.uk/interaction/crizotinib-2.html#bnf_i1567199226666}

_{(9) https://bnf.nice.org.uk/interaction/crizotinib-2.html#bnf_i1567199226697}

1.6.1 The Department is referred to as the Department of Health until January 2018, and the Department of Health and Social Care from January 2018 onwards.

1.6.2 NHS England and NHS Improvement are referred to in the report as separate entities until April 2019, and jointly as NHS England and NHS Improvement post April 2019.

1.6.3 There may be other organisations which are referred to following publication of this report that have subsequently changed their name; this report will not be updated.

3.1.1 The incident relating to the simultaneous prescription of two anticoagulation medications was identified by HSIB and a scoping investigation was initiated to determine the learning potential of a national investigation.

3.2.1 Following an initial scoping investigation, the chief investigator authorised a full investigation as the incident met HSIB’s criteria:

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

The reference event highlighted some of the risks associated with prescribing medicines for patients during a stay in hospital and on discharge, particularly when patients are taking unusual medication treatments at home prior to admission. Such risks are exacerbated when patients move between wards and departments. In addition, support services may not be available seven days a week (NHS England, 2017) to perform tasks such as medicines reconciliation.

According to data published by NHS Improvement on 25 September 2019 , 2,036,657 incidents were reported to the National Reporting and Learning System (NRLS) as occurring in England between April 2018 and March 2019. Of these, 216,154 were classified as medication incidents, with approximately 1% of the medication incidents reported as resulting in moderate harm, severe harm or death.

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

Medicines are prescribed across primary and secondary care using a mixture of electronic and paper systems, but with an increasing reliance on electronic systems.

The widespread use of anticoagulants to treat atrial fibrillation (AF) (Fitzpatrick et al, 2017) is supported by National Institute for Health and Care Excellence guideline CG180 (National Institute for Health and Care Excellence, 2014).

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

There were opportunities for HSIB to share learning on a national level to positively influence processes and practices to reduce prescribing and administrative errors when using ePMA systems, in the context of improving the safe discharge of patients from secondary to primary care.

3.3.1 The investigation interviewed frontline staff and managers within the Trust where the reference event occurred. The investigation reviewed Ann’s clinical records, and Trust policies and national guidance, to understand what actions were taken and how these aligned with both the Trusts’ and national expectations. The investigation identified factors relevant to Ann’s care and treatment.

3.3.2 The findings of this report have been supplemented by insights from subject matter advisors (SMAs), academics and academic literature, in addition to expertise provided by colleagues in national organisations.

_{(19) Accessed via: https://improvement.nhs.uk/resources/national-patient-safety-incident-reports-25-september-2019/}

_{(20) The National Reporting and Learning System (NRLS) is a primarily voluntary confidential reporting system for patient safety incidents. These can be reported by anyone, with most reported by NHS organisations across England and Wales via their local risk management systems. As the system is mainly voluntary, it is not a reflection of incidence, but can provide insight into the types of risks and incidents reported by NHS organisations, as well as patient safety reporting culture.}

The AcciMap model was used to analyse the reference event information and supports the direction of the national investigation. The AcciMap method (Svedung and Rasmussen, 2002) is an incident analysis method that identifies factors within the system that influenced the occurrence of an incident.

The analysis focuses on identifying relationships between the different levels of the system (Figure 3), which include government policy and budgeting; regulatory bodies and associations; local area management; physical processes and actor activities (what staff, people, organisations, systems did); and equipment and surroundings (Stanton et al, 2013).

HSIB has identified a significant safety risk posed by the communication and transfer of information between secondary care, primary care and community pharmacy relating to medicines at the time of hospital discharge. A reference event was identified that resulted in a patient inadvertently receiving two anticoagulant medications at the same time, possibly causing an episode of gastrointestinal (digestive tract) bleeding. Increasingly, healthcare facilities in primary and secondary care are introducing digital solutions (electronic prescribing and medicines administration (ePMA) systems) to improve medicines safety. However, analysis of the reference event identified how ePMA systems can create their own risks – risks that will need to be addressed as these systems become more widespread. Other risk factors relating to prescribing and the discharge of the patient, including medicines reconciliation(21), availability of pharmacy services and weekend working, were identified during the investigation.

_{(21) Medicines reconciliation, as defined by the Institute for Healthcare Improvement, is the process of identifying an accurate list of a person’s current medicines and comparing them with the list currently in use, recognising any discrepancies and documenting any changes.}

National and local context

4.1.1 The investigation was advised by NHS England that, according to data from 2017, not all healthcare services were using ePMA systems and not all functionality was being used, even in the most digitally mature trusts. However, it is expected that the position has improved since then, particularly among the Global Digital Exemplar(22) sites with ePMA systems, and acknowledging that such systems require continuous optimisation beyond the initial baseline implementation.

_{(22) NHS England defines a Global Digital Exemplar as ‘an internationally recognised NHS provider delivering improvements in the quality of care, through the world-class use of digital technologies and information’}

4.1.2 In the Trust where the reference event occurred, an electronic health record (EHR) system had been in use since 2012. The ePMA elements of the system were chosen as they were an add-on module to the existing EHR system.

4.1.3 The Trust implemented the ePMA system in November 2017 using a ‘big bang’ approach (introducing it throughout the Trust at the same time), an approach that the investigation was advised had only been taken by three trusts in England. The remainder of trusts implemented ePMA systems via a phased implementation, at increasing rates as experience grows, adopting an incremental roll-out approach to a smaller number of services.

4.1.4 Prior to implementation of the ePMA system, the Trust drafted a deployment plan which stated that if the availability of resources (staff and finance) could fully support a big bang approach, then this would be the recommended roll-out method to minimise clinical risk. Ultimately, the roll-out resulted in not all services using the electronic system and not all of the system’s functions being used. One of the vendors for an ePMA system advised the investigation that implementation would be expected to take 12 months to two years. This highlights the relative infancy of the system at the time of the reference event.

4.1.5 It was not possible to see Ann’s stay in hospital end to end using the Trust’s EHR because of the way the system was configured. Instead, the admission was represented as individual workflows (processes), accessed via different screens in the system. This meant that each workflow had to be analysed separately to determine the sequence of events and error points.

4.1.6 Electronic reconciliation of medicines was enabled on the Trust’s ePMA system from the outset and so was available at the time of the reference event. However, this function was primarily used only by pharmacy staff (pharmacists and medicine management technicians). Clinicians were not recording medicines on admission on the ePMA system at this time. Medicines reconciliation on discharge was also enabled as a function on the ePMA system from its introduction, and was carried out by pharmacists at the point of checking the ‘To Take Away’ (TTA) medication. However, its functionality was limited where no history of prior medication was completed, and therefore the medicines information was not included in patient discharge letters.

4.1.7 The Trust’s electronic pharmacy stock-control system was separate from its EHR system. This adds to the complexity of systems and makes interoperability more challenging, although it did not impact on the reference event.

4.1.8 The drug chart, TTA medicines and the discharge summary were separate tasks within the ePMA system. The TTA process was as follows:

1. The TTA is created and signed electronically by a doctor.
2. A pop-up window prompts the prescriber to notify pharmacy to dispense the TTA.
3. Once notified, a task is added to the pharmacy patient task list prompting a pharmacist to complete the TTA prescription screen in the ePMA system.
4. Once the screening task is complete, a task is created on the dispensing tasks.
5. The task is completed when the TTA has been dispensed and checked.
6. The TTA is then ready for delivery to the ward.

4.1.9 The ePMA TTA process was for dispensing purposes only and this information did not interface with the discharge summary. At the time of the reference event, the discharge summary was created as a clinical note within the EHR; this contained sections for medications to be added to the ‘summary’ note and the TTA medications could be selected for inclusion in the note. However, this was a manual process and relied on the doctor selecting the correct medications (that is the TTAs, not inpatient prescriptions).

Analysis

4.1.10 It cannot be determined with certainty why medicines reconciliation did not occur in the reference event. However, the Trust has advised that the process is principally carried out by pharmacy staff (pharmacists and technicians) and there was no provision for this as a funded service on Saturdays and Sundays. This limited the number of patients who could receive a full medicines reconciliation over the weekend, the majority needing to wait until a weekday.

4.1.11 There was limited pharmacist cover (for two hours) on the Medical Assessment Unit (MAU) on Saturday, but Ann’s medicines reconciliation was not undertaken during this time. At the weekend, junior doctors on wards were expected to input the medications on to the ePMA system, but again this did not occur, for reasons that could not be determined by the investigation. Aside from the omission, medicines reconciliation relies on staff entering data into the system. In addition, the process was hindered by interoperability issues, as the Trust’s ePMA system was not linked to the electronic systems used in primary care.

4.1.12 The Foundation Year Two (FY2) trainee entered the prescription for the tablet form of anticoagulant (apixaban) on the ePMA system. Immediately prior to this, the anticoagulant injection (dalteparin) was discontinued by the FY2 trainee. The FY2 trainee completed the ePMA TTA medicines section at 08:34 hours so that pharmacy could prepare the medicines and Ann could be discharged.

4.1.13 Ann’s medication history had not been put into the ePMA system, so when the TTA was generated from the ‘Discharge reconciliation’ screen it did not highlight on the drug chart whether any medication had been discontinued. If the medication history had been added to the ePMA, the history would have been visible on the screen and the medications documented as either continued or stopped

4.1.14 A care plan function was available on the ePMA system to provide detailed individual medication information. On Ann’s record there was a care plan for dalteparin; however, the existing care plans for dalteparin had not been fully integrated into the system at the time of the reference event. There was no care plan for apixaban. The investigation was advised that the use of a care plan would guide the prescriber to the correct drug/ dose for a specific condition and prompt the entry of information, which then could be included in the discharge letter. In the case of apixaban, the key piece of information would be the indication for commencing the medication, and this was required information on the TTA, both then and now.

4.1.15 The investigation was given a demonstration of the Trust’s ePMA system, by the lead pharmacist for electronic prescribing and medicines administration, highlighting how a patient’s existing medications could be recorded on the system. It was considered by the investigation that the prescription systems used in primary and secondary care should allow for or enable the importing and exporting of this data

Issues around prescribing, communication, and documentation of take-home medications, leading to over-anticoagulation.

Fig 4 Review by a consultant physician subject matter advisor

National and local context

4.2.1 The investigation was informed by several sources that the concurrent use of paper and electronic systems increases clinical risk. There is an increased risk of error in any transfer of care that requires transcription of medication lists, in this context from paper to ePMA. Therefore, minimising transcribing is the optimal approach, with all services using electronic systems.

4.2.2 The way that the ePMA system was designed, implemented and used at the time of the reference event meant that the electronic system was not implemented across all services. The paediatrics and critical care departments were still using paper records and the emergency department (ED) was using paper healthcare records, alongside ePMA for prescribing medicines administered in the department.

Analysis

4.2.3 When Ann went into hospital, she took a yellow folder containing her previous clinical notes and a basket containing the medications she was self-administering prior to admission (Figure 5).

Fig 5 Ann’s basket of medicines

4.2.4 A record of Ann’s medicines was written in the ED paper records by the clerking (admitting) doctor. The ED notes were scanned digitally but also transferred as a hard copy with the patient to the wards when admitted, as per the Trust’s normal process at the time.

4.2.5 The ED used the ePMA system for prescribing new medicines and prescribing could be viewed as the patient moved to other areas within the Trust. However, rather than all of the admission tasks being undertaken by the ED doctor, the clerking doctor in the reference event Trust was not expected to enter the patient’s existing medicines on to the ePMA system, until they had been admitted to a ward. The existence of the ED paper record of Ann’s drug history resulted in the information becoming dislocated from the electronic system used in other areas of the Trust.

Ann’s medicines needed to be entered on the ePMA system so that the information could be correlated and for adverse drug interaction checks to be undertaken.

HSIB makes the following safety observation

4.2.6 In addition to the anticoagulant medication issues already outlined, discrepancies were identified in relation to other medicines that Ann was prescribed.

4.2.7 Firstly, the investigation established that not all the medications Ann was selfadministering at home were recorded in the paper ED records. The General Practice Summary, from the Health Centre indicates that Ann was prescribed salbutamol 100 micrograms/dose and amoxycillin 500 mg capsules (antibiotics) as a seven-day course, which should have been ongoing throughout her hospital stay. It is not known for certain why these two medicines were not recorded in the ED records. However, the family advised the investigation that Ann rarely administered the salbutamol, so it is likely she left this at home, and possibly the antibiotics too, since they did not appear to be in her basket of medicines (Figure 5).

4.2.8 Secondly, several discrepancies were identified regarding the prescription of medications following Ann’s admission to hospital:

1. Amiodarone was prescribed during the ward round on the cardiology ward and Ann was given two doses before the medication was stopped by the FY2 trainee (following contact by the dispensing pharmacist). This was due to the contraindication with crizotinib (see 1.5.1).
2. The FY2 trainee prescribed 2.5mg apixaban to be administered twice a day, which was subsequently amended to 5mg twice a day by the dispensing pharmacist. The reason for this decision was that Ann did not have any of the risk factors necessitating a lower dose; however, increasing the dose increased the risk of bleeding.
3. The drug dosage of bisoprolol was increased in the discharge summary by the same FY2 trainee. The bisoprolol dose was modified on the TTA at 13:43 hours, but, following this, no notification was sent to alert the pharmacy of the amendment.
4. Not all the medicines that Ann was taking on admission and noted on to the ED paper record, were entered on the ePMA system, one example being crizotinib. Ann self-administered this drug from her own supply since the Trust’s medicine code for oral chemotherapy drugs specified that only senior medical oncology staff could prescribe following review. Another example is pregabalin, which the investigation was unable to find out why it was not entered on the ePMA system, although it was issued as a TTA.

4.2.9 Thirdly, the evidence gathered by the investigation indicated that Ann was self-administering at least one medicine from her own supply of medications during her inpatient stay, in addition to being administered other medicines by the hospital. It was not clear to the investigation whether Ann’s ability to self-administer was assessed and documented. A subject matter advisor (SMA) has advised that a patient should be able to self-administer their own medicines whilst in hospital, unless they are at risk, and providing the appropriate assessments have been undertaken and documented.

4.2.10 The investigation was advised that self-administration may lead to a dual system, which is difficult to maintain in an acute hospital, particularly in respect of ensuring self-administration matches the prescribed record. Unless a discontinued medication is removed from a patient’s own supply, there is always the risk that they may be unaware that a prescription has been amended and/or discontinued and replaced with one for another drug. Unlike in the home environment where drug administration is consistent, treatment in hospital is unstable and subject to adjustments, so potentially confusing for the patient. This is compounded by the fact that patients have limited access to electronic records, compared with paper systems such as the chart at the end of the bed.

National and local context

4.3.1 The design, implementation and use of the ePMA system used at the time of the incident meant that only approximately 80% of functionality had been rolled out, in terms of processes enabled. One example of functionality that was not enabled across the Trust was the specific discharge process(23). This process was considered by the reference event Trust to be the ‘gold standard’ in terms of discharge, and it included an electronic counselling checklist for the prescription of high-risk medications, including anticoagulants. However, the discharge process was only used in one division with in the Trust at the time of the event. The Trust considered that using this discharge process would have improved prescribing and standardised information across secondary and primary care.

4.3.2 Using a standardised discharge process, that includes a discharge summary that can interface with ePMA, provides a synopsis of a patient’s medication on discharge, including standardised sections detailing what medication was stopped, what new medication was started and what medication was unchanged (Figure 6). The Trust accelerated the roll-out of this functionality following the reference event.

_{(23) The specific discharge process is a module within the electronic health record enabling a robust discharge process from hospital, which includes medication reconciliation.}

4.3.3 The Standards for medicines management, published by the Nursing and Midwifery Council (2010), stated the registrant was responsible for ensuring that on discharge, the patient is ‘aware of any changes to their medication, that is, new medicine, dose, brand, route’. The investigation could not verify that this was undertaken during the reference event.

4.3.4 Whilst the Medicines Management Standard (Nursing and Midwifery Council, 2010) was the national guidance in place at the time of the reference event in 2018, it has now been replaced with guidance from the Royal Pharmaceutical Society and the Royal College of Nursing (2019). However, the new guidance does not specifically mention nursing responsibility for handing TTAs over to the patient to administer out of hospital.

4.3.5 The Trust informed the investigation that whilst there is a discharge process that is followed, and was followed at the time of the event, the process did not form part of a Trust policy. The key aspects of the procedure which the Trust used when discharging patients from the ward are as follows:

• ‘Discharge medications received on the ward and then checked against the drug chart on EPMA to ensure accuracy and all present.
• On [the clinical application] within the xxx system print off the discharge medication sheet found on the top toolbar - discharge medication. This can then be electronically signed to say the medications have been given to the patient after the process below.
• Discharge medication then taken to the patient and checked with them.

This involves:

1. Checking all the patient identifier detail is correct.
2. That any new medications are explained as to their use and any potential side-effects discussed.
3. Discussing any issues or concerns the patient may have are addressed and allayed’.
4. If the discharge medication includes any NOAC/DOAC the appropriate counselling form is completed and given to the patient with a copy in the notes
5. Document in the patient MDT notes actions taken.’

Analysis

4.3.6 The discharge summary was completed by the FY2 trainee at 08:49 hours, following the ward round. It included the following medications: apixaban 2.5mg oral tablet twice a day, lansoprazole 30mg once a day, bisoprolol 2.5mg (dose increased to 3.75mg later that day) once a day, and pregabalin 75mg once a day. The discharge summary therefore consisted of one new medication (apixaban) and three existing ones that were continuing (lansoprazole, bisoprolol and pregabalin). It made no reference to the dalteparin being discontinued.

4.3.7 If Ann’s medicines had been entered on the ePMA system as part of a medication history, it would have been apparent when prescribing the medications for discharge that dalteparin had been discontinued and was not on the list of currently prescribed inpatient medications. This would be the case even without the use of the specific discharge process, as the mechanism and screens within which the TTA is composed are the same. The benefit of the discharge process is that it ensures the discharge summary letter contains the most accurate and up to date information when the patient is discharged.

4.3.8 The record on the dispensing system indicates that apixaban was dispensed at 10:26 hours, with an A5-sized patient alert card (from the manufacturer), for Ann to retain and carry with her at all times24. It is documented that a bottle of 56 tablets was dispensed and transferred to the ward in a sealable plastic bag.

4.3.9 The FY2 trainee stated it was normal practice (although not verified), to put only new medications on the discharge summary and not the medicines that had been stopped or were continuing. This practice resulted in the discharge summary and communication being unclear in respect of what medications had been stopped in hospital; it was not explicit in either the discharge summary or medical records that the apixaban had been prescribed to replace dalteparin.

_{(24) Apixaban patient alert card: https://www.medicines.org.uk/emc/rmm/112/Document}

HSIB makes the following safety observation:

4.3.10 Ann was discharged home on day three at 15:11 hours. An entry in Ann’s medical records, made at 14:00 hours by the discharging nurse, notes that medication was given as prescribed. However, there is no further information, other than the family’s recollection of events and the knowledge that Ann was discharged with a supply of apixaban from the hospital pharmacy, in addition to the other three TTAs (as outlined in 4.3.6).

4.3.11 The nurse who discharged Ann at 14:00 hours was working a day shift. The staffing levels on the cardiac ward should be six registered nurses and three healthcare assistants. On the Sunday of Ann’s discharge there were four registered nurses, four healthcare assistants and one agency nurse who was a registered mental healthcare nurse. This meant that there was a shortfall of one registered nurse, although there is no evidence to suggest that staffing levels contributed to the occurrence of the incident.

National and local context

4.4.1 There was a checklist to be completed when counselling a patient on taking a direct oral anticoagulant (DOAC), which was available on the Trust’s intranet. It was not used in the reference event and was considered to be a critical omission. The checklist includes information about anticoagulation, the patient’s drug, taking the medication, consulting the general practitioner (GP) and other advice. The intention was for this to be completed by the healthcare professional providing counselling and signed by the patient/guardian. There was a note to the healthcare professional at the top of the checklist advising that if the patient had previously been on an alternative anticoagulant which had been or would be stopped at a certain point, then the patient/ carer should be made aware of the change.

4.4.2 The investigation received version 2.0 of the checklist, dated May 2018, which was after the reference event. The investigation was not provided with the version of the checklist that was in use at the time of the reference event.

Analysis

4.4.3 The consultant cardiologist recalled seeing Ann for the first time during a ward round at 08:10 hours on day three, for a consultation lasting five to 10 minutes. He told the investigation which bay Ann was in, how she was sitting and that she appeared to be “buoyant”. He talked to Ann about her palpitations and the cardiac arrythmia (abnormal heart rhythm), as the consultant cardiologist noted she had been ‘admitted with atrial fibrillation’.

4.4.4 The consultant changed Ann’s anticoagulant medication from dalteparin to apixaban to ease the self-administration requirements. The only reference to the change of medication in Ann’s notes was ‘apixaban 2.5mg BD’. The consultant cardiologist told the investigation there was one thing Ann mentioned to him, which was: “if I could stop these injections, that would be great”. The consultant cardiologist said he responded in good faith by saying: “I can do something about that”, believing he could improve her comfort by prescribing an oral anticoagulant. He also acknowledged that Ann had a life-limiting condition and so wanted to get her home as soon as possible.

4.4.5 Ann’s family members were not present at the time of the ward round. However, their perception was that Ann did not have an issue with administering dalteparin every day and was accepting of the injections. Ann’s daughter told the investigation that Ann would administer the injection every night before she went to bed. She remembered asking her mother whether it was painful, and Ann’s response was: “No, it’s just something I’ve got to do, and I get on with it.”

4.4.6 The consultant cardiologist’s recollection of his conversation with Ann regarding the dalteparin injections was that she expressed frustration with having to inject the dalteparin, and he considered that he could prescribe an oral alternative. However, the documentation in the medical notes and discharge summary did not fully reflect the discussion between the consultant cardiologist and Ann in relation to this.

4.4.7 The consultant cardiologist and FY2 trainee informed the investigation that they were not aware of the Trust’s counselling checklist for DOACs. The investigation did not establish how widely used the counselling checklist was within the Trust.

HSIB makes the following safety observation:

4.4.8 The consultant cardiologist informed the investigation of his usual process on ward rounds. He reviews the medical records, investigations undertaken and test results. The consultant cardiologist then talks to the patient whilst the more junior doctor is recording in the medical records at the bedside. Decisions are made and the more junior doctor follows up any subsequent actions afterwards. He said that simple tasks such as discontinuing medications might be undertaken immediately, using the nearest computer, whilst tasks such as writing up TTAs would be undertaken after the ward round.

4.4.9 At 14:00 hours on day three of the reference event, the following was documented in Ann’s medical records: ‘…Patient gone home today, husband collect her. Discharge letter and medication given and explained all. Contact number’.

4.4.10 The family informed the investigation that around the time of her discharge, Ann was in the hospital chapel with her husband, whilst her daughter remained on the ward to receive the discharge documentation and a bag of medications. Ann’s daughter did not recall the nurse opening the bag and going through the contents. She recollects being told that in the bag was apixaban, bisoprolol, pregabalin and lansoprazole, with the dosage of bisoprolol having been increased to 3.75mg.

4.4.11 When Ann returned home, she started taking the apixaban tablets as prescribed, but also continued administering the dalteparin injections.

4.4.12 There was limited evidence of the communication with the family in Ann’s notes. The consultant ward round on the cardiology ward took place outside visiting hours at 08:10 hours, so no family members were present.

4.4.13 There was no documentation of direct communication regarding discharge medication with Ann or her family. The investigation identified that there was a discharge process in existence at the reference event Trust. However, the investigation found no evidence of medicines information being shared with Ann or her family, as part of their handover communications prior to discharge.

HSIB makes the following safety observation:

National and local context

4.5.1 A software vendor advised the investigation that if primary care settings use a ‘Health Information Exchange’ (HIE)25 point, then data from the trust’s EHR system could be automatically/systematically pushed to the primary care electronic patient record system. Once GP practices and trusts have signed up to share the information, the EHR systems will permit agreed data to be accessible. HIE is important in aiding the right information to transfer across the patient pathway from primary to secondary care. The Summary Care Record (NHS Digital, 2019) will also provide the latest information about a person’s medication from GP systems.

_{(25) Health Information Exchange enables care professionals to exchange and view patient data, based on an electronic health record system.}

4.5.2 The primary care practice involved in the reference event used an electronic system widely employed in GP practices. There was limited interoperability (see 1.1.3) between the hospital’s ePMA system and the primary care system, with no digital link from the ePMA system to the community pharmacy. Therefore, medication-related information from Ann’s inpatient stay could not be readily transferred or accessed electronically across all systems.

4.5.3 Following Ann’s discharge from hospital, there should have been a reconciled list of her medicines in her GP record within one week of the GP practice receiving the information, and before a prescription or new supply of medicines was issued (National Institute for Health and Care Excellence, 2015). The primary care medical practice employed a practice pharmacist, and part of her role was to undertake medicines reconciliation for hospital patients who had been discharged. However, the practice pharmacist was on holiday when Ann was discharged from hospital.

4.5.4 At the time of the reference event, the Health Centre did not have a policy for medicines reconciliation that set out detailed roles, responsibilities and timescales for completion. The investigation was advised it was “expected practice” that medicines reconciliation would be undertaken by a GP in the absence of a pharmacist, but there was no written policy or procedure to reflect this.

4.5.5 In some parts of the country, web-based platforms (for example, ‘PharmOutcomes’, ‘Refer to Pharmacy’) enable the electronic transmission of medicines-related information. When a patient is discharged from a local hospital, hospital staff will send a referral via the platform to the patient’s pharmacy of choice.

4.5.6 PharmOutcomes was in use at the reference event Trust at the time of the event. However, it was only used to refer patients who required compliance aid devices in the community (which allow medicines to be packaged in individual compartments), to ensure these were in place following discharge.

4.5.7 In primary care, there was a system provided by the local clinical commissioning group that enabled patients to order repeat medications. The patient was required to call the service, following which the prescription would be available at their nominated pharmacy within four working days.

Analysis

4.5.8 Following analysis of the reference event, the investigation identified a lack of interoperability between primary and secondary care electronic prescribing systems and between secondary care and community pharmacies.

4.5.9 The hospital records indicate that a copy of the discharge letter was provided to Ann/ her family at the point of discharge on day three. The discharge summary was received electronically by the Health Centre on the same day (Sunday), reviewed by one of the practice staff the following day, and on day five sent electronically to the GP to action.

4.5.10 As a medicines reconciliation was not undertaken by the pharmacist based in the Health Centre, a medicines review was undertaken by the GP during a face-to-face consultation with Ann. However, this was not until 10 days after she had been discharged.

4.5.11 During the consultation with Ann, the GP added apixaban to her ‘repeat’ medications. When entering this into the electronic system, the GP did not notice an alert notification in respect of dual anticoagulant administration.

4.5.12 Primary care prescribing systems differ in the extent, design, and capabilities of their clinical decision support, and alerts for drug interactions are also different on every system. In the reference event, the primary care electronic system displayed multiple, recurring alerts when the prescribing function was being used. The investigation observed the GP’s computer screen in the Health Centre while elements of the system were demonstrated.

4.5.13 In relation to the drug warnings generated by the primary care system, the investigation was advised there would have been organisation-level control over the way in which the patient-specific warnings are ordered and their visibility (Figure 7). There was a similar control for the individual user, so there was an alternative to adopting the organisational preference. The individual user, in this case the GP, could preview the settings and filter the drug warnings so that all warnings, particular ones or none at all would be displayed.

4.5.14 When a new repeat medication was created on the primary care system, the system would automatically cross-check against the patient’s pre-existing medicines. If there were contraindications, interactions or precautions identified, between the new medicine and a pre-existing one, a patient specific warning would appear on the GP’s computer screen. In the reference event, when the GP added the apixaban to the existing prescription on the electronic system, an interaction warning would have been triggered regarding the dalteparin, as illustrated in Figure 8.

4.5.15 At the reference event Trust, PharmOutcomes was not used to electronically transmit medicines-related information to community pharmacies. Therefore, when Ann was discharged, a referral was not sent to her community pharmacy and this contributed to the delayed detection of the error relating to the anticoagulants.

4.5.16 The community pharmacy was not aware that apixaban had been prescribed by the hospital, because the patient went home with a TTA supplied by the hospital pharmacy and there was no direct electronic link from the hospital to the community pharmacy. If a direct link had been in place, the community pharmacy would have been alerted to the issuing of dalteparin to Ann, as a ‘repeat’, following her discharge from hospital.

4.5.17 When the GP added apixaban to the repeat medications, no apixaban prescription was written and no apixaban tablets were dispensed by the community pharmacy, as Ann still had an adequate supply at home from the 56 tablets she had been issued in hospital.

National and local context

4.6.1 Ann was diagnosed with paroxysmal atrial fibrillation (AF) (an irregular/fast heartbeat) when undergoing hip replacement surgery approximately 10 years prior to the reference event. However, it was only in September 2017, one month after the diagnosis of two cancers (stage 4 incurable lung adenocarcinoma and left renal carcinoma), that she was commenced on an anticoagulant for her AF. The National Institute for Health and Care Excellence (NICE) guideline on atrial fibrillation management was only issued in 2014, so the treatment regime was different when Ann was first diagnosed with AF. The 2014 guidance states that new-onset AF patients should be offered low molecular weight heparin (LMWH) at initial presentation.

Analysis

4.6.2 Ann attended a clinic appointment in September 2017, at which an oncologist and a cardiologist discussed the treatment of her AF. During the consultation with Ann a decision was made to commence an anticoagulant, based on the diagnosis of background metastatic lung cancer (cancer that had spread from another part of her body to her lungs) with the increased risk of a thromboembolic condition (stroke). LMWH injections in the form of dalteparin 15,000 units was prescribed, after checking contraindications.

4.6.3 When the consultant cardiologist saw Ann during a ward round on the cardiac ward on day three, it was the first time he had seen her. The evidence appears to indicate that the focus of the consultation was Ann’s AF and its treatment, which generally consists of heart rate control (or rhythm control) and blood clot prevention.

4.6.4 At the point when the consultant cardiologist changed the anticoagulant medication to apixaban, it was recognised that direct oral anticoagulants (DOACs) are used infrequently in patients with cancer. However, the consultant said that he wanted to help Ann and make “her life a little better” in this respect. It appears that the consultant cardiologist was focused on what would be the most convenient anticoagulant for Ann and not whether any coagulant would be beneficial, or whether apixaban would be as effective as dalteparin. The risks and potential benefits of prescribing an anticoagulant changed after the first decision to prescribe, due to a deterioration in her clinical condition. However, there is no documentary evidence to suggest that the potential benefits and harms of anticoagulation were reassessed when the apixaban was prescribed.

4.6.5 A consultant physician advised the investigation that the decision to prescribe DOACs in patients with advanced cancer needs to be considered carefully. Cancer patients have not been involved in clinical trials of these drugs and the risk of bleeding may be greater in patients with advanced cancer, particularly in those with cerebral (brain) metastases.

Fig 9 Review by a consultant physician subject matter advisor

4.6.6 The investigation considered the reference event in relation to decision making regarding the safety of anticoagulant medications in patients with high-risk, complex conditions. A computed tomography (CT) scan undertaken in March 2018 showed that Ann’s cancer had spread to her brain. One of the major risk factors for bleeding and contraindications to anticoagulation is structural brain disease (Kearnon et al, 2016). Therefore, Ann was at risk of complications due to being on the therapeutic anticoagulant, namely a cerebral bleed (bleed on the brain) (although there is no evidence of this from the clinical findings). It is not apparent that this was considered by the Trust or the doctors who reviewed her case following the CT scan, which included an outpatient review by the acute oncology service. However, the investigation was advised that this service did not routinely review medications due to time pressure, as it is an acute service with rapid throughput.

National and local context

4.7.1 The National Institute for Health and Care Excellence (2016) states pharmacists should undertake medicines reconciliation. However, this did not occur in Ann’s case for the following reasons:

• The reference event Trust had not kept up with the increase in demand for services and there was limited pharmacy cover at weekends. It was a weekend and there was limited pharmacy provision. This resulted in a focus on covering dispensing activities, rather than ward-based work.
• The FY2 trainee did not have time to complete medicines reconciliation on the ward.

4.7.2 Some of the Trust’s services operated at weekends, but this was not universal and there was limited staffing of support services to enact medical decisions and facilitate discharge. Analysis

4.7.3 Ann was in hospital over a weekend, from Friday evening (day one) until Sunday (day three). Ann’s basket of medicines was transferred with her as she moved from the ED to the MAU and finally to the cardiac ward, (although this was only noted in the medical records in the ED and MAU, and not the cardiac ward). As it was a weekend, Ann was not seen by a pharmacist or pharmacy technician and no medicines reconciliation process was completed. The only input from pharmacy was at the dispensing level. Consequently, there was no electronic documentation of the medications that Ann was receiving prior to her admission.

4.7.4 Staffing in the pharmacy dispensary on Sunday (day three) comprised three pharmacists who worked in the morning and one pharmacist who worked all day. The usual support services from pharmacy were not available as ward pharmacists predominately work Monday to Friday. During the week, the ward pharmacists undertake medicines reconciliation.

4.7.5 The investigation was advised by the reference event Trust that the provision of pharmacy services from 09:00 to 17:00 hours on Saturdays and Sundays was significantly reduced compared to those available from Monday to Friday. The service had been unable to keep pace with the 24-hour, seven-days-a-week medical service and increasing numbers of patients discharged during weekends.

4.7.6 The Trust advised the investigation that, whilst pharmacy cover seven days per week would be desirable, the reality is that there is a national shortage of pharmacists, making recruitment to existing vacancies challenging. If the Trust moved to seven-day working, there would be an additional requirement to recruit to the new posts that would be required. Many pharmacists are leaving secondary care for roles in primary care to avoid weekend and on-call working.

4.7.7 The consequence of a reduced weekend pharmacy service in the reference event Trust was that junior doctors, who were already under pressure with an increased workload, had to enter patients’ medication history into the ePMA system. The role of the junior doctor in this context was to document in the medical notes, generate prescriptions on the ePMA and write discharge summaries.

4.7.8 There was no palliative care service or acute oncology service available at the weekend at the Trust where the reference event occurred, so the earliest Ann would have been seen by those services as an inpatient would have been day four. However, she already had an outpatient appointment to attend the acute oncology service on day four and so attended this appointment post discharge. Ideally, Ann should have been reviewed while she was in hospital, which would have been an opportunity to stop her anticoagulants. If her hospital admission had not been over a weekend period, she might have been reviewed by the acute oncology service as an inpatient.

4.7.9 The investigation was advised that weekends were usually the time when clinical teams were most stretched and pressure to discharge patients was often at its greatest, leading to a greater potential for errors. The investigation was also advised that due to workload and staffing at weekends, there was a time pressure on ward rounds.

This was substantiated by a staff member who outlined that a typical weekend workload was as follows:

• The ward round commences at approximately 07:45 hours to 08:00 hours on the coronary care unit (CCU), where there are typically eight patients (the review of the CCU patients on a Sunday tends to be brief if they are stable, as they have been seen on the Saturday too).
• The ward round will then review all new admissions on the cardiac ward (average being five to six new patients).
• Following the ward round, the FY2 trainee would be expected to carry out the jobs and be on call covering half of the hospital’s medical wards and medical outliers for the remainder of the day.
• The consultant cardiologist would go to the cardiac catheterisation suite to do a list there, which the consultant said he would try to start as soon as possible after the ward round.

National and local context

4.9.1 A subject matter advisor (SMA) advised the investigation that most UK hospitals now operate systems whereby medical patients who cannot safely be discharged from the ED are initially admitted to an acute medical unit (AMU) or medical assessment unit (MAU). A proportion of patients will be discharged directly from the AMU/MAU, but those who require ongoing treatment in hospital will be moved to other ward areas.

4.9.2 According to recommendations from the Royal College of Physicians (2013), the ward area selected for transfer should be determined by the patient’s main clinical problem. Therefore, a patient whose primary problem is AF may be deemed suitable for transfer to a cardiac ward, while a patient with lung cancer may be more appropriately managed on an oncology ward or respiratory unit. Challenges arise where patients have combinations of problems affecting more than one body system; this was the situation in the reference event.

Analysis

4.9.3 Upon initial assessment by the Specialist Trainee 4 (ST4) in the ED, the notes listed two diagnoses:

1. ‘AF with fast ventricular rate
2. Decline in last 2 weeks with difficulty swallowing and had nausea/vomiting possibly related to cancer/cancer treatment.’

4.9.4 AF was listed as the primary diagnosis when Ann was clerked by the doctor in the ED. However, her symptoms (difficulty swallowing, nausea and vomiting) did not correlate with the primary diagnosis. It appears that subsequent decision making and specialty referrals were made on the basis of this primary diagnosis.

4.9.5 Ann was reviewed by the medical team; she was seen by the respiratory registrar and by a cardiology consultant with the support of the FY2 trainee. Ann was transferred from the MAU to cardiology (because of the diagnosis of AF). The SMA considered that respiratory medicine would have been a more appropriate specialty for Ann to have been referred to, as it was clear she had advanced cancer.

4.9.6 Ann was seen by a respiratory registrar following the ward round on the MAU and then by a consultant cardiologist prior to discharge. However, she was not seen in hospital by clinicians in oncology or palliative medicine to address her other significant health issues.

4.9.7 Ann was seen in the acute oncology service outpatient clinic at the hospital following discharge, to discuss the findings from a recent CT scan and to relieve distressing symptoms such as difficulty swallowing. The short appointment time meant there was no opportunity to undertake a medicines review. A medicines review would have provided an opportunity to detect the administration error and possibly to question whether Ann needed to remain on anticoagulants at all, given the progression of her cancer.

4.9.8 In summary, the administration of two anticoagulant medicines was not detected until 15 days post discharge from hospital. The opportunity to intervene to prevent harm from the medication error was missed during two contacts with an out-of-hours GP, an appointment with the patient’s own GP and an outpatient visit to an oncology service (Figure 10).

National and international perspective on digitising the secondary care system

5.1.1 The National Programme for IT (NPfIT) aimed to introduce a single electronic health record (EHR), including electronic prescribing, connecting all general practices and hospitals in England. The NPfIT was set up by NHS Connecting for Health (Department of Health, 2005) in 2002 and the UK government announced the dismantling of this NPfIT in September 2011. Consequently, trusts started making their own technology procurement decisions.

5.1.2 The European Commission (2009) investigated the socio-economic impact of interoperable EHR and ePrescribing systems. General findings were that EHR and ePrescribing were beneficial investments in healthcare: the gains to society from such systems exceeded the costs. However, the study found that it typically takes up to nine years before initiatives produce their first positive annual socio-economic return. The study highlighted interoperability as a key contributor to the benefits produced from EHR and electronic prescribing systems.

5.1.3 The National Information Board (2014), established by the Department of Health, stated that all patient and care records would be digital, real-time and interoperable by 2020. In the NHS Long Term Plan, NHS England outlined plans to connect electronic records across primary, secondary and social care by 2024 (NHS England, 2019). Planning for electronic records was managed by NHS England, the National Information Board and the Health and Social Care Information Centre. The three bodies were developing national standards, but local areas could choose their own systems. Challenges to implementation were stated as including interoperability (there were over 100 commercially available electronic record systems), installation and staff training.

5.1.4 In late 2015, the National Advisory Group on Health Information Technology was formed to advise the Department of Health and NHS England on digitising the secondary care system (Wachter, 2016). The advisory group concluded that:

• the NHS should digitise secondary care in stages, so that trusts who were not ready were encouraged and supported to build capacity
• a new digital strategy was needed to achieve a balance between local/regional control and engagement versus centralisation
• systems should be designed with the user in mind, since poorly designed and implemented systems could create opportunities for errors and user dissatisfaction
• national standards for interoperability should be developed and enforced
• all trusts should have a high degree of digital maturity by 2023.

5.1.5 In September 2016, the first 12 Global Digital Exemplar (GDE) acute trusts were announced. GDEs are part of the NHS England’s flagship programme of NHS trusts who are leading the transformation of healthcare technology. The blueprints they develop (structured information and methodologies to help other trusts develop digital capabilities) aim to make it quicker and cheaper for others to reach the same standard (Department of Health and Social Care, 2018a). According to Next Steps on the NHS Five Year Forward View (NHS England, 2017), the intention was that hospitals would choose a GDE to partner with, implementing 80% of the same system and changing 20% to adapt to local needs (NHS England, 2017).

5.1.6 The Nuffield Trust report (Castle-Clarke and Hutchings, 2019) describes how the national policy for digitisation is working from the perspective of acute trusts. The report states that an important role for national bodies is setting digital standards, particularly standards to support access to data across the system, although the national strategy for implementation is often lacking. In relation to the GDE programme specifically, positive feedback was received, particularly regarding the collaborative environment it promoted, which focused on sharing best practice and learning. This was in direct contrast to trusts working ‘in silos’ (on their own), outside the GDE programme (Castle-Clarke and Hutchings, 2019). However, those involved in the programme reported challenges with the reporting requirements.

5.1.7 One of the recommendations arising from the Nuffield Trust report relates to looking beyond blueprints to share best practice, which is a reflection on the GDE and ‘Fast Follower’(26) programme. Significantly, the report says the digital exemplar trusts have had limited success in spreading their expertise to other providers and alternative mechanisms should be considered. However, the report data was gathered in 2018, at which point there were no blueprints and the Fast Follower programme was not established.

5.1.8 NHS England commissioned the University of Edinburgh to independently evaluate the GDE programme. The first-year report (Williams et al, 2018) shows that ‘the GDE Programme has helped to accelerate and focus local efforts of digital transformation in the NHS’. The independent GDE evaluation programme is reviewing both this and the 18-month report, which is now available, and is starting to see some evidence that spread is working in terms of digitisation. The next milestone is the two-year report, which is due in January 2020.

Intended consequences of electronic prescribing and medicines administration (ePMA) systems

5.1.9 There is an international impetus to reduce medication errors, and to implement patient care information systems (PCISs(27)) as a potential means to do so (Ash et al, 2004). Similarly, Shemilt et al (2016) state that ePMA was promoted to reduce prescribing errors (particularly those caused by illegibility) and supported the efficient management of medicines for both patient and trust. Key findings from the Institute of Medicine (2012) also support the concept that health information technology (IT) can improve patient safety (Figure 11) and Mulherin et al (2013) state that electronic prescribing systems are considered a key technology for the prevention of adverse medication events caused by errors.

Fig 11 Interdependent activities for building a safer system for health IT (Institute of Medicine, 2012)

5.1.10 The World Health Organisation’s (WHO’s) third global challenge to reduce medication error is a key national programme of work. Electronic prescribing is seen as a key enabler for this and is an NHS England priority (NHS England, 2018). NHS England has now developed a simulation tool, known as the ePrescribing Risk and Safety Evaluation (ePRaSE). A pilot began in mid-2019, and the tool was launched on 2 September 2019. According to ePRaSE, implementing ePrescribing is associated with a greater than 50% reduction in patient adverse drug events and medication errors.

Introduction of new types of error associated with ePMA systems

5.1.11 However, the complexity of implementation encompassing multiple work processes and cultural factors affects the adoption and use of ePMA systems, causing both intended and unintended consequences (Westbrook et al, 2012). NHS Connecting for Health recognised that although ePMA systems were able to reduce certain prescription errors, they had introduced new types of error.

_{(26) Fast Followers are trusts that have partnered with GDEs.}

_{(27) PCISs are broadly defined as applications that support the healthcare process by allowing healthcare professionals or patients direct access to systems such as ePMA (Ash et al, 2004)}

5.1.12 The extent of the problem is uncertain, however, as there is limited data available relating to the unintended consequences of ePrescribing. In the US, the degree of harm and impact of systems on quality and safety is not well known because of the varied nature of health IT products (Institute of Medicine, 2012), and the lack of a uniform measurement to assess the impact. The position in the UK is similar (Ahmed et al, 2016), with the main problem being that there is no systematic way of collecting and cataloguing incidences of harm caused by ePrescribing. The Department of Health and Social Care policy research group has commissioned a study on optimising ePrescribing in hospitals to address the knowledge gap regarding medication error reduction.

5.1.13 Westbrook et al (2012) say that whilst their study provides persuasive evidence of the current and potential value of commercial e-prescribing systems to significantly and substantially reduce prescribing errors in hospitals, success in achieving this outcome depends on many contextual and organisational factors. Multi-method studies are of great value to understand the mechanisms by which electronic prescribing systems affect prescribing behaviours. Campbell et al (2006) say that in order to ensure clinicians will accept and use a new system, it needs to balance required new work with system-based reductions in the workload associated with old ways of working.

5.1.14 Ash et al (2004) describe that, in practice, prescriptions for medicines are initiated during ward rounds, during discussions among senior and junior doctors and nurses. ‘A case is discussed, a suggestion is made and elaborated on, and it becomes an order [that is, a prescription]. It can also be transformed, renegotiated, or ignored.’ Relevant to the reference event, Ash et al (2004) indicate that the entering of prescriptions on to an ePMA system is the task of the junior doctor, who often undertakes this task alone at a computer, after the patient rounds. The researchers reflect that the junior doctor may enter a series of prescriptions for several patients, copying from the notes made during ward rounds. In such a setting, outside of the actual context in which the patient’s case was discussed, and away from those who could correct any misinterpretations, the researchers state that prescription entry can be prone to errors.

5.1.15 From the UK perspective, the observation of users entering information in batches has more recently been confirmed by Cresswell et al (2017b): ‘…some users recorded patient medication data on paper and entered them into the system in batches’. This information included the prescription of medicines by doctors, as evident in the reference event. The research paper also refers to the delegation of this task to junior doctors, who may not be the decision makers but nevertheless are expected to enter the information on to the system. Both these practices have the potential to create errors in prescribing and are do not reflect the intended contemporaneous use of ePMA systems.

5.1.16 It is evident that implementing ePrescribing is associated with a reduction in prescribing errors, but there are caveats. The reduction in medication errors depends on optimising commercial 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).

HSIB makes the following safety observation:

Categories of error types

5.1.17 Campbell et al (2006) identify that health care organisations in the US often implement ePMA systems as part of their approach to improve medication safety and reduce health care costs. However, the researchers support what others have said about the creation of errors and refer to unintended adverse consequences (UACs) resulting from ePMA system implementation and maintenance. The paper refers to nine major categories of UAC, which were devised as part of a study that focused on US hospitals:

• More/new work for clinicians. ePMA systems may create additional work for clinicians and slow the rate at which they can carry out the clinical documentation and ordering processes (although this loss of efficiency often recovers over time).
• Unfavourable healthcare workflow issues caused by clinical computer technology integration and adjustment.
• Ongoing system requirements. The system must be adjusted, upgraded, tested, interfaced with other systems, and backed up regularly. All employees must be trained in system use and retrained after substantive system changes.
• Electronic medical record systems trend toward ‘going paperless’, whereas in healthcare organisations there remains ‘paper persistence and proliferation in all areas of patient care’.
• Untoward changes in clinical communication patterns and practices. For example, clinicians reported that clinical systems like ePMA may reduce face-to-face communications regarding patient care. This has the potential to increase the likelihood of errors and so requires modifications of computer-based communication systems.
• Negative emotions. This may be helped by training and open communication, to promote better understanding.
• Generation of new kinds of errors. This may be partially overcome by optimising human/ computer interface design.
• Unexpected changes in the organisational structure.
• Overdependence on the technology.

5.1.18 Few studies have applied a systems approach to understanding the causes of specific prescribing errors in the context of hospital electronic prescribing (Puaar and Franklin, 2017), particularly in the UK. A study by Puaar and Franklin (2018), which reviewed how ePrescribing gives rise to new types of error, described three broad groups of latent conditions:

• ePrescribing system functionality and design
• organisational decisions around ePrescribing implementation and use
• prescribing behaviours in the context of ePrescribing.

5.1.19 A further study describing how hospital ePrescribing systems can introduce new areas of risk in addition to offering patient safety benefits, outlined a taxonomy of factors underlying unintended safety risks (Mozaffar et al, 2017):

• Suboptimal system design, including: lack of support for complex medication administration regimens; lack of effective integration between different systems; and lack of effective automated decision support tools. Enhancements in system design by suppliers were considered to be the key enablers of improved patient safety outcomes. These enhancements included better interface designs, and improved integration and transferability of data between systems. The researchers suggest that hospitals need to understand and communicate the safety benefits of ePrescribing as well as unintended consequences and safety threats.
• Inappropriate use of systems (workarounds). Cresswell et al (2017a) define workarounds as behavioural/technological strategies that users of technology employ to overcome issues that hinder them in completing a task. They often arise in the context of changed workflows initiated by the new technologies. Workarounds can create new risks and present new opportunities for improvement in system design and integration.
• Suboptimal implementation strategies, resulting from partial roll-outs/dual systems and lack of appropriate training.

Mitigating risk factors

5.1.20 The causes of errors arising following the introduction of ePrescribing systems are often subtle but potentially serious. Mozaffar et al (2017) highlight the importance of understanding the causes of unintended safety consequences so that the existing and new problems in the development and use of the systems can be addressed. The study demonstrates that the unintended risks that occur in English hospitals are similar to those that occur in the US, although some of the underlying causes are different because of the diversity of systems and greater reliance on overseas suppliers in England.

5.1.21 Puaar and Franklin (2017) found that errors were associated with the design of ePMA systems and integration within the healthcare environment. Their findings suggested that system vendors should focus on revolutionising interface design and usability issues, considering the wider healthcare context in which such software is used. The study also stated that healthcare organisations should draw upon human factors principles when implementing ePrescribing. Consideration of work environment, infrastructure, training, prescribing responsibilities and behaviours should be considered to address local issues identified.

5.1.22 Following the completion of their study in UK hospitals, Mozaffar et al (2017) recommended the creation of technological and organisational strategies, to be implemented by hospitals and the system vendors. To achieve patient safety while using complex technologies means that strategies must be considered that can be used by adopters, implementers and suppliers of health technologies. Work has been undertaken to provide guidelines for best practice in how to create medication prescriptions on electronic systems to mitigate risk (Mozaffar et al, 2017).

HSIB makes the following safety observation:

Learning from ePMA system issues

5.1.23 In the HSIB investigation report Design and safe use of portable oxygen systems(28), the ‘Yellow Card’ system was discussed (see section 5.4). This system offers vendors a viable way to collect and respond to end-users’ findings, thus informing future development. The investigation noted that the ‘Yellow Card’ system was an underused reporting mechanism and is an appropriate reporting method for ePMA systems as they are medical devices.

_{(28) www.hsib.org.uk}

5.1.24 Lichtner et al (2017) state that health information technology is often involved in patient safety incidents, such as incidents involving medication. A sample of reported incidents from the National Reporting and Learning System (NRLS) showed that a variety of digital systems presented common usability problems, problems in data configuration, software coding and supporting infrastructure. The study noted the absence of an effective system at national level to address common issues and enable learning.

5.1.25 The investigation has been advised that the NRLS has been challenging to use in this respect for the collation and IT data aggregation for learning, as there is no taxonomy around IT-related events.

5.1.26 NHS England and NHS Improvement advised the investigation that its project, Development of the Patient Safety Incident Management System (DPSIMS), provided an opportunity to capture technology-related issues, so enhancing learning and the delivery of safe care. Although it is still in the testing phase and may be amended following feedback from users, the DPSIMS will give IT systems and software more prominence than they received in the NRLS. It is possible that aspects of the PSIMS taxonomy will change to allow better classification and identification of IT-related incidents and/or a new category will be created.

5.1.27 Live roll-out of DPSIMS is anticipated to begin during mid-2020. PSIMS will enable better classification and identification of incidents relating to IT systems and software than the NRLS, and the extraction of data about such incidents. However, due to the agile nature of the development of this system, the specific tools that will allow this to happen have not yet been fully developed.

HSIB makes the following safety recommendation:

Lack of standardisation and wide variety of ePMA systems

5.1.28 Mozaffar (2019) describes 18 suppliers of 19 different ePrescribing systems available in hospitals in England (not including specialist oncology and critical care systems). Approximately half of the products were designed and developed by UK based suppliers; the remaining half originated from various other countries, predominately the US. Most of the international systems were designed to cater for the practice needs of their countries of origin and then adapted for the UK. Some systems were developed by trusts as bespoke products; others were commercial ‘off the shelf’ systems and some were a hybrid of both. However, even when the same commercial system has been employed by trusts, some have adapted and configured them differently, negating standardisation and introducing variability. This diversity in the background of the products has resulted in a range of different pre-defined processes, tasks and workflows in the system.

5.1.29 Challenges facing trusts include variable procurement options and a wide range of products with different interoperability and varying functionality. Because of these challenges, Mozaffar et al (2014) reinforce the requirement for further guidance in relation to system selection strategy.

5.1.30 Figure 12 illustrates the main system types used by trusts.

5.1.31 The wide variation in systems and how they are used creates challenges for healthcare professionals who may have to use multiple systems in any given organisation. This is compounded by the need for clinicians to learn and understand new systems if they move between organisations (Barber, 2013). The investigation has been informed that the Royal College of Physicians (RCP) Patient Safety Committee had discussed the need for a standard human-computer interface, so that doctors moving from one trust to another do not have to learn new systems. The RCP Safer Prescribing report highlighted that junior doctors and other prescribers need to be familiar with the functionality of the system and be aware of the potential sources of error in ePrescribing.

HSIB makes the following safety recommendation:

Implementation of ePMA systems

5.1.32 According to the European Commission (2009), there is no single, theoretically correct strategy for implementing interoperable EHRs and ePrescribing systems. It states that decisions to invest in such systems should adopt strategies which fit local settings, adding they are designed to succeed by meeting clearly identified, measurable needs. They identify the use of EHRs and ePrescribing as an essential component of successful change in clinical and working practices, to improve health service delivery and performance.

5.1.33 The primary care sector began digitising in the 1980s, and by the mid-2000s was nearly 100% digital (Wachter, 2016). In 2004, it was recognised that greater use of electronic prescribing in hospitals, bar-coding technology and robotic dispensing had the potential to further reduce the risk of medication errors (Department of Health, 2004). However, full implementation of ePMA systems in secondary care has not yet been achieved (Digital Health, 2017).

5.1.34 IT initiatives are often politically rather than clinically motivated (Cresswell and Sheikh, 2009) which disengages healthcare professionals and other key stakeholders from the outset. The European Commission (2009) adds that EHR and ePrescribing should be approached as a clinical venture, not as an IT project, recognising the significant strategic gains of such systems for healthcare. According to Cresswell and Sheikh (2009), previous research has shown that top-down implementations can contribute to user resistance and subsequently carry a high risk of failure.

5.1.35 Experience has shown that embedding systems into everyday practice is a long-term project. Importantly, the results highlight the need to continually monitor and refine the design of these systems to increase their capacity to improve both the safety and appropriateness of medication use in hospitals (Westbrook et al, 2012). The consequence is that trusts require IT departments with sufficient resources to monitor and adapt systems to local needs, while submitting to external audit and quality control.

5.1.36 The investigation considered the opportunities for improved technological interventions, specifically ePMA systems. The adoption of ePMA systems in hospitals across England (and internationally) has been slow. Data from November 2017 shows that 35% of acute trusts had ePMA supported inpatient prescribing and less than 12% of mental health organisations had an ePMA system (Department of Health and Social Care, February 2018a). Figure 13 illustrates inpatient ePrescribing data from NHS England for acute and mental health/ community trusts from January 2016 and November 2017.

5.1.37 The various ePMA systems are technically complex and the environments in which they have been employed are equally complex and various, with different systems offering different capabilities and functionalities. Successful implementation requires considerable change to working practices across medicine, nursing and pharmacy. These challenges mean that implementing systems can take years rather than months. There is variability in the effectiveness of ePrescribing, which may be attributable to the different systems, interventions or implementation factors (Nuckols et al, 2014).

5.1.38 The wide range of systems available with different interoperability and varying functionality reinforce the requirement for further guidance to help trusts select and implement systems.

5.1.39 In England, multiple electronic and paper systems present significant challenges to the design of systems interfaces, training of the workforce, and the design of safe systems of working (Ahmed et al, 2013, Digital Health, 2017).

5.1.40 The Report of the Short Life Working Group on Reducing Medication-Related Harm (Department of Health and Social Care, 2018a) states that ePMA systems can provide patients with improved information, particularly when patients are transferring between care settings. However, a consequence of patients not being able to view a paper drug chart in hospital is that conversations with healthcare professionals about their medication might be more challenging. One of the key priorities outlined by the Working Group is to improve access to inpatient medication information, particularly in digitised systems.

5.1.41 An ePMA research team funded by the National Institute of Healthcare Research (NIHR) developed an ePrescribing toolkit. This toolkit has been commissioned by NHS England and will continue to support organisations looking to procure and implement an ePMA system. It brings together learning and experience from trusts who have implemented ePMA systems.

Medicines reconciliation and ePMA systems

5.1.42 The Special Report: E-prescribing (Digital Health, 2018) recognises that reconciling the medicine a patient is on when they come into hospital with what they are given whilst in hospital is a significant safety issue. It states:

‘What becomes really powerful is when you link that [ePrescribing] with other systems. Whether that be other systems within the hospital or linking with GP prescribing.’

5.1.43 The guideline NG5 (National Institute for Health and Care Excellence, 2015), states that 95% of medicines reconciliation should be conducted within 24 hours. The investigation was advised that no organisation in England was meeting this standard. This is supported by Safety Thermometer data (Figure 14) which illustrates that in the six year period from October 2013 to (but not including) October 2019, the proportion of patients with medicines reconciliation started within 24 hours of admission has consistently been below the target of 95%. Figure 14 also highlights variability in the proportion achieving the median average of 74.5%, with a decline in performance overall for 2019.

5.1.44 One of the GDE trusts visited informed the investigation that it was achieving 100% of medicines reconciliation on the medical admissions unit, frequently within 14 hours, and approximately 80% trust wide within 24 hours. This was achieved by having an integrated medicines management service encompassing electronic prescribing and medicines administration, electronic care plans and access to the Summary Care Record.

Systems optimisation

5.1.45 In the report Ten key considerations for the successful optimisation of large-scale health information technology, Cresswell et al (2017c) state that optimisation strategies should be ‘continuous organisational efforts to improve HIT [health information technology] systems and the ways they are used’. Further, these efforts might involve the enhancement of functionalities such as decision support systems; the development of data analytic functionalities; and creating robust mechanisms to manage emerging problems, some of which may have been introduced by the HIT. It was anticipated that this will ‘help healthcare organisations in their continuing journey of refining HIT systems to maximise benefits’ (Cresswell et al, 2017c).

5.1.46 The investigation reviewed data about levels of IT adoption within trusts and the use of the available functions offered by IT systems. Three national digital maturity assessments have been undertaken by NHS England in 2015, 2016 and 2017 across all system types (Figure 15). The latest assessment in 2017 found that digital maturity around ePrescribing across trusts ranged from 0% to 71%, with the average being 21%. The information demonstrates that the uptake of systems and the functionality has increased over this period, however, full utilisation of available functionality has still not been completely achieved.

5.1.47 Digital maturity was self-assessed by individual organisations. The aim of the work was to identify how ready trusts were to implement and maintain ePMA systems; identify the available functional and mature capabilities; and find out to what extent these were being utilised. The measures were developed in collaboration with system users and a number of experts in the field and revisited over three years as the initial acceleration of ePMA took place following the distribution of technology funds.

5.1.48 The Report of the Short Life Working Group on Reducing Medication-Related Harm (Department of Health and Social Care, 2018a) finds that systems optimisation remains a challenge. The report states,

‘It is clear from work undertaken by NIHR, and the national digital e-prescribing and medicines administration maturity information, that there are significant challenges and delays with sites optimising systems once they have been implemented.’

Problems with systems optimisation are leading to delays in reaping the benefits of ePMA systems, particularly in relation to medication safety and the oversight of system-generated errors. The challenges are attributable to the technically complex systems, which require considerable change to working practices across medicine, nursing and pharmacy if implementation is to be successful.

5.1.49 NHS England has advised the investigation that whilst it is fully aware of the challenges faced by secondary care when attempting to introduce ePMA systems, until the introduction of ePRaSE, there was no means of evaluating how these challenges were being addressed at a local level. The ePRaSE tool evaluates and supports improvements relating to hospitals’ use of local configuration and clinical decision support (CDS) (see 5.2.8) in ePMA. The aim is to reduce harm caused by adverse drug events and to improve medication and patient safety; the focus being on learning and sharing information to drive optimisation more quickly. If ePRaSE had been in existence at the time of the reference event, the potential for error would have been reduced, by enabling organisational assessment of how well the system had been locally configured.

5.1.50 In summary, attempts to introduce ePrescribing into secondary care have resulted in intended and unintended consequences. There is evidence that ePMAs reduce prescribing errors and improve medicines reconciliation. However, ePrescribing introduces new forms of risk and error. Such risks are compounded by a lack of standardisation in terms of implementation strategies, the wide variety of ePMA systems and inadequate interoperability between systems. Additionally, systems optimisation is hampered by the lack of robust mechanisms to capture, act on and test issues and improvements relating to digital technologies in healthcare.

HSIB makes the following safety recommendation

Medicines reconciliation in primary care

5.2.1 At a strategic level, ‘In primary care, clinical pharmacists are now a key part of the general practice team in primary care networks, working alongside GPs and nurses, seeing patients and using their expertise to get the best health outcomes for people from medicines’ (NHS England, 2019).

5.2.2 At an operational level, the deployment of clinical pharmacists in general practice was launched as a pilot scheme in 2015 (NHS England, 2019). Eighty-nine GP federations were successful in applying for funding from the pilot scheme, which enabled them to recruit more than 490 pharmacists (over 450 whole time equivalent (WTE)) to work across 658 GP practices.

5.2.3 The NHS Long Term Plan (NHS England, 2019) commits to substantially increasing the number of clinical pharmacists working in primary care over the coming years. The Plan states that NHS England will work with government to make more use of community pharmacists’ skills and the opportunities to engage patients.

5.2.4 GPs have usually undertaken the task of medicines reconciliation. However, in recent years some health centres have employed a dedicated practice pharmacist for this task. This is not universal and practice pharmacists are often used for a wide variety of tasks. A subject matter advisor has advised the investigation that the process for medicines reconciliation in primary care is variable.

HSIB makes the following safety observation:

Summary Care Records

5.2.5 Summary Care Records (SCRs) (NHS Digital, 2019) are electronic records of patient information created from GP medical records. Among other information, the SCR holds information about current medication. SCRs can be viewed by authorised staff in other areas of the health and care system involved in the patient’s direct care. Viewing has also been rolled out to community pharmacies.

5.2.6 Access to SCR information means that care in other settings is safer, reducing the risk of prescribing errors (NHS Digital, 2019). However, acute trusts can currently only view the SCR; they cannot update the information or add their own data. The investigation has been advised by NHSX that there are currently no plans to change the SCR or the SCRa (which is the viewing platform for SCR). The Local Health and Care Records programme will be looking to hold this information in the future and NHS England are investing in a number of exemplars to progress the work. This will enable the safe and secure sharing of an individual’s health and care information as they move between different parts of the NHS and social care. It will also enable information to be created and potentially edited by care settings, including secondary care.

5.2.7 This element was not explored further in HSIB’s national investigation. There was no evidence to suggest the inability of the Trust to update or add to Ann’s SCR was an issue in the reference event. It was also evident that work is ongoing at a national level in this area.

Clinical decision support systems: types and functionality

5.2.8 Health IT (2019) defines clinical decision support as follows:

‘Clinical decision support (CDS) provides clinicians, staff, patients or other individuals with knowledge and person-specific information, intelligently filtered or presented at appropriate times, to enhance health and health care. CDS encompasses a variety of tools to enhance decision-making in the clinical workflow. These tools include computerized alerts and reminders to care providers and patients; clinical guidelines; condition-specific order sets; focused patient data reports and summaries; documentation templates; diagnostic support, and contextually relevant reference information, among other tools.’

5.2.9 There are a variety of CDS systems available to guide prescribers when they are clinically managing a patient’s condition. In the context of the reference event, the primary care CDS system generated alerts and drug-drug interaction checks. This included an alert warning the prescriber of an interaction between apixaban and dalteparin.

5.2.10 The investigation undertook direct observation of the electronic system in primary care, by viewing the GP’s computer screen in the Health Centre. In relation to secondary care, evidence was analysed from the literature regarding CDS systems. During field work and observational visits, the investigation heard a wide variety of opinions from clinicians on the usefulness of CDS. The issue was also discussed with stakeholders, academics, and specialist advisors. Both the observational evidence and literature are considered in this section.

5.2.11 Kesselheim et al (2011) highlight that simple CDS systems can provide medication-specific information to users based on a stored repository of clinical information; this repository of clinical information is the knowledge base underlying the CDS system. There are different options for supporting clinical decision making; the most prevalent method is synchronous ‘pop-up’ alerts (alerts that appear immediately when the user selects a field or inputs data) (Mulherin et al, 2013).

5.2.12 The Health Centre involved in the reference event used an electronic system in which all the alerts appeared to be activated. However, other organisations advised their alerts were filtered or turned off altogether.

Alert fatigue

5.2.13 Ash et al (2004) state that decision support overload – the triggering of numerous reminders, alerts, or warning messages – may constantly interrupt the user. Alerts often contain trivial, redundant, or already known information (Campbell et al, 2006). Additionally, alerts present a major workflow process issue, adding to the steps required to enter a prescription. This is supported by Mozaffar et al (2017) who state that ‘the consequent proliferation of alerts, many of which highlighted matters that did not have serious consequences, impaired the usability of the system, as the need to acknowledge these made it harder to complete tasks’.

5.2.14 Campbell et al (2006) state that building and maintaining an appropriate set of alerts for electronic prescribing systems based on current evidence is an onerous, never-ending task, and one that involves calculating specificity versus sensitivity. Mulherin et al (2013) say that for a CDS system to be effective, the ease of using it to provide clinical information and alerts must be balanced with the risk of over-alerting system users. Too many alerts may lead to users paying less attention to them. This is commonly referred to as ‘alert fatigue’.

5.2.15 Alert fatigue is widely agreed to be responsible for high override(30) rates in CDS systems and EHRs; however, the concept itself is not well defined. A 2017 retrospective cohort study found that alert fatigue occurred as a result of cognitive overload(31). Uninformative alerts act as ‘false alarms’, reducing overall responsiveness to alerts, and high numbers of alerts also reduced clinicians’ acceptance of these alerts (Ancker et al, 2017).

_{(30) In this context, if an alert is overridden, the warning is not acknowledged and is dismissed without any action being taken.}

_{(31) Cognitive overload occurs when a large quantity of information is given in insufficient time and without sufficient cognitive resources to distinguish relevant from irrelevant information.}

5.2.16 According to Kesselheim et al (2011), CDS systems are sometimes criticised for issuing too many alerts about possible drug interactions that are of limited clinical usefulness. Users (at departmental/practice level) are often unable to set up the system so that it displays fewer, more specific or more tailored alerts, partly because of the fear of legal consequences if alerts are disabled and harm to a patient ensues. When CDS systems issue excessive alerts about possible drug interactions that are of limited clinical use, this results in 49% to 96% of all alerts being ignored or overridden (Kesselheim et al, 2011)

5.2.17 Campbell et al (2006), Kesselheim et al (2011) and Ancker et al (2017) identify that if clinically appropriate alerts are inadvertently ignored due to alert fatigue, systems’ effectiveness will be diminished, possibly leading to a disproportionately large number (over 25%) of errors and serious adverse consequences for patients.

5.2.18 From a communication perspective, Ash et al (2004) suggest data overload is not the only potential impact of such alerts. The user may experience negative emotions as a result of being constantly interrupted. This is supported by Campbell et al (2006), who state that clinicians report inappropriate alerts as ‘highly frustrating nuisances’. Consequently, healthcare professionals disregard the messages, click them away without looking at the information or turn the warning systems off when they have an opportunity. This was supported by Mozaffar et al (2017), who found that some users created workarounds by switching alerts off, or clicking through them without reading them.

Drug-drug interaction alerts

5.2.19 Phansalkar et al (2012) say that the problem of alert fatigue associated with the use of CDS systems is particularly profound in relation to drug-drug interaction (DDI) alerts. Studies have reported override rates of approximately 90% for DDI alerts. Cho et al (2018) found that alerts relating to potentially harmful DDIs are only moderately effective at preventing harm due to high override rates.

5.2.20 The researchers found that to address the issue of alert overriding, alerts need to be made more specific. This could be achieved by focusing on high-priority DDIs, tiering DDIs according to severity, considering patient factors and co-medications to suppress insignificant alerts, or a combination of these approaches (Cho et al, 2018). Phansalkar et al (2012) concluded that the creation of a standardised list of low-priority DDIs may help to reduce alert fatigue. Despite all this research, override rates remain high and inconsistent across organisations.

5.2.21 The effectiveness of CDS systems was evaluated in US hospitals using simulated prescriptions. The systems detected only 53% of potentially fatal prescriptions and 10% to 82% of prescriptions that would have caused serious adverse drug events (Metzger et al, 2010). The study suggested an increase in national and international collaboration among vendors, implementers, and regulators was required for the co-ordination and oversight of CDS system software.

5.2.22 Ancker et al (2017) found that primary care clinicians became less likely to accept alerts as the numbers increased, particularly more repeated alerts. Complexity of patients’ health conditions was also a factor. Findings were consistent with alert fatigue caused by a high proportion of uninformative alerts combined with complex work, making it challenging to distinguish relevant from irrelevant alerts.

Measures to reduce alert fatigue

5.2.23 To ease alert fatigue, Kesselheim et al (2011) recommend more bespoke or fewer warnings, to include those of the highest importance. However, as mentioned previously, this requires time, expertise, and a continuing commitment.

5.2.24 A study by Kesselheim et al (2011) found that designers, vendors and local organisations limit the ability to modify alerts because of the legal implications of removing a warning that could have prevented a harmful prescribing error. The study identified the solution to be increased national and international collaboration among vendors, implementers, and regulators. It also recommended stronger government regulation of CDS systems and the development of international practice guidelines.

HSIB makes the following safety observation:

Decision support systems in sectors other than healthcare

5.2.25 There do not appear to be any studies that demonstrate the occurrence of alert fatigue in healthcare. However, other sectors, such as transport, have well-developed systems. The Rail Safety and Standards Board (2009) states that good design of alarms and alerts ensures that operators can detect, understand and respond to them appropriately.

5.2.26 Within the aviation industry, the Federal Aviation Administration (FAA) published the Human Factors Team Report on the Interfaces Between Flight Crews and Modern Flight Deck Systems (Federal Aviation Administration,1996). The FAA’s human factors working group reported in 2004 that, ‘Human factors design guidelines already existing within aircraft certification are somewhat fragmented and do not present a unified, coherent approach to the human factors engineering of the modern flight deck.’ Subsequent updates and amendments to the international Certification Specification 25.130232 aim to address human factors and engineering usability of installed equipment on the flight deck, to reduce design-related errors.

5.2.27 Frequently repeated alerts are likely to cause an operator to become familiar with the alert signal, resulting in a reduced response to that alert (known colloquially as the ‘cry wolf effect’), and slower responses in cases where a response is made (cited in Dalton et al, 2018). Dalton et al also discuss the involvement of higher-level cognitive factors. For instance, there is evidence that the ‘cry wolf effect’ may become more obvious in situations of high workload. It has also been demonstrated that being informed that an alert is more reliable is enough to increase the response rates to that alert. Additionally, there is evidence that response rates to an alert increase if other nearby alerts are also active, even if these alerts relate to completely different issues. Collectively, the evidence points to operators using their previous experience of alert reliability, and of their current assessment of the situation, to decide whether to respond. Consequently, in sectors such as aviation, the false alert rate is considered to be a critical factor in the design and evaluation of alerting systems (Dalton et al, 2018).

Development of standards

5.2.28 CDS software sold in the US is subject to oversight by the Food and Drug Administration as medical devices. This body has the authority to ensure systems adhere to general development standards, including engineering usability and human factors evaluation (cited in Kesselheim et al, 2011).

5.2.29 The Australian Commission on Safety and Quality in Healthcare (2016) refers to the country’s national guidelines for on-screen display of clinical medicines information. These guidelines comprise recommendations for clear, unambiguous, standardised onscreen display of medicines information in clinical information systems. The guidelines are intended for those developing, assessing, procuring and implementing IT systems for medication management and electronic prescribing, helping them to understand how design contributes to patient safety, amongst other factors.

5.2.30 In the UK, similar guidelines drafted by the National Patient Safety Agency in 2010 are still in use. The Department of Health and Social Care has already developed software design standards, including standards for CDS systems, which are mandatory for all software suppliers (NHS Digital, 2018).

5.2.31 There is an abundance of literature outlining CDS-related issues, along with initiatives to try and address these issues, by balancing specificity versus sensitivity. However, there are no evidence-based standards for the configuration of software system alerting to ensure users to acknowledge and respond appropriately, using a tiered alert system to determine the type of intervention required (Phansalkar et al, 2012).

_{(32) Certification specification 25.1302 addresses the design and approval of installed equipment intended for flight crewmembers to use from their normally seated positions on the flight deck, encompassing the design and evaluation of controls, displays, system behaviour, and system integration, as well as design guidance for error management.}

HSIB makes the following safety recommendation:

Discharge standards and processes

5.4.1 Discharge is a key point of interface between secondary care, primary care and community pharmacy. However, while the investigation identified over 400 elements (tools, resources and information) in the ePrescribing Toolkit for NHS Hospitals, it was identified that there is minimal focus on integration into safe discharge and/or the interface across primary and secondary care.

5.4.2 The Academy of Medical Royal Colleges (2013) published standards for the clinical structure and content of patient records, including discharge summaries. The publication put forward a strong case for the development and use of standardised electronic health records, in which data is recorded consistently across all contexts. Relevant to the reference event, the Academy of Medical Royal Colleges (2013) stated that the implementation of national standards for the clinical structure and content of electronic health records would enable interoperability between locations and contexts.

5.4.3 Following on from the Academy of Medical Royal Colleges report, the Professional Record Standards Body (2019) produced an eDischarge Summary Standard containing common standards for the information recorded in discharge summaries. The eDischarge standard allows hospitals to transfer standardised clinical information and coded data (SNOMED CT(37) and dm+d(38)) directly into GP IT systems. It ensures all relevant information on diagnoses, medications, procedures and allergies is shared with GPs, to improve quality and consistency of care. Spencer et al (2019) emphasise the importance of the document by stating that correct processing of discharge summaries is essential to ensure patients experience a safe transition of care and not just a hospital discharge.

5.4.4 To enable professionals to deliver the eDischarge standard, the Royal College of Physicians Health Informatics Unit has produced and evaluated a learning package (Royal College of Physicians, 2019). This emphasises the need for complete medication information to be passed on at transfer, and that information should be linked to clinical narrative, as including only new medications is not safe practice.

5.4.5 The Royal Pharmaceutical Society (2012) outlines four core principles for healthcare professionals:

• ‘When transferring a patient, ensure all necessary information about medicines is accurately recorded and transferred with the patient, and that responsibility for ongoing prescribing is clear.
• When taking over care of a patient, check that information about medicines has been accurately received, recorded, and acted upon.
• Encourage patients as active partners in managing medicines when they move, knowing in plain terms why, when and what medicines they are taking.
• Information about medicines should be communicated in a way that is clear, timely, legible and unambiguous – ideally generated and/or transferred electronically.’ There are also three key responsibilities for organisations providing care, which include having safe systems that define roles and responsibilities, ensuring the healthcare professionals are supported to transfer information relating to medicines accurately.

5.4.6 A national Medication Safety Programme was initiated by NHS Improvement in response to the WHO Third Global Patient Safety Challenge: Medication Without Harm. A programme board and general outline of workstreams has been established across four domains, medicines; healthcare professionals; systems and practices; and patients.

HSIB makes the following safety recommendation:

5.4.7 Evidence from the reference event analysis indicates that the discharge process needs to be built into the functionality of ePMA systems from the outset to facilitate safe discharge. In theory each ePMA system could be configured to include the discharge process; however, there is currently no standard to stipulate how the system should be configured, resulting in variation across organisations.

HSIB makes the following safety recommendation:

Clinical standards for seven-day services in hospitals

5.5.1 NHS England’s NHS Services, Seven Days a Week Forum was established in 2013 to consider how NHS services could be improved to provide safer, more responsive and patient-focused services across the seven-day week. The Forum concluded that there was significant variation in outcomes for patients admitted to hospitals at weekends, in terms of mortality rates, patient experience, length of stay and readmission rates (NHS England, 2013; British Medical Association, 2013).

_{(37) SNOMED CT is a structured clinical vocabulary for use in an electronic health record.}

_{(38) The dm+d is a dictionary of descriptions and codes which represent medicines and devices in use across the NHS.}

5.5.2 More recently, a systematic review and meta-analysis has revealed that ‘the weekend effect is unlikely to have a single cause, or to be a reliable indicator of care quality at weekends’ (Chen et al, 2019). This study found that evidence regarding the ‘weekend effect’ was weak and inconsistent in terms of adverse events and length of stay, and was limited on patient satisfaction. Regarding mortality, Chen et al (2019) reported that weekend admissions were associated with a 16% increase; however, the quality of evidence was low and the causes largely attributable to case-mix (type of patients treated) and contextual factors surrounding admissions.

5.5.3 The work of the Seven Day Services Forum (NHS England, 2013) led to the development of 10 clinical standards for seven-day services in hospitals (NHS England, 2017). The standards define what seven-day services should achieve, regardless of when or where patients are admitted.

The four priority standards are:

• time to first consultant review
• diagnostics
• intervention/key services
• ongoing review.

5.5.4 Pharmacy is not one of the priority standards and is part of another standard – the MDT review. This states: ‘All emergency inpatients must be assessed for complex or on-going needs within 14 hours by a multi-professional team, overseen by a competent decision-maker, unless deemed unnecessary by the responsible consultant. An integrated management plan with estimated discharge date and physiological and functional criteria for discharge must be in place along with completed medicines reconciliation within 24 hours.’

5.5.5 The Seven Day Services Clinical Standards (NHS England, 2017) state that ‘support services, both in the hospital and in primary, community and mental health settings must be available seven days a week to ensure that the next steps in the patient’s care pathway, as determined by the daily consultant-led review, can be taken’. Pharmacy is explicitly mentioned in these standards.

5.5.6 In the Seven Day Services Forum (NHS England, 2013), four key themes emerged to support implementation of seven-day services, supporting the statement that: ‘Pharmacy is the golden thread in seven-day services’:

• joining up hospital and community pharmacy services
• pharmacy workforce and ways of working
• targeting where to deliver seven-day services
• affordability and building a case at an organisational level.

5.5.7 One of the supplementary drivers for the change towards the seven-day service was both the clinical and workload support for junior medical staff. The outcome led to a significant shift in work from medical to pharmaceutical staff for the provision of weekend discharge medication. This is one of the benefits of the integrated medicines management service, which the investigation heard about during a visit to a Global Digital Exemplar trust, where pharmacists carry out 98% of all medication requested discharges, seven days per week.

5.5.8 The Deputy Chief Pharmaceutical Officer at NHS Improvement advised the investigation that challenges to the progress of seven-day clinical pharmacy services include ongoing historical service challenges, including issues around the supply of medicines; a lack of service investment; workforce issues; and limitations of digital systems. Agenda for Change contracts (NHS grading and pay system), result in the cost of running services at weekends being significantly higher, particularly on Sundays. This is a significant issue for trusts, which have to balance risk and cost.

5.5.9 During an observation visit to a GDE trust, the investigation was informed about the impact of this variability on the consistency of care provided outside normal working hours. The GDE trust advised that pharmaceutical interventions were being delivered later in the patient pathway and this resulted in reduced rates of medicine reconciliation, delayed discharge and errors.

5.5.10 In most hospitals, support services (including pharmacy) do not match the hospital activity pattern, instead operating Monday to Friday, 09:00 to 17:00 hours. Outside these hours, services are reduced, emergency-only, or absent, with most routine services closing at the point when demand from emergency admissions is at its highest (British Medical Journal, 2016). This finding was supported by Gordon (2017), who said that whilst there was 72% support for the concept of full seven-day services, for this to be properly enacted there must be increased provision and support across the whole hospital. The paper discusses the variable range of support services available out-of-hours and at weekends. Most hospitals reported emergency availability for endoscopy (85%), surgery (82%) anaesthetics (80%) and radiology (77%). However, there were fewer non-medical disciplines available, although pharmacy was one of the exceptions (73%), as there have been established services in hospitals for some years.

5.5.11 According to research by NHS England (2016), limited availability of patient-facing, out-of-hours hospital pharmacy services can lead to lack of support for junior medical and nursing staff, low levels of medicines reconciliation, and poor support for patients at the point of discharge. The paper emphasises that seven-day clinical pharmacy services in hospitals are important to:

• enhance patient experience by allowing patients to discuss medication-related aspects of their care and supporting progress through care pathways
• reduce variation in quality of care by embedding medicines optimisation principles into routine practice, seven days a week
• improve clinical efficiency and patient safety through increased pharmacy staff, able to focus on optimal use of medicines
• address clinical workforce demands by utilising staff such as pharmacist prescribers (Baqir et al, 2014).

5.5.12 The Deputy Chief Pharmaceutical Officer at NHS Improvement agreed that services are stretched due to limited staff numbers. He advised the investigation that consequently there was a need to work differently, for example by allocating more resource at the end of the day when workload tends to be greater. This was a point also made by a director of pharmacy at a GDE trust. During an observation visit, he advised the investigation that it was important to consider workforce efficiency rather than purely focusing on staff numbers. The medical profession has similarly needed to support flexible working to retain consultants in the later stages of their career, in an attempt to ‘stem the drain of expertise and skills from the profession’ (Royal College of Physicians, 2017).

5.5.13 A second GDE trust reported increased efficiency when an integrated medicines management (IMM) service was in place. The integrated management of medication provides a service to the patient from the point of admission to the point of discharge. Integral to its success was the availability of clinical pharmacy services seven days a week, supported by ePMA, electronic care plans and access to the GP system.

5.5.14 To deliver seven-day services, trusts will need to consider a combination of approaches. NHS England (2016) states that planning should include:

• a practical, staged approach to redesigning clinical pharmacy services involving optimal use of technology, workforce and infrastructure
• the provision of targeted services to high-risk patients
• better integration of clinical pharmacists into multidisciplinary teams
• learning and shared practice from exemplar hospitals.

5.5.15 To increase efficiency while maintaining patient safety, an evidence-based clinical prioritisation tool, the Medicines Optimisation Assessment Tool (MOAT), has been developed (Geeson et al, 2019). The tool was developed to target inpatients most in need of pharmacists’ input at two hospitals in England. Data on medication-related problems was collected by pharmacists and information was obtained regarding potential risk factors, such as comorbidities and use of high-risk medicines. Geeson et al (2019) concluded that MOAT has the potential to guide decision making by predicting those patients most at risk of adverse, preventable medication-related outcomes. However, while external validation and further research is required before MOAT is adopted more widely, it has the potential to help trusts to achieve the seven-day service standard. Medicines prioritisation will be explored further in HSIB’s investigation report: ‘I2018/019: Identifying and reducing high-risk prescribing errors in hospital’, which will be published in 2020. Issues relating to the lack of seven-day pharmacy cover.

5.5.16 The Royal Pharmaceutical Society (2014) states that whilst hospital pharmacy teams deliver effective pharmaceutical care and ensure patients use medicines in an optimal way outside normal working hours, there is still variation in the consistency of this care. Limited availability of hospital pharmacy services in some trusts, particularly at weekends, has resulted in a number of reported issues. The two issues relevant to this investigation are prescription errors and the lack of medicines reconciliation.

5.5.17 In relation to addressing medication errors, a large research study (Scullin et al, 2007) identified that pharmaceutical implementation of the integrated medicines management (IMM) pathway reduced errors by an average of 4.2 per patient. In addition, there was a reduction in the length of stay by almost four days for medical admissions and reduced readmission rates. The data also showed faster medication rounds, and timelier and significantly more accurate discharges with a five-fold to eight-fold return on investment.

5.5.18 The second issue relating to the lack of medicines reconciliation has been addressed by one of the GDE trusts visited by the investigation. The trust employs advanced practitioner pharmacists, whose role is to improve patient care and flow at admission and discharge. Additionally, their intervention includes discussions with patients to enhance shared decision making. The wider IMM service is delivered by clinical pharmacists and pharmacy technicians, who undertake medicines reconciliation, verification of medication and clinical review. In the first 18 months following implementation of the seven-day IMM service, there was a demonstrable improvement in the medicines reconciliation rate from 36% in less than 24 hours pre-implementation to around 80% post-implementation.

5.5.19 As previously stated, the medicines reconciliation rate is a significant measure because medicines-related patient safety incidents are more likely to occur when medicines reconciliation is undertaken more than 24 hours after a person is admitted to an acute setting (National Institute for Health and Care Excellence, 2016). Undertaking medicines reconciliation within 24 hours of admission to an acute setting (or sooner if clinically necessary) enables early action to be taken when discrepancies between lists of medicines are identified. This supports seven day working for the pharmacy service.

HSIB makes the following safety observation: