Alexander Reid

Dear Coroner, Re: Regulation 28 Report to Prevent Future Deaths – Alexander Lee Reid who died on 29 June 2021.

Thank you for your Report to Prevent Future Deaths (hereafter “Report”) dated 18 April 2024 concerning the death of Aleander Lee Reid on 29 June 2021. In advance of responding to the specific concerns raised in your Report, I would like to express my deep condolences to Alex’s family and loved ones. NHS England are keen to assure the family and the coroner that the concerns raised about Alex’s care have been listened to and reflected upon. 

I list below the concerns raised within your Report that NHS England can respond to:

• Operation of the COVID-19 Vaccination Programme, including the inclusion criteria for the Shielded Patient List and COVID-19 Population Risk Assessment
• Information provided to patients who were on the Shielded Patient List and subsequently prioritised for vaccination
• Requirements and statutory Clinical Safety duties for GP IT systems suppliers, and our responsibilities for their assurance.

These reflect areas of NHS England’s responsibilities which incorporate some activities formerly operated by NHS Digital, which formally merged with NHS England in February 2023. I address these concerns below.

Shielded Patient List

In February 2021, on the instruction of the Department of Health and Social Care (DHSC), a COVID-19 Population Risk Assessment took place. This risk assessment used data routinely extracted from GP IT records systems – obtained under the General Practice Extraction Service (GPES) – and hospital record systems including Hospital Episode Statistics (HES). These data were used to update identification of the cohort of people to whom infection with Covid-19 could pose a greater risk, and so to whom an invitation for vaccination would be prioritised. The risk assessment used the original – and at that time, latest - version of the QCovid® algorithm. This cohort was identified as “clinically extremely vulnerable”, and for inclusion on the Shielded Patient List (SPL). They would receive further information on shielding and be prioritised for invitation for vaccination. National Medical Director

17 June 2024

Patients identified as Clinically Extremely Vulnerable were managed on an ongoing basis within the SPL, alongside those other patients routinely identified as high risk through:
• the existing clinical methodology
• locally by GPs, or
• by hospital specialists.

As and when patients were no longer considered at high risk by GPs and / or clinicians, their risk status could be updated in the SPL.

Due to the speed and scale of the vaccine rollout, at the time safeguards were introduced into the system to enable curation of the SPL. This included:
• A writeback of the reason for inclusion into the GP records
• A tool to look up the patient’s inclusion for prioritisation
• An enquiry page and email to support patients seeking to understand their inclusion.

Issues surrounding this topic were the subject of considerable media interest and discussion at the time of using QCovid®.To support the safety of the process, safeguards were built into the centrally cohorting systems. Regarding the specific issues raised in your Report, these included:
• Use of the most recently recorded patient BMI within the last 5 years
• Minimum BMI of 15.0 and maximum of 47.0 – data outside this range amended to either 15.0 or 47.0 and marked as data corrected in processing file.
• Default value of 31 used when no BMI data is available.

These safety steps were made publicly available at COVID-19 Population Risk Assessment - NHS England Digital:
• If a person’s BMI was not available, a BMI of 31 was used by default. This was the midpoint BMI in the population data. This would have been utilised in Alex’s case, where the recording of the BMI was more than five years previously.
• If BMI was more than 47, then a BMI of 47 was used to generate the results.
• If BMI was less than 15, then a BMI of 15 was used to generate the results. This was because almost all (99.9%) BMIs in the UK are between 15 and
47. BMIs calculated as being lower than 15 or higher than 47 were likely to be a mistake which would affect the accuracy of the result.

It is therefore unlikely that the BMI of 69.87 incorrectly recorded on the GP record in 2004 caused Alex to be identified as Clinically Extremely Vulnerable by the National COVID-19 Population Risk Assessment.

Local areas were also able to create their own lists of Clinically Extremely Vulnerable patients for shielding. NHS England cannot comment on the processes that were operated at a local level as this would be subject to local governance and controls.

More information can be found about SPL here and details of how the list was created are here. A transparency notice following the closure of SPL provides further information.

Information to Shielded Patients and their GPs

Patients on the SPL were issued with a letter outlining that they had been identified as Clinically Extremely Vulnerable, that they would be invited for vaccination, and were provided additional advice, guidance, and the support available for clinically extremely vulnerable patients at that time. They were informed that their GP would be made aware of their inclusion in the SPL (via updates to the risk flag indicator in the GP record for each patient), and the letter stated that, where patients were unaware why they would be - or did not believe they were - at high risk from COVID-19, they could contact their GP or hospital clinician who could update their records accordingly (as described above). To support this, the details of their registered GP were included in this correspondence.

Clinical Safety of Structured Data Entry into GP IT Systems

Core GP IT records systems are developed - and centrally assured - against standards set out in the GP IT Futures Framework. More information about this can be found here.

Over and above this, all Health IT systems used in the NHS must comply with statutory duties set out in section 250 of the Health and Social Care Act 2012. These duties are described by two Clinical Safety Standards, DCB 0129 and DCB 0160. These standards require that consideration be given to all potential hazards or harms to patients (and users) through use of the product or system. This includes the entry and transmission of patient data.

Depending on the system, a GP IT system can allow height, weight, or BMI to be entered manually. Most systems will offer to calculate BMI when both height and weight are entered/available.

Where there is an automated calculation using other data items (such as is the case for BMI), there is no currently implementable means to apply validation – or define normal – across domains. It should also be noted that the ranges of values that may be possible (or probable) for height, weight and BMI vary by age, gender, and may vary by ethnicity.

We recognise the value in data validation for factors such as BMI. This currently does not exist and the implementation of such tools should be balanced against the sizable implementation challenges, and subsequent clinical risks and hazards that would arise. Such risks and hazards would be considered within the Clinical Safety Case for the product, and a balanced risk assessment would be required to support implementation.

Central Data Extracts, Validation and Future Cohorting

Cohorting of patients drives many systems within the NHS – from vaccinations through to screening. Ensuring accurate and appropriate cohorting ensures the safety and effectiveness of health protection activities.

As described above, in this case, data was extracted from the GP records system via the General Practice Extraction Service (GPES). Safeguards are developed alongside data use for cohorting of patients. This is clinically assured and includes aspects such as:
• exclusion of data that may not be reliable due to time interval since recording
• normalisation/limitation of effect of extremes of recording.

Patient and Digital Clinical Safety in General Practice

NHS England takes Patient Safety extremely seriously. As such we published our Patient Safety Strategy in July 2019 and have developed the Patient Safety Syllabus, which is accessible to all staff, including those working in General Practice and Primary Care. More information on our Patient Safety Strategy is available here.

With the increasing use of digital systems in the NHS, we subsequently published the Digital Clinical Safety Strategy in September 2021. NHS England has developed materials for training on this subject, which includes the safe, effective use of health IT systems, and made this available to all staff. More information on the Digital Clinical Safety Strategy is available here.

All staff should be trained in the safe use of health IT systems that they use routinely to perform their clinical duties. This is a matter for local employers and the development of policies and procedures to support this are monitored by the relevant regulatory bodies – in the case of health care service providers, this is the Care Quality Commission.

Further actions for NHS England

We note the risks that arise from the inaccurate inclusion of data values in cohorting activities and will continue to operate a clinically assured data extraction and validation service to drive future health promotion activities. We will work towards surfacing inclusion data to patients via their NHS App, so that they may discuss their inclusion in health promotions with their care providers.

To support the work of the Primary Care Patient Safety Team, we will promote the Digital Clinical Safety Strategy, and all training modules that are available for staff via these channels.

I would also like to provide further assurances on national NHS England work taking place around the Reports to Prevent Future Deaths. All reports received are discussed by the Regulation 28 Working Group, comprising Regional Medical Directors, and other clinical and quality colleagues from across the regions. This ensures that key learnings and insights around preventable deaths are shared across the NHS at both a national and regional level and helps us pay close attention to any emerging trends that may require further review and action.

Thank you for bringing these important patient safety issues to my attention and please do not hesitate to contact me should you need any further information.

In Practice Systems Ltd. R

Cegedim Rx Ltd.

14/06/2024

RE: Cegedim Healthcare Solutions – Response to Regulation 28 Report to Prevent Future Deaths – Reid

The following comprises Cegedim Healthcare Solutions’ response to the report referenced above that was issued by the Coroner for the area of West Yorkshire (Eastern), dated 18 April 2024. The purpose of this response is to provide details of actions taken or that may be taken by Cegedim Healthcare Solutions to prevent other deaths occurring in similar circumstances in the future. Data entered into Vision clinical systems is subject to validation at two levels, namely, “hard” validation where certain data items may not be entered (e.g. text in a numeric field) and “soft” validation where users are alerted to possible errors in data entry but may still continue and file the data (e.g. range checking of numeric data). Vision clinical systems challenge / alert users at such levels where clinical data values are recorded that are considered out of normal / usual range. For example, in the Vision 3 GP clinical system: Weight (kg) has a “hard” upper limit of 1000. And a “soft” range limit of 0.5 to 180kg. The same data validation as above is applied when recording weight values within the Vision Anywhere clinical system. When using the BMI calculator tool in Vision+, only “hard” data validation is performed on a weight value entered, with a “hard” upper limit of 200kg implemented. However, there are rules in place in Vision clinical systems for a patient’s BMI to calculate.

Vision 3 For a patient’s BMI to calculate, their height must have been recorded after the date of their 15th birthday and both height and weight values recorded for the patient must have a date which is on or after their 16th birthday. Therefore, for the patient who died, although their weight value of 145kg which is considered unusually high for a child aged 11y, the patient’s BMI would not calculate or be available in their record as the patient's height was not recorded after the date of their 15th birthday and the height and weight values recorded for the patient don’t have a date which is on or after their 16th birthday. In this scenario this leaves the patient without a BMI recording in their patient record, forcing the GP to update their records accordingly for this figure to be calculated. Without the recording of a BMI in the patient record, the patient would then not appear in any adhoc reports exported from the system that required the need for BMI as a criteria. It should be noted that for the purposes of Covid, the call/recall was not done by practices, and was done based on data extracted from the clinical systems into national systems. It should also be noted that Childhood BMI charts are implemented in Vision 3 apps (https://help.cegedim- healthcare.co.uk/Vision_3_Apps_Help_Centre/v1.0/Content/C_Apps_Available/Childhood_Growth_Charts_Screen.htm?t ocpath=Help%20Topics%7CApps%20Available%7CPatient%20Apps%20%7C_____5)

Vision Anywhere The following rules are used to determine whether a patient’s BMI is calculated.
1. Is the patient 18y or older?
a. YES: Proceed to 2
b. NO: Do not calculate BMI
2. Is the patient aged 21y or over?
a. YES: Is there a height record on or after the age of 21y?
i. YES: Use most recent height record that has been recorded on or after the age of 21y
ii. NO: Do not calculate BMI
b. NO: Is there a height record in the last 12 months?
i. YES: Use this most recent height record
ii. NO: Do not calculate BMI With regards to the concerned scenario/case, the patient’s BMI would not calculate or be available in their record in accordance with the above rules.

Vision+ (internal system tool of calculators)

For the BMI calculator tool in Vision+, there are no rules in place based on a patient’s age for a patient’s BMI to calculate. As long as a patient has a weight and height recorded with values within the “hard” upper limits set for each (200kg and 2m, respectively).

Vision+ will present a calculated BMI value for the patient. However, upon saving the BMI measurement in Vision+, it is only the height and weight values that are written back to the patient’s record held in Vision 3, but not the BMI value itself.

Vision 3 then may or may not calculate the patient’s BMI based on the height and weight values written back, dependent upon the rules in place in Vision 3 based on the patient’s age (as referenced above).

Therefore, for the patient who died, although their weight value of 145kg which is considered unusually high for a child aged 11y this could still be calculated using the BMI calculator tool in Vision+ and Vision+ would have presented a calculated BMI for the patient.

However, Vision 3, would not then calculate the patient’s BMI in accordance with the rules in place in Vision 3 based on the patient’s age (as referenced above) and therefore a BMI value would not be recorded in the patient notes as the requirements for value recording would not have been met based on the above explained system rules on data entry based on age.

It should be noted that, such enhanced functionality (to additionally take into account patient age when recording weight or height for calculation of BMI, including for patients who are under 16 years of age) is not currently among the NHS requirements in the Framework agreements and therefore has not been developed in our clinical solutions.

Although not currently implemented or on the roadmap for future development work in Vision clinical systems, a potential improvement for implementation in future would be the possible development and implementation of functionality to additionally take into account patient age when recording weight or height for calculation of BMI, including for patients who are under 16 years of age.

This is to enable Vision clinical systems to identify / challenge any obviously erroneous or unusually high or low values when recording this clinical data. It is hoped that the availability of such enhanced functionality in Vision clinical systems would provide further mitigation against the risk of the occurrence of further deaths in similar circumstances.

Sincerely,

Chief Clinical Information Officer

Dear Mr Longstaff

Regulation 28 Report to Prevent Future Deaths - touching on the death of Alexander Lee Reid

Thank you for sharing a copy of your report touching on the tragic death of Alexander Lee Reid. I am responding on behalf of the Royal College of General Practitioners as Honorary Secretary to Council. Firstly, can I convey our sincere condolences to the family and friends of Alexander, I was deeply saddened to read of the circumstances around his death.

The Royal College of General Practitioners (RCGP) is the largest membership organisation in the United Kingdom solely for GPs. It aims to encourage and maintain the highest standards of general medical practice and to act as the ‘voice’ of GPs on issues concerned with education; training; research; and clinical standards. Founded in 1952, the RCGP has just over 54,000 members who are committed to improving patient care, developing their own skills and promoting general practice as a discipline.

The RCGP Special Interest Group in Health Informatics is composed of members of the RCGP who have an interest in and work in the area of Health Informatics. The Joint General Practice Information Technology Committee (JGPITC) comprises of members of the British Medical Association, and members from the Health Informatics Group at the Royal College of General Practitioners; it is recognised by paragraphs 4.31 and 4.40 of the National Health Service General Medical Services (GMS) Contract 2003/04 for General Practice surgeries.

This response on behalf of RCGP draws on and includes the expert opinion of our members in the HIG and Joint GP IT Committee.

We have considered your report and the expert opinion of the GPIT committee and offer our response focusing on the areas where the RCGP might have most significant influence to prevent further deaths. We set out below out the responses to your matters of concern from the GPIT committee followed by actions we can take as RCGP to prevent future deaths.

Matters of concern Issue 1 - When Alex was invited to receive his vaccination early, he did not understand why, and no- one was able to tell him.

We were made aware of frustration voiced by patients, their GP surgeries and those delivering the immunisation programme that the indication for immunisation was not transparent to either party at the point of care. These concerns were discussed with NHS England. The clinical safety hazard is that, because the precise reasons for inclusion in the immunisation programme were not surfaced, neither the patients nor the staff delivering their immunisation were able to identify the incorrect invitation. Those hazard logs may provide insights into what was known about these risks. The lesson learned nationally should be that transparency promotes safety, and in particular that when a patient asks, those providing care should be able to investigate and provide a clear answer.

Issue 2 - The inquest heard expert evidence that the combined vaccination monitoring and recall specification designed to identify vulnerable people for the purposes of inviting them to receive their Covid vaccinations early had identified Alexander as vulnerable from an incorrect BMI of
68.97 recorded in his GP records on 06/02/2004. The mistake was due to the relevant clinician recording Alexander’s height as 145cm and his weight as 145 kg, giving a BMI of 68.97 for an 11-year-old boy whose previously recorded BMI aged 9 had been 14.88.

We note that: a) GP electronic patient record system design typically allows a user to enter all 3 variables manually, but when a height and weight are present the system offers to calculate a body mass index for the user, in turn filing that calculation when the data are saved;

b) The relevant, validated child growth charts are available at the internet page

c) Using those charts as reference:
i. a height of 145 cm is typical for an 11 year and one month old boy (around the 65th Centile for his age)

ii. a weight of 145 kg is very atypical. The 99.6th Centile for him would be approximately 65 kg. This means that his atypical weight was readily computable;

iii. the number 145 is repeated across the height and weight entries, which is readily computable and an unlikely coincidence – user error can be reasonably suspected;

iv. the sudden significant change in weight was readily computable. All sudden apparent significant changes in weight should be of concern to users, but changes incompatible with normal rates of weight change can reasonably be treated as suspicious for data error;

v. a recorded body mass index of 68.97 is much less likely to be entered by a clinician manually, because it is more clearly erroneous. It is a computably very atypical body mass index (especially for a child of 11);

vi. the sudden significant change in body mass index was also readily computable and suspicious for data error; and

vii. BMI is not a routinely calculated item for a child, as it does not offer significant clinical value in normal health, and so any system which interacts with physical health parameters for a child could be optimised for the context of height and weight recordings for paediatric populations and modified to not routinely promote the calculation or recording of a BMI.

We also observe that there was a 16-year latency between the detectably erroneous data being entered, and Alex suffering harm as a consequence. This is relevant to the potential for reducing risk, because there was ample opportunity for the data to have been identified and corrected if there were system tools to help identify such issues and either correct the false data or verify that the data were in fact correct. It also presents an opportunity for improving safety by limiting the data extraction time window to ignore very old data.

Issue 3 - The inquest heard expert evidence that to have built a system that would validate multiple data items in an individual’s GP records for the purposes of ensuring that individuals were not incorrectly identified as vulnerable would not have been feasible within the constraints and context of the Covid-19 programme

The Committee worked with NHS England and other organisations during the pandemic, and we note that many of the systems were developed at a speed and pace commensurate with the overwhelming need to deliver the immunisation programme, balancing speed against safety. The Committee agrees with the expert evidence you were provided.

Issue 4 - The inquest heard expert evidence that an easier and more appropriate option would be to embed validation rules in general practice IT systems that would check such information at the time of data entry.

The Committee agrees that embedding validation rules into GP IT systems would reduce the likelihood that computably erroneous data are entered into patient records.

Issue 5 - If the obviously erroneous BMI had not been recorded or had been challenged at the point of entry by the relevant IT system, Alex would not have been classed as vulnerable, would not have

been offered a vaccine before guidance was published that the under 30’s should not receive the Oxford Astra Zeneca vaccine, and would not have died when he did.

The Committee identifies that the ability to record data that are likely to be erroneous in electronic patient records is a cause of clinical safety hazards including this one. We agree that appropriate mitigations to this hazard include consideration of system design.

a) A warning at the point of entry would be one example of a control mechanism and we agree that it would have been likely that if such a mechanism were in place, those data would not have been recorded in Alex’s record.

b) We note that in some circumstances, thresholds can be set to entirely prevent data entry, such as where numerical variables entered are incompatible with life. In this particular case, a body mass index of 68.97 may not have been appropriate to exclude from data entry.

c) The Committee notes an emergent problem in GP record systems, known as ‘alert fatigue’ which refers to the progressively increasing volume of safety issues presented to users of the GP systems. During time pressured consultations, a user must balance the need to complete a consultation and see the next patient, with tackling system alerts for possible safety / data quality issues which may or may not be clinically necessary to address.

Issue 6 - The consequences of the data input error in this case give rise to a concern that more might be done by way of potential data input errors at the point of entry, with consequential improvements in the reliability of such data and the safety of patients and reducing the risk of other deaths occurring in similar circumstances in the future.

We agree that point-of-entry improvements have the potential to reduce data entry input errors, with the consequential risks of death, and other non-fatal harms.

Comment from the Committee
1. The Committee notes that the most effective management of clinical safety relies on not only the good articulation of hazards, but the holistic appraisal of all of the underlying causes before mitigations are planned. A root cause analysis approach would surface those issues.

2. The absence of a root cause analysis placing this error in context means that the Committee cannot quantify the benefits of improving data input errors.

As RCGP we will ask the GPITC to discuss these matters at the next full Committee meeting, which is due to be held in July 2024. We will ask NHS England to: a) Consider coordinating funded clinical safety workshops to document formally the underlying causes of the hazard, identify existing controls, and define the professional requirements for how the systems should ideally behave to mitigate the risks

b) Report to the Joint GPIT Committee on how these concerns might be addressed in alterations to the existing standards and capabilities set out in the supplier contracts

c) Share your report and ours with the NHS England New Market Entrants programme which is currently working with future GP system suppliers; and

d) We will ask the GPITC to consider what existing reports are available on the quality of clinical data in the UK’s primary care records and advise whether or not there is a need for a national clinical data quality improvement plan.

For our Members we will highlight through our programme of continuing professional development: a) The importance of accurate data entry into the record, the importance for accurate systems for data entry within practices and the implications If this does not occur b) That risks exist if inaccurate data is shared beyond the practice where there are fewer opportunities for it to be rectified

I trust that this reply is helpful and if you have any questions, please do not hesitate to contact me. Our sincere condolences are with Alexander's family.

Dear ,

GPC England acknowledges the tragedy of this case and the unique chain of events, stretching back many years, that led to the death of Mr Reid, ultimately due a rare and unpredictable side effect of the Oxford AstraZeneca vaccine. The data entry error where Mr Reid’s weight was incorrectly entered informed the decision by NHS England to invite Mr Reid for vaccination earlier than his age category would have indicated. Had this error not been present it is likely Mr Reid would have been invited for his vaccination at a different time, when more might have been known about the side effects of the different vaccines which were available, and the outcome may have been different. It could not have been predicted at the time the data entry error was made that there might one day be a future pandemic that would necessitate a vaccination strategy that made use of the weight that had been erroneously entered. It is important for patient trust that medical records are accurate, and that any clinical decisions made based on recorded information are appropriate. If trust is damaged and patients become reluctant to make use of vaccination programmes there is the likelihood of future harm, through illness and possibly death, that might otherwise have been avoided.

With regard to the matters of concern, we propose to discuss these at our next Joint GP IT Committee meeting (which has representatives of the BMA, RCGP and the NHS) to raise awareness of this case and seek a consensus on how systems might evolve to reduce the likelihood of similar data entry errors in the future. Given the complex chain of events of this particularly tragic case, where any one step, had it been different, might have led to a different outcome, we would propose advocating a way forward that addresses the main identified concerns which will transcend this case, namely that patients must know why they have been offered a given treatment and that the decision is sound based on the prevailing medical understanding at the time. It would seem unlikely that the exact same circumstances of this case would ever occur again, but it is clear improvements can be made in the NHS’s IT systems so patients can have greater confidence that they are being offered appropriate treatments that take into account their own unique circumstances.