Ben Kingsley v The Information Commissioner & Anor

1. This decision relates to an appeal brought by the Appellant pursuant to section 57 Freedom of Information Act 2000 . It is in respect of a decision notice (ref: IC-313214-K4P1) issued by the Information Commissioner on 22 August 2024 and concerns a request for information made to the UK Health Security Agency on 26 January 2024. This appeal was heard together with, but not consolidated with, appeal FT/EA/2024/0267.

2. The matter was dealt with by CVP then, as agreed and allowed by rule 32 The Tribunal Procedure (First-tier Tribunal) (General Regulatory Chamber) Rules 2009, thereafter it was dealt with on the papers. W hat follows is a summary of the submissions, evidence and our view of the law. It does not seek to provide every step of our reasoning. The absence of a reference by us to any specific submission or evidence does not mean it has not been considered.

3. Page numbers in brackets refer to the open bundle (or if to the closed bundle then the letters CB are added) and in this decision the following definitions are adopted:- Freedom of Information Act 2000 FOIA Data Protection Act 2018 DPA18 Statistics and Registration Service Act 2007 2007 Act the public interest balance test the PIBT The Tribunal Procedure (First-tier Tribunal) (General Regulatory Chamber) Rules 2009 2009 Rules The Freedom of Information and Data Protection (Appropriate Limit and Fees) Regulations 2004 2004 Regs. Mr Ben Kingsley the Appellant The Information Commissioner the IC The UK Health Security Agency the UKHSA The Department of Health and Social Care DHSC Decision Notice IC-313214-K4P1 the DN Request for information made by the Appellant on 26 January 2024 the Request Office of National Statistics ONS First-tier Tribunal and Upper Tribunal FtT and UT Coronavirus Disease 2019 Covid-19 The National Immunisation Management System NIMS The UK Covid-19 Inquiry the Inquiry appeal FT/EA/2024/0267 from DN ref IC-306332-X9C9 of 17 June 2024. Dr Craig's appeal Guidance issued by the Information Commissioners Office the Guidance The Request

4. On 26 January 2024 (81) the Appellant made the Request to the UKHSA. This is set out in full in the Annex but in short it asked as follows:- "...Please can you provide, compiled or extracted from the raw data that I can see that you have previously confirmed the UKHSA holds (ref Case 04/08/23/LD/309), the following information: For each person aged 20 or over at their recorded date of death occurring between 1 January 2021 and 31 December 2023, whose data consequently is no longer included in the published NIMS database:

1. Age cohort for first dose of Covid vaccination (please use the age cohorts already used for NIMS purposes)

2. Date of each dose of Covid vaccination

3. Date of death"

5. The UKHSA, after providing some information, relied on the exemption at section 38(1)(a) FOIA. Subsequently to this was added reliance on section 12 or alternatively 14 and section 36(2)(c) with 36(4) and also section 41. As appears below we have concluded that section 38 FOIA applies and so have not gone on to consider sections 36 or 41. We have also concluded that section 12 FOIA applies and so have not gone on to consider section 14. Evidence and matters considered

6. We had the Bundle of 186 pages and a closed bundle of 10 pages. Additionally we had:- (a) a 6 page confidential bundle. (b) a 505 page bundle of legal authorities. (c) the Appellant's skeleton argument of 12 June 2025 (in addition to that of 22 January 2025). (d) the UKHSA's skeleton argument of the 17 June 2025. (e) a gist of the closed hearing of 24 June 2025 sent to the parties on 1 July 2025 and a gist of the closed material. (f) the UKHSA's supplementary submissions dated 22 July 2025. (g) an updated version of the Appellant's skeleton argument.

7. Professor Steven Riley and Dr Vahe Nafilyan provided witness statements (175-184) and were in attendance. Professor Riley said (176):- "1.2 I am a Director General and Chief Data Officer for UKHSA and a member of the Executive Committee, reporting to UKHSA's Chief Executive Officer. I lead the Chief Data Officer group within UKHSA. Teams within this group help UKHSA manage the process of obtaining, curating, and analysing data. Additionally, I serve as the Senior Information Risk Officer (SIRO). I have held this or similar roles since the establishment of UKHSA in October 2021. 1.3 I am seconded to UKHSA from Imperial College London, where I am a professor of infectious disease dynamics. My research focuses on respiratory viruses with pandemic potential. In my work, I utilise both primary and secondary data and have helped to establish several cohort studies. As part of my research, I have led analysis of vaccine effectiveness against the transmission of SARS-CoV-2 using data from the REACT-1 study."

8. Dr Nafilyan said (181):- "1. I am a senior statistician employed by the Office for National Statistics (“the ONS”) as the Deputy Director heading up the Health Division. I am responsible for the collection and curation of mortality data, the production of mortality statistics and the epidemiological analyses based on these data. Over the course of the pandemic, I have led numerous publications, both as statistical bulletins on the ONS website and peer-reviewed articles in epidemiology and medical journals. My work has been used by the Scientific Advisory Group for Emergencies (“SAGE”), the Chief Medical Officer, the Cabinet Office, the NHS and the Department for Health and Social Care. "

9. The Appellant indicated that he did not agree everything said in their statements but he did not cross examine either witness. The Confidential Bundle

10. The Tribunal had a small bundle marked "confidential". This included correspondence addressed to the Tribunal about this appeal from Members or former Members of Parliament. The writers of these letters said to the Tribunal for example:- "We wish to confirm for the benefit of the Tribunal that, as serving and former Members of Parliament with an interest in this topic, we consider that transparency for this important data is a matter of significant public interest, as it was already in February 2024"

11. For the reasons set out in the Directions of 24 June 2025 an d h aving considered the Parliamentary Commissioner for Standards note on "MP's Writing to Judges " dated 24 February 2023 this bundle was provided to all parties who were also invited to make any brief comments thought necessary. No comments have been received. This material has not been considered as part of this appeal and has not in any way influenced the Tribunal or this Decision. Background and parties

12. The UKHSA was established in 2021 as part of the UK's response to the Covid-19 pandemic. It is an executive agency of the UK Government and is sponsored by DHSC. Professor Riley in his statement says that (176):- "1.1...UKHSA prevents, prepares for and responds to infectious diseases, and environmental hazards. It provides scientific and operational leadership, working with local, national and international partners to protect the public’s health and build the nation’s health security capability."

13. NIMS is a national vaccine register and in its response (143) t he UKHSA said that it was created "...to record COVID-19 vaccine delivery across England, monitor vaccine coverage, and assess vaccine effectiveness and safety."

14. Dr Nafilyan says in his statement (181):- "2 The ONS is the executive office of the UK Statistics Authority, the UKs largest independent producer of official statistics and National Statistical Institute. The ONS is responsible for collecting and publishing statistics related to the economy, population and society at national regional and local levels..."

3. ONS’s role during the COVID-19 pandemic was to inform decision-makers and the general public with regular data and analytical insights across the economic, social and health themes. ONS published statistics derived from both survey sources, such as its own Covid Infection Study, and from administrative data provided by other departments, such as death registration data from the General Register Office."

15. Professor Riley says in his statement that (179) the data held by UKHSA is also held by the ONS but the UKHSA obtained it from the General Register Office via NHS England.

16. In summary, the Appellant seeks information about deceased people whose details were once included in the NIMS database but removed from it between 2021 and 2023. In his appeal he said (33):- "Towards the end of 2023 I became aware of the UKHSA’s refusal of an FOIA request submitted by a medical professional who had sought anonymised mortality and dosage data that, I understood, could be relevant to an independent expert analysis of an unexplained rise in excess deaths across the UK. As a legal professional, having reviewed the available public documents, including the UKHSA’s explanation of its refusal to release the data which it confirmed that it held, it seemed to me that the UKHSA’s refusal was not legally justified."

17. At the CVP hearing the Appellant also said that he had a personal connection to the issues and that his purpose for seeking disclosure was because in his view the relevant information should be disclosed from a legal perspective. He said that it was not right for a public authority to resist disclosure for what he thinks are policy but not legal reasons. He saw the UKHSA's response as an attempt to suppress data it would rather not have to explain because of their view that the public is not able to understand it.

18. The Appellant was asked about the Inquiry. He said that it had its "hands tied" in some ways including in module 4 on vaccines. He said that the Inquiry was not the place for this issue to be investigated and resolved. This topic was also raised in the Appellant's reply where he said (172):- "32...Though the Covid Inquiry acknowledged that vaccine injuries have occurred, and in module 4 resolved to look at the regime for approving and monitoring the deployment of Covid vaccines, it has expressly not considered any questions about the safety and efficacy profile of the Covid vaccines which could assist those who have suffered serious or fatal injuries. Hugo Keith KC, lead counsel for the Inquiry, said exactly this in his preliminary address at the start of module 4 of the Inquiry on 22 May 2024..."

19. The Appellant said that the possibility that someone somewhere would misunderstand, misrepresent or misuse information put into the public domain was a risk for all information given to the public. However, he said that this information is not needed if one was intent on creating a misrepresentation. He added that if these were valid reasons for a public authority to defeat a request it would make FOIA pointless. In his view if a public authority feared the misuse of data it should be ahead of the request and make it public with explanations. As regards the Request he took the view that if the data were to reveal correlations (say between the vaccine and harms) the answer would not be to say, it could not be disclosed, but to disclose it with whatever explanation was appropriate.

20. In relation to motivation the panel also asked the Appellant whether he was able to satisfy himself from the aggregated data already provided about a potential correlation or theme. He said not. He said that even as a non-statistician he could see that the level of aggregation of the data provided caused it to lose the necessary granularity that would make it capable of being analysed effectively. For the Appellant the utility of the data and his motivation for the Request are linked. He explained that he is motivated to have the data available so that professionals such as Dr Craig and others are able to analyse it, for personal reasons and as someone interested in the theme of accountability. In his skeleton the Appellant said:- "2. This appeal concerns transparency for public health data. I am pursuing it in the public interest and in support of the thousands of families affected by serious adverse consequences (including the death of a loved one) believed potentially to have been caused by a Covid vaccine dose. My hope is that transparency for this data set could assist those families to get closer to understanding, or at the very least ruling in or out, possible or probable causes for their losses. It is also possible that transparency in this case could reveal evidence that might form the basis for future civil claims against public authorities and/or private corporations for remedies which those families are not yet able to access (I shall comment later on the very limited remedy offered by the Vaccine Damage Payment Scheme). For that reason, I believe this may be a paradigm example of a public interest in transparency for official data for which the FOIA framework was envisaged. Barnardisation

21. Barnardisation is referred to in this appeal. It is a technique designed to enable statistical data to be published without the risk of anonymity being undermined or at least achieving the lowering of that risk. This is done by altering data for example by adding or subtracting values from the data concerned. In the Request the Appellant said (81):- "Personal data is defined in legislation as information relating to an identified or identifiable "living" individual. As such, it would seem that by definition the data I have requested, relating only to deceased people, is not personal data. If despite this you consider that it would still be necessary to ensure that disclosure of this data would not breach data protection principles, please compile the dataset I have requested using a Barnardisation process which modifies the date of death for each individual by adding or subtracting (for example) one, two or zero days from the date of death recorded in your raw data thus preventing the identification of any individual."

22. Professor Riley says in his statement (177):- 4.1 Barnardisation is a data perturbation technique that protects privacy by modifying counts in tabular data. Counts are randomly altered by adding 1, subtracting 1, or leaving them unchanged, with probabilities p/2, p/2, and 1-p. The value of p ranges from 0 to 1; higher values increase the number of perturbed counts, making it difficult to distinguish modified counts from reliable ones. 4.2 Deductive disclosure risks often arise from small cell counts in statistical tables. These categories can result in a small population denominator, making it easier to re-identify individuals. A limitation of Barnardisation is that small cells remain small and susceptible to re-identification by malicious actors, even if perturbed. 4.3 The request is not classical Barnardisation. It asks for a list of deaths with age at first dose, vaccination dates, and death dates perturbed by randomly adding or subtracting integers. This does not guarantee anonymity."

23. The Appellant in his reply to the UKHSA (C167) said that in the Request he only suggested a Barnardisation process as an example if thought necessary to protect against re-identification. He said that the UKHSA had:- "...subsequently sought to present my suggestion as a rigid prescription for a specific form of Barnardisation to be carried out (paragraph 15 of the UKHSA Response). This has been used to suggest that it would not now be reasonable for the UKHSA to be required to apply any other form of Barnardisation (paragraphs 15, 25 and 91 of the UKHSA Response).

9. My FOIA request did not prescribe that a specific form of Barnardisation should be applied to the requested data, it simply recognised that the UKHSA might consider Barnardisation necessary in order to protect against re-identification risk, and gave one example of how Barnardisation might be carried out." Role of the Tribunal

24. It is important to note that while the background is helpful to put the appeal into context and may be relevant for example when considering the PIBT the Tribunal's role in an appeal by section 57 FOIA is as set out in section 58 FOIA which provides that:- (1) If on an appeal under section 57 the Tribunal considers— (a) that the notice against which the appeal is brought is not in accordance with the law, or (b) to the extent that the notice involved an exercise of discretion by the Commissioner, that he ought to have exercised his discretion differently, the Tribunal shall allow the appeal or substitute such other notice as could have been served by the Commissioner; and in any other case the Tribunal shall dismiss the appeal. (2) On such an appeal, the Tribunal may review any finding of fact on which the notice in question was based.

25. The Tribunal exercises a full merits appellate jurisdiction and so stands in the shoes of the IC and decides which (if any) exemptions apply ( NHS England -v- Information Commissioner and Dean [2019] UKUT 145 (ACC) para 10). In Peter Wilson -v- The Information Commissioner [2022] UKFTT 0149 the FtT said:- "...the Tribunal’s statutory role is to consider whether there is an error of law or inappropriate exercise of discretion in the Decision Notice. The Tribunal may not allow an appeal simply because it disagrees with the Information Commissioner’s Decision. It is also not the Tribunal’s role to conduct a procedural review of the Information Commissioner’s decision making process or to correct the drafting of the Decision Notice.”

26. On burden of proof in Forstater v Information Commissioner and others [2023] UKUT 303 (AAC) the UT at para 40 said:- (2) the “ordinary presumption” is that it is for an appellant to prove their case. The burden will rest with the appellant except where statute expressly or impliedly provides otherwise: Khan v Custom and Excise Commissioners [2006] EWCA Civ 89 ; [2006] STC 1167 at [73.7]. Neither FOIA nor the Tribunal Procedure (First-tier Tribunal) (General Regulatory Chamber) Rules 2008 contain any express provision about the burden of proof and neither by implication remove the ‘ordinary presumption’ either; and (3) however, the concept of the burden of proof is of secondary importance in tribunal proceedings which involve a full merits review, since to apply strict burdens of proof may prevent the tribunal from properly discharging its responsibility to decide the facts for itself and/or exercise any discretion afresh:..." FOIA

27. FOIA provides that any person making a request for information to a public authority is entitled to be informed in writing if that information is held (section 1(1) (a) FOIA) and if that is the case to be provided with that information (section 1 (1) (b) FOIA). Disclosure of information pursuant to FOIA is "to the world." These entitlements are subject to a number of exemptions which can be absolute or subject to the PIBT namely that “in all the circumstances of the case, the public interest in maintaining the exemption outweighs the public interest in disclosing the information.” Section 38 FOIA

28. Section 38(1)(a) FOIA is subject to the PIBT and provides as follows:- "(1) Information is exempt information if its disclosure under this Act would, or would be likely to— (a)endanger the physical or mental health of any individual, or (b)endanger the safety of any individual."

29. For the exemption to be engaged the disclosure and the endangerment are causally linked and disclosure would be likely to endanger the mental health of any individual. In Lownie v Information Commissioner & The National Archives & The Foreign and Commonwealth Office EA/2017/0087 the FtT said that the words "would be likely "... refers to there being ‘a very significant and weighty chance’ of occurrence, such that the occurrence ‘may very well’ occur, a ‘real risk’ not being enough."

30. The FtT in Lownie also considered the test for mental health and said:- " 49. The Information Commissioner’s position appeared to be, at least at times, that the requisite probability of ‘distress’ to living relatives would be sufficient to meet the requirements of s 38. If that was what was meant, we do not agree. While distress can be a trigger leading to mental ill‐health, we do not consider that distress in itself should be equated with mental ill‐health for the purposes of s 38. A healthy or unhealthy person may experience distress without suffering any, or any additional, mental ill‐health."

31. As regards the type of evidence required in PETA v The ICO & University of Oxford EA/2009/0076 at para 31 the FtT said:- "... It was suggested by PETA that for the Tribunal to be satisfied that there was a danger to mental health that positive evidence from e.g. a psychiatrist as to the clinical impact of the campaign upon the mental health of those affected would be necessary. The Tribunal rejected that contention and was satisfied that the level and nature of the physical threat was sufficient that on a balance of probabilities the effect upon the mental health of those involved would go beyond stress or worry and constitute an endangerment to their mental health."

32. In Calllus v The IC EA/2013/0159 the FtT said:- "60. We have previously explained why we accept that there is a risk of SPoC’s being approached by criminals or others intent on undermining the effectiveness of the RIPA regime. The evidence that the approach may be accompanied by actual or threatened violence, or that a refusal to co-operate may result in violence, was less well supported by evidence. We felt that the witnesses engaged in a degree of speculation in this part of their evidence. The conclusion they feared, however, is one that would flow logically from initiating communications designed to corrupt the relevant official. It is therefore justifiable to regard it is a result that would be likely to flow from the disclosure of names and to conclude that the exemption is engaged."

33. If the exemption is engaged there is then an assessment of the PIBT to conclude whether the public interest in maintaining the exemption outweighs the public interest in disclosure. The relevant date for considering the PIBT is the date the public authority makes its decision on a request which in this matter is 30 May 2024 (see M ontague v ICO and Department for Business and Trade [ 2022] UKUT 104 (AAC )). Section 12(1) FOIA

34. Section 12 (1) (which is not subject to the PIBT) provides as follows:- (1) Section 1(1) does not oblige a public authority to comply with a request for information if the authority estimates that the cost of complying with the request would exceed the appropriate limit.

35. The appropriate limit is set by the 2004 Regs at £600 and the estimated time cost allowable for the tasks in reg 4(3) 2004 Regs is set at £25 per hour. A public authority can only take account of the costs it reasonably expects to incur in relation to a request in carrying the activity specified in the 2004 Regs as further considered in legal authorities such Chief Constable of South Yorkshire Police v Information Commissioner [2011] EWHC 44 (Admin) . We were also referred for example to the decision of the:- (a) UT in Reuben Kirkham v Information Commissioner (Section 12 of FOIA) [2018] UKUT 126 (AAC) (b) FtT in Esslemont -v-The Information Commissioner and Home Office (EA/2020/0008V) in particular paras 39- 44 (c) FtT in Randall v Information Commissioner & Medicines and Healthcare Products Regulatory Agency’ EA/2007/0004 Request to appeal (overview)

36. On 26 January 2024 (81) the Appellant made the Request to the UKHSA. On 30 May 2024 the UKHSA (82) said that they held the information requested. They relied on section 38(1)(a) FOIA but provided some information (87). The UKHSA said (82):- "Given the public interest in this information, we hope that releasing the attached anonymised table, which provides the requested data presented by age band and by week of vaccination and month of death, will satisfy your request." and it provided an explanation of the data saying:- "We have provided additional important below about how the data should be interpreted, to avoid any misinterpretation of the data. We hope this is of help and that you will find how we have presented the data in our disclosure acceptable for your requirements. This dataset contains individuals: • Aged 20 years old and over •That have been vaccinated with at least 1 dose of any Covid-19 vaccine between the 8 December 2020 and the 31 August 2023 • Have a recorded death date up until the 31 August 2023. • Were a resident of England or not confirmable as being a resident of any other country (i.e. possibly England) on the 31 August 2023

37. The UKHSA added an explanation of the data as follows:- " The columns include: • “Age Group At First Dose” - The age group of the person when their first dose (“Dose Number” = 1) of any covid-19 vaccine was administered (range of 5 or 80+) • “Dose Number” – The dose number the record is referring to • “Dose Date Administered Week Commencing” – The Monday commencing the week of the administration of the vaccine “Dose Number” specified • “Date of Death by calendar month” • “Country” – The individuals most recent country of residence (not necessarily at time of vaccination) as of the 31 August 2023. Those recorded as “Not Confirmable” had a postcode but were not able to be linked to a wider region (e.g. transcription error in the postcode, but possibly still a resident of England), and those with “Unknown” did not have any postcode information recorded. • RecordCount- Indicates the number of unique individuals to whom the details in the row apply and the UKHSA went on to add their own commentary on the data.

38. The UKHSA did not provide the actual date of death but only the month of death to avoid the risk that publication to the world could result in the identification of a deceased person referred to in the data.

39. On 4 June 2024 the Appellant asked for an internal review (88). In it he said for example:- "In any event, I write to confirm that your production of a partial data set for which I did not ask is not capable of satisfying my more specific request under the FOI Act. As doubtless the UKHSA is aware, providing mortality data which displays ‘week of vaccination dose’ alongside ‘month of death’ denies the possibility of meaningful analysis of potential trends, patterns or anomalies arising between the two data points"

40. We understand that no review was provided and on 22 June 2024 (92) the Appellant complained to the IC. The IC issued the DN on 22 August 2024 (5) and said:- "2. The Commissioner’s decision is that UKHSA was entitled to rely on section 38(1)(a) of FOIA to withhold some of the requested information. The Commissioner also finds that UKHSA breached section 10(1) and 17(1) of FOIA by failing to communicate the requested information or issue a section 38 refusal notice within 20 working days."

41. On 6 September 2024 the Appellant commenced the appeal (20-32) supported by his submissions (33-56). The outcome sought is (28):- "I ask the Tribunal please to allow my appeal of the ICO's decision notice and to substitute a notice that orders the UKHSA to provide to me the anonymised data set that I requested, which could be compiled on the modified Barnardised basis described in my submission if the UKHSA confirms this is necessary"

42. The IC responded (57- 64) on 11 October 2024. The Appellant then replied (65 – 75) on 24 October 2024. By Directions made on 28 January 2025 (134) this appeal was linked to Dr Craig's appeal and the UKHSA was joined as a party. A response was provided by the UKHSA on 28 March 2025 (140-162) and on 9 April 2025 the Appellant provided a reply (163-174). Professor Riley and Dr Nafilyan provided witness statements on 8 May 2025 (175) and 21 May 2025 (181) respectively. The Closed Hearing

43. As part of the CVP appeal we held a closed hearing. A gist of this was prepared and provided to the parties on 29 July 2025. Summary of the position of the parties on section 38 FOIA Engagement

44. In summary the UKHSA's case is that:- (a) responding with raw data to the Request and even after a process of barnardsiation created a real risk of there being an ability to identify a person whose details were in that data. (b) disclosure of this information with identification in this way to the world would be likely to endanger the mental health of any individual directly and more broadly.

45. When the UKHSA replied to the Request it said that it:- "... considers that disclosure of the requested data may have negative consequences on the mental health of the families of those deceased individuals to whom this data relates. As such, we have deliberated a public interest test in respect of the exemption at section 38(1 ) (a) of the Act . UKHSA trusts that you will understand our decision to present the requested data as described above to mitigate our concerns regarding the families of the deceased."

46. The IC in the DN (8) relied on the content of the DN in the Dr Craig appeal but as a summary listed the issues raised by the UKHSA namely that disclosure:- (a) would present a real risk to the mental health of those linked to the deceased individuals featured in the information (b) would result in a loss of confidence in UKHSA’s ability to protect the well-being of the families and friends of those who have died (c) would risk individuals being identified. UKHSA found that presenting the information by both date of death and week of death could still lead to some individuals being identifiable (c) could lead to the families of the deceased being subjected to unsolicited approaches by the media and anti-vaccine campaigners (d) could lead to misinformation that would have adverse impact on vaccine uptake in the wider public, particularly vulnerable groups.

47. The IC accepted that there was a risk of identification but said that in his view while this would be likely to cause distress it would be difficult to say if that alone would extend to causing harm to the level needed to engage section 38. However the IC did take the view that because of the worry about harm more widely due to the impact on public health caused by misinformation the exemption was engaged.

48. Dr Nafilyn in his evidence (182) said that the ONS was asked to carry out:- "8...penetration testing of a dataset that had been prepared for potential release...to determine whether or not individuals within the dataset remained identifiable following measures taken against re-identification".

49. He said that a number of ONS staff acted as "friendly intruders " to see if this could be done using the data set and other information in the public domain. He said that the result was that:- "10. Using a straightforward internet search with keywords “COVID”, “vaccination”, “deaths” and “England” along with the relevant time period, many cases were found on various websites, notably on the local BBC websites. Many referred to small counts (lower than 5) of individuals on the dataset and one referred to a count of 1. Hence ONS could demonstrate that at least one named individual could be identified in the dataset and, given the short amount of time to do this with limited resources it is very likely that there could be many more identifications had we invested more time and resources in the exercise."

50. He also gave evidence that:- "11. The use of Barnardisation (adding or subtracting a small amount from some values) for the dates here would, in ONS view, be insufficient to protect the one individual – and probably many others, given the perturbation required would not make them clearly indistinct from other potential cases. An intruder would be aware that the method had been used (though not necessarily the parameters of it) and would, with likely high probability, be able to match individual cases with other public sources, typically media reports"

51. In support of this he referred to the use of bardnardisation by the ONS in the 1991 census which had proved insufficient as protection especially for rare events.

52. Professor Riley said (178):- "4.2 The request is not classical Barnardisation. It asks for a list of deaths with age at first dose, vaccination dates, and death dates perturbed by randomly adding or subtracting integers. This does not guarantee anonymity." and that it was his view that (177):- "3.1... disclosure of these data would likely lead to mental health harm to individuals. There were high rates of PTSD symptoms in those who suffered COVID-19 bereavement. Given that these data relate to 1,248,581deaths, it is entirely foreseeable that a motivated intruder could identify individual relatives and make unsolicited approaches based on spurious findings, suggesting a link between vaccination and the death of their late relative"

53. He broadened this risk to include his view that:- "3.2 There is substantial potential harm to vaccine confidence from these data being released. With only minimal potential benefits in terms of additional evidence (see above), there is potential harm to individuals from a reduction in vaccine uptake as a result of these data being released. 3.3 Scares about vaccine safety can damage confidence and lead to long-term declines in coverage. For instance, whooping cough coverage dropped to 30% in the 1970s, while MMR coverage fell below 80% in the late 1990s and early 2000s. These declines have serious consequences, including increased cases, admissions, and deaths, exemplified by the severity of whooping cough before and after the decline in immunisation. Drops in coverage also generate the need for expensive catch-up campaigns."

54. In the appeal the issues raised by the Appellant on the engagement of section 38 FOIA included:- (a) that the UKHSA had not identified any causal link between disclosure of the information requested and a "very significant and weighty chance” of endangering the mental health of individuals. (b) that if the data when disclosed did not "reveal any concerning correlations, or other patterns, as between dosage dates and dates of death...it would be implausible to assert that giving transparency to that data would be very likely to cause psychological disorders among the families of deceased individuals who already know the fact and date of the death of their relative" (c) that if the data "...indicates potentially concerning correlations, or other patterns, which might suggest for example the possibility of a link between the dates of dosage and death in some or all cohorts, one might fairly anticipate that bereaved relatives might then experience distress or anger. However he said that " Such feelings are not though treated as psychological disorders for the purposes of s.38 of the FOI Act and, more importantly and sympathetically, the cause of any distress or anger would be the identification of a possible correlation rather than the fact of the disclosure of the underlying data."

55. This point had been raised by the Appellant previously (49) where he had added :- "More purposively, it would anyway be legally and morally perverse for the UKHSA to argue that the public should be denied access to official public health data that could reveal concerning correlations of this nature on the basis that some portion of the public might become distressed or angry about the possible existence of those correlations. It would be deeply offensive for the UKHSA to make any such assertion so soon after the Infected Blood Scandal report findings have been published"

56. The Appellant also argued against use of this exemption if it was due to a concern about the accuracy of the data or a risk that the public would misled by it. He said:- "The public is not clever enough to understand these data’ or ‘There might be someone somewhere who will misuse or misrepresent these data’ are not lawful bases for resisting disclosure under the FOI Act, and to countenance such a broad interpretation of any exemption under the Act would be a slippery slope and completely at odds with the purpose of the Act ."

57. He also said that the DN had reached an unlawful erroneous conclusion because the IC had conflated:- "A. the possibility that, once the data has been disclosed, someone in the public at large might seek to misuse the data, and that this might result in misinformation which might in turn “cause potential harm to the health of the wider public” and negatively impact on confidence in public health; with B. the disclosure of the data under the Act would be likely to (i.e. that there is a significant and weighty chance that it would) endanger the physical or mental health of individuals." Section 38(1)(a) is not ambiguous in its use of language. It is clear that in order for this exemption to be engaged, it must be established that there is ‘a very significant and weighty chance’ that harm would be caused by the disclosure" The UKHSA has asserted, and the Commissioner accepted, only that unspecified harm might be caused as a result of unidentified third party incompetence or malicious intent at some point after the data had been disclosed for otherwise legitimate and lawful uses. On this basis it is not plausible for the Commissioner to have accepted that s.38(1)(a) has been engaged. If instances of confirmed misinformation (or indeed disinformation) at some later date were to emerge, the appropriate and FOI Act-consistent remedy would be for the UKHSA or other public authorities to publish and promote their own better and more accurate explanations of the data at that time"

58. The UKHSA in its response to this appeal said that while no suggestion was being made that this applied to the Appellant it was concerned:- (a) about the use of the data (in the form requested) by a motivated intruder to identify and contact a bereaved family (b) that such contact without warning " would be likely to have a significant negative impact on mental health of those individuals who have been bereaved." (c) that there was a chance of there being an unsolicited approach to a bereaved and vulnerable person with a suggestion of a link between death and vaccination and that there was a significant and weighty chance of this " endangering the mental health of that person, beyond distress." ( d ) about the risk that harm would arise by disclosure because relatives would not have expected that UKHSA would place such information into the public domain.

59. The UKHSA referred to the very large size of the data set being requested to reach its conclusion that such risks were likely. It also referred to what it called "The spread of mis/disinformation related to the COVID-19 vaccine" which it said was "...not a hypothetical concern." The UKHSA said that it is "...currently monitoring the spread of mis/disinformation related to the COVID-19 vaccine. UKHSA considers that this issue is leading to the ongoing erosion of vaccine confidence in small groups and communities who are already distrustful of the vaccination programmes and receptive to misinformation." The PIBT

60. The parties put forward submissions for the maintenance of the PIBT and for disclosure. In the Appellant's view, as seen in his skeleton, the IC had not carried out this exercise properly and had not taken into account matters raised by the Appellant when he replied to the UKHSA's response (see 73).

61. The arguments in favour of disclosure included:- (a) because the UKHSA and IC were wrong in identifying the risk of disclosure for the purposes of the engagement of the exemption (b) the public interest in increased transparency and in accountability which can serve to increase trust in public institutions. (c) the public interest in there being confidence "in the decisions that are made regarding vaccines and that these decisions are taken based on the best available information." (d) to enable relatives of those who have died following suspected vaccine complications to have an opportunity to understand whether "the death of their relative was an exceptional anomaly or whether it could potentially have been part of a wider pattern for which some form of recognition or accountability might be sought" (e) to enable those who believe they have suffered injury from a Covid-19 vaccine and their wider family to understand the "fullest range of possible explanations for their suffering, and in having the opportunity to rule out or in the possibility that official safety data suggests one or more of their Covid vaccine doses could have played a role" (f) because the vast majority of the UK population had received at least one vaccine dose. (g) because the Inquiry is not tasked with seeking the same information. (h) to avoid there being the damage to the public interest described in the Infected Blood Inquiry Report as "...institutional defensiveness, from the NHS and in particular from government, compounded by groupthink amongst civil servants and ministers, and a lack of transparency and candour." which " drove the response of government over the decades. The institutional defensiveness identified above is damaging to the public interest " (i) to avoid the erosion of vaccine confidence caused by the lack of transparency.

62. Arguments in favour of maintenance of the exemption included:- (a) the public interest in avoidance of the harm referred to when considering the engagement of the exemption (about which the Appellant disagrees) from which it follows that disclosure would only be justified if there were compelling reasons. (b) the public interest in preventing the risk of the data being presented out of context enabling "malicious actors" to undermine public health generally but also " create harmful narratives that would have a detrimental effect on the surviving family members of the deceased." (17) (c) t he public interest in preventing harm caused by " anti-vaccine theorists or campaigners in their efforts encourage members of the public to abstain from vaccination against COVID-19. Therefore, it considers that there is a risk that disclosure of the information in the requested detail would be likely to undermine public adherence in current and future COVID-19 vaccine programs"(18) (d) because of Professor Riley's evidence (176) that the risks of disclosure were substantial but any scientific benefit was negligible and "2.1 Even though this is a large dataset, it has very low utility for the study of adverse effects of vaccination. There is no detail about the cause of death and these data arise from a population with very high levels of vaccination. It seems very unlikely to me that an erstwhile unobserved risk of vaccination could be reliably detected in these data." (e) because of Dr Nafilyan's evidence that (183):-. "13. In all likelihood, the requested data would only permit analysis of limited value from a public health perspective, and the methodological limitations of such an analysis would mean that any inferences are at substantial risk of being over-interpreted and therefore misused." and "18. In ONS view therefore the requested data are not suitable for assessing the relationship between COVID-19 vaccination and mortality, and any analysis performed on data with such limitations risk the drawing of invalid conclusions and the spread of misinformation that could harm public health." ( f ) because the public interest has been lowered or satisfied by the existing provision of information to the Appellant by the UKHSA. (g) because there are very large numbers of people whose family member's data forms part of the requested information and if it was disclosed to the world unless total anonymity could be guaranteed that risked lowering the public's confidence in the way data was held and shared. (h) because the data the subject of the Request when held by the ONS is subject to section 29 of the 2007 Act which if breached can amount to a criminal offence. (i) because (although the ONS did not provide the data to the UKHSA) it could lead to there being an undermining of the relationship for data sharing between ONS and the UKHSA. (j) because while it is in the public interest to seek to assist families with an understanding of what caused the death of a family member and/or compensation such aims can be achieved to some degree in other ways. Those cited were by the holding of an inquest, by the vaccine damage payment scheme and by The Access to Health Records Act 1990 . (k) because of the work being done by the Inquiry and Professor Riley's evidence that (177):- "3.4...Module 4 ‘vaccines and therapeutics’ addresses “issues of recent public concern relating to vaccine safety and the current system for financial redress under the UK Vaccine Damage Payment Scheme.” Any public interest in assessing vaccine safety is therefore already amply served through appropriate and responsible studies and investigations (which will incorporate procedures for protecting privacy etc), as contrasted with misleading narratives around a data set not suited to establishing any relationship between vaccination and death."

63. As can be seen from the gist during the closed hearing the Tribunal were told that the ONS had a Secure Research Service ("SRS") which "provides a secure means for accredited/approved researchers to access de-identified, unpublished data in order to work on accredited research projects for the public good." As the gist says the Tribunal were told that:- "9. The available dataset via SRS (without Barnardisation or aggregation applied) supplements the data requested by the Appellants; and provides greater research utility. This dataset contains person-level data for individuals in England who have died between 8 Dec 2020 and 27 Feb 2023. The dataset includes age at death, underlying cause of death and contributory conditions, sex of the deceased, date the death occurred, date of death registration and Covid-19 vaccination dose type and date for doses 1-6. These data were curated by linking National Immunization Management Information data to death registration data by the deceased NHS number. Access to the data is restricted to COVID19 related research projects."

64. Counsel for the UKHSA (having taken instructions) said that " many researchers have requested and had access to a version of this raw dataset". However as seen in a footnote this was inaccurate and the answer was clarified to say that the SRS had not been accessed for research purposes. On the basis of the SRS it was the UKHSA's submission that:- "13...there exists an alternative means of researchers accessing an enhanced/expanded version of the data sought by the Appellant, outside FOIA, which is subject to safeguards that sufficiently mitigate the risks identified by the Respondents (of a FOIA release to the world)

65. The Appellant set out his view of this in his updated skeleton having seen the gist. He said:- "46...The SRS is a red herring insofar as the data set described is different, and most importantly because access to the SRS is discretionary and expressly not available to the public. The SRS data set therefore is not in any sense reasonably accessible to me, as a non-academic. I note that the UKHSA has not sought to rely on s.21 FOIA (information accessible by other means) to resist disclosure."

66. As regards the balance of the public interest the IC in the DN placed significant weight on preventing the risk of harm to any individual as envisaged by section 38 FOIA. The DN said (18):- "42. The Commissioner has determined that the strength of the arguments favouring disclosure is outweighed by the public interest in maintaining the exemption in order to safeguard individuals’ health and safety. He considers that the information UKHSA has disclosed satisfies the general public interest in transparency to a satisfactory degree. 43.

67. The Appellant in his appeal said (40):- "If a public interest were to be considered relevant I suggest that these data evidently are the subject of very significant public interest and the balance would weigh strongly in favour of disclosure " and in his skeleton said:- "29. The balance of the public interest (in relation both to s.38 and any other applicable FOIA exemption) is overwhelming in favour of transparency, and of the further interrogations and explanations of that data which transparency will enable" The Tribunal's review engagement of section 38 FOIA.

68. There has been a considerable amount said about bardnardisation or other forms of aggregating or manipulating data. We agree with the Appellant that the Request did not require barnardised information but what he said after making the Request was:- "If despite this you consider that it would still be necessary to ensure that disclosure of this data would not breach data protection principles, please compile the dataset I have requested using a Barnardisation process..."

69. The starting point therefore should have been to consider whether the 38 FOIA exemption applied to disclosure of the information in scope of the Request without any form of manipulation. However mostly (with the exception of the Appellant) there has been an assumption that the Request pre-supposed that some sort of disguise of the information would be required. Ultimately this does not present an issue for the Appellant because as regards section 38 FOIA the UKHSA has a more difficult task of demonstrating that barnardised material should be withheld as compared with the raw data.

70. Reference was made to barnardisation in the House of Lords Judgment in Common Services Agency v Scottish Information Commissioner (Scotland) [2008] UKHL 47 (9 July 2008). I n his Judgment Lord Rodger said it was "52.. one method of reducing the risk of identification. It does not guarantee that the risk will be eliminated."

71. From the evidence of the testing done and because of the size of the data set (involving 1,248,581 deaths) we accept the evidence from Professor Riley and Dr Nafilyan that if the UKHSA were to respond to the Request even with barnardised data there would remain a risk of (re)identification. Given this volume of data we accept that someone could be re-identified and this view was reinforced by having seen an actual example in the closed material of the identity of a specific individual being revealed.

72. Covid-19 and the UK's response to it including the development of vaccines is an issue that arouses diverse views. We think it very probable that while it would not be the Appellant's desired outcome there would be very motivated intruders who, with relative ease, could and would identify names from the data and then could and would identify relatives. The numbers of people are so large and the debate in places so polarised that we accept the suggestion that there is a real risk that some people would be contacted about a deceased relative once the data set was published. We conclude that for some (but not for others) being approached about this data or finding out about it, would be distressing and if accompanied by any partisan view it would be very distressing.

73. Views differ about whether such distress would be likely to result in a mental heath issue of the type required for section 38 to be engaged. The IC for example thinks not but Professor Riley (177) based on PTSD rates says he thinks it would. As there has not been an actual release of this data there is no direct evidence to indicate the actual impact nor were we provided with any (if any existed) analogous historical evidence as to what had happened following a previous release of such data. However, while not binding we agreed with:- (a) the FtT in PETA where it was satisfied that:- " the level and nature of the physical threat was sufficient that on a balance of probabilities the effect upon the mental health of those involved would go beyond stress or worry and constitute an endangerment to their mental health." (b) the FtT in Callus where it said: - "...We felt that the witnesses engaged in a degree of speculation in this part of their evidence. The conclusion they feared, however, is one that would flow logically from initiating communications designed to corrupt the relevant official. It is therefore justifiable to regard it is a result that would be likely to flow from the disclosure of names and to conclude that the exemption is engaged."

74. From what was said and from the large numbers involved we concluded that disclosure would be likely to endanger the mental health of some individuals.

75. The UKHSA had a broader position (supported by the IC) that disclosure would also risk harms caused by there being a connection between misinformation created from this requested disclosure and the reduction in the uptake in vaccination more generally. This may or may not be right but to have been able to reach a conclusion that there was the necessary causal link between publication and the harms envisaged we would have needed much more focused evidence. As we had reached our conclusions set out above we did not consider it necessary to seek to reach a conclusion on this aspect.

76. Having concluded that the barnardised information was properly withheld by section 38 FOIA it is worth noting that our conclusion as a result must be that disclosure of the raw data without bardnardisation would clearly also have engaged section 38 FOIA. The PIBT

77. As at the date of response on 30 May 2024 ( Montague ) there was considerable well known, wide-spread public interest, concern, debate in all aspects of Covid-19 and the UK's response to it including the development, deployment, efficacy and safety of covid-19 vaccines. This is clearly not just an issue of what is of interest to members of the public but did amount, as said in the Guidance, to the public good.

78. We agreed with much of what was said by the parties but when weighing up the PIBT certain aspects are worth additional commentary as follows:- (a) as we have concluded that section 38 FOIA is engaged we accept that there was a strong public interest in seeking to prevent the risks identified and gave this aspect considerable weight. (b) there is a strong public interest in the public having information about the Covid-19 vaccination but this was at least partly satisfied because:- (i) of the data released to the world by the UKHSA in its response to the Request (ii) of the establishment of the Inquiry which by May 2024 had commenced even if by then module 4 about "vaccines and therapeutics" had not commenced. (iii) in a barnardised format (which as we conclude would be required) it would have little extra value over and above what had been provided by the UKHSA. (c) there is a strong public interest in enabling families to have a better understanding of what happened to a loved one or to them but this is partly satisfied due to the existence of other ways to seek the same sort of information at least on an individual if not wider basis. (d) we accept that by not publishing the information there is a risk that, against the public interest, some may interpret that as being evidence of institutional defensiveness and an attempt to avoid transparency because something, it might be said, is being withheld from public scrutiny. This however is at least somewhat balanced by the risk (also against the public interest) that if disclosed some may argue that its barnardised format is itself an attempt to hide something from the public. It is also reduced some way by the right of a requester to complain to the IC and then appeal. (e) we did not accept the concern that disclosure by the UKHSA (if it had considered it appropriate) could have undermined trust between the UKHSA and ONS or others. The UKHSA and the ONS are organisations involved in very serious issues and required, and we have no doubt able at both an organisational and individual level, to work constructively and collaboratively. They will know about FOIA. The UT in Davies v IC and The Cabinet Office [2019] UKUT 185 (AAC) said about the "chilling effect" in an analogous situation:- " “In judging the likely consequences of disclosure on officials’ future conduct, we are entitled to expect of them the courage and independence that has been the hallmark of our civil servants since the Northcote-Trevelyan reforms. These are highly-educated and politically sophisticated public servants who well understand the importance of their impartial role as counsellors to ministers of conflicting convictions. The most senior officials are frequently identified before select committees, putting forward their department’s position, whether or not it is their own.” ( f ) we were also not persuaded by the issue raised about the 2007 Act. That is a matter for the ONS who we are told did not actually provide the relevant information to the UKHSA. (g) in so far as it was asserted we gave some but little weight to any suggestion that the public interest favoured maintenance of the exemption because the data might be purposely or accidentally misinterpreted as that is a concern that could potentially be said about many FOIA requests. (h) judged on its own (and outside a consideration of section 14 FOIA) we gave little weight to the submission that what would be disclosed would have very low utility. The motive of a requester is not normally relevant and public authorities might frequently wonder about the utility of what is requested. However as part of the assessment of the balance of the PIBT we gave this some weight.

79. We did not agree with the Appellant's submissions about SRS but we have not considered it because we were not told and we do not know if the SRS portal was in existence with the relevant data when the response was sent. If we knew it had been in existence this would have added further weight to the public interest favouring maintenance of the exemption.

80. While all matters raised were considered (even if not all listed above) in reaching our conclusion our main consideration came down to a balance between a recognition of the considerable public interest in all aspects of Covid-19 and connected questions about vaccines set against preventing the harm envisaged in section 38 FOIA. That alone may have been enough, however, the existence of the data provided with the response (while not being what was requested) did, in our view go some way to satisfy the public interest and adds considerable weight to the arguments in favour of maintenance of the exemption. Summary of the position of the parties on section 12 FOIA

81. In its response the UKHSA said that "39...the raw data set in scope of the request—before any barnardisation was applied—is so large that it cannot be opened, much less processed, via an Excel spreadsheet or similar." The response at para 38 said:- "...UKHSA’s reasonably expects to incur in excess of £600 costs in respect of extracting the information requested from the data set, consisting of (a) the cost of applying the barnardisation to almost 4 million entries and (b) data validation and testing the effectiveness of barnardisation against each entry (estimated 10 minutes per entry or about 76 years total) or even against a sufficiently large sample to meet the requirements of relevant guidance/standards.

82. At para 5.1 of Professor Riley's statement he set out his cost calculation (178) which can be summarised as follows:- (a) 3 hours to find and be granted access to the data . (b) 9 hours to set up and run a process made up of 2 hours for coding, 2 hours to run, 2 hours for output generation and 3 hours for quality assurance. (c) individual-level testing to check results are non-disclosive involving 10 minutes per individual and " Assuming that 12 of the person hours have already been used up, we would only be able to examine another 72 randomly selected individuals before the 24-hour cost threshold is reached."

83. In his response to the UKHSA's position on costs the Appellant said (164) that if the UKHSA were right about their estimates then they could have "...dismissed my request in May 2024 on that single, straightforward, basis" and that section 12 arguments could and should have been raised much earlier. In his updated skeleton the Appellant refers to Esslemont and the South Yorkshire. He also said that:- (a) 3 hours to locate data which UKHSA says it has already shared on an anonymised basis is excessive. (b) he thinks the code will already have been written for other purposes. (c) tasks that have no human intervention cannot be claimed. (d) verification of data/ quality assurance cannot be claimed.

84. As regards the sample penetration testing he referred to South Yorkshire and said:- "37.3 In other words, the process of disentangling disclosable data from data which has not been requested (i.e. that data which it is agreed should be excluded by means of a small cell suppression technique), should not be included. Alternatively, pursuant to the points made above there would be more time available for testing than is asserted, and if the small cell suppression element of the Barnardisation process is calibrated to exclude cells with a value of less than, for example, three or four, penetration testing can then be carried out in a much more focussed way specifically on testing a random selection of cells with a value of four and/or five to validate the effectiveness of that suppression in addressing any reidentification risk " The Tribunal's review on section 12 FOIA.

85. Our primary decision is that section 38 FOIA applies to the Request because even if the information is barnardised the risks could not be excluded. Section 12 therefore only arises if we are wrong on section 38 and (subject to any other exemptions) the UKHSA has to carry out the task of preparing the barnardised material for disclosure. The 2004 Regs allow a public authority for the purposes of section 12 FOIA to take account only of the costs incurred in:- (a) determining whether it holds the information, (b) locating the information, or a document which may contain the information, (c) retrieving the information, or a document which may contain the information, and (d) extracting the information from a document containing it.

86. The costs taken into account can relate to human activity but need not do so as seen in reg 4(4) of the 2004 regs which states that:- "(4)To the extent to which any of the costs which a public authority takes into account are attributable to the time which persons undertaking any of the activities mentioned in paragraph (3) on behalf of the authority are expected to spend on those activities, those costs are to be estimated at a rate of £25 per person per hour."

87. From the 2004 Regs and the legal authorities it is our view that:- (a) the 3 hours to find and locate the data is claimable by reg 4(3)(b) of the 2004 Regs and although the Appellant considers this will already have been done for other purposes this challenge was not put to Professor Riley and we were not satisfied that what had been done before would have entirely removed the need for this task to be done afresh for this specific purpose. (b) the 2 hours of coding is claimable by reg 4(3)(c) and again while noting that the Appellant believes this will already have been done this challenge was not put to Professor Riley and we were not satisfied that what had been done before would have entirely removed the need for this task to be done afresh for this specific purpose. (c) the 4 hours for the process to run and output generation are not attributable to a time cost and so reg 4(4) of the 2004 Regs is not relevant however no other costs have been claimed.

88. The UKHSA claims 3 hours for quality assurance. In South Yorkshire Keith J held :- "28...There is no basis for giving the words extracting the information in regulation 4(3)(d) a wider meaning than that which would otherwise be appropriate simply because complying with a request for information may well involve the completion of other tasks as well. " and the Judgment listed a number of tasks which are not included such as "( iv) determining whether any of the information which has been requested is covered by an exemption in Part II..."

89. UKHSA says that quality assurance is part of the mechanical process of barnardisation to prevent identification (as far as possible) and it is not listed in para 29 of South Yorkshire and it is not analogous, for example, to checking if an exemption applies.

90. In Esslemont ( on which the UKHSA relies) the Home Office had included in its calculations for section 12 FOIA at stage 5 time for "Assurance and verification of the outputs" following the building of a database. Having considered para 29 of South Yorkshire the FtT said:- "43... The Commissioner’s skeleton argument comments that although some degree of checking is involved in any extraction process the HO is required to justify the amount of “verification” and “quality assurance” included in its estimate as reasonable. The Commissioner notes that checking and verification would have been continuing throughout the process from the creation of the search criteria onwards. The Commissioner does not consider this activity to fall within the permitted activities and therefore the estimate is reduced by a further 4.5 hours. "44. However, we note that the checks that occur under Stage 5 are necessary, according to the HO, because ‘Given that this would be a complex and new dataset, it would require running multiple iterations of the data to ensure the database is reporting the data correctly. It would also require comparisons to previous related datasets, checking samples of records and confirming figures are as expected with operational colleagues. These seem to us to be significant tasks and we accept the evidence that, with the systems the HO has in place, they are necessary to ensure that the process is working. Although the estimate of 4.5 hours seems on the high side it does not seem to us to be unreasonable "

91. Esslemont is not binding but we consider "Assurance and verification of the outputs" to be the same as or akin to quality assurance and we agree that the 3 hours of quality assurance checking is a part of extraction and so allowed by reg 5(3)(d) of the 2004 regs.

92. For the same reasons we also accept that randomly checking individual results in this case post barnardisation is part of extraction. We accept the evidence that this would take 10 minutes per individual. Professor Riley made an assumption that 12 hours had been allowed (whereas it is 8) and said "... we would only be able to examine another 72 randomly selected individuals before the 24-hour cost threshold is reached"

93. We did not conclude that it was the UKHSA's position that 72 checks (ie 12 hours) was what would be needed but that 12 hours was all that was left before the "threshold" was reached. The UKHSA's position was as set out by Professor Riley who went on to say that if there were 72 random checks:- "...we could not exclude that the released data could contain over 190,000 potentially identifiable records: any audit of our Barnardisation process would leave a very large absolute number of possibly disclosive records and is unlikely to include enough outliers to ensure compliance with UK GDPR and confidentiality requirements and ISB1523: Anonymisation Standard for Publishing Health and Social Care Data."

94. Based on the size of the dataset concerned and the UKHSA's primary case that no level of bardardisation would prevent identification we understood the UKHSA's position to be that many more than 72 individual records would need to be checked. We agree with this and therefore agree with Professor Riley's starting point that " 5.1 The cost of any reasonable attempt at safe Barnardisation is considerably higher than this indicated reasonable level of 24 person hours (£600 at £25 per hour)." Decision

95. In summary, as regards the reliance by the UKHSA on section 38 FOIA, for the reasons set out above, the DN is in accordance with the law and to the extent that the notice involved an exercise of discretion the IC exercised it correctly.

96. The IC did not consider section 12 FOIA but also, and for the reasons set out above, the UKHSA is entitled to rely upon this section as a lawful reason not to provide the information requested.

97. The appeal is accordingly dismissed. Signed: Judge Heald Date: 14 October 2025 Annex – The Request This is a request for information under the Freedom of Information Act 2000 . Please can you provide, compiled or extracted from the raw data that I can see that you have previously confirmed the UKHSA holds (ref Case 04/08/23/LD/309), the following information: For each person aged 20 or over at their recorded date of death occurring between 1 January 2021 and 31 December 2023, whose data consequently is no longer included in the published NIMS database:

1. Age cohort for first dose of Covid vaccination (please use the age cohorts already used for NIMS purposes)

2. Date of each dose of Covid vaccination

3. Date of death Personal data is defined in legislation as information relating to an identified or identifiable "living" individual. As such, it would seem that by definition the data I have requested, relating only to deceased people, is not personal data. If despite this you consider that it would still be necessary to ensure that disclosure of this data would not breach data protection principles, please compile the dataset I have requested using a Barnardisation process which modifies the date of death for each individual by adding or subtracting (for example) one, two or zero days from the date of death recorded in your raw data thus preventing the identification of any individual. I can see that you have previously sought to resist disclosure of a similar dataset on the grounds that it is or includes personal data, and that compiling a Barnardised form of that data would amount to the creation of new information, which a public authority is not required to do. This is very clearly not the case as a matter of law, and I should draw your attention to the ICO's guidance (https://ico.org/for-organisations/foi) which explains under the heading 'Extracting and compiling information' that if you hold the raw data which enables you to produce a requested dataset without requiring specialist skills, you are obliged to do so. Barnardisation of health data is a standard process for anonymising data with which the UKHSA cannot credibly claim to be unfamiliar or to require specialist skills. Moreover, I should draw to your attention the binding House of Lords decision (Common Services Agency v Scottish Information Commissioner [2008] UKHL 47 ) which is expressly referenced in that same ICO guidance document, and which makes absolutely clear that Barnardisation of health data “would not amount to the creation of new information, nor would it involve the carrying out of any research. It would be to do no more than was reasonable in the circumstances…”. It would thus be unlawful for the UKHSA to assert that any Barnardisation of the data that I have requested would create new information. If any member of the UKHSA Information Rights team were to reassert that earlier erroneous position after having now been made aware of both the House of Lords’ definitive determination of the law in this area, and of the ICO’s guidance, I suggest that this would not only be a breach of the FOIA but might even amount to the blocking of data held by the UKHSA with the intention of preventing its disclosure, which would be a criminal offence under section 77(1), FOIA.