Expert Report to the Infected Blood Inquiry: Statistics

September 2022

© Crown copyright 2022

This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3

Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.

This publication is available at www.infectedbloodinquiry.org.uk

Any enquiries regarding this publication should be sent to us [email protected]

09/22

Printed on paper containing 75% recycled fibre content minimum

Printed in the UK by the APS Group

Table of Contents

• Communicating Uncertainty
• Executive Summary
• Introduction
• Chapter 1 Human Immunodeficiency Virus (HIV) infections in people with bleeding disorders
• Chapter 2 Hepatitis C Virus (HCV) infections in people with bleeding disorders
• Chapter 3 HIV infections in transfusion recipients
• Chapter 4 Hepatitis C Virus (HCV) in transfusion recipients
• Chapter 5 Information from funds
• Chapter 6 Variant Creutzfeldt Jakob Disease (vCJD) infections from blood and blood products
• Chapter 7 Hepatitis B Virus (HBV) infections from blood and blood products
• Acknowledgements
• Authors
• Verifying Statements
• Appendix

Communicating Uncertainty

Few of the quantities in this report are known with certainty. We use two main ways to express the uncertainty resulting from our analyses – numerical ranges and confidence statements.

Numerical ranges have two forms:

Confidence intervals (CI) are a standard statistical technique when estimating single quantities from individual data sources. They express a range of plausible values (usually 95%) that is compatible with the observed data.

Uncertainty intervals arise when building complex models based on a set of assumptions, some judgemental. Uncertainty about the inputs is propagated, using simulation methods, through the model to produce a range (usually 95%) of plausible values for the outputs.

Confidence statements summarise our judgement of the extent that the available data can answer the question of interest, on the following scale: Low, Low/Moderate, Moderate, Moderate/High, High.

Executive Summary

1: HIV infections in people with bleeding disorders

How many people with bleeding disorders were infected with HIV through blood products in the UK between 1970 and 1991?

How many have subsequently died; and of these deaths, how many were due to HIV/AIDS?

• Three independent data-sources lead to broadly similar conclusions.

Table ES1 HIV diagnoses in people with bleeding disorders reported from three different sources up to the end of 2020.

• We conclude that around 1,250 people with bleeding disorders were infected with HIV in the UK between 1970 and 1991. Around three-quarters of these have died, and around half have died of HIV-related causes.
• Confidence that available evidence can answer the questions? Moderate/High.

2: HCV infections in people with bleeding disorders

How many people with bleeding disorders were infected with HCV through blood products in the UK between 1970 and 1991?

How many have subsequently died; and of these deaths, how many were HCV-related?

• In spite of the efforts made by the UKHCDO, it is challenging to establish the number of HCV infections in people with bleeding disorders, due to the limited data on exposure and HCV status.

Table ES2 Number of HCV infections and deaths in people with bleeding disorders in the UK between 1970 and 1991: ‘possible’ infections include those known to have been exposed to at-risk blood products but with unknown HCV status. Additional infections will have occurred in those whose exposure history was unknown and who have not been tested for HCV.

• This table does not include those infected with HIV, who are covered in Chapter 1, though the Skipton Fund number will include some such persons. UKHCDO say it is reasonable to assume that all of those infected with HIV will also have been co-infected with HCV.
• We judge that between 2,400 – 5,000 people with bleeding disorders were infected with HCV, excluding those infected with HIV.
• Confidence that available evidence can answer the questions? Low/Moderate.

3: HIV infections in transfusion recipients

How many people were infected with HIV through blood transfusions in the UK between 1970 and 1991?

How many have subsequently died; and of these deaths, how many were due to HIV/AIDS?

• At least 79, and possibly up to ~100, people were infected with HIV through blood transfusions in the UK between 1970 and 1991, mostly pre-1986. Around 85% have subsequently died, but we do not know the causes of death. This assessment is based on a data set provided by the UK Health Security Agency (UKHSA).
• Confidence that available evidence can answer the questions? Moderate.

4: HCV infections in transfusion recipients

How many people were infected with HCV through blood transfusions in the UK between 1970 and 1991?

How many were chronically HCV-infected, and subsequently died; and, of these deaths, how many were due to their chronic HCV infection?

• Our analysis is based on a complex statistical model of the stages from HCV-infectious donors to recipients becoming chronically HCV-infected following transfusion, and then survival until the end of 2019, taking into account the increased risk associated with chronic HCV-infection. Our primary model was constructed for England, and then adapted for Scotland, Wales and Northern Ireland.
• Numerous sources of evidence have been used in constructing this model, but the many assumptions and inevitable limitations in the data mean that there is considerable uncertainty around our numerical conclusions. We provide central estimates and 95% uncertainty intervals for the results from our ‘baseline scenario’. It is important to note the most likely value is near the central estimate, and that it is very unlikely that the true value is as high as the upper end of the interval, or as low as the lower end.
• Our summary estimates for the UK from our baseline scenario are:

Table ES4 Estimates and 95% uncertainty intervals of the main quantities of interest from the statistical model of HCV transmission from transfusions. Estimates are for the whole of the UK. Females accounted for 64% (95% uncertainty interval: 58% to 69%) of the people chronically infected with HCV by transfusion who survived to the end of 2019.

• Extensive sensitivity analyses identified that assumptions about the effect of both HIV antibody screening and changing guidance for potential blood donors were an important factor: assuming a more limited effect would reduce our estimates by around a quarter. There is essentially no direct evidence regarding this quantity, and so we are reliant on expert judgements.
• Our estimates for the number of infections are reasonably similar to previous estimates based on look-back exercises.
• Confidence that available evidence can answer the questions? Moderate.

5: Information from funds

• The funds will not be a complete record of those infected, but provide a useful ‘calibration’ of results derived from alternative sources.

6: vCJD infections from blood and blood products

How many people were infected with vCJD from blood and blood products in the UK?

• Out of 178 recorded cases of confirmed or probable vCJD in the UK, 5 were blood-borne infections: 3 symptomatic vCJD cases; one asymptomatic person (positive at autopsy for abnormal prion in her spleen) from transfusion-transmitted infection; and one asymptomatic infectee with severe haemophilia A who had been exposed to vCJD-implicated Factor VIII (positive at autopsy). All five have died.
• Confidence that available evidence can answer the questions? High.

7: HBV infections from blood and blood products

How many people were infected with HBV between 1970 and 1991 through blood and blood products, and what were the impacts?

• Due to the limitations in the data available, it is not possible to answer the questions set with any reasonable accuracy, when compared to other infections we investigated. There was a lack of an integrated public health approach at the onset of donor screening in 1971/72 to identify past recipients of transfusion from donors who were infectious HBV carriers. Furthermore, people infected with HBV have never received financial support and so funds are not a source of data.
• Confidence that available evidence can answer the questions? Low.

Introduction

The Inquiry’s Letter of Instruction⁽¹⁾ to its Statistics Expert Group of 25th September 2019 contained an extensive list of questions concerning a wide range of uses of statistics about infections from blood and blood products. However, when the SARS-CoV-2 pandemic began in early 2020, this team of experts in infectious diseases immediately became very heavily engaged in statistical work on COVID-19, and this has continued until now.

Professor Stephen Evans, the Convenor of the Statistics Expert Group, wrote to Sir Brian Langstaff on 27th June 2022⁽²⁾ suggesting a more restricted focus:

• We propose that the report we publish in September should set out, based on the most reliable data available and with clearly stated assumptions, the likely ranges for the numbers of people who have been infected through blood and blood products, and, where possible, their subsequent mortality. We will also publish the statistical models we have created.
• I understand from the Inquiry team that the evidence has touched on a number of statistical issues during the course of the Inquiry’s hearings. We will explain any relevant statistical principles of interest in our report and when we give evidence at a hearing in early October.

This proposal was accepted by Sir Brian.

This report lays out the findings from our investigations into the number of infections from blood and blood products in the UK between 1970 and 1991 (later in the case of vCJD), and the subsequent survival of those infected. We cover HIV, HCV, vCJD and, to a limited extent, HBV. Despite their importance, we do not address how morbidity or quality of life has evolved during disease progression in the past 50 years of pharmaceutical and other advances in care.

It is important to be clear about the limitations of what we can conclude from this, and indeed any other, statistical investigation. In brief, any conclusions we draw from data will depend on:

• Availability: sometimes the information we want was not collected or not retained.
• Definitions: the numbers depend crucially on the criteria in use at the time, for example, what is meant by an antibody-positive HCV test has varied considerably over time.
• Quality: data may not be accurate, particularly when part of routine collections rather than a planned study.
• Completeness: data may be missing for a variety of reasons – and sometimes the fact that data are missing may itself be informative.
• Context: for example, we need to take into account why the data were collected (for example, by review of death certificates; or at autopsy), and what was known at the time.
• Skill in Interpretation: statistics are sometimes represented as hard ‘facts’, but the caveats listed above should make it clear that they can be much ‘softer’ than perhaps perceived. Drawing conclusions from data is not some automatic process, and judgement is always required – the data do not speak for themselves.

As explained previously, we try to be clear about these concerns by not only reporting numerical ranges expressing uncertainty about quantities, but also judgements of our confidence that the available data can answer the primary questions of interest.

An important limitation is that statistics are always an imperfect representation of what we are actually interested in. Tables and graphs of numbers of people infected and their subsequent survival are inevitably a gross, even harsh, summary of the suffering endured by those individuals and the people close to them. We provide no measures of illness, psychological distress, financial harms, family stress, and the many other ways in which infected blood will have damaged lives. In particular, we do not attempt to estimate onward-transmission to partners, children or others. We hope that readers will recognise our understanding that, beneath all the counts and measurements, there are individual human lives. But it is only by summarising all those experiences into bald numbers that we can properly assess the magnitude of what has happened.

This has been a particularly challenging area to investigate as the data, when available, tend to suffer from many of the issues laid out above and so any numerical conclusions we draw are necessarily cautious and approximate. For HIV and vCJD infections, and for people with bleeding disorders, we can actually count cases of interest from databases, although even then we acknowledge possible incompleteness. But we cannot list those who have been chronically infected with HCV from transfusions, and so we rely on statistical models to estimate what we cannot directly observe – these are necessarily dependent on additional assumptions, and this means that any conclusions are even more uncertain.

Chapter 1: HIV infections in people with bleeding disorders

How many people with bleeding disorders were infected with HIV through blood products in the UK between 1970 and 1991?

How many have subsequently died; and of these deaths, how many were due to HIV/AIDS?

Summary findings

• Three independent data-sources lead to broadly similar conclusions.

Table S1 HIV diagnoses in people with bleeding disorders reported from three different sources up to the end of 2020.

• We conclude that around 1,250 people with bleeding disorders were infected with HIV in the UK between 1970 and 1991. Around three-quarters of these have died, and around half have died of HIV-related causes.

Confidence that available evidence can answer the questions? Moderate/High.

Terminology: We follow the relevant Inquiry report⁽³⁾ in using the term ‘people with bleeding disorders’, sometimes shortened to PwBD, to include patients diagnosed with haemophilia (A, B or C) or von Willebrand disease.

Background

The history and current knowledge about HIV have been summarised by an Expert Report to the Inquiry.⁽⁴⁾ Relevant milestones regarding data include:

• 1981: The first recorded case in the UK of what was subsequently known as Acquired Immune Deficiency Syndrome (AIDS) – but human immunodeficiency virus (HIV), as the cause of AIDS, had not yet been identified.
• 1982: The term ‘AIDS’ is used for the first time; a definition is agreed and a proforma for collecting data on each case – the AIDS surveillance form – is introduced in the USA and for use worldwide.
• 1983: The HTLV-III retrovirus, subsequently named “HIV”, was identified in France.
• 1984: Prototype HIV antibody testing introduced in the UK. The UK’s national infectious disease surveillance centres, having introduced the AIDS surveillance form in 1982, establish HIV antibody diagnosis reporting systems to determine the characteristics and extent of infection.
• 1985: HIV antibody screening of the UK’s blood supply from October 1985.

Sources of evidence

There are three main sources of evidence regarding those infected with HIV through blood products. The first source is registrations with the Macfarlane Trust, which provided support to people with bleeding disorders who had been infected with HIV (see Chapter 5).

The second major source is the National Haemophilia Database (NHD), which is run by the United Kingdom Haemophilia Centre Doctors’ Organisation (UKHCDO) and which has evolved from data collection initiated by the Medical Research Council (MRC) in 1969. The NHD data are not used for individual patient care, and therefore are not comparable to the information contained in patient records held by hospitals or haemophilia centres. Our impression is that it has clearly been challenging for the NHD to establish an accurate, up-to-date database without duplicate records.

The NHD received a request dated 6th April 2020 from the Infected Blood Inquiry for analysis of data held by the NHD (“Rule 9 Request”). The resulting report was “Bleeding Disorders Statistics for the Infected Blood Inquiry 2020.”⁽⁵⁾ The primary source for data on HIV infections in people with bleeding disorders is Chapter 8 of UKHCDO (2022) “The number of HIV positive and negative persons with bleeding disorders at each Haemophilia Centre in the UK, broken down by year.”⁽⁶⁾

The third source is the UK Health Security Agency (UKHSA) HIV database, which is a continuation of that held by Public Health Laboratory Service (PHLS), the Health Protection Agency and then Public Health England; this source receives information from Scotland’s HIV database as held currently by Public Health Scotland (originally known as the Scottish Centre for Infection and Environmental Health). Part of the data collected on each individual includes how the infection was probably acquired, and a selection of past published results from this database is shown in Table 1.1. Since 2005, all blood-borne infections have been combined and, in recent years, have been reported under ‘other’.

Table 1.1 A selection of published summaries of HIV blood-borne infection in the UK.

The UKHSA has supplied to the Inquiry⁽¹²⁾ tables that list the number of new HIV diagnoses each year to 2000 in the UK, and the number recorded as being infected by exposure to blood products, as reported by 2020 and re-coded using the 2005 archive when all infections through blood products were merged. ‘Exposure’ was broken into ‘Haemophilia’, ‘Other blood products’, and ‘Blood products (undetermined)’, and further into whether acquired in the UK, acquired outside the UK, or place of acquisition unknown (or not reported). The spreadsheet also reported deaths from all causes.

UKHSA acknowledges the problems in ensuring the accuracy of this database. This is not a clinical record, and – to maintain confidentiality – full patient identifiers are not collected for public health surveillance purposes. However, there is an annual deduplication process, although the great majority (around 98% plus) of identifiers are stable over time. The records are also not necessarily complete and patients’ exposure routes are essentially self-reported.⁽¹³⁾

Although these three databases (Macfarlane, UKHCDO and UKHSA) are concerned with essentially the same cases, there has not been a systematic attempt to reconcile these sources, and therefore they will inevitably arrive at different totals.

Findings – Diagnoses

Source: Macfarlane Trust

Chapter 5 describes how the Macfarlane Trust was set up in 1988 to provide support for people with haemophilia (including women with von Willebrand disease) infected with HIV in the UK. In 2003⁽¹⁴⁾ they reported 1,242 directly infected registrants, in 2007⁽¹⁵⁾ they recorded 1,243 registrants,⁽¹⁶⁾ and in 2013 reported that 353 were known to be alive and registered with the MFET Ltd⁽¹⁷⁾, and therefore 890 were deceased. An additional 14 HIV-infected registrants with the four national funds (see Chapter 5) died between 2017 and 2022, and most will have been people with bleeding disorders.

Limitations:

The Macfarlane Trust may have missed some relevant individuals, for example some of those who died before its establishment, but this is likely to be a small number. Others may have chosen not to register with the Trust. The 1,243 should therefore be considered as a lower bound, although the true total is unlikely to be substantially larger.

Source – UKHCDO

Table 1.2 reproduces Table 8.3.1 of UKHCDO (2022).⁽¹⁸⁾ It records 1,338 people with bleeding disorders diagnosed with HIV between 1979 and 2000 in the UK, the great majority (974) with severe haemophilia A. Forty three were resident abroad. It is assumed that all HIV diagnoses in people with bleeding disorders were due to infection from blood products.

Table 1.2 HIV antibody positive diagnoses in people with bleeding disorders (PwBD) between 1979 and 2000 in UK (source, UKHCDO report).

Limitations:

• The UKHCDO data include people with bleeding disorders with documented results for an HIV test in the NHD, plus five people with bleeding disorders who had AIDS as underlying cause of death recorded on either NHD or NHS Digital. The UKHCDO database excludes partners of people with bleeding disorders, and people with bleeding disorders whose records were labelled as ‘not yet tested’.
• People with bleeding disorders with milder disorders may have been de-registered if their condition improved over time, or they had infrequent contact with their haemophilia centre.
• There does not seem to have been cross-checking between UKHCDO and UKHSA’s register of HIV diagnoses. It is possible that a small number of HIV diagnoses are not known to the NHD.
• Von Willebrand disease is not listed separately, but included in Other Bleeding Disorders.

Table 1.3 shows the 1,338 HIV diagnoses by nation within the UK, according to NHD.⁽²⁰⁾ A further breakdown by Centre is shown in the Appendix Table 1.1. However, for the Penrose Inquiry,⁽²¹⁾ a review of 74 case-histories for patients attributed to Scotland determined that only 60 were likely to have been infected with HIV in Scotland.

Table 1.3 HIV diagnoses up to 2020 in people with bleeding disorders by UK nation (source UKHCDO report).⁽²²⁾

Figure 1.1, which displays the HIV diagnoses by year of UKHCDO-registered diagnosis, indicates the year of the first recorded HIV antibody positive test.⁽²³⁾ Testing began in August 1984 and became more widely available in 1985, and so HIV diagnoses in earlier years arise from testing stored samples. Nine UKHCDO-registered HIV diagnoses were made after 1991, three of them after 1995. The year is unknown for eleven diagnoses.

Figure 1.1 New HIV diagnoses in people with bleeding disorders in the UK by year of first record of a positive test (source UKHCDO).⁽²⁴⁾ Note that the recorded year of diagnosis does not necessarily represent the year of HIV infection due to delays in diagnosis, missing positive tests, and, for those from abroad, confirmatory HIV diagnosis in the UK.

Not all these individuals will have been infected in the UK. UKHCDO report “When a newly immigrated PwBD [person with a bleeding disorder] with HIV infection was registered on the database, the NHD did not record the country where the infection was potentially acquired. Details of HIV tests and the country where it was acquired are only available in the individual’s clinical record held by the centre.”⁽²⁵⁾ Therefore, if based solely on this source, any assessment of the numbers infected in the UK must be an estimate rather than a count.

The Penrose Report recorded that of 74 cases for Scotland, 11 had been infected abroad.⁽²⁶⁾ This provides some information for attempting to estimate how many of the 1,338 cases were infected abroad. Several possibilities arise. First, we note that 63/74 = 85% of the Scottish cases had been infected in the UK. If this proportion were applied generally to the whole of the UK, it would mean that an estimated 0.85 x 1,338 ~1,140 people with bleeding disorders were infected with HIV in the UK. Second, as Scotland had roughly 9% of the UK population in the 1980s, the 11 infected abroad in Scotland could be scaled up to 11/0.09 ~ 122, giving around 1,215 cases infected in the UK. Both these estimates are somewhat below the 1,243 cases recorded by the Macfarlane Trust.

Findings – Deaths

Table 1.4 displays deaths recorded in HIV antibody positive individuals for those with exposure given as ‘haemophilia’ up to 2000 (UKHSA). Also shown are deaths of HIV antibody positive people with bleeding disorders from any cause up to 2020, and from HIV-related causes (UKHCDO).⁽²⁷⁾ Deaths recorded as ‘HIV/AIDS’ or ‘HIV Lymphoma’ are considered as HIV-related.

Table 1.4 Deaths in HIV infected people with bleeding disorders, from any cause and from HIV-related causes (Sources: UKHSA and UKHCDO).⁽²⁸⁾

The UKHCDO data record that 76% of those HIV infected have died (1,017/1,338). As can be seen, the majority (63%, 638/1,017) of these deaths had an underlying cause linked to HIV. Therefore, overall 639/1,338, or 48%, had died from HIV-related causes. Deaths from HIV were greatly reduced after the introduction of more effective treatments in 1995.

Limitations:

The cause of death is taken from the International Classification of Diseases (ICD) ICD-10 code of the underlying cause registered on the death certificate. It is likely that HIV was implicated in other deaths, but not entered as the underlying cause. In particular, it is likely that, particularly at early stages, stigma may have led to HIV not being mentioned on death certificates.

Source – UKHSA

The UKHSA findings⁽²⁹⁾ are summarised in Table 1.5. A total of 1,484 individuals are reported as being infected through ‘blood products’, of whom only 56% have their country of infection recorded.

Table 1.5 New HIV diagnoses in the UK up to 2000 through exposure to infected blood products. HIV diagnosis using the UKHSA 2020 archive, with exposure category based on 2005 archive.

Of those people with assigned exposure routes, the proportion that are people with haemophilia steadily declined over time. If we allocate each year’s ‘undetermined’ total by this proportion, then we would expect another ~124 cases, giving 1,061 + 124 = 1,185. Table 1.5 also reports that at least 19 of these were acquired abroad and so we might estimate, from this UKHSA source, around 1,185 – 19 = 1,166 HIV infections in people with bleeding disorders were acquired in the UK.

As an extreme case, assuming all 182 listed as ‘undetermined’ are people with haemophilia gives 1,061 + 182 = 1,243, a total which matches that of the Macfarlane Trust and has been quoted by a number of sources.⁽³⁰⁾ However, this appears to be a coincidence, as the UKHSA data included those infected abroad.

Limitations:

• Patients’ exposure routes are essentially self-reported.⁽³¹⁾
• It is possible that a proportion of the 241 ‘other blood products’ may refer to people with bleeding disorders who do not have haemophilia, for example patients with von Willebrand disease, and therefore that the ‘haemophilia’ category is an undercount of people with bleeding disorders who are HIV antibody positive.
• A small number of people with bleeding disorders may have been infected with HIV through other routes.

Since the successful claimants to the Macfarlane Trust presumably constitute a lower bound on the true number of UK-acquired infections, this suggests that the UKHSA registry is a less reliable source.

Figure 1.2 New HIV diagnoses in ‘haemophilia’ in the UK by year of first recorded positive test (source UKHSA).⁽³²⁾ Note that the recorded year of diagnosis does not necessarily represent the year of infection due to delays in diagnosis, missing positive tests, and, for those from abroad, confirmatory HIV diagnosis in the UK.

Figure 1.2 shows the year of first HIV antibody positive test for those classified as ‘haemophilia’ in the UKHSA database - more detail is provided in Appendix Table 1.2. The diagnoses recorded by UKHCDO tend to be slightly earlier than those registered with the PHLS and now held by UKHSA, due – we assume – to retrospective testing of stored blood samples in order to establish patients’ earliest HIV antibody positive date.

The UKHSA data include the deaths of 25,999 people with HIV diagnoses recorded before 2000, a recorded mortality rate of 56% up to the end of 2020 (25,999/46,030). Table 1.6 shows the outcomes corresponding to the individuals in Table 1.5 who are reported as having been infected through exposure to blood products.

Table 1.6 Deaths recorded up to the end of 2020 of those infected with HIV before 2000 through exposure to infected blood products: reported also as percentage of the corresponding total of diagnoses in Table 1.5, and so represents the mortality rate. (Source UKHSA)⁽³³⁾

Up to 2020, the overall mortality in the UKHSA database of HIV antibody positive ‘haemophilia’ cases was 820/1,061, or 77%, while the UKHCDO data recorded a similar proportion: 1,017/1,338, or 76%. The fact that the mortality rate in those with ‘undetermined’ exposure is close to that of ‘haemophilia’ adds weight to the assumption that the majority of these are in fact people with bleeding disorders.

Limitations:

• As for UKHCDO, the cause of death is taken from ICD-10 code on death certificate, and may be an undercount for the reasons given previously.
• As noted above, the category ‘haemophilia’ may miss relevant cases that have been either classified as ‘other blood products’ or ‘undetermined’.

Chapter 2: HCV infections in people with bleeding disorders

How many people with bleeding disorders were infected with HCV through blood products in the UK between 1970 and 1991?

How many have subsequently died; and of these deaths, how many were HCV-related?

Summary findings

• In spite of the efforts made by the UKHCDO, it is challenging to establish the number of HCV infections in people with bleeding disorders, due to the limited data on exposure and HCV status.

Table S2 Number of HCV infections and deaths in people with bleeding disorders in the UK between 1970 and 1991: ‘possible’ infections include those known to have been exposed to at-risk blood products but with unknown HCV status. Additional infections will have occurred in those whose exposure history was unknown and who have not been tested for HCV.

• This table does not include those infected with HIV, who are covered in Chapter 1, though the Skipton Fund number will include some such persons. UKHCDO say it is reasonable to assume that all those infected with HIV will also have been co-infected with HCV.⁽³⁴⁾
• We judge that between 2,400 - 5,000 people with bleeding disorders were infected with HCV, excluding those infected with HIV.

Confidence that available evidence can answer the questions? Low/Moderate.

Background

The Archer Inquiry⁽³⁵⁾ clearly summarises the risks facing people with bleeding disorders in the 1970s and 1980s: “The danger of infection from blood products was directly related to the size of the donor pool from which the blood or plasma was collected and to the lifestyles of the communities from which donors were drawn [...] By the end of 1991, therefore, blood products manufactured in the UK were as safe from infection with Hepatitis C as current technology could make them, although this was not necessarily true of all imports.”

Sources of Evidence

Our primary source for data on HCV infections in people with bleeding disorders (PwBD) is the UKHCDO (2022) report “Bleeding Disorders Statistics for the Infected Blood Inquiry 2020,”⁽³⁶⁾ Chapter 9: “The number of people with bleeding disorders at each Haemophilia Centre in the UK showing positive and negative HCV antibody and PCR test result”.

Following the UKHCDO review and record-linkage follow-up described below, we had access to preliminary analyses, carried out at our request by a Manchester statistical team working in collaboration with UKHCDO, for well-documented patients, namely: those who were both born and registered with the National Haemophilia Database (NHD) before 1992 and had been followed-up for mortality via record-linkage to 31 December 2019.

A look-back exercise in 2010 identified over 29,000 people with bleeding disorders registered before universal HCV testing of blood donations was introduced in September 1991, for whom data were requested from the centres.⁽³⁷⁾ This proved challenging. In 2018, a further look-back exercise tried to identify any people with bleeding disorders who might have missed screening for HCV infection despite exposure to products associated with HCV transmission.⁽³⁸⁾

The UKHCDO report considers a patient potentially ‘at-risk of HCV infection’ if there was documentation in the NHD of exposure to a pooled plasma-derived concentrate manufactured before 1988 or a blood component before 1992 – those with no record of exposure on the NHD would tend to have milder disorders. UKHCDO describe the coverage as follows: “All people with bleeding disorders considered at-risk of HCV infection or known to have HCV infection were identified from NHD and allocated to the six groups described (see below). All data from the 2018 look-back exercise and the 2020 update are included in this analysis. Additional at-risk patients were identified when the archived paper records were transcribed into NHD for this analysis and during the review of deceased patients and the causes of death. All at-risk and potentially at-risk patients have been included.”⁽³⁹⁾ The available data therefore come from a variety of efforts to determine the HCV status of those at risk, including those who had died from causes related to HCV.

A discussion with representatives of UKHCDO revealed a belief that the majority of patients with positive HCV test results (at least 95%) would have been reported to the NHD.

Analysis and conclusions: UKHCDO identified 8,752 patients initially considered at-risk of HCV infection, both those who are alive and those known to have died by December 2020. They report six mutually exclusive at-risk HCV categories of people with bleeding disorders based on documentation of infection with HIV and HCV and records of exposure to at-risk products on the NHD. People with bleeding disorders are allocated to each group sequentially, so that, for example, someone who is known to be HIV antibody positive is allocated to the first category regardless of any information about HCV status.

1. HIV antibody positive: The analysis assumed that this group was inevitably co-infected with HCV as many people with bleeding disorders who were HIV infected died before HCV antibody testing was available. This group includes people with bleeding disorders with HIV antibody positive results reported to the NHD or had HIV/AIDS or HIV lymphoma documented as their underlying cause of death on their death certificate.
2. Tested HCV antibody positive: This group includes people with bleeding disorders with a HCV antibody positive test result reported to the NHD or HCV documented on the death certificate. This group includes a small number of people with bleeding disorders who might have acquired the infection abroad.
3. HCV presumed positive: This group includes people with bleeding disorders born before 1992 with HCV-related liver disease or hepatocellular carcinoma (HCC) reported as the underlying cause of death or documented as significant co-morbidity in the absence of any HCV positive antibody result.
4. Tested HCV antibody negative: People with bleeding disorders with an HCV antibody negative test result on the NHD.
5. HCV status unknown, exposed to an at-risk pooled plasma product: This group includes people with bleeding disorders with a record of exposure to at-risk pooled plasma concentrate including exposure to at-risk blood components in the NHD but without HCV antibody test results reported to the NHD. On a precautionary basis, UKHCDO considered this group as at a high risk of HCV-infection (i.e. if indeed patients had not been HCV-tested locally).
6. HCV status unknown, exposed at-risk to blood component: This group includes people with bleeding disorders without HCV antibody test results reported to the NHD and with evidence of exposure to at-risk blood components only.

The remainder were classified as Not known to be at-risk. This group includes people with bleeding disorders with no records of exposure to at-risk pooled concentrates or blood components on the NHD. Some of these people may have been exposed to an at-risk blood product without the NHD being aware.

Table 2.1 HCV status according to UKHCDO in at-risk people with bleeding disorders in the UK (and by nation); and death-rates according to HCV status. UKHDCO have records of 3 persons for whom nation is not recorded.⁽⁴⁰⁾

The first three columns can be taken as confirmed HCV positive, since 1,338 people with bleeding disorders who were HIV-infected are assumed to be co-infected with HCV. The UKHCDO reports 2,178 + 117 = 2,295 confirmed HCV (mono) infections, plus a further 1,469 + 909 = 2,378 people exposed to at-risk products but with unknown HCV status – but who could be considered ‘possible’ HCV infections.

The total of 2,178 people with bleeding disorders in Table 2.1 who were confirmed to be HCV-infected but not HIV-infected is inevitably an underestimate. If 95% coverage is assumed, as discussed above, then we would conclude a minimum of around 2,400 confirmed cases (including the 117 presumed HCV-infected).

The 1,017 deaths in HIV-positive patients matches the data discussed in Chapter 1. In addition, 517 + 117 = 634 confirmed HCV-infected patients have died: the mortality rate in ‘HCV presumed positive’ is 100%, reflecting that these cases have been largely identified through death registrations. In addition, 1,385 (928 + 457) exposed patients have died: a rather higher rate than for those confirmed HCV antibody positive.

UKHCDO report that, of those 634 confirmed and presumed HCV-infected patients (without HIV) who have died, 105 have died from ‘liver failure – HCV’, and 143 have died from HCC.⁽⁴¹⁾ This comprises 248/634 = 39% of all deaths.

Limitations:

• The patients featured in Table 2.1 will not include everyone who was exposed to infected blood products, despite the efforts made to determine their status. Some will have died before 1992 (as Table 2.2 demonstrates), let alone before the 2010 look-back, from causes not related to HCV. Others will have not been followed-up by centres.
• In his Statement to the Inquiry, Professor Hay⁽⁴²⁾ stated that, in the 2010 look-back exercise, of 1,523 patients whose treatment history was reported as ‘unknown’, but who had been HCV tested, 419 (28%) were HCV antibody positive. This will not be a representative sample, as there will have been a reason for selection for HCV testing, but suggests an additional potential source of undercount.

Analysis of well-documented subset of the NHD

The death-rates in Table 2.1 differ not only on account of HIV/HCV co-infection but also because the composition of the groups differ including in terms of sex, bleeding disorder and severity.

The next three tables bring to the fore this additional complexity for the well-documented subset of NHD patients born and NHD-registered before 1992 and for whom record-linkage was successfully achieved so that their survival-status at 31 December 2019 has been ascertained.

Table 2.2 HCV status, separately for male and female patients, together with their survival status at 31 December 1991 and 31 December 2019 respectively. *Sex is missing for two patients.

For males only, Table 2.3 documents the well-documented subset’s composition by bleeding disorder and HCV status. The most common bleeding disorder is Haemophilia A (5 or fewer IU/dl) for men whose HCV status is in the first three rows of Table 2.3, namely: males who were HIV/HCV coinfected, males who tested HCV antibody positive or males who were presumed HCV-infected (generally on account of their cause of death). The fourth row, males whose HCV status is not known but who were exposed to pooled plasma products has a similar pattern in terms of the most common bleeding disorder. The remaining rows are different with “Haemophilia A other” as their most common diagnostic group but notice also von Willebrand disease. The corresponding table for females is in Appendix Table 2.1.

The highest death-rate in Table 2.3 is associated with acquired bleeding disorders.

Table 2.3 Bleeding disorder and its severity for males by HCV status, as described in Table 2.2, together with survivorship to the end of 1991, 1999, 2009, 2013 and 2019.

For males in Table 2.3 who were HIV/HCV co-infected; had tested HCV antibody positive or negative; or whose HCV status was not known but who were exposed to pooled plasma products, Table 2.4 documents how their death-rates, from all causes or from selected major causes of death, have evolved over four epochs of follow-up.

The four epochs of follow-up are:

1. from 1 January 1992 to 31 December 1999 (8 years);
2. from 1 January 2000 to 31 December 2009 (10 years);
3. from 1 January 2010 to 31 December 2013 (4 years);
4. from 1 January 2014 to 31 December 2019 (6 years).

Table 2.4 shows all-cause and cause-specific death-rates per 1000 person-years (pys) of follow-up, each qualified by its 95% confidence interval (CI) which appears in brackets.

Table 2.4 shows that the all-cause death-rate for HIV-infected males reduced from 90 (82 to 98) per 1,000 pys of follow-up in epoch i through 23 (19 to 27) in epochs ii and iii to 13 (8 to 18) per 1,000 pys in epoch iv.

HIV/AIDS or HIV lymphoma accounted for 340/474 (72%) of all deaths which occurred in epoch i for HIV infected males, reducing to 30/144 (21%) in epochs ii and iii to 1/26 (4%) during epoch iv.

The second major cause of deaths in HIV-infected males, all assumed to be HCV-coinfected, was HCC and HCV liver failure which accounted for 70/474 deaths (15%) in epoch i; rose to 49/144 deaths (34%) in epoch ii and iii; but reduced to 2/26 (8%) in epoch iv.

The death-rate for HIV/HCV co-infected males from HCC and HCV liver failure was 13 (10 to 16) per 1,000 pys in epoch i; 8 (5.6 to 10) during epochs ii and iii; and reduced to 1 (0.1 to 3.7) per 1,000 pys in epoch iv.

Per epoch, death-rates from intracranial bleed were similar for those HIV-infected and for those whose HCV status was not reported but who had been exposed to pooled plasma products.

Finally, we comment on the HCC or HCV liver failure death-rates for those who were reported as having tested HCV antibody positive: the rate in epoch i was 2.7 (1.8 to 3.6) per 1,000 pys; rose to 3.9 (3.1 to 4.7) during epochs ii and iii; and to 4.1 (2.7 to 5.4) in epoch iv. The earlier pair of HCC or HCV liver failure death-rates is substantially lower than for males with bleeding disorders who were HIV/HCV co-infected.

We notice also that the epoch-specific death-rates from intracranial bleed were substantially lower for men whose HCV test result was recorded on the NHD than for those HIV/HCV co-infected or for males exposed to pooled plasma products without their HCV status having been registered on the NHD.

Table 2.4 All-cause death-rates and cause-specific mortality during four epochs of follow-up from 1 January 1992 for males in the well-documented subset according to HCV status. The death-rate in an epoch is calculated as the number of deaths that occur during the epoch divided by the person-years of follow-up contributed by persons alive at the start of the epoch. Person-years are the duration of the epoch for persons who survive through to the start of the next epoch. Person-years contributed by those who die during the epoch are counted up to the date of death.

Figure 2.1 Death-rates and cause-specific mortality during four epochs of follow-up from 1 January 1992 for males in the well-documented subset according to HCV status. (A) All-cause mortality. (B) HIV/AIDS, HIV lymphoma related mortality for HIV-infected individuals. (C) Cause-specific mortality.

We note that, of 409 deaths of those testing HCV positive, 163 (40%) were from HCC or HCV liver failure. This matches the proportion calculated earlier on the dataset provided in the UKHCDO report.⁽⁴³⁾

Limitations:

• All people with bleeding disorders who were treated before 1992 should have been counselled about their assessed HCV-risk and offered HCV-testing. However, the information notified to the NHD about patients’ HCV test-history is incomplete – even if HCV-status is unknown to the NHD, the patient may be aware about his or her HCV-status.
• Within the well-documented subgroup of people with bleeding disorders born and NHD-registered before 1992, neither HCC nor HCV liver failure featured as an underlying cause of death for any of 803 males with bleeding disorders, alive at the start of 1992, whose HCV status was unknown to the NHD but who had been exposed to pooled plasma concentrate. The explanation is that UKHCDO re-classifies such patients as “Presumed HCV positive” on account of their cause of death.
• Completeness of information on deaths depends upon whether a register relies upon information on deaths being reported by centres or whether, for better assurance, patients’ survival status has been checked against mortality records. The success of checking against the UK’s death records depends on the quality of information about the patient (such as sex, date of birth, NHS number, first part of postcode of residence) that can be provided to enable the patient to be correctly “matched”.
• Over the period from 1970 to the present day, the ICD has been updated more than once (ICD-10 from 2000 with minor updates since), so that clinicians’ review of death certificates and case-notes to label cause of death has additional merit – provided that consistency can be maintained.

For the Penrose report,⁽⁴⁴⁾ UKHCDO extracted data from the NHD which indicated that 296 living patients and 216 deceased patients treated at Scottish centres had been exposed to HCV, giving a total of 512. After removing duplicates, these counts were revised down to 254 living and 193 deceased patients (total = 447) treated at Scottish centres – these exclude those known to be infected with HIV. Table 2.1 indicates 260 (244 + 16) confirmed HCV antibody positive patients in Scotland, plus a further 146 (36 + 110) exposed to an at-risk product, although with unknown HCV status. This adds to 406 patients (260 + 146), somewhat fewer than the 447 in the Penrose report.

Fund data: In Chapter 5 we report that by 2017 the Skipton Fund had approved 5,529 applications for support for those infected with HCV from blood and blood products, about half of whom (~2,760) would have been for people with bleeding disorders. Around 33% of these claimants (~900) had died.

Chapter 3: HIV infections in transfusion recipients

How many people were infected with HIV through blood transfusions in the UK between 1970 and 1991,

How many have subsequently died; and of these deaths, how many were due to HIV/AIDS?

Summary findings

• At least 79, and possibly up to ~100, people were infected with HIV through blood transfusions in the UK between 1970 and 1991. Around 85% have subsequently died, but we do not know the causes of death. This is based on a data set provided by the UK Health Security Agency (UKHSA).

Confidence that available evidence can answer the questions? Moderate.

Background

Concern about the possibility of HIV-infection through blood transfusions grew in the early 1980s and led to guidance for higher-risk groups to avoid donating blood. Table 3.1 summarises some of the key changes. It is notable that deferral of donations from people who have received blood was not introduced until later.

Table 3.1 Changing advice to potential blood donors during the 1980s.

Sources of evidence: As we saw in Chapter 1, the HIV database (now held by UKHSA) collected information on ‘exposure’ route, and ‘blood/tissue transfer’ was included as a category, but this category ceased in 2005. We have been provided with UKHSA data up to 2000 in which exposure to blood products has been re-assigned, according to the 2005 archive⁽⁵⁰⁾, into ‘Haemophilia’, ‘Other blood products’, and ‘Blood product (undetermined)’. ‘Other blood products’ primarily comprise infection through transfusion, although a few tissue transplants will also be included.⁽⁵¹⁾

Analysis and conclusions: As shown in Table 3.2, there were 241 people with HIV diagnoses up to 2000 who were labelled as having been exposed to ‘other blood products’, of whom 148 were recorded as dying before 2020. Of the 79 people confirmed as HIV-infected through ‘other blood products’ in the UK up to December 2000, 67 had died up to December 2020 (85%).

Table 3.2 Diagnoses of those infected with HIV before the end of 2000 through exposure to ‘other blood products’, and subsequent deaths recorded up to the end of 2020 together with the percentage who died. (Source UKHSA).⁽⁵²⁾

However, it is important to note that, of the 241 confirmed HIV infections via ‘other blood products’, the majority (137, 57%) were acquired abroad. Figure 3.1 shows that the year of diagnosis depends strongly on whether the HIV infection was acquired in the UK or abroad – those infected in the UK declined rapidly after 1986, while those infected abroad increased steadily from that period. Judging by their mortality rate (44%), it seems possible that the great majority of those with ‘country of infection not reported’ were in fact infected abroad, since the mortality rate for people whose infection was acquired in the UK is so much higher (85%).

Figure 3.1 Recorded year of HIV diagnosis of those exposed through ‘Other blood products’, showing difference according to place of acquisition. Note that the recorded year of diagnosis does not necessarily represent the year of infection due to delays in diagnosis, missing positive tests, and, for those from abroad, confirmatory HIV diagnosis in the UK.

However, as shown in Table 1.5, the UKHSA data record 182 people who were infected with ‘blood products (undetermined)’. In 1.22 we suggested around 124 (68%) of these may have been persons with ‘haemophilia’, suggesting around 182 - 124 = 58 could have been additional ‘other blood products’, an extra 24% (58/241) to add to Table 3.2. This would add a further 24% x 79 = 19 people who acquired their HIV infection via transfusion in the UK, making a total of perhaps 98 people infected with HIV from transfusion in the UK before the mid-1980s, of whom around 83 (85%) had subsequently died.

Scotland: The Penrose Inquiry reported⁽⁵³⁾ at least 18 HIV infections from blood transfusions in Scotland, based on data from the Scottish National Blood Transfusion Service (SNBTS) and Health Protection Scotland (HPS), of whom up to 15 had “died from AIDS”. In comparison with our estimates for England, this is somewhat higher than we would expect from simple considerations of population, indicating a higher HIV risk by transfusion in Scotland.

Information from funds. The Eileen Trust was responsible for supporting people infected with HIV who were without a bleeding disorder, although responsibility has now passed to funds for individual nations. Table 5.1 in Chapter 5 shows 74 cases of HIV after transfusion in the UK reported prior to the Eileen Trust, plus 17 cases where country of transfusion was unknown.

Chapter 4: HCV in transfusion recipients

How many people were infected with HCV through blood transfusions in the UK between 1970 and 1991?

How many were chronically HCV-infected, and subsequently died; and, of these deaths, how many were due to their chronic HCV infection?

Summary findings

• Our analysis is based on a complex statistical model of the stages from HCV-infectious donors, becoming chronically HCV-infected following transfusion, and then survival until the end of 2019, taking into account the increased risk associated with chronic HCV infection. Our primary model was constructed for England, and then adapted for Scotland, Wales and Northern Ireland.
• Numerous sources of evidence have been used in constructing this model, but the many assumptions and inevitable limitations in the data mean that there is considerable uncertainty around our numerical conclusions. We provide central estimates and 95% uncertainty intervals for the results from our ‘baseline scenario’. It is important to note that the most likely value is near the central estimate, and that it is very unlikely that the true value is as high as the upper end of the interval, or as low as the lower end.
• Our summary estimates for the UK from our baseline scenarios are:

Table S4 Estimates and 95% uncertainty intervals of the main quantities of interest from the statistical model of HCV transmission from transfusions. Estimates are for the whole of the UK. Females accounted for 64% (95% uncertainty interval: 58% to 69%) of the people chronically infected with HCV by transfusion who survived to the end of 2019.

• Extensive sensitivity analyses identified that assumptions about the effect of both HIV antibody screening and changing guidance for potential blood donors were an important factor: assuming a more limited effect would reduce our estimates by around a quarter. There is essentially no direct evidence regarding this effect, and so we are reliant on expert judgements.
• Our estimates for the number of infections are reasonably similar to previous estimates based on look-back exercises.

Confidence that available evidence can answer the questions? Moderate.

Terminology: HCV infection.⁽⁵⁴⁾ The term ‘HCV infection’ here is used to denote a state in which viral replication is taking place in a patient. All recipients of HCV-contaminated units are assumed to develop HCV infection – see Assumption (a) (paragraph 4.8). A proportion of patients with HCV infection are assumed to clear the virus within 6 months of acquisition – see Assumption (b). Patients who have cleared the infection will test RNA negative but remain HCV antibody positive (anti-HCV positive); they are assumed to be non-infectious and will not transmit HCV, and might be termed HCV-infected. Those who have not cleared the infection will be antigen positive (HCV-RNA positive), and may also be known as having chronic HCV and will be HCV-infectious, unless subsequent antiviral treatment achieves sustained virological clearance.

Our Approach

Previous Chapters have made use of registries containing details of individuals who have been infected with blood products. The numbers infected can therefore be counted, up to the accuracy and completeness of the database. In contrast, we cannot reliably count individuals who were infected with HCV through blood transfusions, since most would never have known they had been infected and would not ever feature in a registry of HCV diagnoses; others do feature but, typically, not until a decade or more after their HCV-implicated transfusion. There are lists of chronically HCV-infected individuals, say fund claimants and from other sources, but these will be substantial undercounts, both through not including those who have been unaware of their HCV infection, and those who died before registering. We therefore need to build a statistical model to estimate the number of HCV infections; such a model necessarily requires numerous assumptions and numerical inputs, which we shall justify using available evidence and judgement.

Previous models, for example Soldan et al. (2002),⁽⁵⁵⁾ have been based on evidence from look-back studies. These involve identifying infected donors, say through people who have been diagnosed as infected (‘trace-back’), and identifying those who had previously received blood from that donor. For example, the 1995 study for England identified 669 people as HCV-infected after receiving a transfusion with a component included in the look-back exercise.⁽⁵⁶⁾ While look-back studies contain valuable information, they are designed to identify individual surviving recipients at increased risk of HCV for further investigation. However, the way they have been identified provides a limited basis for a robust model to estimate overall numbers. The Penrose Inquiry⁽⁵⁷⁾ concluded: “While the look-back exercise was undoubtedly worthwhile, in Dr Gillon’s view, as a means of trying to identify, counsel, test and treat those patients at risk of having contracted Hepatitis C as a result of blood transfusion, it was not a reliable guide to the number of patients likely to have become infected with Hepatitis C through transfusion.”

Our approach closely follows that used by Schnier and Goldberg⁽⁵⁸⁾ (from now on referred to as S&G) in their evidence to the Penrose Inquiry.⁽⁵⁹⁾ S&G created a ‘forward’ model, in which estimates of the main outcomes of interest are generated as a result of a series of stages, as shown in Figure 4.1.

Figure 4.1 Structure of model used to estimate HCV infections following transfusions.

Essentially, by estimating the proportion of HCV-infectious donations (Task 1), and the number of units transfused (Task 2), we can estimate the number of infected recipients (Task 3) in age-sex bands. We then model their survival to 10 years post-transfusion (Task 4), before any mortality-risks due to chronic HCV infection would start becoming apparent. We go on to estimate the number of people surviving to the end of 2019 assuming excess risk from transfusion but not chronic HCV infection (Task 5). Finally, by including an assessment of the excess risk of death in the 21st century from chronic HCV infection, we can estimate how many deaths were linked to HCV (Task 6).

Each stage forms a task in which the following questions are answered both in total, and for each year 1970 to 1991. This modelling is initially for England only.

• Task 1. What proportion of donations were HCV-infectious?
• Task 2. How many units of blood components were transfused?
• Task 3. How many recipients (in age-sex bands) were HCV-infected?
• Task 4. How many chronic HCV-infected recipients survived 10 years after transfusion?
• Task 5. How many chronic HCV-infected 10-year-survivors would have survived to the end of 2019, assuming no excess risk from HCV?
• Task 6. If we allow for excess risk from HCV, how many deaths to the end of 2019 were linked to chronic HCV infection?

Details of each task are given at the end of this Chapter.

We should emphasise that our conclusions depend not only on the modelling assumptions, but the availability of reliable data. It will become clear, in particular, that there are severe limitations in the information collected about blood transfusions over the period of interest.

Baseline model

A baseline model has been constructed for England, adapting the S&G assumptions. This is then adapted to Scotland, Wales and Northern Ireland. The baseline scenario is dependent on multiple assumptions, which we consider plausible but uncertain. The major baseline assumptions are collected together below – see individual Tasks for discussion of these assumptions:

• Assumption (a) Task 1. HCV prevalence in infectious donations is a ‘hybrid’ model with two components. Component 1 is proportional to chronic HCV prevalence in past or current injecting drug users (IDUs) in the year of transfusion (ever-IDUs). Component 2 comes from non-IDUs, and is set at a constant level.
• Assumption (b) Task 1. We assume that a constant proportion of those donors previously infected with HCV (HCV antibody positive) were actively HCV-infectious (RNA positive) and so would transmit the virus in their blood. The proportion is assumed to be around 74%.
• Assumption (c) Task 1. In addition to HIV antibody screening of donors, guidance to prospective donors in the mid-1980s led to a substantial fall (deferral) in HCV-infectious donations. Our baseline assumption is that this occurred in 1985 in England, Wales and Northern Ireland, and in 1984 in Scotland.
• Assumption (d) Task 1. Self-deferral led to around a 67% reduction in donations from HCV-infectious ever-IDUs.
• Assumption (e) Task 1. The contribution to prevalence from non-IDUs (component 2) is a fixed percentage of the proportion of donations that were found to be HCV-infectious when testing began in 1991. The baseline assumption for this percentage is 25%.
• Assumption (f) Task 3. All recipients of HCV-infectious units are assumed to develop HCV infection.
• Assumption (g) Task 3. There is a negligible chance of a transfusion recipient receiving two infected units.
• Assumption (h) Task 4. A proportion (around 18%) of transfusion recipients with HCV-infection is assumed to clear the virus within 6 months of acquisition.
• Assumption (i) Task 4. Chronic HCV infection does not influence recipients’ survival for the first 10 years post-transfusion.
• Assumption (j) Task 5. Having a transfusion increases the annual risk of dying up to 20 years after transfusion.
• Assumption (k) Task 6. Chronic HCV infection increases the annual risk of dying from 10 years post-infection: our baseline assumption is a 53% increased risk.

We first constructed a deterministic model, initially using a spreadsheet and then coded using the statistical programming language R. For any set of specified assumptions about the model, this produces a set of estimates of quantities of interest. We have primarily focused on estimates of the number of transfusion recipients who:

• were infected with HCV;
• would have been chronically infected (were they to survive 6 months);
• were chronically infected, and survived to 10 years post-transfusion;
• were chronically infected, and survived to end 2019 (assuming extra HCV risk);
• were chronically infected, and died by end 2019 (assuming extra HCV risk);
• were chronically infected, and died by end 2019, with death linked to HCV infection.

In what is known as a deterministic sensitivity analysis, we have explored the impact of a wide range of varied assumptions about our model on the outputs listed above. Full details are provided at the end of this Chapter in Table 4.16.

Uncertainty about parameters: On top of the sensitivity analysis to the various major modelling assumptions outlined above, there is additional uncertainty about the quantities entered into any specific models – these are known as parameters. For example, we cannot know the exact effect of chronic HCV infection on mortality, and so a distribution of uncertainty about the central estimate (a 53% increase) is assumed, based on the available evidence. The effect of this uncertainty is assessed through what is known as a stochastic Monte Carlo analysis,⁽⁶⁰⁾ and the results presented as a distribution about the outputs – for example, the number of survivors with chronic HCV infection at the end of 2019 – which can be either graphed or summarised by a median and a 95% interval. This is known as a probabilistic sensitivity analysis.

Results for the baseline scenario are presented in the second row of Table 4.1. The distributional assumptions of the probabilistic model are given in Appendix 4.58. The five rows comprise results from:

• A deterministic baseline model assuming all parameters are fixed at their assumed values.
• A stochastic baseline model, in which parameters and outputs have associated uncertainty.
• Scenario A: A deterministic sensitivity analysis to Assumptions (a) (c) and (d), in which we assume component 1 in HCV-infectious donations comprises past-IDUs rather than ever-IDUs, the deferral effect occurred in 1987 and led to only a 33% reduction in infectious donations.
• Scenario B: A deterministic sensitivity analysis to Assumption (e), in which we assume that there is no component (2) of non-IDU infectious donors. This is analogous to the analysis of S&G.
• Scenario C: A deterministic sensitivity analysis to Assumption (e), in which we assume that component (2) accounts for all infectious donations and that the 1991 proportion of infectious donations applied throughout 1970-1991, i.e., no ‘IDU’ component in the model. This is analogous to the analysis of the England look-back study by Soldan et al. (2002).⁽⁶¹⁾

These additional scenarios have been chosen to illustrate the range of possible estimates.

In this and other Tables in the Chapter, all numbers have been rounded to either 2 or 3 ‘significant figures’ (sf)⁽⁶²⁾ – any greater accuracy would be meaningless given the overall uncertainties in the modelling.

Table 4.1 Estimates for England from baseline deterministic model, baseline stochastic model with 95% uncertainty intervals, and deterministic sensitivity analyses using additional scenarios. See above for specification and sensitivity analysis. The deterministic and median stochastic estimates should be similar, but differ slightly due to the complex non-linear structure of the model. All numbers are rounded to 3 significant figures.

Figure 4.2: Example of the deterministic sensitivity analyses conducted for England. The blue line is the same throughout and corresponds to a deferral effect of 67% in 1985 for ever-IDUs (Assumption (d)) with no proportion of prevalence coming from non-IDUs. (A) Shows the effect of a reduction in the deferral effect to 33%. (B) Shows the impact of the deferral effect occurring in 1986 rather than 1985 (Assumption (c)). Shows the effect of our baseline hybrid model (Assumption (e)), in which there is a constant non-IDU contribution, set at 25% of the prevalence in 1991.

Interpretation:

• Under the stochastic baseline scenario, we estimate around 22,000 transfusion-recipients were infected with HCV in England between 1970 and 1991, with 18,000 becoming chronically infected (were they to survive 6 months). Of these, 6,670 survived for 10 years post-transfusion, when we assume the additional mortality-risk from their chronic HCV infection began.
• Of the 18,000 people with chronic HCV infections, we estimate that 15,800 had died by the end of 2019, and 1,540 of these deaths were HCV-related.
• Of these 15,800 deaths, 11,330 (18,000 - 6,670) had occurred in the first 10 years after transfusion. Of the 4,470 subsequent deaths (15,800 - 11,330), 1,540 (34%) were HCV-related.
• The baseline stochastic scenario shows that there is large uncertainty about all of these numbers, for example the number of chronic infections has a 95% uncertainty interval from 14,100 to 26,600. This reflects the uncertainty in all the quantities that have gone into the model, and implies great caution is needed in quoting the central estimates.
• Deterministic Scenario A focuses on infections from past-IDUs with a reduced deferral effect of 33% in 1987, and with 25% contribution from a constant level of non-IDU infections. This lower deferral effect leads to the conclusion that the historical HCV prevalence was never as high as it might have been if self-deferral had a major impact. The estimated number who were HCV-infected reduces substantially from 22,100 to 15,700, and the number of HCV-related deaths from 1,550 to 1,080.
• Deterministic Scenario B removes the contribution from non-IDUs, so that HCV infections depend solely on ever-IDUs, similar to S&G’s model. This increases the estimated number of HCV-infections modestly to 22,800.
• Deterministic Scenario C assumes a constant level of infectious donations driven by the observed data in 1991 after testing began. This decreases the estimated number of HCV-infections to 20,100, somewhat lower than the 23,500 estimated by Soldan et al. (2002).⁽⁶³⁾

It is clear that the precise proportions of the components in our hybrid model does not have a substantial impact on the conclusions.

Further sensitivity analyses reported at the end of this Chapter confirm that a major driver is Assumption (c), concerning the impact of guidance to prospective donors. Unfortunately, as we explore below in Task 1, there is little direct evidence about this quantity, and so to a large extent we rely on expert judgement.

The uncertainty about the outputs can be visualised as distributions, as in Figure 4.3. These communicate the plausibility of a range of values around the central estimate. For example, the final panel shows the number of excess deaths linked to HCV infection in England; while the bulk of the distribution is around the central estimate of 1,540, there is a reasonable 19% chance that it was more than 2,000 and a modest 12% chance that the number of excess deaths linked to HCV was below 1,000.⁽⁶⁴⁾

Figure 4.3: Distributions for quantities of interest arising from 10,000 simulations from the stochastic Monte Carlo baseline model for England.

A wide variety of estimates can be extracted from the model. For example, Figure 4.4 shows the estimated number of chronically-infected recipients alive each year under three different scenarios. The effect of chronic HCV-infection is considerably larger than that of having had a transfusion in the past.

Figure 4.4: The number of chronically-HCV-infected people in England surviving to each year under three different combinations of ‘hazards’ – the annual risk of dying – in the baseline deterministic model. (1) no increased risk from transfusion or chronic HCV infection, (2) increased risk following transfusion (age-stratified), (3) increased risk following transfusion and infection. There is an increase in the number of people surviving to 2001 (the point at which all people transfused between 1970 and 1991 could have survived to 10 years), and after this there is a steady fall in the number of people surviving.

Numbers identified by funds

Chapter 5 reports data on successful applicants to the variety of funds established to support those HCV-infected by blood or blood products. The most recent and relevant data from the Skipton Fund⁽⁶⁵⁾ seem to be from 2016: there had been 4,165 approved applications in England, with around 50% by persons who did not have bleeding disorders. Hence, we would estimate around 2,080 registered as having been chronically HCV-infected from transfusions in England. Since the Skipton Fund was aware that 619 of its UK-claimants had died, then (pro rata) around 484 of these deaths may have occurred in England which gives a 23% mortality rate (484/2,080), and so around 1,600 chronically HCV-infected claimants (2,080 - 484 = 1,596) in England still surviving in 2016. The current England Infected Blood Support Scheme (EIBSS) does not store registrations separately by blood-disorders or transfusion.

The number of surviving claimants is below our estimate of 2,200 (see Table 4.1) chronically HCV-infected survivors to the end of 2019, but not substantially so.

Adapting the model to Scotland, Wales and Northern Ireland

We assume the same model structure and assumptions for each of the four nations of the UK. Nation-specific data were unavailable for many parameters, and so we have adopted the parameters for England that are discussed in detail below, with the following exceptions:

• Task 1: 1991 HCV infected donor prevalence: Scotland value (0.088%) from S&G.
• Task 1: The trend of HCV-infectious IDUs: Scotland values from S&G.
• Task 1: Year of deferral reduction (Assumption (c)): Scotland value (1984) from S&G.
• Task 2: The number of blood donations: pro rata by population for Wales and Northern Ireland, and S&G values for Scotland.
• Task 5 and 6: The probability of surviving to the end of 2019 assuming survival to 10 years post-transfusion, with and without the effect of chronic HCV infection: these are based on nation-specific life tables as published by the Office for National Statistics⁽⁶⁶⁾.

Further details are provided in Appendix Table 4.1.

The use of blood in Northern Ireland in the 1970s will have been impacted by the Troubles and by the armed forces’ use of blood, which may have been donated by combatants at field hospitals. We were unable to obtain data to account for this and, therefore, our estimates for Northern Ireland may be an overestimation.

Table 4.2 presents the estimates for the baseline models of each of the four nations using the stochastic model. It is clear there is substantial uncertainty around all our UK estimates.

Table 4.2 Estimates for each of the four nations of the United Kingdom from the respective baseline stochastic model (median with 95% uncertainty intervals). Numbers above 1,000 are rounded to 3 significant figures, numbers between 100 and 1000 are rounded to nearest 10 and numbers below 100 are rounded to nearest 5. For the Total UK, we have added the unrounded medians for the four nations, and assessed the multiplicative uncertainty as a weighted average of those for the constituent nations.⁽⁶⁷⁾ We note that 64% of persons in the UK who were chronically infected with HCV through transfusion and survived to the end of 2019 were female (95% uncertainty: 58% to 69%).

Limitations:

• The estimates for Scotland somewhat understate the uncertainty, as we are not including uncertainty in the estimated prevalence of ever-IDUs.

Estimates from other sources – Summary Findings

There has been substantial variation in previous estimates of the number of HCV infections from blood transfusions, some of which have been strongly contested. In this section, we explore the reasons why past estimates differ so much, and contrast them with our own conclusions.

Department of Health (UK)

A widely discussed estimate is from the Department of Health 2011,⁽⁶⁸⁾ who estimated 28,043 post-transfusion HCV-infections for the UK. This is reported as being derived from the estimate by Soldan et al. (2002)⁽⁶⁹⁾ for England, which was 23,500 for 1970-1991. The possible derivation of the 28,043 was discussed in the Penrose Inquiry,⁽⁷⁰⁾ who observed that between 1970 and 1991 England had around 83% of the UK population. Scaling up proportionately gives 23,500/0.83 = 28,300, a close approximation to the claimed 28,043.⁽⁷¹⁾ The Department of Health estimate for Scotland, based on respective populations in 1991 (5.105m/47.88m), would be around 2,500.

Soldan (England)

Soldan’s (2002) estimates are based on analysis of the English look-back programme. When put within the structure of our model, the ideas behind the steps are essentially as follows:

• 0.066% HCV antibody prevalence in donations in late 1991 is assumed to hold throughout 1970-1991.
• 25,864,035 donations collected in England, 1980 – August 1991.
• 41,382,456 components produced (assuming 1.6 components from each donation).
• 26,898,596 components transfused (assuming 65% of components are transfused).
• 17,753 of these were HCV antibody positive (assuming 0.066% of all donations).
• 13,314 HCV RNA positive (i.e. HCV-infectious) components transfused, each leading to an infected recipient (assuming 25% of HCV antibody positive donors are not infectious).

This does not quite match Soldan’s 13,500, as she uses more precise knowledge from the look-back programme.

Soldan then adds: “If the prevalence of anti-HCV amongst blood donors during the 1970s was assumed to be the same as at the end of 1991, inclusion of the 1970s data would generate approximately 10,000 extra HCV-infected blood recipients.” Hence, the total of 13,500 + 10,000 = 23,500 is based on assuming the prevalence of 0.066% held throughout the whole period.

Our central baseline estimate of 22,000 for England is slightly below that of Soldan.

Soldan (Scotland)

For the Penrose Inquiry, Soldan adapted the England methodology estimate,⁽⁷²⁾ factoring in the higher HCV antibody prevalence among blood donors in Scotland in the first six months of HCV screening (0.088%).⁽⁷³⁾ She estimated 3,498 HCV infections following transfusion between 1980 and 1991, which is considerably higher than our estimate.

Penrose (Scotland)

The Schnier and Goldberg (S&G) analysis for the Penrose Inquiry has been discussed previously, and is summarised in Penrose 3.176-3.193.⁽⁷⁴⁾ A major difference between the S&G model and Soldan’s analysis concerned the proportion of HCV-infectious donations each year between 1970 and 1991; Soldan assumed this was constant, while S&G allowed the proportion to vary according to estimated numbers of HCV-infected ever-IDUs in the population.⁽⁷⁵⁾ S&G’s reported estimates, with sensitivity analyses, are shown in Table 4.3.

Table 4.3 Estimates with 90% uncertainty bounds from S&G analysis for the Penrose report.

This Scottish re-analysis shows that the major reason why Soldan’s estimates are higher is the assumption of a constant donor HCV-infection rate throughout the period 1970-1991. By relaxing the assumption, and introducing dependence on the changing prevalence of HCV infected ever-IDUs, we would expect the widely-reported 28,000 UK total to decrease, which has been observed.

Our more complex ‘hybrid’ model results in a central estimate of 2,740 for Scotland, which is higher than Penrose’s baseline, but not as high as Soldan’s ‘constant’ model.

Claims of above 100,000 infections

There have been estimates of up to 400,000 HCV infections from blood transfusions in the UK since 1970.⁽⁷⁶⁾ This appears to be largely based on assuming an HCV-infection prevalence in the UK donor population of at least 2% between 1970 and 1985, falling to 1% between 1986-1991, quoting a PHLS study by Ramsey et al. published in 1998.⁽⁷⁷⁾ Also quoted is Gunson’s 1986⁽⁷⁸⁾ estimate of a 3% incidence of non-A non-B hepatitis, which in turn uses a 1974⁽⁷⁹⁾ study funded by the Medical Research Council (MRC) of 768 patients (474 females, 294 males) who were transfused between 1 July 1969 and 30 June 1971 with 2,824 units of blood and who consented for follow up. During the two-year period, the centre used 8,953 units of blood for transfusion. Hepatitis B antigen was detected in 13 (0.15%) and hepatitis B antibodies in eight (0.09%). Five patients in the MRC survey received blood that was subsequently shown to contain hepatitis B antigen. Gunson also cites supporting evidence from Collins et al. (1983),⁽⁸⁰⁾ who concluded six out of 248 (2.4%) cardiac surgery patients had post-transfusion non-A non-B hepatitis. The 1974 MRC study reports eight cases of post-transfusion “icteric or anicteric post-transfusion viral hepatitis”, an incidence of 1%. Other estimates include Garson et al. (1990)⁽⁸¹⁾ who found six out of 1,100 (0.55%) transfusion recipients were “repeatedly reactive in a commercial assay for antibodies to the C100 protein of hepatitis C virus”. All these estimates are well above the empirical estimate in Section 4.1, based on Soldan’s assessment of a prevalence of 0.066% (532/808,938) HCV antibody positive blood donors tested in England between September and December 1991, after both enhanced self-deferral and HIV antibody screening of blood donors.

There are concerns with the statistics used in the analysis above. For example, it is important to note that the 2.6% prevalence of HCV infection in people given transfusions before 1985, as reported in Ramsey et al. (1998), is not based on a sample of transfusions, but a sample of HCV test results carried out in the 1990s. Such tests tend to be taken if HCV infection is suspected, mainly through symptoms or being in a higher-risk group; indeed, in this study, the rate of HCV infection in those who had not received transfusions was 10% (Table 5⁽⁸²⁾). Due to this ‘targeted’ sampling, the quoted rates cannot be used as an indication of the rates in all transfusion recipients. In addition, Collins et al. (1983) did note some short-term rise in liver function tests but concluded: “The incidence of significant chronic liver disease after blood transfusion possibly attributable to a non-A, non-B hepatitis agent was therefore only 0.4%”. It is important, in general, to realise that before the availability of testing for the virus itself, the use of abnormal liver function tests as a marker of HCV infection cannot be relied upon to give precise estimates.

In her witness statement of October 22nd 2021⁽⁸³⁾ Dr Soldan responded to requests by the Inquiry to critique estimates of HCV-infection rates in transfusion-recipients of 1% (MRC) and 0.55% (Garson). Regarding the MRC 1974 data, she was unsure that the cases reported would have been confirmed as HCV antibody positive if a definitive test had existed – particularly because acute symptoms from HCV would not be typical – and she felt unable to derive an incidence estimate from the data presented. For the Garson data, she noted that only one of the six notional HCV-infected donations went on to infect the recipient, which casts doubt on the accuracy of the assay.

It appears clear that the incidence of post-transfusion HCV infections cannot be properly estimated before reliable tests were adopted in the early 1990s, and so estimates of the number infected, when based on pre-1990 studies, need to be treated with great caution.

Very high numbers of HCV infections have also been quoted in France. An official report on their HCV infections, with lead author Professor Max Micoud, was published in 1993. Discussing the Micoud report, Delorme⁽⁸⁴⁾ states: “If we evaluate this risk (of post-transfusion contamination) at 5%, between 1970 and 1989, 25,000 people have been contaminated each year, potentially leading to chronic hepatitis”, and “In 1992, the number of people infected with HCV in France can be estimated at between 500,000 and 2 million. Of these, 100,000 to 400,000 infections could be of post-transfusion origin”. This estimate of 5% is in contrast to the observed incidence in France when reliable testing was introduced: Nalpas et al. (1998)⁽⁸⁵⁾ reported an incidence of infected blood donations of 0.37% in 1991 after self-deferral and HIV testing – around four times that in the UK, but still substantially below the 5% estimated in the Micoud report.

Task 1 For each year 1970-1991 in England, what proportion of donations was HCV-infectious?

Assumption (a) We adopt a ‘hybrid’ model of HCV-infectious donors that has been explained previously; a first component comprising HCV-infectious ever-injecting drug users (ever-IDUs), and a second constant component comprising HCV-infectious non-IDUs.

This allows us to break the task into a series of smaller questions:

• Task 1.1. Just after screening of blood donors started in September 1991, what proportion of donations were HCV antibody positive?
• Task 1.2. Just after screening started in September 1991, what proportion of donations were HCV-infectious (RNA positive)?
• Task 1.3. How many HCV-infectious ever-IDUs were alive during each year between 1970 and 1991?
• Task 1.4. What was the step-down in donations from ever-IDUs due to guidance and HIV antibody testing in the mid-1980s?
• Task 1.5. What constant level of HCV-infectious (RNA positive) non-IDU donors might it be reasonable to assume between 1970 and August 1991?

Together these steps, and the hybrid model, allow us to estimate the proportion of infectious donations in each year.

Task 1.1. Just after screening of blood donors started in September 1991, what proportion of donations were HCV antibody positive?

Background: Reliable screening of donations for antibodies to HCV began in September 1991 and so could provide good information about the proportion of donations that, earlier in 1991, were HCV antibody positive.

Confidence that available evidence can answer the question? High.

Available sources of evidence: National Blood Authority/PHLS Infection Surveillance Report’s tables⁽⁸⁶⁾ are reported in Soldan’s PhD thesis,⁽⁸⁷⁾ covering England and Wales. The total of 532 HCV antibody positive donor tests is also confirmed in the UKHSA Witness statement to the Inquiry, with subdivision of the total into 94 new donors and 438 repeat donors.⁽⁸⁸⁾

Analysis and conclusions:

Table 4.4 Results of donations that were tested for HCV antibodies between September and December 1991, England and Wales.

Comments and limitations: In their submission to the Penrose Inquiry, S&G reported that between September 1991 and February 1992, 159 of 180,000 donors tested in Scotland were HCV-antibody positive, which is 0.088%⁽⁸⁹⁾, higher than the rate in England and Wales.

Task 1.2. Just after HCV antibody screening of donors started in September 1991, what proportion of donors were HCV-infectious (RNA positive)?

Assumption (b) We assume that a constant proportion of those donors previously infected with HCV (HCV antibody positive) were actively HCV-infectious (RNA positive) and so would transmit the virus in their blood.

Background: Of those infected with HCV, a proportion will spontaneously ‘clear’ the virus naturally, and so stop being infectious. In particular, infected blood donors who clear their virus will not pass on the infection in their donated blood. S&G assumed a value of 75% for the proportion who continue to be infectious.⁽⁹⁰⁾

Available sources of evidence: Micallef et al. (2006)⁽⁹¹⁾ carried out a systematic review of the medical literature and found 31 relevant studies including 675 subjects. They estimated an overall clearance rate of 26% (95% confidence interval: 22% to 29%), with lower rates in males. For transfusion recipients, the clearance-rate was also lower at 18% (95% confidence interval: 13% to 24%).

Confidence that available evidence can answer the question? High

Analysis and conclusions: We assume a donor clearance rate of 26% (95% confidence interval: 22% to 29%) and so a central estimate of 74% of HCV antibody positive blood donors who are infectious. The estimated proportion of HCV-infectious donations in January to August 1991 (i.e. prior to HCV antibody screening of donors) is therefore 0.066% x 0.74 = 0.049%.

Table 4.5. Results of donations that were tested for HCV antibodies between September and December 1991, England and Wales.

Comments and limitations: Since Micallef et al. (2006) is a recent systematic review of the literature, we adopt their overall estimate in the model. Although the 26% includes studies on post-transfusion cases, we use the overall figure as donors may themselves have had transfusions.

Task 1.3 How many chronically HCV-infected ever-injecting drug users (ever-IDUs) were alive during each year between 1970 and 1991?

Background: S&G assume that HCV-infectious donations mainly came from those who had ever injected drugs (‘ever-IDUs’) or had been indirectly infected by injecting drug users, and therefore HCV-infectious donations were related to the prevalence of chronic HCV infection in this population subgroup. MacLennan et al. (1992) reported that “in this preliminary investigation of our HCV-seropositive donor population, IDU appears to be a predominant risk-factor for transmission of HCV.”⁽⁹²⁾

We have broadened the S&G model to a ‘hybrid’ model for England, by assuming HCV-infectious donors are a mixture of ever-IDUs and a group that has a constant HCV antibody prevalence throughout 1970-1991 (Assumption (a)). Different mixture proportions are explored, from 0% to 100% of the HCV antibody prevalence in 1991.

Available sources of evidence: There have been estimates of the number of current injecting drug users,⁽⁹³⁾ but few estimates of the number of HCV-infectious current and past IDUs. We judged the most authoritative to be the Public Health England study, published by Harris et al. (2019).⁽⁹⁴⁾ They used an extended back-calculation approach to recreate the HCV infection process leading to observed disease data. An updated version of this model is used to generate UKHSA’s annual prevalence estimates of HCV.

We have been provided with previously unpublished estimates for HCV-infectious current and past IDUs and their sum (ever-IDUs) for the period 1971-1991 (see Appendix A.6 for estimation for 1970, and Appendix Figure 4.1 for contrast between England and Scotland).

Confidence that available evidence can answer the question? Moderate/High.

Analysis and conclusions: The estimates are shown in column (a) of Table 4.6, while column (b) shows the estimates normalised to be 100 in 1991. Table 4.6 column (c) and Figure 4.5 display the estimated percentage of HCV-infectious donations (assuming ever-IDUs are the sole source), pegged to the estimated HCV-infectious donations in 1991. There was a steady increase in the prevalence of HCV-infectious ever-IDUs over this period, increasing roughly five-fold between 1970 and 1991.

Comments and limitations: These estimates are highly sensitive to the UKHSA’s model assumptions, such as accurate recording of HCV-related outcomes, and that progression probabilities are constant over time. Although the 95% intervals allow for statistical uncertainty, the possibility of bias in the included parameters is not incorporated.

Sensitivity analysis: Current IDUs have poor venous access so that past-IDUs may be more likely to have presented as donors. Hence, we also consider the effect of restricting to past-IDUs (Scenario A in Table 4.1).

Task 1.4 What was the step-down in donations from ever-IDUs due to guidance and HIV antibody testing in the mid-1980s?

Assumption (c) As previously demonstrated in Table 3.1, in the mid-1980s there was strong guidance that those who were currently injecting or had ever injected drugs should avoid giving blood due to the risk of HIV transmission. We therefore assume that in or after 1985 there was a ‘step-down’ in donations from the HCV-infectious population.

As we saw in Table 3.1, the advice and language of the leaflets changed: advice against donations from people who had previously injected drugs was introduced in Scotland in 1984, but only for the rest of the UK in 1987.

Confidence that available evidence can answer the question? Low/Moderate.

Available sources of evidence: There does not seem to be any reliable evidence for the effect of the changed guidance in the mid-1980s. In particular, samples from earlier donations had not been retained, preventing subsequent testing for HCV antibodies. In an analysis of people who had been tested for HCV antibodies,⁽⁹⁵⁾ the proportion who were HCV antibody positive among those who had received transfusions fell by roughly 60% between those transfused before 1985 and those after 1985 (from 2.6% to 1.0%), but this is a highly selected group of people who were subsequently HCV tested, favouring those with disease-progression. Moreover, the pre-1985 date-range was large.

In their submission to the Penrose Inquiry, S&G said that the deferral policy introduced by SNBTS in 1984 was assumed to have reduced the HCV prevalence in the donor population persistently by 66%, saying “This assumption was based on limited local data and expert opinion.”⁽⁹⁶⁾ We also base our assessment on informal discussions with members of Infected Blood Inquiry Expert Groups who were knowledgeable about both the contemporary circumstances and the effects of guidance on behaviour: see Appendix (A.8 onwards) for details.

Analysis and conclusions: Two broad opinions emerged concerning the 67% (two-thirds) reduction (the better supported) or 33% (one-third) reduction. Those who thought such guidance would have very limited effect pointed to the general lack of adherence to health advice leaflets, possible unwillingness to identify as someone affected by drug use, and possible incentive to donate in order to get an HIV antibody test without attending a special clinic: those who tested HIV antibody negative would go on to donate, but could be HCV antibody positive.

Assumption (d) We adopt a 67% reduction in donations from HCV-infectious ever-IDUs as part of our baseline scenario.

Comments and limitations: There are inevitable limitations of using expert judgement.

Sensitivity analysis: England’s messaging on self-deferral by past-IDUs was not explicit until 1987. Hence, we consider the impact of assuming a 33% step-down in 1987 for past-IDUs. Results are shown in Table 4.1 as Scenario A. Further sensitivity analyses are reported in Table 4.16.

Task 1.5. What constant level of HCV-infectious (RNA positive) non-IDU donors might it be reasonable to assume between 1970 and August 1991?

Our ‘hybrid’ model allows for a lower-level constant threshold with the remainder time-varying in accordance with England’s chronically HCV-infected ever (or past) IDUs. IDUs would be expected to be the main but not sole driver of HCV transmission, with other routes including iatrogenic transmission historically, UK personnel being treated overseas, and immigration from countries where HCV was prevalent.

Assumption (e) We assume that a baseline 25% of blood donors’ HCV-infected prevalence in 1991 has been persistent throughout 1970-1991. This is based on judgement (and sensitivity analysis reveals that the precise value has little impact).

Analysis and conclusions: Table 4.6 shows the stages in developing our hybrid model, from ever-IDU prevalence, scaled to the main results are shown in Figure 4.5.

Sensitivity analysis: In Table 4.1 we consider the effect of assuming infections arise solely from ever-IDUs (Scenario B) or at a constant level regardless of IDU prevalence (Scenario C).

Figure 4.5: Stages in estimating the prevalence of infectious HCV donations:

(A) Estimated number of ever-injecting drug users (black columns) and percentage that each year contributes to the value observed in 1991 (red line) – Table 4.6, column (a).

(B) The trend observed in (A) is pegged to an observed value of HCV antibody positive donations in 1991 of 0.049% (red line) – Table 4.6 column (c). The black line shows an estimated effect (67%) of the deferral policy in 1985, which caused prevalence to increase beforehand – Table 4.6 column (d).

(C) The black line in (B) is adjusted to account for the scenario that 25% of the observed 1991 HCV antibody prevalence is not from IDUs and is constant across the period (light blue line). The dark blue line shows the resulting percentage of HCV antibody positive donations from these two sources. – Table 4.6 column (e).

Summary conclusions

Table 4.6. Results from our baseline model for England, showing steps in obtaining estimates of the percentage of HCV-infectious donations each year.

Task 2. For each year between 1970 and 1991, how many units of blood components were transfused?

Approach

Terminology: Unit (from S&G). The term ‘blood component units’ (shortened to ‘units’) is used here to denote any labile component of a whole blood donation, i.e. red cells, platelets, plasma or cryoprecipitate.

There does not appear to be a reliable source for the number of transfusions per year in the period 1970-1991. Counsel to the Inquiry have pointed out that in the 1984 specification of minimum requirements for documentation for blood transfusion, “No attempt was made to prescribe the format of that documentation due to the wide variation in practice across the country.”⁽⁹⁷⁾

We therefore follow S&G in working from the number of donations, and then estimating the resulting number of units transfused. We have therefore broken this task into a series of smaller questions.

• Task 2.1. How many blood donations were there annually in England between 1970 and 1991?
• Task 2.2. Per donation, on average how many units were actually transfused in hospitals?

Together, these provide the estimate of the number of units transfused per year.

Task 2.1. How many blood donations were there annually in England between 1970 and 1991?

Available sources of evidence: Our primary source is a letter (faxed) from the Department of Health to Roger Moore (National Blood Transfusion Service on 10/02/1992),⁽⁹⁸⁾ listing the number of blood and plasma donations each year in England and Wales between 1978 and 1990. These match (for appropriate years) the less precise numbers given in National Blood Transfusion Service Statistics of 1988.⁽⁹⁹⁾ For 1975 and 1976, our source is the Expenditure Review Report of the Blood Transfusion Service, 1978.⁽¹⁰⁰⁾ We have therefore been obliged to estimate the donations in other years through statistical modelling.⁽¹⁰¹⁾

Confidence that available evidence can answer the question? High.

Analysis and conclusions: We have modelled the counts of donations for 1970 – 1974, 1977 and the first 8 months of 1991 (since HCV antibody screening of donors was introduced in September 1991). The results are shown in Figure 4.8, with details of the model fit and extrapolation to other years in the Appendix (A.16 onwards), including estimates for other nations.

The England and Wales totals are then scaled to England by the average population across 1971 – 1991 (94.3%). Results are shown in Table 4.8, column (b).

Comments and limitations:

• We have used limited and imperfect historic data, but they appear to be the only viable data of their kind.
• We are assuming that (a) the rate of donations in the population did not differ between England and Wales; (b) the donations in 1970-1974, 1977 followed the general trend of 1975-1976, 1978-1979; and (c) the donations in 1991 followed the general trend of 1978-1990.

Figure 4.6 Number of blood donations in England and Wales, 1970 to 1991 (Jan-Aug).

Task 2.2. Per donation, on average how many units were transfused?

Available sources of evidence: Our primary source is the National Blood Transfusion Service Statistics of 1988.⁽¹⁰²⁾

Confidence that available evidence can answer the question? Moderate.

Analysis and conclusions: The analysis is illustrated in Table 4.7 for a single year – 1982.

Table 4.7 Worked example for 1982, showing derivation of estimate of the number of units transfused per donation.

Full data on the quantities above were only available for 1982 to 1988, which was a time of rapid change in the transfusion service. We extrapolated back to 1970 and forwards to 1991, assuming the rate of change was 25% of that in the period 1982-1988. Summary results are shown in Table 4.8 and Figure 4.7.

Figure 4.7 Extrapolation of units per donation from observed data in 1982 to 1988.

Comments and limitations:

• The primary limitation is the lack of data on transfusion practice over this period. The 2002 English National Blood Service look-back programme, a 2-year tracing exercise between 1995-1997, was unable to identify the fate of 23% of blood components issued for transfusion.⁽¹⁰³⁾
• A major assumption is that return rates for whole blood and red blood cells (RBC) reflects overall non-usage. Informal discussion with those with contemporary experience suggests there will have been additional ‘wastage’ that varied between hospitals and declined over time, but that these rates may be lower for cryoprecipitate, platelets and ‘plasma for local use’. Overall, our assumption may overstate the units used per donation, but not by a substantial amount.
• Soldan et al. ⁽¹⁰⁴⁾ assumed 1.6 units were available per donation in England, and 65% of available units were transfused, leading to an overall estimate of 1.6 x 0.65 = 1.04 units transfused per donation, corresponding to our estimate for around 1983.
• S&G⁽¹⁰⁵⁾ assumed 1.25 units were available per donation for Scotland, based on “limited local data and expert opinion”, and 56% of units were transfused, giving an overall estimate of 1.25 x 0.56 = 0.70 as a constant rate of units transfused per donation. Penrose concluded the 56% usage could be too low, and at least 66% was more appropriate,⁽¹⁰⁶⁾ which would lead to 1.25 x 0.66 = 0.825, our estimate for the early 1970s.
• Our changing rate should better reflect changing transfusion practices.

Final estimates are shown in Table 4.8. Column (d) is carried forward to the next stage of the model.

Table 4.8 Stages in estimating the number of units transfused in England, 1970-1991 – results reported to 3 significant figures.

*Estimated by interpolating from known data for 1975, 1976, 1978, 1979. (See Appendix for further details).

** Estimated by interpolating from data for 1978-1990: testing introduced in September 1991, and so this is ⅔ of the modelled value. (See Appendix for further details).

*** Estimated by interpolating from known data for 1982 – 1988. (See Appendix for further details).

For the years 1977-80, figures relate to the number of donors reporting at blood transfusion centres.

Task 3. For each year between 1970 and 1991 in England, how many transfusion-recipients (in age-sex bands) were infected with HCV?

The analysis in Task 1 has provided estimates for England, for each year between 1970 and 1991, of the proportion of donations that were HCV-infectious (RNA-positive). Task 2 has estimated the number of units transfused each year. Simply multiplying these numbers together estimates the number of infectious units transfused each year. We then follow S&G⁽¹⁰⁷⁾ in making two assumptions.

Assumption (f). All units from HCV RNA-positive donors are assumed to be equally infectious when transfused to a recipient, and every recipient of a contaminated unit subsequently developed HCV infection.

Assumption (g). There is a negligible chance of a transfusion recipient receiving two infected units. Table 4.5 shows that around 1 in 2,000 units were infectious: even if someone receives 10 units, the chance that more than one of these was infectious is only 1 in 90,000.

Assumptions (f) and (g) imply that every infected unit led to a unique infected recipient. Table 4.9 shows the calculations to give the annual number of individuals infected with HCV by transfusion in England for 1970 to 1991.

Table 4.9 Estimation of the annual number of individuals infected with HCV by transfusion in England, 1970-1991 (totals are the rounded sums of the unrounded values).

Age-sex distribution of those infected by transfusions

Available sources: Wallis et al. (2004)⁽¹⁰⁸⁾ studied 2,899 transfusion recipients during June 1994 in a population of 2.9 million served by a single regional blood transfusion centre (Newcastle). The full data were kindly provided by the researcher-team for more detailed analysis. The age-sex distribution of recipients in the Wallis cohort is shown in Figure 4.8 and Table 4.10.

We were able to compare this to the age-sex distribution of transfusion recipients within the SNBTS record-linkage study in 1999 described later (4.102). Age groups 0 and 1 – 9 from SNBTS were pooled to be 0 – 9 and age groups 80 – 89 and 90+ from Wallis et al. were pooled to be 80+ in order to allow for direct comparison. This showed close agreement of the age-sex distribution between these two independent sources as shown in Figure 4.8, giving confidence to this application across different settings.

Appendix Tables 4.8 – 4.10 present the age-sex distribution of those estimated to be infected in Northern Ireland, Scotland and Wales, respectively.

Figure 4.8 Age-sex distribution of two independent transfusion cohorts: Wallis et al. (2004) 1994 cohort from North West England, and Scottish National Blood Transfusion Service (SNBTS) 1999 cohort.

An alternative English source is the Epidemiology and Survival of Transfusion Recipients (EASTR) by Wells et al.,⁽¹⁰⁹⁾ who studied 68,600 recipients from 29 transfusion centres in the year starting October 2001. Wells et al. provide information by age-band on the survivorship of their transfusion recipients, although their sample is from October 2001 – September 2002, which is further from our period of interest than the June 1994 transfusions in Wallis et al.. Nevertheless, the age-sex distribution of RBC transfusions in EASTR, and the overall distribution in Wallis, are similar.

Assuming the age-sex distribution of Wallis et al., Table 4.10 shows how we would expect the total number of HCV infections via transfusion to be distributed among age-sex bands.

Table 4.10 Estimated age-sex distribution of those infected with HCV through transfusion in England, 1970-1991.

Limitations: Apart from the assumptions listed previously:

• The totals are not broken down by type of transfusion and so our counts do not reflect the risk to individuals: someone receiving many units will necessarily be at higher risk than a person who receives only one unit.
• We assume the age-sex breakdown of transfusion recipients holds for the whole period, whereas transfusion practice will have changed.

Task 4. How many chronic HCV-infected transfusion recipients survived 10 years post-transfusion?

Task 3 provided estimates of the number of individuals infected with HCV via transfusion in each year. We now extend the horizon to 10 years post-transfusion for those who were chronically infected. We obtain estimated 10-year survival rates for different age-sex bands from studies of transfused patients, and apply these survival-rates to the output of Task 3 to obtain estimated numbers of 10-year survivors.

We want to focus attention on those who have been chronically infected, i.e. have not cleared the virus, and so it is convenient to estimate first the nominal number of such cases were they to survive 6 months post-transfusion (some will actually die before this is determined, but this does not affect our longer-term modelling). We have already determined in Task 1.2 the evidence to support the following assumption: Assumption (h) A proportion (18%: 95% confidence interval 13% to 24%)⁽¹¹⁰⁾ of transfusion recipients with HCV-infection is assumed to clear the virus within 6 months of acquisition.

Assumption (i) Chronic HCV infection does not influence recipients’ survival for the first 10 years post-transfusion. Studies of HCV progression often only begin to trace recipients of implicated transfusions several years or more than a decade after the implicated transfusions occurred.⁽¹¹¹⁾ Traced recipients are therefore survivor-selected and, in general, can tell us little about the risk of HCV related death in the first decade of follow-up. More generally, meta-analyses⁽¹¹²⁾,⁽¹¹³⁾ concentrate on earlier stages of HCV morbidity (rather than mortality) and specifically the development of cirrhosis. Progression rate to cirrhosis by 20 years after HCV infection is 7% (4% to 12%), which underscores that mortality from HCV-infection – although not impossible with adverse risk factors – is very unlikely in the first decade of chronic HCV infection.

Confidence that available evidence can answer the question (given previous steps in model)? Moderate/High.

Available Sources of Evidence. We have already used the demographical data of Wallis et al. (2004)⁽¹¹⁴⁾ in Task 3, based on 2,899 transfused during June 1994. S&G also used the published 5-year survival data from Wallis et al. (2004). The full data were kindly provided by the research-team for more detailed analysis.

The follow-up for the Wallis data only goes up to 7 years. Longer follow-up is provided by Morley et al. (2016)⁽¹¹⁵⁾ in the EASTR study previously discussed in Task 3, based on 68,600 recipients of transfusions in the year from October 2001. They provide 10-year survival rates broken down by RBC, fresh frozen plasma (FFP) and platelet transfusion, and by broad age-groups. The EASTR study does not provide survival by the 10-year age-bands required for our analysis, and also their follow-up period of 2001-2011 is not as appropriate as Wallis’s period of 1994-1999. Therefore, we preferred to use the 5-year survival rates from Wallis, and extend them to 10 years using the change in survival between 5 and 10 years in EASTR. The technique for doing this is outlined in the Appendix (A.29), together with comparisons of survival in the two cohorts.

Analysis and conclusions

The resulting estimates are shown in Table 4.11.

Table 4.11 Estimated percentage surviving 5 and 10 years post-transfusion.

Limitations:

• The survival proportions disregard both the reasons for transfusion and type of transfusion. Survival rates for specific groups of people may vary substantially around these averages.
• We are assuming the Wallis et al. (1994) survival rates for the North of England apply to the whole of England.

Table 4.12 summarises the results of this task, which are passed onto further stages of the model; results for the other nations are provided in Appendix Tables 4.16 - 4.18.

Table 4.12 Estimated annual number of individuals with chronic HCV infection by transfusion in England, 1970-1991, and surviving 10 years post-transfusion. Column (a) is obtained by applying the estimated clearance rate (around 18%) to the estimated number of infections in Table 4.9. Column (b) is obtained by applying the survival rates in Table 4.11 to the age-sex profile in Table 4.10.

Task 5 How many chronic HCV 10-year-survivors would have survived to the end of 2019, assuming no excess risk from HCV?

Background: Task 4 estimated the number of survivors 10 years after being chronically infected with HCV at transfusion. We now extend this model to 2019, allowing for the additional risk of having had a transfusion in the past, but not yet allowing for any risk linked to chronic HCV infection. We note that S&G assumed no excess risk for being more than 5-years post-transfusion.

Confidence (assuming previous modelling is appropriate): Moderate/High.

Available evidence: For background mortality, contemporary life-tables for England⁽¹¹⁶⁾ provide the average annual risk (the ‘hazard’) of dying in any year, given age at the start of the year. We then have to model the excess risk of having had a transfusion as a ‘hazard ratio’ that multiplies the annual background hazards, and will depend on years post-transfusion, age at transfusion and sex. A major Dutch study⁽¹¹⁷⁾ followed up over 2 million transfusion recipients in the Netherlands between 1996-2006, but their survival data were not split by sex and used a small number of rather wide age categories. A Scandinavian study⁽¹¹⁸⁾ of over 1 million transfusion recipients in Denmark and Sweden reported average hazard ratios of 1.4 for 10-15 years post-transfusion, and 1.3 for 15-20 years. Their graphs suggest considerable variation by age.

To assist the Infected Blood Inquiry, the SNBTS initiated a record-linkage study which follows up for mortality four 5-yearly cohorts of transfusion recipients, who joined their respective cohort at the date of their first transfusion in the cohort-year. The SNBTS database’s coverage is not Scotland-wide but accounts for around 40% of Scotland's transfused patients. The cohort data are extracted from the SNBTS blood bank Laboratory Information Management System, eTraceline. The 1999 cohort, for example, represented three large teaching hospitals across Scotland and one large general hospital. The cohort-years are 1999 (1999-cohort); 2004 (2004-cohort); 2009 (2009-cohort); and 2014 (2014-cohort) to allow calendar-year trends in the deployment of units to, and survival outcome for, recipients of RBC transfusions to be monitored up to 20 years post-transfusion.

Follow-up for mortality was to 31 December 2019. The vast majority of first transfusion recipients in any cohort-year has received RBC and so we focus on 1999-RBC-cohort, for whom we have 20 years of follow-up post-transfusion; and on the 2004-RBC-cohort, for whom we know survival status for 15 years post-transfusion.

Assumption (j): Lacking evidence beyond 20 years post-transfusion, but mindful that hazard ratios decreased markedly during the second 10 years of follow-up, we assume that there is no residual transfusion-related hazard for those who survived for 20 years post-transfusion.

Analysis and conclusions: Using the appropriate Scottish life-table for each year between 10 years post-transfusion and 2019 (e.g. 1999 – 2001 for the year 2000), annual hazards were extracted per sex and age-band at transfusion (using, for example, age 25 for those aged 20-29 years at transfusion), and updated to reflect the cohort’s age in the year in question. These determined the expected number of deaths in the SNBTS cohorts. The additional risks from 10 years post-transfusion were determined via the ratio of observed to expected deaths from the SNBTS’s record-linkage study (see Appendix A.38 onwards) – this was done with and without stratification by age-band. For example, for those aged 50-59 at transfusion, we estimate a hazard ratio of 2.0 for 11-15 years post-transfusion, and 1.5 for 15-20 years.

The likelihood of surviving 10 years post-transfusion has already been covered in Task 4. Transfusion-adjusted hazard rates were obtained by multiplying the relevant hazards from the nation-specific life-tables for 11-15 years and 16-20 years post-transfusion by the observed/expected ratios calculated previously. The product of one minus the hazards across each year per age-band and sex was used to estimate the probability of survival to 2019, with and without the additional transfusion risk (Figure 4.9 and Appendix Tables 4.20 and 4.27).

These transfusion-adjusted hazard rates were applied to the output from Task 4 to give the estimated number of survivors to the end of 2019, both without and with allowance for the excess mortality of being post-transfusion. This gives rise to columns (b) and (c) of Table 4.13.

Figure 4.9 Estimated probability of surviving until 31 December 2019 post-transfusion in England, given age and transfusion year and assuming excess age-stratified post-transfusion hazard.

Sensitivity analysis: Our baseline scenario pools the broadly similar age-specific post-transfusion hazard ratios for males and females. We explore two other options in the sensitivity analysis below: no additional transfusion hazard (Scenario G), and not allowing hazard ratios to be age-dependent, merely different ratios for 11-15 years (across all ages) post-transfusion versus for 16-20 years (Scenario H).

Task 6 Of those infected with chronic HCV through transfusion between 1970 and August 1991, how many died of HCV-related causes by 2019?

The task: Task 4 estimated the number of those infected with chronic-HCV who survived 10-years after being infected at transfusion, while Task 5 modelled subsequent survival to the end of 2019, assuming additional risk for being post-transfusion but no effect of HCV infection. We now consider the additional risk for being chronically HCV-infected via transfusion; adding the deaths attributable to HCV in each year provides estimates of the deaths related to (treated) chronic HCV-infection.

Available evidence. Authors of the study⁽¹¹⁹⁾ used in Task 1.3 to estimate the number of HCV-infected IDUs also examined the long-term effect on mortality of HCV infection post-transfusion. They used a case-control design, whose participants were traced look-back patients in England who had received an HCV-implicated blood transfusion. We were kindly granted access to the results ahead of publication. This UKHSA team compared all-cause and liver-related mortality of 924 patients infected with HCV via transfusion as the route of infection against 443 patients who had received an HCV-implicated transfusion but were not HCV antibody positive. The team found an overall hazard ratio of 1.53 (95% CI 1.17 – 2.00) for all-cause mortality among HCV-RNA positive patients compared to transfused controls who were not HCV-infected. The estimated hazard ratio of 1.53 was constant across three decades of traced follow-up since the HCV-implicated transfusion, and represents treated natural history since directly acting antiviral therapy will have had a beneficial impact.

Assumption (k) Chronic HCV infection increases the annual risk of dying from 10 years post-infection: our baseline assumption is a 53% increased risk.

Analysis and conclusions: We applied the additional hazard ratio estimated as described above to the hazards assuming additional transfusion risk for chronically HCV-infected patients surviving at least 10 years post-transfusion. This was applied uniformly to each age-group and sex every year from 10 years post-transfusion. The resulting probabilities of survival are shown in Appendix Tables 4.38 to 4.45, and the estimated number of survivors to 2019 shown in column (d) of Table 4.13: estimates for other nations are provided in Appendix Tables 4.46 to 4.54.

Table 4.13 Estimated annual number of individuals infected with chronic HCV by transfusion in England, 1970-1991, and surviving to 2019, both without and with allowing for any effect of chronic HCV infection. Numbers over 100 are rounded to 2 significant figures, whilst those under 100 are rounded to the nearest 5. Note: it is incorrect to estimate the number of HCV-related deaths by subtracting column (d) from (c).

The ‘attributable fraction’⁽¹²⁰⁾ among those exposed to a risk is simply 1 - 1/HR, where HR is the hazard ratio, i.e. the relative risk in each year. For HR = 1.53, we would therefore assume the fraction of deaths attributable to chronic HCV infection 1 – 1/1.53 = 35%.

Table 4.14 presents the estimated number of survivors by age-sex band, pooled over years, and the estimated number of HCV-related deaths.

Table 4.14 Estimated number of chronic HCV infections, survivors until 2019, and number of HCV-related deaths in England – by age-sex band, pooled over years of transfusion 1970-1991. Survival estimates take into account additional risk from both transfusion and HCV. Numbers over 100 are rounded to 2 significant figures, whilst those under 100 are rounded to the nearest 5.

Comment: Table 4.14 shows that females (10,000) outnumbered males (8,000) who were chronically HCV-infected by transfusion during 1970-1991 to a modest extent, whereas females (1,400) greatly outnumber male survivors (800) to the end of 2019 – female proportion is 64% (95% uncertainty interval: 59% to 68%) This reflects both the better female post-transfusion survival (Task 4), and the lower female mortality risk in general.

Also, whereas 70-79 years was the most common age-band at transfusion for both female and male recipients who were chronically HCV-infected, HCV-infected survivors to the end of 2019 would require to have survived at least 28 years post-transfusion, and so are dominated by those under 50 years of age at transfusion (1,350/1,400 female survivors). Of an estimated 1,400 female HCV-infected survivors to the end of 2019, 880 (over half) were aged 20-39 years at transfusion; with a further 190 aged 0-9 years at transfusion. Of the estimated 800 male HCV-infected survivors to the end of 2019, 280 were aged 20-39 years at transfusion; with a further 350 aged 0-9 years at transfusion.

Table 4.15 Estimated age distribution (in years) of people with chronic HCV infection from transfusion between 1970 and 1991 in England, who are alive in December 2019. Numbers over 100 are rounded to 2 significant figures, whilst those under 100 are rounded to the nearest 5.

Table 4.15 represents the same 1,400 female and 800 male HCV-infected survivors by their estimated age-band in completed years at 31 December 2019. Male survivors outnumber females under 50 years (340 males versus 210 females). At older ages, however, males are substantially outnumbered by female HCV-infected survivors: for example, at ages 50-69 years (270 males versus 730 females) and at 70+ years (190 males versus 480 females).

We should expect the female dominance displayed in Table 4.15 to be reflected in surviving claimants who were chronically HCV infected by transfusion and whose age in 2019 was 50+ years. The expected female dominance contrasts sharply with male dominance for people with bleeding disorders who were chronically HCV-infected. See further analysis in Chapter 5.

Further sensitivity analyses

A larger selection of sensitivity analyses is provided in Table 4.16, with the first five rows identical to those in Table 4.1. The rows comprise:

• Results from a deterministic baseline model assuming all parameters are fixed at their assumed values.
• A stochastic baseline model, in which parameters and outputs have associated uncertainty.
• Scenario A: A deterministic sensitivity analysis to Assumptions (a) (c) and (d), in which we assume the first component in HCV-infectious donations comprise past-IDUs rather than ever-IDUs, the deferral effect occurred in 1987, and led to only a 33% reduction in infectious donations.
• Scenario B: A deterministic sensitivity analysis to Assumption (e), in which we assume that there is no second component of non-IDU infectious donors. This is analogous to the analysis of S&G.
• Scenario C: A deterministic sensitivity analysis to Assumption (e), in which we assume that there is a constant proportion of infectious donations 1970-1991, i.e. no ‘IDU’ component in the model. This is analogous to the analysis for the England look-back study by Soldan et al. (2002).⁽¹²¹⁾
• Scenario D: A deterministic sensitivity analysis to Assumption (c), in which we assume that the deferral effect occurred in 1986 rather than 1985.
• Scenario E: A deterministic sensitivity analysis to Assumption (d), in which we assume that the deferral effect led to a reduction of 33% in infectious donations – the lower effect supported by some of the expert respondents.
• Scenario F: A deterministic sensitivity analysis to Assumption (e), in which we assume that the constant level of infectious non-IDU donations is 50% rather than the 25% baseline assumption.
• Scenario G: A deterministic sensitivity analysis to Assumption (j), in which we assume there is no additional risk to the annual hazard of dying from having a transfusion.
• Scenario H: A deterministic sensitivity analysis to Assumption (j), in which we assume a constant rather than age-stratified risk to the annual hazard of dying from having a transfusion.
• Scenario I: A deterministic sensitivity analysis to Assumption (k), in which we assume that there is no increase in the annual risk of dying from HCV.

Table 4.16 Estimates from baseline deterministic model, baseline stochastic model with 95% confidence intervals and deterministic sensitivity analyses for England: results for other nations are shown in Appendix Tables 4.55 to 4.57.

Interpretation: In addition to the discussion of Scenarios A – C given after Table 4.1, there are some noteworthy differences between our deterministic scenarios, although most are minor compared to the uncertainty about the baseline estimates.

• Scenario E demonstrates that the major driver is the assumption of the step-down in infectious donations following the guidance in the mid-1980s: if we assume only a 33% step-down, the number of infections (and subsequent outcomes) reduces by around 28%.
• The final scenarios explore the impact of different assumptions about survivorship in the second and later decade of post-transfusion follow-up.
• Scenario G reverts to our baseline ever-IDU scenario but assumes that there is no additional transfusion hazard beyond 10-years: an estimated reduction of only 200 deaths. The second option (H) is a constant transfusion-related hazard (pooled across all age-groups), separately for 11-15 years (1.3) and lower for 16-20 years (1.1): a more modest estimated reduction of only 100 deaths.

Chapter 5: Information from funds

Summary findings

• The funds will not be a complete record of those infected, but provide a useful 'calibration' of results derived from alternative sources.

Since 1988, a variety of funds have provided financial support to those infected or affected by blood-borne infections: for HIV infections, the Macfarlane Trust and Eileen Trust, with MFET Ltd set up to administer lump sum payments; the Skipton Fund and the Caxton Foundation for those with chronic HCV infections transmitted prior to September 1991. After 2017, these funds were replaced by individual schemes for the four UK nations.

Reported data on successful applicants are summarised in Table 5.1.⁽¹²²⁾

Table 5.1 Summary of historical funds for those infected through blood or blood products.

Comments:

• Registrations to the Eileen Trust are likely to represent nothing more than a lower bound on the number of HIV infections acquired as a result of transfusions within the UK. In 1997, a letter from the Trust explained that under-registration may have occurred because: “some have chosen not to be introduced to the Trust and in some posthumous cases there was no dependant relative who would be eligible [to apply].”⁽¹³⁰⁾ In addition, registration to the Eileen Trust required at minimum clinical proof of HIV infection, inter alia.⁽¹³¹⁾ Given that the majority of transfusion-transmitted HIV infections occurred between 1979 and October 1985⁽¹³²⁾ establishing clinical proof of infection would inevitably have been difficult in some cases. Dr Patricia Hewitt has explained that clinical proof: “required… the Blood Service to say: yes, we know about this case, we have investigated it and it is a case of transfusion-transmitted infection […] It all seemed to me to be quite haphazard”.⁽¹³³⁾ It is possible a proportion of cases where place of transfusion was unknown or where Eileen Trust applications were rejected constitute further cases of HIV infection via transfusion in the UK.
• For the Skipton Fund, we do not have the breakdown with/without bleeding disorder for the cumulative 5,529 registrations when the fund closed in 2017. However, based on the data from 2015, we might assume roughly 50% in each group, i.e., around 2,765 each, with more than 33% deceased of those with bleeding disorders, and more than 24% in those without. There could be a deficit of applications related to those who died earlier.

Recent summary data for the current schemes in the four nations are shown in Table 5.2 (primary infections) and Table 5.3 (secondary infections), based on the ‘Rule 9’ submissions made to the Inquiry (up to July 2022).

Table 5.2. Summary of current UK funds for primary infections through blood or blood products, except WIBSS and NIIBPS, which are combined primary and secondary. The figures for the deceased are included in the totals for each scheme.

Table 5.3 Summary of current UK funds for secondary infections through blood or blood products (excluding WIBSS for which these data are not available and NIIBPS which has records of 2 secondary infections). The figures for the deceased are included in the totals for each scheme.

Comments:

• There was a substantial fall in registrations in the transition between the original funds and the national schemes.
• For consistency, the EIBSS figures do not include numbers of secondary infectees pre-01/11/2017. EIBSS report that before 2017 there were substantial numbers of secondary infectees: 39 with HCV infection (of whom, 2 deceased) and 41 with HIV infection (of whom, 6 deceased).⁽¹⁴⁰⁾
• There are 2,586 HCV-alone registrations in England. Given the relative populations of the four nations in 1991, then we would expect 276 in Scotland, 86 in Northern Ireland, and 155 in Wales. Northern Ireland and Wales match closely to that expected from the English data, but Scotland is substantially higher, possibly reflecting the greater proportion of HCV infections through transfusion than the roughly 50:50 recorded in the Skipton Fund.

Comparison with results from Chapter 4

Our baseline deterministic scenario for England in Chapter 4 provides age-sex bands for survivors from chronic HCV infection by transfusion. We represent here our estimates for survivors at 31 December 2019; and also for survivors at 31 December 2014. In both cases female survivors outnumber males by 1.77 to 1 (1,425/804; 1,614/910); and are older than their male counterparts.

Table 5.4 Estimates from our baseline deterministic scenario for England for the number, sex and age-distribution of survivors from chronic HCV infection by transfusion to 31 December 2014 (910 males; 1,614 females) and to 31 December 2019 (804 males; 1,425 females).

For comparison with the above age-sex distributions, we requested information about Skipton/EIBSS surviving chronically HCV infected claimants without HIV co-infection at 31 December 2014 from an analytical team at the Department of Health and Social Care who have familiarity with those data. We are grateful that this team was able to respond so promptly. First, we document the information provided on survival status at 31 December 2014 for male and female beneficiaries by exposure.

The sex distribution for DHSC/EIBSS surviving beneficiaries whose exposure was Other (mainly by transfusion, we assume) echoes our baseline scenario for survivors from chronic HCV infection by transfusion in that surviving female beneficiaries outnumber males by 1.5 to 1 (718/473); also for surviving people with bleeding disorders among whom male beneficiaries outnumber females by 7.6 to 1.

**Labelled as “haemophiliac”.

Table 5.5 From DHSC/EIBSS, survival status at 31 December 2014 for chronically HCV infected (mono-infected) beneficiaries – by sex and exposure. We have interpreted the DHSC/EIBSS label “haemophiliac” to mean people with a bleeding disorder.

Our goal in Table 5.6 is to compare our baseline scenario’s sex distribution for survivors at 31 December 2014 who were chronically HCV infected by transfusion in England and the corresponding sex distributions for DHSC/EIBSS surviving HCV infected beneficiaries whose exposure was ‘Other’ than a bleeding disorder.

Our baseline deterministic scenario expected around 2,500 survivors at 31 December 2014 who had been chronically HCV infected by transfusion in England. DHSC/EIBSS has around 1200 beneficiaries who were known to be alive at 31 December 2014. Our baseline scenario expected 64% of survivors to be female. This matches well with females representing 60% of DHSC/EIBSS’s beneficiaries whose exposure was other than as people with bleeding disorders (95% CI: 57% to 63%). The DHSC/EIBSS data constitute a similar proportion (around half) of the number of survivors estimated by our model.

We emphasise again the substantial stochastic uncertainty around our estimates from Chapter 4.

Table 5.6 Sex distribution from our baseline scenario for survivors to 31 December 2014 who had been chronically HCV infected by transfusion in England prior to September 1991; and for known-alive surviving DHSC/EIBSS beneficiaries who were not people with a bleeding disorder.

Those who were chronically-HCV-infected via blood products or by transfusion and who are alive at the end of 2014 can benefit from directly acting antiviral therapy. Males predominate among people with bleeding disorders who were chronically HCV-infected and among those known to be alive at the end of 2014. By contrast, females who were chronically HCV infected by transfusion predominate among those who have survived to the end of 2014. Further analysis is contained in Appendix A.53 to A.75.

Chapter 6: vCJD infections from blood and blood products

How many people were infected with vCJD from blood and blood products in the UK?

Summary findings

• Out of 178 recorded cases of confirmed or probable vCJD in the UK, 5 were blood-borne infections: 3 symptomatic vCJD cases; one asymptomatic person (positive at autopsy for abnormal prion in her spleen) from transfusion-transmitted infection; and one asymptomatic infectee with severe haemophilia A who had been exposed to vCJD-implicated Factor VIII (positive at autopsy). All five have died.

Confidence that available evidence can answer the questions? High.

Background

Bovine Spongiform Encephalopathy (BSE) was first officially diagnosed in November 1986, and variant Creutzfeldt-Jakob Disease (vCJD) was first defined in March 1996, based on 10 individuals in the UK.⁽¹⁴¹⁾ In his witness statement to the Inquiry,⁽¹⁴²⁾ Professor James Ironside described vCJD as a unique form of prion disease, from transmission of the BSE agent through food and affecting typically younger people than does sporadic CJD. Once symptomatic, it is a very severe disease with a median of 14 months of illness before death.

Since 1995 and as of October 2021, there have been 178 UK patients with definite or probable vCJD reported by the National CJD Surveillance Unit in Edinburgh (NCJDRSU)⁽¹⁴³⁾ – the last reported case was in 2016. All have died, 55/178 without neuropathological confirmation because consent for autopsy was lacking. The great majority of vCJD cases worldwide has occurred in the UK: France has recorded 28 cases, peaking five years later than in the UK, with 26 others elsewhere in the world than UK or France.

Possible transmission to people with bleeding disorders

UKHCDO report details an exercise in 2004 whereby all people with bleeding disorders in the UK who were alive and had received UK pooled plasma products between 1980 and 2001 were identified as “at-risk of vCJD for public health purposes”.⁽¹⁴⁴⁾ UKHCDO provided summary statistics on 5,147 people with bleeding disorders who were potentially considered at-risk, of whom 785 had received an implicated batch.⁽¹⁴⁵⁾ As we note below (Case 5), one person with haemophilia is thought to have been infected with vCJD through a pooled plasma product.

Possible transmission through blood transfusions

As yet, there is no screening test which can protect the UK’s blood supply against transmission of vCJD. There was, however, early recognition that the disease might be transmitted through blood, leading to the establishment of the Transfusion Medicine Epidemiology Review (TMER). Hence, in 1997, a study was set up to identify recipients of blood that was subsequently found to have been donated by people who developed vCJD. The TMER study reports^{(146) (147)} that 31 vCJD cases were reported to be blood donors, and four more were registered with UKBTS, but only one had given blood. Components from 18 of these 32 vCJD donors were issued to hospitals, and given to 67 identified recipients; six further components are known to have been issued but could not be traced. The current reported status⁽¹⁴⁸⁾ of these 67 individuals is shown in Table 6.1.

Table 6.1 Recorded outcomes of 67 recipients of blood donated by confirmed cases of vCJD. ‘PrPSc deposition’ indicates the abnormal prion protein.

Of the 14 who remain alive, Urwin et al.⁽¹⁴⁹⁾ report that ‘One recipient has moved abroad, and their fate is currently unknown while the remaining 13 have now survived more than 10 years after receiving transfusion from vCJD donors. There have been no new cases of vCJD identified by the NCJDRSU among the recipients of blood from vCJD donors.’

Available evidence on individual cases

The NCJDRSU 2020 Annual Report⁽¹⁵⁰⁾ discusses 4 cases of clinical vCJD following transfusion, while the NHS vCJD website⁽¹⁵¹⁾ reports 5 cases. Professor Ironside confirmed five cases of vCJD via blood/blood products, with four cases caused via blood transfusion (3 clinical vCJD and 1 asymptomatic), and the fifth, a person with haemophilia who was infected after receiving vCJD-contaminated Factor VIII.⁽¹⁵²⁾ Brief details are as follows:

Case 1. Details of the first case of vCJD via blood transfusion were announced in Parliament in December 2003,⁽¹⁵³⁾ and later published in the Lancet.⁽¹⁵⁴⁾ The 24-year-old donor had no signs of vCJD when they gave blood in 1996, although the donor later developed symptoms and died in 2000. A 62-year-old recipient subsequently developed vCJD symptoms and died in 2003, over 6 years after transfusion, and was linked to the donor by the TMER.

Case 2. A second case of vCJD via blood transfusion was found in Glasgow and reported in the Lancet,⁽¹⁵⁵⁾ having been identified by the TMER study. The elderly patient had received blood in 1999 from a donor who died of vCJD in 2001, and subsequently died from a non-neurological disorder in 2003, with vCJD being confirmed at autopsy.⁽¹⁵⁶⁾

Case 3. This young man died in May 2006 after receiving a blood transfusion in September 1997 and developing vCJD 8 years after transfusion – details are contained in a moving witness statement by his father,⁽¹⁵⁷⁾ and published in the Lancet.⁽¹⁵⁸⁾ Before his diagnosis he had been identified as being at-risk.

Case 4.⁽¹⁵⁹⁾ This 65-year-old man received blood from the same donor as case 3 in late 1997,⁽¹⁶⁰⁾ and subsequently developed vCJD symptoms and died in March 2007. He was identified as a potential case by the TMER study, and had been in touch with the National Prion Clinic before developing symptoms.

Case 5. This was a 73 year old male with severe haemophilia who had been treated with vCJD-implicated Factor VIII, and who died in September 2009⁽¹⁶¹⁾ with no history of neurological disease. An autopsy was performed under Her Majesty's Coroner's Instructions. The deceased was positive for the abnormal prion in his spleen and was heterozygous at codon 129 of the prion protein. The most likely route of vCJD infection was through a UK plasma product. Professor Ironside notes this was ‘not proven conclusively on independent review by the DH Health Protection Analytical Team’.⁽¹⁶²⁾

In addition to TMER, and because there is no screening test in blood, the UK has conducted three studies in appendix tissue to assess the prevalence of abnormal prion protein (referred to as PrPSC) by birth-cohort and sex.

Steps taken to protect the UK blood supply included leucodepletion and, from 2004, those who received blood or tissue were asked not to donate. In addition, anyone who has donated to a person who subsequently developed vCJD is asked to abstain from blood and tissue donation. See Professor Ironside’s witness statement for a fuller account.⁽¹⁶³⁾

Chapter 7: HBV infections from blood and blood products

How many people were infected with Hepatitis B Virus (HBV) between 1970 and 1991 through blood and blood products, and what were the impacts?

Summary findings

Due to the limitations in the data available, it is not possible to answer the questions set with any reasonable accuracy when compared to other infections we investigated. There was a‘lack of an integrated approach at the onset of donor screening in 1971/72’⁽¹⁶⁴⁾ to identify donors who were infectious HBV carriers. Furthermore, people infected with HBV have never received financial support, and so funds are not a source of data.

Confidence that available evidence can answer the questions? Low.

Terminology⁽¹⁶⁵⁾

HBsAg (Hepatitis B surface antigen, formerly the Australia antigen/HAA). A “positive” or “reactive” HBsAg test result means that the person is infected with hepatitis B virus. This test detects the actual presence of the hepatitis B virus (the “surface antigen”) in blood. If a person tests positive, further testing is needed to determine if this is a new “acute” infection or “chronic” hepatitis B (HBV carrier). A positive HBsAg test result means that you are infectious and can spread HBV to others through blood and other secretions.

Anti-HBs or HBsAb (Hepatitis B surface antibody) – A “positive” or “reactive” anti-HBs (or HBsAb) test result indicates that a person has antibodies and is protected against HBV. This protection can be from receiving the hepatitis B vaccine or successfully recovering from a past HBV infection. You are not infectious and cannot spread hepatitis B to others. This test is not routinely included in blood screening.

Results of screening donors

Screening for Hepatitis B surface antigen (HBsAg) was routine throughout the UK by December 1972. Before screening, some high rates of HBsAg had been reported; for example, the MRC Working Party on post-transfusion hepatitis⁽¹⁶⁶⁾ described 782 patients who were tested after transfusion of whole blood during 1969-1971, of whom eight tested positive for HBsAg, an incidence rate of 1%. The study population in this report was urban with a high proportion of people with close family links to regions with high incidence of HBV infection during childhood (which increases the risk of developing chronic carriage) – which suggests that 1% may be an upper limit for transfusion risk pre-screening.

Risks remained post-1970, although these are difficult to quantify. It was claimed in 1976⁽¹⁶⁷⁾ that the available tests for HBsAg would detect no more than about 50% of HBV carriers, and so there would have been ‘breakthrough’ HBV infections. Therefore, detected HBsAg infections in donors could give an idea of the likely number of breakthrough HBV infections.

The National Blood Transfusion Service Register of HBsAg Positive Donations⁽¹⁶⁸⁾ provides data for 1971-1991 from all UK Regional Transfusion Centres (RTCs), including total number of donations, total confirmed HBsAg positive, and overall positivity rate in both repeat and new donors. Figure 7.1 shows the HBsAg positive count per 100,000 donations in the UK (not all centres contributed to all years).

Figure 7.1 HBsAg positives per 100,000 donations in the UK. Centres started testing and reporting data in different years, and so not all centres contribute to each year’s results.⁽¹⁶⁹⁾

For example, just after screening began, 9 out of 22 UK RTCs returned results in 1972, and reported 45 HBsAg positive per 100,000 donations. Extrapolating to approximately 1.5m donations suggests that perhaps 800 HBV infectious donations were screened out. Maycock⁽¹⁷⁰⁾ estimated that exclusion of antigen-positive donors in 1970 would have prevented some 575 cases of icteric or anicteric hepatitis.

Between 1983 and 1986, the HBsAg count per 100,000 donations halved, at a time of enhanced self-deferral of donors to reduce HIV-related risks and the introduction of HIV antibody screening of blood donors.

Few HBV cases following transfusion have been reported since 1991. Soldan et al. 2002⁽¹⁷¹⁾ list 14 cases in the UK between 1991-1997, with 11 being the result of chronic infection in donors. In contrast, of the 6 cases listed between 1998 and 2001, all donors had acute rather than chronic infections. These counts understate the true incidence as there will be cases that have been missed. Other fragmentary pieces of evidence include a report of two cases of post-transfusion hepatitis B reported during 1988 from Trent RTC,⁽¹⁷²⁾ and a Public Health Laboratory Service report⁽¹⁷³⁾ that out of 1,061 cases of acute hepatitis B reported during July 1975 to June 1976, 4% (~40) had received blood transfusions.

Risk to people with bleeding disorders

Very limited data are available. In their report to the Inquiry,⁽¹⁷⁴⁾ UKHCDO say “HBV test results have not been collected from centres since vaccination was introduced early in 1980. Jaundice surveys were carried out intermittently over the years, either at the centre level or individual people with bleeding disorders level, but with no consistency. These were not submitted to the NHD.”

Craske et al. (1978)⁽¹⁷⁵⁾ discuss 371 people with haemophilia who were treated with commercial Factor VIII over 1974-1975, in whom 30 cases of HBV were found. Of the original concentrates, 4 of 6 batches were positive for HBsAg. Tables⁽¹⁷⁶⁾ show that of 2,308 people with haemophilia given FVIII and IX in 1977, 18 cases of HBV were recorded.

Comments

The majority of infections are brought into the UK with migration,⁽¹⁷⁷⁾ and the multiplicity of transmission routes makes it almost impossible to assign a cause for individual cases. Where data are available to make infected blood the most likely route, the numbers are small, largely because of the introduction of donor screening soon after the virus was discovered.

If we make assumptions about the sensitivity of the screening test, it might be possible to use the data above to estimate the number of breakthrough infections. Allowing for 90-95% of recipients of infected donations to clear the virus naturally, the number of chronically infected recipients might be estimated in an exercise similar to that of Chapter 4. But, given the quality of the information, the numbers would be very approximate, and in any case would not represent a large number compared to the other infections of interest. We have therefore not conducted a detailed modelling exercise.

Combatting viral hepatitis is part of the United Nations Sustainable Development Goals. As described by Mandal (2019), the UK took a big step forward in August 2017 in its fight against viral hepatitis – specifically hepatitis B – with the incorporation of hexavalent combination vaccine (including against HBV) into the routine childhood immunization programme. This vaccine is now offered to all infants at 8, 12 and 16 weeks, replacing the pentavalent vaccine which protected against diphtheria, tetanus, pertussis, polio and haemophilus influenzae type B. In May 2022, the World Health Organization announced new targets that aimed to eliminate viral hepatitis by 2030 by driving new infections and deaths down to half a million for both hepatitis B and hepatitis C, as well as reducing HBsAG in children under 5 years to below 0.1%.⁽¹⁷⁸⁾

Acknowledgements

As was made clear in the Introduction, this report has been prepared rapidly and under tight deadlines, and would not have been possible without the generous assistance of numerous individuals and organisations. We would particularly like to highlight:

• The Infected Blood Inquiry team – including Moore Flannery, Tom Dunning, Annabel Sharma, Connor Mitton, Michelle Secker, and Elli Brooks – for their extensive advice and support in producing this report, particularly in commissioning three types of linkage study (as advised by Sheila Bird), helping access data from numerous agencies, and in detecting a mass of relevant documents in Relativity.
• UKHSA for providing updated data on HIV diagnoses related to blood and blood products (Chapters 1 and 3).
• The Manchester statistical team for providing a slimmed no-names version of the National Haemophilia Database and for preliminary analyses (Chapter 2).

Mr Ben Palmer, Statistician, National Haemophilia Database

Dr Hua Xiang, Data Analytics Manager, National Haemophilia Database

Dr Matt Gittins, Lecturer in the Centre for Biostatistics, School of Health Sciences, Manchester University

Professor Andy Vail, Director of the Centre for Biostatistics, School of Health Sciences, Manchester University

• The team at the Scottish National Blood Transfusion Service and Edinburgh University for conducting the record-linkage follow-up to 31 December 2019 of four cohorts of transfusion recipients in 1999; 2004; 2019 and 2014 (Chapter 4).

Professor Marc Turner, Director, Scottish National Blood Transfusion Service

Dr Katherine J Forrester, Transfusion Researcher, Scottish National Blood Transfusion Service

Amanda Stewart, Head of Analytics and Planning, Scottish National Blood Transfusion Service

Dr Vanda Inacio, Lecturer in Statistics, School of Mathematics, Edinburgh University

• Colleagues on other Infected Blood Inquiry expert groups who advised on the possible effect of donor guidance in the mid-1980s (Chapter 4).
• The team at UKHSA for unpublished annual data on estimated number [1971-1991] of chronically HCV infected a) ever-IDUs; b) past-IDUs; c) current-IDUs (Chapter 4).

Dr Ross Harris, Senior Statistician, Statistics Modelling and Economics Department, UK Health Security Agency

Professor Daniela De Angelis, Deputy Director, MRC Biostatistics Unit, Cambridge University.

• The team at UKHSA for updated analysis of follow-up on their HCV look-back case control cohort study (Chapter 4).

Dr Ross Harris, Senior Statistician, Statistics Modelling and Economics Department, UK Health Security Agency

Dr. Helen Harris, Clinical Scientist – Epidemiology, UKHSA

Annastella Costella, Hepatitis Scientist, UKHSA

Dr. Sema Mandal, Medical Consultant Epidemiologist, UKHSA

Dr. Monica Desai, Medical Consultant Epidemiologist, UKHSA

Dr. Mary Ramsay, Head of Immunisation Department, UKHSA

also the HCV National Register Steering Group.

• The team from Wallis et al. (2014) for providing their original data (Chapter 4).
• The team at NHSBT for providing detailed survival analyses of EASTR data (Chapter 4).
• The Department of Health and Social Care/EIBSS analytical team who provided demographic and survival information about EIBSS claimants (Chapter 5).

Andrew Parker, Principal Operational Research Analyst, Department of Health and Social Care (DHSC)

Joshua Walden, Senior Economic Adviser, DHSC

Omar Idriss, Deputy Director, DHSC

The analyses for this report were carried out by Ruth McCabe and Dr Sarah Hayes under the immediate direction of Sheila Bird and David Spiegelhalter. We are deeply indebted to Ruth and Sarah for conducting all the modelling for Chapter 4, producing all the graphs and tables, and responding to continual requests for more analyses with good humour and patience.

Stephen Evans convened the Statistics Expert Group; Sheila Bird, Ruth McCabe, Sarah Hayes and David Spiegelhalter were responsible for the initial draft of this report; and then all members of the Statistical Expert Group contributed to critique and final editing.

Authors

Non-SEG Members

Sarah Hayes

Sarah Hayes is a post-doctoral research assistant in the Department of Statistics at the University of Oxford, funded by a National Institute for Health and Care Research grant as part of the Health Protection Research Unit (HPRU) in Emerging and Zoonotic Infections. Her research focuses on the epidemiology of emerging and zoonotic infections. Sarah holds a PhD in Infectious Disease Epidemiology and a Masters in Epidemiology, both from Imperial College London, and a Bachelor in Veterinary Medicine from the Royal Veterinary College.

Ruth McCabe

Ruth McCabe is a doctoral student in the Department of Statistics at the University of Oxford funded by a National Institute for Health and Care Research studentship funded as part of the Health Protection Research Unit (HPRU) in Emerging and Zoonotic Infections. Her research focuses on the modelling and analysis of high-consequence infectious diseases, in particular COVID-19 which she has been working on since early 2020; the application of modelling techniques to inform policy; and methods by which to overcome the issues of missing or non-representative epidemiological data. Ruth has an undergraduate degree in Mathematics, Operational Research, Statistics and Economics from the University of Warwick and a Masters in Statistics from Imperial College London.

SEG Members

Stephen Evans (Convenor)

Stephen Evans is Professor of Pharmacoepidemiology at the London School of Hygiene and Tropical Medicine. He works in the field of safety of medicines and is also interested in detecting scientific fraud and misconduct. He has been an independent expert for the European Medicines Agency Drug Safety Committee and was an advisor to the Bristol Royal Infirmary Inquiry.

Sheila Bird OBE

Sheila Bird is Honorary Professor at the University of Edinburgh and former programme leader at the Medical Research Council (MRC) Biostatistics Unit, Cambridge. She led the MRC Biostatistical Initiative in support of AIDS/HIV studies in Scotland. She has championed the use of statistical methods in evaluating public policies. She has worked on UK dietary exposure to BSE, and on the design and interpretation of surveillance studies for abnormal prion protein; also on the late sequelae of Hepatitis C virus infection.

Penny Chan

Penny Chan was scientific co-ordinator of the Krever Inquiry into the use of infected blood and blood products in Canada. She was subsequently the executive coordinator for the Canadian Advisory Council on Blood Safety and has spent 20 years as a consultant with the World Health Organisation.

Daniela De Angelis

Daniela De Angelis is Professor of Statistical Science for Health at the University of Cambridge and deputy director of the Medical Research Council Biostatistics Unit. Her research develops and applies statistical methods to characterise epidemics, exploiting the complex body of available information on different aspects of the disease of interest. Her work focuses on HIV, Hepatitis C and respiratory viruses.

Christl Donnelly CBE

Christl Donnelly is Professor of Applied Statistics at the University of Oxford and Professor of Statistical Epidemiology at Imperial College London. Her research brings together and develops statistical and biomathematical methods to analyse epidemiological patterns of infectious diseases, including vCJD, HIV and hepatitis A. She is also interested in promoting the public understanding of epidemiology and statistics.

Nicholas P. Jewell

Nicholas P. Jewell is Professor of Biostatistics and Epidemiology at The London School of Hygiene and Tropical Medicine. He has until 2018 been at The University of California in Berkeley. He has worked on statistical issues associated with studies of the natural history of infectious diseases including Human Immunodeficiency Virus (HIV) disease, dengue fever, Ebola Virus Disease, Severe Acute Respiratory Syndrome (SARS), and H1N1 influenza.

Graham Medley OBE

Graham Medley is Professor of Infectious Disease Modelling and former Director of the Centre for Mathematical Modelling of Infectious Disease at the London School of Hygiene and Tropical Medicine. His research uses mathematical models of transmission to inform public health interventions. He has advised the UK Government on HIV, vCJD and pandemic preparedness, and has chaired the SPI-M-O modelling sub-group of SAGE during the COVID-19 pandemic.

Sir David Spiegelhalter OBE

Sir David Spiegelhalter is Chair of the Winton Centre for Risk and Evidence Communication at the University of Cambridge, which aims to ensure that quantitative evidence and risk is presented to people in a fair and balanced way. He contributed to Public Inquiries into both children’s heart surgery at the Bristol Royal Infirmary, and the murders by Harold Shipman.

Verifying Statements

The standard verifying statement is as follows:

All contributing group members confirm that in respect of those parts of the report to which they have contributed:

1. They have made clear which facts and matters referred to in this report are within their knowledge and which are not.
2. Those that are within their knowledge they confirm to be true.
3. The opinions they have expressed represent their true and complete professional opinions on the matters to which they refer.

Sections (i) and (ii) are somewhat unsuitable to statistical analyses, since we rely on data provided from a wide variety of sources, and so cannot say that the information we present is ‘within our knowledge’ or that we can confirm it to ‘be true’.

However, we can confirm that we have exercised our best professional judgement as to the quality and interpretation of the evidence that was available.

Appendix

This document contains additional material related to the Expert Report. Minimal background is provided for the tables, graphs and technical explanations – please refer to the relevant Chapters and Sections in the main report for further explanation, context and references.

Chapter 1

Appendix Table 1.1 HIV Diagnoses of people with bleeding disorders in the UK by year (Source UKHCDO data).

Appendix Table 1.2 New HIV diagnoses in the UK up to 2000, overall and through exposure to infected blood products by year of diagnosis. The numbers reflect the year the diagnoses were made which is not necessarily the same as the year the infections were acquired.

Chapter 2

Appendix Table 2.1 Bleeding disorder and its severity for females by HCV status, as described in Table 2.2, together with survivorship to the end of 1991, 1999, 2009, 2013 and 2019. (Analogous to Table 2.3 in main report).

Chapter 4

Adapting the model to Scotland, Wales and Northern Ireland

The model structure is applicable to each country within the UK, with the only necessary changes coming from specific data inputs. Appendix Table 4.1 presents an overview of the inputs and sources for each of the four nations’ baseline models.

As outlined in Appendix Table 4.1, six inputs were identified as requiring setting-specific data. This was possible with the following exceptions:

• 1991 HCV infected donor prevalence in Northern Ireland.
• Number of HCV infectious ever-IDUs in Northern Ireland and Wales.
• Year of deferral reduction in Northern Ireland and Wales.
• Effect of deferral reduction in Northern Ireland and Wales.
• The number of blood donations in Northern Ireland.

For all missing data, with the exception of the number of blood donations, the values were assumed to be as observed in England. The number of blood donations in Northern Ireland were estimated by assuming population pro-rata of Northern Ireland in comparison to England, as detailed below.

All other parameters were assumed to be setting-independent and could thus be applied to each of the four nations.

Appendix Table 4.1 Model inputs for each of the four nations of the UK.

Chapter 4. Task 1.3 How many chronically HCV-infected ever-injecting drug users (ever-IDUs) were alive during each year between 1970 and 1991?

Ever-IDUs in Scotland compared to England

Appendix Figure 4.1 shows the number of HCV-infectious ever-IDUs in Scotland compared to England, highlighting a greater number per capita in Scotland. While England observes a steady increase in the proportion of HCV-infectious ever-IDUs in comparison to its 1991 value over time, Scotland observes a sharp increase across the 1980s, with extremely low levels observed in the 1970s.

Appendix Figure 4.1 (A) The number of HCV-infectious ever-IDUs in England and Scotland. (B) As in (A) but scaled to per 100,000 population. (C) The trend in HCV-infectious ever-IDUs relative to the value observed in 1991.

Chapter 4. Task 1.4. What was the step-down in donations from ever-IDUs due to self-deferral guidance and HIV antibody testing in the mid-1980s?

A key uncertainty in forward projection of the numbers of people infected with HCV in England by blood transfusion during 1970-1991, is how effective self-deferral information-leaflets for blood donors in the HIV/AIDS era, together with the advent of HIV antibody testing of the blood supply, were in dissuading donations from past and current IDUs who (unknowingly) were HCV-infected.

At short notice, we sought expert opinion from selected members who serve on other expert groups for the Infected Blood Inquiry and were in professional practice in the 1980s. This was relatively successful: only one non-response (on account of annual leave) out of eight, and six provided some quantification.

Six assessments are summarised in Appendix Table 4.2 in terms of the % of HCV-infected ever-IDUs who were dissuaded from donation by self-deferral leaflets, HIV antibody testing or other awareness about the risk of blood-borne infection. Our informants had been given the following instructions:

Please place 20 betting tokens in the betting streets below to express your opinion about the likely percentage-reduction in HCV-infected donations from early in 1985 to 1986.

Appendix Table 4.2 Elicitation on % reduction in HCV-infected donations by ever-injectors across England in 1986 (versus 1985 as baseline) that was achieved by self-deferral advice & HIV antibody-testing.

** Seventh assessor judged that, on account of venous access, the number of current injectors who would have attended as blood donors would be very low – even before HIV antibody testing was introduced. There would have been a further 50% (with 50% confidence).

The resultant distribution is bimodal. Ignoring outlying votes which sum to fewer than 3 [9 of them], the lower mode has accrued 36/110 votes. These 36 votes support a mean reduction of 31% (1130/36; most likely 30%). The upper mode accrued 74/110 votes. These 74 votes support a mean % reduction of 67% (4960/74; most likely 70%). Consistent with the elicited bimodal distribution, our baseline scenario in our simulation model is a 67% reduction, with a sensitivity analysis of a one-third reduction for England; as well as no reduction to mimic the English HCV lookback model assumptions.

In addition, a seventh assessor judged that, on account of venous access, the number of current IDUs who would have attended as blood donors would be very low – even before HIV antibody testing was introduced. There would have been a further 50% (with 50% confidence).

Other contexts that respondents mentioned included:

• greater impact on HCV-infected current IDUs;
• potentially greater impact from HIV antibody testing than from self-deferral leaflets;
• literature on limited impact on public health behaviours via leaflets;
• reluctance of blood donors to be turned down and hence minimization or oversight of historical risk-factors;
• attendance as blood donor specifically to access HIV antibody testing;
• whether potential donors appreciated that past history of having ever-injected mattered (vs being a current injector);
• sensitivity of the HIV antibody test used for initial screening of blood supply;
• how risks to blood supply from non-A, non-B hepatitis infections in ever-IDUs were appreciated in contemporary medical reports, let alone by potential donors;
• in the early 1990s, the proportion of blood donors who, having tested HCV antibody positive, admitted a history of injecting or sexual risk being less than 40%;
• much health education literature requires too high a level of literacy and that written in the 1980s often appeared hectoring.

Past-IDUs

Six respondents offer an opinion on the % of HCV-infected past-IDUs who were dissuaded from donation by self-deferral leaflets, HIV antibody testing or other awareness about the risk of blood-borne infection. See Appendix Table 4.3.

Appendix Table 4.3 Elicitation on % reduction in HCV-infected donations by past-IDUs across England in 1986 (versus 1985 as baseline) that was achieved by self-deferral advice & HIV antibody-testing? ** Until 1987, leaflets asked drug addicts or drug abusers to refrain from donating: they did not specifically encompass past users. Someone who used intravenous drugs occasionally, decades previously might not consider themselves a “drug abuser”. After 1987, leaflets are specific that this applies to an ever-user in the previous decade. Hence, 70 to 80% decrease amongst past users in response (certainty level: 70-80%).

The resultant distribution is again bimodal. Ignoring outlying votes which sum to fewer than 3 [six of them], the lower mode has accrued 48/113 votes. These 48 votes support a mean reduction of 24% (1160/48; most likely 20%). The upper mode accrued 65/113 votes. These 65 votes support a mean reduction of 67%% (4330/65; most likely 65%). Consistent with the elicited bimodal distribution, sensitivity analysis for the simulation model has used 67% and 20% reductions for England’s chronically infected past-IDUs.

Chapter 4. Task 2.1 How many blood donations were there annually between 1970 and 1991?

England and Wales

Data on the number of donations per year in England and Wales were observed for 1975 - 1976 and 1978 - 1990. To estimate the number of donations in years without data, we used Poisson regression as in Schnier and Goldberg (S&G).

There was a distinct difference in the number of donations from 1978 compared to that observed in 1975 - 1976. Similar trends were observed in Scotland during this period.⁽¹⁷⁹⁾ Consequently, we used two separate Poisson regression models (donations as response and year as predictor) in order to accurately capture this difference in trends (Appendix Figure 4.2).

First, we produced a model fitted to observed data from 1975 - 1976 and 1978 - 1979 as shown in Appendix Figure 4.2A, and used this to estimate the donations in 1970 - 1974 and 1977. Secondly, we fitted using the observed data from 1978 - 1990 as shown in Appendix Figure 4.2B and used this to estimate the number of donations in 1991.

The number of donations attributable to England and to Wales separately was obtained by scaling the value in accordance with the respective populations of each country. This makes the assumption that the rate of donations in the population did not differ between England and Wales. Across 1970 - 1991, on average England accounted for 94.3% of the population of England and Wales, and thus 94.3% of donations were attributed to England, with the remaining 5.7% attributed to Wales.⁽¹⁸⁰⁾

Appendix Figure 4.2 Estimating the number of donations in England and Wales in years without data using Poisson regression. (A) Missing years are interpolated using a model fit to 1975 - 1976 and 1978 - 1979. (B) Missing year is interpolated using a model fit to 1978 - 1990.

Northern Ireland

We were unable to obtain specific information on the number of donations in Northern Ireland. Therefore, analogous to the case of England and Wales, we assumed that the number of donations in Northern Ireland could be approximated by assuming that this is proportional to Northern Ireland’s population in comparison to England. Across 1970 - 1991, on average Northern Ireland’s population was 3.3% that of England’s and so the number of donations in Northern Ireland was assumed to be 3.3% the number of those observed in England.⁽¹⁸¹⁾

Scotland

In their analysis, S&G published the number of donations in Scotland between 1970 - 1991 and thus we used these values.

Appendix Table 4.4 presents the estimated number of donations in each of the four nations.

Appendix Table 4.4 The number of blood donations in each nation of the UK across 1970 - 1991.

Chapter 4 Task 2.2. Per donation, on average how many units were transfused?

Data on the number of units transfused per donation were obtained annually for 1982 - 1988 from the National Blood Transfusion Service Statistics (NBTSS) for England and Wales⁽¹⁸²⁾ as shown in Appendix Table 4.5.

We sought to extrapolate the trend observed during this period to 1970 - 1981 and to 1989 - 1991 using a linear model (units transfused per donation as response and years as predictor). However, there is a steep increase observed in the number of units transfused per donation between 1982 - 1988 which is unlikely to hold throughout the entire period 1970 - 1991. Therefore, we scaled the gradient down to 25% of its estimate values for the periods without data and took the values expected under this model (Figure 4.7 in report).

We assume that this is able to be applied to all four nations within the UK. Appendix Table 4.6 shows the resulting estimates of units transfused in each nation.

Appendix Table 4.5 Data from the National Blood Transfusion Service Statistics (NBTSS) from which the number of units transfused per donation was approximated.

*** indicates estimates from extrapolation.

Appendix Table 4.6 The estimated number of units transfused in England, Northern Ireland, Scotland and Wales.

Chapter 4. Task 3. For each year 1970-1991 in England, how many transfusion-recipients (in age-sex bands) were infected with HCV?

Appendix Tables 4.7 - 4.9 present the estimated number of individuals infected with HCV by transfusion in Northern Ireland, Scotland and Wales, respectively.

Appendix Table 4.7 Estimation of the annual number of individuals infected with HCV by transfusion in Northern Ireland, 1970-1991 (totals are the rounded sums of the unrounded values).

Appendix Table 4.8 Estimation of the annual number of individuals infected with HCV by transfusion in Scotland, 1970-1991 (totals are the rounded sums of the unrounded values).

Appendix Table 4.9 Estimation of the annual number of individuals infected with HCV by transfusion in Wales, 1970-1991 (totals are the rounded sums of the unrounded values).

Chapter 4. Task 3. Age-sex distribution of those infected by transfusions

We used the age-sex distribution as observed by Wallis et al. (2004) in the north of England in 1994.⁽¹⁸³⁾ We were able to compare this to the age-sex distribution of transfusion recipients within the SNBTS record-linkage study in 1999. Age groups 0 and 1 - 9 from SNBTS were pooled to be 0 - 9 and age groups 80 - 89 and 90+ from Wallis et al. were pooled to be 80+ in order to allow for direct comparison. This showed close agreement of the age-sex distribution between these two independent sources as shown in Figure 4.8 in the main report, giving confidence to the application of this across different settings.

Appendix Tables 4.10 - 4.12 present the age-sex distribution of those estimated to be infected in Northern Ireland, Scotland and Wales, respectively.

Appendix Table 4.10 Estimated age-sex distribution of those infected with HCV through transfusion in Northern Ireland, 1970-1991.

Appendix Table 4.11 Estimated age-sex distribution of those infected with HCV through transfusion in Scotland, 1970-1991.

Appendix Table 4.12 Estimated age-sex distribution of those infected with HCV through transfusion in Wales, 1970-1991.

Chapter 4. Task 4. How many chronic HCV-infected transfusion recipients survived 10 years post-transfusion?

Extending survival from 5 to 10 years post-transfusion

The probability of surviving to 1, 5 and 7 years post-transfusion, independent of age-group, was broadly similar within the Wallis and EASTR cohorts, as shown in Appendix Tables 4.14 and 4.15. In particular, similar values were observed for Wallis compared to recipients of RBC in the EASTR study. We also compared the aforementioned survival probabilities across the Wallis cohort and EASTR RBC cohort using the EASTR age-groups in both settings. Again, there was broad agreement between the two sources, with the age categories 40 - 59 and 60 - 74 appearing to have the greatest differences.

Appendix Figure 4.3 presents the resulting probabilities of surviving 5 and 10 years post-transfusion, respectively.

Comparison of Wallis et al. (2004) and EASTR

Appendix Table 4.14 Percentage of transfusion recipients surviving to 1, 5, and 7 years post-transfusion using data from the Wallis study and different blood products from the EASTR study without age stratification.

Appendix Table 4.15 Percentage of transfusion recipients surviving to 1, 5, and 7 years post-transfusion using data from the Wallis study and the EASTR study (RBC only) using the age groups (in years) in EASTR.

Appendix Figure 4.3 The probability of surviving to 5 and 10 years post-transfusion by age-sex group.

Appendix Tables 4.16 - 4.18 present the estimated number of people with chronic HCV infection surviving 10 years post-transfusion in Northern Ireland, Scotland and Wales, respectively.