The Age of BLood Evaluation (ABLE) randomised controlled trial: description of the UK-funded arm of the international trial, the UK cost-utility analysis and secondary analyses exploring factors associated with health-related quality of life and health-care costs during the 12-month follow-up

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/144/51. The contractual start date was in September 2011. The draft report began editorial review in September 2016 and was accepted for publication in April 2017. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Julia Boyd reports grants from the University of Edinburgh during the conduct of the study.

Copyright statement

© Queen’s Printer and Controller of HMSO 2017. This work was produced by Walsh et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2017 Queen’s Printer and Controller of HMSO

Background

Red blood cell (RBC) transfusion is a widely practised intervention for critically ill patients. Approximately 85 million RBC units are transfused annually worldwide, and just under 2 million in the UK. In severe cases of anaemia or bleeding, reversal of dangerously low concentrations of haemoglobin can only be achieved by RBC transfusion. However, the practice has recognised risks, including transfusion-associated lung injury, transfusion-associated circulatory overload, bacterial contamination, infection transmission and allergic reactions, although all of these are rare. 1 Transfusion errors resulting in transfusion reactions are also an important cause of morbidity and mortality.

Meta-analysis of randomised controlled trials (RCTs) has demonstrated that ‘restrictive’ transfusion practice (using a lower haemoglobin concentration as a transfusion trigger, typically 70–80 g/l) appears to be safe and leads to a reduction in overall RBC usage. 2–4 The adoption of these restrictive haemoglobin transfusion triggers has resulted in a substantial reduction in RBC use, especially for surgery, and forms a key component of multimodal patient blood-management systems.

Donated RBCs can be stored for up to 35 days in the UK (and up to 42 days in some countries). Historically, the shelf life of RBCs was based on biochemical standards and RBC survival studies conducted in healthy volunteers. Despite the daily use of RBCs in hospitals worldwide, prior to the Age of BLood Evaluation (ABLE) trial (and other trials conducted over a similar time period) there was no high-quality clinical research to determine whether or not older stored RBCs deliver oxygen to tissues as effectively as fresher RBCs. Current standards for approval of RBC products are based on characteristics of the product, especially RBC survival in vivo at 24 hours, but not the ability of cells to transport oxygen to tissues. Prior to the ABLE trial, an accumulating body of laboratory and clinical research had raised the possibility that stored RBCs may be ineffective or even have harmful effects on patients. 5–8 None of this evidence was conclusive, but the signals seen in some uncontrolled observational clinical studies, together with the widespread use of RBC transfusions, meant that this research question was of vital importance to ensure that RBCs are used safely and effectively in the future. The implications of differences in effectiveness and safety of older versus fresher blood also have major practical and financial implications for blood services.

Transfusion laboratories actively manage RBC supplies to minimise wastage, because RBCs are a precious, limited and increasingly costly resource. As a result, RBCs stored for longer periods are frequently supplied to users, and older RBCs tend to be used first. RBCs are used for a wide range of patient groups as part of the management of a wide range of diseases and their complications. Importantly, indications for RBC transfusion range from life-threatening haemorrhage (e.g. trauma, gastrointestinal bleeding, postpartum haemorrhage) to the management of anaemia in otherwise stable patients (e.g. chronic inflammatory conditions, cancer and postoperative anaemia). Remarkably, prior to the ABLE trial, there was no high-quality evidence to reassure clinicians, blood services and patients that RBCs of all licensed storage ages are equally effective for any of these patient groups or clinical indications.

A systematic review undertaken in 2006 found only two small RCTs in adults relating to RBC transfusion. 8–10 Both trials were undertaken by members of the ABLE Canada or ABLE UK study groups. One was a study with physiological end points and the other was the feasibility study for the full ABLE trial. Neither provided conclusive information about the effectiveness and safety of stored blood in comparison with a fresher product (in total, only 89 patients were included).

Why were critically ill patients a suitable population in whom to study this question?

There was a strong rationale for undertaking research to assess the clinical effectiveness and safety of older versus fresher RBCs in critically ill patients. First, anaemia is very common and up to 95% of patients in intensive care units (ICUs) have haemoglobin values below the normal range after 2–3 days. 11,12 Second, RBC transfusion is one of the most common therapeutic interventions in the ICU;11 30–50% of all patients receive at least one RBC transfusion during their ICU stay, and a significant number receive RBC transfusions during the pre- and post-ICU hospital stay; in the UK, 10% of all RBCs are transfused to patients in general ICUs. 13 Third, there is strong evidence from a previous landmark trial [the Transfusion Requirements in Critical Care (TRICC) trial]14 that a restrictive haemoglobin transfusion trigger is safe for most critically ill patients. As a result, the use of RBCs is more consistent in the critically ill, which decreases the potential confounding effect of wide variation in RBC use among similar patient groups. Fourth, the need to improve the evidence base for RBC transfusion in critical care is widely supported by clinicians. A research priority-setting exercise undertaken by the UK Intensive Care Society had identified this research topic among active ICU clinicians as an area of urgent need and importance for further research. This led to a research topic proposal to the Health Technology Assessment (HTA) programme, submitted by the Intensive Care Society on behalf of the clinical community. These factors also indicated a willingness from the clinical community to participate in the ABLE trial in the UK. Fifth, the rationale for RBC transfusion is to restore oxygen-carrying capacity. Critically ill patients frequently have an oxygen supply–demand imbalance, so the biological plausibility of a difference in effectiveness between fresher and older stored RBCs was high. Sixth, mortality from critical illness is high, typically between 20% and 25% in UK ICUs, so this question was of particular relevance to this patient population. The incidence of morbidity that is potentially related to RBC transfusion (e.g. organ failures and infections) is also high, making these useful secondary outcomes. Seventh, critical illness is associated with high health-care costs over a prolonged period, and patients suffer significant long-term disability and reduced health-related quality of life (HRQoL). 15 This is therefore an appropriate population in which to assess cost-effectiveness of this intervention. Eighth, critically ill patients are cared for in dedicated ICUs with specialist staff experienced in the use of RBCs and undertaking complex research protocols, which meant that a trial was feasible and likely to be relatively efficient.

Changes to red blood cells during storage

Several reviews have summarised a large volume of literature characterising well-defined biochemical and cellular changes to RBCs during storage, collectively referred to as the storage lesion. 6,8,16 During storage, RBCs undergo a predictable change in structure, evolving from biconcave discs to spheroechinocytes. These changes are associated with a number of biochemical and biomechanical changes, including a depletion of adenosine triphosphate and 2,3-diphosphoglycerate (2,3-DPG), membrane phospholipid vesiculation and loss, protein oxidation, lipid peroxidation of cell membranes and loss of deformability. It is biologically highly plausible that these changes may have adverse clinical consequences by diminishing oxygen transport through capillary networks. Specifically, structural changes and the loss of cell deformability impair the ability of 8-µm RBCs to navigate the capillary networks, which typically have a smaller diameter. Even if RBCs are able to navigate capillaries, the depletion of 2,3-DPG in the stored RBCs may impair their ability to release oxygen to cells and tissues. Transfusing 2,3-DPG-depleted RBCs in primates, including humans, depletes systemic 2,3-DPG levels and shifts the oxygen dissociation curve leftwards, changes that require 24 hours to several days to reverse. Normal circulating RBCs also facilitate capillary transit by releasing nitric oxide, and this physiological effect is rapidly lost during storage.

The described changes occur at different rates over time, and can be modified but not eliminated with the use of storage solutions. Alterations are well established by 18–21 days, which is the typical age of RBCs currently transfused in the UK and other countries as standard practice. 17–19 At present, the standard used by blood providers to determine storage duration is that 75% of transfused RBCs should survive in the circulation 24 hours after transfusion. This is clearly not a measure of clinical or physiological effectiveness, but rather of physical survival. These measurements are also undertaken in relatively healthy individuals, whereas critically ill patients have an abnormal microcirculation that could further accentuate the clinical importance of the storage lesion.

Prolonged RBC storage changes the supernatant as well as the RBC. In the suspension medium, studies have noted the generation of cytokines and other bioactive substances, including histamine, complement, lipids and cytokines. 8,16 These bioactive substances may stimulate proinflammatory pathways and perhaps change flow patterns in the microcirculation. Other well-documented time-dependent changes include a progressive fall in pH, an increase in plasma potassium levels and release of free haemoglobin from lysed RBCs.

A further relevant factor relates to the use of leuco-reduced prior to storage in the blood bank. In the UK, since 1999, all RBC products must be leuco-reduced prior to storage. This was introduced to minimise the risk of variant Creutzfeldt–Jakob disease transmission, but a substantial body of evidence suggests that this may alter the nature and severity of the storage lesion. In general, evidence would suggest that the storage lesion, based on laboratory assays, is reduced by removing leucocytes from the RBC bag. A substantial proportion of the published observational literature relating to RBC storage and clinical outcomes used populations in which non-leuco-reduced RBCs were used. The generalisability of this literature to the current product was also questionable. Despite this, many countries, including Canada, where the ABLE trial was initiated, have introduced universal leuco-reduced.

In summary, prior to the ABLE trial there was substantial evidence that biochemical and structural changes during storage could decrease the ability of RBCs to transport and release oxygen through effects on oxygen uptake and release, and the process of capillary transit and storage-related accumulation of substances in the supernatant could have adverse clinical effects.

Animal evidence relating to red blood cell storage

Prior to the ABLE trial, animal models had been developed and used to test transfusion efficacy. Earlier studies demonstrated that old stored rat blood did not improve tissue oxygen consumption as compared with fresh RBCs. 20 By bleeding animals and replacing blood with RBC-free solutions (isovolaemic haemodilution), haemoglobin concentrations were so low that each RBC transfusion should result in increased uptake of oxygen by cells (a supply-dependent state). If RBCs worked properly in the supply-dependent state, there should be a noticeable decrease in serum lactate levels and an increase in oxygen consumption. Using this model, studies consistently noted that the transfusion of rat RBCs stored under standard conditions for 28 days when compared with fresh RBCs (stored for < 5 days) did not consistently improve measures of tissue oxygen consumption or hypoxia. However, it had subsequently been shown that these observations were in part explained by low post-transfusion survival of stored rat RBCs, but the experiments demonstrate the potentially limited efficacy of stored RBCs. 20,21 In the most comprehensive animal study, Raat et al. 22 compared the transfusion of 2- to 6-day-old, 2- to 3-week-old and 5- to 6-week-old human blood in a rat isovolaemic exchange model, and showed a decrease in microvascular partial pressure of oxygen in the gut with older RBCs compared with fresh and intermediate blood. However, these changes were not marked, and their clinical relevance to human disease was uncertain.

In summary, animal studies provided some evidence that older RBCs had lower efficacy to transport oxygen than fresh RBCs, but did not allow any conclusion regarding the clinical importance of these effects in humans to be made. The limitations of animal models to human disease supported the need for adequately powered effectiveness trials in a relevant human population, with clinically relevant outcome measures.

Clinical studies examining the importance of the red blood cell storage lesion

Conclusions and summary of background in relation to existing research

Prior to the ABLE trial, the need for a trial was justified by a range of issues: (1) strong animal and laboratory evidence supporting the hypothesis that prolonged storage decreases RBC efficacy as an oxygen transporter and could result in deleterious clinical effects through inflammatory mechanisms; (2) observational studies in human populations reporting a number of associations between prolonged RBC storage and adverse clinical outcomes, including mortality and organ failure, but a clear recognition that this research design was inherently flawed by multiple forms of confounding and bias; and (3) the lack of completed or ongoing adequately powered trials addressing this question. In addition, the widespread use of RBCs in health care generally, and transfusion rates of 30–50% among critically ill patients, justified the need for a large, definitive trial to compare current RBC transfusion practice with exclusive use of fresher RBCs.

Context of the Age of BLood Evaluation trial in the UK

The importance of storage age of RBCs during critical illness was suggested as a topic to the National Institute for Health Research (NIHR) HTA programme and processed through the prioritisation and commissioning boards. Concurrently, the ABLE trial was funded and commenced recruitment in Canada, co-ordinated by the Ottawa Health Sciences Centre and run by the Canadian Critical Care Trials Group. The UK chief investigator (TSW) had been involved in the development and funding of the Canadian grant, but funding did not permit set-up of UK sites. The NIHR HTA programme commissioned a proposal for a UK arm of the ABLE trial through the UK Intensive Care Society, harmonised with the Canadian trial protocol, to ensure the proposed sample size was achieved. The UK proposal included a health economic evaluation, which was unique to the UK arm of the trial. This comprised a cost–utility analysis of fresh blood versus standard-aged blood and a methodological substudy to compare different versions of the EuroQol-5 Dimensions (EQ-5D) HRQoL measure in critical care patients. In addition, given the results of the primary analysis, a further substudy was subsequently added (following agreement with the funder) using the health economic data to evaluate factors associated with costs and quality-adjusted life-years (QALYs) in UK ABLE trial patients. Around the same time that the UK arm of the trial was funded, funding was obtained to extend recruitment to the Netherlands and France such that the ABLE trial became an international critical care trial supported by multiple funders and trial networks, but working to a single sample size and clinical trial protocol.

Objectives

Study end points

Study design

The ABLE study was an international double-blind, multicentre, randomised clinical trial. A summary flow diagram describing the trial is shown in Figure 1.

FIGURE 1.

The design of the ABLE trial in the UK. NOK, next of kin.

Patients were randomised to one of two groups, receiving either:

1. standard-issue RBCs (average storage age: 18–21 days)

2. red blood cells stored for ≤ 7 days (average storage age: 2–7 days).

Following randomisation, all decisions about when to transfuse were determined by caring clinicians unaware of group allocation, and there was no protocol for transfusion decision-making in the trial. The transfusion practice was, therefore, considered to represent current practice in terms of timing, triggers for transfusion and transfusion volumes. The only difference in treatment between the groups was, therefore, the storage age of RBCs used, with both groups receiving currently licensed RBC product. There were no additional blood samples or procedures beyond routine care, and all follow-up data were gathered from the patient record or from questionnaire-based follow-up.

Participant identification and selection

Consent

Most patients were incapacitated at the time of eligibility (critical illness, MV or sedation), such that the Mental Capacity Act 200527 and Adults with Incapacity (Scotland) Act 200028 provided guidance. Because transfusion is usually a time-critical intervention, excessive delays in transfusion that were directly attributable to the trial could result in an enrolment bias by clinicians. In practice, the decision to enter the trial needed to be made within 1–2 hours in most cases (or sooner in cases with bleeding). Several approaches to obtaining consent were used, depending on the urgency of transfusion, mental capacity of the patient and availability of relatives. In addition, the different laws relating to research in incapacitated patients in Scotland versus the remainder of the UK influenced the processes.

• Patients considered to have mental capacity were approached directly, and their informed consent requested.

• For patients who lacked capacity, the approaches differed between Scottish and English/Northern Irish sites in accordance with local legislation regarding the inclusion of incapacitated patients in medical research. This also arose because the ABLE trial was not considered to be a clinical trial of an investigational medicinal product. As such, the laws governing consent differed between UK countries.

In addition, patients who survived their illness and regained capacity were approached for consent to remain in the trial. The processes used, which reflected the differences in law between England/Northern Ireland and Scotland, are summarised in Figures 2 and 3.

FIGURE 2.

Consent flow chart used to determine the appropriate method for obtaining consent in Scotland. WA, welfare attorney.

FIGURE 3.

Consent flow chart used to determine the appropriate method for obtaining consent in England and Northern Ireland.

Randomisation

Management and data collection during the intervention

Follow-up

In the UK, the duration of follow-up extended to 12 months post randomisation, or until a participant died or withdrew from the trial. This duration was longer than for the other international sites, where follow-up for 6 months was undertaken. The longer UK follow-up was undertaken for the health economic analysis.

Premature withdrawal of study participants

There were no predefined withdrawal criteria for patients following trial entry. We aimed to approach all patients for consent to remain in the trial if consent/assent/non-objection for randomisation was provided by a relative, welfare attorney, next of kin, personal consultee or independent clinician. For patients who chose not to remain in the trial or who withdrew during the intervention or follow-up period, no further contact was made. We asked these patients if they were happy for us to use data collected up to that point and data that could be acquired without contacting or involving them further. If they requested for all data to be removed, all data for that patient were destroyed. The same approach was used for situations in which a relative, welfare attorney, next of kin or personal consultee requested withdrawal in the absence of a mentally capacitated patient.

Statistical and data analysis

Adverse events

Adverse events were expected to occur frequently among enrolled patients as a result of ongoing critical illness. AEs that were expected in this population (i.e. events that are in keeping with the patient’s underlying medical condition) were not reported as AEs. Congenital abnormalities/birth defects were not relevant to this trial. Predefined clinically relevant complications recorded in the case report form, such as organ failures and infections, were not additionally reported as AEs, with the exception of acute transfusion reactions. AEs related to blood transfusions were reported, including all acute transfusion reactions occurring during the intervention period. This approach was used consistently across the international recruiting centres.

Serious adverse events (SAEs) were expected in many of the participants, consistent with critical illness. SAEs that were expected in this population (i.e. events that are in keeping with the patient’s underlying medical condition) and those that were collected as outcomes of the trial, including death and organ failure, were not reported as SAEs. Other SAEs were reported at the discretion of the research team at each site. The period of reporting of SAEs by each site research team was from randomisation to 90 days following randomisation for the index hospital admission or until hospital discharge, whichever was sooner.

Trial management and oversight

Ethics approvals

The study was conducted in accordance with the principles of the International Conference on Harmonisation Tripartite Guideline for Good Clinical Practice. A favourable ethics opinion was obtained from the Scotland A Research Ethics Committee (REC; 11/AL/0111) and the Oxford C REC (11/SC/0417). Local research and development (R&D) approvals were obtained at all sites prior to commencing recruitment.

Patient and public involvement in research

As the ABLE trial was funded in the UK to participate in the international trial, we did not involve patients or the public specifically in the UK project. Local patient representatives reviewed the study materials, specifically the information sheets for participants and relatives, and provided feedback on clarity and presentation.

Approvals

The UK ABLE trial required a complex set of approvals to be in place to commence recruitment. These reflected the international nature of the trial, the large number of centres involved, and the need to set up both ICUs sand blood banks to enable randomisation and intervention management (Table 2). In total, over 25 separate contracts/agreements were required to be set up and agreed between the sponsor’s legal department and various organisations. This took considerable time and effort, and resulted in delays while agreements were finalised.

One practical issue related to the definition of the UK protocol. The Canadian co-ordinating centre ran the international trial and acted as sponsor for the Canadian sites, and a Canadian version of the protocol was used. In the UK, with separate funding from the NIHR HTA programme to the University of Edinburgh, the UK co-sponsors (University of Edinburgh and NHS Lothian) required a separate document representing the UK protocol that was legally distinct from the Canadian protocol, despite the unification of protocols into a single international trial. A separate UK protocol therefore needed to be written and approved by the ethics committee.

Ethics considerations

The different legal frameworks for incapacitated patients in England and Scotland required separate ethics applications and approval processes, with trial materials using appropriate terminology for surrogate decision-makers. In many international centres, a true waiver of consent was granted because the two intervention groups were both part of standard care, and the decision to transfuse RBCs was determined by clinicians and not in accordance with the trial protocol. In addition, delays to transfusion were considered unacceptable, as these might have delayed treatment and reduced patient safety. UK law did not allow this approach, but resulted in different approaches in England/Northern Ireland from those used in Scotland (see Chapter 2). These differences made recruitment more difficult for some patients in Scotland, which adversely affected recruitment rates. Approval of consent by telephone was important in Scotland to decrease this impact. In all cases, the patient was approached wherever possible to obtain permission to continue in the trial if they survived their ICU admission and regained capacity. This took considerable research nurse resource and, in some cases, was difficult to achieve before patients were discharged home. The different approaches required to obtain consent or lack of objection to participate and remain in the trial illustrate the complex and time-consuming processes involved in trials recruiting critically ill, incapacitated patients with time-sensitive recruitment windows and interventions.

Site set-up

Site set-up required both the clinical (in the ICUs) and blood bank teams to be trained and SOPs established to execute the randomisation and group allocation. Blood bank set-up was challenging, but was facilitated by specialist blood bank co-ordinators employed for the purpose of set-up, monitoring and support throughout the trial. A detailed set of protocols and procedures was developed to enable potential participants to be rapidly screened to ascertain if allocation to ‘fresh’ RBCs was feasible based on blood group, cross-match and blood availability. In addition, procedures to ensure checking, blinding and modified blood issue procedures were established. There were frequently delays in this set-up as a result of the intense pressure many NHS blood banks worked under, and competing activities such as inspections and audits. A limited number of technician staff were generally available to undertake randomisation and group allocation procedures, which limited recruitment periods to weekdays in most centres. Despite dedicated funding for this activity, many blood banks did not have access to additional staff to support the trial beyond routine NHS work. This resulted in many potential participants being ‘missed’, and limited participation to mainly patients in whom first RBC transfusions were prescribed during weekday working hours. In addition, procedures to ‘tag’ participants on local blood bank systems to ensure that subsequent requests for RBCs maintained both group allocation and blinding were necessary, but did not delay blood issue. A modified blood-checking procedure at the bedside by clinical staff that ensured that national standards were adhered to but that maintained the blinding of RBC storage age was also needed in all centres.

Although the ABLE study was a trial based in the ICU, the major logistic challenges were in the blood banks. This was the first large, multicentre UK trial that required multiple NHS blood banks to allocate trial participants to receive different blood products under emergency conditions while maintaining blinding from clinical teams. The success of the trial reflected very considerable effort and support by blood bank staff and the blood bank co-ordinators working on the trial.

Timelines

A summary of the major trial set-up timelines is shown in Figure 4. We found a wide variation between sites for times to R&D approvals, and times to recruitment of the first patient from final approval (Figure 5). The reasons were multifactorial and varied between centres. Delays with final contracts between the sponsor and study site were prevalent, as a result of the slow responses from legal teams. These resulted in a ‘knock-on’ effect on final R&D approval; this was also delayed in some centres until clinical sites were ready to start screening, to minimise the impact of NIHR metrics that recorded time to first recruit. For clinical set-up, delays in blood bank training were a major source of delay in some centres. Frequent reasons were competing priorities, such as national inspections of the laboratory service (especially the Medicines and Healthcare products Regulatory Agency), unrelated to the ABLE trial, and staff shortages such that the organisational changes required to run the trial were delayed. The funding of dedicated blood bank co-ordinators in the trial was vital to minimise these delays.

FIGURE 4.

Summary of the major logistic milestones during the set-up of the UK ABLE trial. CSP, Coordinated System for gaining NHS Permission.

FIGURE 5.

Summary of the time (days) for R&D approval and the time (days) from approval to first recruitment for the 20 ICUs that participated in the UK ABLE trial. FT, foundation trust.

Trial management

A TMG met every 4–6 weeks throughout recruitment by teleconference. These meetings were supplemented by regular investigator teleconferences with research staff (from both ICUs and blood banks), at which representatives from all sites were encouraged to share both positive and negative experiences of recruitment and trial conduct. These meetings were supplemented by regular newsletters and recruitment tables circulated by e-mail. Screening logs were returned and examined in real time. This was useful for a number of reasons: (1) it provided a rapid indication of when staffing or other problems were affecting recruitment; and (2) it provided valuable data on the reasons that potential patients were ‘missed’, which were collated and shared with all sites. This enabled a focus on solving common recruitment problems that occur across all trial sites. For example, it rapidly became apparent that the most common reason for missed recruitment was that the first transfusion took place out of hours when research staff and/or blood bank staff were unavailable. Several ICUs used local audit to change their practice and minimise night-time transfusions, consistent with current guidance.

These data enabled the reasons for non-enrolment of eligible patients to be clearly tracked during trial conduct (Table 3) and were used to implement improvements in real time.

A continuously updated ‘top tips for recruitment’ checklist and sharing of solutions to problems was a focus of these meetings, which maintained momentum. The ‘top tips’ checklist was used as an audit tool to help each site explore whether or not they could optimise recruitment (Table 4).

Follow-up

A predefined strategy for follow-up at 6 and 12 months post randomisation was used. First, the patient’s GP was contacted to ascertain survival status. Survivors were sent questionnaires by post, accompanied by a £5 gift token. At 12 months, non-response to postal questionnaire was followed by a second postal questionnaire. Failure to respond to postal follow-up was followed by up to three attempted contacts by telephone to complete the questionnaires. In addition, any queries were resolved by telephone contact, with up to three attempted contacts. As follow-up is known to be challenging in critical care survivors, we collected data to summarise the total time needed to achieve follow-up data. We also analysed the effectiveness of different approaches to follow-up in this population.

Audit data indicated that the average time required for each follow-up was 26 minutes at 6 months and 25 minutes at 12 months. Table 5 shows the success rate of the different follow-up strategies. Telephone contact was attempted at least three times, and this significantly improved follow-up response rates, especially at the 6-month time point.

Audit of blood transfusion in participating intensive care units

In order to understand how the patients recruited to the ABLE trial compared with all patients receiving blood transfusion in UK ICUs, we undertook an audit of 489 sequential ICU admissions to 15 out of 20 ABLE trial sites. Each site audited at least 30 sequential admissions, and used blood bank data to ascertain and record all RBC transfusions during the hospital stay, including the numbers of transfusions and whether or not the transfusion occurred pre, during or post ICU care.

Transfusion data were unavailable for six patients. The audit showed that 222 out of 483 (46%) of all patients admitted to the typical UK ICUs received a RBC transfusion during their hospitalisation. Transfused patients received a mean of 6.8 RBC units during hospitalisation. Data showed that 110 out of 483 (23%) patients received RBCs prior to ICU admission [49/483 (10%), exclusively pre ICU admission]. These patients were ineligible for the ABLE trial. A total of 135 out of 483 (28%) patients received RBCs during ICU stay [66/483 (14%), exclusively during ICU stay]. The mean RBC use for transfused patients during ICU care was 4.4 RBC units per patient. A total of 73 out of 483 (15%) patients received RBCs during the post-ICU discharge period [26/483 (5%), exclusively during the post-ICU period]. Many patients received RBCs at multiple time points during their hospitalisation.

These data, collected during the ABLE trial recruitment, indicated that many patients received RBCs but were either not eligible for the ABLE trial or were not included.

Concurrently, we examined the screening logs for the ABLE trial and found that of 3754 patients who received RBCs during their ICU care, only 1035 (28%) were eligible for the ABLE trial.

Together, these data clearly showed that, despite the pragmatic design of the trial, the ABLE trial population was only around 25% of ICU admissions, and this accounted for a minority of patients receiving RBCs during ICU care.

Patients

The international trial cohort was recruited between March 2009 and May 2014. The UK cohort was recruited between January 2012 and October 2014. The UK patients recruited after the international trial sample size was achieved (over a period of 6 months) were included in this summary of the UK cohort and the health economic evaluation. There were 293 patients recruited to the UK trial who were included in the main trial analysis, previously published. 31 In total, 359 patients were recruited in the UK (90% of the revised 400 patient sample size; see Figure 6 for timelines). In total, 354 patients were included in the UK cohort reported here for clinical outcomes (multiple imputation enabled the inclusion of 357 patients in the economic evaluation). These included 58 patients recruited after achievement of the international sample size. In four cases there were no primary outcome data, and in one case a patient withdrew consent.

FIGURE 6.

Recruitment accrual over time to the UK ABLE trial. The revised trial target of 400 patients was agreed with the HTA programme when the end date for the international trial was predicted.

In the international trial, 2510 patients underwent randomisation; 80 (3.2%) were withdrawn after randomisation because primary outcome data could not be obtained, leaving 2430 patients (1211 in the group allocated to receive fresh blood and 1219 in the group allocated to receive standard-aged blood) in the intention-to-treat analysis.

Baseline data were available for 2412 of the 2430 patients with primary outcome data. Of these 2430 patients, 94 (3.9%) did not receive any RBC transfusions. The overall rate of loss to follow-up was 3.2% at 90 days.

For the UK, the Consolidated Standards of Reporting Trials (CONSORT) flow diagram describing the trial is shown in Figure 7. In total, 5989 patients received a blood transfusion during the first 7 days of ICU stay. However, of these, 2081 did not meet the other inclusion criteria. Of the 3908 patients who met all inclusion criteria, 2270 were excluded in accordance with the protocol. By far the most common reason for exclusion, in 2021 patients, was the patient having had a transfusion earlier during their hospital stay. This included transfusions undertaken prior to ICU admission, for example to treat haemorrhage or perioperative blood loss. Of the 1638 patients who were eligible for randomisation, a high proportion (n = 1279, 78%) were excluded for a variety of practical reasons that meant that randomisation and group allocation were not feasible. The most common reason was transfusion at night (n = 957). In these situations, delay of transfusion to enable randomisation was not considered clinically appropriate or safe, but the randomisation procedure could not be organised either by clinical research staff or at the blood bank level. In total, 359 patients were randomised in the UK cohort, out of 1638 (22%) eligible patients according to the trial protocol. Five patients were not included in the primary analysis. In four cases, there were no primary outcome data; in one case, a patient withdrew consent. Of these 354 patients, 179 were randomised to the group allocated to receive fresh blood and 175 patients to the group allocated to receive standard-aged blood.

FIGURE 7.

The CONSORT flow diagram for the UK ABLE trial. a, No admission data from three sites. N/A, not applicable.

Baseline characteristics of the patients for the ABLE trial UK cohort, and for the overall international trial, are shown in Table 6. Patients had an age and sex profile typical of general ICU populations in the UK. The illness severity, based on APACHE II score and requirement for organ support, was high and the majority had significant levels of organ dysfunction based on the MODS score. Almost all patients were non-elective ICU admissions. The majority had a medical diagnosis, most likely to be consistent with the non-eligibility of patients receiving blood transfusion prior to ICU admission, which will include many surgical and trauma patients. The UK ABLE trial cohort was similar to the international cohort, although there were more elective ICU admissions in the UK ABLE trial cohort.

Intervention

Table 7 shows the comparison of transfusion practice between the groups, between the UK ABLE trial and the international trial. In the international trial, a total of 5198 RBC units were given to patients in the group allocated to receive fresh blood and 5210 to patients in the group allocated to receive standard-aged blood. The pre-transfusion haemoglobin concentration was similar in the international cohort to the UK ABLE trial and the values indicated a restrictive transfusion practice consistent with current evidence. Patients received a similar number of RBC units in both groups. This was the case for the overall trial, and for the UK cohort. There was excellent separation of storage-age profile, which was similar in the UK to the international cohort. The difference between the distribution of storage age in each group was clinically relevant (see Figure 8; reproduced from full trial data). Protocol compliance was high. The rate of adherence to the transfusion protocol was 95.4% for all red blood cells transfused, with 100% of patients allocated to receive standard-aged blood receiving only standard-issue RBCs and 84.0% of patients allocated to receive fresh blood receiving only red blood cells stored for ≤ 7 days. In the group allocated to receive fresh blood, only 6.6% of the patients received > 1 RBC unit that had been stored for > 7 days, and only 4.6% received > 2 units that had been stored for > 7 days. Most patients in the group allocated to receive fresh blood [238 of 249 patients (95.6%)] who received only one RBC transfusion received exclusively RBC units that had been stored for < 8 days.

FIGURE 8.

Distribution of RBC units in accordance with length of storage, as transfused to patients allocated to the fresh blood arm (green bars) and to the standard arm (black bars). Source: Lacroix et al. 31

Primary outcome

In the international trial, at 90 days after randomisation, 37.0% of patients in the group allocated to receive fresh blood and 35.3% of patients in the group allocated to receive standard-aged blood had died. The unadjusted and adjusted risk differences are shown in Table 8. There was no evidence of any clinically or statistically important difference between the groups. The proportion of patients who died was similar in the UK ABLE trial cohort.

Secondary analyses

In the international trial, the survival analysis of the time to death showed a hazard ratio in the group allocated to receive fresh blood, compared with the group allocated to receive standard-aged blood, of 1.1 (95% CI 0.9 to 1.2; p = 0.38). No significant difference in mortality was observed between the groups on the basis of follow-up duration, age, number of units transfused, APACHE II score or admission category. Differences between the groups in the UK ABLE trial cohort were similar to those in the international cohort (see Table 8).

In the international trial, no significant differences were observed with respect to major illnesses, duration of respiratory, haemodynamic or renal support, or length of stay in the ICU or hospital. There were also no differences in reported transfusion reactions. Data were comparable for the UK ABLE trial subgroup (see Table 8).

Data for other secondary outcomes are shown in Table 9. There were no differences between the groups in the international trial, and data were similar for the UK ABLE trial cohort.

The per-protocol analysis of the primary outcome, which included only patients who received a transfusion, showed no difference between the groups in the international trial, and data were comparable for the UK ABLE trial cohort. In the sensitivity analysis of the primary outcome, in which outcomes of the patients in the fresh-blood group who received only RBCs that had been stored for ≤ 7 days, were compared with the outcomes in patients in the standard-aged blood group, who received RBCs that had been stored for > 7 days, also showed no differences between the groups. Data are shown in Tables 10 and 11.

Aim

In this study we investigated the cost-effectiveness of fresh blood versus standard-aged blood in critical care patients. The sample comprised patients in the UK ABLE trial.

Methods

Results

In our sample, the mean number of units of RBCs transfused was 4.1 RBC units [standard deviation (SD) 4.9 RBC units] in the group allocated to receive fresh blood (n = 181) and 3.7 RBC units (SD 3.7 RBC units) in the group allocated to receive standard-aged blood (n = 176; Table 12). The mean number of units of fresh-frozen plasma transfused was 1.1 RBC units (SD 5.1 RBC units) in the group allocated to receive fresh blood (n = 181) and 0.6 RBC units (SD 2.1 RBC units) in the group allocated to receive standard-aged blood (n = 176). The mean number of units of platelets transfused was 0.6 RBC units (SD 2.3 RBC units) in the group allocated to receive fresh blood (n = 181) and 0.3 RBC units (SD 1.1 RBC units) in the group allocated to receive standard-aged blood (n = 176).

The mean length of stay in hospital was 36.9 days (SD 28.8 days) in the group allocated to receive fresh blood and 36.3 days (SD 33.2 days) in the group allocated to receive standard-aged blood. Mean length of stay in the ICU was 14.8 days (SD 9.5 days) in the group allocated to receive fresh blood (n = 179) and 15.3 days (SD 10.5 days) in the group allocated to receive standard-aged blood (n = 175). While in the ICU, the mean length of use of invasive MV was 12.3 days (SD 9.4 days) in the group allocated to receive fresh blood (n = 181) and 12.5 days (SD 10.8 days) in the group allocated to receive standard-aged blood (n = 176). The mean number of days on vasopressors and inotropes was 5.5 days (SD 6.8 days) and 1.3 days (SD 3.5 days), respectively, in the group allocated to receive fresh blood and 5.1 days (SD 6.4 days) and 1.5 days (SD 3.9 days), respectively, in the group allocated to receive standard-aged blood. Mean number of days of CRRT was 3.2 days (SD 6.4 days) in the group allocated to receive fresh-blood (n = 181) and 3.2 days (SD 6.4 days) in the group allocated to receive standard-aged blood (n = 176).

Post discharge, surviving patients in both groups used primary care and community care services. The mean number of GP visits in the clinic up to 12 months was 6.3 visits (SD 6.8 visits) in the group allocated to receive fresh blood (n = 76) and 3.9 visits (SD 4.2 visits) in the group allocated to receive standard-aged blood (n = 83). GP visits at home or GP telephone consultations were less frequent. The mean number of district nurse visits was 11.6 visits (SD 32.5 visits) in the group allocated to receive fresh blood group and 12.5 visits (SD 29.6 visits) in the group allocated to receive standard-aged blood. The mean number of home care worker visits was 13.5 visits (SD 57.1 visits) in the group allocated to receive fresh blood and 19.4 visits (SD 107.9 visits) in the group allocated to receive standard-aged blood. Less frequent in both groups were visits to a practice or specialist nurse, physiotherapist, psychologist, occupational therapist, speech and language therapist, dietitian, counsellor or social worker. Post discharge, patients in both groups used secondary care services. The mean number of A&E visits was 0.7 visits (SD 1.3 visits) in the group allocated to receive fresh blood group and 0.9 visits (SD 2.8 visits) in the group allocated to receive standard-aged blood. The mean number of outpatient clinic visits was 4.6 visits (SD 4.6 visits) in the group allocated to receive fresh blood and 6.0 visits (SD 8.2 visits) in the group allocated to receive standard-aged blood. In the group allocated to receive fresh blood, 42% of patients were readmitted to hospital at least once for a mean period of 8.0 days; in the group allocated to receive standard-aged blood, these figures were 34% and 7.9 days.

Accounting for missing data, the mean total cost per patient was £32,346 (95% CI £29,306 to £35,385) in patients receiving fresh blood (n = 181) and £33,353 (95% CI £29,729 to £36,978) in patients receiving standard-aged blood (n = 176; Table 13). In both groups, approximately 85% of the total costs were incurred during the index hospital admission and 15% during follow-up.

Mean utility values were similar for the two groups and over time (see Table 13). Mean total QALYs per patient were 0.207 (95% CI 0.158 to 0.256) in the group allocated to receive fresh blood and 0.213 in the group allocated to receive standard-aged blood (95% CI 0.170 to 0.257).

As expected, at a maximum willingness to pay for a QALY of £20,000, the NMBs for the group allocated to receive fresh blood and the group allocated to receive standard-aged blood were negative: –£28,201 (95% CI –£31,323 to –£25,119) and –£29,102 (95% CI –£32,596 to –£25,609), respectively (see Table 13). Furthermore, at a maximum willingness to pay for a QALY of £30,000, the NMBs for the group allocated to receive fresh blood and the group allocated to receive standard-aged blood were negative: –£26,128 (95% CI –£29,345 to –£22,911) and –£26,978 (95% CI –£30,489 to –£23,465), respectively (see Table 13).

Mean costs and QALYs and NMBs were similar for complete cases (Table 14), with 30–50% missing data on total costs and QALYs.

There were no significant differences in costs between the two groups (mean incremental cost for the group allocated to receive fresh blood vs. the group allocated to receive standard-aged blood: –£231, 95% CI –£4876 to £4415) or in outcomes (mean QALYs gained –0.010, 95% CI –0.078 to 0.057) (Table 15). The INMBs for the group allocated to receive fresh blood versus the group allocated to receive standard-aged blood were not significantly different from zero at a maximum willingness to pay for a QALY of £20,000 (mean £25, 95% CI –£4587 to £4637) and £30,000 (mean –£77, 95% CI –£4821 to £4666).

At a maximum willingness to pay for a QALY of £20,000 and £30,000, the probability that fresh blood is cost-effective was 0.52 and 0.49, respectively (Figure 9); as noted, the probability that standard-aged blood is cost-effective is one minus this figure, indicating that the costs and outcomes for the two treatment groups were similar.

FIGURE 9.

Cost-effectiveness acceptability curve showing the probability that fresh blood is cost-effective vs. standard-aged blood at different values of the maximum willingness to pay for a QALY.

Incremental costs, QALYs gained and INMBs for the group allocated to receive fresh blood versus the group allocated to receive standard-aged blood remained not significantly different from zero when rerunning the analysis without adjusting for potential confounding factors, and using complete cases (see Table 15).

The results were not sensitive to changes in the cost of fresh blood and standard-aged blood, nor to changes of admission and follow-up costs; INMBs did not vary much from baseline values and in every case the INMB was not significantly different from zero (Table 16).

Summary

Our economic analysis to evaluate the cost-effectiveness of using fresh blood units versus standard-aged blood units for transfusions in critically ill patients in the UK showed that there are no differences in terms of costs and outcomes. The univariate sensitivity analysis showed that the results were not sensitive to assumptions made, and the probabilistic sensitivity analysis indicated that the probability that the use of fresh blood was cost-effective was close to the probability that standard-aged blood was cost-effective. The findings mean that there is no reason to prefer fresh blood to standard-aged blood on the basis of differences in quality or length of life, or on cost grounds.

Background and aims

Health-related quality of life is an important outcome in critical care survivors. The EQ-5D is a generic measure of HRQoL that allows the calculation of QALYs for undertaking cost–utility analyses. This is a recommended measure of HRQoL in critical care45 and it has been used in critical care trials. 46–48 Until recently, the EQ-5D-3L has been commonly used, in which each dimension in the descriptive system is assessed using three levels of severity. In order to reduce ceiling effects (the proportion of respondents reporting the best possible health) and to increase sensitivity to changes and differences in health status, the EQ-5D-5L was developed. 49 Several studies have compared the EQ-5D-3L and the EQ-5D-5L in different countries in both general population samples and specific patient groups, but none of these has examined the two measures in critical care patients. 49–59 The aim of the present study was to compare EQ-5D-3L scores with EQ-5D-5L scores in critical care survivors in the UK. The sample for the present study comprised 359 patients treated at any of the 20 participating UK centres.

Methods

Results

In total, 227 of the 359 patients were alive at 6 months, and, of these, 121 returned an EQ-5D questionnaire (53%). A total of 221 patients were alive at 12 months, and 119 returned a questionnaire (54%). The distribution of both sets of utility scores at each time point was wide and discontinuous (Figures 10–13). The mean utility scores at 6 months for the EQ-5D-3L and EQ-5D-5L were 0.51 (SD 0.37) and 0.58 (SD 0.31), respectively (Table 17). The median utility scores were 0.59 (IQR 0.19–0.73) and 0.65 (IQR 0.34–0.82), respectively. At 12 months, the mean values were 0.50 (SD 0.33) and 0.60 (SD 0.33), respectively, and the median values were 0.62 (IQR 0.19–0.69) and 0.70 (IQR 0.42–0.84), respectively. The means, distributions and medians were not significantly different at 6 months (all p-values were ≥ 0.22), and neither was the mean difference at 12 months (p = 0.11), but the differences in distributions and medians at 12 months were significant (both p-values = 0.03).

FIGURE 10.

Distribution of EQ-5D-3L utility scores at 6 months.

FIGURE 11.

Distribution of EQ-5D-5L utility scores at 6 months.

FIGURE 12.

Distribution of EQ-5D-3L utility scores at 12 months.

FIGURE 13.

Distribution of EQ-5D-5L utility scores at 12 months.

There were no differences in observed characteristics between patients receiving the two utility measures at each time point, with the exception that a higher proportion of respondents in the EQ-5D-3L group at 12 months had been allocated to receive fresh blood than that of the EQ-5D-5L group (p = 0.04; Table 18).

After adjusting for patient characteristics, there were no significant differences between the conditional mean and median EQ-5D-3L and EQ-5D-5L utility scores at 6 months (p ≥ 0.17; Table 19), but there were significant differences at 12 months (p < 0.05); mean EQ-5D-5L scores were 0.15 units higher with the EQ-5D-5L than with the EQ-5D-3L, and median EQ-5D-5L scores were 0.20 units higher.

No patients reported being in the worst EQ-5D-3L state or the worst EQ-5D-5L state at 6 or 12 months. Eight patients (15%) reported being in the best health state on the EQ-5D-3L at 6 months compared with six patients (9%) on the EQ-5D-5L. Seven patients (14%) reported being in the best health state on the EQ-5D-3L at 12 months compared with six patients (9%) on the EQ-5D-5L. These figures were not significantly different (p ≥ 0.29).

For the 112 regression models used to determine the extent of variation in utility score according to patient characteristics, in only three models was the coefficient on the patient characteristic significantly different from zero. In all three cases, the significant effect was found using the EQ-5D-5L in patients at the 12-month follow-up using median regression (sample size: n = 70 patients). These analyses showed that immunosuppressive therapy, chronic institutionalisation for > 6 months and deep-vein thrombosis had a significant negative impact on median EQ-5D scores; the marginal effects were –0.23 (95% CI –0.47 to –0.01; p-value = 0.05; number of patients with condition, 8), –0.68 (95% CI –1.22 to –0.13; p-value = 0.02; number of patients with condition, 2) and –0.87 (95% CI –1.42 to –0.31; p-value = 0.01; number of patients with condition, 2), respectively.

At 6 months, there were no unadjusted differences in utility scores (means, distributions and medians) between trial treatment groups using either the EQ-5D-3L or the EQ-5D-5L (p ≥ 0.13; Table 20). At 12 months, there were no unadjusted differences in utility scores between treatment groups using the EQ-5D-3L (p ≥ 0.40), but EQ-5D-5L distributions were significantly different between treatment groups (p-value of –0.03). When we controlled for patient characteristics, there were no significant differences in utility scores between treatment groups for either EQ-5D measure at either time point (p ≥ 0.20; Table 21).

Summary

In conclusion, this study compared the performance of the EQ-5D-3L and the EQ-5D-5L in measuring the HRQoL in critical care survivors in the UK. There was some evidence of difference between the two measures, and that the EQ-5D-5L discriminated between subgroups of patients with major comorbidities whereas the EQ-5D-3L did not, but further research is needed to confirm these findings.

Background and aims

Intensive care units provide potentially life-saving interventions to critically ill patients, but incur considerable costs when treating patients. 61 Length of stay is a driver of ICU costs, and factors affecting this include institutional factors (such as organisational structure and resources), medical factors (especially cause of admission and severity of illness) and psychosocial factors (such as communication between clinical staff, and with the patient’s family). 62 On discharge from the ICU, survivors often have long-term health problems that affect their quality of life and survival, and health service utilisation and costs; factors associated with increased resource use include increasing age, comorbidities, organ dysfunction score and previous resource use. 15,61 The aim of the present study was to investigate the factors associated with hospital costs and total costs and QALYs up to 12 months after ICU admission in critically ill adults in the UK.

Methods

The sample for the present study comprised 359 patients treated at any of the 20 participating UK centres. Two patients had no study measurements and so were not included in the analysis, giving a sample of 357 patients.

Outcome measures

Statistical analysis

Results

In total, 219 of the 357 patients were alive at 12 months, and, of these, 121 returned an EQ-5D questionnaire at 6 months (55% of survivors) and 119 at 12 months (54%). Data on QALYs up to 12 months were available for 96 patients (44%). Including those patients who died, QALY data up to 12 months were available for 228 patients (64% of all patients). The distribution of utility scores at each time point and QALYs up to 12 months was wide and discontinuous (Figures 14–17). The high frequency of zero values in QALYS up to 12 months across all patients reflects the number of patients who died before 6 months (132 patients). The mean utility scores at 6 and 12 months were 0.54 (SD 0.33) and 0.56 (SD 0.33), respectively (Table 23). The medians were 0.62 (IQR 0.28–0.80) and 0.66 (IQR 0.31–0.80), respectively. Mean and median QALYs up to 12 months across all patients were 0.18 (SD 0.26) and 0 (IQR 0–0.40), again reflecting the number of patients who died before 6 months. Values for survivors only were 0.43 (SD 0.23) and 0.46 (IQR 0.27–0.59), respectively.

FIGURE 14.

Distribution of utility scores at 6 months (survivors only).

FIGURE 15.

Distribution of utility scores at 12 months (survivors only).

FIGURE 16.

Distribution of QALYs up to 12 months (all patients).

FIGURE 17.

Distribution of QALYs up to 12 months (survivors only).

Index hospitalisation costs were available for 352 patients (99% all patients), and the mean and median values were £30,803 (SD £20,501) and £26,771 (IQR £15,974–39,744), respectively. Follow-up costs to 12 months were available for 108 survivors (49% of survivors), with mean and median values of £8324 (SD £11,908) and £3819 (IQR £1549–10,617), respectively. Total costs up to 12 months across all patients had mean and median values of £31,867 (SD £21,130) and £26,972 (IQR £16,073–41,590), respectively. The distribution of all three cost variables was positively skewed (Figures 18–20).

FIGURE 18.

Distribution of index hospitalisation costs (all patients).

FIGURE 19.

Distribution of follow-up costs up to 12 months (survivors only).

FIGURE 20.

Distribution of total costs up to 12 months (all patients).

The mean age of patients was 62 years and 53% were male (see Table 23). The prevalence of pre-existing significant comorbid illnesses ranged from 1% (disabling stroke, chronic institutionalisation for > 6 months) to 11% (type 1 or 2 diabetes mellitus with evidence of end-organ damage). The mean APACHE II score at ICU admission was 25 points (SD 7 points) and the mean worst MODS score during ICU stay was 7 points (SD 3 points), with the modal category 5–8 points (50% of patients). Twenty-five per cent of patients died during the ICU stay, and 39% died by the end of the 12-month follow-up.

None of the covariates was a statistically significant predictor of utility scores at 6 or 12 months or of QALYs up to 12 months, either for all patients or survivors only (models 1–4; results not shown).

Index hospitalisation costs were predicted by worst MODS score during ICU stay and whether or not the patient died in the ICU (Table 24). Controlling for whether or not the patient died, hospitalisation costs increased with worsening MODS score; patients with a MODS score of 5–8 points incurred around £10,000 higher mean costs than those with values of 0–4 points, and those patients with scores of ≥ 9 points incurred mean costs that were £16,000–19,000 higher than those with scores of 0–4 points (note that the values for the 9- to 12-point and ≥ 13-point categories were not significantly different from one another). Patients who died incurred, on average, £18,000 lower costs than those who did not die, depending on their MODS score. Survivors with prior chronic institutionalisation incurred costs during the 12 months’ follow-up that were, on average, around £24,000 higher than those without chronic institutionalisation. In the model of total costs up to 12 months, the following covariates were significant: costs were higher with worsening MODS score for those who died in the ICU and those with prior chronic institutionalisation. Controlling for whether or not the patient died, total costs increased with worsening MODS score by, on average, around £8000, £19,000 and £21,000 per patient for those with scores of 5–8, 9–12 and ≥ 13 points, respectively, compared with those with a score of 0–4 points. On average, those who died incurred £30,000 lower costs than those who did not, and those with prior chronic institutionalisation incurred total costs that were, on average, around £24,000 higher than those without chronic institutionalisation. None of the other covariates was individually statistically significant as a predictor of any of the cost measures, nor were they jointly significant when added to the models just described.

Summary

We evaluated the factors associated with hospital costs and total costs and QALYs up to 12 months after ICU admission in critically ill adults in the UK. None of the variables in our data was a significant predictor of utility scores at 6 or 12 months, or of QALYs up to 12 months. The lack of an effect may reflect the limitations of our data set in terms of available measures, sample size and data missingness. Consistent with previous studies, we found that ICU costs were an increasing function of illness severity, measured in terms of MODS score and mortality. 62 Contrary to previous studies, costs from hospital discharge up to 12 months were largely unaffected by acute illness factors, age or comorbidities; however, there was some evidence that chronic institutionalisation, included as an indicator of pre-existing morbidity, was associated with 12-month costs. 15,63

Study conduct

The ABLE UK study was specifically funded to collaborate with the Canadian ABLE trial [funded by the Canadian Institutes of Health Research (CIHR)]. Specifically, the aim was to ensure the recruitment of the intended sample size of 2510 patients, and to include a UK-specific health economic evaluation. At the time of funding, Canadian recruitment was slower than anticipated and there was a risk that the trial would not achieve the target sample, which would have significantly reduced its external validity, generalisability and impact. Funding by the HTA programme, together with recruitment in France and the Netherlands, meant that recruitment was completed to target.

There were several relevant learning issues from this model of funding. First, despite the HTA programme ‘fast-tracking’ the application following the commissioned call, the process of review and final funding, followed by contract finalisation, meant that the ABLE UK trial came ‘on-line’ almost 2 years after recruitment to the international trial began. The further time required to set up recruitment in the UK further delayed contribution to the main trial. Second, several practical issues resulted from the international arrangements: collaboration and data sharing agreements were needed between the UK and Canada; separate sponsorship arrangements for the UK were required; and the trial protocol required adaptation to the UK, mainly as a result of different legal frameworks for the inclusion of incapacitated patients. Third, the legal and contractual burden for the trial was very high as a result of the multiple stakeholders, which created a significant workload for the sponsor and delays in trial set-up, despite attempts to expedite delays. Of relevance, there was no funding allocated to the very high legal/contracting burden for the sponsor (the University of Edinburgh). The consequence of these issues was that, despite intense effort by the trial management team, the set-up time to start recruitment meant that approximately 40% of the international trial sample size had been achieved before UK recruitment began. The recruitment rates in the UK were close to those predicted, and comparable to non-UK centres, but the international sample size was achieved before the UK target of 500 (revised to 400 during the trial) was achieved. A consequence was a risk that insufficient longer-term outcome data would be available for the health economic evaluation. Through collaboration between the international trial team, UK funder, sponsor and ethics committees, we were able to gain approval to extend recruitment in the UK until the database for the international cohort was locked for analysis. This approach allowed us to maximise the use of funds in the UK and obtain a larger data set for economic evaluation. These processes could have been improved and shortened if funding across international collaborators could be considered in a more parallel approach, or the review process for commissioned research could be shortened further once an international trial was under way. In addition, pre-agreed arrangements for sponsorship, shared protocols and data governance between the NIHR and equivalent international funders might have reduced delays and workload.

The UK ABLE trial was the first, to our knowledge, to require considerable research activity from blood bank staff across multiple NHS sites. Blood banks were required to manage group allocation and blinding, which involved implementation of study-specific SOPs and blood inventory management. We encountered several challenges that resulted in significant delays and limited recruitment. First, some blood banks were managed by the blood transfusion service, some by the NHS, and some were contracted partly/wholly to private providers; this resulted in contractual and governance challenges that delayed agreements. Second, almost all blood banks worked under intense pressure with staff shortages and competing activities that took priority over research, notably inspections. As a result, and despite the provision of set-up funding and generous per-patient remuneration, many blood banks were unable to complete training and site set-up procedures in a timely manner, because they were unable to free up dedicated time for these activities. This occurred despite strong support from clinical and technical staff within centres, and enthusiasm to participate in the trial. The nature of the work also required relatively senior staff to implement set-up and training, and the group allocation procedures. Employing blood bank co-ordinators on the project management team, who visited and supported all blood banks, was crucial in minimising these delays and ensuring governance issues were addressed in a consistent manner. Our experience strongly supports the use of similar dedicated individuals in any future trials involving extensive blood bank procedures across multiple sites. We would also recommend identifying methods to protect blood bank staff time early in the trial to ensure set-up times are minimised, as this was the most frequent cause of delays. Finding ways to enable blood bank procedures to occur outside normal working hours would also have significantly increased recruitment, as this was the reason for non-inclusion of otherwise eligible patients in 75% of cases.

The UK ABLE trial required recruitment of incapacitated patients in a time-sensitive manner, as decisions were required within a few hours in order to avoid delays in blood transfusions that had been prescribed for clinical reasons. This resulted in a high level of complexity in the consent procedures, which resulted in significant resource use and limited recruitment. First, different approaches were required in England/Northern Ireland versus Scotland, as a result of different mental capacity legislation. Specifically, in Scotland, patients could not be included unless consent was obtained from a relative/welfare guardian prior to randomisation. This meant that patients for whom a relative/welfare guardian was unavailable during the short recruitment window could not be included in the trial. The provision for telephone consent was essential to decrease the impact of this limitation. The experience of research staff, based on informal feedback, was that this was acceptable to most relatives. In England/Northern Ireland, the ability to ask a professional consultee was available, which reduced the pressure on research staff and made enrolment potentially easier. Experience during the trial suggested that the English legislation improved efficiency and was easier to implement. In all cases (Scotland and England/Northern Ireland), the intention was to discuss ongoing continuation in the trial with survivors once they regained capacity. This was often many days/weeks after randomisation and after discharge from the ICU, which required significant time commitment from research staff and was sometimes unsuccessful if patients had been discharged home. There was only one withdrawal of consent for a patient who regained capacity, suggesting that these processes were acceptable. After seeking clarification from the ethics committee, we amended the protocol (amendment 4) to enable patients to be contacted after hospital discharge with information and for consent to remain in the trial. This appeared to be acceptable to patients; however, when no response was obtained, it meant that the patient remained in the trial based on the agreement provided in hospital by the professional or personal consultee (England/Northern Ireland) or relative/welfare guardian (Scotland). The complexity of these processes illustrates the ethics and governance challenges of critical care research, even when the intervention and control group practices were all within current routine care, as was the case in the trial. Our experience strongly supports the ability to use telephone consent when appropriate (especially in Scotland), and the use of professional consultees when relatives were not available (in England/Northern Ireland). Early clarification of procedures to be used when follow-on consent is not obtained should also be considered in future similar research, including contact post hospital discharge. Our experience also highlights the high resource costs of obtaining consent in critical care trials, which should be recognised in funding applications.

We found the use of regular teleconferences with site representatives useful in maintaining enthusiasm, recruitment rates and, especially, for identifying and sharing common problems and finding solutions. The development of a ‘top tips’ resource and a summary of common blocks to recruitment and solutions that worked well and less well was effective, especially as a ‘living’ document that was regularly updated and circulated. The use of recruitment competitions and prizes for key milestones was also popular and maintained the engagement of research nurses/co-ordinators who were key to ongoing patient identification. Our experience supports the use of multiple strategies to support sites and maintain buy-in, which we believe especially important in critical care trials, which are challenging at multiple levels.

Our experience confirms the challenges found in other critical care trials, which included long-term follow-up for economic evaluation by obtaining questionnaire-based follow-up for up to 12 months. We achieved 61% and 58% completed follow-up at 6 and 12 months, respectively, for questionnaires for surviving patients, which equated to 75% and 74% complete follow-up when death was included as an outcome, respectively. This compared favourably with other recent UK-based critical care trials in which follow-up rates of 30–40% were typical. After a single postal questionnaire, completion rates were only 48% and 49% at 6 and 12 months, respectively. We included a gift token with these questionnaires, which may have improved responses, but this is difficult to assess. The use of a second questionnaire request (at 12 months only) followed by attempted telephone contact improved response rates by 18–20%. Given the increasing interest in longer-term HRQoL outcomes following critical illness, the recognition that these are significantly reduced in many patients, and of their impact on QALYs and cost-effectiveness analysis, obtaining high follow-up rates are important. Our data suggest that including a multifaceted strategy that includes both postal and telephone-based methods, and costing the resources to undertake this, is important in future research. Our embedded audit suggesting an average of 30 minutes per patient follow-up is useful for future grant-costing in this population.

Clinical results

The results of the international ABLE trial were published in March 2015, before the UK ABLE trial follow-up was completed. 31 No significant effects (beneficial or harmful) were found from the use of fresh versus standard-aged RBCs on mortality at 90 days, or any of the predefined secondary outcomes. These included the duration of respiratory, cardiovascular and renal support, length of stay or any predefined AEs, especially transfusion reactions. There were also no differences in outcomes within any of the predefined subgroups, including medical, surgical and trauma admissions. Importantly, the RBC transfusion triggers used were similar to those recommended in current guidelines11,66,67 and a large separation in RBC age was achieved between the groups (mean 6.1 vs. 22.0 days). The standard-aged RBC group was similar to the RBC storage age profile that is used in the NHS. A high rate of compliance with group allocation was achieved, and the findings of sensitivity analyses that included only patients who received RBC transfusions and those that included patients who received only RBCs with storage age according to group allocation were almost identical. The study population had a very high illness severity based on their illness severity score, the degree of organ failure at baseline and during the intervention, and the high mortality rate of 33% at 90 days. This observed control group mortality rate was higher than the 25% on which the sample size was calculated for the international trial. However, the aim was to include patients with a mortality rate that was at least 25%, such that the trial achieved recruitment of the target patient group. The mean length of ICU and hospital stay of 14 days and 32 days, respectively, highlights the extremely high resource use associated with these patients. The 95% CIs around the effect observed in the international trial ranged from 2.1% lower mortality with fresher RBCs to 5.5% lower mortality with standard-aged RBCs. These data provided compelling evidence that fresher RBCs conveyed no clinical benefit to the study population included in the trial, especially as the non-significant direction of effect favoured the standard-aged blood group.

In this report, we analysed the UK cohort according to the predefined analysis plan, and also compared findings with the international cohort. This analysis was underpowered for the clinical outcomes and was undertaken in order to assess if any major differences were present in the UK population in terms of baseline characteristics, transfusion practice, study conduct or clinical outcomes. Our data show that the critically ill patients included in the UK ABLE trial were very similar to those included in the international trial. The only possible exception was a greater representation of surgical patients and those undergoing elective procedures, but these differences were small and unlikely to be clinically relevant. The clinical outcomes were very similar for both the primary and all secondary outcomes. In addition, the process measures, especially the transfusion practice based on the mean pre-transfusion haemoglobin concentration, were very similar.

In summary, the ABLE trial found no clinically or statistically important benefit from transfusing exclusively RBCs stored for ≤ 7 days compared with the current standard practices used in blood banks. There were no important differences between the UK cohort and the international cohort in patient characteristics or outcomes.

Strengths and weaknesses of the clinical trial

The strengths of the international ABLE trial included the sufficiently large sample size to detect a clinically important difference in mortality, and a range of relevant secondary outcome measures. The CIs around the estimations of effect make an important beneficial effect from using fresh RBCs compared with the current standard very unlikely in the population studied. This population included a heterogeneous spectrum of critically ill patients whose baseline characteristics were typical of the sickest patients admitted to ICUs in the UK and other countries. The international participation increased the external generalisability of the trial results. The design minimised the chance of bias through allocation concealment, blinding and low loss to follow-up for the outcomes at 90 days. There was a high rate of protocol compliance, which achieved excellent separation of RBC storage age. The pre-planned sensitivity analyses also all showed similar effects to the main analysis.

The trial has potential weaknesses. The recruited population was heterogeneous, which means that some subgroups in whom different effects may occur could have been under-represented. The enrolment criteria meant that previously transfused patients and those in whom transfusion could not be delayed were not included. This probably explains the relatively low representation of massively transfused patients, trauma patients and patients with primary bleeding diagnoses, such as gastrointestinal bleeding. The embedded audit undertaken during the UK trial showed that 46% of all ICU admissions received RBCs during their hospitalisation, mostly before and during ICU care. Twenty-three per cent of patients received RBCs before ICU admission, and were excluded from the trial for this reason. We also found that only 28% of all transfused patients were potentially eligible for the trial, and only 22% of these were enrolled. By far the most common reason for non-inclusion was transfusion at night or outside usual working hours. These data raise the possibility that the patients not enrolled based on inclusion criteria and/or logistic reasons were systematically different from the trial cohort. As these patients may have been over-represented by diagnoses or complications that mandated more rapid or larger RBC transfusions, this might limit the generalisability to all critically ill patients. The transfusion practice was close to current evidence-based guidelines, which suggest that transfusion triggers close to 70 g/l for most patients, especially in the absence of cardiovascular disease. However, there was clearly some variation between clinicians and centres in the use of RBCs. Although this was inevitable in a pragmatic trial in which this aspect of care was not controlled, it could potentially create some ‘noise’ in the effects observed. Any variation in transfusion decision-making was balanced between the groups, so the impact of this potential confounder was unlikely to be important.

Importantly, the ABLE trial did not test the hypothesis that fresh RBCs were superior to ‘old’ RBCs with a storage duration close to the current maximum storage times of 35–42 days. It is possible that the additional effects of storage during this period could alter the effectiveness of the blood product. The conclusions of the ABLE trial are therefore limited to the comparison of the current typical storage age with exclusively fresher RBCs.

Economic evaluation

We undertook a pre-planned series of analyses that were specific to the UK ABLE trial, and also of potentially more generic value to critical care economic evaluations. These comprised a cost–utility analysis of the value of fresh RBCs compared with standard-aged RBCs, a nested study comparing the EQ-5D-3L and EQ-5D-5L utility scores in critical care survivors and an exploration of factors associated with health-care resource use and QALYs following critical illness. All analyses used the data collected from the UK ABLE trial data set.

Results in context of other research

Several systematic reviews have been published during the conduct of the ABLE trial. A meta-analysis by Wang et al. 69 in 2012 of 18 observational studies and three RCTs, predominantly in cardiac surgery and trauma patients, concluded that older blood was associated with an increased risk of death. A larger meta-analysis by Lelubre and Vincent in 201370 included 55 studies, of which only eight were RCTs. This showed no superiority of fresher RBCs over a range of morbidity outcomes (infection, bleeding, duration of MV, multiorgan dysfunction, length of stay) and mortality. Authors commented on significant confounding factors in interpreting these analyses, namely retrospective designs, heterogeneous populations, small studies and the exposed volume of RBCs. Sicker patients with comorbidities tend to receive more RBC units, such that inferring causality of adverse effects to RBC storage duration is challenging, as evidenced by a meta-analysis, which showed positive results for studies that did not adjust for this compared with those that did.

A Cochrane review of RCTs assessed the effects of age of stored RBCs in people requiring transfusion. 71 Only RCTs were included, with participants of any age (from neonates to adults) requiring transfusion for anaemia of any aetiology. The primary outcome was mortality within 7 days and 30 days after transfusion, with longer-term mortality as a secondary outcome. In total, 16 RCTs were eligible for inclusion in the review with a total of 1864 participants of all ages; five RCTs on neonates, one on paediatric patients and 10 on adults. All included studies followed similar design and analysis strategies; there were two randomised groups of participants, one receiving ‘fresher’ blood than the other, implying dichotomous outcomes, which were examined based on risk ratios when comparing the two groups. There were five trials dealing with transfusion in low-birthweight neonates, with the rest of the trials dealing with anaemia (including malarial anaemia in a paediatric population), cardiac surgery and treatment for critical illness, trauma and the ICU. There were only 2 out of the 16 studies that assigned > 100 patients, with the two larger trials assigning 377 and 910 patients. Five feasibility trials were also included as precursors to larger trials, including the two trials that underpinned the ABLE trial. 9,10 There were differences recorded between the included studies in the additive solutions used to preserve RBCs and whether or not RBCs were leuco-reduced and/or irradiated, and four studies did not report on this information. There were differences between the protocol definitions and the actual age of stored blood used, and many studies reported significant protocol violations, often apparently because of daily fluctuations of stored blood availability. The lack of agreed definitions of the ‘age of blood’ and, as a consequence, the variability of the recorded cut-off points between the included studies were major impediments in exploring the relationship between the length of time of stored blood and the risk of transfusion in this and other reviews. The multiple transfusions required by many patients and the inconsistent reporting of data details regarding the actual age of transfused blood served as compounding factors for the methodological challenges encountered. The studies in the review reported in total three groupings for comparison of the RBC product: ‘fresher’, ‘standard practice’ and ‘actively older’ blood. Eight trials compared transfusion of ‘fresher’ with ‘actively older’ blood, with the remaining eight comparing ‘fresher’ with ‘standard practice’ blood. The definitions of ‘fresh’, ‘standard’ and ‘older’ were not consistent among studies. In many trials, the ranges overlapped such that the distinction between trial groups was small and of questionable relevance. No conclusions could be drawn from the available evidence from RCTs that preceded the ABLE trial.

In summary, these reviews highlight the lack of high-quality evidence prior to the completion and publication of the ABLE trial, and the uncertainty regarding the clinical importance of RBC storage age for clinical outcomes.

An important concurrent RCT to the ABLE trial was RECESS,72 which was a multicentre RCT conducted between 2010 and 2014. Participants were aged ≥ 12 years and undergoing complex cardiac surgery, and likely to require RBC transfusion. Patients were randomly assigned to receive leucocyte-reduced RBCs stored for ≤ 10 days (shorter-term storage group) or for ≥ 21 days (longer-term storage group) for all intraoperative and postoperative transfusions. The primary outcome was the change in MODS score (range of 0–24 points, with higher scores indicating more severe organ dysfunction) from the pre-operative score to the highest composite score through day 7 or the time of death or discharge. The median storage time of RBCs provided to the 1098 participants who received RBC transfusion was 7 days in the shorter-term storage group and 28 days in the longer-term storage group. The mean change in MODS score was an increase of 8.5 points and 8.7 points, respectively (95% CI for the difference: −0.6 to 0.3 points; p = 0.44). The 7-day mortality was 2.8% in the shorter-term storage group and 2.0% in the longer-term storage group (p = 0.43); 28-day mortality was 4.4% and 5.3%, respectively (p = 0.57). AEs did not differ significantly between groups, except that hyperbilirubinaemia was more common in the longer-term storage group. This trial found no evidence of a clinically or statistically important difference in development of organ dysfunction or any of the other secondary outcome measures, associated with the storage age of RBCs. Importantly, RECESS provides similar findings to the ABLE trial in groups at a much lower risk of death, and compared a fresher RBC storage age group with a group allocated to receive RBCs that were, on average, 7 days older than the standard age of RBCs used in the ABLE trial control group.

Another relevant trial, INFORM (INforming Fresh versus Old Red cell Management; ISRCTN08118744),73 was published since the main publication describing the ABLE trial. This was a hospital-wide study randomising any patient requiring a RBC transfusion to receive the freshest versus the oldest available RBCs in the blood bank during their hospital stay in a 1 : 2 ratio. 18 The trial was undertaken in six hospitals and included 20,858 patients with type A or O blood. Of these patients, 6936 were assigned to the short-term storage group and 13,922 to the long-term storage group. The primary outcome was in-hospital mortality, which was estimated by means of a logistic regression model after adjustment for study centre and patient blood type. The mean storage duration was 13.0 days in the short-term storage group and 23.6 days in the long-term storage group (achieving the pre-planned mean separation of 10 days). There were 634 deaths (9.1%) in the short-term storage group and 1213 (8.7%) in the long-term storage group (odds ratio 1.05, 95% CI 0.95 to 1.16; p = 0.34). The results were consistent in three prespecified high-risk subgroups (patients undergoing cardiovascular surgery, those admitted to intensive care and those with cancer) indicating consistent lack of differences in three contrasting groups at different risk of short- and long-term mortality. Importantly, the effects seen in critically ill patients were similar to those found in the ABLE trial, in whom a similar ‘usual-care’ comparator storage age was compared with an overall ‘fresher’ storage age. The risk of death was lower in the INFORM trial ICU population (older storage age, 12.8%; shorter storage age, 13.3%), but these results provide evidence that the ABLE trial findings are likely to be applicable to a large number of ICU patients, despite the narrower inclusion criteria used in the trial.

The TRANSFUSE trial74 (a multi-centre randomised double-blinded Phase III trial of the effect of standard issue red blood cell blood units on mortality compared to freshest available red blood cell units; ClinicalTrials.gov NCT01638416) is a large trial in critically ill patients (n = 5000), in which patients requiring RBC transfusion are randomised to receive either the freshest RBC units available in the blood bank at the time of request or standard-issue RBCs (similar in design to the INFORM trial). 73 This trial is ongoing and the results are not yet available at the time of writing.

Implications for practice

The ABLE trial provides evidence that there is no clinical or economic rationale for transfusing exclusively fresher RBCs to patients at high risk of death in the ICU, when RBC transfusion decisions are at the discretion of the clinician. Blood banks should not change their current practice of providing standard-aged RBCs to this patient group, assuming the blood product is typically aged, on average, 20–22 days, and is leuco-reduced. Taken together with RECESS72 and INFORM73 trial findings, the current evidence base does not support the routine or widespread use of fresher RBCs more than those provided as usual practice (with a mean storage age of 20–22 days).

The ABLE trial does not provide evidence that RBCs with an older storage age are as safe as standard-aged RBCs, especially those with a storage age close to the end of current storage times of 35–42 days. However, most blood banks and blood services actively manage blood stocks such that the proportion of RBCs transfused at the end of storage lifespan is relatively small.

Contributions of authors

Timothy S Walsh (Professor of Critical Care, University of Edinburgh; specialty: critical care and trial design/management) secured funding, contributed to the study design, led the UK trial and contributed to analysis and interpretation.

Simon Stanworth (Consultant Haematologist; specialty: transfusion medicine and trial design/management) secured funding, contributed to the study design, was haematology lead, contributed to trial management and contributed to analysis and interpretation.

Julia Boyd (Trial Manager, Edinburgh Clinical Trials Unit, University of Edinburgh; specialty: trial management) was trial manager and contributed to analysis and interpretation.

David Hope (Critical Care Research Co-ordinator, NHS Lothian; specialty: trial set-up and management) was research co-ordinator and contributed to trial management.

Sue Hemmatapour (Blood Bank Trial Co-ordinator, University of Oxford; specialty: transfusion scientist) was blood bank research co-ordinator and contributed to trial management.

Helen Burrows (Blood Bank Trial Co-ordinator, University of Oxford; specialty: transfusion scientist) was blood bank research co-ordinator and contributed to trial management.

Helen Campbell (Health Economist, University of Oxford; specialty: health economics) secured funding, designed the health economic evaluation and contributed to trial design and management.

Elena Pizzo (Senior Research Associate in Health Economics, University College London; specialty: health economics) contributed to the analysis and interpretation of the health economic evaluation.

Nicholas Swart (Research Associate in Health Economics, University College London; specialty: health economics) contributed to the analysis and interpretation of the health economic evaluation.

Stephen Morris (Professor of Health Economics, University College London; specialty: health economics) led and contributed to the analysis and interpretation of the health economic evaluation.

Publications

The international ABLE trial protocol was published in 2011:

Lacroix J, Hébert P, Fergusson D, Tinmouth A, Blajchman MA, Callum J, et al. The Age of Blood Evaluation (ABLE) randomised controlled trial: study design. Transfus Med Rev 2011;25:197–205.

The results of the international ABLE trial were published in 2015:

Lacroix J, Hébert PC, Fergusson DA, Tinmouth A, Cook DJ, Marshall JC, et al. Age of transfused blood in critically ill adults. N Engl J Med 2015;372:1410–18.

Data sharing statement

The international ABLE trial data are governed by data sharing agreements between the co-investigators. This states that any application to view or utilise the data should be made through the international TSC via the chairperson of the group and ABLE trial chief investigator, Jacques Lacroix.

For the UK health economic evaluation and longer-term follow-up for health-care costs and HRQoL, access to data can be facilitated through the UK chief investigator, Professor Timothy S Walsh, or the UK health economic lead investigator, Professor Stephen Morris.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.

Grant co-applicants

Professor Timothy S Walsh (University of Edinburgh; Chief Investigator), Professor Danny McAuley (Queens University, Belfast; Co-director of Research UK Intensive Care Society), Professor Gavin Perkins (University of Warwick; Co-director of Research UK Intensive Care Society), Dr Simon Stanworth (NHS Blood and Transplant; University of Oxford), Professor Gordon Murray (University of Edinburgh), Mr Douglas Watson (Scottish National Blood Transfusion Service), Professor Duncan Young (University of Oxford), Dr Helen Campbell (University of Oxford), Dr Duncan Wyncholl (St Thomas’ Hospital, London), Professor Rupert Pearse (Queens Mary’s University, London), Dr Jacques Lacroix (ABLE trial Chief Investigator; University of Montreal, QC, Canada), Dr Stephen Wright (Freeman Hospital, Newcastle), Dr Alan Tinmouth (Ottawa Hospital, ON, Canada) and Dr Dean Fergusson (Ottawa Hospital Research Institute, ON, Canada).

The Age of BLood Evaluation UK investigators

UK (English) © 1990 EuroQol Group EQ-5D™ is a trade mark of the EuroQol group

UK (English) © 2009 EuroQol Group EQ-5D™ is a trade mark of the EuroQol group