Restrictive and Liberal Transfusion Strategies in Intensive Care

Anaemia is a common complication of critical illness and 40-50% of all patients in intensive care units (ICUs) receive blood transfusions. However, despite the perceived benefits of correcting anaemia, numerous non-randomised cohort studies and a system review suggest associations between blood transfusions and adverse patient outcomes. The only large randomised controlled trial comparing restrictive with liberal transfusion policies in critically ill patients (the "TRICC study") suggested a trend to greater morality at 30 days in the liberal transfusion group. However the TRICC study does not inform clinical practice with regard to how to manage long stay ICU patients. Also, and importantly, the blood product used in the TRICC study contained white blood cells (non-leucodepleted), which are associated with adverse events, whereas blood in the UK now has the white blood cells removed prior to blood storage (leucodepleted). Due to continuing uncertainty many clinicians continue to use higher Hb transfusion triggers in longer-term ICU patients and we confirmed this recently in a survey of 184 ICU clinicians.

This uncertainty is important for patients, clinicians and the blood services. For example, patients who require longer stays in ICU have long recovery times, use significant hospital resource, and suffer symptoms such as fatigue, breathlessness and reduced Health Related Quality of Life (HRQoL) for many months. It is possible that blood transfusions could alleviate some of these problems. In addition, donated blood is in short supply and is increasingly expensive to produce. As 40% of ICU patients receive transfusions, and use 8% of the Scottish blood stocks, a stronger evidence base could optimise the use of this precious resource.

A Cochrane systematic review (co-authored by one of the applicants, BMcL) noted that current evidence was dominated by the TRICC trial, and concluded that the "....effects of conservative transfusion triggers on functional status, morbidity and mortality, particularly in patients with cardiac disease, need to be tested in further large clinical trials"4. Further studies are clearly required to determine the most effective transfusion practice in critically ill ICU patients.

We have completed an extensive programme of work to inform this study. Based on this data we have identified a patient population in whom an RCT will address many areas of clinical uncertainty. We will study longer-term ICU patients (higher illness severity; higher morbidity; worse long-term functional status; low long-term HRQoL; greater illness costs) who are aged ≥55 (high incidence of IHD; less certainty regarding safety with restrictive approach in TRICC trial). The treatment exposure will last longer than the TRICC trial, potentially including post-ICU care, because anaemia recovers slowly.

This study is a feasibility study comparing the effectiveness of a "restrictive" blood transfusion policy in long-stay ICU patients with a more "liberal" transfusion policy to inform the definitive RCT design. It has been designed to answer the following questions

• What proportion of eligible patients can be recruited to the study?
• What difference in mean Hb concentration and RBCs exposure will occur with our protocol?
• How good is compliance with our proposed study protocol?
• What factors (if any) limit recruitment and result in protocol violations?
• How should the current protocol be modified for a larger definitive trial?
• Are there any strong pro-inflammatory signals or other patient safety concerns associated with liberal RBCs use?
• How should the health economic evaluation be undertaken in the full trial?

This feasibility evaluation includes three elements:

1. A single blind randomised trial comparing patients managed with the restrictive versus the liberal strategy (main study).
2. A comparison of circulating plasma levels of inflammatory biomarkers during the first 24 hours of randomisation in patients managed with the liberal and restrictive strategy.
3. An interview-based qualitative study with clinicians managing patients in the study units.
4. Main Study. The overall hypothesis for this programme of work is that liberal use of RBCs to correct anaemia (HB transfusion trigger ≤90g/L; target Hb range 91-110 g/L) improves clinical outcomes compared with a restrictive transfusion trigger (Hb transfusion trigger ≤70g/L; target Hb range 71-90 g/L) in anaemic critically ill patients requiring prolonged ICU stay (≥4 days). Patients will be randomised to one of two transfusion strategies on a 1:1 basis Group 1 "Restrictive RBC Transfusion group": Patients will receive single unit RBC transfusions with a transfusion trigger of ≤70 g/L with a target Hb concentration of 71-90 g/L during the intervention period. Group 2 "Liberal RBC transfusion group": Patients will receive single unit RBC transfusions with a transfusion trigger of ≤90 g/L with a target of 91-110 g/L during intervention. These patients will all receive a transfusion on the day of randomisation. The intervention will last for 14 days from the time of randomisation or until discharge from the intensive care unit, whichever is longer. The aim is for all surviving patients to receive the randomised intervention for at least 14 days in all cases. Follow-up Complication rates will be ascertained by case note review during the 60 days following randomisation (hospital stay only). The length of their stay in ICU and in hospital will be ascertained from the patient records Longer Term follow up of Patients Patients will be contacted at 60 and 180 days following randomisation. At these time points the following will be determined or measured:

• Survival status
• Physical disability using the Rivermead Mobility Index
• HRQoL using the SF-12 questionnaire
• Health economic questionnaire to determine health care resource use (180 days only).

1. Biomarker Study. The aim of this part of the study is to explore whether a pro-inflammatory signal is associated with RBC transfusion in intensive care patients. A range of pro- and anti-inflammatory markers will be compared between the patients entering the restrictive group, who are not expected to receive RBCs during the 24 hours following randomisation, and the liberal group, who will all receive at least one RBC unit during the 24 hours following randomisation. Blood samples will be collected from patients as follows: Restrictive Group: A blood sample will be drawn following randomisation (time zero), at time zero plus 6 hours, and at time zero plus 24 hours. Liberal Group: A blood sample will be drawn following randomisation and immediately prior to RBC transfusion (time zero), at time zero plus 6 hours, and at time zero plus 24 hours Each blood sample will be 10mL volume (total 30mL per patient). Plasma samples will be batch analysed at the end of the study.
2. Qualitative Study.

The randomised controlled trial (RCT) is widely acknowledged as the "gold standard" in the evaluation of clinical interventions. Recruitment, however, is often problematic, with serious implications for the detection of important treatment differences, the generalisability of research findings to other patient populations and cost. A recent review demonstrated that less than a third of UK studies recruited the required sample size within the timeframe originally specified, and approximately one third required an extended recruitment period. The recruitment and retention of participants is particularly problematic in critical care research, the rationale for which is currently ill-defined.

Review of the existing literature on barriers to research participation suggests that a myriad of methodological/study-specific, organisational and clinician-specific issues exist. The purpose of this exploratory qualitative study, undertaken as part of the RELIEVE programme of work, is to identify and address perceived barriers to recruitment among clinicians prior to a larger definitive trial.

The aims of this study are

• To explore clinicians' experiences of clinical uncertainty in relation to existing guidance (as a default for clinical decision-making) and in relation to recruitment and compliance with the study protocol.
• To explore clinicians' knowledge of and perspectives on existing evidence and guidelines in relation to blood transfusion practice
• To explore clinicians' practice in relation to existing evidence and guidelines
• To explore clinicians' prior perspectives on research participation with particular reference to patient groups (e.g. patients with ischaemic heart disease) or clinical scenarios (e.g. bleeding events) which might elicit clinical uncertainty in relation to recruitment, randomisation or protocol compliance
• To explore clinicians' subsequent experiences of trial participation with reference to previously described concerns
• To develop potential strategies for the improved recruitment and retention of study participants.

Methods We will conduct qualitative semi-structured interviews with clinicians in each of the 6 participating centres. We will invite the Principal Investigator at each site to participate and, in order to minimise selection bias, we will invite a second clinician with a "non-vested interest" in the study to participate. Participants will be interviewed both prior to trial commencement and following trial completion.

Phase I: Prior to study commencement, we will explore (a) clinicians' knowledge and perspectives on existing evidence and guidelines in relation to blood transfusion practice (b) clinicians' practice in relation to existing evidence and guidelines (c) clinicians' perspectives on research participation per se and (d) clinicians' perspectives on the study protocol in relation to patient groups or clinical scenarios which are likely to elicit uncertainty in terms of recruitment, randomisation or protocol adherence.

Phase II: Following study completion, we will invite the same clinicians to describe their experiences and perceptions of trial participation, with particular reference to patient groups or clinical scenarios in which recruitment, randomisation or protocol adherence elicited concern or uncertainty. Using the screening logs, Case Report and Adverse Event Forms kept at each site by the dedicated research nurses, we will explore, where possible, the rationale for (a) clinician refusal of eligible patients and (b) non-compliance with study protocol or (c) withdrawal of recruited patients. We will invite clinicians to identify potential strategies for addressing these issues.

Analysis

The transcribed interviews will be examined and coded for relevant issues using Computer Assisted Qualitative Data Analysis Software (CAQDAS), namely NVIVO8®. Using thematic analysis, we will therefore identify and delineate key issues of concern, as expressed by clinicians, in relation to (a) the recruitment and retention of participants in critical care research (b) the recruitment and retention of eligible patients with reference to the RELIEVE study protocol.

Outcomes

The identification of "generic", protocol-specific and organisational barriers to recruitment will facilitate the identification of clinically and locally relevant strategies to address these issues prior to the larger definitive study. Using both the existing literature and the unique insights provided by this study, we will also develop and refine a screening tool of research barriers which will (a) be used to audit recruitment in the larger definitive study and (b) form the basis of a tool relevant for critical care research per se.