31th July 2014  RBC and Infections in ICU: A Systematic Review and Meta-analysis

Presented By Dr Tommaso Zoerle

We have discussed very few, if any, meta-analyses but the paper this week seems too important to ignore. It is a large meta-analysis of transfusion studies in critically ill patients. At first glance it does not obviously describe neurocritical care patients, even in a subset analysis, but it is an useful study and it is surprising it has not attracted more attention. This is perhaps because the clinical topography around transfusion is so confusing and impenetrable that we feel another meta-analysis of flawed studies will only further fog our thinking.

Transfusion seems to be a popular topic in the critical care literature at the moment which will hopefully clear some of this confusion. We looked at transfusion decisions in SAH patients at a recent journal club and discussed the impact of the TRICC study. We will also be examining a recent TBI study in a few weeks. Moreover, some large trials are on the way, such as EPO-TBI and TRANSFUSE, but none these are really addressing the question which crops up every day – at what Hb trigger should we transfuse critically ill patients with acute brain syndromes?

This week’s journal club looks at the association between health care associated infections (HCAI) and transfusion of packed red blood cells (PRBC). This is not a new idea, ventilator associated pneumonia has been associated with PRBC transfusion as have other infectious complications, for example this paper on blood stream infections, this on PICU patients, and this, more recently, about C. diff. Outside of ICU, this phenomenon has been studied widely, for example, the oncologists and transplant physicians have been interested in transfusion related immune modulation (TRIM) for a long time – transfusion increases recurrence of colorectal cancer but prolongs graft survival in kidney transplant.

TRIM is central to this debate, the immuno-modulatory effects of transfusion have been largely ascribed to leukocytes. Indeed, one of the reasons why TRICC is criticised is because it arguably does not apply to the modern use of leukocyte depleted blood which only became widespread after TRICC was published. However, immune suppression is not the only mechanism through which PRBC exert harmful effects. Alterations in microvascular flow has been implicated and appears to be leukocyte independent. It is a thorny area and some of the the potential mechanisms are discussed here. The editorial which accompanies this week’s paper develops this point.

This question is then further complicated by the evidence that the duration of storage of PBRC influences the incidence of complications, the so called “storage lesion”. See for example this review by Jean-Louis Vincent and this by Rinaldo Bellomo and colleagues in ANZICS. This review tries to tie together the biology of TRIM and and the clinical observation of storage lesion.

All of that aside, we need to work out when to give blood to anaemic non-bleeding critically ill patients with acute brain syndromes. Surprisingly, there is little data which demonstrates a benefit from transfusion, when important outcomes are measured. In contrast there are ample studies which indicate that transfusion can be harmful to neurocritical care patients, for example this retrospective study from Southern California which found that transfusion did not confer a mortality benefit but did increase septic complications. It is hard to draw useful conclusions, however, when mortality is used as the primary outcome.

Multimodal-monitoring has been used to try to shed some light on the problem. For example, this study of our patients in Cambridge perhaps showed that transfusion improved cerebral oxygenation, but it is not a clear signal, see the associated editorial (scroll to page 1166). This study indicates that there are responders and non-responders and that multimodal monitoring could help to identify those groups.

The data that does exist has been used to develop guidelines – some are linked in the background reading for the journal club on transfusion in SAH. In the UK the NHS Blood and Transplant transfusion guidelines recommend 8-10 g/dL for acute brain syndromes while the European guidelines on the management of bleeding following trauma include a good summary of the data supporting transfusion in TBI patients (pp 12-13) and propose a target Hb of 7-9 g/dL.

These guidelines are unsatisfactory, however, for such a complex and heterogenous patient group. We need to develop tools to individualise care. Multimodal monitoring possibly offers the potential to do this: Could we identify patients with impaired cerebral oxygen delivery where the benefit from transfusion outweighs the unambiguous harm? If so, does this mean that we would allow some patients to develop profound anaemia because of adequate oxygen delivery while other patients were transfused despite abundant haemoglobin?

This is our last journal club before the Summer break. We return in September.


The absence of clear data has led to a wealth of meta-analyses, editorials and reviews, some of which are excellent. For example a detailed review by David Zygun on anaemia in neurocritical care, and a nice editorial which helpfully puts the observation that anaemia is associated with worse outcome in TBI patients in the context of the potential harm from transfusion.

meta-analysis of six studies and 537 neurocritical patients in 2012 could not come to a conclusion, not to be deterred the same group are trying again.

This review in the BJA summarises the questions above and puts some physiological muscle behind the argument that anaemia may not be as harmful as we suspect in critically ill patients. Although, neurocritical care patients are underrepresented in the discussion

A review in Current Treatment Options in Neurology covers the background to this question in TBI perfectly and leads the authors to conclude that a Hb threshold of 90 mg/dL should targetted for the first few days after injury.

Review on transfusion in neurocritical care patients.

A good review from Australia on the molecular and pathophysiological basis of transfusion triggers which concludes, like everyone else, that a good trial is needed.