Anaemia in critical care

This chapter is annotations and footnotes for the anaemia entry in Section 2.1.11 (Haematological and Oncological Intensive Care) from the CICM Second Part General Exam Syllabus (First Edition).  The investigations and management of anaemia in ICU have been explored to a great depth in several past paper SAQs:

• Question 15.1 and Question 15.2 from the second paper of 2024
• Question 26.1 and Question 26.2 from the second paper of 2022
• Question 12.1 and Question 12.2 from the first paper of 2019
• Question 18.1 and Question 18.2 from the first paper of 2015
• Question 9 from the second paper of 2005
• Question 4 from the first paper of 2003.

Many of these have generally been more concerned with the interpretation of iron studies and morphological abnormalities of red blood cells. The future is likely to be different, as these easily marked data interpretation and data dump questions are becoming deprioritised in favour of SAQs that test a deeper analysis. 

In summary:

• Definition/classification: 
  • defined according to haemoglobin concentration, as mild (Hb <120/130 g/L), moderate (80-110) or severe (<80).
  • classified by the corpuscular haemoglobin content and RBC size, as well as by pathophysiology.
• Epidemiology: All patients become more anaemic while in the ICU, also:
  • more than 70% will be discharged with anaemia
  • weekly Hb drop is usually 10-30g/L
• Aetiology and pathophysiology:
  • best classified as dilution, increased loss (intravascular and extravascular) or decreased production.
  • loss through bleeding, haemolysis or sequestration 
  • decreased production due to lack of nutritional haematinic factors or loss of proliferative signal (eg. renal failure)
  • In ICU, much anaemia is due to frequent sampling
• Assessment:
  • Examination, looking for nutritional deficiencies, causes (malignancy, jaundce, bleeding) and consequences (weakness, heart failure, hyperdynamic circulation)
  • ​​​​​​​FBC, blood film, iron studies; plus also tests for haemolysis, renal function, and imaging to assess for sources of occult bleeding
• Management, apart from blood transfusion: 
  • Oral iron + haematinic replacement
  • Iron infusion (modest risks of fever/hyperphosphataemia)
  • EPO, supported by some (but not all) guidelines
• Complications:​​​​​​​ slow weaning from the ventilator, extubation failure,  more delirium, increased hospital stay, increased ICU mortality (perhaps only in COPD), exacerbated comorbidities. 

Probably the best single reference for this topic is this ancient article from the Mayo Clinic Proceedings (2003), as well as Newhall et al (2020) for a general physicianly perspective, and Docherty et al (2018) for an ICU focus and a discussion of tranfusion thresholds.  In addition to this generic anaemia chapter, one may also wish to look at the chapter on haemolytic anaemia broadly, and autoimmune haemolysis specifically: these disorders seem to come up rather frequently in the exam papers, more so than one might expect from their community prevalence.

Definition of anaemia

The WHO (2024) define anaemia. To save the reader having to download and read their 79 page document, they set the threshold at a haemoglobin value of 120g/L for women and 130 g/L for men. Their definition for "moderate" anaemia is anything from 80 to 109, and anything below 80 is "severe". The variability of definitions seen in the literature is not very large, which means that sticking with these for the Second Part Exam is reasonable.

Epidemiology of anaemia in ICU

This section can be crudely summarised as "100% of everyone gets the anaemia". This is the general gist of this widely cited study by Warner et al (2020), who found that among patients who came to hospital with normal haemoglobin values, the rate of incident anaemia at ICU admission was 44%, and 71% at ICU discharge. However, they used a haemoglobin value of 120g/L for women and 135 g/L for men, which, some may argue, is unreasonably generous for ICU patients (though meeting the official laboratory definition of anaemia).  Locally, Van der Laan et al (2021) from Fiona Stanley in Perth  used a threshold of 100g/L and found that about 45% of ICU discharges had a lower haemoglobin.  Czempik et al (2022), from Poland, observed that all patients in the ICU lost some haemoglobin, and the average rate of haemoglobin drop over the first week of ICU stay was about 12g/L for already anaemic patients and 28 g/L for the rest. Moreover, the longer you stay in ICU, the more haemoglobin you lose (Juárez-Vela et al, 022)

Aetiology and classification of anaemia

Together with the low haemoglobin result one usually also gets a whole panel of RBC volume and Hb content indices, which immediately allows one to classify the anaemia into groups. These groups are organised by corpuscular haemoglobin content and RBC size:

• Microcytic
  • Microcytic hypochromic, eg. iron deficiency
  • Microcytic normochromic, eg. thalassaemia
• Normocytic
  • Normocytic hypochromic, eg. renal failure
  • Normocytic normochromic, eg. acute blood loss
• Macrocytic
  • Macrocytic hypochromic, eg. malnutrition, B12 deficiency
  • Macrocytic normochromic, eg. liver disease

Alternatively, one could classify anaemia according to the pathophysiological mechanism, as in this article by Berlot et al (2014):

Dilution of RBC concentration

• dilutional anaemia with fluid replacement
• Artifactual anaemia due to sampling of a vessel with diluted content (i.e. the vein that has the fluids running through it).

Increased loss of RBCs

• Extravascular loss
  • Bleeding:
    • Traumatic blood loss
    • Upper or lower GI blood loss, eg. gastric erosion/ulceration
    • Iatrogenic loss through samling and surgery
• Intravascular loss
  • Intravascular haemolysis, eg. autoimmune hemolytic anaemia
  • DIC
  • Mechanical haemolysis, eg. due to extravascular perfusion circuits, dialysis filters, or mechanical heart valves
  • Appropriate reticuloendothelial sequestration of abnormal haemoglobin
    • Thalassaemias
    • Methaemoglobinaemia
    • Sickle cell anaemia

Decreased production of RBCs

• Decreased hematinics
  • B12/folate deficiency
  • Drugs which interfere with B12/folate metabolism
  • Iron deficiency
• Decreased proliferation signals
  • Decreased EPO in renal failure
  • Anaemia of chronic inflammatory states

Causes of anaemia are well covered by Hayter & Thomas (2021). In general papers on aetiology and pathophysiology of anaemia all seem to cover mostly the same ground, eg. this ancient paper from Greenburg (1996) is perfectly applicable even today.

Clinical course of anaemia in the ICU

One might be tempted to say "the course is, we transfuse you" but this would be inaccurate as most anaemic patients in the ICU remain anaemic. Consider that the haemoglobin threshold for transfusion  is 70g/L in the majority of cases. Patients may be discharged from the ICU with haemoglobin concentrations well in the anaemic range, and nobody will be correcting these until their discharge.

What happens after that? Yoon et al (2024) and Van der Laan et al (2021) explored the consequences. In summary:

• The haemoglobin is unlikely to recover back to normal by the time of hospital discharge
• This is completely unsurprising because all the factors that drive the development of the anaemia (inflammation, poor nutrition, depressed bone marrow function) persist after ICU discharge 
• The increased fatigue and exacerbated comorbid conditions that would come with this seem to keep the patient from being discharged, and anaemic ICU survivors seem to have a longer hospital stay.

Pathophysiology of anaemia of inflammation

This has come up twice in past papers, as Question 12.1 from the first paper of 2019 and the almost identical Question 18.1 from the first paper of 2015. In short, for 20% of the marks, the college wanted a brief explanation of how anaemia of inflammation causes the classical pattern of iron study abnormalities. A good article to answer this question is Nemeth & Ganz (2014). The mechanism of anaemia of inflammation can be summarised as follows:

• Decreased iron availability
  • Cytokine release in inflammation stimulates the synthesis of hepcidin, a regulatory molecule which controls the release of iron into the circulation.
  • A high hepcidin level decreases the availability of iron by promoting "iron trapping" within the bone marrow and macrophages. Thus, the iron stores in this form of anaemia are normal, which renders iron infusion pointless.
  • In the absence of circulating iron, erythropoiesis is restricted.
• Increased erythrocyte phagocytosis
  • Cytokine-activated macrophages destroy red cells at an increased rate
  • This reduces the lifespan of erythrocytes.
• Decreased erythropoiesis signals
  • Cytokines act directly on the bone marrow to reduce the rate of erythropoiesis, independent of the levels of circulating erythropoietin
  • Occasionally the level of erythropoietin is also suppressed, which is thought to be a cytokine-related effect acting on its renal secretion.

Assessment of anaemia

For those incapable of picking up their copy of Talley & O'Connor because of paralysing terror, Ingale et al (2017) and especially Alli et al (2017) are excellent for the clinical assessment and investigations.

History could of course be much more comprehensive, but...

• Rate of onset
• Transfusion history (eg. albumin)
• History of fluid resuscitation or trauma
• Dietary history
• Change in bowel habit
• drug history (eg anticoagulants, folate-wrechiong antiepileptics)
• Chronic disease (e.g. TB, HIV, cancer, autoimmune)
• Pregnancy
• Sequelae: lethargy, falls, palpitations, 

Clinical examination

• Pallor
• Nutritional deficiency features:
  • Angular stomatitis 
  • Glossitis in nutritional deficiencies
  • Koilonychia (spoon-shaped nails of iron deficiency)
  • Premature greying (megalblastic anaemia)
• Associated with cause:
  • Scleral ictus and jaundice (haemolysis)
  • "glove and stocking" neuropathy of pernicious anaemia
  • Splenomegaly, lymphadenopathy (malignancy)
  • Bleeding manifestations (petechiae, purpura, ecchymosis)
• Consequences:
  • Muscle weakness
  • Hyperdynamic circulation, flow murmur
  • Features of decompensated heart failure

Investigations for anaemia:

• Resampling to confirm the diagnosis
• Imaging to look for obvious blood loss
• Faecal occult blood test and/or endoscopy to exclude GI blood loss
• Direct Coombs test, formal blood film, conjugated/unconjugated bilirubin, LDH, haptoglobin and reticulocyte count to investigate haemolysis
• RBC folate and B12 levels
• Iron studies
• Coagulation screen (looking for DIC and MAHA)
• Renal function tests and/or EPO levels

This begs the question: if ferritin was such a good marker of iron stores, then how do we interpret it in states of inflammation? Virtually every patient coming to the ICU will have an acute phase response for one reason or another. Is it therefore impossible to make the diagnosis of iron deficiency in such patients? The logical answer would be yes, in the sense that the normal threshold for hypoferritinaemia (as per the WHO) is 15 μg/L, and most of these people would be higher than that.  

Does this imply that we should shift the threshold? Again, the WHO agrees; the normal ferritin level suggested by this body for anyone with a raised CRP is >70 μg/L. Having an inflammation-driven ferritin level higher than this does not preclude the possibility that you have iron deficiency anaemia, but it does make it less likely that you will benefit from iron supplements, which means that functionally speaking you are not iron "deficient" anyway (logically, there is no deficit if your iron stores are proportional to the iron demand).

Management options for anaemia in ICU

It is important to consider whether there is a genuine iron deficiency, or whether the inflammatory state has decreased access to the otherwise normal iron stores. Obviously, in a perfect world the iron-deficient patient would have some iron supplements, and the patient with inflammatory anaemia would have transfusions of blood and some sort of specific management aimed at the cause of inflammation.

However, in critical illness it is possible to have both issues simultaneously.

Iron infusion is covered very well by Geneen et al, 2022. It is difficult to say conclusively that any given critically ill patient will or will not benefit from an iron infusion. Obviously some patients trend to one group or another. For instance, trauma patients can be expected to develop iron deficiency associated with their blood loss. A study which focused on trauma patients (Pieracci et al, 2014) enrolled 150 patients, of whom the majority demonstrated iron studies consistent with iron deficiency. Some were then randomised to receive 100mg of iron sucrose. In them, serum ferritin increased (presumably reflecting that the iron stores were replenished), albeit not quite back to the normal range. Unfortunately, nothing else happened. Neither did the haemoglobin increase, nor did the red cell count, nor was the need for transfusion affected. Mortality and ICU stay were also unaffected.

The IRONMAN trial is mentioned in Question 18.2 from the first paper of 2015; it aimed to answer the question of whether IV iron administered to anaemic ICU patients reduces their need for blood transfusion. The hypothesis was that yes, it will (unless you are severely septic, erythropoiesis should be stimulated by the infusion of 500 mg of ferric carboxymaltose). The authors enrolled seventy patients into each arm. The main positive finding was that the iron-infused patients left hospital with an extra 7g/L of haemoglobin (107 g/L vs 100 g/L), but there was noo overall difference in transfusion requirements. In their answer to Question 12.2 from the first paper of 2019, the college also mentioned an increased rate of adverse events (including infection), which probably refers to the finding that the iron-infused group has an increased nosocomial infection rate (28.6% vs. 22.9%).

Erythropoietin supplementation is covered very well by Litton et al, 2019. Question 18.1 and Question 18.2 from the first paper of 2015 reported erythropoietin levels alongside conventional iron studies. This is odd, because it is not a part of the normal panel. The glorious RPCA Manual lists it as one of the tests "occasionally indicated" to discriminate primary from secondary erythrocytosis. Erythropoietin, however, is one of the potential treatments for anaemia. Recombinant human  EPO was used in this manner by  Corwin et al (2002), whose group was able to achieve a 19% decrease in RBC transfusions. Without transfusions, the treatment group still ended up with a higher haematocrit than the placebo group. This had no impact on mortality outcome or duration of ICU stay, but the meta-analysis by the ESICM transfusion committee did find an absolute mortality improvement of 3.1%, with a modest effect on transfusion (EPO-treated patients received 0.65 units less). 

This decrease in blood transfusion seems to come at the cost of a significantly increased risk of thrombotic events (according to a 2013 meta-analysis by Mesgarpour et al). That is not to mention the actual dollar cost of EPO, which is comparable with the cost of RBC transfusion (if not greater). In any case, EPO is off-label for anything other than the chronic anaemia of renal failure.

Blood transfusion is of course the last option. As such, the pros and cons of  blood transfusion in the ICU are discussed in greater detail elsewhere. The reader is also redirected to Arynov et al (2024), which is an excellent overview of non-Hb transfusion triggers.

Prevention of iatrogenic blood loss: ICU vampirism is discussed by Raurell‐Torredà et al (2024), who recommended that we all use small volume tubes and closed blood-sampling systems that do not require you to discard any volume. Remarkably, though multiple papers observed a reduced volume of blood being drawn as the result (eg. 8ml vs 40ml per day) none were able to associate this finding with an effect on the haemoglobin concentration.

Specific techniques in Jehovah's Witnesses are detailed by Davids et al (2024), who recommended the following:

• Maximise the dissolved oxygen fraction (i.e. high FiO2)
• Use inotropes to increase cardiac output
• Minimise oxygen demand (with sedation paralysis and cooling if necessary)
• EPO, iron folate and B12 (obviously)

Consequences of anaemia for the ICU patient

For something we tolerate down to a value of 70g/L, anaemia sure seems to have a whole host of complications in the critically ill. The compensatory physiological changes in response to anaemia are discussed elsewhere, but in summary:

• Increased respiratory workload, higher minute ventilation
• Increased cardiac demand and myocardial oxygen consumption
• Tachycardia

And, if you can't manage any of those because of poor cardiorespiratory reserve,

• Organ failure due to inadequate oxygen delivery

This translates into:

• Extubation failure (Khamiees et al, 2001; Hb < 100g/L; RR = 5.0) 
• Increased risk of Type 2 MI (Furie et al, 2019; Hb < 80g/L)
• Increased risk of delirium (Joosten et al, 2006; Hb < 120/130, OR = 2.4)
• Increased risk of death (doubled in COPD patients, according to Rasmussen et al, 2010, with a cutoff of 120g/L)

On the other hand, somehow transfusion is even worse? It may seem hard to reconcile the findings of these papers and their haemoglobin values of 100-120 with the trial data that supports transfusion thresholds no higher than 70, but then one remembers that these are trials, whereas the signal of worsening outcomes from anaemia comes from observational and often retrospective data. All we can derive from such studies is the knowledge that the same patients that end up anaemic also tend to end up dead, mad, or slow to wean.

Guidelines and statements for anaemia management

If we insist on a chapter structure where transfusion guidelines are a whole separate discussion, we end up with guidelines that are mostly diagnostic in their focus, or those related to iron infusion and the use of EPO. These would have to be something like:

• "Laboratory Diagnosis of Functional Iron Deficiency" from the British Society for Haematology, which are of limited value to the intensivist
• Van Doren et al (2024), expert consensus on the use of iron (from the US), which are also mostly about the practical administration of iron
• Rizzo et al (2010), the guidelines for the use of EPO in cancer-associated anaemia
• Lasocki et al (2020), French guidelines for management and prevention of anaemia in the critically ill, listed here because they are distinct in calling for the use of EPO in critically ill adults with a haemoglobin less than 100-120 (whereas Vlaar et al (2020), the European guidelines for transfusion, did not recommend EPO, on the grounds that it does not seem to help)