The transfusion threshold is the haemoglobin value at which transfusion will normally be indicated, under stable conditions and in the absence of other clinical signs or symptoms of anaemia. Transfusion should be limited to the smallest amount of blood required to lift the patient above the transfusion threshold. Each hospital laboratory has its own definition of anaemia, based on the normal range for the local population.

A transfusion threshold should be defined as part of an overall strategy to provide optimal patient management.

The transfusion threshold should be viewed as the haemoglobin value below which the patient should not fall during the perioperative period, particularly in the context of ongoing or anticipated blood loss.

3.1 Preoperative

Preoperative anaemia increases the likelihood of allogeneic transfusion⁴⁹ and should be investigated and, where possible, corrected prior to major elective surgery (in this context major surgery refers to procedures for which blood is routinely grouped preoperatively). However, there is limited evidence available on appropriate preoperative haemoglobin concentrations. When a patient refuses a blood transfusion, preoperative haemoglobin is an important determinant of operative outcome, particularly in patients with ischaemic heart disease.^49,50,51 Evidence level 2+

All patients undergoing major elective surgery should have a full blood count performed prior to surgery to avoid short-term cancellation and to allow those patients presenting with anaemia to be investigated and treated appropriately (e.g. iron therapy).

Where possible, anaemia should be corrected prior to major surgery to reduce exposure to allogeneic transfusion.

3.2 Intraoperative

When there is ongoing surgical blood loss, haemoglobin measurements should be interpreted in the context of a multifaceted clinical assessment, which should include clinical evaluation of blood volume status. There is no indication that thresholds should differ during this period, but the use of intraoperative transfusion must reflect the ongoing rate of surgical blood loss, continued haemodynamic instability, and anticipated postoperative bleeding.⁵² Evidence level 3

Accurate measurement of intraoperative blood loss is difficult, although during cardiopulmonary bypass (CPB) frequent haematocrit evaluations are available. Two large prospective observational studies of patients undergoing CPB for primary coronary artery bypass graft (CABG) showed that postoperative mortality and severe ventricular dysfunction were related to low haematocrit during bypass. Though both studies showed increased risk when the haematocrit fell below 0.17, there was no agreement about the safe critical haematocrit value that indicated the need for transfusion.^53,54 Evidence level 2+

Rapid intraoperative measurement of haemoglobin levels using near patient testing may improve safety margins and avoid unnecessary transfusion.^55,56 Prospective assessment of the impact of these new techniques during the intraoperative and immediate postoperative periods is urgently required.

3.3 Postoperative

Three systematic reviews,^52,57,58 five randomised controlled trials,^{59,60,61,62,63} seven cohort studies^{45,49,50,51,64,65,66} and seven consensus statements^{67,68,69,70,71,72,73} were judged to be of an appropriate standard for inclusion in the evidence base for the guideline. All trials and studies were performed using non leucodepleted red cells and differences in haemoglobin and/or transfusion thresholds were described in relation to the intraoperative and postoperative periods only. However, no trial or study has examined transfusion thresholds in patients with chronic disease undergoing elective surgery so it has not been possible to make evidence-based recommendations for this group of patients.

Over the last 12 years, guidelines and consensus statements have consistently expressed the transfusion threshold as a range, usually between 70 and 100 g/l haemoglobin, with clinical indicators further defining the need for allogeneic transfusion in between.^67,69,73 Evidence level 4

Spiess et al⁶⁴ found a statistically significant increase in postoperative myocardial infaction (MI) in CABG patients whose haematocrit was greater than 0.33 on the first postoperative day. However, this was not confirmed by a retrospective assessment of a similar postoperative CABG population in Canada, despite the fact that both studies had a similar overall mortality and postoperative MI rate.⁷⁴ No evidence was found to suggest that cardiovascular function is improved at haemoglobin values >100 g/l. Evidence level 2+

Transfusion is unjustified at haemoglobin levels >100 g/l.

Only limited experimental data and expert opinion were identified on which to base a recommendation on the lower limit of haemoglobin below which transfusion should take place. Experimental data from healthy animals indicates that electrocardiogram (ECG) changes of myocardial ischaemia appear at haemoglobin levels below 50 g/l.⁷⁵ Dogs with experimental stenoses of their coronary artery circulation developed ECG and functional changes at Hb 70 g/l.⁷⁶ During normovolaemic haemodilution in healthy fit resting adults it has been shown that adequate delivery of oxygen was sustained down to a haemoglobin of 50 g/l.⁷⁷

A review of consensus statements⁵⁸ supported a lower limit of 70 g/l and also suggested that patients with cardiovascular problems should have this limit raised to 80 g/l. A large retrospective study of surgical patients confirmed that, allowing for confounding factors, there was no difference in mortality using a lower threshold of either 80 or 100 g/l.⁴⁵ No conclusions could be drawn regarding a lower threshold, as 90% of patients were transfused at Hb <80 g/l. Evidence level 4

Transfusion is required at haemoglobin levels <70g/l.

More evidence exists on which to base an upper limit for the transfusion range. The largest randomised controlled trial (RCT) of transfusion thresholds was performed in over 800 patients admitted to intensive care.⁶³ Patients were randomised to a conservative (70-90 g/l) or liberal (100-120 g/l) threshold and no difference in 30 or 60-day mortality was found. In addition, there was no significance difference in severe ventricular dysfunction, with the overall mortality in this population exceeding 20%. Evidence level 1+

Subgroup analysis indicated that patients under 55 years of age, or with less severe disease, had statistically better survival using the conservative policy, but clearly this requires care in interpretation. A large number of patients (598) were not entered in the study because of physician refusal. In addition, caution should be applied before extrapolating observations in patients in a critical care context to patients having routine surgery, as the patients' characteristics, patterns of morbidity and mortality and levels of physiological monitoring are all different.

In another study,⁶² 428 low risk CABG patients were randomised to a restrictive (<80 g/l) or liberal (>90 g/l) transfusion policy. No difference in mortality, postoperative MI, or significant ventricular complications was seen, nor was there any significant effect on patient rehabilitation. Although a statistically significant lower volume of red blood cells were transfused in the restrictive group, the percentage of patients receiving allogeneic blood was the same in both groups. Evidence level 1+

A small RCT in elderly patients with fractured neck of femur found no difference in either mortality or the achievement of mobilisation targets in patients transfused when symptomatic or with a Hb less than 80 g/l compared with patients whose haemoglobin was maintained above 100 g/l.⁶¹ Another small randomised trial in patients undergoing elective vascular reconstruction found no difference in mortality or morbidity when comparing a transfusion threshold of 90 g/l to 100 g/l.⁵⁹ This is a group of patients in whom the incidence of cardiovascular disease would be expected to be very high. However, both trials had inadequate analytical power to show significant differences in mortality/myocardial events. Evidence level 1-

A small observational study of similar patients found an increase in myocardial ischaemia and myocardial events in patients with a postoperative haemoglobin <90 g/l. An increased incidence of myocardial ischaemia was also detected in an observational cohort study of elderly patients undergoing radical prostatectomy when their haemoglobin fell below 90 g/l. Evidence level 2+

A further retrospective subgroup analysis of the original Transfusion Requirements in Critical Care (TRICC) study population⁶³ identified 357 patients who had a primary or secondary diagnosis of cardiovascular disease, or where cardiovascular disease represented a significant comorbid condition.⁷⁸ Despite having a significantly different mean haemoglobin compared to control patients (85 v 103 g/l), there was no difference in 30 or 60-day mortality, nor in ventricular dysfunction. As with the original study,⁶³ the authors felt that particular care should be exercised when patients had significant peripheral vascular disease, a recent MI or unstable angina. Evidence level 1+

Patients with cardiovascular disease, or those expected to have covert cardiovascular disease (e.g. elderly patients or those with peripheral vascular disease) are likely to benefit from transfusion when their haemoglobin level falls below 90 g/l.