Peri-operative haemoglobin: an overview of current opinion regarding the acceptable level of haemoglobin in the peri-operative period

Introduction The peri-operative use of blood and blood products accounts for two-thirds of all units issued by the blood transfusion services [1]. It is clear from the recent European multi-centre study - Sanguis [2] - that there is a wide variability in peri-operative transfusion practice with regard to the level of haemoglobin (Hb) or haematocrit (Hct) at which transfusion is initiated - even when considering similar patients undergoing similar surgical procedures. Of more concern, in as many as 75% of cases, the indications for the transfusions are not recorded; this may represent casual record keeping or a reluctance to record an ill defined rationale for prescribing. Several reviews of transfusion practice [3-7] between 1965 and 1991 have demonstrated similar variability; when examined retrospectively a large proportion of transfusions were regarded as unnecessary. A recent survey [8] presented hypothetical cases to medical practitioners of varying experience, working in separate specialities in two unlinked medical centres. Once again this demonstrated marked variation in practice not only between the centres but also between specialists and, strikingly, junior doctors were far more likely to transfuse than their senior colleagues. This may reflect an erroneous reliance on 'transfusion triggers' or a failure to appreciate the risks of transfusion coupled with over concern about the risks of anaemia. Furthermore Sanguis [2] highlighted the fact that blood is ordered pre-operatively on the basis of the perceived requirement for a given procedure, rather than on the probability of an individual patient requiring blood (as indicated by age, pre-operative haematocrit, concomitant use of anticoagulants, etc). Thus 50% of patients presenting for cholecystectomy had blood ordered pre-operatively (type and screen or group and cross-match), while only 4% actually had blood transfused. These facts have significant implications given current concern regarding the transmission of infection, immunosuppression, and tumour recurrence associated with blood transfusion and the increasing difficulty and cost of maintaining blood transfusion services. Clearly there are two broad areas of concern: the efficient use of the blood transfusion services (requests for cross-match) and the appropriate use of red blood cells by clinicians. The concept of a 'transfusion trigger' was established in the 1940s by Adams and Lundy [9]. They recommended a minimum acceptable Hb level of 10 g dL−1 (corresponding to Hct 30%). This became widely accepted practice as the '10/30' rule. It has been repeatedly challenged in recent years as too conservative [10,11]. Is there an acceptable minimum for Hb in the perioperative period? Implicit in this is the concept of risk/benefit analysis in balancing the risks incurred as a consequence of anaemia against those arising as a result of transfusion. A rational approach involves establishing a risk profile: Is there an increased mortality/morbidity associated with anaemia? If so, is it related to the degree of anaemia? Are the risks alleviated by transfusion? What are the risks of transfusion? Is there an increased mortality/morbidity associated with anaemia? If so, is it related to the degree of anaemia? Anaemia may be simply defined as a red cell concentration corresponding to an Hb <13.0 g dL−1 (Hct 40%) in males and <11.5 g dL−1 (37%) in females. These values derived from population studies, describe the lower limit of the accepted normal range for that population. Clearly there are two significant variables: the nature of the population studied and the definition of 'normal range'. Furthermore comparing such a value with an individual's Hb may be of limited use; the significant factor being the deviation from the individual's norm not from the population norm. In clinical practice, the important consideration is the functional disturbance and morbidity associated with a given level of Hb rather than a somewhat arbitrary definition of normality. It is important to differentiate between acute and chronic anaemia. This discussion will consider anaemia arising as a consequence of acute blood loss and also chronic anaemia, without considering complex red cell defects, haemoglobinopathies, enzyme deficiencies or autoimmune disorders, which in themselves affect morbidity and mortality. Initial consideration will be given to anaemia as an acute event in the peri-operative period. There is a wealth of experience relating to intentional haemodilution: this technique involves the normovolaemic dilution of all the constituents of blood rather than just a reduction in red cell concentration; this is comparable to the situation intra-operatively in which blood loss is replaced by crystalloid or colloid solutions, or both. What are the physiological effects of haemodilution? Reduction in blood viscosity. Viscosity depends on red cell concentration and the interaction between the red blood cells and plasma proteins. As red cell concentration falls viscosity decreases [12,13]. Reduction in systemic vascular resistance (SVR). Is almost entirely because of the reduction in viscosity described above [14]. Increase in venous return (VR). Probably occurs as a result of an α adrenergic mediated increase in venous tone with movement of blood from the peripheral to central compartment [15,16]. Increase in cardiac output (CO). Occurs principally as a result of an increase in stroke volume (SV) which in part is because of the increase in VR in accordance with Starling's law; if normovolaemia is maintained and the Hct is greater than 25% then increased heart rate is not a feature [17]. Maximal response depends on sympathetic activity which may be attenuated if for example the patient is given β adrenergic blocking agents [18]. Redistribution of flow. From areas of low to high metabolic demand which may cause preferential flow to both the heart and brain. This can lead to a critical reduction in oxygen delivery to other organs such as the kidney [19] during marked haemodilution/anaemia. Increased oxygen extraction. As the absolute amount of oxygen delivered falls demand may be satisfied by an increase in the extraction ratio. At rest this is about 20% but may increase four-fold with rising demand. Thus there is a considerable reserve, the exception being the myocardium in which the extraction ratio is between 50-70% at rest. Release of oxygen in the tissues. May be enhanced by a right shift of the oxy-haemoglobin dissociation curve because of a rise in hydrogen ion concentration occurring as a result of a reduction in red cell buffering capacity. 2-3-Di-Phospho-Glycerate (2-3-DPG). In chronic anaemia oxygen release is further augmented by an increase in the level of 2-3-DPG with subsequent rightward shift of the oxy-haemoglobin dissociation curve. As a net result of all these changes there is an increase in flow as a result of improved flow dynamics and increased cardiac output together with an increase in oxygen extraction ratio such that tissue oxygenation can be maintained, although inevitably the physiological reserve is being diminished. The influence of haemodilution on systemic oxygen transport capacity remains controversial. The reduction in viscosity improves flow in resistance vessels, there is a rise in cardiac output and therefore in bulk flow. The effect in the microcirculation is contentious. Under normal conditions the Fahraeus-Lindqvist effect results in the viscosity in vessels with a diameter of less than 1.5 mm being lower than that in the general circulation. However, rouleaux formation and obstruction by single red blood cells at vessel branch points may effectively increase viscosity. These factors conflict and the resultant effect on flow is difficult to predict. Initial data suggested that maximal oxygen transport capacity would be achieved at an Hct of 30% [20] and did not fall below control values until an Hct of 20%. Although the latter value has been challenged by Lundsgaard-Hansen [21], who advocated a safe lower limit for the Hct of 35%, it has been supported by both clinical [22] and model analysis [23]. What are the limits of normovolaemic haemodilution? Global consideration of oxygen delivery indicates that oxygen consumption (V˙O2) at rest is approximately 140 mL min−1 m−2 whilst oxygen delivery (DO2) is approximately 550 mL min−1 m−2. As oxygen delivery falls V˙O2 becomes dependent on DO2, at which point adequate tissue oxygenation is threatened. The critical value at which this occurs is not known. In one case report [24] V˙O2/DO2 dependency was demonstrated at a DO2 of 184 mL min−1 m−2 in a patient anaesthetized and haemodiluted to an Hb of 4 g dL−1. This is considerably lower than the value of 330 mL min−1 m−2 quoted by Shibutani [25] in anaesthetized patients prior to cardiac surgery. The relevance of such a value is questionable because it may not reflect regional oxygen delivery and consumption. A study of the effect of haemodilution on the splanchnic circulation of anaesthetized pigs [26] showed that although hepatic oxygen delivery was maintained at a haematocrit of 14% there was a marked reduction in portal vein oxygen delivery with impaired tissue oxygenation. Of particular concern is the oxygen supply to the myocardium because it has an extraction ratio of 50-70% at rest. Therefore increasing oxygen supply is almost entirely dependent on increasing flow and therefore on coronary vasodilatation. Thus the presence of stenosis within the coronary vasculature, impaired diastolic relaxation (as a consequence of ischaemia) or myocardial oedema may impair coronary flow sufficiently to cause ischaemia. This problem has recently been addressed in a review by Kettler [27] in which he attempted to define a critical Hct based on functional coronary reserve at different levels of global and myocardial oxygen demand using a theoretical model. Considering myocardial demand during anaesthesia and at a level approximately four times this, as may occur in the post-operative period, he calculated that, in the presence of coronary stenosis, while an Hb of 9.2 g dL−1 was adequate to sustain oxygen delivery under anaesthesia, it would require an Hb of 12.8 g dL−1 when oxygen demand increased. Even in the patient with normal coronary vasculature supply/demand would be critically placed at an Hb of 7.0 g dL−1. In summary when red cell mass is reduced oxygen carrying capacity per unit volume is reduced but global oxygen delivery can be maintained by a combination of increased CO, inter organ flow redistribution and increased oxygen extraction. Fit and healthy individuals may tolerate extreme haemodilution using these compensatory mechanisms. However, this compensation is dependent on: Maintaining normovolaemia; Normal myocardial/ventricular function; An intact autonomic nervous system; Normal vascular beds. These ideal circumstances may not always be met in the post-operative period: The response may be impaired by limited left ventricular function, concomitant medications (β-blockers and α-blockers), residual effects of anaesthesia, and hypovolaemia. Oxygen delivery may be reduced by impaired respiratory function, hypothermia and alkalosis. Oxygen demands may be increased, particularly by shivering during recovery from anaesthesia [28]. In addition the increase in CO results in an increase in cardiac work (volume and rate dependent). It has been argued that this increase is small because of the decrease in afterload arising as a result of the decrease in SVR, but this is a dangerous assumption to make in the post-operative period. At this time afterload may be increased by vasoconstriction (secondary to pain, anxiety, hypothermia) and so further increase myocardial oxygen demand. Thus there are implications for standards of care both intra- and post-operatively if one accepts a lower Hb/Hct. These include: Maintenance of intravascular normovolaemia; Maintenance of adequate oxygenation; Avoidance of hypothermia; Access to rapid reliable estimation of Hb/Hct; Adequate pain control; Rapid and reliable access to blood/blood products when required; In patients with very low Hb levels or in those at particular risk the ability to assess oxygen delivery/consumption may be required. (This may include measurements of serum lactate, mixed venous oxygen saturation (SvO2) and V˙O2/DO2 - remembering of course that these are global measures and may not reflect specific organ function.) The use of on-line ST segment analysis and the development of devices for measuring the redox state of individual tissues (such as Near-Infra-Red-Spectroscopy) may be of benefit in monitoring the effects of anaemia [29]. What clinical evidence is available regarding the effect of anaemia on morbidity and mortality? The following information is derived from retrospective studies involving relatively small numbers of patients, combined reviews of such studies and individual case reports. Carson and Spence [30] reported a study of 125 patients who refused blood transfusion and demonstrated a significant relation between the degree of pre-operative anaemia and mortality (Hb<6 g dL−1=61.5%; Hb>10 gdL−1=7.1%). There was also an independent relation between the degree of blood loss and mortality (>2000 mL = 42.9%; <500 mL = 8%). No patient with a Hb>8 g dL−1 and blood loss <500 mL died. These relations were independent of other potentially confounding variables such as ischaemic heart disease, chronic obstructive airways disease, diabetes mellitus, etc. However, the significance of blood loss is not clearly separated from the extent/duration of surgery and underlying pathology which may independently have affected mortality. The same authors [31] demonstrated that surgery could be performed in patients whose Hb was as low as 6 g dL−1. This study again highlighted the association between operative blood loss and mortality (>500 mL = 8.4%; <500 mL = 0%). They concede that the relatively young age of their patient population (average age 47 years) may have had a beneficial effect, although this was not demonstrated in their initial study. Both of these studies showed a strong correlation between the degree of blood loss and mortality irrespective of the pre-operative Hb. It is possible that this reflects the degree of compensation required in response to an acute fall in Hb; thus not only is the absolute Hb important but also the acute change. A review of experience with Jehovah's Witnesses [32] showed that of 134 cases, who satisfied the criteria of moderate to severe anaemia pre-operatively and who were not transfused peri-operatively, 23 patients died from anaemia (as reported by original authors). In 20 of these the Hb was <5 g dL−1. Conversely 27 patients survived with Hb <5 g dL−1. There were no deaths when Hb was between 5 and 7 g dL−1. There were three deaths in the range Hb 7-8 g dL−1 all following major cardiac surgery. It was noted that of the survivors with a Hb <5 g dL−1 65% were aged less than of those who died were aged over The authors that this group was with regard to age and of these studies were performed under circumstances with regard to the intra- and post-operative monitoring and post-operative so that their results may not be a of clinical studies have that in high risk vascular surgery patients the of myocardial by with an Hct and an Hct a case report involving a patient presenting for vascular surgery demonstrated of pre-operatively in association with a Hct of which when the Hct was by transfusion to Is it possible to make regarding a minimum acceptable level of The discussion has demonstrated that it is possible for individuals to with low levels of Hb. mortality may not be until the level of Hb falls to <5 g dL−1 - for the fact that this may and increased levels of post-operative This is and a basis for discussion particularly with patients for or other transfusion. However, the not be can be achieved by but rather it is to in clinical The be on the of an adequate Hb during the peri-operative as to the quoted Hb. an approach consideration of the patients age, general particular to use of with the autonomic nervous or associated with increased oxygen demands and of blood or Clearly the for the volume of blood required and the time at which transfusion is initiated are by the Hb at the these in would the following as values against which to both practice and clinical without the in clinical for an individual In healthy individuals in the ability to be a Hb of g dL−1 is with or likely to impair their ability to require a Hb of g dL−1. In individuals with chronic anaemia compensatory are a Hb of g dL−1 is with evidence of an increased oxygen demand require a Hb of g dL−1. These are The clinical of the patient be in to as Are the risks associated with anaemia alleviated by transfusion? blood cells be with the of increasing oxygen carrying increasing oxygen delivery at a given cardiac The associated rise in Hct will result in an increase in blood viscosity and SVR, thus increasing myocardial work and oxygen demand. If transfusion is the combination of volume may result in left ventricular and further oxygenation with the risk of myocardial Clearly transfusion is The be to the to oxygen In many clinical of - pain - is a to the of oxygen However, this is not always the Of particular concern are those patients with ischaemic heart in myocardial is to occur in the peri-operative and has been to be associated with a low Hb What are the risks associated with transfusion of Clearly the major risks - are those associated with transfusion which from in patient arising from over transfusion (volume is also to be of from blood transfusion are low clear are the risks arising from the association between blood increased to and the risk of tumour There is evidence that results in a of and that the transfusion of blood may this It would that this depends on the presence of in the transfused blood this the for an increased of post-operative and an impaired ability to cells in an increased tumour recurrence This was by and and a large evidence to this this has been in a retrospective analysis of patients undergoing surgery for Although both showed an increase in tumour recurrence associated with blood transfusion could a cause and effect relation between the two and it may be that other independent variables associated with blood transfusion are A study involving patients with of the and which although with the that transfusion against again could not Furthermore there may be no in recurrence rate of or blood is The demonstrated a point of increase in the of a If one were to that the was because of the of which is clearly this would represent an However, this is not supported by the available therefore of risk is of is also regarding blood transfusions and increased post-operative clinical studies have demonstrated a this has been challenged by A recent review the above that to blood transfusion with an increased rate of post-operative has not been There is evidence only of an association not a and In the of this information is it possible to a risk benefit of a young is different from that of an for a undergoing surgical of a will tolerate a low Hb and can a rate of about with a recurrence rate of about In a with ischaemic heart undergoing a similar is at greater risk of as a result of anaemia and although the risks of transfusion are the same the of those risks are This result in a different for transfusion in the two It be argued that this is no more than it is at an analysis in very broad as has been demonstrated in the discussion there is a of information on which to make this There are no the influence of levels of anaemia on There is regarding the effect of transfusion on tumour recurrence and There is reliable information regarding the of both the transmission of agents and based However, recent experience with that with regard to the of further agents be clinical practice be improved using the information the problem of was However, it would that in areas (such as transfusion is It is clear therefore that The to transfuse red blood cells be a clinical the risks of a low Hb peri-operatively, the risks associated with transfusion and the of the one could risks and make a for At there is information to with risk analysis in other areas of medical practice cardiac risk and that when there are variables the may at be to at risk and a degree of risk to that It would be to on 'transfusion this may which is the of clinical relating to the use of blood in the peri-operative have been These not to be widely this is because of reluctance to such concern over their failure to or simply reluctance to practice is that is more likely if are based on evidence would regard this as a requirement [1]. At it is not possible to these demands with regard to blood transfusions in the peri-operative period. However, the broad approach by and which a value to the evidence level that derived from control to level that derived from case with no control is possible and be if is to be in this of of and etc). If transfusion is to be reduced then of the in those is widely accepted would a basis for In addition although are limited in ability to assess the risks and of the concept of risk benefit analysis is an important one in that it the of a clinical and to that are can be large studies at the relation between anaemia and and transfusion and In the further review of the information available is required. This review was for the of of the of would to and Spence and of for their in the of this