Cardiovascular response to haemorrhage and hypovolaemia

This chapter answers Section G5(i) of the 2023 CICM Primary Syllabus, which expects the exam candidate to "explain the response of the circulation to ...hypovolaemia". Specifically, his chapter deals with haemorrhage, as that is the spiciest form of hypovolemia, and besides there was a previous syllabus item which asked specifically for this topic. There are of course physiological responses to haemorrhage which are non-circulatory, eg. the transcapillary migration of extravascular fluid, the changes in haematocrit, the release of haemopoietic factors, to say nothing of the clotting cascade, etc etc.  For what might appear to be completely arbitrary reasons, those non-circulatory factors are reviewed in another chapter, from the section dealing with body fluids. Here, the focus will be purely on the various ways the cardiovascular system reacts to being emptied.

This topic has appeared a couple of times in the past papers. Question 1 from the first paper of 2010 asked specifically for "the cardiovascular changes that occur following the loss of  1000ml of blood in an adult", an admirably precise question stem. Judging by the college comments, they actually wanted the trainees to only give cardiovascular changes in their answer. In the earlier Question 2 from the second paper of 2009, the more vaguely worded "describe the physiological consequences and responses after an acute haemorrhage of 2.0 litres" invited answers including responses which are non-cardiovascular (eg. secretion of cortisol) as well as changes which are not really "responses" per se, such as fluid shifts. Some might even say that the effect of losing 40% of your blood volume often closely resembles death. For those reasons, that SAQ was banished to the fluid and electrolyte section.

  • The loss of 1000ml of blood corresponds to 20% of the total circulating volume in a 70kg subject, which represents most of the stressed volume.
  • This results in autonomic and neurohormonal effects:
  • Autonomic effects
    • Arterial hypotension causes baroreflex activation.
    • Decreased cardiac output causes chemoreceptor activation.
    • Both reflexes result in autonomic phenomena:
      • Decreased vagal stimulus; thus increased heart rate
      • Sympathetic activation, which has multiple effects:
        • Increased peripheral vascular resistance
        • Redistribution of blood flow away from the cutaneous and splanchnic circulation
        • Stimulation of systemic catecholamine release from adrenal glands, which produces an increased systemic effect in addition to the peripheral sympathetic nervous system effects
        • Stimulation of vasopressin release via the projections from the nucleus of the solitary tract to the hypothalamus
        • Stimulation of renin release by sympathetic stimulation of the juxtaglomerular cells, and due to lower renal perfusion
  • Neurohormonal effects
    • Renin secretion causes:
      • Vasoconstriction (by angiotensin)
      • Increased sodium retention (by aldosterone)
    • Vasopressin release causes:
      • Vasoconstriction (by V1 receptors)
      • Increased water retention (by V2 receptors)
    • Venous hypotension decreases atrial natriuretic peptide secretion,
      which causes:
      • Decreased renal blood flow
      • Decreased urinary water and sodium excretion
    • The net effect is decreased urine output and increased retention of sodium and water
  • Effect of the rate of blood loss
    • A more rapid rate of blood loss places increased stress on the cardiovascular system to maintain haemodynamic homeostasis
    • Healthy individuals will be better able to compensate for more rapid rates of blood loss by increasing their heart rate and cardiac contractility
    • Patients with compromised cardiac function (eg. ischaemic heart disease or heart failure) will have impaired compensatory mechanisms and will not be able to compensate for even relatively slow blood loss

For peer-reviewed reading, The old article by Freeman (1963) is still excellent and hits all the right notes. The reader disturbed by the fact that sixty years have passed since its publication is reminded that the physiological responses to shock are unlikely to have changed over geological time scales. A more modern take, incorporating more modern terminology and some deliciously Guytonian element can be seen in Shen & Baker (2015)

Immediate effects of haemorrhage on blood pressure

Let's ignore the fact that everything happens simultaneously and observe the events following haemorrhage as if they happen in a series of steps. First, let us consider what happens to the pressure of the blood within the circulatory system, putting aside the effects of the heart and any sort of regulatory or compensatory mechanisms. Blood inside vessels is under a pressure which is exerted by its own weight, and by the walls of the vessels it is in. This is usually described as mean circulatory filling pressure (MCFP), and defined as "the pressure that would be measured at all points in the entire circulatory system if the heart were stopped suddenly and the blood were redistributed instantaneously in such a manner that all pressures were equal". This pressure decreases with volume loss. Here is an example of this happening in a famous dog experiment by Guyton (1955), where the circulating volume of the blood was being adjusted with all other factors remaining controlled. 

vascular function curves from Guyton (1955)

The amount of blood which needs to be withdrawn from the circulation for these changes to occur is actually less than you might think, mainly because the relationship between the volume of the circulation and the pressure inside it is not completely linear. Without recapitulating whole swaths of the mean systemic filling pressure chapter, it will suffice to say that only about 20% of the blood volume is thought to be putting pressure on the walls of the circulatory system, and the rest is described as "unstressed volume". 

Unstressed volume in haemorrhage

Anyway: as the volume in the circulation decreases, so the MCFP also falls, and the venous return is diminished. Cardiac output and venous return being essentially the same thing, you could probably leave the discussion at that. But let's assume that the 1000ml haemorrhage had occurred completely from the venous volume, and was so sudden that the left heart and the arterial side of the circulation are still unaffected. The decreased venous volume, in this preposterous scenario, would produce a lower MCFP and therefore a lower right atrial pressure. As the right atrial pressure is an important determinant of preload, this would produce a decrease in the cardiac output because the stroke volume would be lower, provided the heart rate remains the same. It is possible that at some stage someone will need to produce a Guytonian curve for the demonstration of this concept (as these tend to score marks):

changes to cardiac and vascular function curve during haemorrhage

So, that was an overly complicated scenic route, by which we have arrived at the conclusion that the cardiac output will decrease with haemorrhage. And if the cardiac output (i.e. flow of blood) decreases with unchanged peripheral vascular resistance, the pressure in the circuit would decrease, as pressure is the product of resistance and flow:

Pa- Pv = QR,


  • Pa- Pv = the pressure difference between the arterial and venous circulation
  • Q = blood flow, and
  • R = peripheral arterial resistance

In summary, in this completely unrealistic scenario where absolutely no compensation takes place, the cardiac output and blood pressure will decrease in a fairly predictable and linear fashion. This was well demonstrated by  Chien & Billig (1960), who destroyed the sympathetic fibres of eight mongrel dogs and then subjected them to arterial haemorrhage at a rate of 100ml/min. These data are presented here, concatenated onto a single coordinate scale and colourised for clarity:

change in cardiovascular parameters with haemorrhage in the absence of cardiac reflexes

Borrowing a set of circumstances from the college SAQ, the loss of 1000ml  of blood in a 70kg person would represent a 20% decrease in the circulating volume, as that person would ordinarily only be expected to have about 5L of blood in total. From the diagram above (extrapolating from dog data, but we are all mammals here), one would expect this to produce a drop in the stroke volume down to about 35% of the norm, which would halve the cardiac output and drop the MAP by about 30% (i.e down to a MAP of 60 or so).

Obviously, under normal circumstances, none of this would go undetected.

Immediate cardiac reflex response to haemorrhage

As the mechanisms of these cardiac reflexes are discussed elsewhere, in this section it will suffice to demonstrate what they do rather than how they do it. In evolutionary terms, our ancestors have a rich and storied history of haemorrhaging to near-death, and most of the cardiac reflexes have characteristics which look like they were selected by breeding trauma survivors. Well, one exception might be the Barcroft-Edholm vasovagal reflex, which looks like it's designed to kill you, but more on that later. In summary, the most important cardiovascular parameter changes in haemorrhagic shock, and the reflex responses they trigger, are:

  • Arterial hypotension causes baroreflex activation.
  • Decreased cardiac output causes chemoreceptor activation.
  • Venous hypotension decreases atrial natriuretic peptide secretion

The main objective of these rapidly acting reflexes is to: 

  • Increase heart rate
  • Increase cardiac contractility
  • Increase the peripheral vascular resistance (including venous resistance)

In addition to these (or following from them), several neurohormonal events take place

  • Stimulation of systemic catecholamine release from adrenal glands, which produces an increased systemic effect in addition to the peripheral sympathetic nervous system effect
  • Stimulation of vasopressin release via the projections from the nucleus of the solitary tract to the hypothalamus
  • Stimulation of renin release by sympathetic stimulation of the juxtaglomerular cells, and due to lower renal perfusion
  • Decreased atrial natriuretic peptide secretion

The net effect of all of these is decreased urine output and increased retention of sodium and water. Additionally, angiotensin and vasopressin also cause vasoconstriction, as if the systemic catecholamine release was not enough. 

How do we interpret these effects? Leaving aside the effects of heart rate, the adaptive mechanisms

  • Increase venous resistance, which changes the slope of the vascular function curve
  • Increase contractility, which increases the slope of the cardiac function curve
  • Decreased venous compliance, which decreases the size of the unstressed volume and alters the slope of the venous pressure-volume curve

These can also be expressed as a wholesomely Guytonian diagram, which you can see below. For even more detail, one may review Shen & Baker (2015).

unstressed volume and cardiac-vascular function curves in baroreceptor response to haemorrhage

So: this is a boring list of cardiac reflexes and neurohormonal regulatory mechanisms which is already covered in other parts of this site. What does this look like when you are actually looking at the measurable haemodynamic parameters? Generally, at this point in the discussion textbooks tend to flash a version of this diagram at the readers, which is never referenced but which seems to be derived from a rabbit study by Chalmers et al (1966). The original rabbit data and the gentrified textbook version are both presented here for unclear educational reasons. The most important thing to remember is the direction of change, which - to be honest- does not merit a diagram.

 cardiovascular effects of haemorrhage from Chalmers et al, 1966

Response to worsening haemorrhage 

So, initially the loss of volume causes an arterial baroreceptor response. The baroreceptors fire at a certain rate, constantly; as arterial pressure decreases, the firing rate also decreases, and this regulates the central descending control of the autonomic nervous system. Thus with blood loss there is an activation of a normal moment-to-moment control of blood pressure, which causes a reflexive decrease in vagal tone and increase in sympathetic tone. This is nothing extraordinary or pathological, and resembles the effects of standing up to full height from a supine position (which, incidentally, resembles haemorrhage in the sense that it redistributes blood away from the heart).  The heart rate rises and the systemic vascular resistance increases. This continues for a while. One can compensate for the loss of 10-15% of blood volume with this mechanism, and the cardiac output will not suffer very much.

cardiovascular response to 10-15 percent blood loss

During the Second World War the valiant efforts of blood donation collectors have yielded interesting facts. For these early pioneers, it was routine to collect anywhere up to 1000ml of blood from willing volunteers. During these harvesting operations, it was observed that such a volume of blood loss was associated with syncope in some of the donors. In honour of the most obviously famous authors to describe it in 1944, this phenomenon became known as the Barcroft-Edholm reflex, or the "depressor" reflex. It was thought to be a vagal reflex mediated by the sudden loss of right atrial pressure.

cardiovascular response to 15-30 percent blood loss

The diagram above paraphrases the data recorded in the original Barcroft paper from the Lancet (1944). This whole area is dealt with in extraordinary detail in the book entitled "Brain Control of Responses to Trauma". For the satanist who delights in graphs, this book offers much in the way of carefully recorded exsanguination, be it of human or animal blood, with or without ischaemic hind limbs, with neck suction applied, et cetera. 

We will not go into these wild woods. Suffice to say, the human body copes with loss of intravascular volume by a stepwise series of cardiovascular responses, which become more desperate as the blood loss continues. The trend in these cardiovascular responses has been chronicled in a porcine model. After 15-20% blood loss the "depressor" reflex kicks in, overriding the baroreceptor reflex. Bradycardia and vasodilation ensue. One can imagine that this sort of response had evolved in animals who expected to survive huge open wounds; decreased cardiac output in the context of uncontrolled bleeding in this situation is likely to prolong life because it decreases the amount of blood lost, and because a slow circulation promotes clot formation.

cardiovascular response to over 30 percent blood loss

At the end, when blood losses exceed 40% or so, one enters a stage of "irreversible" shock, with massive sympathetic response, maximal heart rate and increased systemic vascular resistance.

But, lets for a moment get back to the "depressor" reflex. Why don't we see this very often? Most of the time blood loss in trauma is accompanied by a steady increase in heart rate.

Well. It turns out tissue injury (for whatever reason) is accompanied by a suppression of the vagal depressor reflex, and a relatively "fixed" increase in the sympathetic activity. Combine this with fear, pain, the sensation of threat to one's life, and the picture becomes more clear.

the abolition of the Barcroft-Edholm reflex

Thus, the depressor reflex is suppressed in pretty much all frequently observed emergency-room scenarios. This might explain why it was first noticed not in trauma victims but in quietly exsanguinated volunteers- there was neither tissue injury nor fight-or-flight sympathetic stimulation.

Staging and classification of haemorrhagic shock

The dismemberment of observed phenomena into smaller subdivisions is a perfectly natural behaviour for the scientist, and haemorrhagic shock has been subjected to in on several occasions. The most famous of these is probably the ATLS classification, which divides haemorrhagic shock into four classes, creating a relationship between the volume lost and the clinical signs. Most authors recognise the papal authority of the ATLS and uncritically reproduce their table in their review articles. Some heretics raise objections that it seems to be divorced from reality. Others have created their own systems, which seem to overlap significantly. One example is provided here, which draws boundaries mainly according to the nature of the compensatory response:

Classification of Haemorrhagic Shock
Classes of Shock (ATLS, 1997)-
from Gutierrez et al (2004) 
Stages of shock
(Jacobsen et al, 1990)

Class 1:

  • Blood loss < 750ml, or <15%
  • HR < 100
  • BP = normal
  • RR 14-20
  • Urine output > 30ml/hr
  • No CNS symptoms

Stage I: compensation

  • Blood loss < 15%
  • Tachycardia
  • Normal BP

Class 2:

  • Blood loss 750ml-1500, 15-30%
  • HR 100-120
  • BP decreased
  • RR 20-30
  • Urine output 20-30ml/hr
  • Anxiety

Stage II: decompensation

  • Blood loss 15-30%
  • Bradycardia
  • BP decreased

Class 3:

  • Blood loss 1500-2000ml, or 30-40%
  • HR >120
  • BP Decreased
  • RR 30-40
  • Urine output 5-15ml/hr
  • Confused

Stage III: terminal sympathetic storm

  • Blood loss >30%
  • Extreme tachycardia
  • Severe hypotension

Class 4:

  • Blood loss >2000ml, 40%
  • HR >140
  • BP decreased
  • RR >35
  • Urine output negligible
  • Lethargic

The effect of bleeding rate on the response to haemorrhage

Without stretching their imagination, even a non-expert will intuitively agree that haemorrhage which is rapid will be more serious and poorly tolerated than one which is slow and controlled. However, an expert will insist on confirming this scientifically, and science demands the sacrifice of animals. Scully et al (2016) subjected eight Merino ewes to different rates of haemorrhage (one group losing approximately 50ml/min, the other 10ml/min) and observed their haemodynamic responses. Their data is reproduced below. The most interesting finding was that, in spite of different rates of haemorrhage, the blood pressure remained relatively stable. The heart rate increased more briskly where the rate of haemorrhage was high, 

effects of different rates of haemorrhage on the cardiovascular response.jpg

The x-axis in this study was unfortunately "percentage of blood volume removed which was required to achieve a drop in MAP of 30 mmHg", and this was quite variable between ewes (some required only 15% blood loss to achieve this drop, whereas others had to lose almost half of their blood volume). This makes the data more difficult to interpret. However, the bottom line is clear: the vagolytic compensatory mechanism has to be much more active in the case of rapid haemorrhage. This has implications for the trauma patient with pre-existing severe LAD stenosis. Not everybody has the cardiovascular wherewithal to compensate for rapid blood loss, and the point of decompensation may be reached well before the conventional ATLS volume thresholds are achieved. 

Cardiovascular response to hypovolaemia

Though to the layperson is likely to perceive some important differences between the loss of body water and the loss of blood, volume is just volume from the point of view of cardiovascular responses. Baroreceptor responses to the loss of water and the loss of blood can be expected to look very similar.  Well, to an extent. There are a couple of important aspects which need to be mentioned before this part of the CICM syllabus is sidelined as "everything same":

  • Chemoreceptor cardiac reflex is not activated. Recall that chemoreceptor cardiac reflex arcs are activated by hypoxia, and that reduced oxygen delivery will result from the loss of haemoglobin but not from the loss of water alone. However, in (for example) dehydration, the loss of body water will actually result in haemoconcentration and the oxygen-carrying capacity of the blood will actually be increased.
  • The vasopressin response will be precede, not follow, the baroreceptor response. Or at least, within the range of volume loss which can be described as nonthreatening the secretion of vasopressin would increase well before the baroreceptor response, in order to defend tonicity. As the body fluids become hypertonic, vasopressin is released to help reclaim free water.

The much more interesting topic of what happens to blood rheology and electrolytes is covered in the chapter about the physiological consequences of the loss of body water. And the cardiovascular adaptation to the isovolaemic loss of haemoglobin is something completely different again.


Kirkman, E., and S. Watts. "Haemodynamic changes in trauma." British journal of anaesthesia 113.2 (2014): 266-275.

Guyton, Arthur C. "Determination of cardiac output by equating venous return curves with cardiac response curves." Physiological reviews 35.1 (1955): 123-129.

Jacobsen, J., et al. "Cardiovascular and endocrine responses to haemorrhage in the pig." Acta physiologica scandinavica 138.2 (1990): 167-173.

Schadt, James C., and John Ludbrook. "Hemodynamic and neurohumoral responses to acute hypovolemia in conscious mammals." American Journal of Physiology-Heart and Circulatory Physiology 260.2 (1991): H305-H318.

Freeman, J. "Physiological effects of haemorrhage." Annals of the Royal College of Surgeons of England 33.3 (1963): 138.

B A Foex  Systemic responses to trauma British Medical Bulletin 1999, 55 (No 4) 726-743

Gutierrez, Guillermo, HDavid Reines, and Marian E. Wulf-Gutierrez. "Clinical review: hemorrhagic shock." Critical care 8.5 (2004): 1-9.

Chien, Shu, and Stuart Billig. "Effect of hemorrhage on cardiac output of sympathectomized dogs." American Journal of Physiology-Legacy Content 201.3 (1961): 475-479.

Evans, David, and H. W. Boyes. "The effects of rapid massive haemorrhage and retransfusion on various cardiovascular parameters in the dog." Canadian Anaesthetists’ Society Journal 16.5 (1969): 385-394.

Shen, Tao, and Keith Baker. "Venous return and clinical hemodynamics: how the body works during acute hemorrhage.Advances in physiology education 39.4 (2015): 267-271.

Mutschler, M., et al. "A critical reappraisal of the ATLS classification of hypovolaemic shock: does it really reflect clinical reality?." Resuscitation 84.3 (2013): 309-313.

Lawton, Luke D., et al. "The utility of Advanced Trauma Life Support (ATLS) clinical shock grading in assessment of trauma." Emergency Medicine Journal 31.5 (2014): 384-389.

Scully, Christopher G., et al. "Effect of hemorrhage rate on early hemodynamic responses in conscious sheep." Physiological Reports 4.7 (2016): e12739.