The oxyhaemoglobin dissociation curve

This chapter is most relevant to Section F7(i) from the 2017 CICM Primary Syllabus, which expects the exam candidates to be able to "explain the oxyhaemoglobin dissociation curve". In fact this exact phrase has appeared in Question 5 from the second paper of 2015. From the college examiner's remarks it is clear that a drawing of the curve was expected, with "appropriate key 
value points (p50, venous and arterial points)". The p50 value, positive cooperativity and the T / R states explaining the curve were all rewarded with extra marks. 
In summary:
  • The oxyhaemoglobin dissociation curve is a sigmoidal relationship between the partial pressure of oxygen and the oxygen saturation of haemoglobin
    • This curve describes the changing affinity of haemoglobin for oxygen which occurs with increasing PaO2
    • The flat upper plateau decreases variability in blood oxygen content even with large changes of PaO2
    • The steep lower part allows the increased release of oxygen from haemoglobin with only a small change in PaO2
  • Positive cooperativity is the main reason behind the non-linearity of this relationship:
    • Haemoglobin is a heterotetramer
    • It is composed of four subunits
    • Each subunit binds oxygen independently.
    • Then, once an oxygen molecule is bound to it, the oxygenated subunit increases the oxygen affinity of the three remaining subunits
    • This is because of a conformational change produced by each subunit binding oxygen, which mediates the transition from the T state to the R state
  • T and R state:
    • This refers to two distinct states of the haemoglobin tetramer molecule
    • The T ("Tense") state is the deoxygenated form (with 0 O2 molecules)
    • The R ("Relaxed") state is the oxygenated state (with 4 O2 molecules)
    • One pair of αβ subunits in the fully oxygenated R-state appears rotated by 15° with respect to the other pair of subunits
    • The binding of each oxygen molecule changes the state of the tetramer, changing the equlibrium contant for the next O2 molecule to bind the next subunit more easily
This is the sigmoidal relationship which describes the changing affinity of haemoglobin for oxygen under a variety of physiological conditions. This chapter tries to explore the fascinating behaviour of haemoglobin, and to discuss it in colourful toddler-like crayon diagrams. Of these diagrams, perhaps the most important is the sigmoidal curve itself, as it is often expected of the exam candidate to discuss its "left shift" and "right shift", a representation of changing the affinity of haemoglobin for oxygen under the influence of various changing physiological parameters. The p50 value as reported by the arterial blood gas analyser represents a mid-point in this curve, and gives us useful information regarding this changing affinity.

A brief history of the oxygen-haemoglobin dissociation curve

The original 1904 article describing the curve, translated for your reading pleasure, is available for free courtesy of the University of Delaware. It was written by Christian Bohr himself (of Bohr Effect fame), with co-authors Karl Albert Hasselbalch (yes, that Hasselbalch) and August Krogh who is best known for Krogh's Principle for the comparative method of biology. It is perhaps under his influence that the investigators chose to use dog and horse blood. Or, perhaps, in the civilized hat-and-coat days of 1904 the idea of harvesting vast quantities of human blood for research was somehow distasteful to these gentlemen of science. Certainly, the blood would have to have been fresh, as at this stage I do not know of any technique available to the investigators which would have kept it liquid for very long.

But let us digress no further. The investigators, armed with an apparatus of their of design, produced a graph of oxygen content of blood at different partial pressures of oxygen and carbon dioxide. This graph is reproduced below entirely unaltered, out of veneration to these founding fathers, and with no permission from anybody whatsoever.

Bohrs original diagram fo oxygen hemoglobin dissociation

One can appreciate from the briefest glimpse that these curves very closely resemble the modern representation of these relationships. Thought the concept of "oxygen saturation" was at this stage undeveloped, the authors did produce a table of their observations, which lists the "oxygen uptake in percent" at various partial pressures of O2 and CO2.

The relationship between haemoglobin saturation and the partial pressure of oxygen was expressed as an equation by J. W. Severinghaus in 1979 (Yes, that Severinghaus). In this paper, Severinghaus used his equation to calculate a table of values for a standard human blood O2 dissociation curve at 37°C and pH 7.40.

One would be surprised as to how difficult this table is to track down online! Everything one gets from Google search results is some bastardised dissociation curve picture with a few "values of interest" thrown over it, or something calculated at 38°C, or worse. In the interest of preserving the intellectual legacy of J.W. Severinghaus, and to make it accessible to the physiology enthusiast, I reproduce his data in the table below.

Oxygen Saturation vs Partial Pressure
PaO2 (mmHg) % saturation PaO2 (mmHg % saturation PaO2 (mmHg % saturation
1 0.6 34 65.16 80 95.84
2 1.19 36 68.63 45 96.42
4 2.56 38 71.94 90 96.88
6 4.37 40 74.69 95 97.25
8 6.68 42 77.29 100 97.49
10 9.58 44 79.55 110 97.91
12 12.96 46 81.71 120 98.21
14 16.89 48 83.52 130 98.44
16 21.40 50 85.08 140 98.62
18 26.50 52 86.59 150 98.77
20 32.12 54 87.70 175 99.03
22 37.60 56 88.93 200 99.20
24 43.14 58 89.95 225 99.32
26 48.27 60 90.85 250 99.41
28 53.16 65 92.73 300 99.53
30 57.54 70 94.06 400 99.65
32 61.69 75 95.10 500 99.72

Armed with Excel, one can plug these values into a graph generator, and arrive at the familiar-looking oxygen-haemoglobin dissociation curve.

oxygen-hemoglobin dissociation curve as per Severinghaus

The advantage of the Severinghaus model is that it has been confirmed experimentally: it fits human blood data to within 0.55% of a saturation percentage point. Furthermore, it is analytically invertible: one can use the equation to calculate a pO2 from an sO2, and vice versa.

Additionally, Dash and Bassingthwaite have published an excellent article in 2010, presenting a set of oxygen-haemoglobin dissociation curves plotted using a mathematical model which they have developed. Their original diagrams can be found here. They have created a Java applet which runs this model, available from This way, anyone anywhere can generate any sort of dissociation curve, under any sort of conditions. For instance, after fifteen minutes of fiddling around, I was able to produce a series of curves to reflect four of the major influences on the shape of the dissociation curve:

a series of oxyhaemoglobin dissociation curves generated using

In spite of the convenience of being able to generate a curve for any set of parameters, the diagrams elsewhere in this site are based on the Severinghaus data, and on the Siggaard-Andersen method of calculating the shape of the dissociation curve. The main reason for persisting with the Siggaard-Andersen method is the fact that Radiometer use it in their ABG machines. The other reason is largely historical, rooted in the authors' initial impression that the Dash-Bassingthwaite model produced inaccurate curves.  Earlier revisions of this chapter incorrectly maintained that at 0.0 mmol/L of 2,3-DPG the p50 is supposed to be about 10mmHg, as opposed to the value generated by the model. However, as the result of a fortunate correspondence with one of the co-authors, this issue was corrected. On top of that, a revised version of the model (Dash et al, 2016) is now available and gives valid results even in truly deranged circumstances (for instance, where pH is above 8.5). 

Anyway, now that we have an idea what the sigmoid graph looks like, we ask: Why does it curve so?

The concept of positive cooperativity

Haemoglobin is a heterotetramer protein. It is composed of two α-haemoglobin and two β-haemoglobin subunits. Each subunit binds oxygen independently. Then, once an oxygen molecule is bound to it, the oxygenated subunit increases the oxygen affinity of the three remaining subunits. This is positive cooperativity: the subunits "cooperate" to enhance each other's binding of oxygen.

Classically, it has been said that haemoglobin exists in two distinct states: the R state and the T state. The R state ("Relaxed") is the oxygenated form; the T state ("Tense") is the deoxygenated form. The distinction between the two states is a change in structure: one pair of αβ subunits in the oxygenated R-state appears rotated by 15° with respect to the other pair of subunits.

I will use a public domain diagram by Tpirojsi to illustrate this concept. It is discussed in much greater detail, and with excellent visual effects, by David Auble at this University of Virginia biochem lecture site (he in turn uses this excellent article as a reference).

R and T states of haemoglobin

The diagram illustrates the "sequential" model of cooperativity, which suggests that one subunit of haemoglobin starts a sequence of conformational changes in the other subunits haemoglobin which increase their affinity for oxygen, and that this happens in a sequence. The "concerted" model in contrast suggests that haemoglobin cannot exist in a "half-relaxed" state, and that it flips rapidly from one state to another as soon as at least one of the haems in each dimeric subunit is oxygenated.

Apparently, the actual behaviour of haemoglobin molecules is some combination of the two. Both rapid switching and gradual allosteric modulation occur simultaneously. In fact, the deeper we dig, the more hitherto unseen complexity we uncover, with some suggestions that besides the R and T states, many more structural variants may exists, and at the moment it is anybody's guess as to how they play a role in the function of haemoglobin.

Anyway. Though in a totally deoxygenated state it exhibits a certain aloof disinterest in oxygen, haemoglobin becomes more and more interested the more oxygen it binds. The heterotetrameric form of haemoglobin is therefore an ideal molecule for oxygen transport, exhibiting a low oxygen affinity in the hypoxic environment of the capillaries, and a high oxygen affinity in the well-oxygenated pulmonary circulation.

Influence of biochemical and physiological factors on cooperativity

Lots of things can influence the normal cooperativity of haemoglobin subunits. For instance, the presence of an abnormal haem subunit in the tetramer can cause havoc. The presence of a ferric (Fe3+) subunit or of a carboxyhaem subunit can abnormally increase the affinity of the whole molecule. The abnormal haem is locked in its R state and cannot revert to a T state in the presence of hypoxia, impairing the normal unloading of oxygen and causing the curve to shift to the left. In contrast, a sulfhaemoglobin subunit will cause the opposite state effect, and shift the curve to the right.

Much of positive cooperativity and the resulting sigmoid shape of the dissociation curve is owed to the influence of allosteric effectors such as pH, pCO2, 2,3-DPG and so forth. The shape of the curve in the absence of all allosteric effectors is actually a rather boring hyperbolic one. 

Important features on the oxygen-haemoglobin dissociation curve

In the answer to Question 5 from the second paper of 2015, the candidates were expected to sketch the curve with "appropriate key value points (p50, venous and arterial points)". What are these value points? Well. In short, there does not appear to be any strong scientific definition for any of them. Arbitrarily, certain values along the curve were selected by somebody to help discuss the influence of allosteric effectors on the shape of the curve. 
oxygen-haemoglobin dissociation curve for CICM SAQs
  • The p50 point, which is discussed in detail elsewhere, is the PaO2 at which haemoglobin is 50% sturated, and corresponds to 24-28 mmHg under normal circumstances. When most people discuss a change in the p50 value, they are usually referring to a "right-shift" or "left-shift" 
  • The "arterial point" is said to correspond to a PaO2 of 100 mmHg by some authors (Taneja & Vaughan 2001), which would give it a saturaton of 98%. This seems consistent across most textbooks (eg. Brandis, Nunn's) but its origin and significance are unclear. 
  • The "mixed venous point" is nominated at a PaO2 of 40 mmHg, and represents the oxygen partial pressure and saturation of mixed venous blood. It is slightly lower than 75%, and - strictly speaking - should not be drawn on the normal curve, as the curve of mixed venous blood is slightly right-shifted by the Bohr effect (mixed venous blood being slightly more acidic and CO2-rich than arterial blood). 

What is the point of these? Difficult to say. Like the p50, the arterial and mixed venous points can be used to discuss shifts of the curve.  Displacement of either point is the consequence of a shift, although none would be as sensitive as p50 because p50 is where the curve is steepest and therefore it is the point which will move the most. However, realistically one could have chosen any arbitrary point in the curve for this purpose. Therefore, for all intents and purposes, the main reason for these points is to help CICM exam candidates draw a neat sigmoid curve.  


Bohr, C., K. Hasselbalch, and A. Krogh. "Concerning a biologically important relationship–the influence of the carbon dioxide content of blood on its oxygen binding." Skand. Arch. Physiol 16 (1904): 402.

- this is an English translation of the original article, which was beautifully titled "Über einen in biologischer Beziehung wichtigen Einfluss, den die Kohlensäurespannung  des Blutes auf dessen Sauerstoffbindung übt".

Severinghaus, John W. "Simple, accurate equations for human blood O2 dissociation computations."Journal of Applied Physiology 46.3 (1979): 599-602.

Mills, Frederick C., and Gary K. Ackers. "Quaternary enhancement in binding of oxygen by human hemoglobin." Proceedings of the National Academy of Sciences 76.1 (1979): 273-277.

Yonetani, Takashi, et al. "Global Allostery Model of Hemoglobin MODULATION OF O2 AFFINITY, COOPERATIVITY, AND BOHR EFFECT BY HETEROTROPIC ALLOSTERIC EFFECTORS." Journal of Biological Chemistry 277.37 (2002): 34508-34520.

Ackers, Gary K., et al. "Molecular code for cooperativity in hemoglobin."Science 255.5040 (1992): 54-63.

Ho, Chien, and Yue Yuan. "Haemoglobin: Cooperativity in Protein–Ligand Interactions." eLS (2010).

Morgan, T. J. "The oxyhaemoglobin dissociation curve in critical illness." Critical Care and Resuscitation 1.1 (1999): 93.

Rovida, E., et al. "Carboxyhemoglobin and oxygen affinity of human blood."Clinical chemistry 30.7 (1984): 1250-1251.

Aberman, A., et al. "An equation for the oxygen hemoglobin dissociation curve."Journal of applied physiology 35.4 (1973): 570-571.

Forbes, W. H., and F. J. W. Roughton. "The equilibrium between oxygen and hæmoglobin I. The oxygen dissociation curve of dilute blood solutions." The Journal of physiology 71.3 (1931): 229-260.

Ackers, Gary K., and Jo M. Holt. "Asymmetric cooperativity in a symmetric tetramer: human hemoglobin." Journal of Biological Chemistry 281.17 (2006): 11441-11443.

Dash, Ranjan K., Ben Korman, and James B. Bassingthwaighte. "Simple accurate mathematical models of blood HbO2 and HbCO2 dissociation curves at varied physiological conditions: evaluation and comparison with other models." European journal of applied physiology 116.1 (2016): 97-113.

MACDONALD, ROSEMARY. "Red cell 2, 3‐diphosphoglycerate and oxygen affinity." Anaesthesia 32.6 (1977): 544-553.

OKADA, Yoshiaki, Itiro TYUMA, and Tsuyoshi SUGIMOTO. "EVALUATION OF SEVERINGHAUS'EQUATION AND ITS MODIFICATION FOR 2, 3-DPG.The Japanese journal of physiology 27.2 (1977): 135-144.

Taneja, Ravi, and Ralph S. Vaughan. "Oxygen." Continuing Education in Anaesthesia, Critical Care & Pain 1.4 (2001): 104-107.