Show Summary Details
Page of

Blood gas analysis in the critically ill 

Blood gas analysis in the critically ill
Blood gas analysis in the critically ill

Gavin M. Joynt

and Gordon Y. S. Choi

Page of

PRINTED FROM OXFORD MEDICINE ONLINE ( © Oxford University Press, 2020. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Medicine Online for personal use (for details see Privacy Policy and Legal Notice).

date: 01 December 2020

Key points

  • Oxygenation is assessed by measuring PaO2 and SaO2 in the context of the inspired oxygen concentration, haemoglobin concentration, and the oxyhaemoglobin dissociation curve.

  • Ventilation is assessed by measuring the PaCO2 in the context of systemic acid-base balance.

  • Acid base assessment requires the integration of clinical findings and a systematic interpretation of arterial blood gas parameters.

  • For clinical use, traditional acid base interpretation rules based on the bicarbonate buffer system or standard base excess estimations, and the interpretation of the anion gap, are substantially equivalent to the physicochemical method of Stewart.

  • The presence of a metabolic acidosis or alkalosis, and its influence on compensatory respiratory responses is important to recognize.


Arterial blood gases provide information that allows the assessment of patient oxygenation, ventilation and acid-base status. Usually, modern blood gas machines directly measure pH, and the partial pressures of carbon dioxide (PaCO2) and oxygen (PaO2) dissolved in arterial blood. These values are then used to calculate the arterial oxygen saturation (SaO2), bicarbonate (HCO3) concentration and base excess (BE). The addition of a co-oximeter allows the direct estimation of haemoglobin content, haemoglobin oxygen saturation (SaO2), and carbon monoxide (COHb) and methaemoglobin (MetHb) saturation. More sophisticated and expensive machines can measure electrolyte concentrations including sodium, potassium, chloride, magnesium, and calcium. Commonly measured parameters and their significance are listed in Table 72.1.

Table 72.1 Common parameters reported on an arterial blood gas


Normal values


pH (m)


  • Measurement of hydrogen ion [H+]:

  • Acidaemia: arterial pH < 7.35

  • Alkalaemia: arterial pH > 7.45

  • Acidosis: an abnormal process or condition that would lower arterial pH

  • Alkalosis: an abnormal process or condition that would raise arterial pH (if there were no secondary changes in response to the primary aetiological factor)

PaCO2 (m)

  • 35–45 mmHg

  • (4.7–6 kPa)

  • Partial pressure of CO 2 in arterial blood:

  • PaCO2 > 45 mmHg (6 kPa) indicates respiratory acidosis

  • PaCO2 < 35 mmHg (4.7 kPa) indicates respiratory alkalosis

HCO3– (c)

22–26 mmol/L

  • The amount of buffer base in arterial blood:

  • <22 mmol/L indicates metabolic acidosis

  • >26 mmol/L indicates metabolic alkalosis

Standard base excess (c)

–2 to +2 mmol/L

The amount of acid (in mmol/L) required to restore 1 L of tested blood to a pH of 7.4

SaO2 (m or c)


The percentage of oxygen bound to haemoglobin

PaO2 (m)

  • 80–100 mmHg

  • (10.7–13.3 kPa)

  • Partial pressure of oxygen in arterial blood:

  • Hypoxaemia is defined as PaO2 < 60 mmHg (8 kPa)

Measured parameters are designated (m) and calculated parameters designated (c).

Assessment of oxygenation

The PaO2 and SaO2 are used to assess oxygenation. These values should be interpreted in conjunction with the fractional inspired oxygen concentration (FiO2). Hypoxaemic respiratory failure is characterized by PaO2 lower than 60 mmHg (8.0 kPa) when breathing room air at sea level. In critically-ill patients, it is usually not necessary or desirable to remove oxygen supplementation to interpret blood gases with the patient breathing room air. Hypoxaemic respiratory failure in the critically ill can be most simply expressed in terms of the PaO2/FiO2 or PF ratio. With the PaO2 measured in mmHg (kPa), the normal PF ratio is approximately 450 (60) and a PF ratio < 200 (25) indicates severe hypoxaemic respiratory failure.

Major pathophysiological mechanisms leading to hypoxaemia include a reduction of inspired oxygen tension, hypoventilation, ventilation–perfusion mismatch, right-to-left shunting, and diffusion impairment. Causes may be elucidated by determining the alveolar–arterial oxygen gradient (A–a PO2) where:

P a O 2   =   ( P B     P H 2 o )   ×   F I O 2     ( P a C O 2 / R )
[eqn 1]

where Pb = barometric pressure, 760 mmHg (101kPa) at sea level, Ph2o = partial pressure of fully saturated water vapour, 47 mmHg (6.3 kPa) at 37°C, PaCO2 = partial pressure of alveolar carbon dioxide ~ PaCO2 because of the ease of exchange of carbon dioxide, R = respiratory quotient (R = 0.8 for a ‘normal’ diet).

The normal (A–a PO2) gradient varies with age and ranges from 7 to 14 mmHg (0.9–1.9 kPa) when breathing room air, but increases with age [1], where:

(Aa  PO 2 )=2.5+0.21×age (in years), in mmHg
[eqn 2]

(A a PO 2 ) = 0.3   +   0.03   ×   a g e ,   i n   k P a
[eqn 3]

Solving the alveolar gas equation helps distinguish whether hypoxaemia is caused by hypoventilation, or is the result of ventilation perfusion mismatch, and/or diffusion abnormalities. In hypoventilation, there will be a normal gradient between alveolar and arterial blood, while other abnormalities result in an increased A–a PO2. Mixed abnormalities often occur, such as in a patient with chronic obstructive airways disease (causing alveolar hypoventilation) and lobar collapse (causing a decreased ventilation perfusion ratio or shunt).

The following formula illustrates that the amount of dissolved oxygen in the blood, measured by PaO2, is relatively small, and the oxygen content of blood is primarily determined by the haemoglobin concentration and the haemoglobin-bound O2 (HbO2), expressed as a saturation.

Blood  O 2  content(mL/100 mL) = Hb(g/dL)× SaO 2 ×1.306+ PaO 2 (mmHg)×0.003
[eqn 4]

The PaO2 should thus always be interpreted in the context of the relationship between PaO2 and HbO2. The oxyhaemoglobin dissociation curve describes this relationship (Fig. 72.1). If the measured saturation is not what you would expect from the PaO2 it may indicate a shift in the oxyhaemoglobin dissociation curve. The PaO2 at which saturation is 50% (P50) is usually used as a reference point. If P50 is increased, it indicates a shift to the right and vice versa. Possible causes of shift are shown in Fig. 72.1. Note that in clinical practice, saturations of <88–90% are considered risky, because below this saturation the content of oxygen carried by arterial blood decreases substantially with relatively small decreases in PaO2, risking tissue hypoxia. When SaO2 is not measured by co-oximetry, but calculated by the blood gas machine on the basis of the oxyhaemoglobin dissociation curve, inaccuracies in SaO2 estimations are likely, especially when PaO2 is relatively low (as occurs in venous samples). Also related to the shape of the curve, saturations greater than 90–92% will increase oxygen content by relatively small amounts. Co-oximetry also allows the quantitative measurement of abnormal haemoglobin complexes, such as COHb and MetHb, both of which can substantially reduce blood oxygen content when present in high concentrations.

Fig. 72.1 The normal oxyhaemoglobin dissociation curve with causes of left and right shift. Relationships at key points are shown; the partial pressure at which saturation is 50% (P50), mixed venous oxygen tension (v), and the inflection point of the oxyhaemoglobin dissociation curve known as the minimum saturation point (M). Below M a small reduction in PaO2 will result in a large decrease in oxygen saturation and oxygen content. To convert to kPa, divide by 7.5 DPG, 2,3-diphosphoglycerate.

Fig. 72.1 The normal oxyhaemoglobin dissociation curve with causes of left and right shift. Relationships at key points are shown; the partial pressure at which saturation is 50% (P50), mixed venous oxygen tension (v), and the inflection point of the oxyhaemoglobin dissociation curve known as the minimum saturation point (M). Below M a small reduction in PaO2 will result in a large decrease in oxygen saturation and oxygen content. To convert to kPa, divide by 7.5 DPG, 2,3-diphosphoglycerate.

Reproduced by permission of G. Joynt/G. Choi, for ICU Web

Assessment of ventilation and acid-base disturbances

Ventilation should always be assessed in combination with acid-base status, as ventilation forms an integral part of acid-base homeostasis. A rise in PaCO2 indicates alveolar hypoventilation, while a decrease indicates alveolar hyperventilation. Given the requirement to maintain a normal pH, functioning homeostatic mechanisms result in metabolic acidosis triggering a compensatory hyperventilation, and metabolic alkalosis a compensatory reduction in ventilation. Similarly, when primary alveolar hypoventilation generates a respiratory acidosis, it results in a compensatory increase in serum bicarbonate that is achieved, in part, by kidney bicarbonate retention. In the same way, respiratory alkalosis induces kidney bicarbonate loss.

It may happen that more than one ‘primary’ abnormality exists, for example, respiratory alkalosis and metabolic acidosis can be caused simultaneously by salicylate poisoning, and diagnosis of combined disorders like this can be challenging. Therefore to assess acid-base and ventilatory disturbances it is necessary to have at least a basic understanding of acid-base physiology, and a systematic approach to the interpretation of the arterial blood gas.

Systematic blood gas interpretation methods

Three commonly used systematic approaches exist to this end [2]‌. The two traditional, established methods use the arterial blood value of pH as a measure of the degree of acidity or alkalinity, PaCO2 as a marker of the respiratory component and bicarbonate concentration (HCO3) or BE as a marker of the non-respiratory or metabolic component of acid-base balance.

The first approach is based mainly on the interpretation of the Henderson–Hasselbach equation:

C O 2 + H 2 O < > H 2 C O 3 < > H + + H C O 3
[eqn 5]

Rearranged this gives:

p H = p K   +   log { [ H C O 3 ] / [ H 2 C O 3 ] }
[eqn 6]

where pK is the dissociation constant for carbonic acid.

This can further be expressed as:

p H   =   p K   +   log { [ H C O 3 ] / [ 0.03   ×   p C O 2 ] }
[eqn 7]

According to this construct, changes in arterial pH occur as a result of changes in either [HCO3] or PaCO2. Primary disorders occur with initial changes in either of the two parameters. Metabolic acidosis occurs with a reduction in [HCO3], whereas an increase in [HCO3] occurs with metabolic alkalosis. An increase in PaCO2 is indicative of respiratory acidosis, while a reduction in PaCO2 indicates respiratory alkalosis.

Homeostasis, however, demands a normal, or near normal, pH and in this system the bicarbonate buffering system accomplishes this goal. The most immediate response is simply that as PaCO2 increases, the law of mass action will result in more [HCO3] production. However, this response by itself may be insufficient so there are additional responses in the kidney that increase [HCO3] production. It is, however, well known that ‘hidden’ buffers like haemoglobin, albumin, and phosphates also buffer excess protons and minimize alterations in pH. One way of addressing this complexity has been to observe normal compensatory changes in humans and animals exposed to acid base disorders, and quantify them in terms of changes in pH, PaCO2 and [HCO3]. This data has provided empirical limits for normal compensation and observed aberrations outside these compensatory limits are assumed to be the result of the presence of other primary abnormalities. Much of this work was completed in Boston, hence, the term ‘Boston Rules’ when this approach is used [3]‌. Box 72.1 provides a guide for using this approach in the clinical setting.

Reproduced by permission of G Joynt/G Choi, for ICU Web

Despite widespread use of the ‘Boston’ method, theoretical limitations are the assumption that PaCO2 and [HCO3] are independent of one another, and that all buffering of metabolic acids is by HCO3. An alternative method of interpretation deals with these theoretical limitations. The ‘standard bicarbonate’ is a CO2-independent index of the HCO3 concentration of a sample when the partial pressure CO2 has been adjusted to 40 mmHg (5.3 kPa) at a temperature of 37°C. Buffer base was introduced as a measure of the concentration of all the buffers present in either plasma or blood, and BE as a measure of how far buffer base has changed from its normal value. BE is defined as the amount of strong acid or base required to return the pH of 1 L of whole blood to 7.4, assuming PaCO2 of 40 mmHg (5.3 kPa), a temperature of 37°C. BE is routinely reported by most blood gas analysing machines. Thus, BE is proposed as a measure of the magnitude of the metabolic disorder, because it assesses all the extracellular buffers (in the blood sample) and is independent of PaCO2. As haemoglobin is a buffer, but is not distributed throughout the extracellular fluid (ECF), BE is calculated for a haemoglobin concentration of 30 or 50 g/L, instead of the actual haemoglobin, allowing the whole ECF buffering capacity (and not just whole blood) to be estimated. With this adjustment, known as standard BE, the differences between in vitro and in vivo behaviour are largely eliminated [4]‌. It is BE variation from normal that forms the basis of the second traditional method of systematic arterial blood gas analysis. The ‘Copenhagen’ rules are usually used in a similar context to the Boston rules (Box 72.1).

More recently, an alternative method developed by Stewart used the physical-chemical principles of electrical neutrality, conservation of mass, and laws of dissociation of electrolytes to explain hydrogen ion disturbances [5]‌. In this model [H+] and [HCO3] are dependent variables and within the system, the concentration of [H+] changes only as a result of the degree of dissociation of water. The degree of dissociation is significantly influenced by three independent factors:

  • The strong ion difference (SID): the difference between strong (always dissociated) cations and strong anions.

  • The total non-volatile weak acids (A TOT): mainly albumin and phosphate (existing in dissociated (A) and non-dissociated (AH) forms).

  • The PaCO2.

Thus, it is the electrochemical forces produced by changes in SID, ATOT and PaCO2 that alter the [H+]. Stewart solved a series of simultaneous mathematical equations describing the relationship between the factors listed above and [H+]. Although more complex, this method produces an accurate description of acid-base abnormalities, and often provides clearer pathophysiological explanations of clinically observed acid-base phenomena than traditional methods [6]‌.

Bedside blood gas interpretation

Attempts to compare the different methods of systematic interpretation of arterial blood gases have failed to demonstrate the clear superiority of any one system. The Boston rules approach (including the use of the anion gap) has been shown to be mathematically similar to the BE method and the Stewart physicochemical method [7,8]. When applied in experimental and clinical settings, the ability of traditional and the Stewart systems to predict the outcome of acid-base alterations is also similar [9]‌. Although the Stewart remains attractive because it provides explanatory information that appears to more precisely explain observed pathophysiological changes, it is currently infrequently used at the bedside. Whether it apparently superior explanatory value will ultimately translate into superior clinical bedside utility remains to be seen [10].

To summarize, all systematic approaches to the diagnosis of acid-base disorders involve three critical steps. The first is a thorough clinical assessment based on history, examination and initial investigations. This should lead to a clinical decision as to what is the most likely acid-base disorder, and what the possible differential diagnoses are. Mixed disorders are often difficult to recognize, and history and examination alone are usually insufficient to make a firm diagnosis. The second step is to perform a systematic evaluation of the arterial blood gas (Box 72.1), including an analysis of the anion gap, and the last is to synthesize all available information and finalize the diagnosis. For the purposes of respiratory monitoring, a diagnosis of respiratory acidosis or alkalosis should lead to the search for associated clinical causes and the implementation of appropriate interventions (Boxes 72.2 and 72.3). Although the detailed diagnosis of metabolic disorders is beyond the scope of this chapter, recognition of the presence of a metabolic acidosis or alkalosis, and its influence on compensatory respiratory responses is important to recognize.


1. Mellemgaard K. (1966). The alveolar-arterial oxygen difference: its size and components in normal man. Acta Physiologica Scandinavica, 67, 10–20.Find this resource:

2. Kellum JA. (2000). Determinants of blood pH in health and disease. Critical Care, 4, 6–14.Find this resource:

3. Severinghaus JW. (1993). Siggaard-Andersen and the ‘Great Trans-Atlantic Acid-Base Debate’. Scandinavian Journal of Clinical Laboratory Investigation, 214 (Suppl.), 99–104.Find this resource:

4. Severinghaus JW and Astrup PB. (1985). History of blood gas analysis. Part II. pH and acid-base balance measurements. Journal of Clinical Monitoring, 1, 259–77.Find this resource:

5. Stewart PA. (1978). Independent and dependent variables of acid-base control. Respiratory Physiology, 33, 9–26.Find this resource:

6. Kellum JA. (2005). Clinical review: reunification of acid-base physiology. Critical Care, 9, 500–7.Find this resource:

7. Kurtz I, Kraut J, Ornekian V, and Nguyen MK. (2008). Acid-base analysis: a critique of the Stewart and bicarbonate-centered approaches. American Journal of Physiology—Renal Physiology, 294, F1009–31.Find this resource:

8. Matousek S, Handy J, and Rees SE. (2011). Acid-base chemistry of plasma: consolidation of the traditional and modern approaches from a mathematical and clinical perspective. Journal of Clinical Monitoring and Computing, 25, 57–70.Find this resource:

9. Gattinoni L, Carlesso E, Maiocchi G, Polli F, and Cadringher P. (2009). Dilutional acidosis: where do the protons come from? Intensive Care Medicine, 35, 2033–43.Find this resource:

10. Kellum JA. (2007). Disorders of acid-base balance. Critical Care Medicine, 35, 2630–6.Find this resource: