Practice Rational Testing

Interpreting arterial blood gas results

BMJ 2013; 346 doi: (Published 16 January 2013) Cite this as: BMJ 2013;346:f16
  1. Nicholas J Cowley, research registrar, anaesthesia and intensive care medicine1,
  2. Andrew Owen, academic clinical fellow1,
  3. Julian F Bion, professor of critical care medicine2
  1. 1Department of Anaesthesia and Critical Care, Queen Elizabeth Hospital Birmingham, Birmingham B15 2WB, UK
  2. 2University Department of Anaesthesia and Intensive Care Medicine, Queen Elizabeth Hospital, Birmingham B15 2TH
  1. Correspondence to: N J Cowley n.j.cowley{at}

Learning points

  • Interpretation of arterial blood gases requires a systematic assessment of oxygenation, pH, standard bicarbonate (sHCO3−) and base excess, partial pressure of carbon dioxide (PaCO2), and additional analytes

  • The P/F ratio (ratio between the PaO2 and the inspired oxygen concentration expressed as a fraction) is a useful guide to the presence and severity of impaired alveolar gas exchange

  • Reassess all acutely ill patients regularly, and consider repeat arterial blood gas analysis

  • Errors in blood gas analysis are dependent more on the clinician than on the analyser

You have been called to see a 69 year old man on a surgical ward because he has become drowsy and short of breath. He had a large bowel resection the previous day, has a background of type 2 diabetes, and is a current smoker. On examination his arterial blood pressure is 104/65 mm Hg, his heart rate 132 beats/min and irregular, and his respiratory rate 22 breaths/min; his oxygen saturations with pulse oximetry are 94% on supplemental oxygen via a 40% Venturi-type mask. He is slightly confused and is complaining of abdominal pain despite using patient controlled analgesia with morphine. His chest is clear on auscultation.

What is the next investigation?

You take a blood specimen for analysis of arterial blood gases for rapid biochemical evaluation to guide diagnosis and initial management. Table 1 shows the results. It is important to adopt a systematic approach to interpreting results of arterial blood gases, as outlined in table 2, preceded by a brief history and focused clinical examination.

Table 1

 Report of arterial blood gases for the hypothetical patient described

View this table:
Table 2

 Guide to systematic approach to analysis of a report of arterial blood gases

View this table:

Step 1: Assess oxygenation

Arterial oxygen tension (PaO2) is the partial pressure of oxygen in arterial blood. The main determinants of PaO2 are the inspired oxygen concentration, alveolar gas exchange, and, to a lesser extent, tissue oxygen consumption. The ratio between the PaO2 and the inspired oxygen concentration expressed as a fraction (FiO2) is termed the PaO2/FiO2 ratio or the P/F ratio. This is a useful index for determining the presence and severity of impaired alveolar gas exchange and is easier to calculate than alternative indices, such as the alveolar-arterial gradient. Estimations of FiO2 based on oxygen flow through a standard facemask are rarely accurate. The FiO2 will vary according to the oxygen delivery device used, the presence of a reservoir, and the patient’s inspiratory flow rate. A healthy individual would be expected to have a P/F ratio above 50, with lower values signifying impaired gas exchange. Patients with acute lung injury or acute respiratory distress syndrome have values below 40 and 26.7 respectively, in addition to other required diagnostic criteria.1

The PaO2 in our example patient (8.9 kPa) is below normal, but as he is breathing supplemental oxygen rather than room air, this represents significant impairment of oxygen uptake, probably from intrapulmonary shunting. Intrapulmonary shunting occurs when areas of lung are perfused without adequate ventilation—for example, after atelectasis, consolidation, fluid accumulation, or acute inflammation of lung tissue. In the calculation of his P/F ratio, the inspired oxygen concentration is determined by the Venturi-type mask (in this case 0.4). Thus, his P/F ratio is calculated as (8.9/0.4 = 22.3), representing marked impairment in gas exchange.

Be aware that the measurement of oxygen saturation using standard pulse oximetry and some arterial blood gas analysers may give misleading results. Oxygen saturations are falsely raised in carbon monoxide poisoning (which produces carboxyhaemoglobin) and depressed in methaemoglobinaemia, which is caused by various drugs or toxins, including nitrate fertilisers, some local anaesthetics, and sulphonamide antibiotics. These conditions cannot be readily distinguished clinically, and analysers using co-oximetry to analyse haemoglobin oxygen saturations will report levels of carboxyhaemoglobin and methaemoglobin.2 However, if oxygen saturation is not available on the analyser, pay close attention to the patient’s clinical history.

Step 2: Assess pH

The pH is usually maintained within a tight range between 7.35 and 7.45, and a small change in the pH will result in a large change in the hydrogen ion concentration, making even modest derangements in the pH of clinical significance. Our example patient has an acidosis (pH of 7.25) or, more accurately, an acidaemia (abnormally low blood pH). In some cases an underlying acid-base disorder can be disguised by compensatory mechanisms that normalise pH, referred to as a compensated acidosis or alkalosis.

Step 3: Assess standard bicarbonate (sHCO3−) and base excess

Most blood gas analysers will calculate values for standard bicarbonate (sHCO3−) and base excess, either of which can be used to isolate metabolic causes of acid-base disturbance. These values are particularly useful when the cause of the acid-base disorder has both metabolic and respiratory components. The contribution of any respiratory acid-base disorder to the sHCO3− concentration and base excess is removed by the analyser’s software, which adjusts the carbon dioxide to the normal value of 5.3 kPa. In the case of metabolic acidosis, we would expect to see a reduction in the sHCO3− concentration, and a more strongly negative base excess (commonly termed a base deficit). For the patient in our example, the acidosis is likely to be metabolic in origin, given the depressed sHCO3− concentration of 18.5 mmol/L, and negative base excess of −7.0 mmol/L. Normal values for standard bicarbonate sHCO3− and base excess exclude metabolic acid-base disturbance, and a raised sHCO3− concentration and positive base excess indicate a metabolic alkalosis. The figure shows the common acid-base disturbances.


Interpretation of arterial blood gases in the presence of acidaemia or alkalaemia, with examples of common diagnoses

A metabolic acidosis can be characterised further by determining the anion gap from the information on the blood gas report. The anion gap is the difference between the anions and cations that are measured as standard (Na+, K+, Cl−, and HCO3−), calculated with the formula: ((Na+) + (K+)) − ((Cl−) + (HCO3−)). A rise from a normal value of 10 (reference range 6-14) mmol/L indicates an excess of unmeasured anions, which are responsible for the underlying acidosis, causes of which include lactic acidosis, ketoacidosis, renal failure, and toxins.3 Many blood gas analysers are able to detect lactate, one of the commonest causes of raised anion gap acidosis, usually caused by inadequate organ perfusion. Trends in lactate concentrations are useful in guiding response to treatment.4 5 A metabolic acidosis with a normal anion gap is usually accompanied by hyperchloraemia, causes of which include iatrogenic saline infusion as well as gastrointestinal loss of bicarbonate from diarrhoea or renal loss of bicarbonate (such as renal tubular acidosis type I and II).

Step 4: Assess arterial partial pressure of carbon dioxide (PaCO2)

The arterial partial pressure of carbon dioxide (PaCO2) should be assessed next to identify any ventilatory component in the acid-base disturbance. A raised PaCO2 value will contribute towards an acidosis, and a low value towards an alkalosis. In our patient the PaCO2 value is not raised, indicating that the acidosis is not respiratory in origin. If respiratory drive were normal, compensatory hypocarbia would be expected. However, in our example, the patient’s PaCO2 (5.9 kPa) is at the upper limit of normal, indicating an inadequate ventilatory response, which could be caused by opioid analgesia, coexistent chronic obstructive pulmonary disease, severe abdominal pain splinting breathing, or incipient ventilatory failure. Thus our patient has a metabolic acidosis without respiratory compensation.

The presence of a normal PaO2 value, or normal values on pulse oximetry, does not rule out respiratory failure, particularly in the presence of supplemental oxygen. An unexpectedly high PaCO2 value is a more sensitive marker of ventilatory failure than pulse oximetry or PaO2, particularly in the presence of supplemental oxygen, as it has a close relationship with depth and rate of breathing.

Step 5: Assess additional analytes

Many “point of care” arterial blood gas analysers can now evaluate electrolytes, haemoglobin, glucose, and lactate. The additional information, available within minutes of the primary assessment, can aid diagnosis and guide early treatment. The patient in our example has hypokalaemia (potassium 3.0 mmol/L). This has probably precipitated atrial fibrillation, which will impair his cardiac output. His haemoglobin concentration of 6.0 g/dL (60 g/L) is low; occult haemorrhage with inadequate tissue oxygen delivery might have caused the metabolic acidosis. This is a particular risk in the postoperative setting when oxygen demand is increased.

Step 6: Reassess

After the start of treatment, regular reassessment will be needed. Repeated blood gas analysis can demonstrate response to treatment and guide further treatment. In a high dependency setting, consider inserting an arterial cannula for obtaining repeated specimens to avoid multiple arterial punctures.


With advances in machine performance and quality assurance,6 7 8 two thirds of errors in point of care analysis of arterial blood gases are now attributable to clinicians.9 10 Attention to detail in sampling technique and processing is thus essential (table 3). If obtaining an arterial sample is difficult, venous blood (taken without a tourniquet) will provide a reasonable substitute for all analytes other than PaO2, although this should be clearly marked as such to avoid confusion in interpretation.

Table 3

Sources of error in analysis of arterial blood gases

View this table:


Our patient received adequate analgesia to allow more comfortable breathing and was monitored closely for evidence of bleeding. He received fluid therapy and a blood transfusion, which coincidentally increased the serum potassium concentration. This treatment caused resolution of his acid-base disturbance on subsequent arterial blood gas analysis, as well as spontaneous reversion to sinus rhythm.


Cite this as: BMJ 2013;346:f16


  • This series of occasional articles provides an update on the best use of key diagnostic tests in the initial investigation of common or important clinical presentations. The series advisers are Steve Atkin, professor, head of department of academic endocrinology, diabetes, and metabolism, Hull York Medical School; and Eric Kilpatrick, honorary professor, department of clinical biochemistry, Hull Royal Infirmary, Hull York Medical School. To suggest a topic for this series, please email us at practice{at}

  • Contributors: All authors have participated in the planning, drafting, and revising of this manuscript. JFB is the guarantor.

  • Competing interests: All authors have completed the ICMJE uniform disclosure form at (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Commissioned; externally peer reviewed.

  • Patient consent not required (patient anonymised, dead, or hypothetical).