01Purpose and principlesWhat the assessment is for and the core concepts behind it.
An ABG answers three separate questions: is oxygen transfer adequate for the inspired oxygen, is alveolar ventilation removing carbon dioxide, and is systemic acid-base balance threatened? It should be interpreted beside respiratory rate, work of breathing, haemodynamics and mental state.
PaCO2 reflects ventilation, whereas PaO2 reflects oxygenation and ventilation-perfusion relationships. Pulse oximetry cannot detect hypercapnia, and a normal SpO2 on supplemental oxygen can coexist with dangerous CO2 retention or worsening acidosis.
Bicarbonate on a blood-gas analyser is calculated; laboratory total CO2 is measured. Small differences are expected, but a large mismatch should prompt a check of sampling, timing and laboratory data.
Key points
- Record the inspired oxygen, device, time and clinical state before interpreting PaO2; an oxygen tension without FiO2 is incomplete information.
- Use one sequence: pH, PaCO2, bicarbonate/base excess, compensation, oxygenation, lactate, then electrolytes and clinical cause.
- Acidaemia is pH below 7.35 and alkalaemia above 7.45; decide which component moves in the direction that explains the pH.
- A high PaCO2 drives respiratory acidosis; a low PaCO2 drives respiratory alkalosis. A low bicarbonate drives metabolic acidosis; a high bicarbonate drives metabolic alkalosis.
- Compensation does not normally overshoot the pH. A near-normal pH with clearly abnormal PaCO2 and bicarbonate may represent compensation or two opposing primary disorders.
- Type 1 respiratory failure is hypoxaemia without hypercapnia; type 2 includes hypercapnia and usually reflects inadequate alveolar ventilation.
- For most acutely ill adults target SpO2 94-98%; if at risk of hypercapnic respiratory failure use 88-92% pending blood gases, unless a patient-specific target applies.
- In COPD, persistent pH below 7.35 with PaCO2 above 6.5 kPa despite optimal initial treatment is a standard trigger to consider NIV; do not delay escalation when the trajectory is adverse.
- An anion gap helps organise metabolic acidosis: calculate Na - (Cl + HCO3), interpret against the laboratory range and consider albumin because hypoalbuminaemia lowers the expected gap.
- Repeat the gas after a material change in oxygen, ventilation or treatment; the trend often matters more than one value.
02Indications, selection and cautionsWhen it is useful, when urgency changes and important limitations.
Raised PaCO2 with pH below 7.35 and limited renal bicarbonate retention, often with drowsiness, exhaustion or reduced respiratory effort. In COPD, PaCO2 above 6.5 kPa with persistent acidosis after initial treatment supports urgent NIV assessment.
Raised PaCO2 with raised bicarbonate and a less depressed pH suggests renal compensation. A new pH fall on a chronically high bicarbonate signals acute decompensation; previous stable gases are valuable.
Low pH and bicarbonate with a raised anion gap suggests added unmeasured acids: lactate, ketones, renal failure or selected toxins. Identify the cause immediately rather than giving bicarbonate reflexively.
Low bicarbonate with reciprocal hyperchloraemia suggests gastrointestinal bicarbonate loss, renal tubular acidosis, saline load or early renal dysfunction; calculate the gap instead of calling every acidosis lactic.
Low PaCO2 with alkalaemia may reflect hypoxaemia, pulmonary embolism, sepsis, pregnancy, liver disease, pain or central stimulation. Anxiety is a diagnosis of exclusion in an acutely breathless patient.
Venous sampling, air bubbles, delayed analysis, excess liquid heparin and a sample drawn soon after changing oxygen can distort results. Re-sample promptly if the numbers do not fit the patient.
03Method and interpretationA systematic approach to the test and its findings.
Read from the initial assessment onwards. Tests may run in parallel in urgent care; first-line, preferred, confirmatory, definitive and gold-standard labels appear only when the chapter explicitly states them.
- 01
ABCDE assessment plus pulse oximetry and documented oxygen deliveryFirst step - Why
- Put gas values into the patient's physiological context.
- Interpretation and limitations
- Record device and flow or FiO2. Normal saturation does not exclude hypercapnia; poor perfusion, motion, dyshemoglobinaemia and skin-pigmentation-related bias can affect oximetry.
- 02
pH, PaCO2, bicarbonate and base excess - Why
- Identify the dominant acid-base process and whether compensation is plausible.
- Interpretation and limitations
- Match the direction of PaCO2 and bicarbonate to the pH. If both could be primary, or compensation is more or less than expected, suspect a mixed disorder and review the clinical timeline.
- 03
PaO2 and oxygenation trend - Why
- Assess gas exchange on the known inspired oxygen.
- Interpretation and limitations
- A low PaO2 on substantial oxygen is more concerning than the same value on air. Use serial gases and specialist measures such as the PaO2/FiO2 ratio in critical care; do not infer severity from PaO2 alone.
- 04
Lactate - Why
- Detect impaired perfusion, severe adrenergic drive, seizures, drug effects or reduced clearance.
- Interpretation and limitations
- A raised lactate is a severity signal but not synonymous with sepsis. Trend it after resuscitation and investigate the mechanism; beta2-agonists can contribute in severe asthma.
- 05
Electrolytes, glucose, ketones, renal function and anion gap - Why
- Define common causes of metabolic disturbance.
- Interpretation and limitations
- Calculate Na - (Cl + HCO3) using values from the same time point. Check the local reference interval and albumin; pair a raised gap with lactate, ketones, renal function and toxin history.
- 06
Co-oximetry when dyshemoglobinaemia is possible - Why
- Measure carboxyhaemoglobin or methaemoglobin that standard pulse oximetry cannot resolve reliably.
- Interpretation and limitations
- In suspected carbon monoxide exposure, PaO2 can be normal because it measures dissolved oxygen; treat the exposure and use co-oximetry rather than reassurance from SpO2.
- 07
Repeat ABG or arterialised capillary gas - Why
- Test response to oxygen, bronchodilation, fluids or ventilatory support.
- Interpretation and limitations
- Repeat after 30-60 minutes in people at risk of hypercapnic failure after oxygen changes, or sooner if deteriorating. A venous gas can screen pH/CO2 trends but cannot replace arterial oxygen assessment when that decision matters.
04Clinical next stepsHow the result changes management or prompts escalation.
01First passA reproducible seven-step interpretationFirst stepEvery arterial gas.+
- 11. Verify patient, arterial source, time, temperature if relevant, oxygen device/flow or FiO2, and whether treatment has just changed.
- 22. Decide acidaemia, alkalaemia or near-normal pH; then identify whether PaCO2 and/or bicarbonate explains it.
- 33. Judge compensation and deliberately look for a mixed disorder when the pH or compensation does not fit one process.
- 44. Assess PaO2 in relation to FiO2 and clinical work of breathing; then review lactate, anion gap, glucose, ketones and electrolytes.
- 55. State a synthesis and action: for example, 'acute-on-chronic hypercapnic respiratory acidosis with persistent hypoxaemia on 28% oxygen—controlled oxygen, optimal COPD treatment and urgent NIV assessment'.
02HypercapniaAcute type 2 respiratory failureRaised PaCO2 with pH below 7.35 or neurological/respiratory deterioration.+
- 1Start or titrate controlled oxygen to 88-92% when hypercapnic risk is present, while treating the reversible cause and involving senior respiratory/critical-care staff.
- 2Review depressant medicines, airway obstruction, COPD/asthma treatment, secretions, pneumothorax, obesity hypoventilation and neuromuscular weakness; do not attribute drowsiness to CO2 without excluding other causes.
- 3In COPD, after about 1 hour of optimal medical therapy, persistent pH below 7.35 with PaCO2 above 6.5 kPa supports NIV if appropriate; severe acidosis, inability to protect the airway or rapid deterioration needs critical-care/intubation planning.
- 4EscalationRepeat the gas promptly after intervention and set ceilings/escalation decisions explicitly; improvement in pH, respiratory rate and mental state is more useful than PaCO2 alone.
03Metabolic acidosisFind and reverse the acid sourceLow pH with low bicarbonate or an unexplained negative base excess.+
- 1Assess perfusion and sepsis, measure lactate, glucose/ketones, renal function and electrolytes, then calculate the anion gap with the local reference range.
- 2Treat the cause: restore perfusion, follow the relevant DKA pathway, manage renal failure or obtain urgent toxicology advice. Correct potassium and glucose safely while monitoring ECG when indicated.
- 3Do not use sodium bicarbonate as routine cosmetic correction. Reserve it for specific specialist-led indications such as selected poisonings, severe hyperkalaemia or profound acidaemia where the cause-specific protocol recommends it.
- 4Repeat pH, bicarbonate, lactate, potassium and clinical perfusion; a falling lactate without haemodynamic improvement does not by itself prove adequate resuscitation.
04Mixed resultWhen the numbers do not fitNear-normal pH despite major abnormalities, compensation outside expectation, or discordant clinical findings.+
- 1Recheck sample type, oxygen documentation, timing, analyser warnings and simultaneous laboratory electrolytes; repeat from a reliable site if necessary.
- 2Use the history to identify simultaneous processes: vomiting plus sepsis, COPD plus diuretics, salicylate toxicity, renal failure plus hyperventilation, or DKA after treatment.
- 3Describe each supported process separately and manage each driver; avoid forcing the result into a single label because mixed disorders often carry greater risk.
05Procedure and medicine safetyRelevant preparation, treatment and contraindications.
Controlled oxygen
Titrate to SpO2 94-98% for most acutely ill adults; if at risk of hypercapnic respiratory failure use 24% Venturi at 2-3 L/min or 28% Venturi at 4 L/min (or nasal cannulae 1-2 L/min) aiming 88-92% pending gases.Oxygen is prescribed to a target, not a fixed saturation maximisation. In critical illness start high-concentration oxygen if needed, then titrate rapidly once reliable oximetry/gases are available; repeat gases after 30-60 minutes in hypercapnic-risk patients.
Naloxone when opioid toxicity is causing ventilatory failure
Titrate IV in small increments in monitored care; a commonly used initial hospital dose is 100-200 micrograms IV, repeated every 2-3 minutes to adequate ventilation, with larger 400 microgram doses used when dependence is unlikely. Follow the current emergency medicines protocol.Aim for adequate breathing, not full arousal. Acute withdrawal, agitation and recurrent toxicity can occur because naloxone may wear off first; infusion and prolonged observation may be needed. Verify the locally approved concentration and protocol.
Sodium bicarbonate 8.4% for selected indications only
No routine dose for undifferentiated metabolic acidosis. In adult cardiac arrest it is not routine; specialist toxicology or hyperkalaemia pathways may use 50 mmol IV with repeat guided by pH/ECG and sodium load.Can generate CO2, increase sodium/osmolality, lower ionised calcium and worsen intracellular acidosis if ventilation is inadequate. Use only when the indication and endpoint are explicit.
06Risks, monitoring and follow-upComplications, safety checks and further assessment.
- Trend respiratory rate, work of breathing, consciousness, SpO2, oxygen device/flow and haemodynamics alongside every repeat gas.
- Repeat ABG 30-60 minutes after oxygen adjustment in patients at risk of hypercapnic failure, and earlier after NIV initiation or any clinical deterioration.
- Track pH and PaCO2 response rather than chasing a normal PaCO2 in chronic retainers; compare with previous stable bicarbonate/gases when available.
- For metabolic acidosis, trend lactate or ketones only with perfusion, urine output, glucose, potassium, renal function and the underlying disease endpoint.
- Document sampling site, FiO2/device and treatment timing so subsequent clinicians can interpret the trajectory safely.
07Special situationsVariants, exceptions and circumstances that change the usual approach.
A normal pH can conceal two diseases
Markedly raised PaCO2 and markedly raised bicarbonate with a near-normal pH may be chronic compensation, but COPD plus vomiting/diuretics can create a superimposed metabolic alkalosis. Timeline and previous results decide.
PaO2 is not oxygen content
Severe anaemia can leave PaO2 and saturation apparently acceptable while oxygen delivery is poor; carbon monoxide can leave PaO2 normal while functional carriage is impaired.
Do not overcorrect chronic CO2 retention
The immediate goal is safer pH, work of breathing and mental state with appropriate oxygenation—not forcing PaCO2 to a normal range that may be neither achievable nor appropriate.
Albumin changes the anion-gap baseline
A 'normal' gap in marked hypoalbuminaemia may hide unmeasured acid. Use the local laboratory approach and interpret the gap as a clue, not a diagnosis.
Lactate can be adrenergic
Severe asthma and repeated beta2-agonist treatment can raise lactate; worsening tachypnoea may then reflect metabolic compensation rather than persistent bronchospasm alone. Reassess airflow and the whole patient before escalating bronchodilator blindly.
08Common pitfallsFrequent interpretation and management errors.
- 01
Interpreting PaO2 without documenting inspired oxygen and device.
- 02
Calling a venous oxygen tension an arterial oxygen measurement.
- 03
Assuming normal SpO2 excludes hypercapnia or acid-base danger.
- 04
Treating the pH with bicarbonate before identifying shock, DKA, renal failure or toxin exposure.
- 05
Describing compensation as a second primary disorder without checking whether it is physiologically plausible—and missing a true mixed disorder when it is not.
- 06
Withholding oxygen from a critically hypoxaemic COPD patient because of fear of CO2 retention; give controlled oxygen, obtain gases and support ventilation.