Interpret an arterial blood gas

Clinical overview

The arterial blood gas (ABG) is the single most decision-changing bedside test you will order in a deteriorating obstetric or post-operative patient. It tells you four things at once that no other test does so quickly: whether the patient is oxygenating, whether she is ventilating, what her acid–base status is, and — increasingly with modern analysers — her lactate, glucose, electrolytes and haemoglobin. For the O&G registrar this matters most in three recurring scenarios: the collapsed or septic mother, the woman with severe pre-eclampsia or pulmonary oedema, and the post-operative patient (post-caesarean, post-laparotomy) who is tachypnoeic, hypotensive or oliguric. In each, an ABG converts a vague clinical impression into a quantified physiological problem you can act on.

The FCOG(SA) examiner is not asking you to be an intensivist. They are testing whether you can take a printed gas result, work through it in a disciplined, reproducible order, name the disturbance correctly, calculate the expected compensation, and translate the numbers into a safe management decision and an escalation call. This is a higher-order thinking skill (HOTS): the marks are in the interpretation and reasoning, not in reciting normal values. The commonest exam failure is reading the pH and pO₂ and stopping there. Pregnancy adds a crucial twist — the normal maternal ABG is not the normal non-pregnant one, and treating a "low" maternal bicarbonate or pCO₂ as pathology is a classic error. Master the systematic approach below and you will interpret any gas, in any patient, every time.

Core knowledge

Figure F3.1 — Pregnancy resets the ABG: progesterone-driven hyperventilation lowers PaCO₂, the kidney excretes bicarbonate, and the result is a chronic compensated respiratory alkalosis — so a "normal" maternal PaCO₂ of 5 kPa signals CO₂ retention.

The normal arterial blood gas — and how pregnancy shifts it

Standard adult arterial reference ranges (textbook physiology, breathing room air at sea level):

Progesterone is a direct respiratory stimulant. From early pregnancy it drives an increase in tidal volume (and hence minute ventilation by ~30–50%), producing a chronic compensated respiratory alkalosis. The pCO₂ falls to roughly 3.7–4.3 kPa, the kidney excretes bicarbonate to compensate (HCO₃⁻ falls to ~18–22 mmol/L), and the pH settles slightly alkalaemic. The practical consequences are large and examinable:

• A maternal pCO₂ of 5.0 kPa — perfectly normal in a non-pregnant adult — represents relative CO₂ retention in a term mother and may signal impending respiratory failure or exhaustion (for example in severe asthma or magnesium toxicity). A "normalising" pCO₂ in a tiring asthmatic is an ominous sign.
• The lowered buffering reserve (lower baseline bicarbonate) means a pregnant woman tolerates a metabolic acidosis less well — she has less bicarbonate to spare. Diabetic ketoacidosis, including euglycaemic DKA, develops faster and at lower glucose levels in pregnancy.
• Maternal alkalosis shifts the oxyhaemoglobin dissociation curve and, with the placenta, helps offload oxygen to the fetus (the maternal–fetal "double Bohr / double Haldane" effect — standard teaching).

Acid–base first principles

pH is governed by the Henderson–Hasselbalch relationship, which simplifies clinically to: pH depends on the ratio of bicarbonate (the metabolic, renal-controlled component) to carbon dioxide (the respiratory, lung-controlled component).

• Acidosis = a process lowering pH. Acidaemia = an actually low blood pH (<7.35). Alkalosis/alkalaemia mirror these. Use the words precisely; a patient can have a normal pH yet harbour two opposing primary disorders.
• A respiratory disturbance is a primary change in pCO₂: a high pCO₂ acidifies (respiratory acidosis), a low pCO₂ alkalinises (respiratory alkalosis).
• A metabolic disturbance is a primary change in bicarbonate/base: a low HCO₃⁻/negative base excess acidifies (metabolic acidosis), a high HCO₃⁻/positive base excess alkalinises (metabolic alkalosis).
• Compensation is the body's attempt to push the pH back toward normal by adjusting the other system. The lungs compensate fast (minutes–hours) by changing ventilation; the kidneys compensate slowly (hours–days) by retaining or excreting bicarbonate. Compensation never fully corrects the pH — if the pH is normal but both pCO₂ and HCO₃⁻ are deranged, suspect either full compensation of a chronic single disorder or a mixed disorder.

The anion gap

The anion gap (AG) separates the causes of a metabolic acidosis and is essential higher-order reasoning:

AG = Na⁺ − (Cl⁻ + HCO₃⁻), normal ≈ 8–12 mmol/L (standard teaching; correct for albumin — add ~2.5 mmol/L to the gap for every 10 g/L the albumin is below ~40 g/L, important in sick hypoalbuminaemic obstetric patients).

• Raised anion gap metabolic acidosis (HAGMA) — unmeasured acid anions accumulate. Mnemonic MUDPILES / GOLD MARK: lactate (sepsis, haemorrhagic shock, ischaemic bowel, tissue hypoperfusion), ketones (DKA, starvation), renal failure (urate, sulphate), toxins (methanol, ethylene glycol, salicylates, paraldehyde). In obstetrics, lactic acidosis from sepsis or haemorrhage is the one you will meet most.
• Normal anion gap (hyperchloraemic) metabolic acidosis (NAGMA) — bicarbonate is lost or chloride gained: large-volume 0.9% saline resuscitation, diarrhoea, renal tubular acidosis, ureteric diversion.

A useful refinement is the delta ratio (Δ anion gap ÷ Δ bicarbonate), which detects a mixed picture — e.g. a HAGMA from sepsis coexisting with a metabolic alkalosis from vomiting.

Oxygenation indices

• The A–a gradient (alveolar–arterial oxygen difference) distinguishes hypoxaemia due to hypoventilation (normal gradient) from V/Q mismatch, shunt or diffusion defect (widened gradient — e.g. pulmonary embolism, pneumonia, pulmonary oedema, ARDS).
• The paO₂/FiO₂ (P/F) ratio grades oxygenation failure and defines ARDS severity (a key consideration in severe pre-eclampsia, sepsis and amniotic fluid embolism).
• Remember that pulse oximetry (SpO₂) and ABG paO₂ measure different things: a normal SpO₂ does not exclude CO₂ retention or a metabolic acidosis. Oximetry also fails in poor perfusion, and is unreliable in carboxyhaemoglobinaemia.

Assessment

Unlock with Pro

You found the good stuff

Unlock the full edition.

Everything behind the blur — and the study system built to walk you into the exam ready.

• Every figure & infographic, full resolution
• ~1,000 exam-format SBAs
• Spaced repetition tuned to your forgetting curve

Plan

Most popularProThe full editionR400 /monthMaxPro + early accessR800 /month

Monthly6 months−17%

🇿🇦 ZAR

Unlock Pro — R400/month→

Secured by Stripe · Cancel anytime · Discount codes apply at checkout

Drawing and handling the sample (do it right or the numbers lie)

• Radial artery is first choice (perform/confirm an Allen test for collateral ulnar flow). Use a heparinised gas syringe; expel all air bubbles immediately (air contamination falsely raises paO₂ and lowers paCO₂).
• Analyse within minutes, or place on ice — at room temperature ongoing leucocyte/erythrocyte metabolism falsely lowers pH and paO₂ and raises lactate. Never bin a sample because the value "looks wrong"; a discordant gas is a clinical signal, not noise — repeat and reconcile, do not silently discard.
• Record the FiO₂ / oxygen delivery at the moment of sampling. A paO₂ of 11 kPa means very different things on room air versus on 15 L/min via a non-rebreather.
• If arterial sampling is difficult or delayed, a venous blood gas (VBG) is a reasonable screen: venous pH runs ~0.02–0.04 lower and venous pCO₂ ~0.5–1 kPa higher than arterial (standard teaching), and venous bicarbonate and lactate track arterial well. A VBG cannot assess oxygenation — use the SpO₂ for that, or get an ABG if oxygenation is the question.

The systematic five-step interpretation (use this exact order, every time)

This disciplined sequence is what earns the HOTS marks. Verbalise it explicitly in a viva.

1. Look at the patient and the context first. Why was the gas taken? Sick, septic, bleeding, breathless, drowsy, pregnant or post-partum? FiO₂? Clinical context narrows the differential before you read a single number.
2. Assess oxygenation. Is the paO₂ adequate for the inspired oxygen? Calculate the A–a gradient or P/F ratio if hypoxaemic. Hypoxaemia on high FiO₂ = a serious gas-exchange problem (think PE, oedema, pneumonia, ARDS, AFE).
3. Look at the pH. Acidaemic (<7.35), alkalaemic (>7.45), or normal? In a term mother remember the normal is ~7.40–7.47, so 7.35 is relatively acidaemic.
4. Identify the primary disorder — match the pH to pCO₂ and HCO₃⁻.
   • Acidaemia + high pCO₂ → respiratory acidosis. Acidaemia + low HCO₃⁻ → metabolic acidosis.
   • Alkalaemia + low pCO₂ → respiratory alkalosis. Alkalaemia + high HCO₃⁻ → metabolic alkalosis.
   • Whichever value moves the pH in the same direction as the pH is the primary problem.
5. Assess compensation and look for a mixed disorder. Is the secondary system moving in the expected direction and by the expected amount? Useful rules of thumb (standard teaching):
   • Acute respiratory acidosis: HCO₃⁻ rises ~1 mmol/L per 1.3 kPa (10 mmHg) rise in pCO₂; chronic, ~3.5 mmol/L.
   • Metabolic acidosis: Winter's formula — expected pCO₂ (mmHg) ≈ 1.5 × HCO₃⁻ + 8 (±2); in kPa, expected pCO₂ ≈ (0.2 × HCO₃⁻) + 1.07. If the measured pCO₂ is higher than expected, there is a coexisting respiratory acidosis (the patient is tiring — escalate).
   • Then calculate the anion gap (and albumin-correct it) on any metabolic acidosis, and check the delta ratio if you suspect a triple disturbance.

Finally, synthesise: name the disorder(s), state the likely cause in this clinical context, and say what you will do. A complete answer sounds like: "This is a high-anion-gap metabolic acidosis with appropriate respiratory compensation by Winter's formula, lactate 6 — consistent with septic shock in this post-caesarean patient; I will resuscitate, take cultures, start antibiotics within the hour and escalate to critical care."

Worked obstetric examples

• Septic post-caesarean woman: pH 7.22, pCO₂ 3.2 kPa, HCO₃⁻ 12, lactate 7, Na 138, Cl 104. → HAGMA (AG = 138 − (104+12) = 22) with respiratory compensation; lactic acidosis of sepsis. Drill: sepsis-six / surviving-sepsis bundle.
• Eclamptic on magnesium, now drowsy and shallow: pH 7.28, pCO₂ 7.5 kPa, HCO₃⁻ 26. → Acute respiratory acidosis (hypoventilation) — suspect magnesium toxicity; check reflexes, stop infusion, give calcium gluconate, support ventilation.
• Hyperemesis, weeks of vomiting: pH 7.50, pCO₂ 5.6 kPa, HCO₃⁻ 34, Cl low, K low. → Metabolic alkalosis (loss of gastric H⁺/Cl⁻); rehydrate with saline and replace potassium.

Management

The ABG does not treat the patient — it directs treatment. Management always runs in parallel with the gas, never after it.

Resuscitate first (ABCDE)

In any acidaemic or hypoxaemic shocked patient, start the ABCDE drill simultaneously with interpreting the gas. Give controlled supplemental oxygen titrated to target saturations; secure IV access; treat the cause of shock. In pregnancy beyond ~20 weeks, 15–30° left-lateral tilt or manual uterine displacement to relieve aortocaval compression is mandatory during resuscitation.

Treat the cause, not the number

• High-anion-gap metabolic acidosis from sepsis or haemorrhage: the treatment is source control, fluids, blood and antibiotics — not bicarbonate. Routine bicarbonate infusion for lactic acidosis is not recommended; the acidosis resolves as perfusion is restored. Lactate clearance on serial gases is a good resuscitation endpoint.
• Obstetric haemorrhage: a rising lactate and worsening base deficit are early markers of inadequate resuscitation and tissue hypoperfusion — escalate the major obstetric haemorrhage protocol, give blood products, and consider tranexamic acid 1 g IV given early (within 3 hours of onset; WOMAN trial). Repeat the gas to track response.
• Respiratory acidosis from hypoventilation: identify why she is not ventilating — opioids, magnesium toxicity, exhaustion, neurological cause — and support ventilation. A rising pCO₂ in a tiring asthmatic mother is an indication to call anaesthetics/ICU for possible ventilatory support, not reassurance.
• DKA in pregnancy (including euglycaemic): fluids, fixed-rate insulin infusion, potassium replacement, and continuous fetal monitoring — the maternal acidosis is mirrored in the fetus and can cause an abnormal CTG that recovers as the mother is corrected. Do not deliver for fetal heart-rate changes until the mother is resuscitated, unless an independent obstetric indication exists.
• Metabolic alkalosis (hyperemesis, gastric loss): saline-responsive — rehydrate and correct potassium/chloride.

South African context and levels of care

• Where ABGs are available: blood-gas analysers and the skill to interpret them are reliably present at regional and tertiary hospitals and in theatre/ICU; many district hospitals and CHCs have no analyser or only an i-STAT/point-of-care device, and primary-care clinics have none. Recognising a sick acidotic patient clinically (tachypnoea, altered consciousness, poor perfusion, high modified-early-warning score) and referring early via the provincial referral pathway is therefore the core skill at lower levels — do not wait for a gas you cannot get before escalating.
• The NDoH National Integrated Maternal and Perinatal Care Guideline (NDoH, 2024) frames the recognition-and-referral of the critically ill mother and the early-warning approach to deterioration.
• Saving Mothers (NCCEMD) repeatedly identifies delayed recognition of severity and delayed referral/transfer as avoidable factors in maternal deaths from obstetric haemorrhage, hypertension and non-pregnancy-related infection (predominantly HIV-associated sepsis/pneumonia). A gas that quantifies severity is only useful if it triggers timely action.
• HIV is hugely relevant: South Africa's high antenatal HIV prevalence means lactic acidosis must be read in context — severe sepsis and Pneumocystis/bacterial pneumonia are common, and there is a (now rare with modern TLD regimens) historical association of NRTI-related lactic acidosis. Always interpret a maternal HAGMA against the HIV and TB background.
• Stock the obstetric resuscitation area per the EML (oxygen, IV fluids, blood-product access, uterotonics, tranexamic acid, magnesium sulphate, calcium gluconate, insulin, potassium).

Red flags / pitfalls

• Reading pH and paO₂ and stopping. Always complete all five steps — the anion gap and compensation calculations are where the diagnosis and the marks live.
• Forgetting pregnancy physiology. A "normal" maternal pCO₂ (~5 kPa) can mean impending respiratory failure; a maternal pCO₂ of 4 kPa and HCO₃⁻ of 19 is normal, not a metabolic acidosis. Always restate the maternal reference range.
• Treating the number, not the cause. Bicarbonate does not fix lactic acidosis; perfusion does. Calcium, not "more magnesium", treats magnesium-toxicity respiratory acidosis.
• A normal pH with deranged pCO₂ and HCO₃⁻ is not "fine" — it is either fully compensated chronic disease or, dangerously, two opposing primary disorders. Calculate compensation.
• A normalising pCO₂ in a previously hyperventilating, tiring patient (asthma, exhaustion, magnesium toxicity) is an emergency, not improvement — escalate.
• Mistaking a venous for an arterial gas, or vice versa — a VBG cannot assess oxygenation and overcalls CO₂ retention.
• Pulse oximetry false reassurance — a normal SpO₂ excludes neither CO₂ retention nor metabolic acidosis, and reads falsely in poor perfusion and dyshaemoglobinaemias.
• Sampling/analytical errors — air bubbles (falsely high paO₂, low pCO₂), delayed/unrefrigerated analysis (falsely low pH/paO₂, high lactate). If the gas is clinically implausible, repeat it correctly; never silently discard a result — a discordant gas is a signal to reconcile, not to ignore.
• Failing to escalate. At district level the right move for a sick acidotic mother is early referral, not waiting for a gas. A rising lactate or worsening base deficit despite resuscitation means call critical care now.

The deteriorating mother — drill

Recognise (tachypnoea, altered consciousness, poor perfusion, oliguria, rising EWS) → ABCDE with oxygen and left-uterine displacement → draw ABG + lactate (and FBC, U&E, glucose, cultures, coagulation) → interpret in five steps → treat the cause (source control + antibiotics for sepsis; blood + TXA + uterotonics for haemorrhage; correct DKA; reverse magnesium toxicity) → repeat the gas to track lactate/base deficit → escalate early to ICU/anaesthetics and, where indicated, refer up the level-of-care pathway. Do not wait for the perfect number to act.

Evidence anchors

• NDoH National Integrated Maternal and Perinatal Care Guideline (NDoH, 2024) — recognition, early-warning and referral of the critically ill / deteriorating mother in the SA levels-of-care system.
• Saving Mothers / NCCEMD (latest triennium) — avoidable factors in maternal death (delayed recognition of severity and delayed referral) across haemorrhage, hypertension and non-pregnancy-related infection (HIV).
• South African National HIV / ART Consolidated Guidelines (2023; TLD first-line) and SAHCS 2023 Adult ART Guidelines — context for HIV-associated sepsis/pneumonia and the historical NRTI lactic-acidosis association.
• WOMAN trial (Lancet 2017) — tranexamic acid 1 g IV given within 3 hours of postpartum haemorrhage onset reduces death from bleeding; supports early TXA in the acidotic, bleeding mother.
• South African EML — Hospital Level (Adults), current edition — oxygen, IV fluids, magnesium sulphate, calcium gluconate, insulin, uterotonics and tranexamic acid for the obstetric emergency.
• The arithmetic of ABG interpretation (Winter's formula, anion-gap and delta-ratio calculations, A–a gradient, expected compensation, and the maternal pregnancy reference ranges) is standard physiology/textbook canon rather than guideline-bound — applied here cautiously and flagged as such.