Arterial blood gas analysis

In short

Arterial blood gas (ABG) analysis measures the pH, partial pressures of carbon dioxide and oxygen, and the derived bicarbonate and base excess of arterial blood, giving a direct read-out of pulmonary gas exchange and acid-base balance. It is the single most used bedside investigation in ICUs, casualty, operation theatres and dialysis units, and the respiratory therapist is usually the person who draws the sample, runs the analyser and reports the interpretation. It is examined in B.Sc Respiratory Therapy / Cardio-Pulmonary Technology papers on respiratory physiology and critical care, and is almost guaranteed in placement technical rounds and viva for hospital technologist posts. Employers test it as a practical skill: a radial puncture with a modified Allen test, correct heparinised syringe handling, and a four-step interpretation of a printed report. Errors are clinically dangerous — a mis-read ABG can lead to wrong ventilator settings or wrong bicarbonate therapy.

Explained

What is measured, what is calculated

• Directly measured (electrodes): pH (glass electrode, potentiometric), PaCO₂ (Severinghaus CO₂ electrode, potentiometric), PaO₂ (Clark polarographic electrode, amperometric). Modern analysers add Na⁺, K⁺, Cl⁻, Ca²⁺, glucose, lactate by ion-selective electrodes/biosensors.
• Calculated (derived): HCO₃⁻, total CO₂, base excess/base deficit, standard bicarbonate, SaO₂ (unless co-oximetry is fitted), A-a gradient, anion gap. HCO₃⁻ is never measured in an ABG — it comes from the Henderson-Hasselbalch equation, so an error in pH or PaCO₂ propagates into it.
• Measured by co-oximetry (spectrophotometry): total haemoglobin, O₂Hb, COHb, MetHb, HHb. Needed when carbon monoxide poisoning or methaemoglobinaemia is suspected — a pulse oximeter and a calculated SaO₂ both read falsely normal there.
• Normal adult arterial values (room air, sea level): pH 7.35-7.45; PaCO₂ 35-45 mmHg (4.7-6.0 kPa); PaO₂ 80-100 mmHg; HCO₃⁻ 22-26 mEq/L; base excess −2 to +2 mEq/L; SaO₂ 95-100%; anion gap 8-12 mEq/L (12-16 if K⁺ included). Venous: pH ≈ 7.32-7.38, PvCO₂ ≈ 42-50, PvO₂ ≈ 35-45, SvO₂ ≈ 70-75%.
• At higher altitude (Shimla, Leh) normal PaO₂ and PaCO₂ are lower because barometric pressure is lower; reference ranges must be altitude-corrected.

Sampling: sites, technique, pre-analytical errors

1. Site: radial artery is first choice (superficial, collateral ulnar supply, compressible); then brachial, femoral (highest complication rate), dorsalis pedis. Neonates: umbilical artery catheter or radial; heel-prick capillary gas is acceptable for pH and PCO₂ but not for PO₂.
2. Modified Allen test before radial puncture: compress radial and ulnar arteries, have the patient clench and open the fist, release the ulnar artery — colour should return to the palm in under 5-15 seconds, confirming collateral flow. A negative (slow) test means pick another site. It screens collateral flow, not puncture success.
3. Equipment: pre-heparinised 1-3 mL ABG syringe (dry lithium/electrolyte-balanced heparin), 23-25 G needle, 30-45° angle for radial, no suction needed — arterial pressure fills the syringe. Expel all air bubbles immediately, cap, mix by rolling (not shaking).
4. Analyse within 15 minutes at room temperature. If delay is unavoidable, cap and store in ice slurry (0-4 °C) for up to 1 hour — but iced plastic syringes allow O₂ to diffuse in and are not recommended for high-PaO₂ samples.
5. Post-puncture: firm pressure 3-5 min (5-10 min if on anticoagulants/thrombolytics), then pressure dressing; observe for haematoma, arterial spasm, distal ischaemia.
6. Pre-analytical errors and their direction:
   • Air bubbles: PaO₂ rises towards 150 mmHg, PaCO₂ falls, pH rises.
   • Excess liquid heparin: dilutional fall in PaCO₂ and HCO₃⁻, pH shifts, K⁺/Ca²⁺ falsely low.
   • Delay at room temperature: leucocyte/RBC metabolism consumes O₂ and makes acid — PaO₂ and pH fall, PaCO₂ rises. Exaggerated in leucocytosis/thrombocytosis.
   • Venous admixture: low PaO₂ with near-normal pH; suspect when PaO₂ is unexpectedly low on high FiO₂.
   • Clot / under-filled syringe: analyser blockage, unreliable values.
   • Temperature: analysers report at 37 °C; in fever or therapeutic hypothermia the uncorrected ("alpha-stat") values are what most ICUs use — state which convention is reported.
7. Contraindications/cautions: absent collateral flow, local infection or burn, AV fistula limb, severe coagulopathy, thrombolytic therapy, Raynaud's/Buerger's disease at that site.

Acid-base physiology and buffering

• Acidaemia/alkalaemia describe the blood pH; acidosis/alkalosis describe the process. Both can coexist in a mixed disorder with a normal pH.
• Buffer systems: bicarbonate/carbonic acid (chief extracellular buffer, open system because CO₂ is exhaled), haemoglobin and plasma proteins, phosphate, and bone carbonate for chronic loads.
• Three lines of defence: chemical buffering (seconds) → respiratory compensation by changing alveolar ventilation (minutes) → renal compensation by H⁺ excretion and HCO₃⁻ reabsorption/regeneration (hours to days, 3-5 days for full effect).
• Respiratory component = PaCO₂ (controlled by alveolar ventilation; PaCO₂ ∝ CO₂ production / alveolar ventilation). Metabolic component = HCO₃⁻ / base excess.
• Compensation never fully corrects pH (except in chronic respiratory alkalosis, which can come close). A pH that has crossed to the other side of 7.40 means a second primary disorder, not overcompensation.
• Base excess (BE) is the amount of strong acid needed to titrate blood back to pH 7.40 at PaCO₂ 40 mmHg and 37 °C — a PaCO₂-independent measure of the metabolic component. Standard base excess (SBE) assumes haemoglobin 5 g/dL to represent extracellular fluid.
• Anion gap detects unmeasured anions in metabolic acidosis; correct it for albumin (add ~2.5 mEq/L per 1 g/dL fall below 4 g/dL) — important in Indian ICU patients who are often hypoalbuminaemic.
• Stewart (physicochemical) approach: pH is determined by PCO₂, strong ion difference (SID) and total weak acid (A_tot); explains dilutional and hyperchloraemic acidosis of large-volume normal saline. Examined descriptively in most syllabi.

Classification of acid-base disorders and causes

• High anion gap metabolic acidosis (MUDPILES/GOLDMARK): methanol, uraemia, DKA, propylene glycol, isoniazid/iron, lactate, ethylene glycol, salicylate. INH poisoning and organophosphate poisoning are common in Indian casualty.
• Normal anion gap (hyperchloraemic) acidosis: diarrhoea, renal tubular acidosis, acetazolamide, ureteric diversion, saline resuscitation, early CKD.
• Mixed disorders: cardiac arrest (respiratory + metabolic acidosis), salicylate poisoning (respiratory alkalosis + high AG acidosis), COPD on diuretics (respiratory acidosis + metabolic alkalosis), sepsis with vomiting.
• Delta ratio / delta-delta (Δ anion gap ÷ Δ HCO₃⁻) distinguishes a pure high-AG acidosis (0.8-2) from a coexisting normal-AG acidosis (<0.8) or metabolic alkalosis (>2).

Step-by-step interpretation

1. Check the sample and context: patient, FiO₂, ventilator settings, temperature, whether arterial, time of draw. Verify internal consistency with Henderson-Hasselbalch before trusting an odd report.
2. Look at pH: <7.35 acidaemia, >7.45 alkalaemia, 7.35-7.45 normal (may still be a mixed or fully compensated disorder).
3. Identify the primary disorder: which of PaCO₂ or HCO₃⁻ has moved in the direction that explains the pH. CO₂ moves with pH in a metabolic disorder; CO₂ moves opposite to pH in a respiratory disorder (the "ROME" mnemonic: Respiratory Opposite, Metabolic Equal).
4. Assess compensation against the expected formula (below). Less than expected or more than expected ⇒ second disorder.
5. If metabolic acidosis, calculate the anion gap (albumin-corrected), then the delta ratio; add lactate and ketones.
6. Assess oxygenation separately: PaO₂ against FiO₂ (P/F ratio), A-a gradient, SaO₂ and the position of the oxyhaemoglobin dissociation curve; then decide oxygen therapy or ventilatory support.
7. Correlate and act: treat the cause, not the number. Bicarbonate is reserved for severe acidaemia (generally pH <7.1-7.2) and specific situations; it is harmful in DKA and in hypercapnia with inadequate ventilation.

Oxygenation indices and the dissociation curve

• Hypoxaemia = low PaO₂; mild 60-79, moderate 40-59, severe <40 mmHg (Indian practice follows these broad bands; thresholds vary by text).
• Five mechanisms of hypoxaemia: low inspired PO₂ (altitude), hypoventilation (raised PaCO₂, normal A-a gradient), diffusion limitation, V/Q mismatch (commonest, responds to O₂), true shunt (poor response to O₂). Hypoxaemia responsive to oxygen = V/Q mismatch; unresponsive = shunt.
• A-a gradient rises with age; a normal young adult on room air has 5-10 mmHg. A normal gradient with hypercapnia points to pure hypoventilation (e.g. opioid overdose); a wide gradient points to lung disease.
• P/F ratio grades ARDS by the Berlin definition with PEEP ≥5 cmH₂O: mild 201-300, moderate 101-200, severe ≤100 mmHg. S/F ratio is used when ABG is unavailable.
• Oxyhaemoglobin dissociation curve: sigmoid; P50 ≈ 26-27 mmHg. Right shift (unloads O₂ to tissue): ↑ temperature, ↑ PaCO₂, ↓ pH (Bohr effect), ↑ 2,3-DPG. Left shift: opposite, plus stored blood, fetal Hb, methaemoglobin, COHb. Key clinical points: PaO₂ 60 ≈ SaO₂ 90%, PaO₂ 40 ≈ 75%, PaO₂ 27 ≈ 50%.
• Haldane effect: deoxygenated haemoglobin carries CO₂ more readily — why CO₂ is loaded in tissues and released in the lung.
• Oxygen delivery depends far more on haemoglobin and cardiac output than on PaO₂; an anaemic patient can have a perfect ABG and tissue hypoxia. Check lactate and ScvO₂.

Analyser quality, standards and Indian practice

• Calibration: one- and two-point automatic calibration with precision gas mixtures (typically 5% CO₂ / 12% O₂ and 10% CO₂ / 0% O₂ balance N₂) and tonometered or aqueous buffers (pH 6.838 and 7.383); internal QC at defined intervals, plus external quality assurance programmes (CMC Vellore EQAS, AIIMS, NABL-recognised schemes).
• Point-of-care analysers (cartridge-based, e.g. handheld units in casualty and ambulances) need the same QC discipline as the main laboratory instrument.
• Standards and regulation: laboratories are accredited to ISO 15189 for medical laboratories (voluntary, through NABL in India); hospitals to NABH standards (voluntary, though linked to CGHS/Ayushman Bharat empanelment). Blood gas analysers are notified medical devices regulated under the Medical Devices Rules, 2017 by CDSCO, with licensing of manufacture/import. Biomedical waste from ABG sampling (needles, syringes, blood) is segregated under the Bio-Medical Waste Management Rules, 2016 as amended — sharps in white puncture-proof containers, blood-contaminated items in red/yellow as per the schedule. Universal/standard precautions apply.
• Allied health practice: the National Commission for Allied and Healthcare Professions Act, 2021 governs registration of respiratory therapy and cardio-pulmonary technology professionals, with state registers and a central register; the detailed scope-of-practice and curriculum documents under it are still being rolled out, so check the current NCAHP notifications for your category.
• Units: SI reporting uses kPa for partial pressures (1 kPa = 7.5 mmHg) and mmol/L for HCO₃⁻; most Indian hospitals print mmHg and mEq/L. Always read the header.

Key points

• ABG - arterial sample measuring pH, PaCO₂, PaO₂ directly; HCO₃⁻, BE and SaO₂ are calculated unless co-oximetry is present.
• Normal values - pH 7.35-7.45, PaCO₂ 35-45 mmHg, PaO₂ 80-100 mmHg, HCO₃⁻ 22-26 mEq/L, BE ±2, SaO₂ 95-100%.
• Modified Allen test - pre-radial-puncture check of ulnar collateral flow; palmar colour returns within about 5-15 seconds.
• Air bubble - the commonest pre-analytical error; raises PaO₂, lowers PaCO₂, raises pH.
• 15-minute rule - analyse promptly; delay lowers pH and PaO₂ and raises PaCO₂ through cell metabolism.
• ROME - Respiratory disorders move PaCO₂ Opposite to pH; Metabolic disorders move PaCO₂ in the same direction (Equal).
• Compensation - respiratory in minutes, renal in 3-5 days; it never overshoots past pH 7.40.
• Base excess - PaCO₂-independent measure of the metabolic component; negative = base deficit = metabolic acidosis.
• Anion gap - Na⁺ − (Cl⁻ + HCO₃⁻), normal 8-12 mEq/L; correct upward for hypoalbuminaemia.
• Delta ratio - ΔAG/ΔHCO₃⁻; <0.8 suggests added normal-AG acidosis, >2 suggests coexisting metabolic alkalosis.
• A-a gradient - normal with hypercapnia means hypoventilation; widened means a gas-exchange problem.
• Shunt vs V/Q mismatch - hypoxaemia that corrects with oxygen is V/Q mismatch; refractory hypoxaemia is shunt.
• P50 ≈ 27 mmHg - right shift (acidosis, fever, high CO₂, high 2,3-DPG) releases O₂ to tissues; left shift holds it.
• PaO₂ 60 = SaO₂ 90% - the steep part of the curve begins; below this SaO₂ falls fast.
• Compensated vs corrected - a normal pH with abnormal PaCO₂ and HCO₃⁻ is compensated, not cured.
• Pulse oximeter limits - cannot see COHb, MetHb, hypercapnia or acid-base status; ABG is needed.
• Bicarbonate therapy - generally for pH below about 7.1-7.2 or specific poisonings; avoid routinely in DKA and uncorrected hypercapnia.
• Regulation - NABL/ISO 15189 for the lab, CDSCO Medical Devices Rules 2017 for the analyser, BMW Rules 2016 (as amended) for the sharps and blood.

Formulas and examples

Example 1 - Compensated COPD (chronic respiratory acidosis). pH 7.34, PaCO₂ 62 mmHg, HCO₃⁻ 32 mEq/L, PaO₂ 58 mmHg on room air, age 64.

• pH 7.34 → mild acidaemia.
• PaCO₂ raised (62) and pH low ⇒ primary respiratory acidosis. PaCO₂ moved opposite to pH.
• Rise in PaCO₂ = 62 − 40 = 22 mmHg. Chronic expectation: HCO₃⁻ rises 3.5 per 10 → 24 + (2.2 × 3.5) = 24 + 7.7 = 31.7 ≈ 32 — matches. Acute expectation would be only 24 + 2.2 = 26.2.
• Answer: chronic (compensated) respiratory acidosis with hypoxaemia. A-a gradient: PAO₂ = 0.21 × (760 − 47) − 62/0.8 = 149.7 − 77.5 = 72.2; A-a = 72.2 − 58 = 14.2 mmHg (expected ≈ 64/4 + 4 = 20) — gradient not widened, so hypoventilation dominates. Controlled O₂ 24-28% targeting SpO₂ 88-92%, treat the exacerbation, consider NIV.

Example 2 - Diabetic ketoacidosis with Winter's formula. pH 7.18, PaCO₂ 22 mmHg, HCO₃⁻ 8 mEq/L, Na⁺ 134, Cl⁻ 98, K⁺ 5.2, albumin 3.0 g/dL.

• pH 7.18 → acidaemia. HCO₃⁻ 8 (low) explains it ⇒ metabolic acidosis; PaCO₂ low, same direction as pH.
• Winter's: expected PaCO₂ = 1.5 × 8 + 8 = 12 + 8 = 20 ± 2, i.e. 18-22. Measured 22 → appropriate respiratory compensation, no added respiratory disorder.
• AG = 134 − (98 + 8) = 28 mEq/L (high). Albumin-corrected = 28 + 2.5 × (4.0 − 3.0) = 30.5.
• Delta ratio = (28 − 12)/(24 − 8) = 16/16 = 1.0 → pure high-AG metabolic acidosis.
• Answer: high anion gap metabolic acidosis with appropriate respiratory compensation — treat with fluids, insulin infusion and potassium replacement; no bicarbonate at this pH.

Example 3 - Cross-check with Henderson-Hasselbalch. Report: pH 7.40, PaCO₂ 40 mmHg, HCO₃⁻ 24 mEq/L.

• 0.03 × 40 = 1.2 mmol/L dissolved CO₂. Ratio 24/1.2 = 20. log₁₀ 20 = 1.301.
• pH = 6.1 + 1.301 = 7.40 — internally consistent. The 20:1 bicarbonate-to-carbonic-acid ratio is what holds pH at 7.4.

Example 4 - Oxygen content and the anaemia trap. Patient A: Hb 15 g/dL, SaO₂ 98%, PaO₂ 95 mmHg. Patient B: Hb 7 g/dL, SaO₂ 98%, PaO₂ 95 mmHg.

• A: CaO₂ = 1.34 × 15 × 0.98 + 0.003 × 95 = 19.70 + 0.285 = 19.98 ≈ 20.0 mL/dL.
• B: CaO₂ = 1.34 × 7 × 0.98 + 0.285 = 9.19 + 0.285 = 9.48 mL/dL — less than half, with an identical-looking ABG.
• If B's cardiac output is 5 L/min, DO₂ = 9.48 × 5 × 10 = 474 mL/min against A's 999 mL/min. Oxygenation is adequate; oxygen delivery is not.

Example 5 - Mixed disorder in salicylate poisoning. pH 7.46, PaCO₂ 20 mmHg, HCO₃⁻ 14 mEq/L, Na⁺ 140, Cl⁻ 102, albumin 4.0.

• pH 7.46 → alkalaemia. PaCO₂ 20 is low and opposite in direction ⇒ primary respiratory alkalosis.
• Expected HCO₃⁻ for an acute 20 mmHg fall = 24 − (2 × 2) = 20; for chronic = 24 − (2 × 4.5) ≈ 15. Measured 14 is below even the chronic expectation.
• AG = 140 − (102 + 14) = 24 mEq/L (high) ⇒ a coexisting high anion gap metabolic acidosis.
• Answer: mixed respiratory alkalosis + high AG metabolic acidosis — the classic salicylate pattern; check salicylate level and lactate, alkalinise urine.

Where it is used

• ICU and HDU: titrating ventilator settings (minute ventilation for PaCO₂, FiO₂/PEEP for PaO₂), weaning decisions, permissive hypercapnia in ARDS, P/F ratio for ARDS grading — in AIIMS, PGIMER, CMC Vellore, Apollo, Fortis, Manipal and district-hospital ICUs.
• Casualty and ambulance services (108/Dial 112 services): poisoning (organophosphate, INH, methanol), DKA, shock, cardiac arrest, trauma — point-of-care analysers.
• Operation theatre and cardiac surgery/perfusion: on-pump gas management, ACT and electrolyte control, post-bypass acid-base correction; ABG every 20-30 minutes during cardiopulmonary bypass.
• Pulmonology and sleep medicine: assessing long-term oxygen therapy eligibility, NIV initiation in hypercapnic COPD and obesity hypoventilation, pre-operative risk assessment.
• Nephrology and dialysis units: uraemic and renal tubular acidoses, dialysate bicarbonate adjustment, post-dialysis alkalosis.
• Job roles: respiratory therapist, cardiac/ICU technologist, perfusionist, anaesthesia technologist, emergency technician, clinical application specialist with analyser companies (Radiometer, Siemens Healthineers, Werfen, Nova, Roche), and NABL/NABH lab quality roles.

Interview questions

• Q: Which ABG values are measured and which are calculated? A: pH, PaCO₂ and PaO₂ are measured by electrodes; HCO₃⁻, base excess, total CO₂ and (without co-oximetry) SaO₂ are calculated, mainly via the Henderson-Hasselbalch equation.
• Q: Why do you perform a modified Allen test? A: To confirm adequate ulnar collateral circulation to the hand before puncturing the radial artery, so that arterial spasm or thrombosis does not cause hand ischaemia.
• Q: What happens if an air bubble is left in the syringe? A: Equilibration with room air raises PaO₂ towards 150 mmHg and lowers PaCO₂, which also falsely raises pH — so bubbles must be expelled immediately.
• Q: How do you tell acute from chronic respiratory acidosis on an ABG? A: By the bicarbonate rise per 10 mmHg of PaCO₂ — about 1 mEq/L acutely and 3.5-4 mEq/L when renal compensation has had 3-5 days.
• Q: State Winter's formula and its use. A: Expected PaCO₂ = 1.5 × HCO₃⁻ + 8 ± 2; if the measured PaCO₂ is higher there is a coexisting respiratory acidosis, if lower a coexisting respiratory alkalosis.
• Q: A patient has pH 7.40, PaCO₂ 60 and HCO₃⁻ 36. What is going on? A: A normal pH with both values far from normal means a mixed disorder — chronic respiratory acidosis with metabolic alkalosis, for example COPD on a loop diuretic.
• Q: Why is the anion gap corrected for albumin? A: Albumin is the main unmeasured anion; hypoalbuminaemia lowers the measured gap and can hide a high-anion-gap acidosis, so add about 2.5 mEq/L per 1 g/dL fall below 4 g/dL.
• Q: How do you distinguish shunt from V/Q mismatch using ABG? A: Hypoxaemia from V/Q mismatch improves markedly with supplemental oxygen; a true shunt shows little improvement despite a high FiO₂, and the A-a gradient stays wide.
• Q: A pulse oximeter shows 99% but the patient is cyanosed and hypoxic. What do you suspect? A: Carbon monoxide poisoning or methaemoglobinaemia — the oximeter and calculated SaO₂ are misleading; run co-oximetry on an ABG.
• Q: What does base excess tell you that bicarbonate does not? A: Base excess quantifies the metabolic component independently of PaCO₂, whereas bicarbonate itself shifts with acute changes in CO₂.
• Q: When would you give sodium bicarbonate? A: Only for severe acidaemia (generally pH below about 7.1-7.2) with adequate ventilation, or specific indications such as tricyclic or salicylate poisoning and hyperkalaemia — not routinely in DKA or lactic acidosis.
• Q: How do you handle a sample that cannot be run within 15 minutes? A: Cap it, note the time, and if the delay exceeds about 15 minutes store it in ice slurry for up to an hour — accepting that iced plastic syringes can falsely raise PaO₂ — and repeat if the result does not fit the patient.
• Q: Why does the oxyhaemoglobin curve shift right in exercising muscle? A: Higher temperature, higher CO₂, lower pH (Bohr effect) and raised 2,3-DPG reduce haemoglobin's affinity, so more oxygen is released to the tissue.

Common mistakes

• Treating a normal pH as a normal ABG and missing a fully compensated or mixed disorder.
• Calling a disorder "overcompensated" when the pH has crossed 7.40 — compensation never overshoots; it is a second primary disorder.
• Forgetting the anion gap in every metabolic acidosis, and forgetting to correct it for low albumin.
• Confusing acidaemia with acidosis, or respiratory compensation with correction of the primary problem.
• Interpreting PaO₂ without recording FiO₂, PEEP, altitude and posture — a PaO₂ of 90 on 60% oxygen is abnormal.
• Technique errors: liquid-heparin dilution, air bubbles, under-filled or clotted syringes, inadequate post-puncture pressure causing haematoma, and puncturing without checking collateral flow or coagulation status.
• Equating a good SpO₂ or PaO₂ with adequate tissue oxygenation while ignoring haemoglobin, cardiac output and lactate.
• Mixing units — using mmHg values in a kPa formula, or mmol/L and mEq/L interchangeably without checking the report header.

Related topics

Acid-base physiology and renal compensation · Pulmonary function tests · Oxygen therapy, aerosol therapy and airway clearance · Mechanical ventilation basics · Capnography and pulse oximetry monitoring · Cardiopulmonary bypass gas management