ResusDoc

BloodGas Pro

Full clinical suite · ED & ICU · Advanced interpretation

Sample type

Photo import

Add a photo of the ABG report

AI extracts values automatically

Patient context

Age (yrs)

Weight (kg)

PaCO₂ (ref 4.7–6.0 kPa)

kPa

HCO₃⁻ (ref 22–26 mmol/L)

mmol/L

Base excess (ref −2 to +2)

mmol/L

FiO₂ (room air = 0.21)

fraction

Temperature (for correction)

°C

PaO₂ (ref art 10.6–13.3 kPa)

kPa

SaO₂ (ref 94–100%)

SpO₂ (pulse oximetry)

Respiratory rate

br/min

All optional — leave blank if unavailable

Haemoglobin (Hb)

Haematocrit (Hct)

Dyshemoglobinaemias

Methaemoglobin (MetHb) (ref <2%)

Carboxyhaemoglobin (COHb) (ref <3%)

Oxyhaemoglobin (O₂Hb)

Lactate (ref <2 mmol/L)

mmol/L

All optional

Sodium (Na⁺) (135–145)

mmol/L

Potassium (K⁺) (3.5–5.0)

mmol/L

Chloride (Cl⁻) (95–110)

mmol/L

Anion gap (measured, ref 8–12 — leave blank to calculate from Na/Cl/HCO₃⁻)

mmol/L

Ionised Ca²⁺ (1.15–1.35 mmol/L)

mmol/L

Phosphate (0.8–1.5 mmol/L)

mmol/L

Albumin (35–50 g/L)

g/L

All optional

Glucose (4.0–7.0 mmol/L)

Urea (2.5–7.8 mmol/L)

Creatinine (60–120 µmol/L)

Bilirubin (<21 µmol/L)

Osmolality (275–295 mOsm/kg)

mOsm/kg

For ventilator calculations & lung-protective targets

Tidal volume (Vt)

Respiratory rate (set)

br/min

PEEP

cmH₂O

Peak pressure (Ppeak)

cmH₂O

Plateau pressure (Pplat)

cmH₂O

E:I ratio

E:I

Ventilator calculations

Run Interpret with weight and ventilator settings entered to populate this page.

Oxygen delivery & consumption

Enter Hb, SaO₂ and weight, then run Interpret to calculate DO₂, CaO₂ and O₂ extraction ratio.

Serial ABG trends

Each interpretation is logged here automatically with timestamp and trajectory arrows.

No ABGs recorded yet — run an interpretation first.

Stewart / physicochemical approach

Enter Na⁺, Cl⁻, albumin and phosphate values, then run Interpret to see Stewart analysis.

Classic clinical scenarios — quick load

Tap any scenario to auto-populate and interpret. Use as teaching anchors or differential reference.

HAGMA decision tree

High anion gap metabolic acidosis workup

1. Lactate >2 mmol/L?→ Lactic acidosis (type A: hypoperfusion; type B: liver/drugs/seizures)

2. Glucose >11 + ketones?→ DKA (also AKA if glucose normal)

3. Creatinine markedly raised?→ Uraemic acidosis

4. Osmol gap >10?→ Toxic alcohol (methanol, ethylene glycol, propylene glycol)

5. Salicylate levels?→ Salicylate toxicity (often HAGMA + respiratory alkalosis)

6. None of above?→ Pyroglutamic acidosis, D-lactic acidosis, paraldehyde

Normal AG metabolic acidosis (NAGMA)

Urine AG positive (>0)?→ RTA type 1 (distal) or type 4; urease UTI

Urine AG negative (<0)?→ GI losses (diarrhoea, fistula, ileostomy)

Post-large saline infusion?→ Dilutional / hyperchloraemic acidosis

K⁺ high + RTA features?→ RTA type 4 (Addison's, ACEi/ARB, trimethoprim)

Teaching — blood gas interpretation

A structured reference for the whole interpretive framework: the six-step read, the compensation formulas the calculator uses, gap arithmetic, oxygenation, Stewart, the pre-analytical traps, and worked cases. Tap a heading to open it. Everything here is unit-agnostic — kPa first, mmHg in brackets.

Read every gas in the same order, every time. The order matters: the pH tells you which way the patient has actually gone, and everything after it is explanation.

Step	Question	What it settles
1 · Oxygenation	Is the patient hypoxaemic, and on how much O₂?	Assessed separately from acid–base. Interpret PaO₂ only against FiO₂ (A–a gradient or P/F).
2 · pH	Acidaemia (<7.35), alkalaemia (>7.45) or normal?	The primary direction. A normal pH does not exclude a disorder — it may be two disorders cancelling.
3 · Respiratory	PaCO₂ high or low?	If the PaCO₂ shift matches the pH shift (↑CO₂ with acidaemia), the problem is respiratory.
4 · Metabolic	HCO₃⁻ / base excess high or low?	If the HCO₃⁻ shift matches the pH shift, the problem is metabolic. Matching both = mixed.
5 · Compensation	Is the secondary change the size you expect?	Under- or over-shoot means a second primary disorder. Formulas in §4.
6 · Gaps	Anion gap (albumin-corrected), delta ratio, osmolar gap, lactate.	Unmasks disorders hidden by a normal pH or normal HCO₃⁻.

Never stop at step 4. The commonest interpretive error in the ED is a "normal" gas in a patient with a high anion gap and a coexisting metabolic alkalosis — vomiting plus sepsis, or DKA plus gastric losses. Steps 5 and 6 are where mixed disorders surface.

Compensation never fully corrects, and never overshoots. If the pH has crossed to the other side of 7.40 from where the primary disorder would put it, that is not compensation — it is a second primary disorder.

Arterial, adult, at sea level

Value	SI (UK)	Conventional (US)	Note
pH	7.35 – 7.45	7.35 – 7.45	H⁺ 35–45 nmol/L. pH is logarithmic — 7.0 is twice the H⁺ of 7.3.
PaCO₂	4.7 – 6.0 kPa	35 – 45 mmHg	The only value under direct minute-ventilation control. 1 kPa = 7.5 mmHg.
PaO₂ (air)	11 – 13 kPa	80 – 100 mmHg	Meaningless without the FiO₂ it was taken on.
HCO₃⁻ (act.)	22 – 26 mmol/L	22 – 26 mEq/L	Calculated from pH and PaCO₂ by Henderson–Hasselbalch, not measured.
Standard HCO₃⁻	22 – 26 mmol/L	22 – 26 mEq/L	HCO₃⁻ the blood would have at PaCO₂ 5.3 kPa (40 mmHg) — strips out the respiratory contribution.
Base excess	−2 to +2 mmol/L	−2 to +2 mEq/L	Titratable acid/base needed to return to pH 7.40. Standard BE (SBE) corrects for whole-body buffering.
Lactate	0.5 – 2.0 mmol/L	4.5 – 18 mg/dL	>4 mmol/L is a mortality marker in sepsis independent of blood pressure.
Anion gap	8 – 12 mmol/L	8 – 12 mEq/L	Local range varies with the analyser; some modern platforms run 4–12. Always albumin-correct.
Na⁺ / K⁺ / Cl⁻	135–145 / 3.5–5.3 / 98–107	same	Gas analysers use direct ion-selective electrodes on whole blood, so a K⁺ is available in about two minutes and avoids the delay and transit haemolysis that inflate a laboratory sample. It is not immune to haemolysis in its own syringe, nor to a prolonged tourniquet or fist-clenching — a surprise result still warrants a formal lab K⁺.

Actual vs standard bicarbonate vs base excess

All three describe the metabolic component; they differ in how they handle CO₂. Actual HCO₃⁻ moves with the PaCO₂, so it rises in pure respiratory acidosis and can mislead. Standard HCO₃⁻ and base excess are CO₂-corrected, so a negative BE is metabolic acidosis regardless of ventilation. A BE more negative than −10 in a trauma patient predicts significant blood loss and transfusion requirement.

Albumin drags the base excess. Hypoalbuminaemia is itself an alkalinising force (§8), so a critically ill patient with albumin 18 g/L can have a "normal" BE while carrying a substantial unmeasured-anion acidosis. Correct the anion gap before you believe a normal gap.

A peripheral venous gas answers most acid–base questions in the ED. It answers no oxygenation question at all.

Value	Peripheral venous vs arterial	Usable?
pH	≈ 0.02 – 0.04 lower	Yes — close enough for triage and for screening acidaemia.
PCO₂	≈ 0.5 – 1.0 kPa (4 – 8 mmHg) higher on average, but the limits of agreement are wide (up to ±1.5 kPa / 11 mmHg), and they widen further in shock	Screening only. A venous PCO₂ <6 kPa (45 mmHg) reliably excludes arterial hypercapnia; a raised one must be confirmed arterially before you act on it.
HCO₃⁻ / BE	≈ 1 – 2 mmol/L higher	Yes.
Lactate	Marginally higher; agreement good if no prolonged tourniquet	Yes — venous lactate is the accepted screening test in sepsis pathways.
K⁺, Na⁺, Cl⁻, glucose, Hb	Clinically interchangeable	Yes.
PO₂ / SO₂	Reflects tissue extraction, not lung function	No. Never report a venous PO₂ as an oxygenation result.

When you still need an arterial sample

Any question about oxygenation: A–a gradient, P/F ratio, ARDS grading, shunt assessment.
Acute hypercapnic respiratory failure — titrating NIV, and any decision hinging on the exact PaCO₂ and the degree of renal compensation.
Carbon monoxide or methaemoglobinaemia (co-oximetry; venous co-oximetry is acceptable for COHb in practice, but confirm locally).
Shock states, where peripheral venous–arterial agreement degrades most.

Central venous and capillary samples

A central venous gas (SVC) tracks arterial values more tightly than a peripheral one and adds ScvO₂ as a global perfusion marker (normal 70–75%; low = inadequate delivery or high extraction, high = impaired utilisation as in cytotoxic shock or sepsis). An arterialised capillary (warmed earlobe or heel) approximates arterial pH and PCO₂ acceptably — the standard in neonates and in long-term NIV clinics — but systematically under-reads PO₂, and is unreliable if the site is cold, shut down or squeezed.

Each primary disturbance generates a predictable secondary change. Measure it. If the observed value falls outside the expected band, there is a second primary disorder.

Primary	Compensation	Expected (mmHg / SI)	Onset
Metabolic acidosis	Hyperventilation ↓PaCO₂	Winter's: PaCO₂ = 1.5 × HCO₃⁻ + 8 ± 2 SI: PaCO₂ kPa = 0.2 × HCO₃⁻ + 1.07 ± 0.27	Minutes → maximal at 12–24 h
Metabolic alkalosis	Hypoventilation ↑PaCO₂	↑0.7 mmHg per 1 mmol/L HCO₃⁻ above 24 (±5) SI: ↑0.09 kPa per 1 mmol/L	Hours; limited by hypoxic drive
Respiratory acidosis — acute	Cell buffering ↑HCO₃⁻	↑1 mmol/L per 10 mmHg (1.33 kPa) rise in PaCO₂	Immediate
Respiratory acidosis — chronic	Renal HCO₃⁻ retention	↑3.5 – 4 mmol/L per 10 mmHg (1.33 kPa) rise	Starts 6–12 h, complete 3–5 days
Respiratory alkalosis — acute	Cell buffering ↓HCO₃⁻	↓2 mmol/L per 10 mmHg (1.33 kPa) fall in PaCO₂	Immediate
Respiratory alkalosis — chronic	Renal HCO₃⁻ excretion	↓4 – 5 mmol/L per 10 mmHg (1.33 kPa) fall	2 – 3 days

Winter's in one line: if the measured PaCO₂ is higher than Winter's predicts, there is a coexisting respiratory acidosis — the patient is tiring, and this is a peri-arrest finding in DKA or salicylate poisoning. If it is lower, there is a coexisting respiratory alkalosis (sepsis, salicylates, pain, PE, liver failure, pregnancy).

A quick bedside cross-check for metabolic acidosis: the expected PaCO₂ in mmHg approximates the last two digits of the pH (pH 7.25 → PaCO₂ ≈ 25 mmHg ≈ 3.3 kPa).

Acute or chronic? The distinction changes the whole management of a hypercapnic patient. A PaCO₂ of 9 kPa (68 mmHg) with HCO₃⁻ 26 is acute and dangerous; the same PaCO₂ with HCO₃⁻ 34 is chronic and the patient has been living there. Target saturations, not a normal PaCO₂ — in known or suspected chronic hypercapnic failure aim for SpO₂ 88–92% pending the result.

Anion gap

AG = Na⁺ − (Cl⁻ + HCO₃⁻) normal 8 – 12 mmol/L

The gap is the unmeasured anions — mostly albumin, with phosphate and sulphate. A raised gap means an acid has been added whose anion the analyser does not see (lactate, ketoacids, salicylate, glycolate, formate, uraemic anions).

Albumin correction — do this every time

Corrected AG = AG + 2.5 × (40 − albumin g/L) / 10

Albumin is the dominant unmeasured anion, so each 10 g/L fall drops the apparent gap by about 2.5 mmol/L. An ICU patient with albumin 20 g/L and a measured gap of 12 has a corrected gap of 17 — a genuine HAGMA that the raw number hides.

Delta ratio — is there more than one metabolic disorder?

Δ ratio = ΔAG / ΔHCO₃⁻ = (corrected AG − 12) / (24 − HCO₃⁻)

Δ ratio	Interpretation
< 0.4	Normal anion gap (hyperchloraemic) acidosis
0.4 – 0.8	Mixed high-gap and normal-gap acidosis (e.g. DKA part-treated with 0.9% saline)
1 – 2	Pure high anion gap metabolic acidosis
> 2	HCO₃⁻ higher than the gap accounts for — a coexisting metabolic alkalosis, or a pre-existing chronic respiratory acidosis with its compensatory high HCO₃⁻

Lactic acidosis characteristically runs a delta ratio nearer 1.6 (lactate distributes and is partly buffered intracellularly), DKA nearer 1.0 (ketoanions are lost in the urine). Use the bands, not the decimal place.

Osmolar gap

Calculated osm = 2 × Na⁺ + urea + glucose (all mmol/L) · gap = measured − calculated

A gap >10 mOsm/kg suggests an unmeasured osmole: methanol, ethylene glycol, propylene glycol (a diluent in IV lorazepam and phenytoin infusions), diethylene glycol, mannitol or ethanol. A normal osmolar gap does not exclude toxic alcohol poisoning — as the parent alcohol is metabolised to its acid, the osmolar gap closes and the anion gap opens. Late presentations show a high anion gap with a normal osmolar gap.

Urine anion gap — for a normal-gap acidosis

UAG = urine Na⁺ + urine K⁺ − urine Cl⁻

A surrogate for urinary ammonium excretion. Negative ("neGUTive") = the kidney is excreting acid appropriately, so the losses are gastrointestinal — diarrhoea, ileostomy, fistula. Positive = failure of renal acid excretion, i.e. renal tubular acidosis.

High anion gap metabolic acidosis — GOLD MARK

Letter	Cause	Clue
G	Glycols — ethylene glycol, propylene glycol	Osmolar gap early; calcium oxalate crystals and AKI in ethylene glycol; propylene glycol from prolonged lorazepam/phenytoin infusions
O	Oxoproline (pyroglutamic acid)	Chronic therapeutic paracetamol in a malnourished, female or CKD patient; unexplained severe HAGMA with everything else normal
L	L-lactate	Type A — hypoperfusion, ischaemic gut, severe hypoxaemia. Type B — metformin, β₂-agonists, adrenaline, linezolid, antiretrovirals, liver failure, thiamine deficiency, seizure, malignancy
D	D-lactate	Short-bowel syndrome; encephalopathy with a normal standard lactate assay (which measures L only)
M	Methanol	Visual disturbance, "snowstorm" vision, basal ganglia infarcts
A	Aspirin / salicylate	Mixed picture: HAGMA plus primary respiratory alkalosis. Tinnitus, pyrexia, agitation
R	Renal failure	Retained sulphate, phosphate, urate; usually a modest gap
K	Ketoacidosis	Diabetic, alcoholic (glucose may be normal or low), starvation (mild), euglycaemic DKA on SGLT2 inhibitors

Euglycaemic DKA. An SGLT2 inhibitor (dapagliflozin, empagliflozin, canagliflozin) can produce full ketoacidosis with a glucose under 11 mmol/L. If you have a high anion gap acidosis with ketones and a normal glucose, check the drug chart before you discount DKA.

Normal anion gap (hyperchloraemic) acidosis — HARDUPS

Hyperalimentation / parenteral nutrition
Acetazolamide, Addison's disease, Ammonium chloride
Renal tubular acidosis (type 1 distal — hypokalaemic, nephrocalcinosis; type 2 proximal — Fanconi, myeloma, acetazolamide; type 4 — hyperkalaemic, diabetes, ACEi/ARB, spironolactone, trimethoprim)
Diarrhoea and other lower-GI losses (the commonest cause worldwide)
Ureteroenteric diversion (ileal conduit, ureterosigmoidostomy)
Pancreatic or biliary fistula / high-output drain
Saline — large-volume 0.9% NaCl resuscitation (iatrogenic, and the reason balanced crystalloids are preferred)

Metabolic alkalosis — split by urine chloride

Urine Cl⁻	Group	Causes	Treatment
< 20 mmol/L	Chloride-responsive	Vomiting, nasogastric aspiration, prior (not current) diuretics, post-hypercapnic, cystic fibrosis sweat losses, villous adenoma	Replace chloride and volume — sodium chloride and potassium chloride
> 20 mmol/L	Chloride-resistant	Primary hyperaldosteronism, Cushing's, current loop/thiazide diuretic, severe K⁺ depletion, Bartter and Gitelman syndromes, liquorice, exogenous alkali	Treat the driver; potassium replacement; consider spironolactone or acetazolamide

Respiratory disorders

Respiratory acidosis — won't breathe (opioids, sedatives, raised ICP, CNS event), can't breathe (neuromuscular disease, Guillain–Barré, myasthenia, high cord lesion, chest wall or pleural disease, obesity hypoventilation), or airway/lung failure (COPD, severe asthma, exhaustion in any cause of respiratory failure).
Respiratory alkalosis — hypoxaemia of any cause, pulmonary embolism, pain and anxiety, sepsis (early), salicylate toxicity, hepatic failure, pregnancy and progesterone, CNS lesion, over-ventilation on a ventilator, altitude.

The five mechanisms of hypoxaemia

Mechanism	A–a gradient	Responds to O₂?	Example
Hypoventilation	Normal	Yes	Opioid overdose, neuromuscular weakness
Low inspired PO₂	Normal	Yes	Altitude, inert-gas asphyxiation
V/Q mismatch	Raised	Yes	COPD, asthma, pneumonia, pulmonary embolism
Right-to-left shunt	Raised	Poorly / not at all	Consolidation, ARDS, AVM, intracardiac shunt
Diffusion limitation	Raised	Yes	Fibrosis, emphysema, exercise

Hypoxaemia that barely moves on 15 L/min via a reservoir mask is shunt until proved otherwise — and shunt is an argument for recruitment (PEEP, CPAP) rather than yet more FiO₂.

Alveolar–arterial gradient

PAO₂ = FiO₂ × (Patm − PH₂O) − PaCO₂ / R A–a = PAO₂ − PaO₂

At sea level with R = 0.8: PAO₂ (mmHg) = FiO₂ × 713 − PaCO₂/0.8, or PAO₂ (kPa) = FiO₂ × 95 − PaCO₂/0.8. The upper limit of normal on room air is roughly (age/4) + 4 mmHg — about 15 mmHg (2 kPa) in a young adult, 20+ in the elderly.

The A–a gradient inflates on supplemental oxygen and the age-based threshold no longer applies. On any FiO₂ above room air, use the P/F ratio instead.

P/F ratio and ARDS severity

Severity	PaO₂/FiO₂ (mmHg)	PaO₂/FiO₂ (kPa)
Mild	200 – 300	26.7 – 40
Moderate	100 – 200	13.3 – 26.7
Severe	≤ 100	≤ 13.3

The Berlin definition also requires onset within one week of a known insult, bilateral opacities not fully explained by effusion/collapse/nodules, respiratory failure not fully explained by cardiac failure or fluid overload, and PEEP or CPAP ≥ 5 cmH₂O. The 2023 global definition extends this to patients on high-flow nasal oxygen at ≥30 L/min and permits an SpO₂/FiO₂ ratio ≤ 315 (when SpO₂ ≤ 97%) where arterial gases or chest imaging are unavailable.

Oxyhaemoglobin dissociation curve

Right shift — offloads O₂ to tissue	Left shift — holds O₂ on the haemoglobin
↑ temperature, ↑ H⁺ / ↓ pH and ↑ CO₂ (Bohr effect), ↑ 2,3-DPG (chronic hypoxia, anaemia, altitude)	↓ temperature, alkalosis, ↓ 2,3-DPG (stored blood), fetal haemoglobin, carboxyhaemoglobin, methaemoglobin

The steep part of the curve sits below an SpO₂ of about 90% (PaO₂ ≈ 8 kPa / 60 mmHg) — which is why a fall from 90% to 85% represents a far larger drop in PaO₂ than 99% to 94%.

Pulse oximetry lies in specific situations. Carboxyhaemoglobin is read as oxyhaemoglobin, so a CO-poisoned patient shows a normal SpO₂ and a normal PaO₂ — only co-oximetry reveals it. Methaemoglobinaemia drives the reading toward a fixed ~85% regardless of true saturation (cyanosis with chocolate-brown blood unresponsive to oxygen). Poor perfusion, motion, nail varnish and darker skin pigmentation all increase the rate of occult hypoxaemia — when the clinical picture and the oximeter disagree, take a gas.

The traditional approach treats HCO₃⁻ as a cause. Stewart argues it is a dependent variable — a consequence of water dissociation — and that plasma pH is determined by only three independent variables:

Independent variable	Definition	Effect of an increase
PCO₂	Respiratory component	Acidifying
SID (strong ion difference)	(Na⁺+K⁺+Ca²⁺+Mg²⁺) − (Cl⁻ + lactate⁻); apparent SID ≈ 40–42 mmol/L	Alkalinising — a fall in SID (usually a rise in Cl⁻) is acidifying
A_TOT (total weak acid)	Albumin and phosphate	Acidifying — so hypoalbuminaemia is alkalinising

SIG (strong ion gap) = SID apparent − SID effective normal ≈ 0 – 2 mmol/L

A raised SIG is the Stewart equivalent of a raised albumin-corrected anion gap: unmeasured anions are present. Where Stewart earns its keep is in the critically ill patient with several forces cancelling — for example a septic patient with lactate 5, albumin 16 and chloride 118, whose pH and base excess look unremarkable because the hypoalbuminaemic alkalosis is masking both a lactic and a hyperchloraemic acidosis. Quantifying each force separately shows three problems where the pH shows none.

Practical bottom line. For most gases the base-excess and anion-gap approach gives the same clinical answer faster. Reach for Stewart when the numbers do not add up: severe hypoalbuminaemia, large-volume crystalloid resuscitation, or an unexplained acidosis with a normal-looking gap.

Sampling and handling

Problem	Effect on the result
Air bubble left in the syringe	PO₂ pulled toward ~20 kPa (150 mmHg) — falsely high if the patient is hypoxaemic, falsely low if on high FiO₂ — and PCO₂ falsely low. Expel bubbles immediately.
Delay at room temperature	Ongoing cellular metabolism: PO₂ falls, PCO₂ rises, lactate rises, glucose falls. Analyse within 15 minutes, or ice the sample.
Marked leucocytosis or thrombocytosis	"Leucocyte larceny" — rapid ex vivo O₂ consumption gives a spuriously low PaO₂. Run immediately.
Excess liquid heparin	Dilutional: falsely low PCO₂, HCO₃⁻ and electrolytes. Use dry balanced-heparin syringes and fill to the mark.
Inadvertent venous sample	The commonest cause of an "inexplicable" gas. Suspect it when the PO₂ is low but the patient is pink with a normal SpO₂.
Prolonged tourniquet or fist-clenching (VBG)	Falsely raised lactate and K⁺.

Interpretive errors

Treating a normal pH as a normal patient. Two opposing primary disorders produce a normal pH. Always run the gaps.
A "normal" PaCO₂ in severe asthma or a severe exacerbation. These patients should be hyperventilating; a normalising PaCO₂ means exhaustion and is a peri-arrest sign.
Calling a raised lactate "hypoperfusion" without thinking. Type B causes — metformin, β₂-agonist infusions, adrenaline, seizures, thiamine deficiency, liver failure, malignancy — need a different response entirely.
Chasing the number in DKA. Fluid and insulin correct the acidosis; bicarbonate is not routinely indicated and worsens intracellular acidosis, hypokalaemia and cerebral oedema risk.
Correcting a chronic hypercapnia to normal. You remove the respiratory drive and unmask a profound metabolic alkalosis. Target saturations and the patient's own baseline.
Ignoring the anion gap when the HCO₃⁻ is normal. A patient vomiting with sepsis can have HCO₃⁻ 24 with a corrected gap of 22.
Blaming the gas for the drug chart. Acetazolamide, topiramate, metformin, SGLT2 inhibitors, propylene-glycol-containing infusions and prolonged normal saline all have signature patterns.

Work each one through the six steps before revealing. Values are given in SI with mmHg in brackets.

Case 1. 19-year-old, two days of vomiting and polyuria, Kussmaul breathing. pH 7.07 · PaCO₂ 1.9 kPa (14) · HCO₃⁻ 4 · Na⁺ 134 · Cl⁻ 96 · K⁺ 5.6 · glucose 31 · albumin 42 · lactate 1.8

Severe HAGMA with appropriate respiratory compensation — DKA. AG = 134 − (96+4) = 34; albumin is normal so no correction. Winter's: expected PaCO₂ = 1.5×4 + 8 = 14 mmHg (1.9 kPa) — the measured value matches exactly, so compensation is appropriate and there is no additional respiratory disorder. Delta ratio = (34−12)/(24−4) = 1.1 → pure HAGMA, no hidden second metabolic process. The K⁺ of 5.6 is a total-body deficit shifted extracellularly by the acidosis and insulin lack — it will fall fast once insulin starts. Fluid and fixed-rate insulin; no bicarbonate.

Case 2. 71-year-old with COPD, two days more breathless, drowsy on arrival, given 15 L/min via reservoir mask in the ambulance. pH 7.21 · PaCO₂ 10.4 kPa (78) · PaO₂ 18 kPa (135) · HCO₃⁻ 31 · BE +4

Acute-on-chronic respiratory acidosis, worsened by uncontrolled oxygen. Acute compensation alone would give HCO₃⁻ ≈ 24 + (78−40)/10 = 28; full chronic compensation would give ≈ 24 + 3.5×3.8 = 37. The measured 31 sits between the two — a chronic retainer with an acute deterioration on top. The PaO₂ of 18 kPa is far above target and is contributing through loss of hypoxic drive, worsened V/Q matching and the Haldane effect. Titrate down to SpO₂ 88–92% with a 28% Venturi, treat the exacerbation, and start NIV if the acidosis persists after initial treatment.

Case 3. 34-year-old found confused, tinnitus, pyrexial, hyperventilating. Empty paracetamol-and-something packet. pH 7.43 · PaCO₂ 2.5 kPa (19) · HCO₃⁻ 12 · Na⁺ 140 · Cl⁻ 102 · albumin 40 · lactate 2.4 · glucose 4.1

Mixed primary respiratory alkalosis and high anion gap metabolic acidosis — salicylate toxicity. The pH is normal, which is the trap. AG = 140 − (102+12) = 26 → a substantial HAGMA. But Winter's predicts a PaCO₂ of 1.5×12 + 8 = 26 mmHg (3.5 kPa), and the measured value is 19 mmHg (2.5 kPa) — far lower than compensation explains, so there is a second primary disorder, a respiratory alkalosis from direct respiratory-centre stimulation. This combination with tinnitus and pyrexia is salicylate until disproved. Send a salicylate level (and repeat it — absorption is erratic), alkalinise the urine, keep the patient breathing, and discuss haemodialysis early. Do not intubate casually: losing the hyperventilation collapses the pH.

Case 4. Day 3 on ICU, septic, has had 6 L of 0.9% sodium chloride. pH 7.33 · PaCO₂ 4.7 kPa (35) · HCO₃⁻ 18 · Na⁺ 143 · Cl⁻ 117 · albumin 18 · lactate 2.1

Hyperchloraemic (normal-gap) acidosis with a concealed high-gap component, unmasked by albumin correction. Raw AG = 143 − (117+18) = 8, apparently normal. Corrected for an albumin of 18: 8 + 2.5×(40−18)/10 = 13.5 — now raised. Delta ratio = (13.5−12)/(24−18) = 0.25, i.e. predominantly a normal-gap process from the chloride load, with a smaller unmeasured-anion contribution. In Stewart terms the SID has fallen (Cl⁻ 117) while a low A_TOT (albumin 18) is pushing the other way, which is exactly why the pH looks milder than the physiology. Switch to a balanced crystalloid and keep treating the sepsis; the acidosis is not a reason to give bicarbonate.

British Thoracic Society / Intensive Care Society. Guideline for oxygen use in adults in healthcare and emergency settings. BTS, 2017 (and subsequent BTS quality standards) — oxygen targets, 88–92% in those at risk of hypercapnic respiratory failure.
ARDS Definition Task Force. Acute respiratory distress syndrome: the Berlin definition. JAMA 2012;307:2526–33.
Matthay MA et al. A new global definition of ARDS. Am J Respir Crit Care Med 2023;207:37–47.
Albert MS, Dell RB, Winters RW. Quantitative displacement of acid–base equilibrium in metabolic acidosis. Ann Intern Med 1967;66:312–22 — the original of "Winter's formula".
Stewart PA. Modern quantitative acid–base chemistry. Can J Physiol Pharmacol 1983;61:1444–61; and Kellum JA, Elbers PWG (eds). Stewart's Textbook of Acid–Base. 2nd ed, 2009.
Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med 1998;26:1807–10 — the 2.5 per 10 g/L correction.
Byrne AL et al. Peripheral venous and arterial blood gas analysis in adults: a systematic review. Respirology 2014;19:168–75.
Batlle DC, Hizon M, Cohen E, Gutterman C, Gupta R. The use of the urinary anion gap in the diagnosis of hyperchloremic metabolic acidosis. N Engl J Med 1988;318:594–9.
Joint British Diabetes Societies. The management of diabetic ketoacidosis in adults. JBDS-IP, current edition — including euglycaemic DKA and SGLT2 inhibitors.
Balanced crystalloid vs saline, for iatrogenic hyperchloraemic acidosis: SMART (Semler MW et al. N Engl J Med 2018;378:829–39), BaSICS (Zampieri FG et al. JAMA 2021;326:1–12), PLUS (Finfer S et al. N Engl J Med 2022;386:815–26).

Formulas implemented in the calculator are listed in full in §4, §5 and §7 so that any result this tool produces can be reproduced by hand. Where UK and international practice differ, UK practice is followed.

⚕ Teaching material for registered clinicians. Clinical decision support only — always correlate with the patient in front of you, local guidelines and senior review.

Blood Gas Interpreter — ABG, VBG and CBG Analysis with Teaching Notes