Mixed Acid–Base Disorders & the Delta–Delta — The 5-Step Approach · ICU Topics

Overview & definition

A mixed acid-base disorder is the presence of two or more primary acid-base disturbances simultaneously. The single most important principle: the body never over-compensates. The physiological compensation is always in the direction of the primary disorder, and it is limited — if the measured compensation is more or less than the expected, a second primary disorder is present. The mixed disorders are common in the ICU (the sepsis, the multi-organ failure, the vomiting, the diarrhoea, the drugs) and the systematic 5-step approach uncovers them.[1]

Clean medical illustration of an arterial blood-gas analyser readout showing a mixed-disorder pattern, a clinical workstation displaying the anion-gap and the delta-delta ratio calculations, clinical-blue lighting, a careful analytical mood — FigureThe mixed acid-base disorders — the 5-step approach and the delta-delta. The body never over-compensates; if the compensation is off, a second disorder is present.

The 5-step approach

Three-panel infographic on a white clinical-blue background: LEFT steps 1-3 (pH to acidosis/alkalosis; pCO2/bicarbonate to primary; compensation Winter formula expected pCO2 equals 1.5 times bicarbonate plus 8 plus-or-minus 2, if actual higher then concurrent respiratory acidosis, if lower then respiratory alkalosis); CENTRE steps 4-5 the gaps (anion gap equals Na minus (Cl plus bicarbonate) 8 to 12 albumin-corrected; delta-delta ratio of delta AG to delta bicarbonate: about 1 pure high-AG, under 1 concurrent normal-gap hyperchloraemic acidosis, over 2 concurrent metabolic alkalosis); RIGHT common patterns (sepsis lactic high-AG plus respiratory alkalosis; salicylate respiratory alkalosis plus high-AG; DKA plus vomiting high-AG plus metabolic alkalosis; COPD plus diuretic chronic respiratory acidosis plus metabolic alkalosis). Banner 'The body never over-compensates — if the compensation is off, a mixed disorder'. Flat vector illustration, crisp typography. — FigureThe 5-step approach and the common mixed patterns. The delta-delta uncovers the concurrent metabolic disorder in the high-anion-gap acidosis.

Steps 1 and 2 — the primary disorder

The pH — the acidosis (under 7.35) or the alkalosis (over 7.45).[1]
The pCO2 and the bicarbonate — the respiratory (the pCO2 moves opposite to the pH) vs the metabolic (the bicarbonate moves with the pH).[1]

Step 3 — the compensation (is it appropriate?)

The metabolic acidosis — the Winter formula: the expected pCO2 = (1.5 × HCO3) + 8 (±2). If the actual pCO2 is higher than expected → a concurrent respiratory acidosis (the under-ventilation). If the actual pCO2 is lower → a concurrent respiratory alkalosis (the over-ventilation).[1]
The metabolic alkalosis — the expected pCO2 rises about 0.7 mmHg per 1 HCO3 (the hypoventilatory compensation, the limited by the hypoxia).[1]
The respiratory acidosis/alkalosis — the acute: the HCO3 rises 1 per 10 pCO2 in the acidosis (the falls 2 per 10 in the alkalosis); the chronic: the HCO3 rises 4 per 10 pCO2 in the acidosis (the renal compensation over the days).[1]

Step 4 — the anion gap

The anion gap (AG) = Na − (Cl + HCO3). The normal 8 to 12 (the albumin-corrected: the AG rises 2.5 per 10 g/L fall in the albumin).[1]
The high AG → the unmeasured anions (the lactate, the ketones, the toxins — the methanol, the ethylene glycol, the salicylate; the renal failure).[1]
The normal AG → the hyperchloraemic acidosis (the RTA, the diarrhoea, the saline).[1]

Step 5 — the delta-delta (ΔAG/ΔHCO3)

The ΔAG = (the measured AG) − 12 (the normal). The ΔHCO3 = 24 − (the measured HCO3) (the fall).[1]
The ratio = ΔAG / ΔHCO3:[1]
- About 1 — the pure high-AG metabolic acidosis (the AG rise matches the HCO3 fall).[1]
- Under 1 (e.g. 0.7) — the AG rise is less than the HCO3 fall → a concurrent normal-gap (hyperchloraemic) metabolic acidosis.[1]
- Over 2 (or over 1.5 to 2) — the AG rise exceeds the HCO3 fall → a concurrent metabolic alkalosis.[1]
The equivalent: the corrected HCO3 = the measured HCO3 + ΔAG. If over 24 → a concurrent metabolic alkalosis; if under 24 → a concurrent metabolic acidosis.[1]

The Henderson–Hasselbalch equation — the chemistry under the numbers

The bicarbonate buffer system — the dominant extracellular buffer — is governed by the Henderson–Hasselbalch equation: [1]

pH = 6.1 + log₁₀( [HCO₃⁻] / (0.03 × PaCO₂) ) [1]

where 6.1 is the pKa of the carbonic acid system and 0.03 is the solubility coefficient of CO₂ in plasma (the mmol/L per mmHg).[1][8] Three clinically decisive consequences follow:

The pH is set by the RATIO, not the absolute values. A bicarbonate of 12 with a PaCO₂ of 27 (ratio 12 / 0.81 = 14.8) gives a pH of 7.28; a bicarbonate of 24 with a PaCO₂ of 54 (the same ratio) gives the IDENTICAL pH of 7.28. The first is a metabolic acidosis with appropriate compensation; the second is a respiratory acidosis with appropriate compensation. The number you must hold in your head is the ratio of bicarbonate to (0.03 × PaCO₂).
The denominator (PaCO₂) is the respiratory variable; the numerator (HCO₃⁻) is the metabolic variable. A primary change in either moves the pH predictably, and the body's compensatory adjustment of the OTHER variable tries to restore the ratio. This is the mathematical heart of "the body never over-compensates" — compensation always pulls the ratio back towards 20:1 (the normal ratio of HCO₃⁻ 24 to dissolved CO₂ 1.2).[8]
The equation explains why a normal pH in a sick patient is suspicious. In the healthy state the ratio is ~20:1. If a critically ill patient has a PaCO₂ of 30 (low — respiratory alkalosis) and a HCO₃⁻ of 17 (low — metabolic acidosis), the ratio is 17 / 0.9 = 18.9 and the pH is near-normal (~7.40). A near-normal pH in a sick ICU patient almost always means TWO opposing primary disorders, not wellness.[5]

Winter's formula — derivation, application, and the limits of compensation

Winter's formula — the expected PaCO₂ = (1.5 × HCO₃⁻) + 8 (± 2) mmHg — is the bedside tool that tests whether the respiratory compensation for a metabolic acidosis is appropriate. It was derived by Albert, Dell and Winters (1967) from the observed compensatory hyperventilation in patients with a metabolic acidosis, and it is the single calculation that most often unmasks a mixed disorder.[1][1]

The four compensation rules — apply the one that matches the primary disorder

Primary disorder	Expected compensation	Rule	If the actual is OFF →
Metabolic acidosis	↓ PaCO₂	Winter's: PaCO₂ = 1.5 × HCO₃ + 8 ± 2[1]	Higher → concurrent respiratory acidosis; Lower → concurrent respiratory alkalosis
Metabolic alkalosis	↑ PaCO₂	PaCO₂ rises ~0.7 mmHg per 1 mmol/L rise in HCO₃ (limited by hypoxia)	Higher → concurrent respiratory acidosis; Lower → concurrent respiratory alkalosis
Acute respiratory acidosis	↑ HCO₃	HCO₃ rises ~1 mmol/L per 10 mmHg rise in PaCO₂	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis
Chronic respiratory acidosis (COPD)	↑ HCO₃	HCO₃ rises ~4 mmol/L per 10 mmHg rise in PaCO₂ (renal, over days)	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis
Acute respiratory alkalosis	↓ HCO₃	HCO₃ falls ~2 mmol/L per 10 mmHg fall in PaCO₂	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis
Chronic respiratory alkalosis	↓ HCO₃	HCO₃ falls ~4–5 mmol/L per 10 mmHg fall in PaCO₂ (renal, over days)	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis

The acute-versus-chronic distinction for the respiratory rules is critical: a COPD patient with a chronically elevated PaCO₂ of 70 has a chronically raised HCO₃ (the renal compensation). If that HCO₃ is LOWER than the chronic rule predicts, a concurrent metabolic acidosis is present (a common finding in the COPD exacerbation with a lactic acidosis or a hyperchloraemic acidosis).[2]

The Stewart (strong ion difference) approach — the physical-chemical view

The traditional Henderson–Hasselbalch approach treats the bicarbonate as the metabolic variable. The Stewart approach (Peter Stewart, 1981) re-derives the pH from three independent variables, treating bicarbonate as a dependent (derived) quantity. The two frameworks describe the same chemistry; Stewart simply makes explicit what the traditional model can only describe.[6][7]

The PaCO₂ — the respiratory variable, identical in both frameworks.
The strong ion difference (SID) — the difference between the fully dissociated (strong) cations and anions. In practice this is dominated by Na⁺ − Cl⁻, with a normal value of about 40 mmol/L. The strong ions cannot combine with hydrogen, so the SID acts like an electrical charge gap that water dissociation must fill: a small SID (a fall in sodium or, far more commonly, a rise in chloride — as with large-volume 0.9% saline) drives water dissociation towards more free H⁺ and an acidosis; a large SID (chloride loss from vomiting or diuretics) drives an alkalosis.[6][13]
The total concentration of weak acids (Aₜₒₜ) — predominantly albumin and, to a lesser extent, phosphate. A fall in Aₜₒₜ (hypoalbuminaemia, ubiquitous in critical illness) is itself an alkalinising process, which is why a low albumin masks an underlying high anion-gap acidosis — the basis for the albumin correction of the AG.[7][13]

The Stewart framework earns its place at the bedside by explaining three things the traditional model can only label:[6][16]

Why large-volume 0.9% saline causes a hyperchloraemic acidosis — it loads chloride without bicarbonate, shrinking the SID. This is the mechanism behind the SMART trial's finding that balanced crystalloids (which preserve the SID) modestly outperform saline.[19][12]
Why hypoalbuminaemia hides a high anion-gap acidosis — it lowers Aₜₒₜ (alkalinising), offsetting the acid; the albumin-corrected AG restores the diagnostic sensitivity.[7]
Why a metabolic alkalosis accompanies chloride depletion (vomiting, diuretics) — the chloride loss widens the SID.[13]

In Stewart's terms, the "unmeasured anions" of critical illness (lactate, ketones, toxin metabolites) are quantified as the strong ion gap (SIG) — the difference between the apparent SID (SIDa, calculated from all the measured strong ions) and the effective SID (SIDe, calculated from the buffer bases and the unmeasured anions). A normal SIG is zero; a positive SIG signals unmeasured anions, the Stewart equivalent of a high anion gap, but corrected for albumin and phosphate automatically.[7][12]

The delta-delta in depth — the delta ratio and the corrected bicarbonate

The delta-delta (also called the delta ratio, ΔAG/ΔHCO₃⁻) is the calculation that uncovers a concurrent metabolic disorder hiding inside an apparent single high-anion-gap metabolic acidosis. Wrenn (1990) introduced it as the bedside approach to the mixed metabolic disorders, and it remains the single most reliable screen for a second metabolic process.[4][1]

The principle: in a pure high-AG metabolic acidosis, every molecule of unmeasured anion (lactate, ketone) that accumulates consumes one molecule of bicarbonate — so the rise in the AG (ΔAG) should equal the fall in the HCO₃⁻ (ΔHCO₃⁻), and the ratio is about 1. Any deviation means a second metabolic disorder is pushing the bicarbonate up or down independently.[4]

The delta-delta — what each ratio means

ΔAG / ΔHCO₃⁻ ratio	Interpretation	Concurrent disorder	Classic clinical setting
~1 (0.8–2.0)	The AG rise matches the HCO₃⁻ fall	None — a pure high-AG metabolic acidosis	Straightforward lactic acidosis or DKA
< 1 (e.g. 0.7)	The HCO₃⁻ has fallen MORE than the AG has risen	Concurrent normal-gap (hyperchloraemic) metabolic acidosis	DKA resuscitated with saline; diarrhoea + lactic acidosis; renal failure
> 2	The AG has risen MORE than the HCO₃⁻ has fallen	Concurrent metabolic alkalosis	DKA + vomiting; lactic acidosis + vomiting/diuretics; uraemia + vomiting
< 0.4	The AG is near-normal but the HCO₃⁻ is very low	Predominant / pure normal-gap acidosis (the high-AG component is minor)	Pure hyperchloraemic acidosis from saline or RTA

[1]

The equivalent and often simpler bedside method is the corrected (or "gap-gap") bicarbonate: [1]

Corrected HCO₃⁻ = measured HCO₃⁻ + ΔAG [1]

If the corrected HCO₃⁻ is greater than 24, a concurrent metabolic alkalosis is present (the bicarbonate has been "pushed up" by the alkalosing process). If it is less than 24, a concurrent metabolic acidosis (additional acid load) is present. The corrected bicarbonate and the delta ratio give the same answer by two routes; many clinicians find the corrected bicarbonate more intuitive at the bedside.[4][5]

Mixed metabolic acidosis — the combined high-AG and normal-AG acidosis

A mixed metabolic acidosis — a high-anion-gap acidosis AND a normal-anion-gap (hyperchloraemic) acidosis occurring together — is one of the commonest mixed disorders in the ICU, and it is the disorder most reliably missed without the delta-delta.[4]

The mixed metabolic acidosis — how the delta-delta catches it

Start with a high-AG metabolic acidosis

The patient has a high AG (e.g. 22, lactate 4) with a low HCO₃⁻ (e.g. 14). Na 140, Cl 104.

Calculate the delta-delta

ΔAG = 22 − 12 = 10. ΔHCO₃⁻ = 24 − 14 = 10. Ratio = 10/10 = 1.0 → a PURE high-AG acidosis. So far, so good.

Now resuscitate with 3 litres of 0.9% saline

Saline loads chloride (Cl rises to 112) without bicarbonate. A hyperchloraemic (normal-AG) acidosis is superimposed. The HCO₃⁻ falls further (to 11) but the AG barely moves (lactate clearing: AG now 16).

Re-check the delta-delta

ΔAG = 16 − 12 = 4. ΔHCO₃⁻ = 24 − 11 = 13. Ratio = 4/13 = 0.3 → a CONCURRENT normal-gap acidosis. The chloride (112) and the falling HCO₃⁻ against a near-normal AG confirm the hyperchloraemic component.

The corrected HCO₃⁻ confirms

Corrected HCO₃⁻ = 11 + 4 = 15 (well under 24) → a concurrent metabolic acidosis. The two methods agree.

The clinical lesson

The delta-delta under 1 (or the corrected HCO₃⁻ under 24) signals the saline-induced hyperchloraemic component. Switch to a balanced crystalloid (Plasma-Lyte, Hartmann) — SMART shows measurable renal and outcome benefit over saline.<Cite id="19" />

The classic settings for a mixed metabolic acidosis are: DKA during recovery (the resolving ketoacidosis + the saline-induced hyperchloraemia from aggressive resuscitation — the AG normalises but the acidosis persists, exactly because the chloride has replaced the ketones); diarrhoea with a lactic acidosis (sepsis, mesenteric ischaemia); renal failure with a superimposed lactic acidosis; and any patient resuscitated with large volumes of 0.9% saline who was already acidaemic.[4][12]

Mixed metabolic alkalosis + metabolic acidosis — the delta ratio over 2

A concurrent metabolic alkalosis and metabolic acidosis is the mirror-image mixed metabolic disorder. The clinical clue is a high-anion-gap metabolic acidosis whose bicarbonate is higher than the AG rise would predict — the alkalosing process (vomiting, diuretics, nasogastric suction) has "rescued" the bicarbonate partially.[4][5]

The DKA-plus-vomiting — the classic metabolic alkalosis + acidosis

The presentation

A young type-1 diabetic with several days of vomiting and polyuria. The vomiting causes a metabolic alkalosis (loss of HCl and volume); the insulin deficiency causes DKA (a high-AG metabolic acidosis from ketoacids). Both are primary.

The blood gas

pH 7.20, HCO₃⁻ 13, PaCO₂ 30. Na 138, Cl 90, K 3.0.

Step 4 — the anion gap

AG = 138 − (90 + 13) = 35 (high-AG acidosis). The delta-delta: ΔAG = 35 − 12 = 23; ΔHCO₃⁻ = 24 − 13 = 11. Ratio = 23/11 = 2.1 → OVER 2.

The corrected HCO₃⁻ confirms

Corrected HCO₃⁻ = 13 + 23 = 36 (well over 24) → a concurrent metabolic alkalosis. The low chloride (90) and low potassium (3.0) point to the vomiting.

Why the pH is "better than expected"

Without the vomiting, the HCO₃⁻ would be much lower (around 11) and the pH correspondingly lower. The alkalosing process partially offsets the acid — the pH 7.20 is "less acidic" than a pure DKA at this AG would produce.

Management implication

Treat the DKA (insulin, fluids) but DO NOT over-correct the potassium — the metabolic alkalosis of vomiting and the impending insulin-driven intracellular K shift conspire to drop the K dangerously. Replete K early. The metabolic alkalosis resolves as the volume is restored and the chloride repleted.

The corrected bicarbonate and the delta ratio over 2 are the two routes to the same diagnosis. The low chloride and the hypokalaemia are the biochemical signatures of the vomiting (or diuretic) alkalosis layered on top of the high-AG acidosis.[2][3]

Triple disorders — the three-way mixed disturbance

A triple disorder is the simultaneous presence of three primary acid-base disturbances. The classic and most examinable is the metabolic alkalosis + metabolic acidosis + respiratory alkalosis — the textbook scenario being the septic patient with vomiting, a lactic acidosis, and hyperventilation.[1]

The classic triple disorder — sepsis + vomiting + lactic acidosis + hyperventilation

The scenario

A 60-year-old with pyelonephritis and septic shock, 2 days of vomiting, now tachypnoeic. Three primary disorders are coexisting.

Disorder 1 — the high-AG metabolic acidosis

The lactic acidosis of septic shock (the lactate 5). A primary metabolic acidosis: the HCO₃⁻ falls.

Disorder 2 — the metabolic alkalosis

The vomiting (the loss of HCl and volume, the chloride depletion). A primary metabolic alkalosis: the HCO₃⁻ rises independently.

Disorder 3 — the respiratory alkalosis

The hyperventilation (septic encephalopathy, fever, the early direct medullary stimulation, hepatic failure). A primary respiratory alkalosis: the PaCO₂ falls.

The blood gas may look deceptively simple

pH 7.39, HCO₃⁻ 16, PaCO₂ 27. The pH is near-normal — three opposing processes nearly cancelling. Na 138, Cl 92.

Step 3 — Winter reveals the respiratory component

Expected PaCO₂ = 1.5 × 16 + 8 = 32 ± 2. The actual PaCO₂ (27) is LOWER than expected → a concurrent respiratory ALKALOSIS.

Step 4 — the anion gap

AG = 138 − (92 + 16) = 30 (high-AG acidosis).

Step 5 — the delta-delta reveals the metabolic alkalosis

ΔAG = 30 − 12 = 18; ΔHCO₃⁻ = 24 − 16 = 8. Ratio = 18/8 = 2.25 → OVER 2 → a concurrent metabolic ALKALOSIS. Corrected HCO₃⁻ = 16 + 18 = 34 (well over 24) — confirms.

The synthesis

Three primary disorders: a high-AG metabolic acidosis (lactate) + a metabolic alkalosis (vomiting) + a respiratory alkalosis (hyperventilation). Each was found by a single calculation — Winter (the respiratory), the delta-delta (the metabolic alkalosis).

The triple disorder is the acid-base diagnosis that most rewards the systematic 5-step approach. A single number (the near-normal pH) conceals three processes; only by working through pH → primary → compensation → anion gap → delta-delta do all three emerge. The salicylate-intoxicated patient who then vomits is another triple-disorder classic (the respiratory alkalosis + the high-AG acidosis of salicylate + the metabolic alkalosis of vomiting).[1][5]

The common ICU mixed patterns

The high-AG acidosis + the respiratory alkalosis: the sepsis (the lactic acidosis + the hyperventilation), the salicylate toxicity (the early respiratory alkalosis — the direct medullary stimulation + the high-AG acidosis — the salicylate), the hepatorenal syndrome, the severe liver disease.[1]
The high-AG acidosis + the metabolic alkalosis: the vomiting (the metabolic alkalosis) + the lactic acidosis or the DKA (the "DKA plus the vomiting"). The delta-delta over 2.[1]
The high-AG + the normal-gap (hyperchloraemic) acidosis: the DKA recovery (the resolving ketoacidosis + the saline-induced hyperchloraemia), the diarrhoea + the lactic acidosis, the renal failure.[1]
The mixed respiratory + the metabolic: the COPD (the chronic respiratory acidosis) + the metabolic alkalosis (the diuretic); the cardiac arrest (the respiratory + the metabolic acidosis); the respiratory failure + the sepsis.[1]
The triple disorder: the sepsis + the vomiting + the lactic acidosis + the hyperventilation (the metabolic acidosis + the metabolic alkalosis + the respiratory alkalosis); the salicylate + the vomiting (the respiratory alkalosis + the high-AG acidosis + the metabolic alkalosis).[1]

The one-paragraph exam answer

A mixed acid-base disorder is two or more primary disturbances simultaneously. The key principle: the body never over-compensates — if the compensation is off, a second disorder is present. The 5-step approach: (1) the pH; (2) the pCO2/HCO3 to the primary; (3) the compensation (the Winter formula for the metabolic acidosis: the expected pCO2 = 1.5 × HCO3 + 8 ±2 — if the actual is higher → a concurrent respiratory acidosis; if lower → a respiratory alkalosis); (4) the anion gap = Na − (Cl + HCO3) (8 to 12, albumin-corrected); (5) the delta-delta (ΔAG/ΔHCO3): about 1 the pure high-AG, under 1 the concurrent normal-gap hyperchloraemic acidosis, over 2 the concurrent metabolic alkalosis. The common ICU patterns: the sepsis (the lactic high-AG + the respiratory alkalosis), the salicylate (the respiratory alkalosis + the high-AG), the DKA + the vomiting (the high-AG + the metabolic alkalosis), the COPD + the diuretic (the chronic respiratory acidosis + the metabolic alkalosis). The Stewart approach re-expresses the same chemistry as three independent variables — the PaCO2, the strong ion difference (the Na minus Cl, about 40), and the total weak acids (the albumin) — and explains the saline-induced hyperchloraemic acidosis (the low SID), the masking of a high-AG acidosis by the hypoalbuminaemia (the low A-tot), and the alkalosis of the chloride depletion. The triple disorder (the metabolic alkalosis + the metabolic acidosis + the respiratory alkalosis — the classic in the sepsis with the vomiting and the lactic acidosis and the hyperventilation) is found by working all five steps: the Winter formula unmasks the respiratory, the delta-delta unmasks the metabolic alkalosis.

[1]

Worked examples — the 5-step approach applied

Worked example 1 — the septic patient with a near-normal pH (a mixed disorder in disguise)

The gas and the electrolytes

pH 7.38, PaCO₂ 30, HCO₃⁻ 17. Na 140, Cl 105, albumin 30 g/L. Lactate 4.5.

Step 1–2 — the pH and the primary

pH 7.38 is just within normal. The PaCO₂ is low and the HCO₃⁻ is low. Both move in the direction of an acidosis for the HCO₃⁻ and an alkalosis for the PaCO₂ — a near-normal pH from two opposing processes.

Step 3 — the compensation

If the primary is a metabolic acidosis, expected PaCO₂ = 1.5 × 17 + 8 = 33.5 ± 2. The actual (30) is LOWER → a concurrent respiratory ALKALOSIS.

Step 4 — the albumin-corrected anion gap

AG = 140 − (105 + 17) = 18. Corrected for albumin: the albumin is 30 (10 below 40), so add 2.5 → corrected AG ~20.5 (high-AG acidosis). The lactate (4.5) accounts for most of the ΔAG.

Step 5 — the delta-delta

ΔAG = 20 − 12 = 8; ΔHCO₃⁻ = 24 − 17 = 7. Ratio = 8/7 = 1.14 → about 1, a pure high-AG acidosis (no concurrent metabolic disorder).

The diagnosis

A high-AG metabolic acidosis (lactic, from sepsis) + a respiratory alkalosis (hyperventilation). Two primary disorders — the classic septic pattern.

[1]

Worked example 2 — DKA that will not clear (the saline-induced hyperchloraemic component)

The setting

12 hours into DKA treatment. The ketones have cleared, the glucose is normalising, but the HCO₃⁻ is still low and the patient was resuscitated with 4 litres of 0.9% saline.

The gas and the electrolytes

pH 7.30, PaCO₂ 28, HCO₃⁻ 14. Na 140, Cl 110, albumin 35.

Step 3 — Winter

Expected PaCO₂ = 1.5 × 14 + 8 = 29 ± 2. Actual 28 → within range. The compensation is appropriate (no concurrent respiratory disorder).

Step 4 — the anion gap

AG = 140 − (110 + 14) = 16 (corrected ~17.5). The AG is only mildly raised — the ketoacidosis has largely resolved.

Step 5 — the delta-delta

ΔAG = 17 − 12 = 5; ΔHCO₃⁻ = 24 − 14 = 10. Ratio = 5/10 = 0.5 → UNDER 1 → a concurrent normal-gap (hyperchloraemic) acidosis. Corrected HCO₃⁻ = 14 + 5 = 19 (under 24) — confirms.

The diagnosis and the lesson

The persistent acidosis is now a hyperchloraemic (normal-AG) acidosis from the saline, NOT ongoing ketoacidosis. The high chloride (110) is the give-away. Stop the saline, switch to a balanced crystalloid, and the HCO₃⁻ will normalise — no bicarbonate, no more insulin beyond maintenance.

The strong ion difference in practice — when Stewart changes management

Management infographic for mixed acid-base disorders: five-step approach, treat drivers including perfusion ventilation chloride and toxins, prefer balanced fluids — FigureName every process, then treat causes — not the pH number alone.

The Stewart variables — what each does to the pH

Variable	Normal value	A change towards acidosis	A change towards alkalosis	Clinical correlate
PaCO₂	40 mmHg	↑ (hypoventilation)	↓ (hyperventilation)	The respiratory variable — common to both frameworks
Strong ion difference (SID)	~40 mmol/L (Na⁺ − Cl⁻)	↓ — chloride rises (saline) or sodium falls	↑ — chloride falls (vomiting, diuretics)	The dominant metabolic variable in Stewart's terms; explains saline acidosis and vomiting alkalosis
Total weak acids (Aₜₒₜ)	albumin ~40 g/L	↑ (hyperalbuminaemia — rare) or hyperphosphataemia	↓ — hypoalbuminaemia (very common in ICU)	Hypoalbuminaemia is alkalinising and MASKS a high-AG acidosis; this is why the albumin correction of the AG matters
Unmeasured anions (SIG)	~0 (zero)	↑ — lactate, ketones, toxins, uraemia	—	The Stewart equivalent of the high anion gap; a positive SIG is the acid load

[1]

The practical use of the Stewart framework at the bedside is not to replace the traditional approach but to explain the cases that the traditional model labels but cannot account for: the hyperchloraemic acidosis of saline (a low SID, not a "bicarbonate loss"), the alkalosis of hypoalbuminaemia (a low Aₜₒₜ), and the unmeasured anions quantified as the SIG rather than the cruder AG.[6][7][16]

The albumin correction of the anion gap — the hidden acid load

The albumin correction is essential in the ICU because hypoalbuminaemia is near-universal in critical illness, and an uncorrected anion gap will MISS a high-AG acidosis in the hypoalbuminaemic patient. The rule: the AG falls by approximately 2.5 mmol/L for every 10 g/L fall in the albumin below 40 g/L.[1][7]

Corrected AG = measured AG + 2.5 × ( (40 − albumin g/L) / 10 )
A patient with a measured AG of 14 (apparently normal) and an albumin of 20 g/L has a corrected AG of 14 + 2.5 × 2 = 19 — a high-AG acidosis that the uncorrected gap completely conceals. [1]

This is the single most common reason a high-AG acidosis is missed at the bedside. Always correct the AG for albumin in the ICU patient.[7][20]

Short-answer questions

SAQ — Systematic ABG interpretation in the septic, vomiting patient

10 minutes · 10 marks

A 64-year-old woman is admitted to the ICU with pyelonephritis and septic shock. She has vomited for 48 hours and is tachypnoeic (RR 32). Arterial blood gas on 6 L/min nasal specs: pH 7.39, PaCO₂ 27 mmHg, HCO₃⁻ 16 mmol/L, PaO₂ 94 mmHg. Sodium 138, chloride 92, potassium 3.1, albumin 28 g/L, lactate 5.2 mmol/L. Interpret this blood gas systematically and outline the immediate management.

[1]

SAQ — Compensation limits and the body that never over-compensates

10 minutes · 10 marks

A 72-year-old man with severe COPD (baseline PaCO₂ 68 mmHg, HCO₃⁻ 36 mmol/L at last clinic visit) is admitted with a COPD exacerbation and recent frusemide 80 mg daily for cor pulmonale. Arterial blood gas: pH 7.38, PaCO₂ 60 mmHg, HCO₃⁻ 35 mmol/L. Sodium 140, chloride 88, potassium 2.9. Assess the appropriateness of the compensation and explain the principle that governs it.

[1]

Clinical pearls

High-yield mixed acid-base pearls for the CICM / FFICM / EDIC exam

The body NEVER over-compensates. This single principle is the foundation of the entire topic. If the measured compensation is more or less than the predicted (Winter for a metabolic acidosis; the acute/chronic rules for the respiratory), a SECOND primary disorder is present. Always calculate the expected compensation and compare it to the actual — never accept a single disorder at face value.[1][8]
A near-normal pH in a sick ICU patient is a mixed disorder until proven otherwise. Two opposing primary disorders (e.g. a metabolic acidosis + a respiratory alkalosis) pull the pH in opposite directions and produce a deceptively "normal" looking gas. The abnormalities are in the PaCO₂ and the HCO₃⁻, not the pH. If the patient is sick and the pH is 7.40, work the five steps.[5]
Winter's formula is the single calculation that most often unmasks a mixed disorder. Expected PaCO₂ = 1.5 × HCO₃⁻ + 8 ± 2. It takes ten seconds and it catches the concurrent respiratory acidosis or alkalosis that the pH alone will never reveal. Albert, Dell and Winters derived it in 1967 from observed compensatory hyperventilation — it remains the bedside standard.[1]
The delta-delta (ΔAG/ΔHCO₃⁻) catches the concurrent METABOLIC disorder that Winter cannot. Winter tests the respiratory compensation; the delta-delta tests for a second metabolic process. A ratio under 1 = a concurrent normal-gap acidosis; over 2 = a concurrent metabolic alkalosis. The corrected HCO₃⁻ (measured HCO₃⁻ + ΔAG) gives the same answer by a simpler route.[4]
Always correct the anion gap for albumin in the ICU. The AG falls ~2.5 mmol/L per 10 g/L fall in albumin. A hypoalbuminaemic ICU patient (albumin 20) with a "normal" measured AG of 14 actually has a corrected AG of 19 — a high-AG acidosis the uncorrected gap hides. The albumin correction restores the diagnostic sensitivity and is non-negotiable in critical illness.[7][20]
The salicylate toxicity is the classic mixed disorder — respiratory alkalosis + high-AG acidosis from the start. Salicylate directly stimulates the medulla (an early respiratory alkalosis) AND uncouples oxidative phosphorylation (a high-AG metabolic acidosis with lactate and ketones). The two coexist from the early presentation. Treat with urine alkalinisation (sodium bicarbonate to trap the salicylate) and haemodialysis for the severe case. Do not miss salicylate in the unexplained mixed disorder — check the level.[1][9]
The septic patient often has THREE disorders — the lactic acidosis + the respiratory alkalosis + (if vomiting) the metabolic alkalosis. This is the classic triple disorder. The pH may be near-normal (three opposing processes). Only the systematic approach — pH, primary, Winter (the respiratory), AG, delta-delta (the metabolic alkalosis) — uncovers all three.[1]
The DKA-plus-vomiting is the textbook mixed metabolic disorder — the delta ratio over 2. The vomiting (metabolic alkalosis) partially rescues the bicarbonate that the ketoacidosis (high-AG acidosis) has consumed. The corrected HCO₃⁻ is over 24. Watch the potassium — the alkalosis and the insulin-driven intracellular shift conspire to drop it dangerously.[2][4]
The "DKA that will not clear" after 12 hours is usually a saline-induced hyperchloraemic acidosis, not ongoing ketoacidosis. The AG has normalised (the ketones have gone) but the HCO₃⁻ stays low and the chloride is high. The delta ratio is under 1. Switch from 0.9% saline to a balanced crystalloid (SMART shows measurable benefit) and the HCO₃⁻ will recover — no bicarbonate, no extra insulin.[19][12]
0.9% saline causes a hyperchloraemic acidosis by lowering the strong ion difference (Stewart). Saline loads chloride (154 mmol/L) without bicarbonate, shrinking the SID (Na⁺ − Cl⁻) from ~40 towards ~30. This is the physical-chemical explanation for the clinical finding that balanced crystalloids (Plasma-Lyte, Hartmann) outperform saline in the SMART and SALT-ED trials. Prefer balanced crystalloids for resuscitation; reserve saline for the hyponatraemic or chloride-losing patient.[19][6]
Hypoalbuminaemia is itself an alkalinising process — it masks the high-AG acidosis and lowers the apparent AG. This is why the albumin correction matters, and it is the Stewart "low Aₜₒₜ" effect. A low albumin also explains why some ICU patients have a higher-than-expected pH for their apparent acid load.[7][13]
The metabolic alkalosis can delay weaning from mechanical ventilation. A chronic respiratory acidosis (COPD) with a superimposed metabolic alkalosis (diuretics, vomiting) leaves the patient with a high bicarbonate and a high baseline PaCO₂. When the ventilator is reduced, the PaCO₂ must rise to maintain the pH, and the patient fails to wean. Correct the metabolic alkalosis (replete chloride and potassium, stop the diuretic) before weaning.[3]
In a high-AG acidosis, a falling lactate with a PERSISTENT acidaemia signals a concurrent normal-gap (hyperchloraemic) acidosis. The delta ratio falls below 1 as the lactate clears. This is the commonest reason a "recovering" septic patient stays acidaemic — the saline resuscitation has superimposed a hyperchloraemic component. Check the chloride and the corrected HCO₃⁻.[4][17]
The corrected HCO₃⁻ (measured HCO₃⁻ + ΔAG) is the bedside shortcut to the delta-delta. Over 24 → a concurrent metabolic alkalosis; under 24 → a concurrent metabolic acidosis. It is mathematically equivalent to the delta ratio and most clinicians find it faster to compute at the bedside. Use whichever you find more intuitive — but use ONE of them in every high-AG acidosis.[4][5]
Sodium bicarbonate is NOT indicated for a metabolic acidosis outside the peri-arrest, hyperkalaemic, or tricyclic-antidepressant setting. BICAR-ICU found no overall outcome benefit for bicarbonate in severe metabolic acidaemia (pH ≤ 7.20). Treat the cause (restore perfusion, dialyse, insulin, fomepizole) — the acidosis corrects when the underlying process reverses. The exception is the Acute Kidney Injury Network (AKIN) subgroup of BICAR-ICU, where bicarbonate reduced the need for renal replacement therapy.[18][9]
The osmolar gap + the high-AG acidosis = a toxic alcohol (methanol or ethylene glycol) until proven otherwise. The parent alcohol raises the osmolar gap; the metabolites (formic, glycolic, oxalic acid) raise the anion gap. The injury is done by the metabolites, so fomepizole (to block alcohol dehydrogenase) and haemodialysis are time-critical. A normal osmolar gap does NOT exclude a late presentation once the parent alcohol has been fully metabolised.[1][9]
The Henderson–Hasselbalch equation (pH = 6.1 + log(HCO₃⁻ / (0.03 × PaCO₂))) sets the pH by the RATIO, not the absolute values. A HCO₃⁻ of 12 with a PaCO₂ of 27 gives the same pH as a HCO₃⁻ of 24 with a PaCO₂ of 54 — the first is a compensated metabolic acidosis, the second a compensated respiratory acidosis. The ratio (normally ~20:1) is the number to hold in your head.[8]
Base excess and standard base excess quantify the metabolic component, corrected for the respiratory. The standard base excess (SBE) is the metabolic component referenced to a PaCO₂ of 40, removing the respiratory influence. A negative SBE is a metabolic acidosis; a positive SBE a metabolic alkalosis. Schlichtig (1998) defined the expected compensation in terms of the SBE — useful when the HCO₃⁻ is itself confounded.[14][11]

The metabolic alkaloses — chloride-responsive and resistant

A mixed disorder involving a metabolic alkalosis requires the alkalosis itself to be characterised, because the management differs. The first question is whether the alkalosis is chloride-responsive (the urine chloride under 10–20 mmol/L — vomiting, nasogastric suction, diuretics, post-hypercapnia) or chloride-resistant (the urine chloride over 20 — mineralocorticoid excess, severe hypokalaemia).[2][3]

The metabolic alkalosis — chloride-responsive vs resistant

Feature	Chloride-RESPONSIVE (urine Cl⁻ < 20)	Chloride-RESISTANT (urine Cl⁻ > 20)
Mechanism	Volume and chloride depletion → the kidney retains bicarbonate	Mineralocorticoid excess or severe K⁺ depletion → bicarbonate generation
Common causes	Vomiting, NG suction, diuretics, post-hypercapnia, villous adenoma	Primary hyperaldosteronism, Cushing's, exogenous mineralocorticoids, severe hypokalaemia, Bartter/Gitelman
Urine chloride	< 10–20 mmol/L (low)	> 20 mmol/L (high)
Blood pressure	Usually normal or low	Usually HIGH (mineralocorticoid) — except Bartter/Gitelman
Treatment	Normal saline + KCl repletion (restores volume and chloride)	Treat the cause (spironolactone, remove the adenoma); K⁺ repletion
Relevance to the mixed disorder	The common metabolic alkalosis in the ICU mixed picture (vomiting, diuretics)	Rare but consider in the hypertensive ICU patient with an alkalosis

[1]

In the mixed-disorder context, the chloride-responsive alkalosis (vomiting, diuretics) is by far the commonest, and it is the one that elevates the delta-delta above 2 and the corrected HCO₃⁻ above 24. The treatment is chloride and potassium repletion — which also helps resolve the delta-delta towards 1.[2]

Trials and evidence

Albert, Dell & Winters — the origin of Winter's formula (PMID 6016545)

Source

Annals of Internal Medicine, 1967 — the foundational observational study from which Winter's formula is derived

Question

What is the quantitative relationship between the serum bicarbonate and the compensatory PaCO₂ in a metabolic acidosis?

Method

Analysis of patients with metabolic acidosis of varying severity, plotting the observed PaCO₂ against the serum total CO₂ (bicarbonate) content

Key result

The compensatory PaCO₂ fell predictably with the bicarbonate, fitting the regression PaCO₂ = 1.5 × HCO₃⁻ + 8 (± 2). This is the equation every intensivist still uses at the bedside

Enduring legacy

Over 50 years later, Winter's formula remains the standard test of compensation for a metabolic acidosis — the calculation that most often unmasks a concurrent respiratory disorder

[1]

Wrenn K — The delta gap: an approach to mixed acid-base disorders (PMID 2240729)

Source

Annals of Emergency Medicine, 1990 — the bedside method that brought the delta-delta into routine clinical use

Concept

In a high-AG metabolic acidosis, the rise in the AG (ΔAG) should match the fall in the bicarbonate (ΔHCO₃⁻). A mismatch reveals a concurrent metabolic disorder

The method

Compare the ΔAG to the ΔHCO₃⁻. A ratio of ~1 = a pure high-AG acidosis; under 1 = a concurrent normal-gap acidosis; over 2 = a concurrent metabolic alkalosis

Clinical impact

Gave clinicians a single calculation that exposes the concurrent metabolic disorder hiding inside an apparent single high-AG acidosis — the cornerstone of the mixed-metabolic-disorder diagnosis

[1]

Jaber et al — BICAR-ICU: sodium bicarbonate in severe metabolic acidaemia (PMID 29910040)

Design

Multicentre, open-label, randomised controlled trial — 389 critically ill adults with severe metabolic acidaemia (pH ≤ 7.20, HCO₃⁻ ≤ 20)

Intervention

Sodium bicarbonate 4.2% infusion to maintain pH above 7.30 vs no bicarbonate (standard care)

Primary outcome

Composite of death by day 28 or at least one organ failure at day 7 — NO significant difference in the primary intention-to-treat population

Subgroup signal

In the AKIN stage 2–3 (acute kidney injury) subgroup, bicarbonate REDUCED the primary composite outcome AND the need for renal replacement therapy. This is the modern evidence base for selective bicarbonate use

Bottom line

Bicarbonate is not routinely indicated for severe metabolic acidaemia, but may be considered in the patient with significant acute kidney injury. Treat the cause; do not reflexively give bicarbonate for the number

[1]

Semler et al — SMART: balanced crystalloids vs saline in critically ill adults (PMID 29768150)

Design

Pragmatic, cluster-crossover, single-centre randomised trial — 15,802 critically ill adults across 5 ICUs at Vanderbilt

Intervention

Balanced crystalloids (lactated Ringer's or Plasma-Lyte) vs 0.9% saline for all intravenous fluid

Primary outcome

Composite of death, new renal replacement therapy, or persistent renal dysfunction at 30 days

Key result

Balanced crystalloids modestly but significantly REDUCED the composite outcome (14.3% vs 15.4%, p = 0.04). The benefit is attributable to the avoidance of the saline-induced hyperchloraemic (low-SID) acidosis

Acid-base relevance

The trial is the clinical validation of Stewart's prediction: saline lowers the strong ion difference and causes acidosis and renal harm; balanced crystalloids preserve the SID. Prefer balanced crystalloids for resuscitation

[1]

Moviat et al — Conventional vs physicochemical (Stewart) approach in ICU metabolic acidosis (PMID 12793889)

Source

Critical Care, 2003 — a head-to-head comparison of the traditional and Stewart approaches in the ICU

Question

Does the Stewart (strong ion difference) method detect mixed disorders the traditional approach misses?

Finding

The Stewart approach detected additional disorders (particularly the hypoalbuminaemia-induced alkalosis and the saline-induced acidosis) that the traditional method under-appreciated. The two methods largely concurred for the high-AG component

Lesson

The Stewart framework adds sensitivity for the SID-driven (saline, chloride) and Aₜₒₜ-driven (albumin) disorders; it complements rather than replaces the traditional approach

[1]

Schlichtig, Grogono & Severinghaus — PaCO₂ and standard base excess compensation (PMID 9671365)

Source

Critical Care Medicine, 1998 — the definitive compilation of human acid-base compensation limits

Concept

Defined the EXPECTED compensation for each primary disorder in terms of the standard base excess (SBE), and the limits beyond which a mixed disorder must be present

Key contribution

Established that compensation is predictable and bounded — the basis for the rules (Winter, the acute/chronic respiratory) used at the bedside. When the measured value lies outside the predicted band, a second primary disorder is certain

[1]

Rastegar A — Clinical utility of Stewart's method (PMID 19520748)

Source

Clinical Journal of the American Society of Nephrology, 2009 — the accessible review that brought Stewart into mainstream nephrology and critical care teaching

Core teaching

Re-derives the pH from the three independent variables (PaCO₂, SID, Aₜₒₜ) and demonstrates that the traditional and Stewart approaches agree when all variables are accounted for — but Stewart makes the saline acidosis, the albumin effect, and the chloride-depletion alkalosis mechanistically explicit

Bedside relevance

The review clinicians cite when justifying the albumin correction of the AG and the preference for balanced crystalloids over saline

[1]

Adrogué & Madias — Management of life-threatening acid-base disorders (PMID 9414329, 9420343)

Source

New England Journal of Medicine, 1998 (two parts) — the authoritative clinical review of acid-base management

Part 1 (9414329)

Covers the metabolic acidoses — the high-AG and normal-AG, the diagnosis by the anion gap and the delta-delta, and the management (treat the cause, dialysis, the role of bicarbonate)

Part 2 (9420343)

Covers the metabolic alkaloses (chloride-responsive and resistant), the respiratory acidoses and alkaloses, and the mixed disorders — with the compensation rules and the delta-delta method

Enduring status

Still the most-cited single reference for the systematic approach to acid-base disorders; the framework the examiners expect you to reproduce

[1]

Red flags

References

[1]Albert MS, Dell RB, Winters RW Quantitative displacement of acid-base equilibrium in metabolic acidosis Ann Intern Med, 1967.PMID 6016545
[2]Adrogue HJ, Madias NE Management of life-threatening acid-base disorders. First of two parts N Engl J Med, 1998.PMID 9414329
[3]Adrogue HJ, Madias NE Management of life-threatening acid-base disorders. Second of two parts N Engl J Med, 1998.PMID 9420343
[4]Wrenn K The delta (delta) gap: an approach to mixed acid-base disorders Ann Emerg Med, 1990.PMID 2240729
[5]Haber RJ A practical approach to acid-base disorders West J Med, 1991.PMID 1843849
[6]Emmett M Stewart Versus Traditional Approach to Acid-Base Disorders Anesth Analg, 2016.PMID 27636585
[7]Rastegar A Clinical utility of Stewart's method in diagnosis and management of acid-base disorders Clin J Am Soc Nephrol, 2009.PMID 19520748
[8]Kraut JA, Madias NE Approach to patients with acid-base disorders Respir Care, 2001.PMID 11262558
[9]Kraut JA, Madias NE Treatment of acute metabolic acidosis: a pathophysiologic approach Nat Rev Nephrol, 2012.PMID 22945490
[10]Gomez H, Kellum JA Understanding Acid Base Disorders Crit Care Clin, 2015.PMID 26410149
[11]Berend K Diagnostic Use of Base Excess in Acid-Base Disorders N Engl J Med, 2018.PMID 30067929
[12]Moviat M, van Haren F, van der Hoeven H Conventional or physicochemical approach in intensive care unit patients with metabolic acidosis Crit Care, 2003.PMID 12793889
[13]Story DA, Kellum JA New aspects of acid-base balance in intensive care Curr Opin Anaesthesiol, 2004.PMID 17021538
[14]Schlichtig R, Grogono AW, Severinghaus JW Human PaCO2 and standard base excess compensation for acid-base imbalance Crit Care Med, 1998.PMID 9671365
[15]Moviat M, Terpstra AM, Ruitenbeek W, Kluijtmans LA, Pickkers P, van der Hoeven HG Contribution of various metabolites to the unmeasured anions in critically ill patients with metabolic acidosis Crit Care Med, 2008.PMID 18176310
[16]Story DA Acid-Base Analysis in the Operating Room: A Bedside Stewart Approach Anesthesiology, 2023.PMID 37934110
[17]Al-Jaghbeer M, Kellum JA Acid-base disturbances in intensive care patients: etiology, pathophysiology and treatment Nephrol Dial Transplant, 2015.PMID 25213433
[18]Jaber S, Paugam C, Futier E, et al Sodium bicarbonate therapy for patients with severe metabolic acidaemia in the intensive care unit (BICAR-ICU): a multicentre, open-label, randomised controlled, phase 3 trial Lancet, 2018.PMID 29910040
[19]Semler MW, Self WH, Wanderer JP, et al Balanced Crystalloids versus Saline in Critically Ill Adults N Engl J Med, 2018.PMID 29768150
[20]Berend K, Duits AJ The role of the clinical laboratory in diagnosing acid-base disorders Crit Rev Clin Lab Sci, 2019.PMID 30917291

Overview & definition

The 5-step approach

Steps 1 and 2 — the primary disorder

The pH — the acidosis (under 7.35) or the alkalosis (over 7.45).[1]
The pCO2 and the bicarbonate — the respiratory (the pCO2 moves opposite to the pH) vs the metabolic (the bicarbonate moves with the pH).[1]

Step 3 — the compensation (is it appropriate?)

The metabolic acidosis — the Winter formula: the expected pCO2 = (1.5 × HCO3) + 8 (±2). If the actual pCO2 is higher than expected → a concurrent respiratory acidosis (the under-ventilation). If the actual pCO2 is lower → a concurrent respiratory alkalosis (the over-ventilation).[1]
The metabolic alkalosis — the expected pCO2 rises about 0.7 mmHg per 1 HCO3 (the hypoventilatory compensation, the limited by the hypoxia).[1]
The respiratory acidosis/alkalosis — the acute: the HCO3 rises 1 per 10 pCO2 in the acidosis (the falls 2 per 10 in the alkalosis); the chronic: the HCO3 rises 4 per 10 pCO2 in the acidosis (the renal compensation over the days).[1]

Step 4 — the anion gap

The anion gap (AG) = Na − (Cl + HCO3). The normal 8 to 12 (the albumin-corrected: the AG rises 2.5 per 10 g/L fall in the albumin).[1]
The high AG → the unmeasured anions (the lactate, the ketones, the toxins — the methanol, the ethylene glycol, the salicylate; the renal failure).[1]
The normal AG → the hyperchloraemic acidosis (the RTA, the diarrhoea, the saline).[1]

Step 5 — the delta-delta (ΔAG/ΔHCO3)

The ΔAG = (the measured AG) − 12 (the normal). The ΔHCO3 = 24 − (the measured HCO3) (the fall).[1]
The ratio = ΔAG / ΔHCO3:[1]
- About 1 — the pure high-AG metabolic acidosis (the AG rise matches the HCO3 fall).[1]
- Under 1 (e.g. 0.7) — the AG rise is less than the HCO3 fall → a concurrent normal-gap (hyperchloraemic) metabolic acidosis.[1]
- Over 2 (or over 1.5 to 2) — the AG rise exceeds the HCO3 fall → a concurrent metabolic alkalosis.[1]
The equivalent: the corrected HCO3 = the measured HCO3 + ΔAG. If over 24 → a concurrent metabolic alkalosis; if under 24 → a concurrent metabolic acidosis.[1]

The Henderson–Hasselbalch equation — the chemistry under the numbers

The bicarbonate buffer system — the dominant extracellular buffer — is governed by the Henderson–Hasselbalch equation: [1]

pH = 6.1 + log₁₀( [HCO₃⁻] / (0.03 × PaCO₂) ) [1]

where 6.1 is the pKa of the carbonic acid system and 0.03 is the solubility coefficient of CO₂ in plasma (the mmol/L per mmHg).[1][8] Three clinically decisive consequences follow:

The pH is set by the RATIO, not the absolute values. A bicarbonate of 12 with a PaCO₂ of 27 (ratio 12 / 0.81 = 14.8) gives a pH of 7.28; a bicarbonate of 24 with a PaCO₂ of 54 (the same ratio) gives the IDENTICAL pH of 7.28. The first is a metabolic acidosis with appropriate compensation; the second is a respiratory acidosis with appropriate compensation. The number you must hold in your head is the ratio of bicarbonate to (0.03 × PaCO₂).
The denominator (PaCO₂) is the respiratory variable; the numerator (HCO₃⁻) is the metabolic variable. A primary change in either moves the pH predictably, and the body's compensatory adjustment of the OTHER variable tries to restore the ratio. This is the mathematical heart of "the body never over-compensates" — compensation always pulls the ratio back towards 20:1 (the normal ratio of HCO₃⁻ 24 to dissolved CO₂ 1.2).[8]
The equation explains why a normal pH in a sick patient is suspicious. In the healthy state the ratio is ~20:1. If a critically ill patient has a PaCO₂ of 30 (low — respiratory alkalosis) and a HCO₃⁻ of 17 (low — metabolic acidosis), the ratio is 17 / 0.9 = 18.9 and the pH is near-normal (~7.40). A near-normal pH in a sick ICU patient almost always means TWO opposing primary disorders, not wellness.[5]

Winter's formula — derivation, application, and the limits of compensation

The four compensation rules — apply the one that matches the primary disorder

Primary disorder	Expected compensation	Rule	If the actual is OFF →
Metabolic acidosis	↓ PaCO₂	Winter's: PaCO₂ = 1.5 × HCO₃ + 8 ± 2[1]	Higher → concurrent respiratory acidosis; Lower → concurrent respiratory alkalosis
Metabolic alkalosis	↑ PaCO₂	PaCO₂ rises ~0.7 mmHg per 1 mmol/L rise in HCO₃ (limited by hypoxia)	Higher → concurrent respiratory acidosis; Lower → concurrent respiratory alkalosis
Acute respiratory acidosis	↑ HCO₃	HCO₃ rises ~1 mmol/L per 10 mmHg rise in PaCO₂	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis
Chronic respiratory acidosis (COPD)	↑ HCO₃	HCO₃ rises ~4 mmol/L per 10 mmHg rise in PaCO₂ (renal, over days)	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis
Acute respiratory alkalosis	↓ HCO₃	HCO₃ falls ~2 mmol/L per 10 mmHg fall in PaCO₂	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis
Chronic respiratory alkalosis	↓ HCO₃	HCO₃ falls ~4–5 mmol/L per 10 mmHg fall in PaCO₂ (renal, over days)	Higher HCO₃ → concurrent metabolic alkalosis; Lower → concurrent metabolic acidosis

The Stewart (strong ion difference) approach — the physical-chemical view

The PaCO₂ — the respiratory variable, identical in both frameworks.
The strong ion difference (SID) — the difference between the fully dissociated (strong) cations and anions. In practice this is dominated by Na⁺ − Cl⁻, with a normal value of about 40 mmol/L. The strong ions cannot combine with hydrogen, so the SID acts like an electrical charge gap that water dissociation must fill: a small SID (a fall in sodium or, far more commonly, a rise in chloride — as with large-volume 0.9% saline) drives water dissociation towards more free H⁺ and an acidosis; a large SID (chloride loss from vomiting or diuretics) drives an alkalosis.[6][13]
The total concentration of weak acids (Aₜₒₜ) — predominantly albumin and, to a lesser extent, phosphate. A fall in Aₜₒₜ (hypoalbuminaemia, ubiquitous in critical illness) is itself an alkalinising process, which is why a low albumin masks an underlying high anion-gap acidosis — the basis for the albumin correction of the AG.[7][13]

The Stewart framework earns its place at the bedside by explaining three things the traditional model can only label:[6][16]

Why large-volume 0.9% saline causes a hyperchloraemic acidosis — it loads chloride without bicarbonate, shrinking the SID. This is the mechanism behind the SMART trial's finding that balanced crystalloids (which preserve the SID) modestly outperform saline.[19][12]
Why hypoalbuminaemia hides a high anion-gap acidosis — it lowers Aₜₒₜ (alkalinising), offsetting the acid; the albumin-corrected AG restores the diagnostic sensitivity.[7]
Why a metabolic alkalosis accompanies chloride depletion (vomiting, diuretics) — the chloride loss widens the SID.[13]

The delta-delta in depth — the delta ratio and the corrected bicarbonate

The delta-delta — what each ratio means

ΔAG / ΔHCO₃⁻ ratio	Interpretation	Concurrent disorder	Classic clinical setting
~1 (0.8–2.0)	The AG rise matches the HCO₃⁻ fall	None — a pure high-AG metabolic acidosis	Straightforward lactic acidosis or DKA
< 1 (e.g. 0.7)	The HCO₃⁻ has fallen MORE than the AG has risen	Concurrent normal-gap (hyperchloraemic) metabolic acidosis	DKA resuscitated with saline; diarrhoea + lactic acidosis; renal failure
> 2	The AG has risen MORE than the HCO₃⁻ has fallen	Concurrent metabolic alkalosis	DKA + vomiting; lactic acidosis + vomiting/diuretics; uraemia + vomiting
< 0.4	The AG is near-normal but the HCO₃⁻ is very low	Predominant / pure normal-gap acidosis (the high-AG component is minor)	Pure hyperchloraemic acidosis from saline or RTA

[1]

The equivalent and often simpler bedside method is the corrected (or "gap-gap") bicarbonate: [1]

Corrected HCO₃⁻ = measured HCO₃⁻ + ΔAG [1]

Mixed metabolic acidosis — the combined high-AG and normal-AG acidosis

The mixed metabolic acidosis — how the delta-delta catches it

Start with a high-AG metabolic acidosis

The patient has a high AG (e.g. 22, lactate 4) with a low HCO₃⁻ (e.g. 14). Na 140, Cl 104.

Calculate the delta-delta

ΔAG = 22 − 12 = 10. ΔHCO₃⁻ = 24 − 14 = 10. Ratio = 10/10 = 1.0 → a PURE high-AG acidosis. So far, so good.

Now resuscitate with 3 litres of 0.9% saline

Re-check the delta-delta

The corrected HCO₃⁻ confirms

Corrected HCO₃⁻ = 11 + 4 = 15 (well under 24) → a concurrent metabolic acidosis. The two methods agree.

The clinical lesson

Mixed metabolic alkalosis + metabolic acidosis — the delta ratio over 2

The DKA-plus-vomiting — the classic metabolic alkalosis + acidosis

The presentation

The blood gas

pH 7.20, HCO₃⁻ 13, PaCO₂ 30. Na 138, Cl 90, K 3.0.

Step 4 — the anion gap

AG = 138 − (90 + 13) = 35 (high-AG acidosis). The delta-delta: ΔAG = 35 − 12 = 23; ΔHCO₃⁻ = 24 − 13 = 11. Ratio = 23/11 = 2.1 → OVER 2.

The corrected HCO₃⁻ confirms

Corrected HCO₃⁻ = 13 + 23 = 36 (well over 24) → a concurrent metabolic alkalosis. The low chloride (90) and low potassium (3.0) point to the vomiting.

Why the pH is "better than expected"

Management implication

Triple disorders — the three-way mixed disturbance

The classic triple disorder — sepsis + vomiting + lactic acidosis + hyperventilation

The scenario

A 60-year-old with pyelonephritis and septic shock, 2 days of vomiting, now tachypnoeic. Three primary disorders are coexisting.

Disorder 1 — the high-AG metabolic acidosis

The lactic acidosis of septic shock (the lactate 5). A primary metabolic acidosis: the HCO₃⁻ falls.

Disorder 2 — the metabolic alkalosis

The vomiting (the loss of HCl and volume, the chloride depletion). A primary metabolic alkalosis: the HCO₃⁻ rises independently.

Disorder 3 — the respiratory alkalosis

The hyperventilation (septic encephalopathy, fever, the early direct medullary stimulation, hepatic failure). A primary respiratory alkalosis: the PaCO₂ falls.

The blood gas may look deceptively simple

pH 7.39, HCO₃⁻ 16, PaCO₂ 27. The pH is near-normal — three opposing processes nearly cancelling. Na 138, Cl 92.

Step 3 — Winter reveals the respiratory component

Expected PaCO₂ = 1.5 × 16 + 8 = 32 ± 2. The actual PaCO₂ (27) is LOWER than expected → a concurrent respiratory ALKALOSIS.

Step 4 — the anion gap

AG = 138 − (92 + 16) = 30 (high-AG acidosis).

Step 5 — the delta-delta reveals the metabolic alkalosis

ΔAG = 30 − 12 = 18; ΔHCO₃⁻ = 24 − 16 = 8. Ratio = 18/8 = 2.25 → OVER 2 → a concurrent metabolic ALKALOSIS. Corrected HCO₃⁻ = 16 + 18 = 34 (well over 24) — confirms.

The synthesis

The common ICU mixed patterns

The high-AG acidosis + the respiratory alkalosis: the sepsis (the lactic acidosis + the hyperventilation), the salicylate toxicity (the early respiratory alkalosis — the direct medullary stimulation + the high-AG acidosis — the salicylate), the hepatorenal syndrome, the severe liver disease.[1]
The high-AG acidosis + the metabolic alkalosis: the vomiting (the metabolic alkalosis) + the lactic acidosis or the DKA (the "DKA plus the vomiting"). The delta-delta over 2.[1]
The high-AG + the normal-gap (hyperchloraemic) acidosis: the DKA recovery (the resolving ketoacidosis + the saline-induced hyperchloraemia), the diarrhoea + the lactic acidosis, the renal failure.[1]
The mixed respiratory + the metabolic: the COPD (the chronic respiratory acidosis) + the metabolic alkalosis (the diuretic); the cardiac arrest (the respiratory + the metabolic acidosis); the respiratory failure + the sepsis.[1]
The triple disorder: the sepsis + the vomiting + the lactic acidosis + the hyperventilation (the metabolic acidosis + the metabolic alkalosis + the respiratory alkalosis); the salicylate + the vomiting (the respiratory alkalosis + the high-AG acidosis + the metabolic alkalosis).[1]

The one-paragraph exam answer

[1]

Worked examples — the 5-step approach applied

Worked example 1 — the septic patient with a near-normal pH (a mixed disorder in disguise)

The gas and the electrolytes

pH 7.38, PaCO₂ 30, HCO₃⁻ 17. Na 140, Cl 105, albumin 30 g/L. Lactate 4.5.

Step 1–2 — the pH and the primary

Step 3 — the compensation

If the primary is a metabolic acidosis, expected PaCO₂ = 1.5 × 17 + 8 = 33.5 ± 2. The actual (30) is LOWER → a concurrent respiratory ALKALOSIS.

Step 4 — the albumin-corrected anion gap

AG = 140 − (105 + 17) = 18. Corrected for albumin: the albumin is 30 (10 below 40), so add 2.5 → corrected AG ~20.5 (high-AG acidosis). The lactate (4.5) accounts for most of the ΔAG.

Step 5 — the delta-delta

ΔAG = 20 − 12 = 8; ΔHCO₃⁻ = 24 − 17 = 7. Ratio = 8/7 = 1.14 → about 1, a pure high-AG acidosis (no concurrent metabolic disorder).

The diagnosis

A high-AG metabolic acidosis (lactic, from sepsis) + a respiratory alkalosis (hyperventilation). Two primary disorders — the classic septic pattern.

[1]

Worked example 2 — DKA that will not clear (the saline-induced hyperchloraemic component)

The setting

12 hours into DKA treatment. The ketones have cleared, the glucose is normalising, but the HCO₃⁻ is still low and the patient was resuscitated with 4 litres of 0.9% saline.

The gas and the electrolytes

pH 7.30, PaCO₂ 28, HCO₃⁻ 14. Na 140, Cl 110, albumin 35.

Step 3 — Winter

Expected PaCO₂ = 1.5 × 14 + 8 = 29 ± 2. Actual 28 → within range. The compensation is appropriate (no concurrent respiratory disorder).

Step 4 — the anion gap

AG = 140 − (110 + 14) = 16 (corrected ~17.5). The AG is only mildly raised — the ketoacidosis has largely resolved.

Step 5 — the delta-delta

ΔAG = 17 − 12 = 5; ΔHCO₃⁻ = 24 − 14 = 10. Ratio = 5/10 = 0.5 → UNDER 1 → a concurrent normal-gap (hyperchloraemic) acidosis. Corrected HCO₃⁻ = 14 + 5 = 19 (under 24) — confirms.

The diagnosis and the lesson

The strong ion difference in practice — when Stewart changes management

The Stewart variables — what each does to the pH

Variable	Normal value	A change towards acidosis	A change towards alkalosis	Clinical correlate
PaCO₂	40 mmHg	↑ (hypoventilation)	↓ (hyperventilation)	The respiratory variable — common to both frameworks
Strong ion difference (SID)	~40 mmol/L (Na⁺ − Cl⁻)	↓ — chloride rises (saline) or sodium falls	↑ — chloride falls (vomiting, diuretics)	The dominant metabolic variable in Stewart's terms; explains saline acidosis and vomiting alkalosis
Total weak acids (Aₜₒₜ)	albumin ~40 g/L	↑ (hyperalbuminaemia — rare) or hyperphosphataemia	↓ — hypoalbuminaemia (very common in ICU)	Hypoalbuminaemia is alkalinising and MASKS a high-AG acidosis; this is why the albumin correction of the AG matters
Unmeasured anions (SIG)	~0 (zero)	↑ — lactate, ketones, toxins, uraemia	—	The Stewart equivalent of the high anion gap; a positive SIG is the acid load

[1]

The albumin correction of the anion gap — the hidden acid load

Corrected AG = measured AG + 2.5 × ( (40 − albumin g/L) / 10 )
A patient with a measured AG of 14 (apparently normal) and an albumin of 20 g/L has a corrected AG of 14 + 2.5 × 2 = 19 — a high-AG acidosis that the uncorrected gap completely conceals. [1]

This is the single most common reason a high-AG acidosis is missed at the bedside. Always correct the AG for albumin in the ICU patient.[7][20]

Short-answer questions

SAQ — Systematic ABG interpretation in the septic, vomiting patient

10 minutes · 10 marks

[1]

SAQ — Compensation limits and the body that never over-compensates

10 minutes · 10 marks

[1]

Clinical pearls

High-yield mixed acid-base pearls for the CICM / FFICM / EDIC exam

The body NEVER over-compensates. This single principle is the foundation of the entire topic. If the measured compensation is more or less than the predicted (Winter for a metabolic acidosis; the acute/chronic rules for the respiratory), a SECOND primary disorder is present. Always calculate the expected compensation and compare it to the actual — never accept a single disorder at face value.[1][8]
A near-normal pH in a sick ICU patient is a mixed disorder until proven otherwise. Two opposing primary disorders (e.g. a metabolic acidosis + a respiratory alkalosis) pull the pH in opposite directions and produce a deceptively "normal" looking gas. The abnormalities are in the PaCO₂ and the HCO₃⁻, not the pH. If the patient is sick and the pH is 7.40, work the five steps.[5]
Winter's formula is the single calculation that most often unmasks a mixed disorder. Expected PaCO₂ = 1.5 × HCO₃⁻ + 8 ± 2. It takes ten seconds and it catches the concurrent respiratory acidosis or alkalosis that the pH alone will never reveal. Albert, Dell and Winters derived it in 1967 from observed compensatory hyperventilation — it remains the bedside standard.[1]
The delta-delta (ΔAG/ΔHCO₃⁻) catches the concurrent METABOLIC disorder that Winter cannot. Winter tests the respiratory compensation; the delta-delta tests for a second metabolic process. A ratio under 1 = a concurrent normal-gap acidosis; over 2 = a concurrent metabolic alkalosis. The corrected HCO₃⁻ (measured HCO₃⁻ + ΔAG) gives the same answer by a simpler route.[4]
Always correct the anion gap for albumin in the ICU. The AG falls ~2.5 mmol/L per 10 g/L fall in albumin. A hypoalbuminaemic ICU patient (albumin 20) with a "normal" measured AG of 14 actually has a corrected AG of 19 — a high-AG acidosis the uncorrected gap hides. The albumin correction restores the diagnostic sensitivity and is non-negotiable in critical illness.[7][20]
The salicylate toxicity is the classic mixed disorder — respiratory alkalosis + high-AG acidosis from the start. Salicylate directly stimulates the medulla (an early respiratory alkalosis) AND uncouples oxidative phosphorylation (a high-AG metabolic acidosis with lactate and ketones). The two coexist from the early presentation. Treat with urine alkalinisation (sodium bicarbonate to trap the salicylate) and haemodialysis for the severe case. Do not miss salicylate in the unexplained mixed disorder — check the level.[1][9]
The septic patient often has THREE disorders — the lactic acidosis + the respiratory alkalosis + (if vomiting) the metabolic alkalosis. This is the classic triple disorder. The pH may be near-normal (three opposing processes). Only the systematic approach — pH, primary, Winter (the respiratory), AG, delta-delta (the metabolic alkalosis) — uncovers all three.[1]
The DKA-plus-vomiting is the textbook mixed metabolic disorder — the delta ratio over 2. The vomiting (metabolic alkalosis) partially rescues the bicarbonate that the ketoacidosis (high-AG acidosis) has consumed. The corrected HCO₃⁻ is over 24. Watch the potassium — the alkalosis and the insulin-driven intracellular shift conspire to drop it dangerously.[2][4]
The "DKA that will not clear" after 12 hours is usually a saline-induced hyperchloraemic acidosis, not ongoing ketoacidosis. The AG has normalised (the ketones have gone) but the HCO₃⁻ stays low and the chloride is high. The delta ratio is under 1. Switch from 0.9% saline to a balanced crystalloid (SMART shows measurable benefit) and the HCO₃⁻ will recover — no bicarbonate, no extra insulin.[19][12]
0.9% saline causes a hyperchloraemic acidosis by lowering the strong ion difference (Stewart). Saline loads chloride (154 mmol/L) without bicarbonate, shrinking the SID (Na⁺ − Cl⁻) from ~40 towards ~30. This is the physical-chemical explanation for the clinical finding that balanced crystalloids (Plasma-Lyte, Hartmann) outperform saline in the SMART and SALT-ED trials. Prefer balanced crystalloids for resuscitation; reserve saline for the hyponatraemic or chloride-losing patient.[19][6]
Hypoalbuminaemia is itself an alkalinising process — it masks the high-AG acidosis and lowers the apparent AG. This is why the albumin correction matters, and it is the Stewart "low Aₜₒₜ" effect. A low albumin also explains why some ICU patients have a higher-than-expected pH for their apparent acid load.[7][13]
The metabolic alkalosis can delay weaning from mechanical ventilation. A chronic respiratory acidosis (COPD) with a superimposed metabolic alkalosis (diuretics, vomiting) leaves the patient with a high bicarbonate and a high baseline PaCO₂. When the ventilator is reduced, the PaCO₂ must rise to maintain the pH, and the patient fails to wean. Correct the metabolic alkalosis (replete chloride and potassium, stop the diuretic) before weaning.[3]
In a high-AG acidosis, a falling lactate with a PERSISTENT acidaemia signals a concurrent normal-gap (hyperchloraemic) acidosis. The delta ratio falls below 1 as the lactate clears. This is the commonest reason a "recovering" septic patient stays acidaemic — the saline resuscitation has superimposed a hyperchloraemic component. Check the chloride and the corrected HCO₃⁻.[4][17]
The corrected HCO₃⁻ (measured HCO₃⁻ + ΔAG) is the bedside shortcut to the delta-delta. Over 24 → a concurrent metabolic alkalosis; under 24 → a concurrent metabolic acidosis. It is mathematically equivalent to the delta ratio and most clinicians find it faster to compute at the bedside. Use whichever you find more intuitive — but use ONE of them in every high-AG acidosis.[4][5]
Sodium bicarbonate is NOT indicated for a metabolic acidosis outside the peri-arrest, hyperkalaemic, or tricyclic-antidepressant setting. BICAR-ICU found no overall outcome benefit for bicarbonate in severe metabolic acidaemia (pH ≤ 7.20). Treat the cause (restore perfusion, dialyse, insulin, fomepizole) — the acidosis corrects when the underlying process reverses. The exception is the Acute Kidney Injury Network (AKIN) subgroup of BICAR-ICU, where bicarbonate reduced the need for renal replacement therapy.[18][9]
The osmolar gap + the high-AG acidosis = a toxic alcohol (methanol or ethylene glycol) until proven otherwise. The parent alcohol raises the osmolar gap; the metabolites (formic, glycolic, oxalic acid) raise the anion gap. The injury is done by the metabolites, so fomepizole (to block alcohol dehydrogenase) and haemodialysis are time-critical. A normal osmolar gap does NOT exclude a late presentation once the parent alcohol has been fully metabolised.[1][9]
The Henderson–Hasselbalch equation (pH = 6.1 + log(HCO₃⁻ / (0.03 × PaCO₂))) sets the pH by the RATIO, not the absolute values. A HCO₃⁻ of 12 with a PaCO₂ of 27 gives the same pH as a HCO₃⁻ of 24 with a PaCO₂ of 54 — the first is a compensated metabolic acidosis, the second a compensated respiratory acidosis. The ratio (normally ~20:1) is the number to hold in your head.[8]
Base excess and standard base excess quantify the metabolic component, corrected for the respiratory. The standard base excess (SBE) is the metabolic component referenced to a PaCO₂ of 40, removing the respiratory influence. A negative SBE is a metabolic acidosis; a positive SBE a metabolic alkalosis. Schlichtig (1998) defined the expected compensation in terms of the SBE — useful when the HCO₃⁻ is itself confounded.[14][11]

The metabolic alkaloses — chloride-responsive and resistant

The metabolic alkalosis — chloride-responsive vs resistant

Feature	Chloride-RESPONSIVE (urine Cl⁻ < 20)	Chloride-RESISTANT (urine Cl⁻ > 20)
Mechanism	Volume and chloride depletion → the kidney retains bicarbonate	Mineralocorticoid excess or severe K⁺ depletion → bicarbonate generation
Common causes	Vomiting, NG suction, diuretics, post-hypercapnia, villous adenoma	Primary hyperaldosteronism, Cushing's, exogenous mineralocorticoids, severe hypokalaemia, Bartter/Gitelman
Urine chloride	< 10–20 mmol/L (low)	> 20 mmol/L (high)
Blood pressure	Usually normal or low	Usually HIGH (mineralocorticoid) — except Bartter/Gitelman
Treatment	Normal saline + KCl repletion (restores volume and chloride)	Treat the cause (spironolactone, remove the adenoma); K⁺ repletion
Relevance to the mixed disorder	The common metabolic alkalosis in the ICU mixed picture (vomiting, diuretics)	Rare but consider in the hypertensive ICU patient with an alkalosis

[1]

Trials and evidence

Albert, Dell & Winters — the origin of Winter's formula (PMID 6016545)

Source

Annals of Internal Medicine, 1967 — the foundational observational study from which Winter's formula is derived

Question

What is the quantitative relationship between the serum bicarbonate and the compensatory PaCO₂ in a metabolic acidosis?

Method

Analysis of patients with metabolic acidosis of varying severity, plotting the observed PaCO₂ against the serum total CO₂ (bicarbonate) content

Key result

The compensatory PaCO₂ fell predictably with the bicarbonate, fitting the regression PaCO₂ = 1.5 × HCO₃⁻ + 8 (± 2). This is the equation every intensivist still uses at the bedside

Enduring legacy

Over 50 years later, Winter's formula remains the standard test of compensation for a metabolic acidosis — the calculation that most often unmasks a concurrent respiratory disorder

[1]

Wrenn K — The delta gap: an approach to mixed acid-base disorders (PMID 2240729)

Source

Annals of Emergency Medicine, 1990 — the bedside method that brought the delta-delta into routine clinical use

Concept

In a high-AG metabolic acidosis, the rise in the AG (ΔAG) should match the fall in the bicarbonate (ΔHCO₃⁻). A mismatch reveals a concurrent metabolic disorder

The method

Compare the ΔAG to the ΔHCO₃⁻. A ratio of ~1 = a pure high-AG acidosis; under 1 = a concurrent normal-gap acidosis; over 2 = a concurrent metabolic alkalosis

Clinical impact

Gave clinicians a single calculation that exposes the concurrent metabolic disorder hiding inside an apparent single high-AG acidosis — the cornerstone of the mixed-metabolic-disorder diagnosis

[1]

Jaber et al — BICAR-ICU: sodium bicarbonate in severe metabolic acidaemia (PMID 29910040)

Design

Multicentre, open-label, randomised controlled trial — 389 critically ill adults with severe metabolic acidaemia (pH ≤ 7.20, HCO₃⁻ ≤ 20)

Intervention

Sodium bicarbonate 4.2% infusion to maintain pH above 7.30 vs no bicarbonate (standard care)

Primary outcome

Composite of death by day 28 or at least one organ failure at day 7 — NO significant difference in the primary intention-to-treat population

Subgroup signal

Bottom line

[1]

Semler et al — SMART: balanced crystalloids vs saline in critically ill adults (PMID 29768150)

Design

Pragmatic, cluster-crossover, single-centre randomised trial — 15,802 critically ill adults across 5 ICUs at Vanderbilt

Intervention

Balanced crystalloids (lactated Ringer's or Plasma-Lyte) vs 0.9% saline for all intravenous fluid

Primary outcome

Composite of death, new renal replacement therapy, or persistent renal dysfunction at 30 days

Key result

Acid-base relevance

[1]

Moviat et al — Conventional vs physicochemical (Stewart) approach in ICU metabolic acidosis (PMID 12793889)

Source

Critical Care, 2003 — a head-to-head comparison of the traditional and Stewart approaches in the ICU

Question

Does the Stewart (strong ion difference) method detect mixed disorders the traditional approach misses?

Finding

Lesson

The Stewart framework adds sensitivity for the SID-driven (saline, chloride) and Aₜₒₜ-driven (albumin) disorders; it complements rather than replaces the traditional approach

[1]

Schlichtig, Grogono & Severinghaus — PaCO₂ and standard base excess compensation (PMID 9671365)

Source

Critical Care Medicine, 1998 — the definitive compilation of human acid-base compensation limits

Concept

Defined the EXPECTED compensation for each primary disorder in terms of the standard base excess (SBE), and the limits beyond which a mixed disorder must be present

Key contribution

[1]

Rastegar A — Clinical utility of Stewart's method (PMID 19520748)

Source

Clinical Journal of the American Society of Nephrology, 2009 — the accessible review that brought Stewart into mainstream nephrology and critical care teaching

Core teaching

Bedside relevance

The review clinicians cite when justifying the albumin correction of the AG and the preference for balanced crystalloids over saline

[1]

Adrogué & Madias — Management of life-threatening acid-base disorders (PMID 9414329, 9420343)

Source

New England Journal of Medicine, 1998 (two parts) — the authoritative clinical review of acid-base management

Part 1 (9414329)

Covers the metabolic acidoses — the high-AG and normal-AG, the diagnosis by the anion gap and the delta-delta, and the management (treat the cause, dialysis, the role of bicarbonate)

Part 2 (9420343)

Covers the metabolic alkaloses (chloride-responsive and resistant), the respiratory acidoses and alkaloses, and the mixed disorders — with the compensation rules and the delta-delta method

Enduring status

Still the most-cited single reference for the systematic approach to acid-base disorders; the framework the examiners expect you to reproduce

[1]

Red flags

References

[1]Albert MS, Dell RB, Winters RW Quantitative displacement of acid-base equilibrium in metabolic acidosis Ann Intern Med, 1967.PMID 6016545
[2]Adrogue HJ, Madias NE Management of life-threatening acid-base disorders. First of two parts N Engl J Med, 1998.PMID 9414329
[3]Adrogue HJ, Madias NE Management of life-threatening acid-base disorders. Second of two parts N Engl J Med, 1998.PMID 9420343
[4]Wrenn K The delta (delta) gap: an approach to mixed acid-base disorders Ann Emerg Med, 1990.PMID 2240729
[5]Haber RJ A practical approach to acid-base disorders West J Med, 1991.PMID 1843849
[6]Emmett M Stewart Versus Traditional Approach to Acid-Base Disorders Anesth Analg, 2016.PMID 27636585
[7]Rastegar A Clinical utility of Stewart's method in diagnosis and management of acid-base disorders Clin J Am Soc Nephrol, 2009.PMID 19520748
[8]Kraut JA, Madias NE Approach to patients with acid-base disorders Respir Care, 2001.PMID 11262558
[9]Kraut JA, Madias NE Treatment of acute metabolic acidosis: a pathophysiologic approach Nat Rev Nephrol, 2012.PMID 22945490
[10]Gomez H, Kellum JA Understanding Acid Base Disorders Crit Care Clin, 2015.PMID 26410149
[11]Berend K Diagnostic Use of Base Excess in Acid-Base Disorders N Engl J Med, 2018.PMID 30067929
[12]Moviat M, van Haren F, van der Hoeven H Conventional or physicochemical approach in intensive care unit patients with metabolic acidosis Crit Care, 2003.PMID 12793889
[13]Story DA, Kellum JA New aspects of acid-base balance in intensive care Curr Opin Anaesthesiol, 2004.PMID 17021538
[14]Schlichtig R, Grogono AW, Severinghaus JW Human PaCO2 and standard base excess compensation for acid-base imbalance Crit Care Med, 1998.PMID 9671365
[15]Moviat M, Terpstra AM, Ruitenbeek W, Kluijtmans LA, Pickkers P, van der Hoeven HG Contribution of various metabolites to the unmeasured anions in critically ill patients with metabolic acidosis Crit Care Med, 2008.PMID 18176310
[16]Story DA Acid-Base Analysis in the Operating Room: A Bedside Stewart Approach Anesthesiology, 2023.PMID 37934110
[17]Al-Jaghbeer M, Kellum JA Acid-base disturbances in intensive care patients: etiology, pathophysiology and treatment Nephrol Dial Transplant, 2015.PMID 25213433
[18]Jaber S, Paugam C, Futier E, et al Sodium bicarbonate therapy for patients with severe metabolic acidaemia in the intensive care unit (BICAR-ICU): a multicentre, open-label, randomised controlled, phase 3 trial Lancet, 2018.PMID 29910040
[19]Semler MW, Self WH, Wanderer JP, et al Balanced Crystalloids versus Saline in Critically Ill Adults N Engl J Med, 2018.PMID 29768150
[20]Berend K, Duits AJ The role of the clinical laboratory in diagnosing acid-base disorders Crit Rev Clin Lab Sci, 2019.PMID 30917291