Overview — Why a System Beats Pattern-Recognition
"The most common error in acid–base interpretation is stopping at the first abnormality. A pH of 7.36 looks reassuring, yet may conceal a severe metabolic acidosis fully compensated by respiratory alkalosis with a coexisting metabolic alkalosis — three disorders in one 'near-normal' gas. A disciplined, stepwise method uncovers every process present and, more importantly, points to the diagnoses driving them."
Washington Manual of Critical Care, 4th Ed. Wolters Kluwer 2023. Chapter: Acid–Base Disorders.The Vocabulary — Get It Right First
- Acidaemia / Alkalaemia — the actual blood pH (<7.35 / >7.45). Describes the blood.
- Acidosis / Alkalosis — the process that tends to lower/raise pH. You can have an acidosis without acidaemia (if a competing alkalosis dominates).
- Primary disorder — the initiating process. Compensation — the predictable, secondary physiological response that limits the pH change but never fully corrects or overcorrects it.
The body never overcompensates. If you calculate "compensation" that would push the pH past normal to the other side, a second primary disorder is present. Likewise, compensation outside the predicted range (too much or too little) signals a mixed disorder. This single principle is what turns acid–base from guesswork into diagnosis.
Marino Physiology — The Henderson–Hasselbalch Backbone
"All of clinical acid–base analysis rests on a single relationship between the three measured quantities: pH is determined by the ratio of bicarbonate to the partial pressure of carbon dioxide. A metabolic process changes the numerator (HCO₃⁻); a respiratory process changes the denominator (PaCO₂). Every acid–base problem is, at heart, a question of which term moved first and how the other responded."
Marino PL. The ICU Book, 5th Ed. Acid–Base Interpretation. Wolters Kluwer; 2025.↑HCO₃⁻ or ↓PaCO₂ → alkalaemia · ↓HCO₃⁻ or ↑PaCO₂ → acidaemia
Respiratory compensation (changing ventilation to adjust PaCO₂) begins within minutes and is complete within hours. Renal (metabolic) compensation (retaining or excreting HCO₃⁻) takes 2–5 days to be complete. This is why respiratory acidosis has separate acute and chronic compensation expectations — the kidney needs days to catch up. Knowing the tempo tells you the chronicity of the disturbance.
The anion gap exposes acids whose anion isn't on the routine panel (lactate, ketoacids, etc.). Because albumin is the dominant unmeasured anion, a low albumin lowers the "normal" gap: for every 1 g/dL fall in albumin below 4.5, the expected gap falls by ~2.5 mEq/L. The hypoalbuminaemic ICU patient can have a "normal" gap of 4–6 — so a measured gap of 12 in that patient is actually a high-gap acidosis. Always use the albumin-corrected anion gap in the critically ill.
The Method That Never Misses a Disorder
Look at the pH — acidaemia or alkalaemia?
pH <7.35 = acidaemia; >7.45 = alkalaemia. Even a "normal" pH (7.35–7.45) can hide balanced mixed disorders — keep going regardless.
Identify the primary process — metabolic or respiratory?
Find which of HCO₃⁻ and PaCO₂ explains the pH. If pH and PaCO₂ move in opposite directions → primary respiratory. If they move in the same direction → primary metabolic. (e.g. low pH + low HCO₃ = metabolic acidosis; low pH + high PaCO₂ = respiratory acidosis.)
Check compensation — is it appropriate?
Apply the expected-compensation formula (step 4). If the actual value sits outside the predicted range → a second primary disorder is present. Compensation never normalises the pH.
Calculate the anion gap (albumin-corrected)
AG = Na − (Cl + HCO₃); correct for albumin. A high gap = a high-anion-gap metabolic acidosis is present — even if the pH or HCO₃ look normal. Always calculate the gap on every gas with electrolytes.
If high gap → check the delta-delta (Δ/Δ) ratio
Compares the rise in anion gap to the fall in bicarbonate. Reveals a coexisting normal-gap acidosis (ratio <1) or metabolic alkalosis (ratio >2) hiding behind the high-gap acidosis.
Find the cause — and (if high gap) the osmolar gap
Build the differential (MUDPILES / GI vs renal). In high-gap acidosis of unclear cause, calculate the osmolar gap to screen for toxic alcohols (methanol, ethylene glycol). Then treat the underlying disease — not the number.
💡 Let the tool do the arithmetic: the ABG Analyser performs all six steps automatically — primary disorder, Winter's/compensation check, corrected anion gap, delta-delta ratio, osmolar gap, and an acute management plan. This page is the "why" behind what the tool computes.
The Numbers You Must Know
(1.5 × HCO₃) + 8 ± 2Actual > expected → concurrent respiratory acidosis · Actual < expected → concurrent respiratory alkalosis
0.7 × (HCO₃ − 24) + 40 ± 2 (rises ~0.7 mmHg per 1 mEq/L HCO₃)
+1 mEq/L | Chronic: +3.5–4 mEq/LRespiratory alkalosis — expected HCO₃ fall per 10 mmHg ↓PaCO₂ Acute:
−2 mEq/L | Chronic: −4–5 mEq/L
Na − (Cl + HCO₃) (normal 8–12)Corrected AG =
AG + 2.5 × (4.5 − albumin g/dL)
(measured AG − 12) ÷ (24 − measured HCO₃)<0.4 pure NAGMA · 0.4–0.8 mixed high+normal gap · 1–2 pure high-gap · >2 high-gap + metabolic alkalosis (or chronic resp acidosis)
2×Na + glucose/18 + BUN/2.8 (mg/dL units)Osmolar gap = measured − calculated. >10–15 → toxic alcohol (methanol, ethylene glycol), mannitol, or other unmeasured osmole.
Differentials You Can Recall at the Bedside
Lactate and ketoacidosis are by far the commonest in the ICU. Always send a lactate and ketones; calculate the osmolar gap if the cause isn't obvious.
The urine anion gap (Na + K − Cl) separates the two big groups: negative UAG → GI loss (appropriate renal NH₄⁺ excretion); positive UAG → renal cause (RTA).
Saline-resistant (urine Cl >20): hyperaldosteronism, Cushing's, severe hypokalaemia, Bartter/Gitelman. Treat the cause; saline won't fix it.
Putting the 6 Steps Together
ABG: pH 7.38, PaCO₂ 28, HCO₃ 16, Na 140, Cl 96, albumin 4.0.
Step 1–2: pH near-normal but low HCO₃ + low PaCO₂ → a metabolic acidosis with respiratory response.
Step 3 (Winter's): expected PaCO₂ = 1.5×16 + 8 = 32 ± 2 (30–34). Actual 28 is lower → an additional respiratory alkalosis.
Step 4 (AG): 140 − (96 + 16) = 28; corrected +2.5×(4.5−4.0)=+1.25 → ~29. High-gap acidosis.
Step 5 (Δ/Δ): (28−12)/(24−16) = 16/8 = 2.0 → at the upper limit; a coexisting metabolic alkalosis is possible.
Conclusion: high-gap metabolic acidosis + respiratory alkalosis (± metabolic alkalosis) — e.g. salicylate poisoning or sepsis. The near-normal pH concealed three processes.
ABG: pH 7.20, PaCO₂ 24, HCO₃ 9, Na 138, Cl 110, albumin 4.5, glucose high, ketones positive.
Step 1–3: acidaemia, metabolic acidosis. Winter's expected PaCO₂ = 1.5×9+8 = 21.5 ± 2 (19.5–23.5); actual 24 is slightly high → respiratory compensation is mildly inadequate (early fatigue).
Step 4 (AG): 138 − (110 + 9) = 19 → high gap (ketoacidosis).
Step 5 (Δ/Δ): (19−12)/(24−9) = 7/15 = 0.47 → <1 → a coexisting normal-gap acidosis (here, hyperchloraemia from saline resuscitation / urinary ketone loss).
Conclusion: high-gap (ketoacidosis) + normal-gap (hyperchloraemic) metabolic acidosis. Explains why the HCO₃ stays low even as ketones clear — anticipate this during DKA treatment.
When the Traditional Model Falls Short
"The Stewart approach treats pH as a dependent variable determined by three independent factors: the strong ion difference, the total weak acid concentration (largely albumin and phosphate), and PaCO₂. Its great clinical contribution is explaining the acidosis of large-volume saline — not a gain of acid, but a fall in the strong ion difference as chloride rises relative to sodium."
Marino PL. The ICU Book, 5th Ed. Acid–Base Interpretation (Stewart approach). Wolters Kluwer; 2025.Strong Ion Difference (SID) = (Na + K + Ca + Mg) − (Cl + lactate). Normal ~40 mEq/L. A fall in SID (e.g. ↑Cl from saline, ↑lactate) causes acidosis; a rise in SID causes alkalosis. Albumin and phosphate are the weak acids — hypoalbuminaemia is independently alkalinising (it explains why malnourished ICU patients often have a metabolic alkalosis you can't otherwise account for). You don't need to compute Stewart at the bedside, but it gives a coherent explanation for the three commonest ICU puzzles: saline (hyperchloraemic) acidosis, dilutional acidosis, and hypoalbuminaemic alkalosis.
The traditional Henderson/anion-gap method and Stewart usually agree on the bedside diagnosis. Use the anion-gap method day-to-day (fast, validated), and reach for Stewart concepts to explain the otherwise-puzzling gas: the saline-resuscitated trauma patient with a hyperchloraemic acidosis, or the hypoalbuminaemic patient whose "metabolic alkalosis" is really just a low weak-acid state. Switching to balanced crystalloids (Plasma-Lyte / Ringer's lactate) avoids the iatrogenic hyperchloraemic acidosis of large-volume saline.
Common Mistakes in Acid–Base Interpretation
A pH near 7.40 is not "normal acid–base" until you have calculated the anion gap and checked compensation. Balanced mixed disorders (e.g. metabolic acidosis + metabolic alkalosis) produce a deceptively normal pH. Always complete all six steps.
In the hypoalbuminaemic ICU patient, the "normal" anion gap is reduced by ~2.5 per 1 g/dL fall in albumin. An uncorrected gap of 10–12 may actually represent a significant high-gap acidosis. Always use the albumin-corrected anion gap in critical illness.
A high-gap acidosis can coexist with a normal-gap acidosis or a metabolic alkalosis that the anion gap alone hides. Skipping the Δ/Δ ratio means missing the third process — e.g. the DKA patient who is also vomiting (alkalosis) or saline-loaded (hyperchloraemic acidosis).
A chronic CO₂ retainer (COPD) has a compensatory high bicarbonate — that is appropriate compensation, not a separate metabolic alkalosis. Use the acute vs chronic compensation formulas before labelling the bicarbonate abnormal, and beware over-ventilating these patients to a "normal" CO₂ (causes post-hypercapnic alkalosis).
BICAR-ICU showed no benefit from bicarbonate in lactic acidosis (a possible signal only in concurrent AKI). Bicarbonate raises CO₂, can worsen intracellular acidosis, and lowers ionised calcium. Treat the cause (perfusion, sepsis, toxin) — reserve bicarbonate for pH <7.1 with specific indications.
In an unexplained high-gap acidosis, a normal osmolar gap does not fully exclude toxic alcohol late (the parent alcohol is metabolised to acid, closing the osmolar gap as the anion gap opens), but a high osmolar gap early is a red flag. Send the osmolality and act before confirmatory levels return — fomepizole/ethanol and dialysis are time-critical.
References
- Marino PL. The ICU Book, 5th Ed. Acid–Base Interpretation; Organic Acidoses; Metabolic Alkalosis (Ch.31–33). Wolters Kluwer; 2025.
- Berend K, de Vries APJ, Gans ROB. Physiological Approach to Assessment of Acid–Base Disturbances. N Engl J Med 2014;371:1434–1445.
- Kraut JA, Madias NE. Lactic Acidosis. N Engl J Med 2014;371:2309–2319.
- Jaber S, Paugam C, Futier E et al. (BICAR-ICU). Sodium bicarbonate therapy for patients with severe metabolic acidaemia in the ICU. Lancet 2018;392:31–40.
- Stewart PA. Modern quantitative acid–base chemistry. Can J Physiol Pharmacol 1983;61:1444–1461.
- Semler MW, Self WH, Wanderer JP et al. (SMART). Balanced Crystalloids versus Saline in Critically Ill Adults. N Engl J Med 2018;378:829–839.
- Irwin RS, Rippe JM. Irwin and Rippe's Intensive Care Medicine, 8th Ed. Acid–Base Disorders. Wolters Kluwer; 2018.
- Washington Manual of Critical Care, 4th Ed. Chapter: Acid–Base Disorders. Wolters Kluwer 2023.