Acid-Base Disorders
A systematic 6-step approach to every ABG. pH → primary disorder → compensation → anion gap → delta-delta. Master this and you'll never be stumped by an ABG again.
🧪 6-Step ABG Interpretation
Step 1 -Look at the pH
| pH | Primary Process |
|---|---|
| < 7.35 | Acidemia (acidosis is the dominant process) |
| 7.35–7.45 | Normal (or mixed disorder with complete compensation) |
| > 7.45 | Alkalemia (alkalosis is the dominant process) |
Step 2 -Identify the Primary Disorder
| If pH < 7.35 (acidemia) | If pH > 7.45 (alkalemia) |
|---|---|
| PCO₂ > 40 → respiratory acidosis (hypoventilation) | PCO₂ < 40 → respiratory alkalosis (hyperventilation) |
| HCO₃ < 22 → metabolic acidosis | HCO₃ > 26 → metabolic alkalosis |
Step 3 -Is Compensation Appropriate?
| Primary Disorder | Expected Compensation |
|---|---|
| Metabolic acidosis | Winter's formula: expected PCO₂ = 1.5 × [HCO₃] + 8 (± 2). If actual PCO₂ ≠ expected → additional respiratory disorder. |
| Metabolic alkalosis | Expected PCO₂ = 0.7 × [HCO₃] + 21 (± 2). Or: PCO₂ rises ~0.7 for each 1 mEq/L rise in HCO₃. |
| Acute respiratory acidosis | HCO₃ rises 1 per 10 mmHg ↑ PCO₂ |
| Chronic respiratory acidosis | HCO₃ rises 3.5–4 per 10 mmHg ↑ PCO₂ |
| Acute respiratory alkalosis | HCO₃ falls 2 per 10 mmHg ↓ PCO₂ |
| Chronic respiratory alkalosis | HCO₃ falls 4–5 per 10 mmHg ↓ PCO₂ |
If compensation is more or less than expected → there's a second (mixed) disorder. Example: metabolic acidosis with PCO₂ lower than Winter's predicted → concurrent respiratory alkalosis (sepsis?). Pick the acute or chronic row deliberately. Acute is minutes to hours and reflects tissue buffering alone; chronic means renal compensation has run to completion, which takes 3 to 5 days and shifts bicarbonate several times further. Using the acute formula on a chronic CO₂ retainer invents a metabolic alkalosis that is not there, and using the chronic formula on an acute one hides a real second disorder.
Step 4 -Calculate the Anion Gap
AG = Na − (Cl + HCO₃). Normal: about 3–11 on the ion-selective electrode analyzers most labs now use, which read chloride higher than older methods; the classic 12 ± 4 comes from older analyzers and is still the teaching anchor for MUDPILES and delta-delta. Use your own lab's reference range when calling a gap high or low. Correct for albumin: for every 1 g/dL albumin below 4 → add 2.5 to the AG.
MUDPILES (AG Elevated > 12)
| Cause | Mechanism |
|---|---|
| M ethanol | Alcohol dehydrogenase converts methanol to formic acid, which inhibits cytochrome oxidase and damages the retina (optic disc edema, blindness). Treat with fomepizole + dialysis. |
| U remia | Failing kidneys cannot excrete sulfates, phosphates, urate, and other organic anions. Develops at GFR < 20. |
| D KA (also alcoholic KA, starvation KA) | Insulin deficiency (or alcohol-induced NADH excess) drives lipolysis and hepatic ketogenesis. Accumulating β-hydroxybutyrate and acetoacetate are the unmeasured anions. |
| P ropylene glycol | Solvent in IV lorazepam, phenobarbital, and diazepam infusions. Metabolized by alcohol dehydrogenase to lactic acid. Classic iatrogenic ICU cause after high-dose benzo drips. |
| I soniazid / Iron | INH depletes pyridoxine (B6) and inhibits GABA synthesis → refractory seizures + lactic acidosis. Treat with high-dose pyridoxine 1:1 (g per g of INH ingested). Iron OD disrupts mitochondrial oxidative phosphorylation → lactate. |
| L actic acidosis #1 cause in hospital | Type A: tissue hypoperfusion or hypoxia (sepsis, shock, ischemia, severe anemia, CO poisoning). Type B: drugs (metformin, linezolid, propofol, NRTIs, β2 agonists), liver failure, malignancy (Warburg effect), thiamine deficiency, D-lactic from short bowel. See full breakdown in Differentials tab. |
| E thylene glycol | Antifreeze. Metabolized to glycolic acid (AG acidosis) and oxalic acid (calcium oxalate crystalluria → AKI). Treat with fomepizole + dialysis. Bedside clues: oxalate crystals in urine, fluorescence under Wood's lamp (some products). |
| S alicylates | Uncouples mitochondrial oxidative phosphorylation → AG acidosis. Also directly stimulates the medullary respiratory center → early respiratory alkalosis. The classic mixed disorder (AGMA + resp alkalosis) is the giveaway. Tinnitus is the bedside tell. |
HARDUPS (AG Normal / Hyperchloremic)
| Cause | Mechanism |
|---|---|
| H yperalimentation (TPN) | Cationic amino acids (lysine, arginine, histidine) in TPN generate HCl as they are metabolized. Modern TPN includes acetate buffer to offset this; classic boards still tests it. |
| A ddison's / Acetazolamide | Addison's (adrenal insufficiency): low aldosterone → impaired distal H+ and K+ excretion (mimics Type 4 RTA, hyperK). Acetazolamide: blocks proximal carbonic anhydrase → bicarbonate wasted in urine. |
| R TA | Type 1 (distal): α-intercalated cells cannot secrete H+ → urine pH > 5.5, hypoK, kidney stones (Sjögren, SLE). Type 2 (proximal): impaired HCO3- reabsorption → Fanconi syndrome, hypoK. Type 4: hypoaldosteronism (DM, ACEi/ARB, TMP-SMX, heparin, spironolactone) → hyperK. Most common RTA. |
| D iarrhea #1 cause of NAGMA | Direct loss of bicarbonate-rich small bowel and colonic secretions. Urine AG is negative (kidneys appropriately excrete NH4+), distinguishing it from RTA. |
| U reteral diversion (ileal conduit, ureterosigmoidostomy) | Bowel mucosa exchanges urinary Cl- for HCO3-. The longer the urine sits in contact with bowel, the more HCO3- is lost. |
| P ost-hypocapnia / Pancreatic fistula | Post-hypocapnia: chronic respiratory alkalosis caused renal HCO3- wasting as compensation; when PaCO2 abruptly normalizes (e.g., post-intubation), bicarb is now inappropriately low. Pancreatic fistula or biliary drain: direct loss of HCO3-rich secretions. |
| S aline (NS resuscitation) | Large-volume 0.9% NS delivers a Cl- load that displaces HCO3- extracellularly = hyperchloremic dilutional acidosis. Switch to LR or PlasmaLyte for ongoing resuscitation. SMART 2018 |
Low or Negative Anion Gap
A low gap (≤ 3 on modern analyzers) or a negative gap is a clue, not a curiosity. The gap is only Na − (Cl + HCO₃), so it falls when unmeasured anions fall, when unmeasured cations rise, or when the lab over-reads chloride or under-reads sodium. Once albumin is accounted for, a negative gap should send you looking for lab error, a chloride artifact (salicylate, bromide, iodide), a paraprotein or lithium. The danger is not the low number itself but what it hides: an artifactually low gap can mask a high-gap acidosis, most importantly salicylate poisoning.
| Cause | Mechanism and Why It Matters |
|---|---|
| Laboratory error Most common cause | A small error in any of three measured values moves the gap directly (a Cl reported 4 high lowers the gap by 4). Repeat the panel before chasing a diagnosis, especially when the gap is negative and nothing else fits. |
| Hypoalbuminemia Most common real cause | Albumin is the main unmeasured anion. The gap falls about 2.5 mEq/L for every 1 g/dL fall in albumin, and chloride rises to fill the space. Cirrhosis, nephrotic syndrome and critical illness routinely drop the baseline, so always correct: a "normal" gap of 12 with albumin 2.0 is really about 17 and hides a gap acidosis. |
| IgG paraprotein (multiple myeloma) and polyclonal gammopathy (cirrhosis, HIV, CKD) | IgG has an isoelectric point above physiologic pH, so it circulates positively charged and acts as an unmeasured cation balanced mainly by chloride. IgA can do the opposite and raise the gap; IgM has no consistent effect. The gap was low in only ~22% of IgG gammopathies, so it cannot screen for myeloma, but an unexplained low gap earns an SPEP with immunofixation. Very high protein can also under-read sodium (below). |
| Salicylate (pseudohyperchloremia) | Salicylate is read as chloride by many chloride ion-selective electrodes, and it is now the most common cause of pseudohyperchloremia. The error grows with the salicylate level and the age of the electrode, and can produce a very small or negative gap that conceals the high-gap acidosis of salicylate poisoning. A low gap never excludes salicylate toxicity: send a level whenever the story fits. |
| Bromide and iodide (pseudohyperchloremia) | Halides are read as chloride, most markedly by ion-selective electrodes (colorimetric methods are affected too). This produces the most profoundly negative gaps (one case of bromism reported a gap of −65). Suspect bromism when a negative gap arrives with confusion, ataxia or psychosis, and ask the lab to measure chloride by another method. |
| Pseudohyponatremia (severe hypertriglyceridemia, marked hyperproteinemia) and sodium above the assay limit | Indirect ion-selective electrodes dilute the sample, so a large lipid or protein fraction makes sodium read falsely low and the gap with it. Direct potentiometry (the blood gas analyzer) gives the true sodium. Sodium above the analyzer's upper limit (often ~170) is also under-reported. |
| Lithium (severe toxicity) | Lithium is an unmeasured cation balanced by chloride and bicarbonate, while only sodium enters the formula. It needs very high levels to matter (a gap of −2 at lithium 14.5 mmol/L in one case), so a therapeutic level does not move the gap, but a low gap in a lithium user is a reason to check the level urgently. |
| Marked hyperkalemia, hypercalcemia, hypermagnesemia | More unmeasured cations push chloride up. The effect is usually small, and the cation itself is visible on the panel, so this is rarely the diagnostic question. |
| Polymyxin B | A polycationic antibiotic that lowers the gap by roughly 1–2.5 mEq/L. Worth knowing so a small drop on therapy is not over-read. |
Working up a low or negative gap, in order: (1) repeat the panel, because error is the most common cause; (2) correct for albumin; (3) look at K, Ca, Mg, triglycerides and total protein; (4) send salicylate and, if relevant, lithium levels and ask about bromide exposure, because these are the causes that hide a dangerous poisoning; (5) if chloride looks implausible, ask the lab to rerun it by a different method; (6) if still unexplained, SPEP with immunofixation for an IgG paraprotein.
Step 5 -Delta-Delta (if AG elevated)
Delta-delta = ΔAG / ΔHCO₃ = (AG − 12) / (24 − HCO₃)
| Ratio | Interpretation | What To Do |
|---|---|---|
| < 0.4 | Essentially a pure non-AG (hyperchloremic) acidosis (the bicarb fell but almost none of it is accounted for by unmeasured anions, so the gap is not the story -think diarrhea, saline, RTA) | Replace the bicarbonate. This is the one band where alkali is genuinely therapeutic, because chloride is not metabolizable, so the lost bicarbonate will never regenerate itself the way lactate and ketones do. Switch large-volume 0.9% NS to a balanced crystalloid (LR, Plasma-Lyte) and replace ongoing GI losses. For RTA, oral alkali (sodium bicarbonate or citrate), with K repletion in types 1 and 2 (both waste potassium, and alkali therapy drives it lower still) and attention to hyperkalemia in type 4. |
| 0.4–1 | AG metabolic acidosis + concurrent non-AG metabolic acidosis (the bicarb dropped more than expected from the AG alone → additional acid or bicarb loss) | Treat the gap cause, which is the dangerous half, but expect the bicarbonate to stay low after it clears. The classic case is DKA resuscitated with large-volume saline: the ketones clear while the chloride load holds the bicarbonate down. Follow beta-hydroxybutyrate, not the bicarbonate, so the insulin drip is not run on against an acidosis that is now chloride rather than ketones. The same logic applies to a diarrheal illness layered on a lactic acidosis. |
| 1–2 | Pure AG metabolic acidosis (the drop in bicarb matches the rise in AG, because a pure gap acidosis consumes one bicarbonate for each unmeasured anion added) | Treat the cause and do not reach for bicarbonate. Lactate and ketones are metabolized back to bicarbonate once the process stops, so the acidosis self-corrects while alkali only buys overshoot alkalosis, ionized hypocalcemia and a sodium load. Insulin and fluids for DKA; perfusion and source control for lactic acidosis; fomepizole plus dialysis for toxic alcohols; thiamine before glucose in alcoholic ketoacidosis (glucose alone can precipitate Wernicke). Reserve bicarbonate for severe acidemia, where the signal is confined to AKI. BICAR-ICU, 2018 |
| > 2 | AG metabolic acidosis + concurrent metabolic alkalosis (vomiting, diuretics, bicarb administration), or a pre-existing chronic respiratory acidosis. Both mean the same thing mathematically: the baseline bicarb was already above 24, so measuring the fall from 24 understates it. | Name the second process before treating either, because the pH can sit near normal while both are severe, and correcting the acidosis alone unmasks the alkalosis. Vomiting or NG suction: 0.9% NS with KCl, since the deficit is chloride and potassium, not free water. Diuretics: hold or reduce. Exogenous bicarbonate: stop it. If instead this is a chronic CO₂ retainer, do not normalize the PaCO₂ quickly -renal compensation takes days to unwind, so you are left with a post-hypercapnic metabolic alkalosis and a suppressed respiratory drive. |
Step 6 -Correlate with the Patient
Match the numbers to the story at the bedside: vomiting or NG suction, DKA, sepsis, COPD, toxic ingestion, diuretics, large-volume saline, diarrhea. A gas that does not fit the history means you have the wrong history or an extra disorder you have not named yet, and it is worth resolving before you act on the numbers. This is also the step that stops the arithmetic from overriding the patient: a delta-delta of 1.8 in someone who has been vomiting for three days is telling you to go back and look for the alkalosis, not to sign out a pure gap acidosis.
📋 Differentials
Metabolic Alkalosis
CLEVER PD (Causes of Metabolic Alkalosis)
| Cause | Mechanism |
|---|---|
| C ontraction (volume depletion) SALINE-RESP | Loss of Cl⁻-rich extracellular fluid (vomiting, diuretics, sweating) makes the proximal tubule reabsorb Na⁺ paired with HCO3- instead of Cl⁻. Volume contraction also activates RAAS → aldosterone drives distal H⁺ secretion. Together: rising and maintained HCO3-. |
| L icorice (glycyrrhizic acid) SALINE-RESIST | Inhibits renal 11β-HSD2, which normally inactivates cortisol to cortisone in the collecting duct. Unopposed cortisol activates the mineralocorticoid receptor → apparent mineralocorticoid excess (Na retention, K⁺ and H⁺ wasting). Mimics primary hyperaldosteronism but with low renin AND low aldosterone. |
| E ndocrine (Conn's, Cushing's, Bartter, Gitelman) SALINE-RESIST | Conn's: autonomous aldosterone → distal Na/K and Na/H exchange. Cushing's: excess cortisol overwhelms 11β-HSD2. Bartter: defective NKCC2 in TAL = "endogenous loop diuretic" (hypoK, hypoCl, alkalosis, normal BP). Gitelman: defective NCC in DCT = "endogenous thiazide" (adds hypoMg). |
| V omiting / NG suction SALINE-RESP Common | Direct loss of HCl from gastric secretions raises serum HCO3- (every H⁺ lost = one HCO3- generated). Concurrent volume and Cl⁻ depletion add a contraction component, perpetuating the alkalosis until both are repleted. |
| E xcess alkali SALINE-RESP | Direct HCO3- load: IV sodium bicarbonate, milk-alkali syndrome (CaCO3 antacids + milk → hyperCa, AKI, alkalosis; full breakdown in Hypercalcemia topic), massive transfusion (citrate metabolized to bicarb by the liver), TPN with acetate buffer overload. |
| R efeeding SALINE-RESP | Carb load triggers insulin surge → cellular uptake of K⁺, phos, Mg, and intracellular H⁺ shift. The resulting hypokalemia and intracellular acidosis maintain extracellular alkalosis. Watch closely in malnourished patients restarting nutrition. Full breakdown → Refeeding Syndrome topic. |
| P ost-hypercapnia SALINE-RESP | Chronic respiratory acidosis (COPD) caused renal HCO3- retention as compensation. When PaCO2 abruptly normalizes (post-intubation overventilation, NIV, mechanical hyperventilation), HCO3- is left inappropriately high → unmasked metabolic alkalosis. Avoid by lowering ventilator support gradually. |
| D iuretics (loop, thiazide) RESIST (active) RESP (off) | Block Na/Cl reabsorption proximal to the collecting duct → increased distal Na⁺ delivery + secondary hyperaldosteronism → distal Na/K and Na/H exchange. Plus volume contraction, hypoK, and hypoCl. While dosed: urine Cl > 20 (resistant). After stopping: urine Cl falls, becomes responsive. K⁺-sparing diuretics (spironolactone, eplerenone, amiloride) do NOT cause alkalosis (and can correct it). |
| Urine Cl⁻ | Category | Causes | Treatment |
|---|---|---|---|
| < 20 mEq/L | Chloride-responsive (saline-responsive) | Vomiting/NG suction (#1), diuretics (after stopping), post-hypercapnia | NS (volume + chloride repletion). Correct the deficit. |
| > 20 mEq/L | Chloride-resistant | Hyperaldosteronism (Conn syndrome), Cushing's, Bartter/Gitelman, active diuretic use, severe hypokalemia | Treat underlying cause. K⁺ repletion. Spironolactone if hyperaldosteronism. |
Why NS and K⁺ correct chloride-responsive metabolic alkalosis:
- NS (volume + Cl⁻): Volume depletion (from vomiting, NG suction, diuretics) makes the proximal tubule avidly reabsorb Na⁺. With Cl⁻ also depleted, Na⁺ has to be reabsorbed paired with HCO₃⁻ instead of Cl⁻, which keeps serum HCO₃⁻ high. Volume contraction also activates aldosterone, which drives distal Na⁺/H⁺ exchange and generates more HCO₃⁻. Giving NS restores volume (turns off RAAS) and supplies Cl⁻ so the proximal tubule pairs Na⁺ with Cl⁻ again, letting the kidney finally dump excess HCO₃⁻ in the urine.
- K⁺ repletion: Hypokalemia maintains the alkalosis through three mechanisms. (1) Transcellular shift: K⁺ leaves cells in exchange for H⁺ moving in, producing intracellular acidosis but extracellular alkalosis. (2) Distal tubule: when K⁺ is low, the principal cell swaps Na⁺ for H⁺ instead of K⁺, dumping more H⁺ in the urine. (3) Proximal tubule: hypokalemia stimulates ammoniagenesis (NH₄⁺ excretion), and each NH₄⁺ excreted generates a new HCO₃⁻. Replacing K⁺ shuts all three down. You cannot fully correct metabolic alkalosis without correcting K⁺.
Renal Tubular Acidosis (RTA)
| Type | Defect | pH | K⁺ | Classic Association |
|---|---|---|---|---|
| Type 1 (Distal) | Can't secrete H⁺ in distal tubule | Urine pH > 5.5 (can't acidify) | ↓ (hypoK) | Sjögren, SLE, nephrocalcinosis, amphotericin B |
| Type 2 (Proximal) | Can't reabsorb HCO₃ in proximal tubule | Urine pH < 5.5 (once threshold exceeded) | ↓ (hypoK) | Fanconi syndrome, multiple myeloma, carbonic anhydrase inhibitors (acetazolamide) |
| Type 4 (Hypoaldo) | ↓ Aldosterone or tubular resistance | Urine pH < 5.5 | ↑ (hyperK) | Most common RTA. Diabetic nephropathy, ACEi/ARBs, spironolactone, TMP-SMX, heparin |
Pearl: Always calculate the corrected AG (for albumin) and delta-delta ratio to identify mixed disorders.
Lactic Acidosis: Types A and B
Cohen-Woods classification. Two types only: A (hypoperfusion) and B (everything else). B is subdivided into B1, B2, B3. Every named cause (MALA, propofol infusion, D-lactic acidosis) sits inside one of these buckets.
| Type | Mechanism | Causes |
|---|---|---|
| Type A Most common | Tissue hypoperfusion or hypoxia. Anaerobic glycolysis surges. | Shock (septic, cardiogenic, hypovolemic, obstructive), regional ischemia (mesenteric, limb, compartment syndrome), severe hypoxemia, severe anemia, CO poisoning, post-arrest, seizures, heavy exertion. |
| Type B1 Underlying disease | Disease impairs lactate clearance or shifts metabolism without overt hypoperfusion. | Liver failure (impaired clearance), malignancy (Warburg effect; classic in leukemia/lymphoma), sepsis (overlaps with A), thiamine deficiency, diabetes, pheochromocytoma, short bowel / SIBO → D-lactic acidosis. |
| Type B2 Drugs / toxins | Mitochondrial toxicity, β2 stimulation, or impaired clearance. | Metformin (MALA, especially with AKI), linezolid (> 2 wks), propofol infusion syndrome, NRTIs (didanosine, stavudine, zidovudine), β2 agonists (albuterol, epinephrine, terbutaline), salicylates, cyanide, methanol, ethylene glycol, acetaminophen (late), cocaine, alcohol. |
| Type B3 Inborn errors | Genetic mitochondrial or enzyme defects. | Mitochondrial disorders (MELAS), pyruvate dehydrogenase deficiency, G6PD-related defects. |
D-lactic acidosis (bacterial overgrowth): Short bowel syndrome and SIBO let gut bacteria (Lactobacillus, Streptococcus bovis, Bifidobacterium) ferment unabsorbed carbohydrate into D-lactate, the stereoisomer humans cannot metabolize efficiently (we lack effective D-lactate dehydrogenase). Triggered by carb-rich meals. Presents as episodic encephalopathy (slurred speech, ataxia, confusion; looks intoxicated) with a wide anion gap. The trap: the standard lactate assay measures L-lactate only, so it returns normal despite the gap. You must specifically order a D-lactate level. Treatment: low-carb diet, oral non-absorbable antibiotics (metronidazole, neomycin, vancomycin, rifaximin) to suppress overgrowth, IV fluids, bicarb if severe. Sits inside Type B1 in Cohen-Woods (the "D" refers to the molecular isomer, not a separate Cohen-Woods type).
Bedside framing: Lactate > 4 with hypotension → Type A, resuscitate first. Lactate elevated but patient looks well-perfused → Type B (drugs, liver, malignancy, thiamine). Persistent lactate despite resuscitation → recheck for missed ischemia (mesenteric, limb), failed source control, then consider metformin, linezolid, thiamine deficiency, malignancy.
Toxic Alcohols (suspect when AG ↑ + osmolar gap > 10)
Osmolar gap = measured osm − (2×Na + glu/18 + BUN/2.8). Normal < 10. The trap: the gap shrinks as the parent alcohol is metabolized to its toxic acid, so a late presentation can show a wide AG with a deceptively normal osmolar gap. Order both early and trend.
| Toxin | Source | Toxic metabolite | Bedside clue | Treatment |
|---|---|---|---|---|
| Ethylene glycol | Antifreeze, brake fluid | Glycolic acid (AG ↑) then oxalic acid (calcium oxalate → AKI) | Calcium oxalate crystalluria, hypocalcemia (Ca binds oxalate), AKI, Wood's lamp fluorescence (some products contain fluorescein) | Fomepizole 15 mg/kg IV load → 10 mg/kg q12h. Hemodialysis if pH < 7.25, AKI, end-organ damage, or level > 50 mg/dL. |
| Methanol | Windshield washer, moonshine, industrial solvents | Formic acid (AG ↑, inhibits cytochrome oxidase) | Visual changes (snowstorm vision, optic disc edema, blindness), abdominal pain, putaminal hemorrhage on imaging | Fomepizole + emergent dialysis (lower threshold than EG: pH < 7.30 or any visual changes). Folinic acid 50 mg IV q4h enhances formate clearance. |
| Isopropanol | Rubbing alcohol, hand sanitizer | Acetone (NOT an acid) | Osmolar gap WITHOUT AGMA, ketones positive without acidosis, fruity breath, CNS depression. Mimics DKA but glucose normal. | Supportive only. Fomepizole NOT indicated (no toxic acid metabolite). Dialysis only if hemodynamic instability or coma. |
| Propylene glycol | IV solvent in lorazepam, phenobarbital, diazepam infusions | Lactic acid (Type B lactic acidosis) | ICU patient on prolonged benzo drip with rising lactate without hypoperfusion. AKI possible. | Stop the offending infusion. Fomepizole if severe. Dialysis for refractory cases. |
When to give fomepizole empirically: Suspected toxic alcohol ingestion + any of (osmolar gap > 10, pH < 7.30, AG > 16, visual changes, AKI). Don't wait for the level. Fomepizole blocks alcohol dehydrogenase, halting metabolism to the toxic acid. Loading dose 15 mg/kg IV over 30 min. Ethanol drip is the fallback when fomepizole is unavailable (target ethanol level 100–150 mg/dL).
Respiratory Acidosis (PaCO2 > 40)
Hypoventilation fails to clear CO2. Acute (HCO3 rises 1 per 10 mmHg ↑ PaCO2) vs chronic (HCO3 rises 3.5–4 per 10) distinction matters: chronic is renally compensated, acute is dangerous and may need ventilation.
| Mechanism | Causes | Bedside flag |
|---|---|---|
| CNS depression | Opioids, benzodiazepines, alcohol, barbiturates, brainstem stroke, encephalitis, ICH, post-ictal | Pinpoint pupils + slow RR → naloxone trial. AMS + bradypnea. |
| Lung / airway | Severe COPD/asthma exacerbation (eventual fatigue), end-stage ILD, severe pneumonia, ARDS, pulmonary edema, upper airway obstruction | "Normalizing" PaCO2 in a previously tachypneic asthmatic = pre-arrest. Intubate before the next ABG. |
| Neuromuscular | GBS, myasthenic crisis, ALS, polymyositis, severe hypoK/hypoPhos/hypoMg, organophosphates, botulism, high cervical cord injury | NIF less negative than −20 cmH2O, FVC < 15 mL/kg, paradoxical breathing → intubate before ABG declares failure. |
| Chest wall / restriction | Severe kyphoscoliosis, morbid obesity (OHS), large pleural effusion, flail chest, abdominal compartment syndrome, tense ascites | Restrictive PFT pattern, low ERV. OHS often missed on the ward. |
| Iatrogenic | Permissive hypercapnia (asthma, ARDS protective ventilation), under-set vent rate or tidal volume, sedation oversedation, equipment failure | Check vent settings before treating the patient. Always. |
Acute vs chronic at the bedside: Pure acute respiratory acidosis means HCO3 ≈ 24 + (PaCO2 − 40)/10. If HCO3 is much higher than predicted, the kidneys have had days to compensate (chronic). A COPD patient with pH 7.36, PaCO2 65, HCO3 36 is at chronic baseline; the same numbers in a previously healthy patient mean impending failure.
Respiratory Alkalosis (PaCO2 < 40)
Hyperventilation. Almost always centrally driven. Most common acid-base disorder in hospitalized patients, and frequently part of a mixed picture (sepsis, salicylates, hepatic encephalopathy).
| Trigger | Mechanism / Notes |
|---|---|
| Hypoxemia | Carotid body drives ventilation. PE, pneumonia, pulmonary edema, high altitude, severe anemia. Always check SpO2 and consider CTPA in unexplained tachypnea. |
| Sepsis | Cytokine-mediated central drive. Often the earliest acid-base abnormality, before lactate or hypotension appear. |
| Pain / anxiety | Voluntary or involuntary hyperventilation. Pure psychogenic hyperventilation is a diagnosis of exclusion (rule out PE, sepsis, ASA first). |
| Salicylate toxicity | Direct medullary stimulation + uncoupling of oxidative phosphorylation. Classic mixed AGMA + respiratory alkalosis. Tinnitus is the giveaway. |
| Hepatic encephalopathy / cirrhosis | Ammonia and progesterone-like metabolites stimulate the medullary center. Almost universal in advanced cirrhosis. |
| Pregnancy | Progesterone is a respiratory stimulant. Normal pregnant PaCO2 is 28–32 with compensatory low HCO3 (~18–20). |
| CNS lesions | Brainstem stroke, tumor, infection, head injury → central neurogenic hyperventilation. |
| Iatrogenic | Over-ventilation on the vent (rate or tidal volume too high). Re-check settings before treating. |
Compensation rule: Acute → HCO3 drops 2 per 10 mmHg ↓ PaCO2. Chronic → HCO3 drops 4–5 per 10. So a chronic respiratory alkalosis can present with a markedly low HCO3 (15–18) that mimics metabolic acidosis. Always check the pH first; alkalemia rules out a primary metabolic acidosis.
🧪 Workup
Workup
- ABG/VBG -VBG acceptable for pH/HCO₃
- BMP -AG calculation
- Albumin -corrected AG = AG + 2.5×(4−albumin)
- Lactate -#1 AG acidosis in hospital
- Ketones/BHB
- Osmolality (serum + calculated) -osmol gap >10 = toxic alcohol
- Urine AG -positive=RTA, negative=GI loss
- APAP + salicylate -always in unexplained AG acidosis
🚨 Management
Management
- AG acidosis: DKA→insulin. Lactic→treat shock. Toxic alcohols→fomepizole+dialysis. Uremia→dialysis. Stewart Approach, Kellum 2009
- Non-AG acidosis: Diarrhea→IVF. RTA→bicarb. NS-induced→switch to LR.
- Resp acidosis: COPD→BiPAP. Opioid→naloxone. NMD→intubation.
- Met alkalosis: Cl-responsive→NS+KCl. Cl-resistant→treat cause.
💊 Medications
Medications
| Drug | Dose | Route | Notes |
|---|---|---|---|
| NaHCO₃ | 50-150 mEq | IV | Only pH < 6.9. Monitor K⁺. BICAR-ICU, Jaber 2018 |
| Fomepizole | 15 mg/kg load | IV | Toxic alcohols |
| Acetazolamide | 250 mg q6-12h | IV/PO | Refractory met alkalosis |
| KCl | 20-40 mEq q2-4h | IV | Must correct K⁺ to fix met alkalosis |
📋 On Rounds
pH 7.28, PCO₂ 24, HCO₃ 11, Na 140, Cl 100, AG 29. Interpret this ABG.
Step 1: pH 7.28 = acidemia. Step 2: HCO₃ 11 (low) = metabolic acidosis (primary). Step 3: Winter's formula: expected PCO₂ = 1.5(11) + 8 = 24.5 ± 2. Actual PCO₂ = 24 → appropriate respiratory compensation. Step 4: AG = 140 − (100 + 11) = 29 (elevated) → AG metabolic acidosis. Think MUDPILES: DKA? Lactic acidosis? Toxic ingestion? Step 5: Delta-delta = (29−12)/(24−11) = 17/13 = ~1.3 → pure AG metabolic acidosis.
Why do you correct the anion gap for albumin?
Albumin is a negatively charged protein that contributes ~12 mEq/L to the unmeasured anions (the normal AG). In hypoalbuminemic patients (cirrhosis, nephrotic syndrome, malnutrition, critical illness), the baseline AG is lower because there are fewer albumin anions. A "normal" AG of 12 in a patient with albumin of 2 actually represents a hidden AG elevation of ~5. Correction: for every 1 g/dL albumin below 4, add 2.5 to the calculated AG.
Walk me through the 6-step approach to ABG interpretation.
(1) pH: acidemia (< 7.35) or alkalemia (> 7.45)? (2) Primary disorder: the pH and pCO₂ move opposite in a respiratory disorder, the pH and HCO₃ move together in a metabolic one. (3) Compensation: Winter's (expected pCO₂ = 1.5 × HCO₃ + 8 ± 2); higher than expected → concurrent respiratory acidosis, lower → respiratory alkalosis. (4) Anion gap on every gas, corrected for albumin. (5) Delta ratio if the gap is raised. (6) Correlate with the patient -a gas that does not fit the story means an extra disorder.
How do you calculate the corrected anion gap for albumin?
The anion gap is falsely low in hypoalbuminemia (albumin is the main unmeasured anion). Corrected AG = Calculated AG + 2.5 × (4.0 − measured albumin). For each 1 g/dL drop in albumin below 4, add 2.5 to the AG. Example: Na 140, Cl 108, HCO₃ 20, albumin 2.0 → raw AG = 140-108-20 = 12 (looks normal). Corrected AG = 12 + 2.5 × (4.0-2.0) = 17 (elevated!). Without correction, you'd miss a significant anion gap acidosis.
How many types of lactic acidosis are there?
Two (Cohen-Woods): Type A = tissue hypoperfusion or hypoxia (shock, ischemia, severe hypoxemia, anemia, CO poisoning) and Type B = everything else. Type B has 3 subtypes: B1 underlying disease (liver failure, malignancy, sepsis, thiamine deficiency, short bowel/SIBO), B2 drugs/toxins (metformin, linezolid, propofol, NRTIs, β2 agonists), B3 inborn errors (mitochondrial, PDH deficiency).
Confused, tachypneic patient on chronic aspirin. Na 140, Cl 118, HCO₃ 20, so the anion gap is 2. Does that rule out salicylate toxicity?
No. Many chloride ion-selective electrodes read salicylate as chloride (pseudohyperchloremia), so Cl is falsely high and the gap falsely small or negative, hiding the high-gap acidosis. Send a salicylate level and ask the lab to rerun chloride by another method. Other causes of a low or negative gap: lab error, hypoalbuminemia, bromide or iodide, IgG paraprotein, severe lithium toxicity.
Short bowel patient with episodic confusion and slurred speech after meals, wide anion gap, but lactate is normal. Diagnosis?
D-lactic acidosis from bacterial overgrowth. Gut bacteria (Lactobacillus, Strep bovis, Bifidobacterium) ferment unabsorbed carbs into D-lactate, which humans cannot clear (we only metabolize L-lactate efficiently). The standard lactate assay measures L-lactate only, so it reads normal despite the gap. Order D-lactate level to confirm. Treat: low-carb diet, oral non-absorbable antibiotics (metronidazole, neomycin, rifaximin). Cohen-Woods: Type B1.
Lactate is 8 but the patient is well-perfused, BP normal, no shock. What's on your differential?
Type B lactic acidosis. Think drugs (metformin/MALA, linezolid > 2 wks, propofol infusion syndrome, NRTIs, β2 agonists like albuterol or epinephrine), liver failure (impaired clearance), malignancy (Warburg effect, classic in leukemia/lymphoma), thiamine deficiency (especially alcoholics, refeeding), D-lactic acidosis in short bowel. Sepsis can also cause Type B physiology even before overt hypotension.
Clinical Examples
📋 Case 1, Triple Acid-Base Disorder in Sepsis
Patient: 58M with septic shock from pneumonia. ABG: pH 7.22, pCO₂ 30, HCO₃ 12. Na 142, Cl 98, albumin 2.0. Lactate 8.4. Also receiving NS resuscitation for 24h.
Key findings: Step 1: acidemia. Step 2: low HCO₃ = metabolic acidosis. Step 3: Winter's: expected pCO₂ = 1.5(12)+8 = 26 ± 2. Actual 30 → pCO₂ HIGHER than expected → concurrent respiratory acidosis (tiring out?). Step 4: corrected AG = (142-98-12) + 2.5(4-2) = 32+5 = 37 → AG metabolic acidosis. Step 5: delta-delta = (37-12)/(24-12) = 25/12 = 2.1 → hidden metabolic alkalosis (from vomiting? contraction?).
Management:
- Triple disorder: AG metabolic acidosis (lactic from sepsis) + respiratory acidosis (respiratory failure) + metabolic alkalosis (contraction from volume depletion)
- Address the lactate: aggressive IVF (switch to LR, NS worsens non-AG acidosis), source control, antibiotics
- The rising pCO₂ suggests respiratory fatigue → may need intubation soon (normally sepsis causes hyperventilation/low pCO₂)
- Correct the albumin, always use corrected AG in ICU patients (hypoalbuminemia is nearly universal)
- Recheck ABG in 2h, if pCO₂ continues to rise, intubate before arrest
Teaching point: A "normal" pCO₂ in a septic patient is alarming, they should be hyperventilating. A pCO₂ that is higher than Winter's predicted suggests respiratory muscle fatigue and impending respiratory arrest. This is a pre-intubation sign.
📋 Case 2, Non-Anion Gap Metabolic Acidosis (NAGMA)
Patient: 45F with chronic diarrhea from Crohn's disease. pH 7.30, pCO₂ 28, HCO₃ 14. Na 138, Cl 112, AG 12 (normal). Urine: Na 30, K 20, Cl 60.
Key findings: Normal AG + metabolic acidosis = NAGMA (non-anion gap metabolic acidosis). Hyperchloremic (Cl elevated). Use urine anion gap (UAG) to distinguish GI vs renal cause: UAG = Na+K-Cl = 30+20-60 = -10 (NEGATIVE).
Management:
- Negative UAG = kidneys ARE appropriately excreting acid (NH4⁺) → GI bicarbonate loss (diarrhea is the cause)
- If UAG were positive: kidneys CANNOT excrete acid → renal tubular acidosis (RTA)
- Treat the diarrhea: optimize Crohn's therapy (5-ASA, immunomodulators, biologics per GI)
- Oral sodium bicarbonate 650 mg TID for symptomatic acidosis (bridge while treating diarrhea)
- Replace potassium (diarrheal losses), hypokalemia is common with GI bicarbonate loss
Teaching point: The urine anion gap is the key to NAGMA workup. Negative UAG = GI loss (diarrhea, kidneys working fine). Positive UAG = renal problem (RTA, kidneys can't excrete acid). Remember: "Negative = Normal kidneys, Positive = Problem in kidney."
📋 Case 3, Mixed Metabolic Alkalosis + Respiratory Alkalosis
Patient: 72F with COPD on chronic steroids, admitted for COPD exacerbation with NG tube on suction × 3 days. ABG: pH 7.58, pCO₂ 32, HCO₃ 38. K⁺ 2.8, Cl 88.
Key findings: Severely alkalemic (pH 7.58, dangerous). Two alkalosis disorders: metabolic alkalosis (HCO₃ 38, from NG suction losing HCl + volume contraction) AND respiratory alkalosis (pCO₂ 32, should be 48-52 for this HCO₃ level as compensation, but is much lower → concurrent respiratory alkalosis from anxiety/pain).
Management:
- Replace volume: NS is chloride-rich (restores renal ability to excrete bicarb, "chloride-responsive" metabolic alkalosis)
- Replace potassium aggressively: K⁺ 2.8 is dangerous + hypokalemia maintains metabolic alkalosis (kidneys retain H⁺ to excrete K⁺)
- Review NG suction necessity, can it be discontinued or placed to gravity?
- Address respiratory alkalosis: treat pain/anxiety (may be driving hyperventilation), optimize bronchodilators
- Danger: pH > 7.55 causes seizures, arrhythmias, and shifts the oxygen dissociation curve left (impaired O₂ delivery)
Teaching point: Metabolic alkalosis is maintained by three things: volume depletion (kidneys reabsorb Na with HCO₃), hypokalemia (kidneys excrete H⁺ instead of K⁺), and chloride depletion. Fixing all three (NS + KCl) corrects "chloride-responsive" alkalosis. Urine Cl < 20 = chloride-responsive.
📣 Sample Presentation
One-Liner
"Mr. Young is a 32-year-old with type 1 DM presenting with ABG: pH 7.18, pCO₂ 22, HCO₃ 8. Na 140, Cl 102, AG 30. Consistent with anion gap metabolic acidosis with appropriate respiratory compensation."
Key Points to Cover on Rounds
6-step ABG: (1) Acidemia (pH 7.18), (2) Primary metabolic acidosis (low HCO₃ 8), (3) Respiratory compensation: expected pCO₂ by Winter's = 1.5(8)+8±2 = 18-22 → actual 22 (appropriate), (4) AG 30 (elevated, corrected for albumin). (5) Delta ratio: ΔAG=18, ΔHCO₃=16 → 1.1 (pure AG acidosis). (6) Correlates with the history: DKA (glucose 420, ketones positive), no vomiting and no diuretics, so no hidden alkalosis to explain away. Plan: treat DKA -insulin drip, fluids, electrolyte monitoring.
Monitoring
- ABG q2-4h during correction
- BMP q4-6h -K⁺ shifts with pH
- AG trend
- Lactate q2-4h if lactic acidosis
- UOP
⚡ Summary
Summary
Work Every Gas the Same Way
Six steps in the same order every time: pH, primary disorder, compensation, anion gap, delta ratio, correlate with the patient. The discipline is the point: skipping straight to a pattern you recognize is how a second, hidden disorder gets missed in a patient who has three.
Steps 1 and 2
pH below 7.35 is acidemia, above 7.45 is alkalemia, and it names the primary disorder because compensation never fully corrects the pH. Then match the direction: a low pH with a low bicarbonate is metabolic acidosis, a low pH with a high pCO₂ is respiratory acidosis, and so on. A normal pH with abnormal numbers means two opposing disorders, not a normal patient.
Check Whether Compensation Fits
Metabolic acidosis: Winter's formula, expected pCO₂ = 1.5 × HCO₂ + 8 ± 2. Acute respiratory: bicarbonate moves 1 per 10 mmHg of CO₂; chronic: 3.5–4 per 10, because renal compensation has had 3 to 5 days to run. Compensation outside the predicted range is a second primary disorder, and this is the step that finds the patient who is tiring toward respiratory failure while their metabolic acidosis is being treated.
Always Calculate the Anion Gap
Na minus Cl minus HCO₂, normal about 12. Correct it for albumin: add 2.5 for every 1 g/dL the albumin is below 4, because albumin is the main unmeasured anion. An uncorrected gap in a hypoalbuminemic patient hides a real gap acidosis, which is the commonest reason lactic acidosis and ketosis are missed in the sick and the malnourished.
Know the Two Cause Lists
MUDPILES for a raised gap: Methanol, Uremia, DKA, Propylene glycol, Isoniazid or Iron, Lactate, Ethylene glycol, Salicylates. HARDUPS for a normal gap: Hyperalimentation, Acetazolamide, RTA, Diarrhea, Ureteroenteric fistula, Pancreatic fistula, Saline. Diarrhea and RTA are the two that matter at the bedside, separated by the urine anion gap.
Delta-Delta Finds the Third Disorder
Compare the rise in anion gap with the fall in bicarbonate. A ratio of 1–2 is a pure gap acidosis. 0.4–1 means a coexisting non-gap acidosis, and below 0.4 it is essentially all non-gap. Above 2 means a coexisting metabolic alkalosis, classically the DKA patient who has also been vomiting, or a chronic CO₂ retainer whose baseline bicarbonate was already high. The ratio also decides whether alkali helps: bicarbonate is therapeutic in the non-gap bands, because chloride is not metabolizable and the lost bicarbonate never comes back on its own, but usually unnecessary in a pure gap acidosis, because lactate and ketones are metabolized back to bicarbonate once the cause is treated. Skipping this step is how a mixed picture gets treated as a simple one.
Check the Osmolar Gap When the Gap Is Unexplained
A raised anion gap with an osmolar gap above 10 means a toxic alcohol: methanol, ethylene glycol or propylene glycol. Treat with fomepizole and dialysis, and do not wait for confirmatory levels, which are send-outs. Calcium oxalate crystals and acute kidney injury point to ethylene glycol; visual loss points to methanol.
Separate Type A From Type B Lactate
Type A is hypoperfusion, shock, ischemia or hypoxia, and it is the one that means the patient is dying. Type B has no perfusion deficit: metformin, alcohol, thiamine deficiency, malignancy, linezolid, propofol, epinephrine infusion. Treating a type B lactate with more fluid does nothing, so the mechanism has to be identified before the number is chased. Also remember the ABG is only one input: correlate it with the patient in front of you.
📄 One Pager
Text version
Nephrology / Critical Care · One Pager
ABG Interpretation
6 steps: pH → primary disorder → compensation → AG (corrected for albumin) → delta ratio → correlate with the patient. MUDPILES for AG acidosis. Winter's formula.
6-Step Approach
(1) pH (acid vs alkalemia). (2) Primary disorder (pCO₂ vs HCO₃). (3) Compensation appropriate? (Winter's formula for met acidosis, and pick acute vs chronic deliberately). (4) AG on every gas, corrected for albumin. (5) Delta ratio (ΔAG/ΔHCO₃) if the gap is raised. (6) Correlate with the patient -a gas that does not fit the story means an extra disorder.
AG Acidosis -MUDPILES
M ethanol · U remia · D KA · P ropylene glycol · I NH/Iron · L actic acidosis · E thylene glycol · S alicylates Check lactate, ketones, osmol gap, tox screen.
Key Formulas
Winter's: expected pCO₂ = 1.5 × HCO₃ + 8 ± 2. AG = Na − Cl − HCO₃. Corrected AG = AG + 2.5 × (4 − albumin). Delta-delta: > 2 = hidden met alk, < 1 = hidden non-AG met acidosis.
Key Drugs
Bicarb drip150 mEq in D5W (if pH < 6.9 in DKA)
Fomepizole15 mg/kg load (toxic alcohols)
Insulin drip0.1 u/kg/hr (DKA)
DialysisRefractory acidosis or toxic ingestion
Pitfalls
- Not correcting AG for albumin (misses hidden AG acidosis in ICU patients)
- Treating number without finding cause
- Bicarb for DKA (except pH < 6.9)
- Missing mixed acid-base disorder
Related Topics
Calciphylaxis (Calcific Uremic Arteriolopathy)Chronic Kidney Disease (CKD)Contrast-Induced NephropathyDialysis IndicationsHypernatremiaHyponatremia