Bedside Snapshot
- Normal (Adult)
- Lab-specific; often 5–12 mEq/L (same number in mmol/L). Potassium usually omitted. Some labs list an upper limit of 15–16.
- Critical Values
- No universal critical value: interpret by trend and clinical context. Screening trigger ≥15 mEq/L (proposed, not a guideline).
- High Suggests
- Lactate: shock, sepsis, bowel ischemia, metformin
- Ketoacids: DKA, alcoholic or starvation ketosis, SGLT2 inhibitor DKA
- Kidney failure (later stage)
- Toxic alcohols, salicylate, 5-oxoproline (chronic acetaminophen)
- Low Suggests
- Lab error or hypoalbuminemia (most common)
- IgG paraprotein (myeloma)
- Lithium, bromide or iodide toxicity
- Act Now If
- Gap ≥15 mEq/L with low bicarbonate, low pH, or rising lactate or ketones: work up as possible high-gap acidosis; get pH, lactate, ketones, glucose and creatinine; notify receiving team or medical control.
- Watch Out For
- A near-normal bicarbonate can hide a large gap: vomiting alkalosis can mask lactic acidosis. Correct the gap for albumin when albumin is low.
Reference & Critical Ranges
Lab-specific ranges: Every lab sets its own interval, and it depends on the analyzer and the population. Use the range printed on your report. The values below are typical only.
Reference ranges
| Population | Conventional | SI | Notes |
|---|---|---|---|
| Adult | Lab-specific; often 5–12 mEq/L | Same number (mmol/L) | Potassium is usually left out of the calculation. Older methods gave 8–16 or 9–17 mEq/L, and many labs still list an upper limit of 15–16. |
| Pregnancy | No pregnancy-specific interval found | — | Use the laboratory's adult range and read it with albumin and the clinical picture. |
| Children | Age-specific range from your lab | — | No verified pediatric interval for this page. The calculation is the same as in adults. |
Severity & critical values
| Level | Conventional | SI | What it means |
|---|---|---|---|
| Low | Below your lab's lower limit | Same | Unmeasured cation (IgG paraprotein, lithium, bromide or iodide), hypoalbuminemia, or laboratory error. |
| Elevated | Above your lab's upper limit | Same | Usually a high-gap acidosis when bicarbonate is also low. Metabolic alkalosis, hyperphosphatemia, paraprotein or laboratory error can also raise it. |
| Screen positive | ≥15 mEq/L | ≥15 mmol/L | Cutoff proposed for screening high-gap acidosis in a single-centre 2026 study of inpatients tested for organic acids (sensitivity 98%, specificity 34%). Not a guideline threshold; expect a raised gap in advanced CKD. |
| DKA pattern | >12 mEq/L | >12 mmol/L | The 2024 ADA-led consensus says a gap above 12 mmol/L fits DKA. It is not one of the diagnostic criteria. |
| Critical | None | None | No universal panic value. Act on the trend, pH, lactate or ketones, creatinine and perfusion. |
Critical values: No universal critical value exists for the anion gap. Act on its parts and on the patient: pH, bicarbonate, lactate, ketones, creatinine, potassium and perfusion. A gap of 15 mEq/L or more is a trigger to screen. It is not a diagnosis.
Conversion: You don't need one. Na⁺, Cl⁻ and HCO₃⁻ each carry a single charge, so mEq/L and mmol/L read the same.
Anion gap
AG = Na⁺ − (Cl⁻ + HCO₃⁻)
Normal is your lab's range, shown in the table above. Potassium is usually left out.
Albumin-corrected anion gap
Corrected AG = measured AG + 2.5 × (4.0 − albumin in g/dL)
Normal is the same as the measured gap. With albumin in g/L, the factor becomes 0.25 × (40 − albumin). Correct the gap every time albumin is low.
Causes of Abnormal Results
High (gap above your lab's upper limit)
A raised gap means an unmeasured anion is present. Your job is to find out what is adding it.
| Mechanism | Common causes | Bedside clue |
|---|---|---|
| Lactate | Tissue hypoperfusion (shock, sepsis, bowel or limb ischemia); metformin; linezolid; nucleoside reverse transcriptase inhibitors; liver failure with reduced clearance | Hypotension, abdominal pain, poor perfusion; lactate above your lab's limit |
| Ketoacids | Diabetic ketoacidosis; alcoholic ketoacidosis; starvation ketosis; euglycemic DKA on an SGLT2 inhibitor | Vomiting, abdominal pain, Kussmaul breathing, confusion; glucose may be below 200 mg/dL |
| Kidney failure | Advanced renal insufficiency (retained sulfate and phosphate) | Rising creatinine and BUN, low urine output; early CKD usually gives a normal gap |
| Toxic alcohols | Methanol, ethylene glycol; less often diethylene glycol, or propylene glycol from prolonged high-dose lorazepam infusion | Visual symptoms, altered mental status, raised osmol gap; with a lorazepam infusion, think propylene glycol (see the toxic alcohols page) |
| Salicylate | Aspirin toxicity | Unexplained gap; check a salicylate level |
| D-lactate | Short bowel syndrome with carbohydrate malabsorption | Bowel resection history; request D-lactate when standard lactate does not explain the gap |
| 5-oxoproline | Chronic acetaminophen use, most often in older adults | Gap with normal lactate and ketones; ask about chronic acetaminophen |
Do not stop at lactate: In one 2026 cohort, an estimated quarter of patients with a raised gap were left unexplained by raised lactate, ketones or salicylate. Keep renal failure, toxins and 5-oxoproline on the list.
Low (gap below your lab's lower limit)
A low gap is rare, and it still matters. Think unmeasured cation, low albumin or lab error.
| Mechanism | Common causes | Bedside clue |
|---|---|---|
| Laboratory error | Analyzer or calibration problem, or a wrong sample | Result conflicts with the patient; repeat on a fresh sample |
| Hypoalbuminemia | Critical illness and other low-albumin states | Low albumin on the same panel; correct the gap |
| Unmeasured cations | IgG paraprotein (myeloma); lithium; polymyxin B; hypercalcemia and hypermagnesemia (variable effect) | High calcium or a paraprotein on the work-up; lithium level if on lithium |
| Halide toxicity | Bromide or iodide intoxication, including some bromide-containing drugs | Chloride reads falsely high with a low or negative gap |
A low or negative gap is a clue, not noise: Repeat the sample first. Then look for myeloma, lithium or bromide toxicity before you dismiss the result.
Normal gap (hyperchloremic) acidosis
In this pattern, bicarbonate is lost directly or diluted by chloride. No unmeasured anion gets added. Chloride, urine studies and the fluid history show the route.
| Mechanism | Common causes | Bedside clue |
|---|---|---|
| GI bicarbonate loss | Diarrhea, pancreatic fistula, urinary diversion (ileal conduit) | Stool, drain or fistula losses; a urine anion gap helps separate GI from renal loss |
| Renal bicarbonate loss | Renal tubular acidosis; early renal insufficiency; carbonic anhydrase inhibitors; hypoaldosteronism | Hyperchloremia with a normal gap; review drugs and kidney history |
| Chloride load | Large volumes of chloride-rich fluid, such as 0.9% saline | Bicarbonate falls as chloride rises after liters of fluid |
| Ketoacid clearance | Recovery from ketoacidosis, when the ketoacid anion is excreted | Falling bicarbonate with a normal gap during DKA recovery |
Interpretation & Bedside Approach
Step-by-step approach
- Is it real? Start with the sample itself. The pH, PaCO₂ and bicarbonate should agree with the Henderson–Hasselbalch equation (pH = 6.1 + log [HCO₃⁻ ÷ (0.03 × PaCO₂)]). If the gap surprises you, repeat it on the same analyzer.
- How sick is the patient? Those four go first: perfusion, mental status, breathing and potassium.
- Acute, chronic or trend? Acute acidosis usually comes from organic acids, such as lactate or ketoacids. Chronic acidosis more often means bicarbonate is being wasted, or the kidneys are not acidifying urine properly. Trust the trend over any single value.
- Calculate the gap and correct it for albumin: Use the formulas in Reference & Critical Ranges. With low albumin, act on the corrected gap.
- Classify with your lab's range and the screening cutoff: Above your upper limit means a raised gap. A gap of 15 mEq/L or more is a high-sensitivity screening trigger from a selected inpatient cohort (specificity 34%). Work up the cause. Don't treat the number. Use the albumin-corrected gap, expect a raised gap in advanced CKD, and state which threshold you used.
- Check for a second disorder: Compare the rise in the gap with the fall in bicarbonate (table below). The 1:1 relationship varies between patients. Treat a mismatch as a prompt, not a diagnosis.
- Find the cause and escalate: Work through lactate, ketones, creatinine, toxic alcohols and salicylate. Call the receiving team or medical control with the numbers.
Pattern & what it suggests
| Pattern | Suggests | Next step |
|---|---|---|
| High gap, near-normal bicarbonate | High-gap acidosis masked by a concurrent metabolic alkalosis, such as vomiting | Check lactate and ketones; a near-normal bicarbonate does not mean a stable patient |
| Gap rises more than bicarbonate falls | Possible added metabolic alkalosis (vomiting, volume contraction) | Check chloride, potassium and the history of losses |
| Bicarbonate falls more than the gap rises | Possible added normal-gap acidosis (saline, diarrhea, DKA recovery) | Check chloride and the fluids given |
| Gap normal, bicarbonate low, chloride high | Normal-gap (hyperchloremic) acidosis | Find the route: GI loss, renal loss or chloride load |
| Gap below your range | Unmeasured cation, hypoalbuminemia, halide toxicity or laboratory error | Correct for albumin, repeat the sample, look for a paraprotein |
| High gap, normal lactate and ketones | Toxins (toxic alcohols, salicylate), kidney failure or 5-oxoproline | Osmol gap, salicylate level, creatinine, acetaminophen history |
DKA criteria (2024 consensus)
| Criterion | Threshold | Note |
|---|---|---|
| Glucose | ≥200 mg/dL (11.1 mmol/L) or known diabetes | About 10% of DKA presents with glucose below 200 mg/dL (euglycemic DKA). SGLT2 inhibitors account for most of these. |
| Ketones | β-hydroxybutyrate ≥3.0 mmol/L or ketonuria ≥2+ | Blood β-hydroxybutyrate is the more specific test. Urine strips detect acetoacetate. |
| Acid–base | pH <7.30 and/or bicarbonate <18 mmol/L | Mixed acid–base disorders occur in about one-third of DKA presentations. |
Mixed disorders are common: Two processes at once are common. Diuresis, vomiting or hyperventilation can each add a second. Read the gap next to the pH, PaCO₂ and bicarbonate.
Special Populations
| Population | What changes | Clinical implication |
|---|---|---|
| Neonates & infants | No verified neonatal interval for this page | Use the age band on your lab report. |
| Children & adolescents | Same calculation; age-specific reference intervals from the lab | Interpret with the pediatric range and the clinical picture, not the adult 12 mEq/L line. |
| Pregnancy | No pregnancy-specific interval found | Vomiting adds metabolic alkalosis that can mask a high gap. Correct for albumin and check lactate and ketones. |
| Older adults | Renal insufficiency and chronic acetaminophen use are more common causes of a raised gap | Ask about acetaminophen and check creatinine. 5-oxoproline is easy to miss. |
| CKD / dialysis | Acidosis is usually normal-gap early and high-gap later. A gap of 15 mEq/L is less specific as GFR falls: about 44%, 26% and 10% specific at GFR ≥60, 30–59 and <30 mL/min/1.73 m² | In kidney failure a raised gap is expected and does not identify a second cause. Look for uremia and the other causes in the table. |
| Liver disease | Low albumin lowers the measured gap; reduced lactate clearance raises lactate | Correct for albumin. Check lactate and ketones when acidosis develops. |
| Critically ill | Hypoalbuminemia is common: 49% of ICU patients had albumin below 20 g/L in one cohort | Correct every gap. A normal measured gap can hide a high-gap acidosis. |
| Other (SGLT2 inhibitors, metformin) | SGLT2 inhibitors cause euglycemic DKA; metformin is linked to lactic acidosis | Check ketones even when glucose is normal, and lactate in any metformin user with a raised gap. |
Interfering Factors & Pitfalls
Collection & handling
| Factor | Effect | Why | Avoid / fix |
|---|---|---|---|
| Sample left open to air | False + (gap reads high) | CO₂ escapes and measured bicarbonate falls | Cap the tube and run it promptly |
| Mixing analyzers or labs | Gaps not comparable | Reference intervals and calibration differ between analyzers | Trend on one analyzer; never compare gaps across devices |
Medications & toxins
| Factor | Effect | Why | Avoid / fix |
|---|---|---|---|
| Lithium | False − (true low gap) | Lithium is an unmeasured cation | Check a lithium level in a low gap on lithium |
| Bromide or iodide drugs | False − (pseudohyperchloremia) | Halides read as chloride | Ask about bromide-containing drugs when chloride is high with a low gap |
| Metformin | ↑ lactate and a true high gap | Impaired oxidative phosphorylation | Check lactate in any metformin user with a raised gap |
| SGLT2 inhibitors | True high gap with normal glucose | Ketoacidosis at glucose below 200 mg/dL | Measure blood β-hydroxybutyrate when DKA is possible |
Physiologic & biologic
| Factor | Effect | Why | Avoid / fix |
|---|---|---|---|
| Hypoalbuminemia | False − if uncorrected | Albumin is the main unmeasured anion | Correct the gap for albumin |
| Severe hyperphosphatemia | True ↑ (modest) | Phosphate is an unmeasured anion | Expect a small rise, mainly in advanced kidney failure; check phosphate and creatinine before attributing the gap to it |
| Severe hypertriglyceridemia (lipemic sample) | False + (enzymatic bicarbonate method) | Lipid interferes with photometric bicarbonate, which can read low and raise the gap | Check the lab's method; repeat on a blood gas or direct-ISE sample and do not act on a lipemic gap alone |
| Hyperalbuminemia (severe volume depletion) | False + | Albumin raises the unmeasured anion | Interpret with the clinical picture |
| IgA paraprotein | False + | Less common than IgG; unmeasured anion | Repeat and consider a paraprotein work-up |
| IgG paraprotein (myeloma) | True low gap | Unmeasured cation | Work up myeloma when the gap is low |
| Hyperglycemia | No change in the gap | Low measured sodium is translocational | Correct sodium separately (see the sodium page) |
Assay & device-specific
| Factor | Effect | Why | Avoid / fix |
|---|---|---|---|
| Bicarbonate method | Gap error follows bicarbonate error | Proficiency-testing allowances are wide: about ±4 mmol/L for sodium, ±5% for chloride and ±20% for total CO₂ (bicarbonate) | Trend on one device; treat a borderline gap as a prompt to repeat |
| Point-of-care cartridge (potassium-inclusive gap) | ↑ by about the potassium value; cutoffs do not transfer | Some point-of-care analyzers add potassium to the gap and report total CO₂ instead of bicarbonate | Read the device's printed formula and range; do not apply K-free 5–12 or 15 cutoffs |
Common misinterpretations
- A normal gap rules out acidosis: Wrong. Normal-gap acidosis and mixed disorders still need a pH and a bicarbonate result.
- Potassium is part of the formula: It is usually left out, as most reference intervals assume.
- A high gap always means DKA: Outside DKA, lactate, toxins and kidney failure cause most high gaps.
- The delta ratio settles mixed disorders: Its 1:1 relationship varies between patients, so don't lean on it alone.
Physiology & Test Basics
What it measures
Sodium is the main measured cation. Chloride and bicarbonate are the main measured anions. The anion gap is the difference: sodium minus chloride and bicarbonate. It estimates the unmeasured anions: mainly albumin, phosphate, sulfate and organic acids. Unmeasured cations pull it back down. This page is about the calculated gap. Sodium, chloride and bicarbonate each have their own page.
Physiology & regulation
- Charge balance: Plasma cations equal anions. So any unmeasured anion shows up as a gap.
- Normal gap: Albumin is the main unmeasured anion. That is why the normal gap is positive.
- Buffering: Bicarbonate soaks up acid. The Henderson–Hasselbalch equation ties pH, bicarbonate and PaCO₂ together.
- High gap: Lactate, ketoacids and other acid anions use up bicarbonate. They also add unmeasured anion. So the gap rises as bicarbonate falls.
- Normal gap: Bicarbonate is lost directly or diluted by chloride. Chloride rises to keep the charge balanced.
How it is measured
The lab works it out from sodium, chloride and bicarbonate on the same sample, usually serum or plasma from the chemistry panel. Potassium is usually left out, though some point-of-care analyzers include it. Some labs report bicarbonate as total CO₂. Some blood gas analyzers calculate a gap too. Each analyzer has its own result, so use your lab's reference interval and trend on one device. A point-of-care gap is only as good as its bicarbonate.
Why it matters
The gap is the fastest first sort for metabolic acidosis. High gap: lactate, ketones, toxins and kidney failure must be checked. Normal gap: the route of bicarbonate loss matters. Low albumin can hide a high-gap acidosis. A low gap can point to myeloma, lithium or bromide toxicity.
Clinical Pearls
Calculate it on every panel: A bicarbonate that reads only a little low can hide a big gap. One published illustrative case: bicarbonate 18 mEq/L, sodium 130 and chloride 66. That worked out to a gap of 46, from lactic acidosis plus vomiting alkalosis.
Correct it for albumin in sick patients: Every 1 g/dL that albumin drops takes about 2.5 mEq/L off the measured gap. A measured gap of 12 with albumin at 2 g/dL corrects to 17 mEq/L.
Transport: Write down the device and the time with every gap. Some point-of-care analyzers include potassium and report total CO₂. Their gap runs higher and has its own range. Don't apply K-free cutoffs to it. Bicarbonate varies more between analyzers than sodium or chloride. Compare trends on one device only, and repeat the test on arrival.
Monitoring after DKA treatment: Normal-gap (hyperchloremic) acidosis from saline and ketoacid clearance can look like DKA that won't go away. Read the gap, ketones and chloride together before you change therapy.
A low gap is a clue: Myeloma, lithium, bromide or iodide toxicity, and hypoalbuminemia, all push the gap down. Rule out lab error first. Then take the history.
Urine ketones miss early DKA: Urine strips pick up acetoacetate, not β-hydroxybutyrate. When DKA is suspected, measure blood β-hydroxybutyrate.
Look past lactate: A raised gap with normal lactate and ketones still needs a search. Check for toxic alcohols, salicylate, kidney failure or 5-oxoproline from chronic acetaminophen. See the toxic alcohols page.
Know which threshold you used: The 15 mEq/L screening cutoff comes from a single-centre 2026 study. It is not a guideline. Many labs still report 12 mEq/L, or their own upper limit. Say which threshold you used.
References
- Kraut JA, Madias NE. (2007). Serum anion gap: its uses and limitations in clinical medicine. Clinical Journal of the American Society of Nephrology, 2(1), 162–174. https://doi.org/10.2215/CJN.03020906
- Umpierrez GE, Davis GM, ElSayed NA, et al. (2024). Hyperglycemic crises in adults with diabetes: A consensus report (ADA, EASD, JBDS, AACE, DTS). Diabetes Care, 47(8), 1257–1275. https://doi.org/10.2337/dci24-0032
- Chionh CY, Tien CS, Yeon W, et al. (2026). Diagnostic threshold and performance of anion gap in screening for high anion gap metabolic acidosis. Singapore Medical Journal, 67(3), 162–169. https://doi.org/10.4103/singaporemedj.SMJ-2023-009
- Fong JMN, Tang GXJ, See KC. (2026). The anion gap: basic chemistry, powerful clinical implications. Singapore Medical Journal, 67(3), 132–134. https://doi.org/10.4103/SINGAPOREMEDJ.SMJ-2026-101
- Figge J, Jabor A, Kazda A, et al. (1998). Anion gap and hypoalbuminemia. Critical Care Medicine, 26(11), 1807–1810. https://doi.org/10.1097/00003246-199811000-00019
- Mehta AN, Emmett JB. (2008). GOLD MARK: an anion gap mnemonic for the 21st century. The Lancet, 372(9642), 892. https://doi.org/10.1016/s0140-6736(08)61398-7
- Stein H. (2019). Spuriously low serum bicarbonate levels in patients with hyperlipidemia: a report of 4 cases. American Journal of Kidney Diseases, 73(1), 131–133. https://doi.org/10.1053/j.ajkd.2018.04.016
- Zar T, Yusufzai I, Sullivan A, Graeber C. (2007). Acute kidney injury, hyperosmolality and metabolic acidosis associated with lorazepam. Nature Clinical Practice Nephrology, 3(9), 515–520. https://doi.org/10.1038/ncpneph0573
Medical Disclaimer
- For Educational Purposes Only: This content is intended for educational reference and should not be used for clinical decision-making.
- Not a Substitute for Professional Judgment: Always consult your local protocols, institutional guidelines, and supervising physicians.
- Verify Before Acting: Users are responsible for verifying information through authoritative sources before any clinical application.
AI Assistance Notice
AI was used to assist in organizing, paraphrasing, and formatting this information.