Lab and Bedside Calculators

Each row is one calculation. Take every input from a single specimen and a single draw time, then check each unit against the formula.

1. Use one specimen. Take Na⁺, Cl⁻, HCO₃⁻, glucose, BUN, albumin and blood gas values from the same draw before you calculate a gap or ratio.
2. Check the units. Glucose, BUN and creatinine go in mg/dL here. Albumin goes in g/dL. Convert any mmol/L result before you apply a formula.
3. Calculate the primary value first. Calculate the anion gap, then its albumin correction. Correct calcium or sodium before you call a level low or normal.
4. Compare with your laboratory range. Every cutoff in the table is a typical value. Your lab's reference interval and local protocol govern.
5. Recalculate after treatment. Repeat the inputs after fluids, insulin or ventilator changes so the value reflects the current patient.

Every formula gets one worked example. The values are made up. None of them is patient data.

Example: Plug in Na⁺ 138, Cl⁻ 102 and HCO₃⁻ 14, and the AG is 22 mEq/L. Albumin of 2 g/dL then pushes the corrected AG to 22 + 2.5 × 2 = 27 mEq/L.

Example: Measured osmolality 320 with Na⁺ 140, glucose 90 mg/dL and BUN 14 mg/dL. Calculated osmolality is 280 + 5 + 5 = 290, so the gap is 30 mOsm/kg. Ethanol pushes measured osmolality up. Some labs subtract an ethanol term too (ethanol in mg/dL ÷ 3.7). Use your lab's form.

Example: Ca 7.2 mg/dL with albumin 2 g/dL gives 7.2 + 0.8 × 2 = 8.8 mg/dL. The 0.8 factor is the common adjustment. It is not a universal constant, so use the factor your lab reports. In SI units the factor is 0.02 mmol/L per g/L of albumin shortfall. Corrected calcium is an estimate. In critical illness, with an acid-base disturbance, massive transfusion or very low albumin, measure ionized calcium instead of trusting the correction.

Example: Na⁺ 128 mEq/L with glucose 500 mg/dL gives 128 + 2.4 × 400 ÷ 100 = 137.6, about 138 mEq/L. Using the conventional 1.6 factor instead, the result is 128 + 1.6 × 400 ÷ 100 = 134.4, about 134 mEq/L. The study average comes from a 1999 study of six healthy subjects. Somatostatin drove glucose above 600 mg/dL, and the average fall was 2.4 mEq/L per 100 mg/dL. That same study found the fall is nonlinear. The 1.6 factor worked up to about 400 mg/dL. Above 400 mg/dL, about 4.0 per 100 mg/dL fit better. Use the factor your lab and protocol name. State it in the handoff.

Example: FiO₂ 0.40, PaCO₂ 40 mmHg, PaO₂ 150 mmHg gives PAO₂ = 0.40 × 713 − 50 = 235 mmHg. The A–a gradient is 85 mmHg. On room air, the upper limit of normal is about (age ÷ 4) + 4 mmHg, or about 19 mmHg at age 60. The rule was derived on room air. It gives no limit at a higher FiO₂, where the gradient normally widens. The formula assumes a respiratory quotient of 0.8 and sea-level barometric pressure (760 mmHg). Altitude changes PAO₂.

Example: PaO₂ 60 mmHg on FiO₂ 0.60 gives P/F 100. That is the severe range under the Berlin categories. P/F alone does not diagnose ARDS. Berlin also requires acute onset within one week. It requires bilateral opacities that effusion, collapse or nodules do not fully explain. It requires respiratory failure that cardiac failure or fluid overload does not fully explain. It requires PEEP or CPAP of at least 5 cmH₂O, so record the PEEP with the sample. PaO₂ runs lower at altitude. The Berlin definition corrects P/F by barometric pressure ÷ 760 above 1000 m.

Example: With HCO₃⁻ at 10 mEq/L, the expected PaCO₂ is 23 ± 2 mmHg, or 21 to 25. A measured 30 mmHg sits above that band, so an added respiratory acidosis is present.

Example: 70-year-old woman, 80 kg, SCr 1.5 mg/dL: (70 × 80) ÷ (72 × 1.5) × 0.85 = 44 mL/min. The drug label or pharmacy policy decides whether actual, ideal or adjusted weight goes in. Cockcroft-Gault is a steady-state estimate. It is unreliable in AKI, obesity and cachexia. Drug labels may still key to it. CKD staging, though, uses eGFR (see eGFR).

Example: Urine Na⁺ 20 mEq/L, plasma Na⁺ 140 mEq/L, plasma creatinine 2 mg/dL, urine creatinine 40 mg/dL: (20 × 2) ÷ (140 × 40) × 100 = 0.71%. The result lands in the prerenal range. Keep the Pitfalls caveats in mind: diuretics, contrast, rhabdomyolysis, sepsis, glomerulonephritis and CKD.

• Pediatrics: The same formulas apply. Use pediatric reference ranges for electrolytes and blood gases. For kidney function, use a pediatric method, not Cockcroft-Gault. In children, the bedside Schwartz estimate is eGFR ≈ 0.413 × height (cm) ÷ serum creatinine (mg/dL). This page sets no pediatric cutoff for the gaps or ratios.
• Pregnancy: The baseline shifts. Normal PaCO₂ is about 28–32 mmHg and bicarbonate about 18–21 mEq/L, from physiologic respiratory alkalosis. So a value that looks normal may be abnormal, and Winter’s formula must be read against pregnancy values. Use the lab's pregnancy reference intervals where they are provided. Serum creatinine runs lower in pregnancy. Cockcroft-Gault is not validated there.
• Older Adults: Cockcroft-Gault leans on age and weight. The drug label or pharmacy policy picks the weight. Very low or very high body weight shifts the result.
• Renal Impairment: When the kidney injury is unstable, creatinine-based results (CrCl and FENa) may not describe current function. Diuretics make FENa unreliable as well (see Pitfalls).
• Hepatic Impairment: No specific adjustment is established for the liver. Albumin feeds two corrections, and it is often low in liver disease. Check the albumin value before you correct anything. The albumin-corrected anion gap still applies.

• Error: Mixing mmol/L with mg/dL is the trap. Glucose and BUN go into the osmolality and sodium formulas in mg/dL. A glucose of 25 mmol/L is 450 mg/dL. Check the unit on the report first.
• Error: Calling a gap normal with no albumin in hand is the trap. Low albumin can hide a high anion gap. Calculate the albumin-corrected gap before you call it normal.
• Error: Reading sodium off the uncorrected value is the mistake. So is treating one glucose factor as exact. Corrected sodium runs higher than measured when glucose is high. Decide on sodium therapy from the corrected value. The conventional factor is 1.6. The 1999 study average is 2.4 (134 vs 138 in the example). The fall is nonlinear above 400 mg/dL. State which factor your lab and protocol use.
• Error: Trusting albumin-corrected calcium in critical illness. Corrected calcium can mislead in a critically ill patient. For massively transfused, septic or acid-base-disturbed patients, measure ionized calcium. Don't lean on the correction alone.
• Error: Trusting FENa with diuretics or in non-prerenal injury. Diuretics raise FENa, and a prerenal patient can then look like acute tubular necrosis. FENa below 1% also shows up in contrast nephropathy, rhabdomyolysis, early sepsis and glomerulonephritis. FENa is less reliable in CKD. FEUN is an alternative. A 2026 meta-analysis found moderate accuracy for intrinsic versus prerenal AKI in hospitalized patients (pooled sensitivity 0.74, specificity 0.78). This page sets no cutoff.
• Error: Swapping in SpO₂. A P/F ratio needs PaO₂. The 2024 global ARDS definition uses SpO₂/FiO₂ only when SpO₂ is 97% or lower.
• Error: Entering a guessed FiO₂. Use the FiO₂ the patient is receiving when the sample is drawn. A nasal cannula flow rate does not set a known FiO₂.
• Error: Applying Winter's formula during a rapidly changing acidosis. Compensation takes time. A fast-changing patient may not show the expected PaCO₂ yet.

• EMS: Field teams usually record what the device shows, along with the draw time. The gaps, corrections and ratios get calculated at the receiving lab or in the ED.
• Transport: Point-of-care analyzers report electrolytes, gases and calculated values, and their units and reference ranges can differ. Write the FiO₂ and PEEP on every blood gas, so the P/F ratio can be read at handoff. Cabin altitude lowers PaO₂ at the same FiO₂. Read P/F with altitude in mind.
• Handoff: Send the calculated values with the raw inputs and the draw time. Any treatment decision built on them varies by state and agency protocol.