Three independent variables
Stewart describes pH through PaCO₂, strong ion difference (SID), and total weak acids (mainly albumin and phosphate). Bicarbonate is a dependent variable. This app uses SBE to quantify the net metabolic component, the Stewart–Figge measurements to explain its contributors, and empirical bicarbonate-based rules to assess compensation. SBE is a derived summary, not a Stewart independent variable.
Carbon dioxide
PaCO₂ describes the respiratory component.
Strong ions
The difference between strong cation and anion charge.
Weak acids
Albumin and phosphate contribute pH-dependent charge.
Core equations
Concentrations are mmol/L unless stated. Charge is reported in mEq/L, which incorporates valence. Gas equations assume 37°C and PaCO₂ in mmHg.
- Calculated bicarbonate · Henderson–Hasselbalch
- HCO₃⁻ = 0.0307 × PaCO₂ × 10(pH − 6.105)
- Estimated standard base excess · Van Slyke form
- SBE = 0.9287 × [HCO₃⁻ − 24.4 + 14.83 × (pH − 7.4)]
- Apparent strong ion difference
- SIDa = Na + K + 2Ca + 2Mg + Li − Cl − lactate − 2sulfate
- Albumin and phosphate charge · Figge
- Alb⁻ = albumin (g/L) × (0.123 × pH − 0.631)
Pi⁻ = phosphate × (0.309 × pH − 0.469) - Effective strong ion difference & strong ion gap
- SIDe = HCO₃⁻ + Alb⁻ + Pi⁻
SIG = SIDa − SIDe - Anion gap · potassium excluded throughout
- AG = Na − Cl − HCO₃⁻
AGcorrected = AG + 0.25 × [40 − albumin (g/L)]
Delta ratio = (AGcorrected − local AG midpoint) / (24 − HCO₃⁻)
Chart methods
Expected compensation
The main interpretation plots the actual value against the equation’s expected range. Acute and chronic ranges stay separate when the time course is unknown. Additional screening margins are described in the equation notes; they do not expand the plotted range. The chart uses the same compensation assessment as the written result.
Estimated base-excess contributions
The published simplified Fencl–Stewart method by Story et al. estimates the sodium–chloride and albumin effects. The remainder is explicitly a residual.
Na–Cl effect = Na − Cl − 38
Albumin effect = 0.25 × (42 − albumin in g/L)
Other effects = SBE − Na–Cl effect − albumin effect
The residual includes lactate, phosphate, minor ions, unmeasured ions, and approximation error. It is not SIG and cannot identify a particular acid. Opposing effects can produce a near-normal total.
Ion changes from the reference panel
The individual-ion graph is an algebraic charge decomposition, relative to the illustrative panel below. Strong cation changes are positive; strong anion changes are negative. Albumin and phosphate terms subtract their actual charge changes, incorporating both concentration and pH. Subtracting the change in SIG closes the balance.
ΔHCO₃⁻ = ΔSIDa − ΔAlb⁻ − ΔPi⁻ − ΔSIG
This identity is exact within the model. It is not a validated per-ion SBE partition or a prediction of treatment response. Changing an electrolyte while fixing pH and PaCO₂ will produce a compensating change in SIG.
The illustrative panel is a mathematical graph anchor, not a population-normal specimen. Its calculated SIG can exceed the conventional 0–2 mEq/L screen even though pH and SBE are within range.
Illustrative comparison panel — not laboratory reference intervals| Measurement | Value | Unit |
|---|
| pH | 7.400 | — |
|---|
| PaCO₂ | 40 | mmHg |
|---|
| Sodium | 140 | mmol/L |
|---|
| Potassium | 4 | mmol/L |
|---|
| Chloride | 104 | mmol/L |
|---|
| Lactate | 1 | mmol/L |
|---|
| Albumin | 40 | g/L |
|---|
| Phosphate | 1 | mmol/L |
|---|
| Ionized calcium | 1.2 | mmol/L |
|---|
| Magnesium | 0.8 | mmol/L |
|---|
| Lithium | 0 | mmol/L |
|---|
| Sulfate | 0 | mmol/L |
|---|
The reference SIG is a mathematical baseline from the illustrative panel, not a clinical normal range. The raw SIG always remains visible.
Charge balance
The charge-balance plot compares cation and anion equivalents. A positive SIG is drawn as net residual anionic charge; a negative SIG as net residual cationic charge. This closes the plot by construction and does not validate the input measurements.
Differential selection
The interpretation separates the primary disorder, compensation, measured contributors and possible causes. Cause names and clinical clues stay visible in Brief view; Detailed adds checks, equations and limits. Appropriate compensation does not create a second disorder’s differential.
Low albumin is shown separately when its alkalinising effect offsets acidosis. A raised lactate does not remove other possible gap causes, and an incomplete panel keeps unassessed mechanisms explicit. Causes are not ranked by probability.
The cause categories draw on the LITFL acid–base worksheet. Calculations use the Stewart–Figge model, source-specific intervals and the compensation rules below.
Ordering the causes
Metabolic groups are ordered by the size of their estimated charge effect. Acidifying and alkalinising effects both count; an opposing effect retains its label. Respiratory disorders and findings without a comparable estimate are listed separately.
For this comparison, albumin and phosphate charge are evaluated at pH 7.40. The mathematical anchors are Na − Cl 36 mmol/L, lactate 1 mmol/L, albumin 40 g/L and phosphate 1 mmol/L. A classified SIG is compared with the midpoint of the selected interval: the conventional screening range or a supplied local interval. These anchors do not replace laboratory reference intervals or diagnostic cutoffs.
Sodium–chloride: Na − Cl − 36
Lactate: 1 − lactate
Albumin: (40 − albumin) × 0.2792
Phosphate: (1 − phosphate) × 1.8176
SIG: selected interval midpoint − SIG
Effects are expressed in mEq/L: negative is acidifying, positive is alkalinising. The charge estimates guide display order; they are not individual SBE contributions, changes from this patient's baseline or predictions of treatment response. Their size depends on the stated anchors. The AG is a separate screen and receives no score. Equal magnitudes at one decimal place retain a stable order. Causes within a group require clinical assessment; they are not ranked by probability.
Interpretation rules
Complete pH correction means pH is 7.35–7.45 and the selected compensation rule is compatible. Incomplete pH correction means compensation is compatible but pH remains outside that range. Mixed disorders are never labelled fully compensated solely because pH is normal. These labels describe pH correction separately from adequacy of compensation.
Classification uses SBE, PaCO₂, pH, and the selected clinical course. When opposing metabolic and respiratory components coexist at a normal pH, both compensation hypotheses are checked. If both fit or the course changes the conclusion, the primary process is left uncertain.
- pH 7.35–7.45, PaCO₂ 35–45 mmHg, HCO₃⁻ 22–26 mmol/L, and the default SBE −2 to +2 mEq/L are screening ranges, not universal laboratory limits. The SBE cutoff can be changed to match the local laboratory convention. Values within 0.05 mEq/L of that cutoff are shown as borderline at displayed precision; this allowance is not measurement uncertainty. Changing this cutoff does not change the compensation equations or their margins.
- The history selects the acute or chronic respiratory rule. With an unknown course, both possibilities are shown separately. An SBE deviation can represent compensation and does not alone establish a second primary disorder.
- Winter’s expected PaCO₂ = 1.5 × HCO₃⁻ + 8 ±2 for metabolic acidosis. In metabolic alkalosis, the expected PaCO₂ rise is 0.6–0.75 per 1 mmol/L HCO₃⁻ above 24, usually no higher than 55 mmHg.
- For each 10 mmHg PaCO₂ rise above 40, expected HCO₃⁻ rises 1–2 acutely or 3–4 chronically. For each 10 mmHg fall, HCO₃⁻ falls 1–2 acutely or 4–5 chronically. Values start from HCO₃⁻ 24.
- A 2-unit margin is used outside the respiratory bicarbonate bands and the metabolic-alkalosis PaCO₂ band before suggesting an extra process. This is a screening convention, not a validated cutoff. Results outside the published band but within that extra margin are labelled borderline; they are not classified as complete compensation. A 0.05-unit numerical allowance avoids changing a classification through displayed rounding alone. Linear rules become less reliable at extremes and during rapid changes.
- The delta ratio is withheld for isolated respiratory compensation or an uncertain primary process, and uses the displayed local AG midpoint; it supports a pattern, never proves a single diagnosis. A high ratio may also reflect a previously raised bicarbonate.
- SIG is net residual charge, not a direct measurement or a specific diagnosis. Methods, assay errors, omitted ions, and local ranges affect it. The default SIG range is 0–2 mEq/L, a conventional screen described by Herdman et al. (2015). It is not a national or method-validated local interval. Replace it with your laboratory’s method-matched interval when available, or clear both limits to withhold comparison. Values within 0.05 mEq/L of either limit are shown as borderline at displayed precision. Assays and the total-magnesium approximation can shift SIG; an increase alone does not establish metabolic acidosis.
Measurements & model limits
- Use adult blood gases measured at 37°C. Venous, capillary and unknown samples retain quantitative calculations with their limits stated; arterial compensation and oxygenation conclusions are withheld. This app does not convert venous to arterial values or correct for patient temperature.
- Use ionized calcium. Total magnesium is a clinical approximation: it includes bound magnesium and can raise the calculated gap; ionized magnesium may be selected when measured. Method changes affect SIG comparability.
- All divalent strong ions are multiplied by two when converting mmol/L to mEq/L. Phosphate charge is already handled by the Figge equation and is not multiplied again.
- Unmeasured lithium and sulfate stay within the residual gap. Measured lactate, lithium, magnesium, and sulfate are never described as unmeasured components.
- Total protein is retained as optional context. Globulin charge is not independently modeled; adding total protein to albumin would double-count protein effects.
- The albumin linear charge approximation was fitted over pH 6.9–7.9. Extreme results and sample inconsistencies require independent review. Input guardrails are data-entry limits, not survivability limits.
- Calculations run in this file. Cases and a lab profile are stored only when you select Save on this device. Browser storage depends on the browser and file location; export a backup for portability. Clearing browser data removes saved cases. There is no analytics or remote calculation. Do not enter patient identifiers.
Medical disclaimer
ABG Bench supports clinical assessment and education. It does not replace clinical judgement, local protocols or a full assessment of the patient. Check inputs, units and results against the original measurements. Clinicians remain responsible for clinical decisions and patient care.
Errors or omissions may occur. To the extent permitted by applicable law, the provider accepts no liability for loss or harm arising from use of this tool. This does not exclude rights or liabilities that cannot lawfully be excluded.
References
- Rocktaeschel et al. · Critical Care, 2003
Clinical strong-ion calculations, charge-equivalent units, and Henderson–Hasselbalch constants.
- Figge–Fencl · Albumin titration model
The pH-dependent albumin charge approximation and its fitted range.
- Story, Morimatsu & Bellomo · BJA, 2004
Simplified Fencl–Stewart base-excess partition. DOI: 10.1093/bja/aeh018.
- Ortner et al. · Strong ion and weak acid analysis, BJA
The SBE equation and comparison with physicochemical analysis.
- Figge et al. · Anion gap and hypoalbuminemia, 1998
Albumin correction of the anion gap.
- Merck Manual · Compensation in simple acid–base disorders
Clinical compensation rules and acute/chronic ranges.
- Stewart · Independent and dependent variables of acid-base control
Independent variables and the dependent role of bicarbonate.
- González et al. · Secondary response to chronic respiratory acidosis
Observed variation in chronic compensation, including pH values within the usual reference interval.
Reference intervals and local methods
These UK and Australian sources provide adult laboratory intervals and clinical screening ranges. They are not a single national guideline. Use the interval for the local laboratory, specimen and assay.
SIDa and SIDe use supplied method-matched intervals. SIG starts with an editable conventional screening range of 0–2 mEq/L; a matching local interval takes priority. A simplified bedside SIG equation and the full Stewart–Figge calculation are not interchangeable. Published research comparisons are not applied as universal cutoffs.
Oxygenation and simulation
Oxygenation calculations require an arterial sample and a usable FiO₂. The result identifies estimated FiO₂ and displays the barometric-pressure and respiratory-quotient assumptions. The P/F ratio alone cannot establish ARDS.
The simulator solves electroneutrality for pH while holding residual SIG fixed. It has no fluid distribution, renal response, ventilation response or time course. It describes model sensitivity, not a treatment response.
Use and limitations
Interpret results alongside the clinical history, examination and laboratory findings. Software checks cover equations, interpretation rules and unit conversions; they do not replace independent clinical validation, which has not been completed.
Worked examples are synthetic teaching cases generated by the calculator’s equations.