How this instrument works
Winters' formula estimates the PaCO₂ a person's lungs should reach once they have fully compensated for a metabolic acidosis. Multiply the measured bicarbonate by 1.5 and add 8, and the result is the expected carbon dioxide tension in mmHg — the point respiratory compensation should land on if the lungs are the only thing responding to the falling pH. Albert, Dell, and Winters published the regression behind it in 1967, having plotted paired bicarbonate and PaCO₂ readings from patients with metabolic acidosis and fitted the line that best described how one moved with the other.
The formula never claimed to predict one exact number — scatter around that regression line was real, so the accepted convention wraps the expected value in a band of plus or minus 2 mmHg. A measured PaCO₂ landing inside that band means the lungs are doing precisely what a metabolic acidosis alone would call for, nothing more. Comparing a measured value against that predicted band, rather than against the raw expected number alone, is the entire clinical use of the tool.
A measured value sitting outside the band is the useful finding. Too high, and the lungs are not blowing off as much carbon dioxide as expected — a second, independent problem is holding the level up, namely a respiratory acidosis riding along with the metabolic one. Too low, and the patient is breathing faster than the acidosis alone explains, pointing instead to a concurrent respiratory alkalosis. Either way, catching a mixed disorder that a bicarbonate and pH reading alone would hide is the whole point of running this arithmetic.
- Enter the measured serum bicarbonate, HCO₃⁻, in mEq/L from the metabolic panel.
- Enter the measured PaCO₂ in mmHg from the same blood gas draw.
- Read the expected PaCO₂ and its ±2 mmHg band — the working block shows the multiplication and addition in full.
- Check whether the actual PaCO₂ falls within that band; a value outside it flags a second acid-base disorder.
Worked example — HCO₃⁻ 12 mEq/L
A bicarbonate of 12 mEq/L: 1.5 × 12 = 18, plus 8 gives an expected PaCO₂ of 26 mmHg, with an acceptable band of 24 to 28 mmHg. A blood gas drawn on the same patient reads 26 mmHg — dead center of the band. The lungs are compensating exactly as much as this metabolic acidosis alone would predict, with no second disorder needed to explain the number.
Keep that same bicarbonate of 12 mEq/L but change the measured reading to 35 mmHg, well above the 24-28 band. The lungs are not clearing nearly enough carbon dioxide for how acidotic this bicarbonate is, so a concurrent respiratory acidosis — ventilation not keeping pace with the metabolic problem — is the likely explanation. A higher bicarbonate tells a calmer story: at 18 mEq/L, the expected figure climbs to 35 mmHg (range 33 to 37), and a measured value of 34 mmHg again falls comfortably inside its band.
Questions
What does it mean if the measured PaCO₂ falls above the expected range?
It means the lungs are retaining more carbon dioxide than this metabolic acidosis alone would explain — a second, independent respiratory acidosis is present, whether from sedation, neuromuscular weakness, airway obstruction, or plain exhaustion of the drive to breathe. The metabolic and respiratory problems are stacking, and the reading is worse than compensation alone would ever produce.
What does it mean if the measured PaCO₂ falls below the expected range?
The opposite problem: the patient is hyperventilating past what the bicarbonate alone calls for, pointing to a concurrent respiratory alkalosis — pain, anxiety, sepsis, salicylate toxicity, and pregnancy are common drivers. Two forces are pulling the carbon dioxide level down at once, the expected metabolic compensation plus an independent respiratory drive layered on top.
Why does the formula add a constant of 8 instead of starting from zero?
Because the relationship between bicarbonate and PaCO₂ is not a straight line through the origin. Albert, Dell, and Winters fit their regression to real patient measurements, and the best-fitting line crossed the axis at 8, not 0. Both the slope of 1.5 and the intercept of 8 come directly from that 1967 dataset rather than from a theoretical derivation.
Why is there a tolerance band instead of one exact predicted number?
Because no regression line passes through every data point exactly — real patients scattered around the fitted relationship even while compensating normally for their acidosis. A band around the expected value absorbs that ordinary variation, so a reading anywhere inside it counts as consistent with compensation alone rather than as proof of a second disorder.
Does Winters' formula apply to any acid-base disturbance?
No — it was built specifically for metabolic acidosis and the respiratory response that follows it. Other primary disorders use separate compensation rules entirely: respiratory acidosis and alkalosis have their own expected bicarbonate shifts, and metabolic alkalosis has a different expected rise in carbon dioxide tension. Plugging numbers from a different primary disorder into this arithmetic produces a meaningless answer.
Is this the same as the bedside rule comparing pH digits to PaCO₂?
It is related but not identical. That bedside trick compares the trailing digits of arterial pH against the measured carbon dioxide reading as a fast mental shortcut, and it can drift from Winters' actual regression in more extreme cases. This calculator applies the original 1.5 × HCO₃⁻ + 8 arithmetic directly, which is the steadier version to lean on when precision matters.
References
- Albert, Dell & Winters 1967, Ann Intern Med — original study (PubMed)
- StatPearls — Physiology, Acid Base Balance (NCBI Bookshelf)
Read this first: This instrument computes a screening figure from population formulas — it is not a diagnosis, and it cannot see the whole picture a clinician can. Use it to inform a conversation, not to replace one.