The Cockcroft-Gault Equation Explained

In 1976, two Canadian physicians published a short paper in Nephron describing a way to predict creatinine clearance from a serum creatinine value, an age, a sex and a weight. Fifty years later, that equation is still the figure most renal drug-dosing guidance is written against — and still the one most people apply without quite knowing what it assumes.

This article takes it apart.

The equation

CrCl (mL/min) = [(140 − age) × weight(kg)] / (72 × Scr)   × 0.85 if female

Where Scr is serum creatinine in mg/dL. In SI units, with creatinine in µmol/L:

CrCl (mL/min) = [(140 − age) × weight(kg) × F] / Scr
F = 1.23 (male) · F = 1.04 (female)

The two forms are the same equation. The SI constants exist only because 1 mg/dL equals 88.4 µmol/L, and 72 ÷ 88.4 works out at roughly 1.23. Our CrCl calculator converts exactly rather than using the rounded constants, so both unit paths give an identical answer.

Where it came from

Cockcroft and Gault derived the equation from 249 patients — 236 men and 13 women — with measured 24-hour creatinine clearances. They were looking for something a clinician could compute at the bedside, in an era before automated laboratory reporting of estimated filtration rates.

Two facts about that cohort matter enormously and are almost never mentioned:

• It was overwhelmingly male. The female correction factor was not derived from a large female sample; it was an adjustment applied to account for lower average muscle mass.
• It was small, and it was 1976. Creatinine assays have changed since — modern IDMS-traceable enzymatic assays report lower values than the Jaffe methods of the 1970s.

Neither fact makes the equation useless. Both are reasons to treat its output as an estimate with a real error bar rather than a measurement.

What each term is doing

(140 − age)

This encodes the observation that filtration declines with age. Creatinine clearance falls by roughly 1 mL/min per year from about age 40 in otherwise healthy adults. Subtracting age from a fixed constant is a crude linear approximation of that decline — but across the age range where the equation is used, it holds up reasonably well.

At age 140, this term reaches zero. That is an artefact of curve-fitting, not a biological claim.

× weight (kg)

This is the term doing the most work, and the one most often got wrong.

Weight is standing in for muscle mass, because muscle mass determines how much creatinine the body produces, and therefore how much the kidneys have to clear. In a normal-weight adult, total body weight is a decent proxy for muscle mass.

In an obese patient it is not. Adipose tissue produces very little creatinine. Feed actual body weight into the equation for someone with a BMI of 39 and you will credit them with clearance they do not have — often by 30% or more. This is the single largest source of avoidable error in routine CrCl calculation, and we cover it in detail in adjusting CrCl for obese patients.

÷ (72 × Scr)

Serum creatinine sits in the denominator, which has an important consequence: the relationship between creatinine and clearance is curvilinear, not linear.

| Serum creatinine (mg/dL) | CrCl for a 70 kg, 60-year-old man | | --- | --- | | 0.8 | 97 mL/min | | 1.0 | 78 mL/min | | 1.5 | 52 mL/min | | 2.0 | 39 mL/min | | 3.0 | 26 mL/min |

Notice the shape. Moving from 0.8 to 1.0 — a change many clinicians would not blink at — costs 19 mL/min. Moving from 2.0 to 3.0, a much larger absolute change, costs only 13 mL/min.

The practical implication: small rises in creatinine within the reference range represent large losses of kidney function. A patient whose creatinine drifts from 0.8 to 1.1 over two years has lost roughly a quarter of their estimated clearance while every laboratory report came back flagged as normal.

The 72 is simply the constant that made the units work out against the measured clearances in the original cohort.

× 0.85 if female

A flat 15% reduction, applied to account for lower average muscle mass and therefore lower creatinine production in women.

It is a population-average adjustment applied to an individual, which is a conceptual weakness it shares with almost every clinical estimating equation. A heavily muscled woman and a frail man are both poorly served by it.

What the equation assumes

Every one of these has to hold for the result to be meaningful:

1. Renal function is stable. The equation assumes serum creatinine is at steady state. In acute kidney injury it is not — creatinine lags the true filtration rate by a day or more, so a patient whose kidneys failed this morning still has a reassuring creatinine this afternoon. In AKI, Cockcroft-Gault is not merely imprecise; it is wrong, and wrong in the dangerous direction.
2. Creatinine production is proportional to body weight. Fails in obesity, cachexia, amputation, paraplegia and extreme muscularity.
3. The patient produces creatinine normally. Fails on a vegetarian diet (less dietary creatine), with creatine supplementation, and in advanced liver disease.
4. Nothing is interfering with the assay or with tubular secretion. Trimethoprim and cimetidine both raise serum creatinine by blocking tubular secretion without changing filtration at all.

Why it is still used

Given all that, it is fair to ask why anyone still uses a 1976 equation when CKD-EPI 2021 exists.

The answer is not clinical superiority — CKD-EPI is more accurate. The answer is regulatory continuity. The pharmacokinetic studies that established renal dose adjustments for most drugs on the market used Cockcroft-Gault creatinine clearance as their measure. The thresholds printed on the label — "reduce dose if CrCl < 30 mL/min" — mean Cockcroft-Gault CrCl.

Substituting an eGFR figure into a threshold that was derived against CrCl introduces a mismatch. The two are in different units, and even after de-indexing they do not agree. We work through this in GFR vs CrCl: which should guide dosing.

So: use CKD-EPI for staging kidney disease, and Cockcroft-Gault for dosing drugs whose labels specify creatinine clearance. Not because the older equation is better, but because it is the one the dosing evidence was built on.

How accurate is it?

Studies comparing Cockcroft-Gault against measured clearance consistently find that it overestimates by roughly 10–30%. Some of that is the equation; some is that measured creatinine clearance itself overestimates true GFR, because the renal tubules secrete creatinine in addition to filtering it.

For most dosing decisions, an estimate within 30% is adequate. Where it is not — aminoglycosides, chemotherapy, a patient sitting right at a threshold — a timed urine collection measures the patient in front of you rather than estimating from a 1976 cohort — provided the collection is complete, which is the part that decides whether the result means anything.

In practice

• Choose the weight basis before you calculate. It matters more than anything else you will do.
• Do not read a serum creatinine and assume. Calculate.
• Check whether the patient is in a steady state. If they are not, the number is not valid.
• Know which measure the drug label specifies, and use that one.

You can run all of this on the CrCl calculator, which shows actual, ideal and adjusted body weight side by side with the working displayed in full.