Two formulas, two different questions
Most calculators give you one number and leave you to work out what it means. The two equations here are not interchangeable, and using the wrong one has consequences.
| Cockcroft-Gault | CKD-EPI 2021 | |
|---|---|---|
| Reports | Creatinine clearance | Estimated GFR |
| Units | mL/min | mL/min/1.73 m² |
| Normalised to body size | No | Yes |
| Published | 1976 | 2021, race-free |
| Primary use | Renal drug dosing | CKD staging and referral |
| Needs weight | Yes | No |
The critical difference is normalisation. CKD-EPI reports what the kidney function would be if the patient had an average-sized body of 1.73 m². That is exactly what you want when comparing one patient against population thresholds. It is exactly what you do not want when deciding how much drug this particular body will clear.
Why Cockcroft-Gault survives a 1976 publication date
By modern standards Cockcroft-Gault is crude. It was derived from 249 men, it uses weight as a proxy for muscle mass, and it applies a flat 0.85 multiplier for women. It is less accurate than CKD-EPI at estimating true GFR.
It persists because drug labels were written against it. When a package insert says to reduce the dose below 30 mL/min, that threshold came from pharmacokinetic studies that calculated clearance with Cockcroft-Gault. Substituting a different equation means comparing a number against a cut-off that was never defined on that scale. The two can disagree by 20% or more in the same patient, and near a dosing threshold that is the difference between two dose levels.
The practical rule most pharmacy references give: use the equation the label used.
The weight question, which changes the answer most
Cockcroft-Gault needs a weight, and which weight you feed it moves the result more than any other choice on this page.
- Ideal body weight (Devine). The usual default. The formula was derived in a non-obese population, and adipose tissue generates almost no creatinine, so it should not count toward clearance capacity.
- Actual body weight. Reasonable when the patient is at or below ideal weight — using ideal weight in an underweight patient would overestimate clearance.
- Adjusted body weight. Ideal plus 40% of the excess over ideal. Commonly applied when actual exceeds ideal by more than roughly 30%, on the reasoning that obesity adds some lean mass along with the fat, just not proportionally.
In a 120 kg patient whose ideal weight is 66 kg, these three choices can produce results differing by nearly a factor of two. That is not a rounding difference; it can be a whole dosing band. The calculator shows which weight it used so the number is never ambiguous.
Removing the race coefficient
The 2009 CKD-EPI equation included a multiplier of roughly 1.16 for patients recorded as Black. In practice it reported higher kidney function for those patients, which pushed them further from the thresholds that trigger nephrology referral, medication adjustment and transplant waitlist eligibility.
In 2021 a joint task force of the National Kidney Foundation and the American Society of Nephrology recommended a refit equation without the coefficient, on the reasoning that race is a social classification rather than a physiological variable, and does not belong as a term in an equation about filtration. The version implemented here is that race-free 2021 equation.
Reading the staging
KDIGO stages chronic kidney disease by eGFR band. Staging uses the normalised figure, not the de-indexed one:
| Stage | eGFR (mL/min/1.73 m²) | Description |
|---|---|---|
| G1 | 90 and above | Normal or high |
| G2 | 60–89 | Mildly decreased |
| G3a | 45–59 | Mild to moderately decreased |
| G3b | 30–44 | Moderately to severely decreased |
| G4 | 15–29 | Severely decreased |
| G5 | Below 15 | Kidney failure |
A stage on its own is not a diagnosis of chronic kidney disease. CKD requires the abnormality to persist for at least three months, and staging is completed with albuminuria, which this calculator does not take. A single eGFR of 55 in an otherwise well person is a reason to repeat the test, not a diagnosis.
When creatinine misleads
Every equation here estimates filtration from a single blood marker whose production depends on muscle. That assumption fails in identifiable situations:
- Acute kidney injury. Creatinine lags real change by a day or more. In a rapidly falling GFR, any estimate based on it reads far too high.
- Atypical muscle mass. Amputation, paraplegia, severe malnutrition and advanced cirrhosis lower creatinine production, so filtration looks better than it is. Bodybuilders and people on creatine supplements show the reverse.
- Drugs that block tubular secretion. Trimethoprim, cimetidine, dolutegravir and cobicistat raise serum creatinine without reducing filtration at all. The rise is real, the loss of kidney function is not.
- Pregnancy. Plasma volume and GFR both change substantially; neither equation was validated for it.
When the estimate and the clinical picture disagree, cystatin C or a measured clearance resolves it. The formula is a screening estimate, not a measurement.
Units and privacy
Serum creatinine is reported in mg/dL in the United States and in µmol/L in most of the rest of the world, related by a factor of 88.4. Switching the unit here converts the value you already typed rather than reinterpreting it — reinterpreting 1.2 mg/dL as 1.2 µmol/L would be an error of two orders of magnitude, and it is a real failure mode in tools that simply relabel the field.
Everything runs in your browser. No patient value is transmitted, logged or stored, and there is no account. You can disconnect from the network and the calculator still works.
Frequently asked questions
Which result should I use for drug dosing?
Cockcroft-Gault, in mL/min, unless the drug label says otherwise. Most renal dosing thresholds in package inserts were derived from studies that used Cockcroft-Gault, so matching the formula to the label keeps you on the same scale the threshold was defined on. CKD-EPI is normalised to 1.73 m² of body surface area, which is the right thing for staging chronic kidney disease and the wrong thing for dosing a specific body.
Why does CKD-EPI 2021 no longer include a race coefficient?
The 2009 equation multiplied results for Black patients by about 1.16, which systematically reported higher kidney function and could delay referral or transplant listing. In 2021 a joint NKF-ASN task force recommended removing it, on the grounds that race is a social category rather than a biological variable that belongs in a physiological equation. The 2021 equation used here is the race-free version.
Should Cockcroft-Gault use actual, ideal or adjusted body weight?
Ideal body weight is the common default. The formula was derived in a population that was not obese, and fat tissue produces very little creatinine, so feeding actual weight into it in an obese patient overestimates clearance and can lead to overdosing. Adjusted body weight — ideal plus 40% of the excess — is widely used when actual weight exceeds ideal by more than about 30%. There is no single universal rule, which is why the tool lets you choose and shows which weight it used.
When is a creatinine-based estimate unreliable?
Whenever creatinine is not in steady state or muscle mass is atypical. That includes acute kidney injury, where creatinine lags behind the actual change by a day or more; pregnancy; amputation; paraplegia; severe malnutrition; cirrhosis; very muscular or very frail patients; and anyone on a drug that blocks tubular creatinine secretion, such as trimethoprim or cimetidine, which raises creatinine without changing filtration at all.
Is any of this sent to a server?
No. Every calculation runs in your browser as arithmetic. Nothing is uploaded, nothing is logged and there is no account, which matters because the inputs are patient data.