eGFR and Creatinine Clearance Calculator

Three equations, three different quantities

eGFR is the glomerular filtration rate, the volume of plasma the kidneys filter each minute, estimated from serum creatinine, age and sex; this calculator gives it by CKD-EPI and MDRD alongside the Cockcroft-Gault creatinine clearance. They are not three attempts at the same number, and treating them as interchangeable is the most common error on this page.

• Cockcroft-Gault estimates creatinine clearance in mL/min. It uses body weight and is not indexed to body surface area, so it scales with the person in front of you.
• MDRD and CKD-EPI estimate glomerular filtration rate indexed to 1.73 m² of body surface area. Neither takes weight at all. An indexed result is a rate per standard body size, which is what makes it comparable between people and wrong for weight-based dosing.

So comparing a clearance in mL/min against a GFR in mL/min/1.73 m² compares two different things. The tool shows CKD-EPI de-indexed to the patient’s own surface area for exactly this reason: that figure is the one you may put alongside Cockcroft-Gault. For someone of average size the two forms are close, which is precisely what makes the distinction easy to forget.

Why creatinine is a late and blunt marker

Look at the curve. It is a hyperbola, not a line, and the consequence is stark: at 55 years old, a creatinine moving from 0.6 to 1.2 mg/dL, which stays inside most reference intervals the whole way, corresponds to an estimate falling by more than a third.

The reason is that the remaining nephrons compensate. As filtration falls the surviving glomeruli each filter more, so creatinine barely rises until a substantial fraction of function is gone. Only then does it climb steeply, which is the flat part of the curve on the right.

It also lags. After an acute injury creatinine takes a day or more to accumulate, so an early value can look reassuring in a kidney that has already stopped working. That is why acute kidney injury is defined by a change in creatinine and by urine output, not by a single absolute value.

Why the three disagree, and in which direction

MDRD is a single power law. One exponent means one slope everywhere, and it was fitted in people who already had chronic kidney disease. So it underestimates at near normal function, by roughly 14 percent at a creatinine of 0.9 in the tool’s default patient, and it converges on CKD-EPI as function falls, which is the range it was derived in. Below about 0.7 mg/dL it has nothing to flatten it and diverges upward instead. Reporting MDRD above 60 is discouraged for both reasons, and the tool marks it when it happens.

CKD-EPI has two slopes, joined at a sex-specific creatinine of 0.7 for women and 0.9 for men. Shallow below, steep above. That is the whole fix for MDRD’s problem: one equation can then stay accurate across both normal and severely reduced function.

Cockcroft-Gault separates from both for an unusual body size, because it is the only one that takes weight. Change the mass slider and watch it move while the other two do not budge. For a 140 kg patient it nearly doubles, and that is not a better estimate of filtration: it is a different quantity that happens to be useful for dosing.

The race coefficient is gone, and why

The original MDRD and the 2009 CKD-EPI both carried a coefficient that raised the estimate for Black patients. It was removed in the 2021 revision of CKD-EPI, and this tool implements the race-free version with no race input to give.

The reasoning was twofold. Race is a social category rather than a biological determinant of creatinine production, so the coefficient had no mechanistic basis. And its effect was to report higher function for Black patients at the same creatinine, which delayed referral to nephrology and delayed transplant listing. Removing it was a decision about harm, not only about accuracy.

When an estimate from creatinine should not be used at all

Every equation here assumes creatinine is in a steady state and that its production tracks a typical relationship with age and sex. Where either assumption fails, the estimate is unreliable in a direction you can usually predict.

• Acute kidney injury. Creatinine is still rising, so any estimate overstates function.
• Unusual muscle mass. A bodybuilder produces more creatinine and a cachectic or amputee patient produces less, so function is under- and overestimated respectively.
• Pregnancy, where filtration rises substantially and none of these equations was fitted.
• High protein or creatine intake, and drugs such as trimethoprim and cimetidine that block tubular secretion of creatinine without changing filtration at all.
• Children. All three are adult equations. Paediatric practice uses the Schwartz formula, which is why the tool’s age input starts at 18.

In those situations a measured clearance, or an estimate based on cystatin C, is the better tool.

Reading the category, and what it does not mean

The KDIGO categories run from G1 above 90 down to G5 below 15. The line at 60 matters most, because below it a reduced eGFR is a stage of chronic kidney disease on its own, and above it other evidence of damage such as albuminuria is needed for the diagnosis.

A category is not a diagnosis. KDIGO requires the abnormality to have been present for at least three months, and a single blood test cannot establish that. Staging also has an albuminuria axis that nothing on this page touches, and it carries at least as much prognostic weight as the filtration category.

Converting the units, which is where the real danger is

Multiply mg/dL by 88.4 to get µmol/L, and divide to go back. The factor comes from creatinine’s molar mass of 113.12 g/mol together with the change from decilitres to litres.

British, Canadian and Australian laboratories usually report µmol/L; American and Indian ones report mg/dL. Every equation here expects mg/dL. Entering 90 where the formula wants 1.02 makes Cockcroft-Gault wrong by about ninetyfold, and CKD-EPI and MDRD wrong by roughly two hundredfold, because both divide by creatinine raised to a power above 1. That is the single most dangerous mistake available on this page, and it is why the unit is a visible choice in the tool rather than an assumption.

Common mistakes

• Comparing an indexed GFR with a clearance. One is per 1.73 m², the other is not. De-index first, which the tool does for you.
• Using an indexed eGFR for weight-based drug dosing. It describes a standard body, not this one. That is why Cockcroft-Gault survives.
• Getting the units wrong. Ninetyfold to two hundredfold depending on the equation, and the arithmetic looks fine.
• Trusting an eGFR in acute kidney injury. Creatinine has not caught up yet.
• Reading a normal creatinine as normal function, particularly in an elderly or low muscle mass patient. The curve is steepest inside the reference interval.
• Quoting MDRD above 60. It is unreliable there by construction and the tool flags it.
• Calling a category a diagnosis. Three months, plus the albuminuria axis, plus a clinical picture.