CKD-EPI vs MDRD vs Cockcroft-Gault: Which GFR Formula Is Most Accurate?

CKD-EPI, MDRD and Cockcroft-Gault are three different equations that turn the same blood creatinine value into three different kidney-function numbers. For one 65-year-old woman with a serum creatinine of 1.0 mg/dL, the equations on this page give 62.5 (CKD-EPI 2021), 58.9 (CKD-EPI 2009), 55.6 (MDRD) and 62.0 mL/min (Cockcroft-Gault, assuming 70 kg). That spread straddles the common cutoff of 60, which is why it matters which equation your lab used. The National Kidney Foundation (NKF) and the NIDDK recommend the 2021 CKD-EPI creatinine equation for adults.

The four equations side by side

All four use serum creatinine in mg/dL. The first three are indexed to a standard body surface area of 1.73 m2; Cockcroft-Gault is not, and it estimates creatinine clearance rather than GFR. The coefficients below are taken from the NKF and NIDDK equation pages, the Tufts CKD-EPI equation page, and the published 2009 CKD-EPI equation.

Worked example: one woman, four answers

Inputs: female, age 65, serum creatinine 1.0 mg/dL (88.4 umol/L). Cockcroft-Gault also needs body weight, so we assume 70 kg and show 60 kg and 80 kg to illustrate how sensitive it is. CKD-EPI 2009 is shown without the race multiplier. Results are rounded to one decimal place and were computed directly from the equations above.

The 2021 equation lands just above 60 while the older two land just below it. A difference of 3 to 7 points does not mean her kidneys changed; it means the same creatinine was converted with different curves. Cockcroft-Gault swings by almost 18 mL/min across a 20 kg weight range because body weight is a direct input. The NKF categories used here (G1 90 or above, G2 60 to 89, G3a 45 to 59, G3b 30 to 44, G4 15 to 29, G5 below 15) come from the NKF CKD classification page, which also notes that G1 and G2 do not by themselves meet the criteria for CKD without evidence of kidney damage.

More cases: how the gap changes

Two patterns show up in these four rows. First, in every row the MDRD value is below the 2021 CKD-EPI value; both the NKF and the NIDDK state that CKD-EPI is more accurate than MDRD, particularly at higher GFR. Second, Cockcroft-Gault is well above both for the heavier men and below for the 75-year-old, because it responds to weight and age differently. The 30-year-old man in the last row has a creatinine of 1.4; if he is very muscular, creatinine can run high without any kidney problem, and clinicians then often turn to cystatin C.

What changed between 2009 and 2021

The 2021 equation removed the race variable. Tufts, which hosts the CKD-EPI equations, states that compared with 2009 the 2021 version gives lower eGFR for Black individuals and higher eGFR for non-Black individuals. Computing both equations across ages 20 to 90 and creatinine values that give an eGFR between 15 and 120, we find the 2021 result is about 5 to 14 percent lower than the 2009 result with the 1.159 multiplier, and between about 0 and 11 percent higher than the 2009 result without it. Our 65-year-old woman is a non-Black example: 58.9 becomes 62.5, which moves her from G3a to G2. See our post on the race-free equation for the full discussion.

The same Tufts page says the 2021 equation is somewhat less accurate than 2009 against measured GFR in both groups but remains acceptable for clinical use in many circumstances. The 2021 development paper (Inker and colleagues, New England Journal of Medicine) reported that 85 percent or more of eGFR values in Black and non-Black participants fell within 30 percent of measured GFR for all the equations it tested, and that creatinine-cystatin C equations without race were the most accurate and showed the smallest differences between groups.

When each equation is used

• CKD-EPI 2021 creatinine: recommended by the NKF and NIDDK for estimating GFR in adults, and the equation our GFR calculator uses.
• CKD-EPI 2021 creatinine-cystatin C: NIDDK states that using both markers is preferred and more accurate than creatinine alone, and the NKF suggests cystatin C confirmation for some borderline creatinine results.
• MDRD: a legacy equation. Some older reports and software may still show it.
• Cockcroft-Gault: still named in many drug labels. NIDDK notes that dosing recommendations were developed using several methods, including Cockcroft-Gault creatinine clearance, and that eGFR can be converted for dosing as eGFR x BSA / 1.73. With an assumed body surface area of 2.0 m2, our 62.5 becomes 72.3 mL/min. That conversion matters mainly when body size is very different from average.

Common mistakes when comparing results

• Comparing numbers from different equations as if they were one trend. A move from 58.9 to 62.5 may be the equation, not the kidney.
• Mixing units. Cockcroft-Gault is in mL/min; CKD-EPI and MDRD are in mL/min/1.73 m2.
• Entering creatinine in the wrong unit. Divide umol/L by 88.42 to get mg/dL.
• Staging from a single test. Under KDIGO 2024, CKD means abnormalities of kidney structure or function present for more than 3 months.
• Assuming the equation fixes the muscle-mass problem. Creatinine-based equations are less reliable at extremes of muscle mass, in pregnancy, and when kidney function is changing quickly.

Frequently asked questions

Why does my eGFR change when my lab switches formula?

The equations use different curves fitted to different datasets. For the same creatinine, age and sex, our examples differ by 3.5 to 12 points between CKD-EPI 2021 and MDRD. Ask your lab or clinician which equation produced each result before comparing them.

Is Cockcroft-Gault obsolete?

It is not used for staging CKD, but it still appears in drug labelling because many dosing studies used it. NIDDK also recommends considering the combined creatinine-cystatin C equation when eGFR may directly affect dosing.

Does this calculator use MDRD?

No. Our calculator uses the 2021 CKD-EPI creatinine equation only. The MDRD and Cockcroft-Gault values on this page were computed separately from the published formulas.