eGFR calculator (CKD-EPI 2021) for eligibility screening

The CKD Epidemiology Collaboration (CKD-EPI) published its first creatinine equation in 2009, with a multiplier for Black race. In 2021 Inker and colleagues (New England Journal of Medicine) refit the equation in a pooled data set without the race term and published the race-free creatinine equation, together with a creatinine-cystatin C version. A joint task force of the National Kidney Foundation and the American Society of Nephrology recommended it for immediate adoption, and US laboratories have been moving to it since. It is the equation this calculator uses, for adults whose creatinine is calibrated to the IDMS reference method.

The shape is the same as before: two segments joined at a knee. Below the knee (creatinine at or below κ, which is 0.7 mg/dL in women and 0.9 mg/dL in men), eGFR depends weakly on creatinine through the exponent α; above it, eGFR falls with creatinine to the power −1.200. Age enters as 0.9938 per year, about a 0.6% fall in eGFR per year of age, and women get a factor of 1.012. Remember that the constants are different from the 2009 equation. Mixing the two sets of numbers is a known source of error in hand-built spreadsheets.

The equation estimates glomerular filtration rate from a single creatinine value. It assumes a steady state, so it is not valid when creatinine is changing quickly, as in acute kidney injury. Creatinine also reflects muscle mass and diet, and a few drugs, such as trimethoprim and cimetidine, raise creatinine without changing true GFR by blocking its tubular secretion. In people with unusual muscle mass, extreme body size or amputations, or where kidney function decides a dose, a cystatin C equation or a measured GFR is the better route, and the protocol will say so.

Kidney function appears in the eligibility criteria of many trials, for safety (a drug cleared by the kidney), for dosing, or because the condition under study depends on it, as in heart failure, diabetes and transplant studies. A usable criterion names five things: the equation (CKD-EPI 2021 creatinine, an older equation, or Cockcroft-Gault), the threshold and its direction, whether the value is indexed to body surface area (mL/min/1.73 m²) or absolute (mL/min), the sample (local or central laboratory, and how recent), and what happens when the screening value is borderline.

The equation matters more than people expect. For the same person, CKD-EPI 2021, MDRD and Cockcroft-Gault can differ by 10 or more mL/min, and the Cockcroft-Gault result also moves with body weight. Some protocols name Cockcroft-Gault because the drug’s dosing was derived using creatinine clearance; others use CKD-EPI. The calculator’s optional weight field shows the Cockcroft-Gault value next to the eGFR so you can see both. Absolute and indexed values also differ for people far from the average body size.

Borderline results deserve a rule. A creatinine a tenth of a mg/dL higher is within the analytical and biological variation of the test, yet it can move a participant across a cut-off. The calculator shows the eGFR at ±0.1 mg/dL so a coordinator can see how close the call is. The protocol should say whether a repeat is allowed within the screening window and which value counts. See the eligibility screening eCRF template for how criteria and their source values are recorded.

This is a reference calculation, not clinical advice. It reproduces a published equation so that a protocol writer, data manager or coordinator can check a number. It is not a medical device, it has not been validated as clinical software, and it must not be used to decide a treatment, a dose, or whether a real person is eligible for a trial. Those decisions belong to the protocol and the investigator, using the laboratory’s own reported value and local procedures.

It does not apply to children, for whom paediatric equations such as the bedside Schwartz are used, or in acute kidney injury, pregnancy or dialysis. It does not use cystatin C, race or an ethnicity coefficient. It does not stage chronic kidney disease, which needs a persistent abnormality over more than three months and albuminuria. If your data come from several laboratories, record which laboratory and which method produced each creatinine, see lab data and reference range management.

For a data manager, the other side of the same coin is consistency. A calculated field applies one formula to every participant, which removes hand calculation and the class of query it generates. It cannot fix a wrong input, so the quality of the creatinine, age and sex entries still decides the result. The calculated fields and skip logic guide covers how to build and test derived fields in general.

In heart failure trials and transplant trials eGFR is collected at screening for eligibility and again throughout follow-up as a safety or efficacy measure, so the same calculated field is reused across visits, and the same formula should be applied at every one. Diabetes studies often use eGFR as a stratification factor or a safety check. The vital signs eCRF template carries the other derived values, such as BMI and mean arterial pressure, that sit alongside it.