Kidney function equations

CKD-EPI estimates GFR for staging; Cockcroft-Gault estimates creatinine clearance for drug dosing; KFRE predicts kidney failure.

Side-by-side comparison of model characteristics
Characteristic	CKD-EPI 2021	Cockcroft-Gault	KFRE
Clinical question	What is the estimated glomerular filtration rate?	What is the estimated creatinine clearance for drug dosing?	What is the 2- and 5-year risk of kidney failure in CKD G3–G5?
Outcome predicted	Estimated measured GFR	Creatinine clearance	Treated kidney failure (dialysis or transplantation)
Time horizon	Point in time	Point in time	2 and 5 years
Intended population	Pooled research and clinical cohorts (CKD-EPI consortium).	249 adult male Veterans Affairs inpatients (1976).	Canadian referred CKD G3–G5 cohorts (derivation 2011); validated in 31 multinational cohorts across 4 continents (2016).
Evidence level	Strong evidence	Moderate evidence	Strong evidence
Validation	—	—	Multinational validation (Tangri 2016): Calibration adequate in North American cohorts; the original equation overestimated risk in some non-North American cohorts, corrected by lowering baseline risk by 32.9% (2 years) and 16.5% (5 years).
Discrimination	—	—	—
Calibration	—	—	—
Geographic applicability	—	—	North America (original calibration); recalibrated for other regions.
Guideline status	Recommended by the NKF-ASN Task Force (2021) and KDIGO 2024 CKD guideline.	—	KDIGO 2024 CKD guideline recommends validated risk equations such as the KFRE to guide referral and planning.
Important limitations	Not valid in AKI or unstable creatinine. Affected by muscle mass, diet and some drugs.	Derived before creatinine standardisation; overestimates in obesity with actual weight. Female correction (0.85) was assumed, not measured.	Does not account for the competing risk of death, so risk may be overestimated in older or multimorbid patients. Non-North American populations need the recalibrated baseline.

Inputs required

Which variables each model uses. Models with more inputs are not automatically better — validation in your population matters most.

Input variables used by each model
Input	CKD-EPI 2021	Cockcroft-Gault	KFRE
Age	●used	●used	●used
Sex	●used	●used	●used
Serum creatinine	●used	●used	–not used
Actual body weight	–not used	●used	–not used
Height (optional)	–not used	●used	–not used
eGFR	–not used	–not used	●used
Urine ACR	–not used	–not used	●used

Comparison shows model characteristics only. Different models predict different outcomes in different populations, so their numerical results are not directly comparable.