Kinetic estimated glomerular filtration rate (keGFR)

Clinical background

In acute kidney injury the serum creatinine changes rapidly, and conventional eGFR formulae (CKD-EPI, MDRD, Cockcroft–Gault) presuppose a steady state. The result is systematic error: when the creatinine is rising, the standard formulae overestimate the actual GFR, and during recovery they underestimate it. This has direct clinical relevance for drug dosing, where overestimation can lead to toxicity and underestimation to subtherapeutic levels, particularly for antibiotics with a narrow therapeutic window such as vancomycin.

The kinetic estimated glomerular filtration rate (keGFR) was developed to solve this problem. The formula uses two consecutive creatinine values and the interval between them to calculate the clearance actually operating during the change, instead of waiting for a steady state to be reached.

Calculating the keGFR

The formula rests on the principles of creatinine balance: at steady state, production equals elimination, and the amount eliminated per unit time gives the clearance. When the creatinine is not stable, clearance can be derived from how much creatinine accumulates or disappears over a known time interval, if the production rate and the volume of distribution are known.

$$\text{keGFR} = \frac{S_{cr,\text{base}} \times \text{eGFR}$${\text{base}}}{\overline{S}{cr}} \times \left[1 - \frac{24 \times (S_{cr,2} - S_{cr,1})}{\Delta t \times \text{max},\Delta S_{cr}/\text{day}}\right]

where $S_{cr,\text{base}}$ is the baseline (steady-state) serum creatinine, $\text{eGFR}${\text{base}}eGFRbase​ is the baseline (steady-state) eGFR, $\overline{S}${cr} is the mean of the first and second serum creatinine, $S_{cr,2} - S_{cr,1}$ is the difference between the two creatinine values, $\Delta t$ is the time between the two creatinine values in hours, and max $\Delta S_{cr}$/day is the maximum rise in creatinine per day if the GFR were zero (often set to 1.5 mg/dL/day).

The last variable reflects anuric creatinine production and acts as a physiological ceiling: it states how fast the creatinine would rise if the kidneys stopped filtering altogether. Its value depends on muscle mass and catabolism, but 1.5 mg/dL/day is the accepted standard value for adults [1].

The formula was derived from first principles of creatinine balance, not from an empirical cohort. Chen published it in 2013 in the Journal of the American Society of Nephrology and demonstrated its properties through scenario analyses of different AKI patterns and of renal recovery [1]. Because the derivation is theoretical, all validation rests on external studies.

Interpretation in practice

The keGFR is interpreted like any GFR estimate in mL/min/1.73 m², but with one important difference: the value applies to the specific interval between the two samples, not as a general level. A rising creatinine pulls the keGFR below what the steady-state eGFR would show; a falling creatinine raises it. This is the essence of the formula and not an error, but a reflection of the fact that renal function is genuinely worse (or better) than a single creatinine value suggests.

Clinically, keGFR is used mainly for two decisions:

In a prospective observational study of 107 patients with AKI in a medical intensive care unit, 65% of patients required a dose change for at least one drug when keGFR was used instead of CKD-EPI. Vancomycin was the drug most often affected (36%), followed by aciclovir (23%) and meropenem (23%) [3]. In a larger secondary analysis of 946 ARDS patients from the FACTT trial, keGFR produced a change in dosing category in 23% of all patients and in 34 to 38% of those with AKI [2].

Validation and performance

The only prospective multicentre study to compare keGFR with measured GFR (Tc-DTPA) included 119 patients with AKI, 63% of them from intensive care and 71% with stage 1 AKI. keGFR correlated moderately with measured GFR (r = 0.68), comparable to the Jelliffe equation (r = 0.73). Both equations systematically underestimated the measured GFR, and the Bland–Altman limits were wide, over 40 mL/min/1.73 m², indicating low precision at the individual level. keGFR performed better in patients with chronic kidney disease and in the elderly [4].

In a retrospective analysis of AmsterdamUMCdb with 2,492 septic intensive care patients, keGFR was evaluated as a tool for early AKI diagnosis against the KDIGO criteria as the reference. keGFR combined with urine output achieved a sensitivity of 93%, a specificity of 73% and an accuracy of 86% (kappa = 0.77). keGFR identified stage 1 AKI a median of 13 hours earlier than KDIGO, and stages 2 and 3 approximately 5 to 8 hours earlier [5].

Taken together, the studies show that keGFR has reasonable discrimination for identifying AKI and for influencing dosing decisions, but that its precision for estimating an exact GFR at the individual level is limited. This is not unique to keGFR: in a study by Bragadottir et al., MDRD, CKD-EPI and Cockcroft–Gault also had an error margin of approximately 68% compared with measured GFR in early AKI [2].

Limitations

The formula presupposes that the baseline eGFR and baseline serum creatinine are known and represent a genuine steady state. Where a reliable baseline is lacking, which is often the case in acute illness, the estimate becomes uncertain. A reasonable estimate of the baseline serum creatinine can then be made using the KDIGO back-calculation from earlier values, but this introduces further uncertainty.

The maximum rise in creatinine per day (default 1.5 mg/dL/day) is an average value that does not suit everyone. With a large muscle mass, rhabdomyolysis or increased catabolism, the production rate may be higher, and with low muscle mass, advanced age or liver disease, lower. If the true value differs from 1.5, the keGFR will be systematically wrong, but the direction of the error depends on whether the creatinine is rising or falling.

keGFR does not apply to patients on dialysis or with anuria, since the formula then gives no meaningful information. Nor is it validated for children, pregnant women or patients with an extremely low or high body surface area. With large fluid shifts, common in intensive care, the serum creatinine is affected by haemodilution and haemoconcentration, and the keGFR may then give misleading values if fluid balance is not corrected for [2].

Finally, keGFR is an estimate of the average GFR over the chosen time interval, not a snapshot. If renal function changes rapidly within the interval, the estimate may lag behind reality.

References

1. Chen S. Retooling the creatinine clearance equation to estimate kinetic GFR when the plasma creatinine is changing acutely. J Am Soc Nephrol 2013;24(6):877–88. PMID: 23704286
2. Kwong YD, Chen S, Bouajram R et al. The value of kinetic glomerular filtration rate estimation on medication dosing in acute kidney injury. PLoS One 2019;14(11):e0225601. PMID: 31770424
3. Dinakar D, Chandan G, Sreedhara R et al. Kinetic estimated glomerular filtration rate and drug dosing in critically ill patients with acute kidney injury: a prospective observational study. Sci Prog 2025;108(1):478817. PMID: 39885773
4. Pelletier K, Lafrance JP, Roy L et al. Estimating glomerular filtration rate in patients with acute kidney injury: a prospective multicenter study of diagnostic accuracy. Nephrol Dial Transplant 2020;35(11):1886–93. PMID: 33151336
5. Lijović L, Pelajić S, Hawchar F et al. Diagnosing acute kidney injury ahead of time in critically ill septic patients using kinetic estimated glomerular filtration rate. J Crit Care 2023;75:154276. PMID: 36774818