Choosing Creatinine Clearance, eGFR, or Cystatin C for Dosing

Why the Renal-Function Estimate Matters

A dosing model may use renal function as a covariate for drug clearance, but the term “renal function” can refer to different estimates with different definitions and units. Cockcroft–Gault estimates creatinine clearance; commonly reported eGFR equations estimate glomerular filtration rate (GFR), often indexed to a standard body surface area (BSA); cystatin C provides another filtration marker. These estimates are related, but they are not interchangeable measurements.

The choice matters because a model’s renal covariate is meaningful only in the form in which it was developed or validated. Substituting one estimate for another, or using an indexed value where an absolute value is expected, can change the model input even if the laboratory result itself has not changed. The earlier material on renal function covariates in clearance models provides the broader PK context; here the focus is on selecting and interpreting the renal-function estimate.

Definition: Creatinine clearance is the volume of plasma cleared of creatinine per unit time. Cockcroft–Gault provides an estimate of creatinine clearance, usually in mL/min. Estimated GFR (eGFR) is an estimate of filtration rate calculated from filtration markers and other patient characteristics. Many laboratory eGFR results are reported in mL/min/1.73 m², meaning the estimate is normalized to a standard BSA. A cystatin C-based eGFR uses cystatin C, alone or together with creatinine, in a validated estimating equation.

What Each Estimate Represents

Cockcroft–Gault creatinine clearance. The conventional Cockcroft–Gault equation estimates creatinine clearance from age, body weight, serum creatinine, and a sex factor. With age in years, weight in kg, and serum creatinine in mg/dL, the result is in mL/min:

\text{CrCl}_{\text{CG}} = \frac{(140-\text{age})\times \text{weight}}{72\times S_{\text{cr}}} \times \begin{cases} 1, & \text{male}\\ 0.85, & \text{female} \end{cases}

This is an estimate, not a direct measurement. Its original formulation assumes a reasonably stable serum creatinine and uses body weight as an input. The weight descriptor used in a particular clinical or modeling context therefore matters, especially when body size or composition is unusual. Do not silently change the weight convention if the drug label or model specifies one.

Creatinine-based eGFR. A laboratory typically calculates eGFR with a validated equation using serum creatinine, age, and sex. The equation and assay matter, and the reported value is commonly indexed to 1.73 m². Creatinine is influenced by muscle mass, diet, and tubular secretion, as well as filtration. Thus, a creatinine-based eGFR is not a direct measurement of GFR and may be misleading when creatinine generation is atypical.

Cystatin C-based estimates. Cystatin C is a filtration marker that is less dependent on muscle mass than creatinine. Equations may estimate GFR from cystatin C alone or combine it with creatinine. Cystatin C is not free of nonrenal influences: factors such as glucocorticoid use, thyroid dysfunction, inflammation, smoking, and adiposity can affect its concentration. A combined estimate can be useful when creatinine may be unreliable, but it remains an estimate with its own assumptions and sources of error.

Units and Body-Surface-Area Indexing

An indexed eGFR is expressed relative to a standard BSA; it is not the patient’s absolute filtration rate in mL/min. If an application calls for an absolute value and the relevant label or model permits use of eGFR, de-indexing scales the reported estimate by the patient’s BSA:

\text{eGFR}_{\text{absolute}}\;(\text{mL/min}) = \text{eGFR}_{\text{indexed}}\;(\text{mL/min/1.73 m}^2) \times \frac{\text{BSA}\;(\text{m}^2)}{1.73\;\text{m}^2}

For example, a patient with BSA below 1.73 m² will have an absolute eGFR below the indexed number, while a patient with BSA above 1.73 m² will have an absolute eGFR above it. This conversion changes the scale and units; it does not improve the underlying estimate’s accuracy. One commonly used BSA approximation is the Mosteller formula, with height in cm and weight in kg:

\text{BSA}\;(\text{m}^2)= \sqrt{\frac{\text{height}\;(\text{cm})\times\text{weight}\;(\text{kg})}{3600}}

Cockcroft–Gault is already expressed in mL/min and is not indexed to 1.73 m². Do not multiply or divide it by BSA merely to make it resemble an indexed eGFR. Conversely, do not treat an indexed eGFR as an absolute mL/min value without checking what the dosing application requires.

Worked Example: Calculating Cockcroft–Gault Creatinine Clearance

Consider a hypothetical 68-year-old male weighing 72 kg with serum creatinine of 1.20 mg/dL. Assume serum creatinine is stable and use the conventional Cockcroft–Gault equation.

Substitute the values.

\text{CrCl}_{\text{CG}} = \frac{(140-68)\times72}{72\times1.20} = \frac{5184}{86.4} = 60.0\ \text{mL/min}

The estimated creatinine clearance is 60.0 mL/min. This value is an estimate based on the equation and its inputs, not a measured GFR. If a drug’s dosing guidance or a validated model specifically uses Cockcroft–Gault, this is the type of estimate to calculate, with the required weight convention checked in advance.

Worked Example: De-Indexing a Laboratory eGFR

A hypothetical adult is 170 cm tall and weighs 70 kg. The laboratory reports a creatinine-based eGFR of 42.0 mL/min/1.73 m². Calculate BSA using Mosteller, then calculate absolute eGFR.

First calculate BSA.

\text{BSA} = \sqrt{\frac{170\times70}{3600}} = \sqrt{3.3056} = 1.818\ \text{m}^2

Then de-index the eGFR.

\text{eGFR}_{\text{absolute}} = 42.0\times\frac{1.818}{1.73} = 44.1\ \text{mL/min}

The indexed laboratory value is 42.0 mL/min/1.73 m²; the calculated absolute value is about 44.1 mL/min. These are different representations of an estimate, not two independent measurements. Use the absolute value only when the dosing context calls for that scale.

Matching the Estimate to a Dosing Model

The primary selection rule is to use the renal-function definition, equation, units, and indexing convention that match the drug label or the model’s development and validation. If a model was fit using Cockcroft–Gault CrCl in mL/min, entering indexed eGFR without conversion changes the covariate and its interpretation. Even after de-indexing, eGFR is not automatically equivalent to Cockcroft–Gault CrCl: the underlying markers, equations, and physiologic quantities differ.

A model may use renal function directly as a covariate, or it may use a normalized quantity relative to a reference value. In either case, the covariate’s definition is part of the model specification. For example, if the model uses \(R=\text{CrCl}/\text{CrCl}_{\text{ref}}\), both values must be defined on the same scale and in the same units. Replacing CrCl with eGFR changes the input unless the model has been shown to support that substitution.

When the label specifies a particular estimator, follow that specification rather than choosing whichever number appears most clinically plausible. If the model documentation is unclear, resolve the ambiguity before using the output for a dosing decision. A recorded value without its equation, units, indexing status, and relevant input conventions is difficult to reproduce and may not be suitable for model application.

Worked Example: Investigating Discordant Estimates

A hypothetical 80-year-old adult weighs 55 kg and has serum creatinine of 0.60 mg/dL. The laboratory reports creatinine-based eGFR of 78 mL/min/1.73 m², cystatin C-based eGFR of 39 mL/min/1.73 m², and a combined creatinine–cystatin C eGFR of 54 mL/min/1.73 m². The patient’s BSA is 1.60 m². A dosing model’s documentation specifies Cockcroft–Gault CrCl.

Calculate the specified covariate. Assuming stable serum creatinine, and using the female factor of 0.85 for this hypothetical patient:

\text{CrCl}_{\text{CG}} = \frac{(140-80)\times55}{72\times0.60}\times0.85 = \frac{3300}{43.2}\times0.85 = 64.9\ \text{mL/min}

The model-specified input is therefore 64.9 mL/min under the stated assumptions. For comparison, de-indexing the combined eGFR gives \(54\times(1.60/1.73)=49.9\) mL/min. The difference does not establish that either estimate is “the true” value: the measures use different markers and equations, and the model explicitly specifies Cockcroft–Gault.

The discrepancy warrants a clinical and data check rather than averaging the values or substituting one without justification. Low muscle mass could make creatinine-based estimates appear higher than filtration-based estimates would suggest; cystatin C can also be affected by nonrenal factors. Confirm the units, equation, assay context, stability of renal function, and relevant patient characteristics. If uncertainty remains clinically important, the prescriber or clinical pharmacist may need to consider an appropriate confirmatory assessment and the limits of applying the model.

Assumptions, Diagnostics, and Alternatives

These estimates are most interpretable when renal function is sufficiently stable for serum-marker-based equations to apply. During rapidly changing kidney function, serum creatinine may lag behind the change in filtration; a value calculated from it can therefore misrepresent current function. The same caution applies to interpreting cystatin C estimates as instantaneous measurements. In unusual body composition, severe illness, or other settings where marker generation or concentration is altered, question whether the equation’s assumptions are reasonable.

Before applying a dosing model, check the estimate’s provenance and inspect for avoidable errors. Useful checks include:

Confirm the serum creatinine units, age, sex factor where relevant, weight, and weight descriptor used in Cockcroft–Gault.
Confirm whether a laboratory eGFR is indexed and whether the label or model requires indexed or absolute units.
Check that the patient’s height and weight used for BSA are plausible and measured in the stated units.
Review whether creatinine is changing, muscle mass is atypical, or cystatin C may have important nonrenal influences.
Investigate large disagreements among estimates rather than choosing the highest, lowest, or average value by default.

Where a more direct assessment is needed, measured GFR or measured urinary creatinine clearance may be considered in an appropriate clinical setting. These approaches have their own practical and measurement limitations; they are alternatives to consider, not automatic replacements for a model’s validated input. The appropriate choice depends on the clinical question and the evidence supporting the dosing method.

Common pitfall: “The laboratory reports eGFR in mL/min, so it can replace Cockcroft–Gault CrCl in a model.” This reasoning overlooks both the estimate’s definition and possible BSA indexing. Check the exact units and equation first; convert indexed eGFR to absolute units only when appropriate, and do not treat the converted eGFR as equivalent to Cockcroft–Gault unless the model or dosing guidance supports that substitution.

Why the Distinction Matters

A renal estimate is not merely a number to insert into a dosing calculation. Its definition determines what that number represents, and its units determine the scale on which a model interprets it. Confusing CrCl with GFR, or indexed with absolute eGFR, can create a mismatch between the patient’s estimate and the model’s covariate definition. Careful selection does not eliminate estimation error, but it makes the input consistent with the evidence on which the dosing recommendation depends.

The estimates also differ in the patient characteristics that can undermine them. Creatinine-based estimates require attention to muscle mass and steady-state assumptions; cystatin C offers a useful complementary marker but can be influenced by nonrenal factors. Discordance is therefore a diagnostic signal to examine context and inputs, not permission to select whichever value produces a preferred recommendation.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions below.

What is the key distinction between Cockcroft–Gault CrCl and a laboratory eGFR?
A laboratory reports eGFR as 60 mL/min/1.73 m² for a patient with BSA 1.50 m². What is the corresponding absolute eGFR, if de-indexing is appropriate?
Why should indexed eGFR not automatically be entered into a model specified for CrCl in mL/min?
Name one reason creatinine-based estimates may be unreliable and one nonrenal influence on cystatin C.
A patient’s creatinine- and cystatin C-based estimates disagree substantially. What is a better response than averaging them?

Solutions. (1) Cockcroft–Gault estimates creatinine clearance, while eGFR equations estimate GFR; they are related but not interchangeable quantities. (2) \(60\times(1.50/1.73)=52.0\) mL/min, to three significant figures. (3) The indexed eGFR is normalized to 1.73 m², and it is also based on a different equation and filtration marker; both scale and definition may mismatch the model. (4) Low or unusual muscle mass can distort creatinine-based estimates; glucocorticoid use, thyroid dysfunction, inflammation, smoking, or adiposity can influence cystatin C. (5) Check units, equations, inputs, renal-function stability, and patient factors; then follow the model or label’s specified measure and seek appropriate clinical review if the uncertainty matters.