 |
 |
Lynda A. Szczech, MD,
MSCE
Published:
Medscape 05/28/2009
Ann Intern Med. 2009;150:604-612
Levey AS,
Stevens LA, Schmid CH, et al
Creatinine has been, is now, and
will be for the foreseeable future our easiest way to monitor kidney function.
However, the inadequacies and limitations of creatinine as such a marker are
well described. Due to the nonlinear relationship between creatinine and kidney
function, as well as the effect of a patient's muscle mass on the association
between absolute value of creatinine and kidney function, formulas to
approximate kidney function using serum creatinine and proxies for muscle mass
are essential.
The Cockcroft-Gault equation was
the first widely used formula.[1] Although creatinine clearance
using this equation is relatively simple to calculate, it was recognized to
overestimate kidney function, and many have raised concerns on the basis of its
limited generalizability. Investigators from the Modification of Diet in Renal
Disease (MDRD) study subsequently derived a number of formulas to approximate
kidney function using that study that performed better than the Cockcroft-Gault
formula.[2,3] The formula has been widely
applied in both clinical care and research since its publication, but clinicians
continue to discuss whether it can be used for patients with clearance of less
than 60 mL/min and without diabetes mellitus.
Given these continued
generalizability limitations, Levey and colleagues collaborated with multiple
investigators to obtain data from studies in which the glomerular filtration
rate (GFR) was measured using exogenous filtration markers and serum creatinine.
Ten studies were available and were randomly divided into 2 groups for
development of the new estimation equation and the internal validation of the
new equation. At the Cleveland Clinical Laboratory, serum creatinine values were
recalibrated to standardized creatinine measurements using the Roche enzymatic
method. The equation was developed with potential clinical predictor variables
of serum creatinine, age, race (black vs white and other), and sex. Additional
variables available in some models include diabetes mellitus, previous organ
transplantation, and weight.
The dataset used for development of
the equation contained 5504 subjects. The mean serum creatinine was 1.65 mg/dL
with a mean GFR measured at 68 mL/min, providing a robust representation of
patients with milder kidney disease than was possible with the MDRD study.
Approximately 29% of participants had diabetes mellitus and 32% were African
American. The mean age was 47 years with limited representation of the elderly
(9% between 66 and 70 years, and 3% > 71 years).
The Chronic Kidney Disease
Epidemiology Collaboration (CKD-EPI) formula developed is represented as the
equation below, in which the values of the constants of a, b,
and c vary on the basis of race, sex, and serum
creatinine.
GFR = a × (serum
creatinine/b) c × (0.993)age
The variable a takes on
the following values on the basis of race and sex:
The variable b takes on
the following values on the basis of sex:
The variable c takes on
the following values on the basis of sex and creatinine
measurement:
The investigators compared the
abilities to identify patients with various degrees of kidney disease of the new
CKD-EPI formula with the MDRD formula. Although both formulas performed
similarly well, in which subjects were misclassified in terms of category of
kidney disease, the CKD-EPI formula was more often correct (P <
.001). In the separate subset of patients designated to validate the new
formula, the CKD-EPI formula demonstrated less bias and improved precision and
greater accuracy than the MDRD formula (P < .001), particularly for
those patients with an estimated GFR > 60 mL/min. The median difference
between measured and estimated GFR was 2.5 mL/min for the CKD-EPI formula and
5.5 mL/min for the MDRD formula (P < .001).
Because of the reduced error of
overestimation of kidney function, the CKD-EPI equation yielded a lower
estimated prevalence of kidney disease in the National Health and Nutrition
Examination Survey (NHANES) dataset. Where the MDRD equation estimated the
prevalence of kidney disease at 13.1%, the CKD-EPI formula estimated it slightly
lower at a prevalence of 11.5%. This limited extent to which the equation
overestimates GFR reflects the major advantage of this newer
formula.
The use of serum creatinine is a cheap and easy way to estimate kidney function. Although the manner in which to transform serum creatinine into an accurate measure of absolute kidney function remains a problem, this analysis takes use 1 step closer. Researchers anticipate that it will be quickly adopted in the clinical community to allow an even more accurate equation of kidney function.
[ CLOSE
WINDOW ]
1.
Cockcroft DW, Gault MH. Prediction
of creatinine clearance from serum creatinine. Nephron. 1976;16:31-41.
2.
Levey AS, Greene T, Kusek J, et al.
A simplified equation to predict GFR from serum creatinine. J Am Soc Nephrol.
2000;11:155A. Abstract.