Background: The calibration of serum creatinine values to standardized creatinine and the commutability of serum creatinine across surveys are essential to the correct use of National Health and Nutrition Examination Survey (NHANES) data for kidney function and for generating estimates of the burden of kidney disease in the United States.
Study design: Calibration study of serum creatinine in NHANES III (1988-1994) and NHANES 1999-2000, 2001-2002, and 2003-2004 to directly compare creatinine measurements from the original surveys with standard creatinine measured using an assay traceable to known gold-standard methods. We also assessed predictors of differences between methods (potential interferences) in this general population.
Setting & participants: The NHANES are ongoing cross-sectional surveys of the civilian noninstitutionalized population of the United States. We selected random samples of approximately 200 stored specimens from persons aged 60 years or older from each survey (NHANES III, 1999-2000, 2001-2002, and 2003-2004).
Measurements: Stored serum specimens from the 4 NHANES surveys were analyzed for serum creatinine by using a Roche enzymatic assay implemented at the Cleveland Clinic Research Laboratory (CCRL). The Roche assay is traceable to gold-standard reference methods. The original NHANES serum creatinine values were obtained using the Jaffé method (kinetic alkaline picrate) implemented in several different laboratories.
Results: Overall agreement between the original NHANES values (Jaffé method) and CCRL measurements (Roche enzymatic) was high, but substantial biases were observed in NHANES III and 1999-2000. No bias was observed in NHANES 2001-2002 and 2003-2004. Final calibration equations to correct serum creatinine values in the relevant surveys are provided. Assay differences were independent of sex, race/ethnicity, and bilirubin and triglyceride levels, but weakly related to age and glucose concentration.
Limitations: We were not able to examine drift in measurements over time within each survey or directly evaluate freeze-thaw effects.
Conclusions: The magnitude of differences in serum creatinine measurements in NHANES III and 1999-2000 from standard creatinine would result in large differences in estimates of kidney function (10% to 20%). Thus, correction of original creatinine values in NHANES III and 1999-2000 is essential, but no correction is needed for NHANES 2001-2002 or 2003-2004.