OBJECTIVE
Biological variation consists of within-person (WP) and between-person (BP) variation. These components of biological variation are used to set analytical goals for imprecision and bias, evaluate serial changes for individual analytes, and assess the clinical utility of population-based reference intervals. Estimates of WP coefficients of variation (CVw) and BP coefficients of variation (CVg) for laboratory analytes were estimated from the 1999-2002 National Health and Nutrition Examination Survey (NHANES).
METHODS
NHANES is a survey of the noninstitutionalized civilian U.S. population that uses a stratified, multistage probability design to collect a nationally representative sample. Between- and within-person variations were estimated for 34 laboratory analytes, including general biochemical, nutritional, and environmental analytes. Between-person variation was estimated taking into account the complex sample design of NHANES by Taylor series linearization. For WP variation, a nonrandom sample was obtained with an average of 18.8 days (range: 3-51 days) between two analyte measurements. Data outliers were excluded using Tukey's method to obtain more stable estimates of variation.
RESULTS
The BP and WP variations were estimated on as many as 18,761 and 853 sample persons, respectively. When compared with the BP sample, the WP sample was older (mean age: 39 compared with 30 years) and had more non-Hispanic white (45% compared with 37%) and fewer Mexican-American (19% compared with 30%) persons. There was no statistically significant difference in gender proportions between the BP and WP samples. Serum sodium had the lowest CVg (1.8%) and the lowest CVw (1.0%). The index of individuality (CVw/CVg) ranged from 0.20 for blood lead to 0.76 for serum iron. The CVg exceeded the analytical method coefficient of variation for all analytes. Within-person variation was also compared between males and females, and several analytes revealed significant differences (p < 0.01). Serum ferritin had the greatest difference for CVw (males, 17.9% compared with females, 28.8%).
[1]
C G Fraser,et al.
Generation and application of data on biological variation in clinical chemistry.
,
1989,
Critical reviews in clinical laboratory sciences.
[2]
K. Wolter.
Introduction to Variance Estimation
,
1985
.
[3]
John W. Tukey,et al.
Exploratory Data Analysis.
,
1979
.
[4]
Callum G. Fraser,et al.
Biological Variation: From Principles to Practice
,
2001
.
[5]
C Ricós,et al.
Current databases on biological variation: pros, cons and progress.
,
1999,
Scandinavian journal of clinical and laboratory investigation.
[6]
Steve P. Verrill,et al.
Confidence Bounds and Hypothesis Tests for Normal Distribution Coefficients of Variation
,
2007
.
[7]
M. Borel,et al.
Day-to-day variation in iron-status indices in healthy men and women.
,
1991,
The American journal of clinical nutrition.
[8]
Edward R. Ashwood,et al.
Tietz Textbook of Clinical Chemistry and Molecular Diagnostics
,
2005
.
[9]
C G Fraser,et al.
Biological variation in clinical chemistry. An update: collated data, 1988-1991.
,
1992,
Archives of pathology & laboratory medicine.
[10]
M. Carroll,et al.
Estimate of biological variation of laboratory analytes based on the third national health and nutrition examination survey.
,
2005,
Clinical chemistry.
[11]
E. K. Harris.
Effects of intra- and interindividual variation on the appropriate use of normal ranges.
,
1974,
Clinical chemistry.
[12]
C. Fraser.
Inherent biological variation and reference values
,
2004,
Clinical chemistry and laboratory medicine.