Within-person variance in biochemical indicators of iron status: effects on prevalence estimates.

The effect of within-person variance on prevalence estimates from population distributions based on a single measurement was examined for hematologic and iron-status indicators collected in the Hispanic Health and Nutrition Examination Survey (HHANES). Within-person to between-person variance ratios (W:Bs) were estimated for 11 indicators by using data from 80 persons who provided blood twice in HHANES. Distributions of selected iron-status indicators from 20-44-y-old Mexican American females from HHANES were adjusted for within-person variance by using the W:B estimates, and prevalences of low values based on the original and adjusted distributions were compared. W:B were less than 1 for the majority of the indicators. Nonetheless, prevalences of low values from the original distributions were inflated when commonly used cutoff values were applied. Within-person variance in serum analytes needs to be controlled when the prevalence of a condition in populations is assessed.