Separate estimation of biological and analytical variance components when quantities and reagents are unstable

A model for reliable estimation of variance components for biological within- and between-subject variation as well as for analytical variation when both the quantity and the reagents are unstable has been established. This model was applied to the alpha-interferon receptor on human leucocytes which involves two major problems. First, the receptor has to be quantified within a few hours after blood sampling, and second, the reagents for the measurement procedure must be used within 2 weeks. For the number of receptors per cell the biological estimates of coefficients of variation were 14% for within-subject variation and 20% for between-subject variation, respectively. For the dissociation constant both estimates were zero as expected. The model is robust and applicable to other systems with unstable quantities and reagents.