Usefulness of reference limits and evaluation of significant differences. An example of the biological variation of serum rheumatoid factors.

To establish analytical quality standards based on biological variations and to evaluate the usefulness of reference values for rheumatoid factors (RF), analytical and biological variations in samples from 34 volunteers between 23 and 46 years old were estimated at monthly intervals for six months. The samples were processed in duplicate on a Hitachi 717 automatic analyzer using the manufacturer's reagents. The within- and between-subject coefficients of variation associated with RF were 8.5% and 24.5% respectively, and the individuality index was 0.45, indicating the high individuality of this parameter. The analytical standards of quality estimated as coefficients of variation (%) were: imprecision, 4.2% and inaccuracy, 6.5%. The heterogeneity index of the within-subject variances was 0.17, thus not rejecting the homogeneity hypothesis. The RF may be considered useful for monitoring purposes because of their high individuality and relatively small critical difference (25% for alpha = 0.10). However, when RF levels are normal it is difficult to interpret them because of their high between-subject variability.