Effects of intra and interindividual variation on the appropriate use of normal ranges

Abstract

Normal ranges based on the distribution of single samples from a large number of individuals reflect both intra and interindividual variation. If the average ratio of these two sources of variation is small, then, assuming gaussian distributions, the conventional normal range will usually include a larger than expected proportion of an individual's distribution of values. When the average ratio exceeds 1.4, the normal range will include a proportion either larger or smaller than expected, depending on whether the individual's variability is less than or greater than average intraindividual variation. Investigation of multivariate normal regions in certain cases where calculations are feasible produced similar results. With these numerical guidelines, data from recent blood chemistry studies indicate that conventional normal ranges are likely to be less sensitive than desired to significant changes in an individual's biochemical state. This analysis supports the continued development and use of cumulative (in time) systems for reporting laboratory test results for individuals.