DISTINGUISHING PHYSIOLOGIC VARIATION FROM ANALYTIC VARIATION.

WHEN REPEATED measurements of an individual’s blood chemistry or other physiologic variable are made over an extended period of time, the problem arises of separating long-term analytic or ‘laboratory’ deviation from the truly physiologic component of the total observed variation. In a normal person, the latter component represents deviations from homeostasis which may arise from short-term trends, cyclic changes, or simply transient events. This paper discusses the pros and cons of several experimental strategies for quantitatively estimating the physiologic component of variation in a series of results from a single individual. In addition, mathematical methods are proposed for inferring the statistical characteristics of this component over a population of individuals. This requires, of course, that repeated samples be taken from each of a set of N individuals, preferably a random sample from some pre-defined population. Both physiologic and analytic variation depend largely on the time-scale over which they are measured. In healthy persons, observations over a 5-yr period at 6-month intervals are likely to contain a higher proportion of physiologic variance than weekly measurements over 2-3 month. On the other hand, a body chemical with a strong circadian rhythm will exhibit less variation in weekly samples taken at a fixed time of day than in samples taken every few hours throughout one day. Analytic deviation. too, will have both short and long-term components. In terms of routine clinical chemistry operations, it is probably sufficient to divide analytic deviation into two general sub-components : (a) that associated with replicate determinations from the same sample and (b) longer term variation during the study period, due to variable laboratory conditions, even if the general method of analysis remains unchanged. The first sub-component can be estimated easily as long as the separate determinations are performed independently. It is the second sub-component that provides the greatest problem in interpreting repeated measurements of an individual’s physiologic state.