All components measured from body fluids (such as blood or urine) have normal, physiological fluctuations. These fluctuations or variations have been recognized since at least 1970 for human clinical pathology and are known as biological variation. The first veterinary reports of biological variation of clinical pathology measurands were in the late 1980’s with the early 1990’s seeing a large number of studies from the University of Copenhagen, primarily for dogs but also cattle. There are now over 40 studies relating to veterinary clinical pathology biological variation across 11 species (a full list can be seen on the Citations page). These studies have shown that biological variation varies across different species so even more work in this area is required.
How is biological variation determined?
Essentially, samples are collected on repeated occasions from a number of individuals. Ideally, these samples are run at the same time in duplicate. For example, 20 ferrets (of known health status) might be tested weekly for 8 weeks, the samples are stored/frozen (if appropriate). Typically, after the end of the sampling, all the samples are tested at the same time in duplicate. The amount of variation that occurs within-individuals (CVI), between-individuals (CVG) and attributable to the analyser, analytical variation (CVA) is determined by analysis of variance (ANOVA). The full details of what is required and how to determine these results are on the Study Guidelines pages of this website.
Biological variation data can be used to determine the clinical significance of two consecutive results from an individual. This is because, by knowing the amount of physiological change, we can see if there is a change beyond normal physiology. The amount of allowable change is called a Reference Change Value (RCV) and beyond this, a change is considered medically significant. This medical significance is typically calculated to a 95% probability. The appropriateness of using RCV (as opposed to traditional, population-based reference intervals) is determined by each measurand’s Index of Individuality (II). When measurands have a high amount of individuality, a medically significant change may occur that remains within a population-based reference interval. Other measurands are very tightly controlled (such as electrolytes and albumin) and remain similar across a species (and sometimes between species); for these, correctly determined population-based reference intervals are appropriate. Traditionally, for human clinical pathology, II is calculated such that a lower value indicates greater variability. Within the veterinary literature, the recent trend is to use the inverse formula such that (more intuitively), a higher value indicates greater individuality (and appropriateness to use RCV). This website continues this trend. We have a link to a RCV calculator that is hosted by VIN.
We can use the knowledge of how much physiological variation occurs to set standards of how much analyser variation should be allowable. For example, measurands with low physiologic variation will have any variation ‘drowned out’ if there is too much analyser variation. It has been demonstrated that analyser variability less than or equal to half of the amount of within-individual variability (CVA < 0.5 x CVI) results in CVA contributing to no greater than 12% of the total variability; this standard is known as ‘desirable imprecision’. Desirable bias is determined as 0.25 x √(CVI2 + CVG2) and if this standard is reached, then results can be compared between different analysers. It is appropriate to apply ‘minimum’ standards for both imprecision and bias when technology cannot reach desirable quality standards for closely regulated measurands (such as electrolytes or albumin). Conversely, optimum quality standards can be applied for poorly regulated analytes (such as bilirubin).