Current and emerging concepts in biological and analytical variation applied in clinical practice

Abstract A single laboratory result actually represents a range of possible values, and a given laboratory result is impacted not just by the presence or absence of disease, but also by biological variation of the measurand in question and analytical variation of the equipment used to make the measurement. Biological variation refers to variability in measurand concentration or activity around a homeostatic set point. Knowledge of biological and analytical variation can be used to facilitate interpretation of patient clinicopathologic data and is particularly useful for interpreting serial patient data and data at or near reference limits or clinical decision thresholds. Understanding how biological and analytical variation impact laboratory results is of increasing importance, because veterinarians evaluate serial data from individual patients, interpret data from multiple testing sites, and use expert consensus guidelines that include decision thresholds for clinicopathologic data interpretation. The purpose of our report is to review current and emerging concepts in biological and analytical variation and discuss how biological and analytical variation data can be used to facilitate clinicopathologic data interpretation. Inclusion of veterinary clinical pathologists having expertise in laboratory quality management and biological variation on research teams and veterinary practice guideline development teams is recommended, to ensure that various considerations for clinicopathologic data interpretation are addressed.


Individuality and HSP calculation examples
Three studies have assessed BV of alkaline phosphatase (ALP) in cats. [41][42][43] All three found CV G to be approximately 3-fold greater than CV I , which means variation between individuals is substantially greater than variation within individuals and that a pRI for feline ALP is likely to be diagnostically insensitive for detecting medically important change within an individual cat. High individuality of feline ALP can be shown mathematically by the index of individuality

Reference change value, CD, and iRI calculation examples
Using data from 4 years of annual wellness testing, an individual cat's ALP HSP (the mean ALP value calculated from historical wellness data) is 36 U/L. Reference change value for ALP is 31%, as calculated using median feline CV I for ALP. 32 As above, CV A is imprecision of the analyzer being used for these assessments, as determined using historical control data. Reference change value 31% means that an increase in ALP activity greater than, or a decrease less than, 31% from 36 U/L is considered medically significant.
Critical difference is determined by multiplying the HSP by the RCV, or in this example, 36 × 0.31 = 11 U/L. A CD of 11 U/L means an increase in ALP activity greater than, or a decrease less than, 11 U/L from 36 U/L is considered medically significant for this cat only.
Individualized reference intervals can be calculated for this cat by using CD and HSP to determine a range. Alkaline phosphatase activities outside this range are considered abnormal for this cat only. In this example, 36 U/L ± 11 U/L = iRI 24 to 47 U/L.
Because ALP in cats is a highly individual analyte (with II approximately 3), iRI for a given individual cat is expected to be considerably narrower than any pRI. Because iRI is calculated from an individual cat's HSP, the resulting range applies only to that individual cat and should not be applied to other cats. In contrast to ALP and other highly individual analytes, iRI of analytes with low individuality (low II) is expected to approximate pRI more closely.

Dispersion calculation examples
Dispersion around a measured result is influenced by imprecision of the analyzer (CV A ) and by the number of samples measured and used to estimate the patient's concentration. In general, (1) for an analyte with a given CV I , dispersion is expected to increase as assay CV A increases and (2) for assays with similar CV A , estimating patient concentration from multiple samples is most beneficial for analytes with larger CV I . F I G U R E A 1 Serum or plasma creatinine concentration dispersion (mg/dL) and IRIS CKD decision thresholds in cats. Green zones indicate serum or plasma creatinine concentrations for which confidence in correct patient classification is highest (at least 95% statistical probability). Gray zones indicate serum or plasma creatinine concentrations for which confidence in correct patient classification is less, because dispersion of creatinine measurement means overlap of results for patients at or near decision thresholds is possible F I G U R E A 2 Serum or plasma creatinine concentration dispersion (mmol/L) and IRIS CKD decision thresholds in cats. Green zones indicate serum or plasma creatinine concentrations for which confidence in correct patient classification is highest (at least 95% statistical probability). Gray zones indicate serum or plasma creatinine concentrations for which confidence in correct patient classification is less, because dispersion of creatinine measurement means overlap of results for patients at or near decision thresholds is possible