Pitfalls in interpreting laboratory results

Editor, – I have read Dr Pat Phillips' article (Aust Prescr 2009;32:43-6) with interest. He points out that an individual's laboratory result may be abnormal for them, but still lie within the reference interval. This can occur when the individual's biological or 'intra-individual' variance is small compared with the 'inter-individual' or group variance. The 'index of individuality' which is the ratio of the intra-individual coefficient of variation (CV_i) to the group CV (CV_g) is used to estimate this variance. If the index is less than 0.6, the population-derived reference interval will not be of great use and the variable is said to show high individuality. If it is greater than 1.4 it should be useful.

The example used in the article on alkaline phosphatase is unfortunate, as this variable shows high individuality and the population reference interval is of limited value. For a variable such as ionised calcium, where the intra-individual variation is close to the inter-individual variation and therefore has a high index of individuality, it will be useful.

Another detail worth mentioning is that the appropriate CV for calculating the least significant difference is the combined intra-individual and analytical CV. This is obtained by squaring the respective CVs to obtain the variances, adding them and taking the square root to obtain the combined CV.

There is much published information on these sources of variation.^1,2 The possibility that individuals vary significantly in their intra-individual variances is recognised. Nevertheless taking these combined values into consideration can be helpful, as Dr Phillips shows, in interpreting successive laboratory results in patients on treatment.

John Masarei
Chemical Pathologist
Mount Pleasant, WA

Dr Pat Phillips, author of the article, comments:

I appreciate Dr Maserei identifying the 'index of individuality' as an objective way to tell when a test result may be within the relevant reference range (based on a group of people) but outside the individual's healthy range (which may be much narrower). This distinction can be clinically important. For example, a freeT4 may be within the laboratory range (that is, normal) but be biologically high for the individual and associated with a suppressed or increased thyroid stimulating hormone. This is the pathophysiology of the real clinical syndromes 'subclinical' hyper-and hypothyroidism.

Unfortunately, the only measure of result variability given by most laboratories is the laboratory reference range, which includes many components of variability as well as that occurring within one individual. In these situations, one has little choice and must interpret the individual result in the context of the general laboratory range.

However, when interpreting sequential results in one individual, one does not consider the laboratory reference range but the total variability within that individual (CV_i). I suggested that the least significant change should be considered a true signal of biological change over and above the background 'noise' of variability and is approximately 2CV_i.

The major point was that when interpreting laboratory results, one is trying to identify a clinical signal against the background variability. For single results the only information about the background variability is the laboratory reference range, but for sequential results the appropriate measure of variability is the variability within the individual and the least significant change.