Dear Editor,

We thank Dr Walton for her insightful comments regarding our recent article “Biological variability of C-reactive protein and specific canine pancreatic lipase immunoreactivity in apparently healthy dogs”.[1]

We clearly indicated in our paper that the index of individuality was calculated as CVG/(CVI2 + CVA2)1/2, rather than (CVI2+CVA2)1/2/CVG as originally described in Harris's original publication.[2] The use of the reciprocal ratio CVG/(CVI2 + CVA2)1/2 addresses a semantic difficulty arising from Harris's original description, whereby data sets with high individuality (ie, with low intra-individual variation, CVI, and high inter-individual variation, CVG) have a numerically small value for the index of individuality.[3]

Fraser & Harris recognized this semantic difficulty in their 1989 publication, cited by both Dr Walton and us. The paragraph in Fraser & Harris’ paper immediately following that quoted by Dr Walton reads as follows:

“Use of this index [i.e., the index of individuality calculated as (CVI2+CVA2)1/2/CVG] has semantic difficulties in that,[sic] analytes with a high degree of individuality have, numerically, a low index. Adoption of the reciprocal index, namely SG/SA+I, would have advantages because analytes with a high degree of individuality would have a high index of individuality. The critical values for the utility of population based reference values in patient management would become SG/SA+I >1.7, population-based reference values are of quite limited usefulness; SG/SA+I <0.7, use of such reference values is appropriate.”[3]

Note that Fraser & Harris’ paper refers to variance, S, rather than coefficient of variation, CV, in this paragraph; however, elsewhere in their paper they also use coefficients of variation (ie, CVG/CVA+I) in reference to the calculation of index of individuality.

The higher degree of individuality of canine C-reactive protein compared to Spec cPL is readily appreciated by assessment of the figures in our paper. Note that individual dog data sets for C-reactive protein (Fig 1) tend to be more closely grouped within the individual, and in many cases all values for one individual dog show no overlap with values for other individual dogs. This higher degree of “individuality” is most apparent when comparing the data sets from dogs 1, 2, 4, and 6. By comparison, in the Spec cPL data (Fig 2), the data sets from most dogs overlap substantially, ie, each dog shows less “individuality”.

Early drafts of our paper featured one to two sentences explicitly stating that the calculation of index of individuality was in this reciprocal form originally proposed by Fraser & Harris. These were removed in response to a reviewer's suggestion. This appears to have had the unfortunate effect of potentially misleading the reader, and we apologize for any confusion this may have caused.


  1. Top of page
  2. References
  • 1
    Carney PC, Ruaux C, Suchodolski J, et al. Biological variability of C-reactive protein and specific canine pancreatic lipase immunoreactivity in apparently healthy dogs. JVIM 2011;25:825830.
  • 2
    Harris EK. Effects of intra- and interindividual variation on the appropriate use of normal ranges. Clin Chem 1974;20:15351542.
  • 3
    Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. CRC Critical reviews in clinical laboratory sciences 1989;27:409437.