We professionals in laboratory medicine agonize over provision of appropriate population-based reference intervals (RI). Good quality RI, in addition to flags on results triggered by their limits, are required by laboratory accreditation bodies and standards such as ISO 15189. And we do know that many clinical users of laboratory results do not necessarily understand the subtleties of what we call RI, partitioned when necessary by age, sex, and other factors, and often use what they tend to term “normal ranges” from a variety of sources as definitive criteria of health or disease. Although the well-known, authoritative guidelines from the Clinical Laboratory Standards Institute (CLSI) and the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) assist us to develop RI in which we have considerable confidence, there are still a number of major problems with RI, even if partitioned to take significant affecting factors into account.
One of the most important problems is that, for most analyses in laboratory medicine, the within-subject biological variation is less than the between-subject variation. Consequently, individuals have a range of values that span only a part of the RI. They can have important changes in results when these all lie within the RI. We would usually consider such changes as unremarkable, and so would our users: in consequence, important clinical information might be missed. In addition, results can change from inside to outside the RI (and vice versa) without having any clinical importance although we flag all of those outside the RI.
The journal Clinical Chemistry and Laboratory Medicine celebrates its 50th anniversary this year. The first issue contains an article titled: The theory of reference values: an unfinished symphony. This can be downloaded free of charge at www.degruyter.com/view/j/cclm. The many important unanswered questions about RI are discussed in what we think is essential reading for all in laboratory medicine. Alternatives to traditional RI are also discussed. Interestingly, it is stated that, wherever it will apply, individual reference values and reference change values (RCV) have their place. Individual RI have not been much used and are difficult to construct, but, since by far and away the majority of investigations in laboratory medicine are performed for monitoring rather than diagnosis, we should all be applying RCV much more in our laboratories in the objective assessment of the significance of changes in serial results from individuals.
The strategies to derive RCV have been very well documented, as described in Clin Chem 2011;57:1635-7: the calculation requires:
- a decision on the probability that the difference is to be detected and whether this is change (two-tailed) or rise or fall (one-tailed),
- information on analytical imprecision – which is known by every laboratory for every analysis, and
- data on within-subject biological variation.
It is unrealistic to suggest that laboratories derive their own data on biological variation for every test in the repertoire, and we are fortunate that databases of published values are available, particularly those of Carmen Ricos and colleagues, which are regularly updated at www.westgard.com. However, we should always remember that the values documented are medians of the data found in the publications accepted for inclusion in the derivation of estimates in the database. Moreover, although this database is widely used and cited, as it should be, the data do have a number of important limitations, and the need for a standardized approach to derivation and publication of information on the components of biological variation is becoming increasingly recognised.
Thus, although we strongly advocate the use of RCV and encourage flagging of significant changes in serial laboratory results from individuals, as we have done in our own laboratories since the last millennium, we encourage those developing RCV to go back and asses the source publications in order to determine the applicability of the data to their own situation, analytical systems, and population, so as to get the most appropriate RCV. Often, it will be found that published studies on biological variation are somewhat deficient. However, similar to the current widely-used guidelines such as STARD, STROBE, and PRISMA, the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) Working Group on Biological Variation (http://efcclm.eu/science/wg-biological-variation) is well under way in developing a check-list to enable assessment of suitability of data for publication, assist in the development of new studies for the production of reliable estimates of biological variation following well-defined protocols, and assess the accuracy of existing published work (http://www.biologicalvariation.com/Tools.html). We look forward to its early publication and hope that you will apply it in your work to improve the interpretation of laboratory test results.
Recommended reading: Siest G, et al. The theory of reference values: an unfinished symphony. Clin Chem Lab Med. 2013;51:47-64.