Abstract
Reference ranges provide critical information to guide clinicians in their decision-making. Despite this, there appears to be little conformity in laboratory practice and there are few topics in clinical biochemistry that raise as many hackles as reference ranges. In fact, the International Federation of Clinical Chemistry (IFCC) has previously published a series of papers on its proposed methodology for sample collection and statistical analysis and in this issue of the Annals, Ceriotti et al. 1 propose ‘The way forward’. They discuss the classical IFCC approaches and the factors that need refining. There have, however, over the years been other methods employed for the development and derivation of reference ranges. These need to be included in the current debate and consideration should also be given to the way in which reference ranges are utilized by clinicians who often require decision-making values rather than ‘normal values’, for example when interpreting thyroid-stimulating hormone, cholesterol and glucose results.
The classical approach advocated by the IFCC and Ceriotti et al. demands the collection of a precisely defined cohort of subjects who have been sampled under exacting conditions. A minimum sample size of 120 subjects is proposed to ensure appropriate mathematical precision. There are many circumstances when this approach is essential as real differences occur for age, gender, ethnicity, etc. Indeed, a number of analytes have recently been shown to vary significantly between different cities on the eastern Pacific coast. 2 In view of this, even careful selection of appropriate cohorts can be problematic. Should separate reference populations be selected with healthy subjects for screening tests, and symptomatic subjects for unwell populations? The issue of ethnicity is becoming indistinct too as our multicultural societies have populations of mixed race with disease prevalences that do not necessarily reflect the original endogenous pattern. Further precision in subset design may be required for liver enzymes which vary in relation to abdominal obesity as measured by waist circumference. 3 Even after collecting a library of samples, how long will they be stable in storage for future studies?
An alternative method that has been advocated is data warehousing or the use of patient data that has been analysed and stored in laboratory computers. This method has the advantage of enormous data sets that extend over the entire age range. Moreover, it also accesses samples from populations that are typically difficult to reach, such as children. In addition, it excludes the bias introduced by the requirement for patient consent. The disadvantage is that case mix and preanalytical aspects are unknown and because there is ‘contamination’ by the effects of disease, there is uncertainty whether outlier removal protocols are appropriate. However, case-mix and preanalyticals reflect the local community and phlebotomy practice and therefore the samples that are received for diagnostic purposes. These two ill-defined factors of case-mix and phlebotomy practice are considered to be fatal flaws by Ceriotti et al. but I would argue are advantages. In recent issues of the Annals this technique has been valuably used to develop reference ranges for children 4 and across the entire age range for serum alkaline phosphatase. 5
A further approach is being developed in the UK inline with our National Health Service. Currently, each laboratory has its own analytical methods and reference ranges with variations between neighbouring laboratories. This was exposed by the introduction of the clinical guidelines for chronic kidney disease. The bias between creatinine methods, and therefore estimated glomerular filtration rate (eGFR), had the potential to distort access to clinical treatments by misclassifying patients in a way that was not compatible with a national service. The solution was the introduction of a method-related normalizing factor into the eGFR calculation. This highlighted the need for uniformity across all analytical services and the task has been grasped by the Pathology Harmony group. 6 This project is attempting a process of consensus between laboratory professionals to develop a common set of reference ranges and clinical protocols that can be used widely across the UK. The philosophy behind this scheme is pragmatic and assumes that the minor differences in methods between laboratories are of little clinical significance. The debate between the purists and pragmatists has to be tempered against the clinical risk created by a plethora of different reference ranges. Unsurprisingly, the NORIP project, which measured the routine blood analytes from 3035 subjects across Scandinavia to derive a common set of Scandinavian reference ranges, 7 has values that are not dissimilar to the Harmony ranges!
The Harmony project will not at present be able to bridge the analytical variations in immunoassay methods and we still need method-related ranges for analytes such as troponin 8 and tumour markers, but hopefully the laboratory community will soon be able to agree on international standards which will allow further progress in harmonization.
Finally, we should consider how blood tests are used by clinicians. Blood tests are performed for screening and diagnosis/disease monitoring. The approach advocated by Ceriotti et al. certainly gives the most precise ranges for a defined cohort of subjects and might be the best approach for ‘wellness screening’. Jorgensen et al. 9 have explored the concept of using 99th centile ranges for healthy subjects for whom disease exclusion is required. They suggest that 95th centile ranges should then be used for disease monitoring. This proposal has merit and should be explored. However, it might not be suitable to the warehousing methodology as 99th centile values would be most liable to errors introduced by erroneous outlier exclusion protocols.
In summary, there are many approaches to reference ranges none of which will satisfy both purists and pragmatists. However, the ultimate aim of the laboratory is to provide reference ranges which will help clinicians manage patients and we should do everything in our powers to help.