The trouble with reference ranges

Xchange newsletter - Summer 2007

Harmonising reference ranges within laboratory testing has long been a difficult area. Results from healthy individuals are provided alongside patient test results to aid assessment of whether a patient's results fall above, below or within the 'normal' range. However, there is no internationally agreed standard for reference ranges. 'Normal' cholesterol results may vary by 5% and there can be a ten-fold difference in troponin results. Here we talk to Dr Julian Barth, chair of the ACB's Reference Ranges Standardisation Project, about the challenge of standardising results across the country.

"It is important to consider the end user of these reference ranges and how they are using them," explains Julian. "Traditionally, they were generated for clinicians, but increasingly they are being interpreted by nurses and other healthcare professionals, pharmacists, patients and even insurance companies. These different users have a range of questions they want answering - Is the result normal? Is there anything wrong? Is it different from before? Perhaps we should consider whether different reference ranges are appropriate for different clients."

Current variation in reference ranges means that if a patient moves around the country, new test results may not correlate to their previous ones. In extreme cases, this could compromise patient management. Likewise, the trend towards networking laboratories also calls for reference ranges to be standardised across laboratories. It should be possible to compare test results and reference ranges regardless of where the test was performed. With electronic patient records in the offing, this will become even more important. Same unit reporting (mass units or molar units) would be one progressive step in this and the UK laboratory community has led the way for this to happen with therapeutic drugs and toxins.

There is so much variation in troponin testing for example, it would be impossible to have a single column on the electronic record to inform troponin results. "We are now in a situation where we really must move to common reference intervals," says Julian. "A number of different projects are underway to decide on how we can best achieve this."

About reference ranges

If reference ranges are to be standardised there are numerous factors that need to be considered:

What is the analyte?

A study of creatinine results by the International Measurement Evaluation Programme (IMEP) (www.irmm.jrc.be/imep) shows considerable bias in creatinine measurement. When reviewed by method type, there is still significant scatter.

What is the interval?

Usually, an interval of 95% is used although some suggest using an interval of 99% for screening purposes as the probability of detecting 'true' negatives remains high. See figure 1.

What is the population?

Many reference ranges are based on results from healthy blood donors but is it always appropriate to use such populations? Should we use homogenous or diverse populations for studies? Some advocate using data warehousing so that all results are kept and used as a representative sample of the population you are serving. This, however, wouldn't be commutable across the country.

What is the decision limit?

Some tests (such as those for diabetes, eGFR and thyroid disease) have decision limits that have been set by clinicians who may only have experience of one test without knowledge of variations. Does the decision limit suit all possible tests?