Skip to main navigation
Skip notifications
Access to POEMs and Essential Evidence Plus will no longer be included in CMA membership as of Dec. 1, 2023.
Search
Laboratory results are not an exact science
Clinical Question
How much variation is there in clinical laboratory values, and how should we account for it?
Bottom line
"The numbers don't lie." How many times have you said that to yourself or to a patient? Although numbers may not lie, they may not tell the whole truth. For example, a single HbA1c test result of 6.3% (45 mmol/L) could actually be as low as 5.5% (39 mmol/mol) or as high as 7.1% (51 mmol/mol). This range is due to the inherent variability in the analytic process (4.3%), as well as to biologic variability (ie, the variation in the same person over the course of days caused by physiologic changes). As a result, it's hard to tell if a change in a blood level over time is real or just a reflection of these sources of variability. Bookmark this website: www.bmj.com/content/368/bmj.m149 — it's a tool that determines the variability that can be expected in a single laboratory value and calculates the reference change value needed to signify a real difference between the values in the same person over time. 5
Reference
Study design: Other
Funding: Self-funded or unfunded
Setting: Not applicable
Synopsis
Just how precise are those test results we read about all the time? Each particular assay has inherent analytical variation, ranging from less than 2% for a bone density test to up to 20% for a single alkaline phosphatase measure. However, the larger contributor to variation in laboratory results is biological variation, the fluctuation in individual patients over time due to physiologic changes. Combined, these challenges to precision can make a single iron, bilirubin, or triglyceride level be inaccurate by as much as 50%. When testing the same person over time, the inherent variation can require at least a 50% change in the levels to be considered valid (that is, outside of the limit of biologic and analytical variability). Fortunately, the authors of this study have created a website that graphs the effect of these variations on many laboratory tests. The website determines the "reference change value," which is the difference that must be seen between 2 test results for the change to be considered "real" and not due to assay variability.
Reviewer
Allen F. Shaughnessy, PharmD, MMedEd
Professor of Family Medicine
Tufts University
Boston, MA
Comments
Variability in annual cholesterol testing
Such an important point... still doing annual cholesterol tests? commenting on perceived up or down changes... I was struck by the comprehensive and 100% Canadian review of cholesterol testing inaccuracy by our friends at Best Science who explained that the 7% variability in cholesterol results, test to test meant there was more variability than possible meaningful change in annual cholesterol blood work... yet another reason not to do it... for a great review see: https://therapeuticseducation.org/podcast/episode-200-please-stop-measu…