Getty Images / Jason Butcher
A new statistical model based on estimates of within-subject biological variation, analytical variation, and previous test results simplifies the process for clinical labs to report personalized reference intervals. Investigators detailed their findings in Clinical Chemistry.
Labs could use the model to report both population-based reference intervals and personalized intervals to clinicians, Abdurrahman Coşkun, MD, professor of clinical biochemistry at Acibadem Mehmet Ali Aydinlar University School of Medicine in Istanbul, Turkey, and the study’s corresponding author, told CLN Stat. “Thus, clinicians can see the difference between population-based reference intervals and personalized reference intervals. This will help them to manage patients more effectively.”
Physicians need reference data to interpret test results correctly, but this data isn’t typically available. Instead, physicians compare the individuals’ test results with population-based reference intervals, which don’t necessarily represent all individuals correctly, said Coşkun. “To aid in better interpretation of laboratory tests to improve patient outcome and safety, patients’ test results should be compared with their own personal reference intervals.”
The lack of personalized reference intervals worldwide provided the impetus for this study.
The personalized reference interval of a test has a set point with an upper and lower limit. A test result located within this personalized threshold is considered normal. Increasing the number of previous measurements increases the accuracy of the set point.
To develop their statistical model, Coşkun and his colleagues combined the within-subject biological variation of glucose and the analytical variation from measurement instruments, obtaining the total variation around the set point of the analyte. For each analyte, investigators included at least 10 previous measurements results taken within the last 15 years. “We calculated the set point of the test from the mean of the previous test results obtained from the individual in a steady state situation. The total variation around the set point of the test gives the ‘personalized reference interval’ of the test,” explained Coşkun.
The investigators applied the model to lab test results from 784 adult patients, calculating personalized reference intervals for 27 commonly used clinical chemistry and hematology measurement quantities.
“We found that the personalized reference intervals can be easily calculated in any healthcare service; we do not need complicated mathematical equations, sophisticated technologies, expensive instruments, or serial experiments,” Coşkun summarized. In a surprising finding, only a few previous test results were sufficient to calculate reliable personal reference intervals. “This is very important because such data can be obtained for many individuals, and in practice it is very easy for an individual to have his/her own personalized reference interval established.”
Clinical labs all have ample patient data in their laboratory information systems. “The critical point is that the data used to calculate the set point must be obtained when the patients are in a steady state. Such results must be identified and marked. Then, it is easy to calculate the personal reference intervals of the analyte” with this model, advised Coşkun. Looking ahead, his team plans to calculate personalized reference intervals based on prospectively collected data from thoroughly characterized study participants. The goal is to get more high-quality data for developing the personalized reference interval model and its applications.
The European Federation of Clinical Chemistry and Laboratory Medicine’s (EFLM) Working Group on Biological Variation and Task Group for the Biological Variation Database have set up the EFLM Biological Variation Database and the European Biological Variation Study to increase the reliability of biological variation data for analytes in clinical use. “This work is ongoing,” said Coşkun.