According to a viewpoint article published in JAMA, there currently is debate regarding the definition of “normal” values of common laboratory tests for the application in precision medicine research and big data assessment. Additionally, the term “healthy individuals” for referencing laboratory values on a comparative basis can also be challenging to define.

Defining normality and ensuring multiplicity across population strata are just 2 of the issues involved in precision medicine research. Furthermore, the act of defining a generally healthy population for estimating a “normal” variation range across the population is an inherent challenge in applying clinical laboratory testing in research. Health is considered a relative condition that lacks a standardized definition, and the definition of what “health” comprises varies among laboratories.

Current disparities exist between guidelines from the Clinical and Laboratory Standards Institute (CLSI) and clinical research practice with regard to how many reference individuals should be enrolled in a study to establish or verify reference intervals for laboratory analytes. Although the CLSI recommends 120 reference individuals, researchers often use fewer than 120 individuals to verify an existing reference range, with some laboratories only using approximately 20 individuals.

Continue Reading

Additionally, multiplicity of analytes (eg, hemoglobin A1C) distributed across the population and subpopulations likely present differences if no corrections are made for the number of performed comparisons. Computing reference intervals in large-scale data sets (eg, electronic health records) while attempting to overcome the obstacles in multiplicity can often result in suboptimal test utilization across the population, resulting in reductions in sensitivity, specificity, and clinical utility.

Related Articles

Electronic health records, however, may offer solutions by helping to facilitate systematic analyses across laboratories and other data sets while concurrently adjusting for the scale of multiple testing. Computationally derived genetic ancestry in addition to laboratory testing could also help overcome the conflation of race and ancestry that is pervasive in data.

Multiple groups, including researchers and laboratories, can help achieve proposed goals by actively “sharing estimated reference intervals across demographic strata and documenting design choices such as outlier procedures and inclusion criteria, allowing other researchers to reproduce their calculations.” Additionally, healthcare institutions may also be highly valuable in these efforts by making “consented patient data available to compute reference ranges for their populations.”


Manrai AK, Patel CJ, Ioannidis JPA. In the era of precision medicine and big data, who is normal? [published online April 23, 2018]. JAMA. doi:10.1001/jama.2018.2009.