Validation of an approach using only patient big data from clinical laboratories to establish reference intervals for thyroid hormones based on data mining

Abstract

BackgroundWhile many studies have established reference intervals (RIs) for thyroid hormones using patient data, this approach has not been validated. Therefore, in this study, we aimed to validate an approach for establishing RIs for thyroid hormones only using patient data from clinical laboratories. MethodsWe established two derived databases: derived database* and derived database#. Reference individuals in derived database* were selected using strict exclusion criteria, and the RIs established by the database were considered standard RIs (RIs*). Individuals in derived database# were the physical examination population, whose information was downloaded directly from the Laboratory Information System, and RIs established from this database were evaluated (RIs#). The comparative confidence interval (CI) method and consistency of the decision results based on external databases were used to compare RIs* and RIs#. ResultsRIs# and RIs* for the thyroid hormones tested were similar. The 90% CIs of the upper and lower limits of RIs for most thyroid hormones overlapped between RIs# and RIs*, and the limit of RIs# was within the 90% CI of RIs*. The consistency rates for the results of the RIs* and RIs# in the external database were greater than 98% for all thyroid hormones tested. ConclusionIt was possible to establish RIs for thyroid hormones using only patient data from clinical laboratories after adopting appropriate statistical methods.