Dataset Information

Establishing laboratory-specific reference intervals for TSH and fT4 by use of the indirect Hoffman method.

ABSTRACT:

Background

The results of examinations of laboratory parameters are the basis of appropriate medical decisions. The availability of reliable and accurate reference intervals (RIs) for each laboratory parameter is an integral part of its appropriate interpretation. Each medical laboratory should confirm their RIs. Up-to-date reference intervals for thyroid function hormones are still a matter of ongoing controversy. The aim of the study was the application of the indirect Hoffman method to determine RIs for TSH and fT4 based on the large data pools stored in laboratory information systems and the comparison of these RIs to generally used RIs.

Material and methods

The TSH and fT4 routine examination results of hospitalized and outpatient populations were collected over five years (2015-2019), and reference limits were established by the improved Hoffmann method after the exclusion of outliers. Comparative verification of established RIs was conducted with the RIs values provided by test manufacturers and literature data.

Results

Various RIs were observed in different age groups in the examined populations. For TSH, RIs varied between different age groups, with a narrower range of RIs in the studied adult population and a shift of both reference boundaries toward higher values in comparison to manufacturers' data among children. RIs estimated for fT4 were very similar to the manufacturer and literature data.

Conclusion

Thyroid hormone levels change during a person's lifetime and vary between sexes, but this difference does not always influence the clinical interpretation of laboratory results in the context of RIs. The use of indirect methods is justified due to the ease and low cost of their application.

SUBMITTER: Placzkowska S

PROVIDER: S-EPMC8741008 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Background: Serum thyroid-stimulating hormone (TSH) and free thyroxine (FT4) change dynamically during pregnancy. Differences in geographic regions, populations, and manufacturer's methodologies can affect the reference intervals for thyroid function tests. The 2017 guidelines of the American Thyroid Association (ATA) recommended 4.0 mU/L as the cut-off point for the upper limit of serum TSH in early pregnancy. A systematic review is called for to establish practical, gestational-specific TSH and FT4 reference intervals for pregnant Chinese women and to explore whether the criteria are suitable for China. Methods: English and Chinese articles published from inception to Aug 2017 were searched in the PubMed, EMBASE, and SCIE English-language databases and the CNKI, WanFang, and CQVIP Chinese databases. The relative descent or ascent rates of serum TSH and FT4 were calculated, after which Comprehensive Meta-Analysis V2.0 software was used to analyze the data. Results: Eleven studies (6 in English and 5 in Chinese), five kits and 11,629 Chinese women from nine cities were considered in this meta-analysis. Compared with the reference ranges provided by manufacturers, serum TSH decreased in the first trimester, with the upper limit declining by 21.7% (5.0-36.6%), to a value close to 4.0 mU/L, and the lower limit declining by 85.7% (73.5-97.1%). It continued decreasing in the second trimester, with the upper limit declining by 24.0% (6.4-40.9%) and the lower limit declining by 40.7% (9.0-85.7%). For FT4, the upper limit fluctuated slightly, and the lower limit increased by 6.8% (1.0-14.6%) in the first trimester. Serum FT4 dropped gradually, with the upper limit declining by 21.8% (2.5-31.8%) and the lower limit declining by 12.7% (2.6-19.6%) in the second trimester. During the third trimester, the upper limit decreased by 25.1% (12.7-35.0%), while the lower limit decreased by 20.9% (14.8-27.3%). Conclusions: Various regions, kits and test methods affect the gestational TSH and FT4 levels. The non-pregnant serum TSH upper limit minus 22% is very close to 4.0 mU/L, which can be used as a sub-optimal approach to represent the cut-off value for pregnant Chinese women in the first trimester.

Project description:BackgroundReference intervals represent the expected range of physiological test results in a healthy population and are essential to support medical decision making. Particularly in the context of pediatric reference intervals, where recruitment regulations make prospective studies challenging to conduct, indirect estimation strategies are becoming increasingly important. Established indirect methods enable robust identification of the distribution of "healthy" samples from laboratory databases, which include unlabeled pathologic cases, but are currently severely limited when adjusting for essential patient characteristics such as age. Here, we propose the use of mixture density networks (MDN) to overcome this problem and model all parameters of the mixture distribution in a single step.ResultsEstimated reference intervals from varying settings with simulated data demonstrate the ability to accurately estimate latent distributions from unlabeled data using different implementations of MDNs. Comparing the performance with alternative estimation approaches further highlights the importance of modeling the mixture component weights as a function of the input in order to avoid biased estimates for all other parameters and the resulting reference intervals. We also provide a strategy to generate partially customized starting weights to improve proper identification of the latent components. Finally, the application on real-world hemoglobin samples provides results in line with current gold standard approaches, but also suggests further investigations with respect to adequate regularization strategies in order to prevent overfitting the data.ConclusionsMixture density networks provide a promising approach capable of extracting the distribution of healthy samples from unlabeled laboratory databases while simultaneously and explicitly estimating all parameters and component weights as non-linear functions of the covariate(s), thereby allowing the estimation of age-dependent reference intervals in a single step. Further studies on model regularization and asymmetric component distributions are warranted to consolidate our findings and expand the scope of applications.