Statistical models for pediatric endocrinology

Abstract Background: Modelling references for biomarkers as reference curves enables calculation of patient z-scores that are adjusted for both gender and covariance with age during childhood.Aim: To establish biomarker reference curves using the ‘LMS’ method previously employed for conventional growth charts.Design, Setting and Participants: Cross-sectional study of healthy Norwegian girls (n=647) and boys (n=465) in ages 6 to 16 years were recruited in the Bergen Growth Study (2016). Blood samples were analyzed using state-of-the-art instruments.Results: Reference curves for several biomarkers were established in the ‘LMS’ framework and provided here for clinical implementation. Conclusion: The ‘LMS’ reference framework is already used to create conventional growth charts and may also be applied in clinical biochemistry.