Establishing Reference and Optimal Curves for Hemoglobin and Ferritin: Methodological and Computational Frameworks to Evaluate and Synthesize Existing and Emerging Evidence

Abstract

Introduction: Reference intervals (RIs) provide benchmarks for interpreting laboratory test results; however, inconsistencies in their development complicate clinical decision-making, especially in pediatrics where biomarker levels change rapidly with age and development. My dissertation synthesized existing evidence on pediatric RIs and reference curves (RCs) for hemoglobin and ferritin and established RCs and optimal curves (OCs) for these biomarkers using new methodological and computational frameworks. Although developed for hemoglobin and ferritin, these frameworks can be applied to other pediatric biomarkers and tailored to other populations. Methods: I conducted two systematic reviews with meta-analyses to understand existing evidence and methodological challenges in generating pediatric hemoglobin and ferritin RIs and RCs. I evaluated heterogeneity using forest plots, heatmaps, web-based visualization tools, and the I² statistic. I also proposed a standardized age partitioning approach to enable quantitative synthesis. I then analyzed data from the TARGet Kids! cohort study of healthy Canadian children ages 2-weeks to 10-years. I developed RCs for pediatric hemoglobin and ferritin, overcoming excessive partitions and small sample sizes in early childhood. I also introduced novel pediatric OCs derived from a sub-sample of participants meeting predefined optimality criteria. Finally, I created web-based tools for both the systematic reviews and cohort analyses, to explore heterogeneity, visualize data, and aid in interpretation. Findings: Both systematic reviews showed substantial heterogeneity across studies due to differing age intervals, population characteristics, and analyzer types, with limited data available for very young children. Lower limits from many published RIs differed from published World Health Organization (WHO) thresholds. Among participants in our full cohort analysis and those meeting predefined optimality criteria, the proportion of pre-adolescents who would be classified as iron deficient according to the WHO thresholds and the American Society of Hematology (ASH) thresholds varied substantially with age. Conclusion: RCs and OCs provide essential benchmarks for interpreting results, evaluating thresholds, and assessing population iron status. This work identified key methodological limitations, addressed several of these, and established frameworks for synthesizing evidence and developing pediatric RCs and OCs. Future studies should evaluate the clinical value of the tools developed and extend these approaches to other key laboratory biomarkers.