Continuous reference curves for common hematology markers in the CALIPER cohort of healthy children and adolescents on the Sysmex XN‐3000 system

Abstract

Introduction: Clinicians and healthcare professionals rely heavily on health-associated standards, such as reference intervals (RIs), for appropriate laboratory test result interpretation. RIs are commonly partitioned into discrete age/sex bins based on statistical and/or clinical significance. In pediatric hematology, such partitioning does not adequately represent complex variation in analyte concentrations throughout maturation. The objective of this study was to establish continuous RIs for common hematological parameters in the healthy pediatric Canadian Laboratory Initiative on Pediatric Reference Intervals (CALIPER) cohort.Methods: Data from healthy CALIPER children and adolescents (6 months-<19 years) were used to generate continuous RIs (ie, 2.5th and 97.5th quantiles) for 19 hematological parameters. Continuous curves were statistically established with nonparametric quantile regressions. Flagging rate analysis was completed for the established continuous upper and lower reference limits and subsequently compared to previously published discrete CALIPER reference intervals for all parameters.Results: Continuous RIs were established for 19 hematology parameters, where seven required sex-specific reference curves. Based on flagging rate assessment, continuous RIs appear to more accurately estimate hematological reference limits over the pediatric age range, especially for analytes with complex age-and sexspecific reference value patterns.Conclusions: This is the first study to generate continuous RIs for a breadth of hematological markers in a healthy pediatric Canadian population. The increased power of continuous reference intervals to accurately estimate the complex relationship between hematological analyte concentration and age during a time of extensive growth and development is expected to improve laboratory test result interpretation and, subsequently, pediatric clinical decision-making.

A pipeline for the fully automated estimation of continuous reference intervals using real-world data

“…In order to put our generated results into context, we compared the results to RIs established in other projects. Here, we included results from the CALIPER studies 21 , 22 , 38 , the multi-center PEDREF project 13 , 14 , as well as RIs stated in the package inserts provided by the manufacturers 23 – 27 . The comparisons are shown in Fig.…”

“…The CALIPER project has established age- and sex-specific RIs for children from birth to adulthood for over 170 biomarkers using population-based methods. For a subset of these, continuous RIs are reported as well 21 , 22 , 37 , 38 . However, due to a limited number of samples from healthy neonates and young children, and the extensive dynamics during these early stages of life, the first 6 or 12 months of life are excluded.…”

“…However, due to a limited number of samples from healthy neonates and young children, and the extensive dynamics during these early stages of life, the first 6 or 12 months of life are excluded. For ALP, CREA, and HB, continuous RIs were generated using nonparametric quantile regression with penalized splines, and the resulting RIs (2.5th and 97.5th percentiles) are reported for each half year of life 21 , 22 . Additionally, for ALP and CREA, continuous RIs have recently been established using LMS-based models 38 .…”

“…To showcase the application of the pipeline, we estimate continuous RIs with CIs using pediatric routine measurements of three biomarkers known for their extensive pediatric dynamics and compare the results to established RIs from the CALIPER studies 21 , 22 , the PEDREF 13 , 14 project as well as from the package inserts provided by the manufacturers 23 – 27 .…”

Establishing pregnancy-specific haematological reference intervals in Ghana; a three-center cross-sectional study

“…Therefore, future longitudinal study designs should consider simultaneously comparing discrete and continuous haematological reference intervals to ascertain which better reflect the dynamics in pregnancy. Recently, the CALIPER cohort study [23] concluded that continuous reference intervals better represented the complex changes in haematological variables within the paediatric age range instead of the traditionally used discrete partitioning of these reference intervals. A similar data exploration would likely improve haematological test interpretation as well as improve intra-partum and post-partum decision making process.…”