Statistical agreement and cost–benefit: Comparison of methods for constructing growth reference charts

Abstract

Background: Growth reference charts are important tools for adequate paediatric decisions.Aim: In view of the workload required to construct empirical growth reference charts we debate practicable and less demanding alternatives and took the recent national 2000–2002 Lithuanian growth charts as an example. Two options appeared reasonable: (1) applying international WHO child growth standards and WHO growth reference data for 5–19 years that are recommended for global use; or (2) replacing the costly empirical method of deriving national growth references by more convenient low-cost statistics, e.g. the method of generating synthetic references for the Lithuanian population.Methods: We analysed the degree of agreement between the 2000–2002 Lithuanian growth charts, and the international WHO child growth standards and WHO growth reference data for 5–19 years and synthetic references for the Lithuanian population using the Bland–Altman method.Results: Synthetically generated references for the Lithuanian population slightly surpassed the national Lithuanian reference for body height (males +0.3 (SD 0.9) cm; females +0.2 (SD 0.6) cm) particularly at young age, which may be regarded clinically irrelevant. WHO international child growth standards and the WHO growth reference data for 5–19 years, however, failed to match the Lithuanian references as they underestimated mean height in boys by –2.8 (SD 1.4) cm and in girls by –2.9 (SD 1.1) cm, with extremely discrepant estimates of more than –6 cm occurring in several adolescent cohorts.Conclusions: The analysis revitalizes the debate on clinically relevant and at the same time practicable but less demanding alternatives for constructing growth reference charts, and for economic reasons, strongly suggests replacing the traditional empirical methods by synthetic growth references.