Linear growth and mid-childhood cognitive outcomes in three birth cohorts of term-born children: an approach to integrating three growth models to explore critical windows. Issue 8 (26th August 2020)

• Record Type: Journal Article
• Authors: Leung, Michael; Krishna, Aditi; Yang, Seungmi; Bassani, Diego G; Roth, Daniel E
• Is Part Of: BMJ open. Volume 10:Issue 8(2020)
• Publication Date: 2020-08-26
• Subjects: epidemiology -- community child health -- public health -- statistics & research methods; Medicine -- Research -- Periodicals; 610.72
• DOI: 10.1136/bmjopen-2020-036850
• Languages: English
• ISSNs: 2044-6055

Abstract

Abstract : Objective: To illustrate that a mediation framework can help integrate inferences from three growth models to enable a comprehensive view of the associations between growth during specific developmental windows and mid-childhood IQ. Design: We analysed direct and indirect associations between mid-childhood IQ and length/height growth in five early-life age intervals bounded by conception, birth, early, mid and late infancy, and mid-childhood using estimates from three growth models (lifecourse, conditional change and change score) applied to three historical birth cohorts. Participants and setting: 12 088 term-born children from the Collaborative Perinatal Project (CPP) in the USA (n=2170), the Promotion of Breastfeeding Intervention Trial (PROBIT) in Belarus (n=8275) and the Cebu Longitudinal Health and Nutrition Survey (CLHNS) in the Philippines (n=1643). Primary outcome measure: Mid-childhood IQ. Results: Our analyses revealed cross-cohort and cross-interval variations in the direct and indirect effects of foetal and early childhood physical growth on mid-childhood IQ. For example, in CPP, there was a direct association of prenatal growth with IQ that was not evident in the other cohorts, whereas in PROBIT and CLHNS, we observed that foetal and early growth-IQ associations were mediated through size in later periods. Conclusion: Lifecourse, conditional change and change score growth models yield complementary inferences when appropriately interpreted. Future longitudinal studies of associations of early-life growth with later outcomes would benefit from adopting a causal mediation framework to integrate inferences from multiple complementary growth models.