Study design and participants
We analyzed data from a prospective birth cohort of children born to mothers who participated in a double-blind, cluster-randomized controlled trial in rural western China, which has been described in detail elsewhere.14 Briefly, pregnant women from every village in two counties were randomized to daily consume a capsule of folic acid, iron/folic acid, or multiple micronutrients between 1st August 2002 and 28th February 2006. 1400 singleton births from the parent trial after 2004 were enrolled for long-term follow-up, and 1388 were followed during the first two years of age. Among them, 745 were followed at early adolescence aged 10-14 years between 1st June and 31st December 2016 (see Supplementary online material: Figure S1). Data analyses took place from 1st October 2020 to 30th April 2021. The trial and follow-up evaluation protocols were approved by the Ethics Committee in Xi’an Jiaotong University Health Science Center. Written informed consent was obtained from the biological parents or caregivers, while verbal consent was obtained from all the participants depending on their age.
Measurements
Anthropometric measurements
We measured HC in maximum frontooccipital circumference at birth, 1, 3, 6, 9, 12, 18, and 24 months of age. Field workers from public health graduates implemented all these measurements using a flexible tape by standardized procedures. We standardized all these measures into z-scores using the INTERGRROWTH-21st and WHO growth standard for birth and postnatal HCs, respectively.15,16
Cognitive development and emotional and behavioral health assessments
We administered the scale of Wechsler Intelligence Scale for Children, Fourth Edition (WISC-IV) among adolescents in a school meeting room. We derived the age-standardized full-scale intelligence quotient (FSIQ) to represent adolescent general cognitive ability and verbal comprehension (VCI), perceptual reasoning (PRI), working memory (WMI), and processing speed index (PSI) to assess other aspects of cognitive development.17 At the same day, we applied the Youth Self-Report (YSR-2001) of the Achenbach System of Empirically Based Assessment (ASEBA) to assess the emotional and behavioral outcome which was completed by adolescents themselves after uniform instructions.18 Eight narrow-band scales or syndromes i.e., Anxious/Depressed, Withdrawn, Somatic Complaints, Social Problems, Thought Problems, Attention Problems, Delinquent/Rule-Breaking Behavior and Aggressive Behavior, and three broad-band scales i.e., scores of Internalizing Behavior Problems (constituting by Anxious/Depressed, Withdrawn and Somatic Complaints), Externalizing Behavior Problems (constituting by Delinquent/Rule-Breaking Behavior and Aggressive Behavior) and Total Behavior Problems, were derived. All these scales were localized and qualified in Chinese norms with satisfying reliability.
Covariates
The following covariates were potentially included in the present study. Socioeconomic characteristics at enrollment included parent age, education, occupation and household wealth. The wealth index was constructed by principal component analysis using the 16 household assets or facilities, and then categorized into low-, medium-, and high-income households. Antenatal randomized regimens with durations were classified into folic acid or folic acid plus iron <180 days, folic acid plus iron ≥180 days, multiple micronutrients <180 days, and multiple micronutrients ≥180 days. Maternal reproductive history was indicated by parity. Birth outcomes mainly included birth weight-for-gestational age and sex z-score calculated by INTERGRWOTH, preterm birth (<37 gestational weeks), low birth weight (<2500 g), small-for-gestational age (SGA, <10th by INTERGROWTH)15 and sex. All the information obtained from the parent trial by face-to-face interview, and/or standard procedures were detailed elsewhere.14
Statistical analyses
To identify infant HC z-score trajectories across birth, 1, 3, 6, 9, 12, 18, and 24 months of age, we conducted a group‐based trajectory modelling that could efficiently segment the sample into mutually exclusive and exhaustive subgroups of individuals with similar growth patterns.19 We decided the final trajectories by comprehensively accounting for the following principles as well as parsimoniously summarizing the data features: i) statistically significant tests of parameter estimates for linear, quadratic, cubic or quartic terms, ii) lower Bayesian and Akaike information criterion values, ii) average of the posterior probabilities of group membership for individuals assigned to each group >0.7, and iv) odds of correct classification based on the posterior probabilities of group membership >5.
We then compared the characteristics among the trajectory groups using Chi-squared tests or analysis of variance. Further, the multivariable multinomial logistic regressions were used to examine the predictors of HC trajectories. In addition, we performed generalized estimating equations to estimate the associations of HC trajectories with adolescent cognitive development and emotional and behavioral health. For general interests, we examined the associations of HC z-score at single-time point with adolescent developmental outcomes.
To help identify the sensitive periods of HC growth, we calculated the conditional growth of HC in the following periods: 0-1, 1-3, 3-6, 6-12, 12-18, 18-24, 0-3, 0-6, 0-12, 0-24 and 12-24 months,20 and then examined their associations with adolescent developmental outcomes using generalized estimating equations.
We took the FSIQ and scores of total behavior problems, internalizing and externalizing behavioral problems as primary outcomes, and the other aspects of cognitive development and mental health as secondary outcomes, respectively. All statistical analyses were conducted by using Stata 15.0 (Stata Corp, College Station, Texas, USA). A two-sided P value < 0.05 was considered statistical significance.