Table 1.

Recently published children’s health study pooled analyses

First Author (Publication date)	Study description	Primary study aim	Author identified strengths and limitations
Casas (2015) [11]	9377 mother–child pairs enrolled in 14 study populations from 11 European birth cohorts	Explore exposure–response relationship between PCB-153 and p-p′-DDE and birth outcomes; to evaluate whether any no exposure–effect level and susceptible subgroups exist; and to assess the role of maternal gestational weight gain.	Able to harmonize common potential confounders but with loss of some particular and valuable cohort characteristics. Only one PCB congener 153 was measured in all cohorts. There are 209 PCB congeners that differ in structure and mechanism of action and hence may have different health outcomes. Large sample size allowed better description of the shape of the relationship and identification of effect modifiers.
Iszatt (2015) [12]	Up to 2,487 children from 7 European birth cohorts	Using biomarker concentrations of polychlorinated biphenyl 153 (PCB-153) (n = 2,487), and p,p′-dichlorodiphenyldichloroethylene (p,p′-DDE) (n = 1,864), estimate prenatal and postnatal persistent organic pollutants (POPs) exposure using a validated pharmacokinetic model and examine POP exposure association with infant growth from birth to 24 months in singleton term children.	The largest study to date using pooled data across larger samples of individuals with heterogeneous and distinct prenatal/postnatal exposure profiles. Compared with single-cohort studies, the pooled design had: ○Better control for unmeasured confounding because the underlying confounder structure varies across cohorts. ○reduced or eliminated reporting bias by showing results for all eligible cohorts. Found significant heterogeneity when pooling the cohorts. The estimate of the average effect does not account for the magnitude of variation among the cohorts. Although variance inflation factors were < 5, variance doubled when prenatal and postnatal were mutually adjusted, suggesting collinearity.
Buckley (2016) [13]	707 children from three prospective cohort studies enrolled in the US between 1998 and 2006	Examine associations of prenatal urinary phthalate metabolite concentrations and body mass index (BMI) assessed in children between ages 4 and 7 years and evaluated differences by child’s sex.	Pooling data from three independent cohorts with notable variation in population characteristics strengthened the robustness of the findings. Provided a large sample size to assess heterogeneity of associations by hypothesized modifying factors. Potential bias due to missing covariate data and loss to follow-up may exist.
Engel SM. (2016) [14]	pooled analysis of four birth cohorts (children’s centers; n = 936)	Evaluate associations of prenatal exposure to organophosphorus pesticides (OPs) with mental and psychomotor development of children 24 months of age, taking into account both genetic and demographic susceptibility factors.	Both confounder adjustment and examination of heterogeneity were limited to covariates and characteristics that were shared by all centers. Although pooling these cohorts afforded more power to investigate gene–environment interactions, power was still limited in stratified analyses. Pooled analyses improved the ability to determine whether there is an overall effect of OP exposure experienced in diverse settings on child neurodevelopment both for policy and for research purposes.
Stratakis (2016) [15]	Multicenter, population-based birth cohort study from 1996 to 2011 in 9 European countries and the US 26,184 pregnant women and their children	To examine whether fish intake in pregnancy is associated with offspring growth and the risk of childhood overweight and obesity.	Strengths include large sample size, centralized data analysis following a consensus protocol, standardized exposure definition, and harmonized information about child outcomes and potential confounders. Potential confounding variables were defined as similarly as possible among the cohorts.