RESUMO
PURPOSE: Associations between prenatal earthquake exposure and children's mental health remain unclear. Moreover, there is a paucity of research using quasi-experimental statistical techniques to diminish potential selection bias. Thus, this study aimed to explore the impact of prenatal exposure to the Chilean earthquake of 2010 on children's behavioural and emotional problems between 1½ and 3 years old using propensity score matching. METHODS: Participants included 1549 families from the Encuesta Longitudinal de la Primera Infancia cohort in Chile. Maternal reports using the Child Behaviour Checklist (CBCL) were used to assess behavioural and emotional problems between 1½ and 3 years old. Information on prenatal earthquake exposure was collected via maternal report. The Kernel matching estimator was used to compare the average treatment effects of children who were exposed to the earthquake compared to those who were not. RESULTS: Five of the seven CBCL outcomes were statistically significant after matching and adjustment for multiple testing, suggesting greater difficulties for exposed children which included emotional reactivity, anxious/depressed, sleep problems, attention problems, and aggression (mean difference of 0.69, 0.87, 0.73, 0.85, 3.51, respectively). The magnitude of the effect was small to medium. CONCLUSION: Findings contribute to the potential causal inferences between prenatal earthquake exposure and increased behavioural and emotional problems in early childhood. Results suggest that in utero experiences may have long-term consequences for infants' well-being, supporting the need for specific interventions in pregnancy after natural disasters.
Assuntos
Transtornos do Comportamento Infantil , Terremotos , Lactente , Feminino , Gravidez , Criança , Humanos , Pré-Escolar , Chile , Ansiedade , Transtornos do Comportamento Infantil/psicologia , AgressãoRESUMO
A secondary analysis was conducted on longitudinal data collected from ELPI, a representative Chilean survey to model Chilean infant's receptive language using contextual, maternal and prenatal factors. The sample for the current study comprised children aged between 36 and 48 months (n = 3921). The sample was re-assessed when children were aged 60-72 months (n = 3100). Linear regression analyses were conducted. At the first time point, all the predictors included were significant (living area, health system provision, maternal intelligence and education, adolescent pregnancy, maternal medical appointments during pregnancy, and presence of a significant other at childbirth), except for smoking during pregnancy. The model explained 13% of the variance. However, when timepoint one receptive language scores were included in the analyses for when children were aged 60-72 months, only two variables remained as significant predictors: previous receptive language scores and maternal education, explaining 21% of the variance. Findings and implications are discussed.