Mixed hidden Markov quantile regression models for longitudinal data with possibly incomplete sequences.

Quantile regression provides a detailed and robust picture of the distribution of a response variable, conditional on a set of observed covariates. Recently, it has be been extended to the analysis of longitudinal continuous outcomes using either time-constant or time-varying random parameters. However, in real-life data, we frequently observe both temporal shocks in the overall trend and individual-specific heterogeneity in model parameters. A benchmark dataset on HIV progression gives a clear example. Here, the evolution of the CD4 log counts exhibits both sudden temporal changes in the overall trend and heterogeneity in the effect of the time since seroconversion on the response dynamics. To accommodate such situations, we propose a quantile regression model, where time-varying and time-constant random coefficients are jointly considered. Since observed data may be incomplete due to early drop-out, we also extend the proposed model in a pattern mixture perspective. We assess the performance of the proposals via a large-scale simulation study and the analysis of the CD4 count data.

Cognitive ability, neighborhood deprivation, and young children's emotional and behavioral problems.

PURPOSE: To examine if cognitive ability moderates the effect of area (neighborhood) deprivation on young children's problem behavior. METHODS: Data from the first two sweeps of the Millennium Cohort Study (MCS) in the UK were used. Children were clustered in small areas in nine strata in the UK and were aged 9 months at Sweep 1 and 3 years at Sweep 2. Neighborhood deprivation was measured with the Index of Multiple Deprivation at Sweep 1. Overall and specific problem behavior was measured with the Strengths and Difficulties Questionnaire at Sweep 2. To explore moderator specificity we used three indices of ability (verbal cognitive ability, non-verbal cognitive ability, and attainment of developmental milestones). Adjustment was made for child's age and sex, and for Sweep 1 family adversity (number of adverse life events), family structure, mother's social class and psychological distress, and family socio-economic disadvantage. RESULTS: We found both support for our main hypothesis, and evidence for specificity. Neighborhood deprivation was, even after adjustment for covariates, significantly associated with children's peer problems. However, verbal and non-verbal cognitive ability moderated this association. CONCLUSIONS: Neighborhood deprivation was related to peer problems even at preschool age. Although the effect of neighborhood deprivation on externalizing problems was mediated by family poverty and parental socio-economic position and although its effect on internalizing problems was mediated by parental mental health, its effect on difficulties with peers was independent of both parental and child characteristics. Cognitive ability moderated the effect of neighborhood deprivation on preschoolers' peer relationships difficulties.

Modeling risks: effects of area deprivation, family socio-economic disadvantage and adverse life events on young children's psychopathology.

BACKGROUND: The effects of contextual risk on young children's behavior are not appropriately modeled. AIMS: To model the effects of area and family contextual risk on young children's psychopathology. METHOD: The final study sample consisted of 4,618 Millennium Cohort Study (MCS) children, who were 3 years old, clustered in lower layer super output areas in nine strata in the UK. Contextual risk was measured by socio-economic disadvantage (SED) at both area and family level, and by distal and proximal adverse life events at family level. Multivariate response multilevel models that allowed for correlated residuals at both individual and area level, and univariate multilevel models estimated the effect of contextual risk on specific and broad psychopathology measured by the Strengths and Difficulties Questionnaire. RESULTS: The area SED/broad psychopathology association remained significant after family SED was controlled, but not after maternal qualifications and family adverse life events were added to the model. Adverse life events predicted psychopathology in all models. Family SED did not predict emotional symptoms or hyperactivity after child characteristics were added to the model with the family-level controls. CONCLUSIONS: Area-level SED predicts child psychopathology via family characteristics; family-level SED predicts psychopathology largely by its impact on development; and adverse life events predict psychopathology independently of earlier adversity, SED and child characteristics, as well as maternal psychopathology, parenting and education.

Psychopathology and prosocial behavior in adolescents from socio-economically disadvantaged families: the role of proximal and distal adverse life events.

The study investigated if proximal contextual risk (number of adverse life events experienced in the last year) or distal contextual risk (number of adverse life events experienced before the last year) is a better predictor of adolescent psychopathology and prosocial behavior. It also tested for the specificity, accumulation and gradient of contextual risk in psychopathology and prosocial behavior, and for the interaction between proximal and distal contextual risk in psychopathology and prosocial behavior. The sample was 199 11-18 year old children from a socio-economically disadvantaged area in North-East London. The Strengths and Difficulties Questionnaire (SDQ), which measures four difficulties (hyperactivity, emotional symptoms, conduct problems, and peer problems) and prosocial behavior, was used. Confounders were age, gender, and maternal educational qualifications. To model the relationship between the five SDQ scales and contextual risk multivariate response regression models and multivariate response logistic regression models that allow the error terms of the scale specific models to be correlated were fitted. This study highlighted the importance of proximal contextual risk in predicting both broad and externalizing psychopathology, and the importance of considering risk accumulation rather than specificity in predicting psychopathology. By showing that the number of proximal adverse life events experienced had a steady, additive effect on broad and externalizing psychopathology, it also highlighted the need to protect adolescents experiencing current risk from further risk exposure. By showing that the number of distal adverse life events experienced did not affect the proximal risk's impact on either broad or externalizing psychopathology, it highlighted the need to protect all adolescents, irrespective of experience of early life adversities, from risk.