Adolescent Alcohol Use is Predicted by Childhood Temperament Factors Before Age 5, with Mediation Through Personality and Peers

MUCH OF THE research on risk factors for alcohol use has focused on adolescence, due to the importance of this period for initiation and escalation of alcohol use (Windle et al., 2009). However, most risk and protective factors for alcohol use do not arise de novo in early adolescence but rather have roots in developmental processes stretching back into early childhood (Zucker et al., 2008). Clarifying the roots of these developmental patterns will be vital in acquiring a more complete understanding of the origins of risk factors for the acquisition and maintenance of patterns of alcohol use. During their very early years, children begin to develop skill sets and behavioral styles for regulating their emotions and behaviors and interacting with others. Through interactions between early temperamental styles and the child’s experience, the individual enters adolescence with personality characteristics and life experiences that have been accumulated over the first decade of life. A number of studies now suggest that temperamental and behavioral characteristics from childhood predict much later alcohol use outcomes. Several large, longitudinal studies have found that externalizing behavior measured at ages 7 to 9 predict alcohol consumption in late adolescence and young adulthood (Dubow et al., 2008; Englund et al., 2008; Maggs et al., 2008; Pitkanen et al., 2008). A small number of studies have also found behavioral characteristics even earlier in life that predict subsequent alcohol-related outcomes. For example, in a longitudinal epidemiological sample of children born in the city of Dunedin from 1972 to 1973, observer ratings of undercontrol and of behavioral inhibition at age 3 predicted the diagnosis of alcohol use disorders at age 21 (Caspi et al., 1996). Studies by Zucker and colleagues of children of alcoholics have also found that early externalizing behavior and measures of undercontrol in preschool age children were related to alcohol use outcomes in early adolescence (Martel et al., 2009; Zucker, 2006). Thus, the handful of studies that have examined risk factors in childhood have found significant relationships with adolescent and adult alcohol use and/or misuse; however, most of these studies have analyzed data from late childhood (e.g., ages 8 to 9), with only a few notable exceptions examining earlier behavioral development. The findings from this small number of studies indicate that further exploration of early childhood risk factors and the pathways by which they are associated with the eventual emergence of risky behaviors in adolescence and young adulthood is an important area of study. Data from the Avon Longitudinal Study of Parents and Children (ALSPAC) provide a rare opportunity to address questions about longitudinal pathways of risk for alcohol use. This ongoing project has followed a large cohort of children and their parents from early in the mother’s pregnancy through childhood and adolescence, with assessments ongoing as the participants currently enter young adulthood. The study enrolled all pregnant mothers resident in a defined geographical area with expected dates of delivery between April 1, 1991, and December 31, 1992, and has collected comprehensive health-related information, from both the mothers and (when they were old enough) children. With this rich existent data set, we sought to ask: How early can we predict risk for adolescent alcohol involvement? We make use of measures of temperament collected from the mother at multiple assessments from 6 months through 5 years of age. Although we use the word temperament throughout this manuscript, we adopt a broad perspective and make use of all measures that index emotional and behavioral styles through the age of 5. From these measures, we attempt to predict self-reports of alcohol use and alcohol-related problems from the adolescents at age 15.5. Additionally, we sought to understand factors that may mediate associations between very early childhood temperament and adolescent alcohol use. We focused on personality and peers. We examined personality because temperament is thought to encompass the affective, activational, and attentional core of personality. Temperament represents the individual differences in reactivity and self-regulation that arise very early in life, while personality consists of the more stable patterns of thoughts, emotions, and behaviors that subsequently emerge (Rothbart and Bates, 1998). Accordingly, we sought to examine whether any associations that might exist with very early childhood temperamental characteristics would be mediated through personality dimensions as measured later in life (early adolescence). A considerable literature connects personality traits to alcohol use (Littlefield and Sher, 2010), providing further rationale for testing whether personality mediates associations between early temperament and subsequent alcohol use. Second, we tested whether associations between early temperament and subsequent alcohol use would be mediated through peer characteristics. Studies have consistently found that peer substance use and deviant peer affiliation is one of the strongest predictors of adolescent substance use (Dick et al., 2007; Simons-Morton et al., 2001; Wang et al., 1995; Windle, 2000). Association with deviant peers is not random, but rather reflects in part selection by the adolescent into antisocial peer groups based partly on their own predispositions (Dick et al., 2007; Harden et al., 2008; Kandel, 1978; Rose, 2002). Other aspects of peer relationships have also been shown to be important. For example, social rejection by peers predicts adolescent substance use in prospective studies (Dishion et al., 1995; Dodge et al., 2009). Transactional theory (Sameroff, 2009) suggests that features of the child’s interpersonal style and behavior may contribute to peer group rejection. The association between child characteristics and affilation with antisocial peers and/or having difficulty with peers, and the subsequent association of these variables with substance use led us to test whether aspects of the peer group may mediate potential associations between temperamental characteristics and substance use. Finally, we ask whether there are gender differences in the mediational pathways from early child characteristics to adolescent alcohol-related outcomes. Mean differences in levels of alcohol use have been shrinking in industrialized countries as gender roles have begun to equalize (Rose et al., 2001); however, less clear is whether the relationships between risk and protective factors are the same for the sexes (Dodge et al., 2009). While several studies failed to find robust differences in predictors of alcohol involvement (Zucker, 2008), other studies report evidence for sex-specific risk factors, particularly with respect to the relative importance of internalizing factors and emotional distress as they relate to alcohol use in females (Hussong et al., 2008; Kokkevi et al., 2007). Many studies are underpowered to test for robust sex differences. With the large cohort of children followed in ALSPAC, we are able to address this question.