Strand 1: High Risk Behaviours in Adolescence

Leader: Glyn Lewis Strand Publications

A range of “high risk behaviours” become relatively common in adolescence: we shall examine risky sexual behaviour, cannabis use, problematic alcohol use and self harm. Each of these will be considered in its own right, though Jessor (1998) and McGue (2005) have suggested that all high risk behaviours can be thought of as a single phenomenon and we will examine this. We also need to distinguish high risk behaviours from the universal tendency of adolescents to try new things. Antisocial behaviour will be studied in Maughan’s MRC grant (G0500953) so is not included separately here but Maughan will bring her expertise to our research. The analysis will take into account other important factors which influence adolescent behaviour (e.g. gender, puberty, peer influences, parenting). We will examine what mechanisms account for the relationship between SES and high risk behaviours in adolescence.

1.1 Describing the relationship between SES and high risk behaviours

Researchers: Lewis, Low, Sterne, Ness, Lawlor, Hickman, Wiles, Sayal, Gunnell, Evans

Our first step is to describe the relationship between SES and high risk behaviours. There is some rather inconsistent evidence for an association between risky sexual behaviour, alcohol and drug use and SES (Wellings 2001; Low 2001; Daniel 2006) though the evidence for self- harm is stronger (Patton et al 2004). We will use multiple measures of SES including parental income, education, occupational status as well as measures of wealth. These might influence outcome via different mechanisms (Braveman 2005, Lewis 1998, Araya 2003). We will also look at how SES measures change over time as repeated measures are available in ALSPAC (e.g. Burgess et al 2004; Galobardes 2004). We will describe the co-occurrence of high risk behaviours and examine the association between the different high risk behaviours in longitudinal analyses. We will examine whether there are consistent developmental relationships between our outcomes in relation to puberty. Factor analysis will be used to study the correlations between behaviours and cluster analysis to study groupings within individuals.

(Back to top)

1.2 School & neighbourhood influences on high risk behaviours

Researchers: Lewis, Low, Sterne, Ness, Lawlor, Hickman, Wiles, Sayal, Gunnell, Evans, Vignoles, Duckworth

A number of factors are thought to influence high risk behaviours relate to the school and neighbourhood context, notably academic achievement, peer relationships and “school connectedness” (Fergusson et al 2003, Green 2005, Hawton & James 2005). Kairouz (2005) did not find differences in alcohol or cannabis use though Schneiders (2006) did report neighbourhood differences in child behaviour problems in the Netherlands. Maes (2003) and Kumar (2002) found some evidence for school effects on alcohol use in Belgium and the US. Our aim is to identify possible contextual effects – specifically schools and neighbourhoods - on behaviour, and distinguish them from those that act through individuals, using the multilevel techniques developed in strand 5 and the measures of neighbourhood in strand 4.

(Back to top)

1.3 Mental health and high risk behaviour

There is evidence that SES is associated with depression and anxiety in adolescence and childhood (eg Brooks-Gunn 1997; Fergusson 2000, see also Costello 2003). These projects will examine whether mental health acts as a mediating variable between SES and high risk behaviours.

i) High risk behaviours and depression and anxiety in adolescence

Researchers: Lewis, Low, Sterne, Ness, Lawlor, Hickman, Wiles, Sayal, Gunnell, Evans, Maughan

There is some evidence suggesting depression and anxiety can lead to highrisk behaviours (eg Henry 1993, Rohde 1996) but also that high risk behaviours might increase risk of depression and anxiety (eg Hallfors 2005). Using repeated measures of mental health and high risk behaviours we can investigate the direction of causality in the cohort.

ii) Hyperactivity and high risk behaviours

Researchers: Sayal, Propper, Maughan, Tharpar

There is a clear overlap between high-risk behaviours, conduct disorders and hyperactivity (Taylor 1996). Although hyperactivity can be conceptualised as a continuum within the population, previous research has mainly focused on the extreme end with children meeting criteria for Attention Deficit/Hyperactivity Disorder (ADHD). We will investigate whether subthreshold levels of hyperactivity in children aged 4 and 7 years has long term adverse consequences on high risk behaviours. We will also investigate theuse of Mendelian randomisation methods in relation to polymorphisms in DRD4 (see strand 5).

(Back to top)

1.4 Individual Psychological characteristics and high risk behaviours

i) Sensation seeking and high risk behaviours

Researchers: Lewis, Low, Sterne, Ness, Lawlor, Hickman, Wiles, Sayal, Gunnell, Evans, Munafo, Maughan, Thapar

We will investigate whether sensation seeking, a facet of the broader personality trait of extraversion, partially mediates the relationship between SES and high risk behaviour. Sensation seeking is associated with high risk behaviours in adolescence (Kelley 2004) and with polymorphisms of the dopamine DRD4 receptor (Munafo 2003; Munafo 2006). This may be due to variation in executive functioning, alerting, and orienting processes (Swanson, 2003). In ALSPAC executive function has been measured using the Stop Signal Test at age 10 and sensation seeking (Arnett 1994) has been measured at age 11 and 13 and will be measured again at 15. The evidence on genetic variation in dopaminergic candidate genes can be used for the instrumental variable approach as well as studying this relationship in its own right.

ii) Does impaired autobiographical memory mediate the relationship between SES and self-harm

Researchers: Gunnell, Evans, Kidger, Hawton, Williams, Lewis

Childhood trauma and parental separation are associated with self-harm (Dube 2001, Fergusson 2003) and may mediate SES effects. We will test the hypothesis that early trauma in childhood results in autobiographical memory deficits that impair problem-solving and leads to self- harm in times of crisis (Williams 1986; Pollock 2001). Autobiographical memory has been assessed in the 12 year-old clinic using a modified version of the Williams task (1996). George Patton, Melbourne, and Michael Lynskey, St Louis will be invited to participate. 

(Back to top)