Exploring the association between anxiety and conduct problems in a large sample of Twins aged 2-4.

Anxiety and conduct problems are common areas of difficulty in child psychopathology (Anderson, Williams, McGee, & Silva, 1987). Although each may appear alone these problems also co-occur in both epidemiological and clinic-referred samples (Fergusson & Horwood, 1993; Russo & Beidel, 1994). This is interesting because at first glance the symptoms of anxiety appear to be counter indicative of conduct problems, and the presumed causes of anxiety and conduct problems would seem to imply a negative association. Although estimates as to the degree of overlap between anxiety and externalizing problems vary between studies, a review of the literature has reported that between 2 and 21% of children refered for anxiety had externalizing problems (Russo & Beidel, 1994). Some of the studies leading to these estimates combine child and adolescent data, and co-occurrence between anxiety and externalizing problems may be greater in preadolescent as compared to adolescent samples. For example, in a study of children diagnosed with overanxious disorder, 61% of those aged 5-11 were also diagnosed with an externalizing disorder, but only 15% of those aged 12-19 received such a diagnosis (Strauss, Lease, Last, & Francis, 1988).

Over the past decade there has been interest in using genetic data to examine the origins of associations between traits (for a textbook on behavioral genetics see Plomin, DeFries, McClearn, & McGuffin, 2001). Such data can be used to estimate the extent to which associations are mediated by genes, shared environment (any environmental influence contributing to the resemblance of family members) and nonshared environment (environmental factors not contributing to the resemblance of family members). An example of genetic mediation of the association between anxiety and conduct could be via the serotonin system--which has been associated with both anxiety and conduct problems (Eley, Collier, & McGuffin, 2002; Unis et al., 1997), although as discussed later most genetic research suggests that genetic factors operate differently for anxiety and conduct. Shared environmental factors, such as parenting style could also influence both anxiety and conduct (Barrett, Rapee, Dadds, & Ryan, 1996). Finally, nonshared environmental factors, such as peer relationships, could influence both anxiety and conduct. For example, associating with a poorly behaved peer group may result in higher levels of conduct problems (see Windle & Windle, 1996), and higher levels of anxiety could result from fear of the consequences of such behavior. Understanding the origins of associations between traits is important as it can help direct future research. For example, if behavioral genetic studies show the association between two traits to be entirely mediated by genetic factors, searching for shared environmental influences on this association is not warranted.

Genetic research examining the association between emotional and behavioral problems has tended to focus on problems either within internalizing arenas (e.g., anxiety and depression: Eley & Stevenson, 1999; Silberg, Rutter, Neale, & Eaves, 2001; Thapar & McGuffin, 1997) or within externalizing arenas (e.g., conduct and hyperactivity: Silberg et al., 1996; aggressive and nonaggressive antisocial behavior: Eley, Lichtenstein, & Moffitt, 2003). Such studies have shown that associations within these domains are largely explained by genetic factors. Only a handful of genetic studies have explored associations between internalizing and externalizing problems in childhood and adolescence and these studies have examined a range of phenotypes and age groups and have yielded conflicting results. A common genetic liability was found to be the most significant indicator of the association between depressive symptoms and antisocial behavior in an adolescent sample (O'Connor, McGuire, Reiss, Hetherington, & Plomin, 1998; O'Connor, Neiderhiser, Reiss, Hetherington, & Plomin, 1998) and between internalizing and externalizing problems in 4- to 11-year-old twins (Schmitz & Mrazek, 2001). In contrast, shared environment was most influential in the association between internalizing and externalizing problem composites in a sample of 2- to 3-year-olds (Schmitz, Cherny, Fulker, & Mrazek, 1994). This was also found to be the case in 5-9 and 12-to 15-year-olds--especially for the younger group (Gjone & Stevenson, 1997). The mixed results may in part be due to age and measurement differences between the studies (for a discussion of the effects of age and phenotype on genetic estimates on anxiety, see Eley and Gregory, 2003). However, as they stand, these studies suggest the tentative hypothesis that overlap between anxiety and conduct is due early in development to shared environmental influences and, later in development, to genetic influences.

Other areas of research provide few clues as to whether genetic, shared environmental, or nonshared environmental influences are important in the positive association between anxiety and conduct problems. For example, although genetic influences on the serotonin system may be important in the association between anxiety and conduct problems, much of the previous research emphasises serotonin differences between these two problems. For example, whole blood serotonin (5-HT) correlates positively with conduct symptoms, but negatively with symptoms of anxiety (Pliszka, Rogeness, Renner, Sherman, & Broussard, 1988). Similarly, there appears to be an assumption in the literature that shared environmental factors that affect anxiety--such as emotional overinvolvement (for a recent review of family processes in the development of anxiety see Dadds & Roth, 2001)--are different from those that affect conduct problems, such as having a delinquent sibling (Farrington, 1995). Although few studies have explored the effects of family life on anxiety and conduct problems, available studies have yielded contrasting results, with some finding evidence for risk factors impacting on both problems (Barrett et al., 1996; Dadds, Barrett, Rapee, & Ryan, 1996), and others finding evidence of specificity (Rosenstein & Horowitz, 1996). Finally, a clear picture of the effects of nonshared environmental factors on the association between anxiety and conduct is lacking. For example, although an increase in stressful life events has been associated with an increase in both emotional and behavioral problems (Berden, Aalhaus, & Verhulst, 1990), other types of life events can increase one type of problem and diminish another. This was found in a study with a large sample of midadolescents that showed an association between positive daily events, including social engagements with friends, and higher levels of delinquent activity (e.g., peer-associated, age-normative deviant behaviors) and lower levels of depressive affect (Windle & Windle, 1996).

In summary, although there are many genetic and environmental reasons to think that anxiety and conduct problems are different, a modest positive association exists between them. The few genetic studies of this association have provide mixed results, although they suggest the possibility that shared environmental influences are more important in childhood than later in life. This study provides the first behavioral genetic analysis of the association between anxiety and conduct problems in a sample of preschoolers.

METHOD

Participants

Participants in this study were members of the Twins Early Development Study (TEDS), a longitudinal study of development in which parents of all twins born in England and Wales during 1994-95 were invited to participate. Assessments were made at 18 months, 2, 3, and 4 years. At the first assessment 10,932 families were sent booklets to complete. A total of 9,442 of these families (86%) provided data at the first assessment. From this initial sample, twin pairs were excluded if there were extreme pregnancy or perinatal difficulties (e.g., gestation <32 weeks: n = 204) or if either of the twins had an extreme medical condition (e.g., chromosomal abnormalities such as Down's Syndrome and Cerebral Palsy: n = 212).

The present sample included families providing data on anxiety and conduct problems at one or more times during the 2, 3, and 4-year-old assessments. This resulted in a final sample of 6,783 twin pairs: 1,071 monozygotic male (MZM); 1,154 dizygotic male (DZM); 1,217 monozygotic female (MZF); 1,115 dizygotic female (DZF); and 2,226 opposite sex (DZO) twin pairs. Attrition analyses reveal a number of differences between families providing data at 18 months and those providing data at 2, 3, and 4, namely: families remaining in the study are slightly more likely to …