Clinical Child and Family Psychology Review, Vol. 2, No. 1, 1999

Behavioral Genetics as a Tool for Developmental Psychology: Anxiety and Depression in Children and Adolescents Thalia C. Eley1,2

Over the past decade there has been a huge increase in the number of behavioral genetic studies looking into anxiety and depression in children and adolescents. There are now enough data in this area to make a review of the results useful. This paper begins with an outline of the methods used in such research and moves on to review the results in extant studies. Overall, these studies indicate modest to moderate genetic influence on both anxiety and depression. However, behavioral genetic methods are also paramount for exploring environmental influences in addition to genetic influences. Shared environment (that which makes family members resemble one another) is rarely identified in adult studies of personality or psychopathology and does not appear to be a significant influence for depression but it is for anxiety. Nonshared environment, which makes family members differ from one another, is found to be a significant influence for both anxiety and depression. Patterns within these results due to rater effects, age effects, sex effects, the precise phenotype measured, and the study design are explored. KEY WORDS: Twin studies; adoption studies; anxiety; depression, children.

BEHAVIORAL GENETIC ANALYSIS OF TWIN DATA

Shared environment factors are defined as those that make members of a family similar to one another, whereas nonshared environment factors are those which make members of a family different from one another. It is important to note that it is not simply that the experience of the environment itself is shared or not shared, but that the effect of that influence results in resemblance between family members or in differences between family members. Monozygotic or MZ (identical) twins share both their entire genome and also, by definition, their shared environment. Dizygotic or DZ (fraternal) twins only share half their additive genetic material on average—the same as any other sibling pair. In addition they entirely share their shared environment. The correlations for members of MZ and DZ twin pairs can be partialled as summarized in Eqs. (2) and (3).

The majority of data quantifying genetic and environmental influences on depression and anxiety in children and adolescents has come from twin studies. As the analysis of these data may be unfamiliar to some, this paper begins with a presentation of the concepts and analysis of twin data in behavioral genetic studies. In these studies, variance in the phenotype is divided into three latent factors: heritability or additive genetics (A2), common or shared environment (C2), and nonshared environment (E2), as illustrated by the following equation: 1

Social, Genetic, and Developmental Psychiatry Research Centre, Institute of Psychiatry, University of London, London, United Kingdom. 2 Address all correspondence to Thalia C. Eley, Social, Genetic, and Developmental Psychiatry Research Centre, Institute of Psychiatry, De'Crespigny Park, London SE5 8AF, United Kingdom; e-mail: [email protected]

21 1096-4037/99/0300-0021$16.00/0 © 1999 Plenum Publishing Corporation

Eley

22

These simultaneous equations allow one to estimate heritability as can be seen below. From this equation the contributions of shared and non-shared environment influences can be calculated. Three examples of how different pairs of correlations result in very different parameter estimates are given in Table I. In all of the pairs of data in Table I the DZ correlation is never less than half of the MZ correlation. This is because these were all examples where any genetic influence was additive. All DNA can occur in variant forms called alleles. Additive genetic influences are those in which having two copies of the risk confers twice as much risk as having only one copy. However, there are genes in which the action of one allele is dominant over the others, and in this case having either one or two copies of this allele results in the same outcome. This is called dominance genetic influence. MZ twins once again share all their dominance genetic influence, but DZ twins only share one-quarter. Genetic dominance results in increased similarity of MZ pairs compared to DZ pairs, because they share all their genetic dominance variance, whereas DZ twin pairs only share one quarter. Thus, whenever such a pattern of correlations is seen (with DZ correlations less than half those for the MZ) this indicates the presence of genetic dominance. Unfortunately, because we have only three observed statistics (variance in the phenotype, and the correlations for MZ and DZ twins), we can only estimate three parameters at any one time. We therefore tend to estimate dominance only when there is no evidence for shared environment (C2), and this parameter can therefore be dropped from the analyTable I. Some Examples of Different Pairs of Monozygotic and Dizygotic Twin Correlations and the Different Parameter Estimates Calculated from Thema TMZ

rDz

A2

C2

E2

.6

.3

.4 .6

.6 .4 .0

.0

.6 .6

.4 .4 .4

a

.2 .6

r Mz = monozygotic twin correlation, rDZ = dizygotic twin correlation, A2 = additive genetic influence, C2 = shared environment influence, E2 = nonshared environment influence.

ses and replaced by D2. Such analyses use the following equations:

Consider the example where the MZ correlation is .6, and the DZ correlation is .2. If we used Eq. (4) to calculate A2 we would obtain a value of .8, which is greater than the MZ correlation, and therefore not possible. In this case we therefore use the formulas given above which result in estimates of .2 for A2 (4 X .2 = .8 and .8 - .6 = .2), .4 for D2, and .4 for E2. The adoption design works from the same principles. Members of an adoptive family share their shared environment but not their genes. In contrast, members of nonadoptive biological families share their shared environment and 50% of their genes. The correlations between dyads within such families are calculated from these principles. For example, the correlation for a trait within a biological sibling pair, or biological parent-child pair is 1/2A2 + C2, the same as for a pair of DZ twins. In contrast, the correlation between adoptive siblings or an adoptive parent-child dyad is just C2 as they share no genes. The contrasting correlations between these two are used comparatively in just the same way as correlations between MZ and DZ twins. One of the advantages of the adoption study is the direct estimate of C2 obtained from adopted family members, as this parameter has to be inferred indirectly from twin data, and a direct estimate is thus less likely to be contaminated with error. Correlations can provide a rough estimate of heritability but cannot test whether each of the terms is significantly different from zero. Furthermore, they do not take into account the variance in the measures; they simply use the correlations alone. To take into account the variance within the measures, variancecovariance matrices are parameterized using modelfitting analysis (see Neale & Cardon, 1992). In such analyses the covariation within a twin pair is parameterized rather than just the correlation (which has been standardized removing all influence of variance). Figure 1 is a diagrammatic representation of Eqs. (1) to (3). The variance in each twin is accounted for by the paths from the three latent factors: A, C, and E. The correlation between the twins is accounted for by two sets of paths: genetic (A X 1.0 X A = A2

Behavioral Genetics: Anxiety and Depression in Children

23

In this way it is possible to ascertain which factors are significant in the etiology of the phenotype being studied. Two other fit indices commonly used are the comparative fit index which should be between 0.9 and 1.0 and Akaike's Information Criterion which should be large and negative. The model of best fit is said to be that which has the fewest paths in it, without significant deterioration in the fit of the model. Limitations Fig. 1. Model fitting with twin data.

for MZ pairs and A X 0.5 X A = 1/2A2 for DZ pairs); and shared environment (C X 1.0 X C = C2 for all twins). It can be seen from this figure that the correlation between MZ twins, for example, is therefore A2 + C2 as given in Eq. 2. Models are fitted to the variance-covariance matrices for the MZ and DZ pairs. The fit of the model is tested using chi-square statistics. Chi square is a test of the extent to which the observed data differ from the measured data. The conventional use of chisquare statistics is in two-group experimental designs, such as pre- and posttreatment. In such a design one would want to show group differences and would therefore look for large and significant chi-square values. In contrast, with model-fitting analyses the aim is to find a model that provides a good fit to the data (i.e., a model that explains the data well). If the model fits the data well, the chi-square values will be small and nonsignificant. In other words the model will not differ significantly from the data. The significance of any path can be tested by comparing the chi-square value of a model with that path to a model that does not include that path. For example, to establish whether the shared environment parameter is significant in the model, the full model is first run, which includes the C terms, followed by a reduced model without the C terms. In univariate analyses, an ACE model and an AE model differ in their degrees of freedom by one, as one path has been dropped. The chi-square value for the difference in fit between the two models is the difference between the ACE model chi-square and that for the AE model. If the difference in chi-square value between the two models is significant for 1 degree of freedom then the path is significantly different from zero and cannot be dropped from the model.

It should be noted that both the twin and adoption designs are not without problems. A recent article robustly defended the twin method as "the perfect natural experiment" (Martin, Boomsma, & Machin, 1997), while also presenting its main limitations. Limitations include the possibility that MZ twins encounter more similar environments to one another than DZ twins which would artificially inflate heritability estimates. However, a series of studies using a variety of methods appear to demonstrate that the more similar environments experienced by MZ pairs are due to their own genetic similarity which leads them to elicit or produce more similar environments (e.g., Kendler, Neale, Kessler, Heath, & Eaves, 1993a). Another feature of the twin design which might artificially inflate heritability is that of chorionicity. The chorion is the sack within the placenta. While two thirds of MZ twins are monochorionic (i.e., develop within one sack), all DZ twins are dichorionic (developing in separate sacks), and as such it could be argued that a large proportion of MZ twins have a more similar gestational environment to one another than DZ twins. However, monochorionicity also tends to lead to more of the problems associated with twin births such as shared infection, shared vasculature, and other anomalies of sharing a crowded chorion (Martin et al., 1997), although other studies indicate that this may not be the case (Christensen, Vaupel, Holm, & Yashlin, 1995), and as such the effects of chorionicity could go either way. Adoption studies are generally regarded as having fewer flaws and sources of uncertainty in the data than twin studies. However, there are three main difficulties with this design (see Plomin, DeFries, McClearn, & Rutter, 1997). The first is that the families involved in adoption (both biological and adoptive) may not be representative of the general population. Second, as the biological mothers provide the child's environment for 9 months this could result in

24

additional resemblance between the biological mother and her adopted-away child in addition to any genetically produced resemblance. This could artificially inflate heritability estimates. Finally, there is the problem of selective placement. This is the process by which attempts are made to match characteristics (such as SES, height, coloring, etc.) of the adoptive parents to those of the biological parents. This results in artificially high levels of similarity within the adoptive family for any traits on which selection occurred. As the only factor accounting for resemblance within the model for adoptive families is the shared environment (C2), selective placement results in inflated estimates of this parameter. However, most studies appear to show little evidence of selective placement (see Plomin et al., 1997). In summary, neither the twin design nor the adoption design is flawless. As they both have different strengths, the combination of the two makes a unique and valuable contribution to our understanding of human development. For example, the direct estimate of the shared environment from adoption studies is a positive addition to any results from twin studies where the shared environment is inferred and may be subject to errors.

TWIN AND ADOPTION STUDIES OF ANXIETY AND DEPRESSION IN CHILDREN AND ADOLESCENTS Over the last decade there has been a proliferation of twin, sibling, and adoption studies of anxiety and depression symptoms and disorders in children and adolescents. These studies have produced a wide variety of results, and their pattern is not entirely clear. These differences may be due to several factors associated with anxiety and depression during childhood and adolescence. The first and most obvious is that several different phenotypes have been examined in these studies. It is possible that different types of anxiety and different types of depression have different origins, in addition to any generic differences between anxiety and depression as broad categories. Second, it is possible that changing prevalence rates of these symptoms with age reflect developmental changes in etiology. This would mean that samples with differing age groups might be more likely to produce discrepant findings. Similarly, sex effects on etiology could also cause differences be-

Eley tween studies, especially if the sample is not large enough to analyze such effects specifically. Another difficulty observed in the design of such studies during childhood is the problem of who should be used as the rater. Child-reported symptoms may differ in several ways from those reported by a parent. Specifically, parents appear to be relatively poor at distinguishing between anxiety and depression in their children (Achenbach, 1991) and, as such, their ratings of anxiety and depression may not distinguish adequately between the two. If anxiety and depression symptoms differ considerably in their etiology, this could also lead to discrepancies within the group of studies of anxiety and within the group of studies of depression, in that individual studies may have dealt with this issue to differing extents. In other words parent-report measures may not tap as accurately into anxiety alone or depression alone as they appear to, and as such, if there are differences in the etiology of the two types of symptoms this may lead to inaccurate results. This leads to one final issue, the correlation between anxiety and depression, and the comorbidity seen between the two (Seligman & Ollendick, 1998). Given the high correlation between these two types of symptoms, it is not surprising that distinguishing between them, particularly via parent report, is difficult. Bivariate genetic analyses address this issue, by indicating whether it is predominantly genetic or environmental factors that influence both anxiety and depression that are responsible for their correlation. Finally, there may be study design effects in that the twin design may yield broadly different results from other designs such as the adoption design. This review therefore begins by considering the different phenotypes that have been considered in this area: fear and phobia symptoms, anxiety symptoms as assessed by questionnaire, mixed anxiety/ depression symptoms as assessed by a parent-report questionnaire, depression symptoms assessed by questionnaire (see Table II for a summary of the results regarding symptom measures), Separation Anxiety Disorder (SAD), Overanxious Disorder (OAD), and Major Depressive Disorder (MDD) (see Table III for a summary of the results regarding symptom measures). This is followed by sections documenting results on age-related change, sex differences, rater effects, and the correlation between anxiety and depression. The samples in all of the twin studies reviewed here consisted of roughly half MZ pairs and half DZ same-sex pairs so only total sample size is given.

Behavioral Genetics: Anxiety and Depression in Children

25

Anxiety and Depression Phenotypes

Anxiety Symptoms

Fear and Phobia Symptoms

The first study of the full range of anxiety symptoms involved 376 pairs of twins ages 8 to 16 years (Thapar & McGuffin, 1994). In this study a parent-report anxiety questionnaire was completed for the whole sample, and the adolescents (ages 12-16 years) also completed a self-report questionnaire. Results from these two data sets were very different. The parent report of anxious symptomatology was found to have an estimated heritability of 59% with no age effect, whereas the adolescent self-report measure had no significant genetic component. The shared environment showed the reverse pattern, being large and significant for the adolescent self-report measures only. Among the authors' hypothesized reasons for this discrepancy was the suggestion that the parents are rating an enduring trait, whereas the adolescents' self-report may reflect current state. This possibility could be tested by using the Spielberger State-Trait Anxiety Inventory for Children (STAIC; Spielberger, 1973) which specifically measures these two aspects of anxiety symptoms separately. One of the largest studies in this area is the Virginia Twin Study of Adolescent Behavioral Development (VTSABD). This study is of 1,412 samesex twin pairs ages 8 to 16 years. There are two papers that present results on anxiety from this study that give somewhat different results. The first paper presented child-reported data only (Topolski et al., 1997) and included results for the Revised Children's Manifest Anxiety Scale (R-CMAS; Reynolds & Richmond, 1979), a measure of self-reported anxiety symptoms. The etiology of the questionnaire symptoms showed significant sex and age effects (described later), but overall this study indicated both shared environment and genetic influence on anxiety symptoms. The second paper from the VTSABD (Eaves et al., 1997) included mother-, father-, and self-reported anxiety symptoms, as assessed by a factor score created from questionnaire-rated anxiety symptoms. Overall these results indicated moderate genetic influence, with no significant influence of shared environment (except for child-reported questionnaire symptoms, for which the shared environment accounted for 33% of the variance in males). The differences between these two papers are likely to be due to the use of different formulations of both symptom measures and diagnoses. It is unclear at this time the precise reason for these differences, and as this was the

The earliest twin study of anxiety used the Fear Survey Schedule for Children-Revised (FSSC-R: Ollendick, 1983), a self-report measure of specific fear or phobia symptoms in 319 same-sex twin-pairs aged 8 to 18 years (Stevenson, Batten, & Cherner, 1992). The total fear score was found to have moderate heritability (A2 = .29), with some specific fear factors showing higher heritability (e.g., fear of the unknown, A2 = .46) and some showing negligible heritability (e.g., fear of medical procedures, A2 = . 14). The influence of shared environment was moderate and significant for all types of fear symptom (C2 = .23 to .59). The nonshared environment factor was significant for all areas, but was particularly important for fear of medical procedures (E2 = .47), and the authors suggest that this could be because such experiences are by their nature child specific. If so, the environmental factors responsible for the fear they produce would be due in large part to nonshared environmental influences. Fear of failure, which as the authors noted, is the fear dimension most closely related to social fear, has a large shared environment factor (C2 = .51), which suggests that such fears may be learned in the family setting. The authors also examined the influence of genetic and environmental factors on very high fear scores and found that their influences were of very similar magnitude to those found for individual differences in the normal range. This suggests that the same etiological influences may account for both normal and disorder levels of symptoms. A subsequent study of parent-rated levels of DSM-IV fear and phobia symptoms in 1,106 twin pairs, 7-9 years old, found a significant role for both genetic and shared environment influences (Lichtenstein & Annas, 1997). Heritability estimates for animal, situational, and mutilation fears ranged from 16 to 55%, with shared environment accounting for between 31 and 60% of the variance. The finding of significant influence of shared environment is relatively rare in twin studies of psychopathology, and has been absent or minimal in studies of the etiology of adult phobias (Kendler, Neale, Kessler, Heath, & Eaves, 1992b; Kendler et al., 1995). This makes it particularly interesting that such influences are significant in predicting variance in the normal range of fear symptoms in children and adolescents. The specific factors this shared environment influence consists of are discussed below.

26

Eley

Behavioral Genetics: Anxiety and Depression in Children

27

Eley

28

first study to assess the etiology of anxiety and depression in children and adolescents at disorder level, replication is needed before firm conclusions can be drawn. Finally, a recent study of 395 same-sex child twin pairs ages 8-16 years examined the etiology of a factor score of anxiety that had been created specifically to reduce the level of depressive symptomatology included in the measure (Eley & Stevenson, in press). Purer measures of both anxiety and depression were created by using factor analysis of items from two common self-report measures that correlate highly, partly due to symptom overlap between the measures. The initial stage entered all items from the Children's Depression Inventory (GDI; Kovacs, 1985) and all items from the Trait scale of the STAIC (Spielberger, 1973) into a factor analysis. Ten factors from this analysis were then subjected to a further factor analysis which resulted in just two second-order factors: one representing anxiety, the other depression. The correlation between these two factors was .27 as compared to a correlation of .68 between the original scales. The heritability of this measure of anxiety was found to be low (A2 = .10) with moderate shared environment influence (C2 = .39). In summary, as with the two studies of fear symptoms, these studies indicate a role for shared environment in anxiety symptoms, although this is greater in some studies than others. There are also indications of genetic influence, which varies rather more greatly across studies. Nonshared environment is a moderately large and significant influence on these measures. Anxiety/Depression Symptoms Ratings by Parents The first adoption study to examine anxiety and depression in children presented data from 111 adopted sibling pairs and 221 unrelated adopted sibling pairs ages 10 to 15 years (van den Oord, Boomsma, & Verhulst, 1994). The measure used was the Anxious/Depressed subscale of the Child Behavior Checklist (CBCL; Achenbach, 1991). The heritability of this measure was low (h2 = .13), but the shared environment accounted for around one third of the variance (c2 = .31). More recently, the same measure was used in a twin study of 181 same-sex twin pairs ages 7-15 years (Edelbrock, Rende, Plomin, & Thompson, 1995). This study also found that the shared environment accounted for around one third

of the variance (c2 = .31), but in addition, genetic factors accounted for a further third (h2 = .34). The third study to use this type of measure is a sibling-cousin study, which reported data on 436 pairs of full siblings, 119 pairs of half siblings, and 122 pairs of first cousins (all pairs were same-sex), ages 4 to 10 years (van den Oord & Rowe, 1997). The measure used was the Anxious/Depressed subscale from the Behavior Problems Index (BPI; Peterson & Zill, 1986), a short-form of the CBCL (Achenbach, 1991). This study found less familiality (within pair resemblance or combined genetic and shared environment influences) than the other two studies, with genetic factors accounting for about one quarter of the variance and shared environment factors around one fifth. Finally, parent-reported anxiety symptoms as assessed by the Anxious/Depressed subscale of the CBCL (Achenbach, 1991) were also collected on 395 same-sex twin pairs ages 8 to 16 years (Eley, 1996). Genetic factors accounted for almost half the variance in this measure, with negligible and nonsignificant shared environment influence. Although these studies assessed mixed anxiety/ depression, they generally provide further evidence for shared environment influence on anxiety in children, with genetic factors also playing a role. However, the results also indicate great variability across studies. Depression Symptoms There are now at least eight published papers reporting genetic and environmental parameter estimates for depression symptoms in children and adolescents, although the first of these had too small a sample to warrant interpretation (Wierzbicki, 1987). The first results using an adequate sample came from the study of Nonshared Environment and Adolescent Development (NEAD), which includes 707 same-sex pairs of twins, full-, half-, and unrelated-siblings in both intact and broken families (Rende, Plomin, Reiss, & Hetherington, 1993). This study gave a heritability estimate for self-reported depressive symptoms of .34, with negligible shared environment. Following this, self-reported data from 108 adolescent twin pairs produced the very high heritability estimate of .76, with no shared environment (Thapar & McGuffin, 1994), and parent-reported symptoms for the entire child and adolescent sample (N = 218 pairs) gave a heritability estimate of .48.

Behavioral Genetics: Anxiety and Depression in Children A very similar heritability estimate of .46 with no shared environment was obtained for children ages 7-12 years within a study of early and middle childhood (Murray & Sines, 1996). Self-reported data from 395 same-sex pairs ages 8 to 16 years again produced a similar heritability estimate of .48, with low and nonsignificant shared environment (Eley, 1997a). The heritability estimates from the VTSABD were varied, ranging from .1 for childreported symptoms to .7 for father-reported symptoms (Eaves et al., 1997). These differences appear to be largely due to rater effects which are discussed below. Results from the only adoption study to include self-reported depression symptoms indicated negligible genetic or shared environment influences (Eley, Deater-Deckard, Fombonne, Fulker, & Plomin, 1998). Finally, more recent results from the NEAD indicate moderate genetic influence with nonsignificant shared environment for a composite measure including observation scores in addition to mother-, father-, and self-report (O'Connor, Neiderhiser, Reiss, Hetherington, & Plomin, 1998). Two studies explored the etiology of extreme depression (high scores) as compared to individual differences in the normal range (Eley, 1997a; Rende et al., 1993). The results were similar, with the extreme group heritability estimates being around .25, and extreme group shared environment around .3. Although the shared environment influence of .3 was greater at the extremes than that found in studies of depression symptoms in the normal range, this difference was not significant due to lack of statistical power. Genetic influences did not differ for the extremes as compared to the normal range, suggesting that the magnitude of genetic influence is no different for very high depression to normal levels of depression. Overall these data indicate moderate genetic influence on depression symptom scores in children and adolescents, with negligible shared environment, although there is some evidence for shared environment on very high depression scores. Nonshared environment influence is also a significant contributor to variance in these measures. Separation Anxiety Disorder (SAD) The two papers from the VTSABD also present data on SAD (Topolski et al., 1997; Eaves et al., 1997). In the first paper, Topolski et al. (1997) looked at self-report measures only, variance in child-reported

29

SAD was due to shared and nonshared environment with no significant influence of genes. In contrast, in the second paper, Eaves et al. (1997), reporting on mother-, father-, and self-reported children's anxiety, child-reported SAD was influenced by additive genetic and nonshared environment factors. Motherand father-reported SAD showed very high nonshared environment influence, with little genetic or shared environmental variance. Thus, in the latter paper, shared environment was not significant in predicting SAD, for all three sources, including childreport in contrast to the child-reported results in the earlier paper. There are no clear reasons for this discrepancy, apart from the probable use of different formulations of the diagnosis of SAD in the two papers. Another large study of 2,043 same-sex twin pairs ages 3-18 years found both age and sex effects on maternally rated SAD (Feigon, Waldman, Levy, & Hay, 1997, in press). The results indicated both shared environmental and genetic influences on SAD. Similar results were found for retrospectively reported separation anxiety symptoms in a study of 200 same-sex twin pairs ages 17 to 66 years (Silove, Manicavasagar, O'Connell, & Morris-Yates, 1995). Although this study had a very biased sample (recruited from media advertising), it adds to the findings on sex differences in SAD which are discussed in more detail below. The results for SAD generally indicate that the familiality of this disorder results from both genetic and shared environment influences. Nonshared environment is also a significant influence. These studies further demonstrate that anxiety in children and adolescents, unlike most other behavioral phenotypes, is influenced by the shared environment. Overanxious Disorder (OAD) OAD was only looked at in the VTSABD study (Eaves et al., 1997; Topolski et al., 1997). The results for this disorder are rather more similar across the two papers than those for SAD, and indicate additive genetic and nonshared environment as significant influences on OAD in children and adolescents, for mother-, father-, and self-reported data (Eaves et al., 1997; Topolski et al., 1997). Major Depressive Disorder (MDD) The last phenotype to have been considered is MDD which was only looked at in the VTSABD

30

Eley

Behavioral Genetics: Anxiety and Depression in Children (Eaves et al., 1997; Topolski et al., 1997). The etiology of this disorder varied considerably depending both on the rater and the sex of the child, and these influences are discussed in more detail below. Considered together the results indicate moderate genetic influence on MDD, with no influence of the shared environment. In summary, the etiology of anxiety and depression in children differs somewhat according to the definition. There have been too few studies in this area to draw clear distinctions between these different phenotypes, but there are some indications that while nonshared environment is of roughly similar magnitude for both SAD and OAD, SAD may be more influenced by the family environment than OAD. This finding could perhaps be interpreted in the light of the role attachment may play in SAD (Warren, Huston, Egeland, & Sroufe, 1997). However, it should be noted that the questionnaire measures also revealed significant influence of shared environment, and these questionnaires are meant to tap the kinds of symptoms that would form the basis of a diagnosis such as OAD. Furthermore, overall it appears that while shared environment is important for anxiety it has a negligible role for depression. However, it is clear from these results that the quantification of genetic and environmental influences varies considerably depending on several aspects of the design of the study, which are next explored in greater depth. Rater Effects Child Report Several of the studies discussed above used the child as the reporter, sometimes with parent-report in addition (Eaves et al., 1997; Eley & Stevenson, in press; Stevenson et al., 1992; Thapar & McGuffin, 1995; Topolski et al., 1997). It is difficult to conclude whether any differences between child report and parent report are due to the reporter, as they are confounded by age—child report being used for children 8 years or older only. However, in general the child data on anxiety indicate roughly equal contributions of genes and shared environment, although there are some indications that shared environment influences may account for slightly more variance. Between them the two account for around one half of the variance, with the remainder due to nonshared environment and error. In contrast, for depression,

31

the heritability of the self-reported data are highly inconsistent, with estimates varying from .00 to .76, although shared environment remains low and negligible in all studies. Nonshared environment is a strong and significant influence on depression in all reported studies. In summary, the child-reported data are relatively inconsistent which may be indicative of the difficulty of obtaining self-reported data from children as young as 8 years. Further work on appropriate measures will help in this area. Parent Report Most of the studies cited have included parent report, whether or not they also ascertained child report (Eaves et al., 1997; Edelbrock et al., 1995; Eley, 1996; Feigon et al., 1997, in press; Lichtenstein & Annas, 1997; Thapar & McGuffin, 1995; van den Oord et al., 1994; van den Oord & Rowe, 1997). For anxiety, the results for the whole group vary considerably but overall, as with the child report, they suggest that genetic and shared environment factors account for around one half of the variance in anxiety between them. However, for the parentreported data it looks as if genetic influences may be slightly greater than shared environment, which is particularly interesting given that rater effects due to one parent rating two children would inflate the shared environment rather than the genetic parameter (it would tend to inflate both MZ and DZ correlations). This suggests that parents and children may be rating somewhat different phenomena, as seen by the very low correlations between parent and child ratings of anxiety (commonly in the .3 range), a finding demonstrated in several nongenetic studies of anxiety in children as well (e.g., Engel, Rodrigue, & Geffken, 1994). For depression, the results are rather more consistent than those from the child-reported data, indicating heritability in the .5 to .7 range in almost all studies with negligible shared environment. Age Effects Three studies directly tested for and demonstrated age effects in anxiety. The first paper from the VTSABD found both sex and age effects on the RCMAS (Reynolds & Richmond, 1979) (Topolski et al., 1997). Results indicated higher heritability of anxiety symptoms in early adolescence (age 11-13

Eley

32

years) as compared to late childhood (8-10 years) or midadolescence (14-16 years) for the girls, and decreasing familiality (both genetic and shared environment influences) over time for the boys. Second, was the study of SAD in 2,043 same-sex twin pairs ages 3 to 18 years (Feigon et al., 1997, in press). In this study heritability was found to increase with age. The third study also found both sex and age effects (Eley & Stevenson, 1998b) on child-reported anxiety symptoms. In this sample, heritability was higher in adolescents (12-16 years) as compared to children (8-11 years), particularly for girls. At least two other studies tested for but did not find age effects (Eaves et al., 1997; Thapar & McGuffin, 1995). There is some evidence from these results for agerelated changes in the etiology of anxiety symptoms. It appears that heritability of anxiety increases with age, though this may be over a discrete period during adolescence. There are three papers that indicate increasing heritability of depression with age, with an accompanying decrease in shared environment. The first is a study of parent-reported depressive symptoms in 364 same-sex twin pairs ages 4 to 6 years or 7 to 12 years (Murray & Sines, 1996) which found heritability and shared environment of .46 and .00 for the older group and .00 and .29 for the younger group, respectively. However, it should be noted that these differences were not formally tested. In the second study, heritability and shared environment estimates for parentreported depressive symptoms were .18 and .60, respectively, in 114 child twins pairs (ages 8-11 years), and .78 and .04 for the 108 adolescents (12-16 years) (Thapar & McGuffin, 1994). The third study found both sex and age effects (Eley & Stevenson, 1998b) on child-reported depression symptoms. In this sample heritability was higher in adolescents (12-16 years) as compared to children (8-11 years), particularly for girls. Although these age-related changes are interesting, and relatively consistent with increasing heritability with age, in order to clarify exactly what is occurring a more systematic approach to the issue is needed. The studies cited above used different methods, and, notably, several used very different criteria for dividing samples into different age groups (Topolski et al., 1997). However, given the changes seen during adolescence and puberty, many of which are governed by genes, the hypothesis that influences on psychological dimensions such as anxiety may increase over this period is highly plausible and the small amount of evidence available suggests this may

be the case. Longitudinal studies, which are able to establish the causes of continuity and change in symptomatology over time, will be essential in understanding the developmental psychopathology of anxiety. Sex Effects Sex effects on anxiety have been addressed and identified in six of the studies reviewed here (Eaves et al., 1997; Silove et al., 1995; Eley & Stevenson, 1998b; Feigon et al., 1997, in press; Lichtenstein & Annas, 1997; Topolski et al., 1997). Of these, five found greater heritability estimates for girls as compared to boys (Eaves et al., 1997; Eley & Stevenson, 1998b; Feigon et al., 1997, in press; Silove et al., 1995; Topolski et al., 1997). This is the most consistent result yet to appear from the literature on behavioral genetic analyses of anxiety in children, and is one which may have implications for molecular genetic research in anxiety. For example, there may be greater successes in identifying genes for anxiety if a female sample is used. Furthermore, any genes found to influence anxiety may have a greater effect in females than they do in males. Far fewer studies of depression have explored sex differences. The VTSABD (N = 1,412 samesex twin pairs ages 8-16 years) found only minor differences between the parameter estimates for boys and girls (Eaves et al., 1997), and just one of these reached statistical significance (father-reported major depressive disorder was more heritable in boys than in girls). A smaller study (N = 364 same-sex twin pairs) also found higher heritabilities for parentreported depression symptoms in boys than in girls, but no statistical test was made to establish whether this difference was significant (Murray & Sines, 1996). Finally, in exploring both age and sex effects, a recent twin study of 496 pairs ages 8 to 16 years obtained higher heritability estimates for boys as compared to girls, especially in male adolescents (Eley & Stevenson, 1998b). Shared environment influence was moderate in the children, and highest in the adolescent females (c2 = .56). The model fitted the data significantly better when allowed to differ for all four groups (Eley & Stevenson, 1998b). These studies provide modest evidence for increased heritability of depression symptoms for boys as compared to girls. However, the data can only be considered exploratory, and given the apparent difference between age-related changes in the etiology of anxiety as com-

Behavioral Genetics: Anxiety and Depression in Children pared to depression in females, the results need replication and clarification.

33

tently found that the genetic influences on anxiety and depression, both as symptoms and disorders, are almost entirely shared, in contrast to the environmental influences which are largely specific.

Comorbidity Anxiety and Depression Two studies have addressed the origin of the correlation between anxiety and depression in children within a behavioral genetic design. The first investigated the causes of covariation between maternally reported anxiety and depression symptoms in 172 twin pairs ages 8 to 16 years (Thapar & McGuffin, 1997). This study found that most of the covariation between anxiety and depression symptoms could be explained by genetic factors that influenced both. There was some specific genetic influence on depression, but most of the genetic variance was shared. Some of the covariation was also accounted for by nonshared environment influences relating to both measures, but this could have been due to correlated measurement error. Questionnaire measures of anxiety and depression in children tend to include some overlapping items which produces artificial inflation of the correlation between the scales (Seligman & Ollendick, 1998). A more recent study using data from a sample of 490 child twin pairs ages 8-16 years was designed specifically to reduce artifactual covariation between the measures of anxiety and depression (Eley, 1997b; Eley & Stevenson, in press). A factor analytic procedure was used to produce relatively uncorrelated anxiety and depression scores (r = .27), as compared to the high correlation found between the total scale scores for the two questionnaire measures used (r = .67). The production of these purer factors allowed for the identification of shared and specific etiological factors, having removed artifactual overlap between the measures. Bivariate genetic analyses indicated that all genetic variance was shared across the two measures, and this shared genetic factor accounted for 80% of the phenotypic correlation. In other words, all genes influencing anxiety in this population were also influencing depression, and these shared genetic influences accounted for 80% of the association between the two measures. In contrast, the environmental factors were almost entirely measure specific. These results are in line with research in adults (Kendler, Heath, Martin, & Eaves, 1987; Kendler, Neale, Kessler, Heath, & Eaves, 1992a; Roy, Neale, Pedersen, Mathe, & Kendler, 1995), which has consis-

CONCLUSIONS Genes, Anxiety and Depression in Children One clear finding from these studies is that there is genetic influence on both anxiety and depression in childhood, which accounts for around one third of the variance in most cases. However, this contribution differs across definitions and raters. It also appears that the genetic contribution may increase with age. Furthermore, for anxiety, heritability appears to be greater for girls than boys, with the reverse pattern for depression. This suggests that molecular genetic research needs to consider carefully the groups used in designing studies aimed at identifying specific genetic risk factors. For example, it may be that adolescent girls provide the best chance of detecting genes for anxiety, as this is the group for whom genetic factors appear to have the greatest influence on anxiety. However, more evidence on these issues is required before firm conclusions of this kind can be drawn. Another hypothesis relevant to molecular genetic research is that when genes are found for anxiety they will tend to be associated also with depression. In the child literature the high genetic correlation between anxiety and depression has only been shown with regard to questionnaire measures of symptoms and needs to be extended to disorder. However, adult work suggests that the same situation will be found there, although in adults the genetic correlation varies by anxiety disorder type (Kendler et al., 1992a; Kendler, Neale, Kessler, Heath, & Eaves, 1993b; Kendler et al., 1995; van den Oord & Rowe, 1997). Finally, the recent publication of an association between a marker in the promoter of the serotonin transporter gene and the emotional triad of anxiety, depression, and neuroticism offers some support for this hypothesis (Lesch et al., 1996). This study indicated that the relationship between a specific genetic marker held not only for anxiety but also for depression and for neuroticism. This is a clear example of shared genetic influence on both depression and anxiety, as demonstrated in twin studies.

34

Shared Environment and Anxiety in Children The clearest findings in this review of the literature are that shared environment plays little or no role in depression, whereas there is significant shared environment influence on most definitions of anxiety in children and adolescents. This is particularly exciting given the rarity with which this parameter has been found to be significant for most other behavioral disorders, in both childhood and adult disorders. In particular, adult studies of anxiety disorders have been notable in the lack of significant shared environment (e.g., Kendler et al., 1992b). This emphasizes the need for separate research into anxiety disorders in children and adolescents where the shared environment has been demonstrated to be a significant influence. Shared environment influences are defined as those that make family members resemble one another. It is therefore likely that a significant proportion of shared environment influences consist of shared family experiences during childhood, and that, as such, these have greater effect during childhood than in later stages of life. There are several published studies that investigate the role of environmental stressors in the development of anxiety during childhood. Perhaps the most obvious candidate for a shared environment factor is maternal psychopathology. The difficulty with this is that it is unclear from current research whether the high incidence of anxiety in children of anxious or depressed mothers is due to shared genes or shared environmental influences. Studies are needed of biological and adoptive mothers of children with high anxiety in order to unravel this relationship. However, there is considerable evidence for relationships between several possible environmental mediators of parental psychopathology and anxiety in the child. For example, insecure attachment has been shown to be more prevalent in infants of depressed mothers (Radke-Yarrow, Cummings, Kuczynski, & Chapman, 1985). Moreover it has also been shown to be associated with anxiety and depression in the child (Armsden, McCauley, Greenburg, Burke, & Mitchell, 1990). Furthermore, the absence of a good confiding relationship in a mother's life is associated with emotional disorders in children (Goodyer, Wright, & Altham, 1988). Undesirable parenting, such as inconsistent or restrictive patterns of behavior are also associated with anxiety in children (Kohlmann, Schumacher, & Streit, 1988; Krohne & Hock, 1991). Finally, there is also considerable evidence that early childhood bereavement predicts later anxi-

Eley ety (e.g., Kranzler, Shaffer, Wasserman, & Davies, 1989; Goodyer & Altham, 1991; Goodyer, Germany, Gowrusankur, & Altham, 1991), an effect which may itself be mediated by parental psychopathology or altered parenting style. Divorce and other life-changing events, especially those involving separation from the mother (e.g., long-term hospitalization for illness) have also been associated with later anxiety (see Goodyer, 1990, for a review). Finally, demographic factors such as poor housing or poverty may be relevant. It has been hypothesized that adverse physical environments may increase the prevalence of life events that are known to be associated with anxiety (Goodyer, 1990). Nonshared Environment, Anxiety and Depression in Children The literature on environmental risk factors for anxiety and depression includes some candidates that would act as nonshared environment factors. For example, recent stressful life events, which have been demonstrated to be associated with emotional disorders including anxiety and depression, are likely to be child-specific (Goodyer, Kolvin, & Gatzanis, 1987). More specifically, friendships and the absence of school-based achievements have also been shown to be related to emotional disorders (Goodyer, Wright, & Altham, 1990). One particularly interesting result described above is that the environmental influences on anxiety and depression appear to be symptom-specific. Much of the research on environmental risks described above relates both to anxiety and depression. However, one study of adults found a specific relationship between risks characterized by loss with depression as the outcome, and events characterized by threat of future danger, which were followed by an onset of anxiety (Finlay-Jones & Brown, 1981). A study of child and adolescent twin pairs in which at least one of the twins was highly anxious or highly depressed attempted to replicate this finding (Eley, 1996; Eley & Stevenson, 1998a). Events characterized by danger (or threat) were distinguished from those characterized by loss. Danger events independent of the behavior of the child were of higher prevalence in anxious children as compared both to their nonanxious cotwins and to other nonanxious children. Loss events independent of the behavior of the child were found to be more prevalent in depressed children as compared to nondepressed children. Furthermore,

Behavioral Genetics: Anxiety and Depression in Children chronic experiences of 4 weeks or more characterized as schoolwork stressors, family relationship problems, or friendship difficulties were all significantly more common in depressed than nondepressed children. This study was the first to specify the associations between life events, anxiety, and depression in children and adolescents. In summary, the application of behavioral genetic designs to the study of anxiety and depression in children and adolescents is beginning to provide some interesting insights into the relative importance of genetic and environmental influences. This approach is not limited merely to the identification of genetic influence on a phenotype, but is also able to clarify the role of environmental risk factors, some of which may act in conjunction with genetic influences. These methods are also central to identifying how the influences on anxiety differ between the two sexes, and to exploring the etiology of continuity and change in symptoms with age and over time. The combination of behavioral genetics and developmental psychopathology will in time lead to far greater understanding of the origin and development of anxiety and depression in children and adolescents.

REFERENCES Achenbach, T. M. (1991). Manual for the Child Behavior Checklist and 1991 profile. Burlington: University of Vermont, Department of Psychiatry. Armsden, G. C., McCauley, E., Greenburg, M. T., Burke, P. M., & Mitchell, J. R. (1990). Parent and peer attachment in early adolescent depression. Journal of Abnormal Child Psychology, 18, 683-697. Christensen, K., Vaupel, J. W., Holm, N. V., & Yashlin, A. I. (1995). Mortality among twins after age 6: foetal origins hypothesis versus twin method. British Medical Journal, 310, 432-436. Costello, A., & Angold, A. (1988). Scales to assess child and adolescent depression: Checklists, screens, and nets. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 726-737. Eaves, L. J., Silberg, J. L., Meyer, J. M., Maes, H. H., Simonoff, E., Pickles, A. , Rutter, M., Neale, M. C., Reynolds, C. A., Erickson, M. T., Heath, A. C., Loeber, R., Truett, K. R., & Hewitt, J. K. (1997). Genetics and developmental psycopathology: 2. The main effects of genes and environment on behavioral problems in the Virginia Twin Study of Adolescent Behavioral Development. Journal of Child Psychology and Psychiatry, 38, 965-980. Edelbrock, C., Rende, R. D., Plomin, R., & Thompson, L. A. (1995). A twin study of competence and problem behavior in childhood and early adolescence. Journal of Child Psychology and Psychiatry, 36, 775-785. Eley, T. C. (1996). The aetiology of emotional symptoms in children and adolescents: Depression and anxiety in twins. Doctoral thesis, University of London. Eley, T. C. (1997a). Depressive symptoms in children and adolescents: etiological links between normality and abnormality: a

35

research note. Journal of Child Psychology and Psychiatry, 38, 861-866. Eley, T. C. (1997b). General genes: A new theme in developmental psychopathology. Current Directions in Psychological Science, 6, 90-95. Eley, T. C., Deater-Deckard, K., Fombonne, E., Fulker, D. W., & Plomin, R. (1998). An adoption study of depressive symptoms in middle childhood. Journal of Child Psychology and Psychiatry, 39, 337-345. Eley, T. C., & Stevenson, J. (1998a). Specific life events and chronic experiences differentially associated with depression and anxiety in young twins. Submitted for publication. Eley, T. C., & Stevenson, J. (1998b). Using genetic analyses to explore the covariation between anxiety and depression symptoms: Effects of age and sex. Submitted for publication. Eley, T. C., & Stevenson, J. (in press). Using genetic analyses to clarify the distinction between depressive and anxious symptoms in children and adolescents. Journal of Abnormal Child Psychology. Engel, N. A., Rodrigue, J. R., & Geffken, G. R. (1994). Parent-child agreement on ratings of anxiety in children. Psychological Reports, 75, 1251-1256. Feigon, S. A., Waldman, I. D., Levy, F., & Hay, D. A. (1997). Genetic and environmental influences on various anxiety disorder symptoms in children [Abstract]. Behavior Genetics, 27, 588. Feigon, S. A., Waldman, I. D., Levy, F., & Hay, D. A. (in press). Genetic and environmental influences on separation anxiety disorder symptoms and their moderation by age and sex. Journal of Abnormal Psychology. Finlay-Jones, R., & Brown, G. W. (1981). Types of stressful life events and the onset of anxiety and depressive disorders. Psychological Medicine, 11, 803-815. Goodyer, I. M. (1990). Family relationships, life events and childhood psychopathology. Journal of Child Psychology and Psychiatry, 31, 161-192. Goodyer, I. M., & Altham, P. M. E. (1991). Lifetime exit events and recent social and family adversities in anxious and depressed school-age children and adolescents-I. Journal of Affective Disorders, 21, 219-228. Goodyer, I. M., Germany, E., Gowrusankur, J., & Altham, P. M. E. (1991). Social influences on the course of anxious and depressive disorders in school-age children. British Journal of Psychiatry, 158, 676-684. Goodyer, I. M., Kolvin, I., & Gatzanis, S. (1987). The impact of recent undesirable life events on psychiatric disorders in childhood and adolescence. British Journal of Psychiatry, 151, 179-184. Goodyer, I. M., Wright, C., & Altham, P. M. E. (1988). Maternal adversity and recent stressful life events in anxious and depressed children. Journal of Child Psychology and Psychiatry, 29, 651-667. Goodyer, I. M., Wright, C., & Altham, P. M. E. (1990). Recent achievements and adversities in anxious and depressed school age children. Journal of Child Psychology and Psychiatry, 31, 1063-1077. Kendler, K. S., Heath, A. C., Martin, N. G., & Eaves, L. J. (1987). Symptoms of anxiety and depression: Same genes, different environments? Archives of General Psychiatry, 44, 451-457. Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1992a). Major depression and generalized anxiety disorder. Same genes, (partly) different environments? Archives of General Psychiatry, 49, 716-722. Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1992b). The genetic epidemiology of phobias in women: The interrelationship of agoraphobia, social phobia, situational phobia, and simple phobia. Archives of General Psychiatry, 49, 273-281.

36 Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1993a). A test of the equal-environment assumption in twin studies of psychiatric illness. Behavior Genetics, 23, 21-27. Kendler, K. S., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1993b). Major depression and phobias: the genetic and environmental sources of comorbidity. Psychological Medicine, 23, 361-371. Kendler, K. S., Walters, E. E., Neale, M. C., Kessler, R. C., Heath, A. C., & Eaves, L. J. (1995). The structure of the genetic and environmental risk factors for six major psychiatric disorders in women: Phobia, generalized anxiety disorder, panic disorder, bulimia, major depression, and alcoholism. Archives of General Psychiatry, 52, 374-383. Kohlmann, C. W., Schumacher, A., & Streit, R. (1988). Trait anxiety and parental child-rearing behaviour: Support as a moderator variable? Anxiety Research, 1, 53-64. Kovacs, M. (1985). The Children's Depression Inventory (CDI). Psychopharmacology Bulletin, 21, 995-1124. Kranzler, E. M., Shaffer, D., Wasserman, G., & Davies, M. (1989). Early childhood bereavement. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 513-520. Krohne, H. W., & Hock, M. (1991). Relationships between restrictive mother-child interactions and anxiety of the child. Anxiety Research, 4, 109-124. Lesch, K. P., Bengel, D., Heils, A., Zhang Sabol, S., Greenburg, B. D., Petri, S., Benjamin, J., Muller, C. R., Hamer, D. H., & Murphy, D. L. (1996). Association of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory region. Science, 274, 1527-1530. Lichtenstein, P., & Annas, P. (1997). The heritability of specific fears and phobias in children [Abstract]. Behavior Genetics, 27, 598. Martin, N., Boomsma, D., & Machin, G. (1997). A twin-pronged attack on complex trait. Nature Genetics, 17, 387-392. Murray, K. T., & Sines, J. O. (1996). Parsing the genetic and nongenetic variance in children's depressive behavior. Journal of Affective Disorders, 38, 23-34. Neale, M. C., & Cardon, L. R. (1992). Methodology for genetic studies of twins and families. Dordrecht, The Netherlands: Kluwer. O'Connor, T. G., Neiderhiser, J. M., Reiss, D., Hetherington, E. M., & Plomin, R. (1998). Genetic contributions to continuity, change, and co-occurrence of antisocial and depressive symptoms in adolescence. Journal of Child Psychology and Psychiatry, 39, 323-336. Ollendick, T. H. (1983). Reliability and validity of the Revised Fear Survey Schedule for Children (FSSC-R). Behaviour Research and Therapy, 21, 685-692. Peterson, J. L., & Zill, N. (1986). Marital disruption, parent-child relationship, and behavior problems in children. Journal of Marriage and the Family, 48, 295-307. Plomin, R., DeFries, J. C., McClearn, G. E., & Rutter, M. (1997). Behavioral Genetics (3rd ed.). New York: W. H. Freeman. Radke-Yarrow, M., Cummings, E. M., Kuczynski, L., & Chapman, M. (1985). Patterns of attachment in two- and three-year-olds in normal families and families with parental depression. Child Development, 56, 884-893.

Eley Rende, R. D., Plomin, R., Reiss, D., & Hetherington, E. M. (1993). Genetic and environmental influences on depressive symptomatology in adolescence: Individual differences and extreme scores. Journal of Child Psychology and Psychiatry, 34, 1387-1398. Reynolds, C. R., & Richmond, B. O. (1979). Factor structure and construct validity of 'What I Think and Feel': The Revised Children's Manifest Anxiety Scale. Journal of Personality Assessment, 43, 281-283. Roy, M. A., Neale, M. C., Pedersen, N. L., Mathe, A. A., & Kendler, K. S. (1995). A twin study of generalized anxiety disorder and major depression. Psychological Medicine, 25, 1037-1049. Seligman, L. D., & Ollendick, T. H. (1998). Comorbidity of anxiety and depression in children and adolescents: An integrative review. Clinical Child and Family Psychology Review, 1, 125-144. Silove, D., Manicavasagar, V., O'Connell, D., & Morris-Yates, A. (1995). Genetic factors in early separation anxiety: Implications for the genesis of adult anxiety disorders. Acta Psychiatrica Scandinavica, 92, 17-24. Spielberger, C. (1973). Preliminary test manual for the State-Trait Anxiety Inventory for Children. Palo Alto, CA: Consulting Psychologists Press. Stevenson, J., Batten, N., & Cherner, M. (1992). Fears and fearfulness in children and adolescents: a genetic analysis of twin data. Journal of Child Psychology and Psychiatry, 33, 977-985. Thapar, A., & McGuffm, P. (1994). A twin study of depressive symptoms in childhood. British Journal of Psychiatry, 165, 259-265. Thapar, A., & McGuffin, P. (1995). Are anxiety symptoms in childhood heritable? Journal of Child Psychology and Psychiatry, 36, 439-447. Thapar, A., & McGuffin, P. (1997). Anxiety and depressive symptoms in childhood—a genetic study of comorbidity. Journal of Child Psychology and Psychiatry, 38, 651-656. Topolski, T. D., Hewitt, J. K., Eaves, L. J., Silberg, J. L., Meyer, J. M., Rutter, M., Pickles, A., & Simonoff, E. (1997). Genetic and environmental influences on child reports of manifest anxiety and symptoms of separation anxiety and overanxious disorders: A community-based twin study. Behavior Genetics, 27, 15-28. van den Oord, E. J. C. G., Boomsma, D. L, & Verhulst, F. C. (1994). A study of problem behaviors in 10- to 15-year-old biologically related and unrelated international adoptees. Behavior Genetics, 24, 193-205. van den Oord, E. J. C. G., & Rowe, D. C. (1997). Continuity and change in children's social maladjustment: A developmental behavior genetic study. Developmental Psychology, 33, 319-332. Warren, S. L., Huston, L., Egeland, B., & Sroufe, A. (1997). Child and adolescent anxiety disorders and early attachment. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 637-644. Wierzbicki, M. (1987). Similarity of monozygotic and dizygotic child twins in level and lability of subclinically depressed mood. American Journal of Orthopsychiatry, 57, 33-40.