C 2005) Journal of Psychopathology and Behavioral Assessment, Vol. 27, No. 4, December 2005 ( DOI: 10.1007/s10862-005-2406-8
Parent and Child Anxiety Sensitivity: Relationship in a Nonclinical Sample Jennie C. I. Tsao,1,4 Cynthia D. Myers,2 Michelle G. Craske,3 Brenda Bursch,1 Su C. Kim,1 and Lonnie K. Zeltzer1 Accepted November 10, 2004
Little is known about the relationship between parent and child anxiety sensitivity (AS), particularly in nonclinical samples. The present study examined this association in 207 healthy parent-child pairs including 244 children (mean age = 12.3 years, 103 girls) and 226 parent figures (175 mothers). Sequential multiple linear regression revealed that parent AS was significantly associated with child AS in girls, but not in boys. Parent AS accounted for 9% of incremental variance in child AS, after controlling for child age, parent depression, and parent anxiety. Analyses of parent AS dimensions indicated that the social concerns dimension accounted for 14% of incremental variance in child AS in girls only. Parent anxiety and depression showed no association with child AS, once parent AS was taken into account. The findings indicate that parent AS, especially AS social concerns, demonstrates a significant relationship with child AS in this nonclinical sample of girls. KEY WORDS: anxiety sensitivity; anxiety transmission; sex differences; children; parents.
Anxiety sensitivity (AS) refers to the specific tendency to interpret anxiety-related bodily sensations (e.g., shortness of breath, rapid heartbeat) as leading to harmful physical, psychological, or social consequences (Reiss, Peterson, Gursky, & McNally, 1986; Taylor, 1999). AS is related to, but conceptually distinct from, trait anxiety (McNally, 1996) or, the more general tendency to react anxiously to anxiety-provoking stimuli (Lilienfeld, Turner, & Jacob, 1993). In adults, there is extensive support for the empirical distinction between AS and trait anxiety, as well as for the association between AS and clinical anxiety, particularly panic disorder (Taylor, 1999). The concept of AS in children and adolescents has been
more controversial. Chorpita, Albano, and Barlow (1996) reported data challenging the validity of the construct of AS among children under the age of 12 years. The authors posited that younger children lacked the cognitive capacity to associate internal bodily sensations with future, external dangers (e.g., going crazy, death). Other studies however, have supported the construct of AS in children as young as age 6. Weems, HammondLaurence, Silverman, and Ginsburg (1998) found that the CASI predicted incremental variance in both trait anxiety and fear, in a large sample of clinic-referred youth aged 6–17. Further support for the AS construct in children was provided by Rabian, Embry, and MacIntyre (1999) who reported that the CASI significantly predicted anxiety and fear in response to a behavioral challenge test in children aged 8–11 years, after controlling for state and trait anxiety. More recently, AS has been found to predict experience of panic attacks in older adolescents (mean age = 15.4 years) (Weems, Hayward, Killen, & Taylor, 2002). Compared to adolescents with stable low AS, those with stable high and escalating AS were more likely to report a panic attack over a period of 4 years. It is not known however, whether AS in younger children predicts panic attacks in a similar fashion.
1 Pediatric
Pain Program, Departments of Pediatrics, Anesthesiology, Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine at UCLA, Los Angeles, California. 2 H. Lee Moffitt Cancer Center and Research Institute, Department of Interdisciplinary Oncology at the College of Medicine, University of South Florida, Florida. 3 Department of Psychology, University of California, Los Angeles, California. 4 To whom correspondence should be addressed at UCLA, 10940 Wilshire Boulevard, Suite 1450, Los Angeles, California 90024; e-mail:
[email protected].
259 C 2005 Springer Science+Business Media, Inc. 0882-2689/05/1200-0259/0
260 Despite the growing literature supporting the utility of the AS construct in children and adolescents, very little is known about how AS develops. One study found that adult first-degree relatives of patients with panic disorder demonstrated higher AS than normal controls, supporting the notion that high AS runs in families (van Beek & Griez, 2003). On the other hand, Mannuzza et al. (2002) found no differences in AS for 261 children (aged 6–17 years) of parents with lifetime anxiety and/or mood disorders and controls, suggesting that parent anxiety/depression do not predispose children to high AS. Parent AS was not measured in this study so the extent of correspondence between parent and child AS is not known. Results from a twin study found that the heritability of AS is estimated at 45% (Stein, Jang, & Livesley, 1999), suggesting that more than half of the variance in AS levels may be attributable to environmental factors such as childhood learning experiences. Additional work in the same twin sample revealed that the heritability of AS was confined to women only (Jang, Stein, Taylor, & Livesley, 1999). In women, genetic factors accounted for 37–48% of the variance in AS whereas for men, environmental factors accounted for all of the variance (Jang et al., 1999). These findings suggest sex-differentiation in routes of transmission for AS. Research findings regarding environmental factors in the development of AS has been mixed. Ehlers (1993) found that adult anxiety disorder patients retrospectively reported more childhood learning experiences related to parental and personal anxiety symptoms, compared to normal controls. In contrast, other retrospective studies have found that childhood learning history related to somatic symptoms generally, rather than anxiety symptoms specifically, was associated with high AS in young adults (Stewart et al., 2001; Watt & Stewart, 2000; Watt, Stewart, & Cox, 1998). Similar findings were reported in healthy adolescents aged 12–14 years (Muris, Merckelbach, & Meesters, 2001). More recent work suggests that interpersonal factors such as attachment style may be involved in the development of AS (Weems, Berman, Silverman, & Rodriguez, 2002). Silverman and Weems (1999) posited that selective and distorted encoding of threatening information, characteristic of the cognitive response styles of insecurely attached persons, may predispose such individuals to misinterpret anxiety symptoms as catastrophic. Thus, feelings of personal unworthiness and/or interpersonal distrust may exacerbate the tendency to appraise anxiety sensations as dangerous. For instance, expectations that others will be rejecting and untrustworthy may lead to the notion that anxiety symptoms might result in social censure. In addition, if anxiety symptoms are interpreted
Tsao, Myers, Craske, Bursch, Kim, and Zeltzer at harbingers of catastrophe, distress is heightened if there are pre-existing beliefs that others will not help or that attempts at self-help will be inadequate. In accord, Weems, Berman et al. (2002) reported that insecurely attached high school and university students, specifically those classified as preoccupied and fearful, had significantly elevated AS relative to securely attached individuals. The authors suggest that early attachment relations may play a role in the eventual development of AS, particularly the social aspects of AS (e.g., fear of social censure/ridicule for anxiety sensations). Most prior work on the relationship between parent and child AS has used retrospective designs, which may be subject to memory biases. One of the few studies that have directly examined this relationship found that children (aged 7–15 years) of parents with panic disorder reported greater fear of physical symptoms than children of parents with animal phobias and controls (Unnewehr, Schneider, Margraf, & Jenkins, 1996). Unfortunately, this study used a 5-point rating scale with unknown psychometric properties to measure child AS. Silverman and Weems (1999) reported that in 144 clinic-referred children (aged 6–17 years), parent depression was significantly related to child AS, after controlling for parent AS and parent anxiety. They found no clear association however, between parent AS and child AS. Little research to date has been conducted in nonclinical samples and such work may promote understanding of the development of AS in children without existing psychopathology. Evidence of a strong relationship between parent and child AS in nonclinical samples would for example, support the implementation of prevention efforts directed at modifying parent AS to reduce potential risk for future anxiety disorders in children. In one of the few relevant investigations to date, an unpublished study by Hale and Calamari (1999) reported that parent AS was a significant predictor of child AS in 50 healthy parent-child pairs after controlling for parent trait anxiety. However, this study focused only on global AS and did not examine AS dimensions (physical concerns, mental concerns, and social concerns). Previous research supports the distinction and clinical utility of these AS dimensions (Zinbarg, Barlow, & Brown, 1997; Zinbarg, Brown, Barlow, & Rapee, 2001). Hale and Calamari also did not examine parent-child relationships separately by child sex, and as discussed above prior work has shown sex differences in the heritability of AS (Jang et al., 1999). The present study therefore sought to replicate and extend the work of Hale and Calamari in a large, ethnically diverse healthy sample. Relationships between parent and child AS were examined separately for boys and girls, taking into account child age, parent anxiety, and depression
Parent and Child Anxiety Sensitivity symptoms. The relative contribution of parent AS dimensions to the prediction of child AS was also explored, as well as the specific association of parent depression and of parent anxiety to child AS, after controlling for parent AS. In addition, associations among child AS dimensions and parent AS were investigated, as well as the potential moderating effects of child age on parent-child AS relationships.
METHOD Participants and Procedure There were 244 healthy child and adolescent participants (124 girls; 50.8%) with a mean age of 12.7 years (SD = 3.11, range = 8–18). The ethnic composition was: 40.2% Caucasian, 13.9% African American, 9.8% Asian American, 23.8% Hispanic, 12.3% Other. Parent socioeconomic status (Hollingshead, 1975) was: unskilled workers 3.7%, semiskilled workers 4.1%, clerical/sales 11.9%, technical 41.8%, professional 34.8%. Participants were recruited from a major urban area through mass mailing, posted advertisements, and classroom presentations as part of a larger study examining laboratory pain response. Institutional Review Board approved informed consent forms were completed by parents, and children provided written assent. The majority of parent participants were mothers (n = 177; 79.7%), followed by fathers (n = 33; 14.9%), legal guardians (n = 2, 0.9%), and other (n = 10; 4.5%). The “other” category consisted of an aunt, a grandmother, a stepmother, an older brother, and older sisters. Data regarding ethnicity of parent participants were not available. Because the focus of the present study was on parent-child relationships, data for legal guardians and “other” participants were excluded from the analyses. Participants with incomplete data (19 parents and 6 children) were also excluded. The final sample consisted of 207 child-parent pairs [child participants (mean age = 12.3 years; 103 girls); parent participants (175 mothers, 84.5%)]. Details of the larger study from which the current data are drawn have been described elsewhere (Tsao et al., 2004). The questionnaires included in the present study were part of a set of measures administered to participants in the larger study (for details see Tsao et al., 2004). Child/adolescent participants completed a measure of AS (see below) in a quiet room at the health science center. Parents completed questionnaires (described below) either in a separate quiet room at the health science center or at home. Parents were not with their child while he or she completed the questionnaire.
261 Measures Child Questionnaire The Childhood Anxiety Sensitivity Index (CASI) (Silverman, Fleisig, Rabian, & Peterson, 1991) is an 18item scale that measures the tendency to view anxietyrelated bodily sensations as dangerous. Items are scored on a 3-point scale (none, some, a lot); total scores are calculated by summing all items. The CASI has demonstrated high internal consistency (α = .87) and adequate test-retest reliability (range = .62–.78 over 2 weeks) (Silverman et al., 1991). The CASI correlates well with measures of trait anxiety (r‘s = .55–.69) but also accounts for variance in fear not attributable to trait anxiety measures (Weems et al., 1998). Earlier work (Silverman, Ginsberg, & Goedhart, 1999) revealed a hierarchical multidimensional factor structure for the CASI similar to that found for the adult Anxiety Sensitivity Index (see below), consisting of three first-order factors: physical concerns, mental incapacitation concerns, and social concerns. However, recent confirmatory factor analysis of the CASI identified four factors: disease concerns, unsteady concerns, mental illness concerns, and social concerns (Silverman, Goedhart, Barrett, & Turner, 2003). Whereas the mental illness and social concerns factors were identical to those derived in the earlier analyses, Silverman et al. maintain that disease concerns and unsteady concerns comprise two specific facets of AS physical concerns. The best fitting model was obtaining using 13 items; addition of the remaining 5 items resulted in some diminution of model fit, although the four factors were still reliably discriminable. Accordingly, data for CASI total scores included all 18 items whereas data for the CASI dimensions disease concerns and unsteady concerns are derived from the 13-item version of the subscales (the mental and social concerns dimensions comprised the same items in both the three- and four-factor solutions). The magnitude of the relationships did not change substantially when the 18-item versions of the disease and unsteady concerns subscales were used.
Parent Questionnaires The Anxiety Sensitivity Index (ASI) (Reiss et al., 1986) a 16-item measure of the tendency to interpret physical sensations as harmful, asked participants to indicate on a 5-point scale (very little, a little, some, much, very much) the degree to which each statement was true for them. The ASI has been shown to have adequate internal consistency, (α = .82) (Telch, Shermis, & Lucas,
262 1989), and high test-retest reliability (.71 over 3 years) (Maller & Reiss, 1992). Three lower-order factors in the ASI have been found to replicate most consistently across various populations: physical concerns, mental incapacitation concerns, and social concerns (Zinbarg et al., 1997). The Symptom Checklist-90-R (SCL-90-R) (Derogatis, 1994), is a widely used instrument that measures nine symptom dimensions: somatization, interpersonal sensitivity, obsessive-compulsive, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. Items are scored on a 5-point scale of distress ranging from “not at all” to “extremely.” Good internal consistency (α range = .77–.90) and adequate test-retest reliability (range = .68–.83 over 10 weeks) for the nine dimensions have been reported (Derogatis, 1994).
RESULTS Statistical Analyses Bivariate correlations among the parent and child measures were calculated, controlling for child sex and age. For the CASI, in addition to the total score, dimensions corresponding to the following lower-order factors were examined: disease concerns, unsteady concerns, physical concerns, mental illness concerns, and social concerns (Silverman et al., 2003). For the ASI, in addition to the total score, dimensions corresponding to the three lower order factors were examined: physical concerns, mental incapacitation concerns, and social concerns (Zinbarg et al., 1997). For parent depression, the SCL-90R depression dimension was used. For parent anxiety, the SCL-90-R anxiety and phobic anxiety dimensions were used—the former measures generalized anxiety and panic attacks, and the latter measures agoraphobia (Derogatis, 1994). For multivariate analyses, sequential multiple regression analysis was used. To evaluate the relationship between CASI total scores and ASI scores, child age was entered in the first step of the regression analyses, followed by parent depression and anxiety (Step 2), and ASI scores (Step 3). To evaluate the relationship between the ASI subscales and total CASI scores, the regression model was run with the ASI subscales entered in place of ASI total scores in Step 3. To examine the incremental predictive value of parent depression and of parent anxiety to child AS, the regression model was run with ASI total scores and parent anxiety (or parent depression) in Step 2, and parent depression (or parent anxiety) in Step 3. To evaluate the association between the CASI subscales and total
Tsao, Myers, Craske, Bursch, Kim, and Zeltzer ASI scores, separate analyses were conducted to compare the three and four lower-order factors. For these analyses, child age was entered first followed by the three (or four) subscale scores entered simultaneously. The above multivariate analyses were conducted separately for boys and girls. To examine the potential moderating effects of child age on the relationship between parent and child AS, following Baron and Kenny (1986), the moderator (age), predictor (total CASI), and predictor by moderator interaction term (CASI × age) were analyzed in relation to ASI scores. Age was dichotomized as ages 6–11 years and ages 12+ years. Moderator effects are considered evident if the interaction term is significant (Baron & Kenny, 1986). These analyses were conducted separately for boys and girls. In the total sample, sex by age interaction effects on the relationship between parent and child AS were also explored. Residuals were examined for violation of assumptions and outliers. Two cases were identified as having standardized residuals greater than an absolute value of 3, and were excluded from the multivariate analyses. A standard probability level of .05 was used for all analyses.
Descriptive Statistics and Reliability Table I shows the means and standard deviations for the CASI by sex and for the total sample. Mean and subscale scores for the CASI did not differ by sex (t‘s < 1). Means for the CASI total score were slightly elevated compared to previous nonclinical samples but still in the nonclinical range. Silverman et al. (1991) found mean scores of 25.5–26.5 in 81 healthy children (mean age = 13.3 years). However, most prior studies have utilized predominantly Caucasian samples, and the present sample was nearly 60% non-Caucasian. Rabian et al. (1999) reported a mean of 34.0 in a nonclinical sample with a majority (64%) of African American children (mean age = 9.5 years), and a community sample of African American adolescents (mean age = 15.6 years) reported means of 27.7–28.4 (Ginsburg & Drake, 2002). Higher means on the ASI have been reported for Asian American and Latino adolescents, compared to Caucasians (Weems, Hayward et al., 2002). A recent study also reported higher means for the CASI in Hispanic/Latino youth compared to European Americans (Pina & Silverman, 2004). Review of extant literature reveals no other studies that have published means for the CASI in samples of Asian Americans. Given the wide age range of the sample, CASI scores were examined by age, dichotomized as 6-11 years and 12+ years. There were 100 children aged 6–11 years
Parent and Child Anxiety Sensitivity
263
Table I. Means (Standard Deviations) and Reliability Coefficients (Cronbach’s Alpha) for the Parent and Child Measures
Children CASI Total score CASI Subscales Physical concerns Mental concerns Social concerns Disease concerns Unsteady concerns Parents ASI Total score ASI Subscales Physical concerns Mental concerns Social concerns SCL-90-R Depression Anxiety Phobic Anxiety
Total sample (N = 207)
Girls (n = 103)
Boys (n = 104)
α
27.6 (5.3)
28.1 (5.9)
27.1 (4.7)
.80
10.3 (2.7) 3.9 (1.2) 6.2 (1.4) 5.4 (1.5) 4.9 (1.6)
10.4 (3.1) 4.0 (1.3) 6.2 (1.3) 5.5 (1.7) 4.9 (1.8)
10.2 (2.4) 3.9 (1.1) 6.1 (1.5) 5.3 (1.4) 4.9 (1.4)
.73 .55 .52 .62 .57
Total Sample (N = 207)
Mothers (n = 175)
Fathers (n = 32)
α
9.8 (9.0)
9.6 (8.9)
10.7 (9.7)
.89
4.3 (5.7) 1.3 (2.2) 4.2 (2.8)
4.1 (5.5) 1.3 (2.3) 4.2 (2.8)
5.3 (6.5) 1.4 (1.8) 4.0 (2.8)
.89 .73 .65
coefficients (Cronbach’s alphas) for the measures were in general quite high (i.e., above .70), with the exception of the CASI subscales. These results replicate prior findings of high internal consistency for the measures used in the present study. Bivariate Analyses
.41 (.55) .24 (.43) .16 (.34)
.42 (.55) .25 (.46) .18 (.36)
.31 (.50) .92 .15 (.21) .90 .09 (.16) .74
Note. CASI = Childhood Anxiety Sensitivity Index; ASI = Anxiety Sensitivity Index; SCL-90-R = Symptom Checklist 90-Revised.
(52 girls; 48 boys) and 106 children aged 12 and older (51 girls; 55 boys). Age was not available for one participant. Total CASI was significantly higher in younger children (M = 28.6, SD = 5.6), compared to older children (M = 26.7, SD = 4.9) (t(204) = 2.51, p < .02). Means and standard deviations for the parent measures by sex and for the total sample are displayed in Table I. The overall mean for the ASI was relatively low compared to norms for nonclinical samples (M = 19.0, SD = 9.1; average scores were 19.8 for women and 17.6 for men) (Peterson & Reiss, 1993). However, a similar mean ASI score of 9.2 (SD = 7.3) was recently reported in first-degree relatives of patients with panic disorder (van Beek & Griez, 2003). Moreover, ASI scores have been shown to predict panic attacks, anxiety symptoms, and functional impairment in adults with a mean score as low as 4.0 (SD = 2.9) (Schmidt, Lerew, & Jackson, 1997), suggesting that even in groups with low mean ASI scores, the measure is still predictive of panic and anxiety (Peterson & Plehn, 1999). Table I also displays the means and standard deviations for the SCL-90-R dimensions depression, anxiety, and phobic anxiety. Mean scores for these dimensions were well within the normative range for nonpatient samples (Derogatis, 1994). As shown in Table I, reliability
Correlations (corrected for child sex and age) among the child and parent measures are shown in Table II. CASI total scores were modestly but significantly correlated with ASI total scores, but not with parent depression or anxiety. Total CASI scores were correlated with the ASI social concerns subscale but not ASI physical or mental concerns subscales. Of the CASI subscales, only the disease concerns subscale was correlated with total ASI scores. In addition, the ASI social concerns subscale was significantly related to CASI disease concerns, unsteady concerns, and social concerns subscales. ASI scores were highly correlated with parent depression and anxiety scores. Correlations between total CASI and total ASI scores were examined by child age (dichotomized as 6–11 years and 12+ years) and sex. In younger children, parent and child scores were not correlated (r = .08, NS), but among children 12 and older, this correlation was significant (r = .22, p < .03). Examination of these correlations by sex revealed that in younger children, the correlation between parent and child scores was not significant for either boys or girls (r‘s = .03 and .12, respectively, NS). However, in older children, the correlation between parent and child scores was significant for girls (r = .41, p < .01) but not for boys (r = −.02, NS). Multivariate Analyses Relationship of Parent Measures and CASI Scores When total CASI scores were regressed on the parent measures, the regression model was significant for girls only [F(1, 5) = 2.95, p < .02] [for boys, F < 1, NS], after controlling for child age. As displayed in Table III, neither child age (Step 1) nor parent depression and anxiety scores (Step 2) accounted for a significant portion of the variance in total CASI scores. However, entry of ASI total scores (Step 3) resulted in a significant incremental increase in the prediction of total CASI scores, accounting for an additional 9% of the variance (see Table III). The full model containing all predictors accounted for 13% (9% adjusted) of the variance in total CASI scores for girls.
264
Tsao, Myers, Craske, Bursch, Kim, and Zeltzer Table II. Bivariate Correlations Among the Parent and Child Measures (Controlling for Child Sex and Age)
1 Total CASI 2 CASI disease 3 CASI unsteady 4 CASI mental 5 CASI social 6 CASI physical 7 Total ASI 8 ASI physical 9 ASI mental 10 ASI social 11 Parent dep. 12 Parent anxiety 13 Parent phobic anxiety
1
2
3
4
5
6
0.79∗ 0.72∗ 0.60∗ 0.54∗ 0.86∗ 0.15∗∗ 0.10 0.12 0.19∗ 0.09 0.04 0.03
0.53∗ 0.39∗ 0.28∗ 0.87∗ 0.14∗∗ 0.09 0.13 0.16∗∗ 0.12 0.09 0.07
0.27∗ 0.18∗ 0.88∗ 0.10 0.07 0.06 0.14∗∗ 0.08 0.01 0.01
0.18∗ 0.38∗ 0.10 0.09 0.09 0.06 0.04 0.00 0.04
0.26∗ 0.07 0.01 0.07 0.16∗∗ 0.04 0.09 0.04
0.14 0.09 0.10 0.17 0.11 0.05 0.04
7
8
9
0.93∗ 0.79∗ 0.72∗ 0.38∗ 0.42∗ 0.46∗
0.47∗ 0.34∗ 0.40∗ 0.44∗
0.45∗ 0.42∗ 0.45∗ 0.47∗
10
11
12
0.22∗ 0.19∗ 0.22∗
0.88∗ 0.80∗
0.87∗
Note. Total CASI = Childhood Anxiety Sensitivity Index total score; Total ASI = Anxiety Sensitivity Index total score; Parent dep. = SCL-90-R Depression subscale; Parent anxiety = SCL-90-R Anxiety subscale; Parent phobic anxiety = SCL-90-R Phobic Anxiety subscale. ∗ p < .01; ∗∗ p < .05.
Relationship of ASI Subscales to Total CASI Scores To examine the relationship between the ASI subscales and CASI total scores, the ASI subscales were entered simultaneously in the final step instead of the
ASI total score. The resultant regression model was significant for girls [F(1, 7) = 4.04, p < .01] but not so for boys (F < 1, NS). As shown in Table III, the ASI social concerns subscale predicted significant, incremental variance in total CASI scores, explaining an additional 14% of
Table III. Sequential Multiple Regression of Child Age, ASI, Parent Depression, and Anxiety on Total CASI Scores for Girls
β
sr2 (incremental)
Age
−.04
.00
Step 2
Parent depression Parent anxiety Parent phobic anxiety
.34 −.14 −.14
.03 .01 .00
Step 3
ASI total score
Step CASI (DV) Step 1
CASI (DV) Step 1
Variables entered
.32∗
Model R2
Adj. R2
Change in R2
.00
−.01
.00
.05
.01
.05
.13
.09
.09∗
.00
−.01
.00
.05
.01
.05
.23
.18
.19∗
.09
Age
−.01
.00
Step 2
Parent depression Parent anxiety Parent phobic anxiety
.29 −.08 −.17
.02 .00 .01
Step3
ASI social concerns ASI mental concerns ASI physical concerns
.45∗ .07 −.08
.14 .00 .00
Note. β = Standardized regression coefficient; sr2 (incremental) = incremental contribution of IV to R2 in the total set of IV’s; Model R 2 = Coefficient of determination (goodness of fit) for overall regression model after entry of each independent variable; Adjusted R2 = R2 adjusted for number of IV’s and sample size; Change in R2 = incremental contribution of an independent variable to R2 in the total set of independent variables. ∗ p < .01.
Parent and Child Anxiety Sensitivity
265
incremental variance, after controlling for the other variables. Thus, for girls the AS social concerns dimension accounted for all the variance in CASI scores explained by parent ASI scores. The full model containing all predictors accounted for 23% (18% adjusted) of the variance in total CASI scores.
sion models were not significant for girls or boys. Given that AS social concerns subscale accounted for all the variance in CASI scores explained by parent ASI scores in girls, the analyses were also conducted with this subscale as the dependent variable. The results of these analyses were significant for girls only. As shown in Table IV, when the three CASI lower-order factors were entered into the model, both the physical concerns and social concerns subscales accounted for significant incremental variance in the ASI social concerns subscale, explaining an additional 4 and 5%, respectively of incremental variance. The full model with all predictors accounted for 20% (17% adjusted) of the variance in the ASI social concerns subscale. When four first-order factors were entered instead, only the CASI social concerns subscale explained significant incremental variance in the ASI social concerns subscale, accounting for an additional 4% of the variance (see Table IV). The full model with all predictors explained 18% (14% adjusted) of the variance in the ASI social concerns subscale.
Relationship of Parent Depression/Anxiety Scores to CASI Scores, Controlling for ASI Scores Analyses examining the specific relationship of parent depression scores and of parent anxiety scores to total CASI scores after controlling for parent ASI scores, revealed that neither variable predicted incremental variance in boys or girls. In boys, the overall regression models were not significant (F‘s < 1, NS). In girls, the regression models were significant due to the inclusion of parent ASI scores, however the incremental change in the prediction of CASI scores following the entry of parent depression or anxiety scores did not reach significance. Thus, the inclusion of parent depression scores in the final step, after entry of parent anxiety scores and parent ASI scores, did not significantly improve model fit [F change (1, 95) = 3.31, p = .07]. Similarly, the entry of parent anxiety scores in the final step after parent depression scores and parent ASI scores, did not significantly change model fit [F change (2, 95) = 1.04, p = .36].
Moderating Effects of Age on the Relationship Between Parent and Child AS Scores and Sex by Age Interaction Effects Analyses examining the moderating effects of child age on the association between total CASI scores and total ASI scores did not reveal evidence of moderation in girls or boys. Similar results were obtained when the ASI social concerns subscale was examined in place of total ASI scores. Multivariate analyses in the total sample did not reveal a significant age by sex interaction effect
Relationship of CASI Subscales to ASI Scores When total ASI scores were regressed on either the three or the four lower-order CASI subscales, the regres-
Table IV. Sequential Multiple Regression of Child Age and CASI Subscales on ASI Social Concerns Subscale Scores for Girls
Step
Variables entered
ASI social concerns (DV) Step 1 Age Step 2
CASI physical concerns CASI mental concerns CASI social concerns
ASI social concerns (DV) Step 1 Age Step 2
∗p
CASI disease concerns CASI unsteady concerns CASI mental concerns CASI social concerns
< .01; ∗∗ p < .05.
β −.03 .22∗∗ .17 .23∗∗
−.03 .16 .10 .15 .20∗∗
sr2 (incremental)
Model R2
Adj. R2
Change in R2
.00
−.01
.00
.20
.17
.00
−.01
.18
.14
.00 .04 .02 .05
.20∗
.00 .01 .01 .02 .04
.00
.18∗
266 on the relationship between global ASI scores and total CASI scores, or on the relationship between the AS social concerns subscale and total CASI scores. DISCUSSION The present results revealed that in this ethnically diverse, nonclinical sample, there was a significant association between parent and child AS in girls only after controlling for child age, as well as for parent depression and anxiety. Global parent AS explained 9% of incremental variance in child AS in girls. Analyses of AS dimensions revealed that the parent AS social concerns dimension accounted for all of the variance in child AS explained by parent AS for girls only (see Table III). Parent AS social concerns accounted for 14% of incremental variance in child AS in girls, after accounting for the other variables. Notably, there was no relationship between parent and child AS in boys. In addition, there was no evidence of a specific relationship between parent depression or parent anxiety and child AS for children of either sex, once parent AS was controlled for. These findings are consistent with the notion of parental transmission of AS, particularly parent AS social concerns, in the development of AS in healthy girls. Examination of the CASI subscales revealed that the subscales did not predict global parent AS for children of either sex. But, the CASI subscales did predict parent AS social concerns in girls only. When three CASI lower-order factors were included, physical and social concerns explained significant incremental variance (i.e., 4 and 5%, respectively) in parent AS social concerns. However, when four lower-order factors were included, only CASI social concerns accounted for significant incremental variance (i.e., 4%) in parent AS social concerns suggesting that the division of the CASI physical concerns dimension into disease concerns and unsteady concerns appears to attenuate the relationship with parent AS social concerns. Regarding age effects, bivariate analyses indicated that parent and child AS was correlated in older children (12+ years) only. But, multivariate analyses did not reveal evidence of the moderating effect of age on the relationship between parent and child AS. Additional bivariate analyses by sex revealed that the correlation between parent and child AS was significant in older girls only. Multivariate analyses however, did not show significant age by sex interaction effects for the association between parent and child AS. It is possible that there may be a threeway interaction between child age, child sex, and child AS in relation to parent AS, but there was insufficient power in the present sample size for this interaction to
Tsao, Myers, Craske, Bursch, Kim, and Zeltzer reach statistical significance. Future studies may explore this possibility in larger samples. The present findings are consistent with Hale and Calamari (1999) who reported an association between parent and child AS in a small healthy sample, although these authors did not examine parent-child relationships separately by child sex. On the other hand, the results diverge from Silverman and Weems (1999) who found no clear association between parent and child AS in a large sample of clinic-referred children. However, the Silverman and Weems findings were from preliminary analyses of data in a book chapter and it is unclear what proportion of their sample were girls. It is possible that separate analyses by child sex, as was conducted in the present study, might have revealed different results. Another possibility is restricted range in scores of clinicreferred children and their parents, compared to a healthy sample. The most common effect of range restrictions is a reduction of the correlation coefficient (Howell, 1992). It may also be that analysis of total parent AS in the earlier study may have overshadowed the contribution of individual AS dimensions, for example the parent social concerns dimension. Nevertheless, the inconsistency among the findings to date points to the possibility, still to be tested, that the nature of the relationship between parent and child AS differs in clinical and nonclinical populations. Review of extant literature reveals no other studies investigating the links between individual parent AS dimensions and child AS, including the examination of these relationships by child sex. The current findings point to potential sex differences in these associations that are particularly evident for parent AS social concerns—results that are noteworthy given evidence that interpersonal factors, specifically insecure attachment style, may play a role in the development of AS (Weems, Berman et al., 2002), particularly the social aspects of AS (e.g., fear of social ridicule for anxiety sensations). This possibility leads to speculation as to why parent AS social concerns might especially influence girls’ own beliefs about anxiety symptoms. It could be that parents’ concerns over controlling their outward expressions of anxiety, particularly in the context of social situations, are more easily transmitted to girls compared to concerns about internal symptoms of anxiety. Another possibility is that parental expressions of concern over external anxiety symptoms and their social consequences may be of particular relevance to girls during developmentally sensitive periods (e.g., puberty). Such possibilities are speculative and require further study in clinical and nonclinical samples. Notably, the AS social concerns subscale consists of only three items: “It is important to me not to appear nervous,” “It is important to me to stay in control of my
Parent and Child Anxiety Sensitivity emotions, “It embarrasses me when my stomach growls.” Prior research has noted the relatively weak nature of this subscale in factor analytic studies (e.g., Mohlman & Zinbarg, 2000), which may in part be attributable to the small number of items. It is interesting that despite these limitations, this dimension still exhibited the strongest link to child AS in the present sample of girls. The present findings do not speak of the mechanisms by which parent AS might influence child AS, which are still to be determined. Genetic processes have been estimated to account for about 45% of the variance in AS (Stein et al., 1999), suggesting that environmental influences predominate. Examination of these effects by sex revealed that AS appears to be heritable only in women, whereas in men, environmental factors accounted for all of the variability (Jang et al., 1999). Other work has indicated that AS physical concerns and social concerns, as well as ASI total scores have a significant heritable component across sex (Stein et al., 1999). Jang and colleagues (Jang et al., 1999) suggest that these AS factors may be heritable in women only. Thus, further investigation regarding sexspecific aspects of AS transmission is warranted. Regarding environmental factors in AS, social learning, operant conditioning (e.g., parental reinforcement of symptoms), and informational transmission have all been implicated as possible means by which parents might shape children’s AS. Research to date strongly suggests that childhood learning experiences related to AS are likely not specific to anxiety symptoms but involve parental concerns about somatic symptoms in general and/or the reinforcement of sick-role behaviors related to bodily symptoms of all kinds (Muris et al., 2001). Such findings support the possibility that AS is a risk factor for a range of disorders including hypochondriasis (Watt & Stewart, 2000) and chronic pain (Asmundson, 1999). However, existing research has not examined the extent to which high AS parents may be more likely to engage in behaviors that promote AS in children, both in controlled environments such as the laboratory and in naturalistic settings. Future research may also focus on the degree to which such parental behaviors and/or parent AS predict child AS in a longitudinal fashion. Limitations of the present findings should be mentioned. First, the correlational nature of the study design precludes any statements regarding causality or the means by which parent AS might affect child AS. The current sample of parents consisted of mostly mothers, and there were insufficient numbers of fathers to conduct separate analyses by parent group. Assuming a two-parent family, it is conceivable that such influences may differ. Prior work has revealed sex differences in AS (Stewart, Taylor, & Baker, 1997), and it is unknown whether the sex of
267 the parent interacts with the sex of the child to modulate the relationship between parent and child AS. It may be that the null findings for boys were due to the relatively small number of fathers in the present sample. Future studies may investigate these possibilities in samples with adequate numbers of fathers and mothers. Child/adolescent participants were not formally screened for anxiety disorders, so it is not known what portion of the sample may have met criteria for an anxiety disorder. However, all participants were attending school, were not taking psychotropic medications, and scored in the normative range on well-validated measures of anxiety and depression. Relatedly, there was no formal assessment for clinical anxiety and/or depression in parents. Inspection of SCL-90-R scores, however, indicated that parent participants were in the normative range on all subscales. In sum, parent AS, particularly parent AS social concerns, was significantly associated with child AS in this sample of healthy girls. The present results suggest that parent beliefs, particularly those of mothers, about the social consequences of anxiety symptoms may be most salient to AS development in girls. On the other hand, parent depression and parent anxiety did not show a clear relationship to child AS in children of either sex, once parent AS was accounted for. The current findings are consistent with the notion that parent AS, especially maternal AS, may play at least some role in the development of child AS in nonclinical populations of girls. But notably, the present study was not able to identify a link between parent and child AS in healthy boys, perhaps due to the small number of fathers in the parent sample. Whether such an association exists and what factors modulate this possible relationship awaits further study. In light of evidence that AS, at least in older adolescents, is a clear risk factor for panic symptoms (Weems, Hayward et al., 2002), continued examination of the developmental trajectory of AS is warranted. Such investigations may aid in the reduction of AS and consequent risk of possible future anxiety disorders in children. ACKNOWLEDGMENTS This study was supported by R01 DE12754-01A1, awarded by the National Institute of Dental and Craniofacial Research, and by UCLA General Clinical Research Center Grant MO1-RR-00865 (PI: Lonnie K. Zeltzer). REFERENCES Asmundson, G. J. G. (1999). Anxiety sensitivity and chronic pain: Empirical findings, clinical implications, and future directions. In
268 S. Taylor (Ed.), Anxiety sensitivity: Theory, research and treatment of the fear of anxiety (pp. 269–285). London: Erlbaum. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Chorpita, B. F., Albano, A. M., & Barlow, D. H. (1996). Child Anxiety Sensitivity Index: Considerations for children with anxiety disorders. Journal of Clinical Child Psychology, 25(1), 77–82. Derogatis, L. R. (1994). SCL-90-R Symptom Checklist-90-R. Minneapolis, MN: National Computer Systems. Ehlers, A. (1993). Somatic symptoms and panic attacks: A retrospective study of learning experiences. Behaviour Research Therapy, 31, 269–278. Ginsburg, G. S., & Drake, K. L. (2002). Anxiety sensitivity and panic attack symptomatology among low-income African-American adolescents. Journal of Anxiety Disorders, 16, 83–96. Hale, L. R., & Calamari, J. E. (1999). Parental Symptomatology and Child Anxiety: The Role of Trait Anxiety and Anxiety Sensitivity. Paper presented at the 19th National Conference of the Anxiety Disorders Association of America, San Diego, CA. Hollingshead, A. B. (1975). Four-factor index of social status. Tampa, Florida: Department of Sociology, Yale University. Howell, D. C. (1992). Statistical Methods for Psychology (3rd ed.). Boston: PWS-Kent. Jang, K. L., Stein, M. B., Taylor, S., & Livesley, W. J. (1999). Gender differences in the etiology of anxiety sensitivity: A twin study. The Journal of Gender-Specific Medicine, 2, 39–44. Lilienfeld, S. O., Turner, S. M., & Jacob, R. G. (1993). Anxiety sensitivity: An examination of theoretical and methodological issues. Advances in Behaviour Research and Therapy, 15, 147–183. Maller, R. G., & Reiss, S. (1992). Anxiety sensitivity in 1984 and panic attacks in 1987. Journal of Anxiety Disorders, 6, 241–247. Mannuzza, S., Klein, R. G., Moulton, J. L., Scarfone, N., Malloy, P., Vosburg, S. K., et al. (2002). Anxiety sensitivity among children of parents with anxiety disorders: A controlled high-risk study. Journal of Anxiety Disorders, 16, 135–148. McNally, R. J. (1996). Anxiety sensitivity is distinguishable from trait anxiety. In R. M. Rapee (Ed.), Current controversies in the anxiety disorders (pp. 214–227). London: Guilford. Mohlman, J., & Zinbarg, R. E. (2000). The structure and correlates of anxiety sensitivity in older adults. Psychological Assessment, 12, 440–446. Muris, P., Merckelbach, H., & Meesters, C. (2001). Learning experiences and anxiety sensitivity in normal adolescents. Journal of Psychopathology and Behavioral Assessment, 23, 279–283. Peterson, R. A., & Plehn, K. (1999). Measuring anxiety sensitivity. In E. Steven Taylor (Ed.), Anxiety sensitivity: Theory, research, and treatment of the fear of anxiety (p. xii, p. 370). London: Erlbaum. Peterson, R. A., & Reiss, S. (1993). Anxiety sensitivity index revised test manual (2nd ed.). Worthington, OH: IDS. Pina, A. A., & Silverman, W. K. (2004). Clinical phenomenology, somatic symptoms, and distress in Hispanic/Latino and European American youths with anxiety disorders. Journal of Clinical Child and Adolescent Psychology, 33, 227–236. Rabian, B., Embry, L., & MacIntyre, D. (1999). Behavioral validation of the Childhood Anxiety Sensitivity Index in children. Journal of Clinical Child Psychology, 28, 105–112. Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency and the predictions of fearfulness. Behaviour Research and Therapy, 24, 1–8. Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1997). The role of anxiety sensitivity in the pathogenesis of panic: Prospective evaluation of spontaneous panic attacks during acute stress. Journal of Abnormal Psychology, 106, 355–364.
Tsao, Myers, Craske, Bursch, Kim, and Zeltzer Silverman, W. K., Fleisig, W., Rabian, B., & Peterson, R. A. (1991). Child Anxiety Sensitivity Index. Journal of Clinical Child Psychology, 20, 162–168. Silverman, W. K., Ginsberg, G. S., & Goedhart, A. W. (1999). Factor structure of the childhood anxiety sensitivity index. Behaviour Research and Therapy, 37, 903–917. Silverman, W. K., Goedhart, A. W., Barrett, P., & Turner, C. (2003). The facets of anxiety sensitivity represented in the Childhood Anxiety Sensitivity Index: Confirmatory analyses of factor models from past studies. Journal of Abnormal Psychology, 112, 364–374. Silverman, W. K., & Weems, C. F. (1999). Anxiety sensitivity in children. In S. Taylor (Ed.), Anxiety sensitivity: Theory, research, and treatment of fear of anxiety (pp. 239–268). Mahwah, NJ: Erlbaum. Stein, M. B., Jang, K. L., & Livesley, W. J. (1999). Heritability of anxiety sensitivity: A twin study. American Journal of Psychiatry, 156, 246–251. Stewart, S. H., Taylor, S., & Baker, J. M. (1997). Gender differences in dimensions of anxiety sensitivity. Journal of Anxiety Disorders, 11, 179–200. Stewart, S. H., Taylor, S., Jang, K. L., Cox, B. J., Watt, M. C., Fedoroff, I. C., et al. (2001). Causal modeling of relations among learning history, anxiety sensitivity, and panic attacks. Behaviour Research Therapy, 39, 443–456. Taylor, S. (1999). Anxiety sensitivity: Theory, research and treatment of the fear of anxiety. London: Erlbaum. Telch, M. J., Shermis, M. D., & Lucas, J. A. (1989). Anxiety sensitivity: Unitary personality trait or domain-specific appraisals? Journal of Anxiety Disorders, 3, 25–32. Tsao, J. C. I., Myers, C. D., Craske, M. G., Bursch, B., Kim, S. C., & Zeltzer, L. K. (2004). Role of anticipatory anxiety and anxiety sensitivity in children’s and adolescents’ laboratory pain responses. Journal of Pediatric Psychology, 29, 379–388. Unnewehr, S., Schneider, S., Margraf, J., & Jenkins, M. (1996). Exposure to internal and external stimuli: Reactions in children of patients with panic disorder or animal phobia. Journal of Anxiety Disorders, 10, 489–508. van Beek, N., & Griez, E. (2003). Anxiety sensitivity in first-degree relatives of patients with panic disorder. Behaviour Research and Therapy, 41, 949–957. Watt, M. C., & Stewart, S. H. (2000). Anxiety sensitivity mediates the relationships between childhood learning experiences and elevated hypochondriacal concerns in young adulthood. Journal of Psychosomatic Research, 49, 107–118. Watt, M. C., Stewart, S. H., & Cox, B. J. (1998). A retrospective study of the learning history origins of anxiety sensitivity. Behaviour Research Therapy, 36, 505–525. Weems, C. F., Berman, S. L., Silverman, W. K., & Rodriguez, E. T. (2002). The relation between anxiety sensitivity and attachment style in adolescence and early adulthood. Journal of Psychopathology and Behavioral Assessment, 24, 159–168. Weems, C. F., Hammond-Laurence, K., Silverman, W. K., & Ginsburg, G. S. (1998). Testing the utility of the anxiety sensitivity construct in children and adolescents referred for anxiety disorders. Journal of Clinical Child Psychology, 27, 69–77. Weems, C. F., Hayward, C., Killen, J., & Taylor, C. B. (2002). A longitudinal investigation of anxiety sensitivity in adolescence. Journal of Abnormal Psychology, 111, 471–477. Zinbarg, R. E., Barlow, D. H., & Brown, T. A. (1997). Hierarchical structure and general factor saturation of the Anxiety Sensitivity Index: Evidence and implications. Psychological Assessment, 9, 277–284. Zinbarg, R. E., Brown, T. A., Barlow, D. H., & Rapee, R. M. (2001). Anxiety sensitivity, panic, and depressed mood: A reanalysis teasing apart the contributions of the two levels in the hierarchical structure of the Anxiety Sensitivity Index. Journal of Abnormal Psychology, 110, 372–377.