|9 of 11|
|Journal Format-PDF (122 K)|
Copyright © 2001 Elsevier Science Ltd. All rights reserved.
Parental problem drinking and anxiety disorder symptoms in adult offspring
Examining the mediating role of anxiety sensitivity components
Paula S. R. MacPherson, Sherry H. Stewart, and Lachlan A. McWilliams
Department of Psychology, Dalhousie University, Life Sciences Centre, 1355 Oxford Street, Halifax, Nova Scotia, Canada B3H 4J1
Available online 21 November 2001.
Paula S. R. MacPherson, Sherry H. Stewart, and Lachlan A. McWilliams
Preliminary studies have implicated childhood exposure to parental problem drinking as a possible factor in the development of anxiety sensitivity (AS). The present retrospective study was designed to examine the role of exposure to distressing parental problem drinking behaviors, over and above the role of parental alcoholism, in the development of various AS components (psychological, physical, and social concerns) in the offspring. We also examined the possible mediating role of AS components in explaining relations between parental drinking problems and anxiety-related symptoms in the adult offspring. A sample of 213 university students provided retrospective reports of both distress related to parental drinking [Children of Alcoholics Screening Test (CAST)] and parental alcoholism [maternal and paternal forms of the Short Michigan Alcoholism Screening Test (SMAST)]. Participants also reported on their own current AS levels [AS Index (ASI)], general anxiety symptoms [State-Trait Anxiety Inventory-Trait subscale (STAI-T)], and lifetime history of uncued panic attacks [Panic Attack Questionnaire-Revised (PAQ-R)]. Scores on the CAST predicted AS psychological and physical concerns (but not social concerns) over and above participant gender and parental alcoholism measured by the SMASTs. Moreover, AS psychological concerns proved a consistent modest mediator of the relations between parental problem drinking on the CAST and both general anxiety and uncued panic outcomes in the offspring. Thus, exposure to distressing parental problem drinking behavior may be one factor that contributes to elevated AS psychological concerns in the child, which in turn may contribute to the development of anxiety disorder symptoms in the offspring.
Author Keywords: Alcoholism; Children of alcoholics; Anxiety sensitivity; Panic attacks; Family history
Anxiety sensitivity (AS) is an individual difference variable characterized by a fear of anxiety (Reiss, 1991). Research has provided evidence that there are three empirically distinct, but intercorrelated, AS dimensions: (1) physical concerns (fears of physical arousal symptoms), (2) psychological concerns (fears of loss of control), and (3) social concerns (fears of embarrassment) (Stewart and Zinbarg). AS has been suggested as a risk factor for the development of anxiety symptoms (e.g., panic attacks) and anxiety disorders (Reiss, Peterson, Gursky, & McNally, 1986). Consistent with theoretical prediction, numerous studies have demonstrated elevated levels of AS in panic disorder patients (see review in Peterson & Reiss, 1992). Elevated AS levels have also been found in other anxiety disorders, such as generalized anxiety disorder (for a review, see Cox, Borger, & Enns, 1999). Subsequent longitudinal research has further demonstrated that AS plays a causal role in the development of panic attacks (e.g., Schmidt, Lerew, & Jackson, 1997) and anxiety disorders (e.g., Maller & Reiss, 1992).
Until recently, the origins of elevated AS levels have received relatively little research attention. A twin study investigated the magnitude of genetic influences on AS (Stein, Jang, & Livesley, 1999). It was determined that AS has a strong heritable component, with genetic factors accounting for 45% of the variance in AS levels. However, more than half (i.e., 55%) of the variance was attributable to environmental influences, implying that experience also plays a strong role in shaping AS. These findings are consistent with Reiss and McNally's (1985) original suggestion that AS could arise from either genetic factors and/or learning experiences.
Watt, Stewart, and Cox (1998) investigated the relationship between childhood learning experiences and the development of AS in a nonclinical university student sample using retrospective questionnaires. Watt et al. found levels of AS in young adulthood to be positively related to instrumental and vicarious conditioning experiences in childhood. Specifically, high AS students reported more instances than controls where they were rewarded by parents for displaying sick role behavior in response to childhood anxiety symptoms. Similarly, high AS students reported more instances where parents modeled and were rewarded for fear reactions to their own anxiety symptoms and/or verbally transmitted their beliefs about the harmfulness of these symptoms to the child. These findings were replicated in two subsequent studies (i.e., Stewart and Watt).
An additional childhood learning experience that may be involved in shaping AS is exposure to distressing parental problem drinking behaviors. Watt et al. (1998) also found that high AS individuals reported significantly more episodes of observing parental uncontrolled behavior due to drunkenness and/or anger than controls. This is suggestive of a possible connection between exposure to parental problem drinking and the concerns of high AS individuals regarding loss of control (i.e., AS "psychological concerns"). In another retrospective study investigating the developmental antecedents of AS, Scher and Stein (1999) investigated the role of exposure to parental threatening, hostile, and rejecting behavior in the development of AS and its lower-order components. Degree of childhood exposure to these types of parental behaviors together predicted a significant proportion of the variance in overall AS levels among a sample of nonclinical young adults. Moreover, different types of parental behaviors predicted distinct AS components. For example, greater exposure to parental hostile and rejecting behaviors predicted higher levels of AS psychological concerns (Scher & Stein, 1999). The "hostile and rejecting" category of parental behavior included behaviors that could occur as a consequence of parental intoxication (Scher & Stein, 1999). These two studies thus provide preliminary evidence that childhood exposure to parental problem drinking behaviors might contribute to the development of high AS¯¯particularly AS "psychological concerns"¯¯in the offspring.
A 1987 study by Graves (see also citation in International Diagnostic Systems (IDS), 1989) attempted to examine more directly whether high levels of AS were associated with childhood exposure to parental alcohol problems. In this study, a sample of nonclinical young adults was administered a measure designed to detect the presence of an alcoholic parent that used items adapted from the Children of Alcoholics Screening Test (CAST; Jones, 1983). AS levels were significantly positively related to modified CAST scores, suggesting that high AS may indeed be related to exposure to parental problem drinking. A substantial limitation of Graves' (1987) study, however, was its sole use of the CAST¯¯a measure that has been criticized on a number of grounds (for a review, see Sher, 1991). One concern pertains to the nature of the construct being tapped by the CAST. It has been suggested that, despite its name, the CAST may be most accurately conceptualized as a measure of the child's distress associated with exposure to parental problem drinking behaviors rather than as a measure for detecting parental alcoholism. Parental alcoholism and distress due to parental drinking behavior are two conceptually distinct constructs. Although normally overlapping, they may diverge in certain cases. A child could be distressed by the drinking behavior of a parent who does not meet criteria for an alcohol disorder. Conversely, a parent may be alcoholic and yet the child not exposed to distressing parental behaviors as a result of the alcohol disorder. Thus, given the nature of the CAST, the association identified by Graves between AS levels and CAST scores likely reflects a history of more childhood exposure to distressing behaviors pertaining to a parent's drinking among high AS individuals.
Rather than implicating childhood exposure to distressing parental problem drinking behavior in the development of high AS, Graves' (1987) findings could instead reflect greater rates of alcohol disorder in the parents of high AS young adults. In other words, high AS might be a marker for alcoholism that is passed on genetically from the alcoholic parent to their offspring. If AS was such a genetic "marker," it must (1) distinguish between people who already have alcoholism and those that do not, (2) prove to be a stable characteristic over time, and (3) identify more people among the biological relatives of alcoholics than among people in the general population (Adler and Iacono). Preliminary evidence suggests that AS may meet several of these criteria. With respect to the first criterion, unusually high levels of AS have been shown to be associated with the diagnosis of alcohol dependence (e.g., Karp, 1993). With respect to temporal stability, the AS Index (ASI) has a very high test¯retest reliability over periods of up to 3 years (e.g., Maller & Reiss, 1992; see also review in Peterson & Reiss, 1992). Finally, Graves' study provides preliminary evidence suggestive that elevations on the ASI may identify more people among the biological relatives of alcoholics than among people in the general population. In order to make a stronger argument that it is childhood exposure to parental problem drinking behavior that contributes to the development of AS in the offspring, research needs to control for the alternative possibility that AS may be a marker for alcoholism that is passed on genetically.
High AS acquired through exposure to distressing parental problem drinking may also help explain the often-observed intergenerational link between alcoholism and certain anxiety disorders. Parental alcoholism has been found to be a risk factor for the development of a wide range of psychological disorders in the offspring, including anxiety-related disorders (e.g., Kessler, Davis, & Kendler, 1997). For example, parental alcoholism has been found to be associated with higher rates of panic disorder in the offspring, in both community (Kessler et al., 1997) and clinical (el Guebaly, Staley, Rockman, & Leckie, 1991) samples. Nonetheless, the exact nature of this intergenerational link between alcoholism and anxiety disorders is not clear (Kushner and Merikangas). Upon repeatedly observing the uncontrolled, hostile behavior of a problem-drinking parent, a child might learn to fear her/his own symptoms of arousal and to fear loss of control (cf. Ehlers, 1993). In turn, this acquired AS could place the child at risk for later developing anxiety symptoms and anxiety-related disorders (cf. Reiss et al., 1986). Thus, elevated AS might play a "mediating" or intervening role in explaining relations between exposure to distressing parental drinking behavior and anxiety disorder symptoms in the offspring. This mediating role of AS may operate regardless of whether the adult offspring's parent(s) had a diagnosable alcohol disorder.
In the present study, a retrospective design was used to further investigate the role of childhood exposure to distressing parental problem drinking behavior as a potentially relevant learning experience in the development of the various dimensions of AS in young adulthood. The CAST (Jones, 1983) was utilized in an effort to replicate Graves' (1987) findings and was used as a measure of exposure to distressing parental problem drinking behavior. The Short Michigan Alcoholism Screening Tests for both mother and father (M-SMAST, F-SMAST; Sher & Descutner, 1986) were also administered as indicators of the presence of a diagnosable alcohol disorder in each parent and were used to control for the alternate possibility that parental alcoholism is responsible for elevated AS levels in adult offspring. If a significant relationship was found between CAST scores and AS levels, after controlling for parental alcoholism on the SMASTs, this would provide stronger evidence than any available to date for a role of childhood exposure to parental problem drinking behavior in the development of AS in the offspring. With respect to relations with particular AS components, it was expected that, even after controlling for parental alcoholism, the psychological concerns component of AS would be significantly associated with exposure to distressing parental problem drinking behavior on the CAST (cf. Scher & Stein, 1999).
Finally, the present study used mediator regression analyses (Baron & Kenny, 1986) to examine whether any of the AS components would serve an intervening role in explaining the associations between childhood exposure to parental problem drinking behavior and the development of anxiety disorder symptoms in the young adult offspring. It was hypothesized that AS psychological concerns would mediate the expected association between exposure to parental problem drinking and anxiety disorder symptoms (general anxiety and uncued panic) in the young adult offspring.
Two hundred and thirteen (164 female, 49 male) undergraduate students from a second-year psychology course at Dalhousie University served as participants. Students' mean age was 21.0 (S.D.=2.9) years. All students in the class were invited to participate in the study voluntarily for partial course credit compensation. Students were informed that the researchers were investigating the effects of parental drinking practices on adult children. Any student not wanting to participate in the study but who wished to receive the credit points had the option of writing a brief (i.e., one page) paper on a related topic. None of the invited students declined to participate.
An author-compiled questionnaire was used to collect demographic information (i.e., age and gender).
The PAQ-R (Cox, Norton, & Swinson, 1992) is a measure that is often used in the study of nonclinical panic. It was included in the present study to assess participants' lifetime history of uncued (or "spontaneous") panic attacks. The PAQ-R begins with a description of a panic attack as defined in the DSM-IV (American Psychiatric Association (APA). Respondents are first asked whether they have ever experienced such an attack. Those indicating a lifetime history of at least one prior panic attack are asked to indicate whether they have ever experienced such an attack "out of the blue." Those indicating a history of uncued panic on the latter question were coded as positive for uncued panic (scored as 1) and all others were coded as negative for uncued panic (scored as 0).
This 20-item self-report measure assesses an individual's general anxiety levels using items such as "I feel nervous and restless" and "I worry too much over something that really does not matter." Each item is rated on a four-point Likert scale assessing relative frequency of occurrence of the symptom in question. The scale ranges from 1 (almost never) to 4 (almost always). There is considerable evidence indicating the STAI-T is a psychometrically sound measure (Spielberger et al., 1983).
This 16-item self-report questionnaire assesses an individual's fear of anxiety based on beliefs that anxiety experiences have aversive consequences. Each item is rated on a five-point Likert scale ranging from 0 (very little) to 4 (very much). There is support for the excellent psychometric properties of the ASI, including its high internal consistency, high test¯retest reliability, and construct validity (Peterson & Reiss, 1992). We scored the ASI according to its three lower-order subscales (physical, psychological, and social concerns) using the scoring system recommended by Zinbarg et al. (1999). The physical concerns subscale involved eight items [Items 3, 4, 6, 8¯11, and 14; e.g., "It scares me when I feel `shaky' (trembling)"]. The psychological concerns subscale involved four items (Items 2, 12, 15, and 16; e.g., "When I cannot keep my mind on a task, I worry that I might be going crazy"). Finally, the social concerns subscale involved the remaining four items (Items 1, 5, 7, and 13; e.g., "It embarrasses me when my stomach growls"). Given the differing number of items, ASI subscales were scored as item means (possible RANGE=0¯4).
This 30-item self-report measure was developed to identify "adult children of alcoholics" (Jones, 1983). Items tap the adult child's psychological distress associated with parental drinking, their perceptions of marital discord between parents related to parental drinking behavior, their attempts to control a parent's drinking, and their efforts to escape from parental drinking. Items also assess exposure to family violence related to parental drinking behavior, the adult child's tendencies to perceive a parent as an "alcoholic," and their desire for professional counseling related to these issues. Sample items are "Did you ever argue or fight with a parent when he or she was drinking?" and "Did you ever protect another family member from a parent who was drinking?" Participants respond to each item by indicating either "yes" or "no." The CAST is scored by summing all "yes" responses resulting in a score ranging from 0 (no experiences with parental alcohol misuse) to 30 (multiple experiences with parental alcohol misuse).
The original SMAST is a 13-item self-report questionnaire designed to detect the presence of an alcohol disorder in the test taker (Selzer, Vinokur, & Van Rooijan, 1975). The F-SMAST and M-SMAST were adapted from the SMAST by Sher and Descutner (1986) to screen for the presence of an alcohol disorder in the test taker's father and mother, respectively. Specifically, the 13 SMAST items were reworded to refer to the father's and mother's drinking behavior, for the F-SMAST and M-SMAST, respectively. Participants respond to each item by indicating either "yes" (1) or "no" (0). Sample items are "Was your father (mother) able to stop drinking when he (she) wanted to?" and "Has your father (mother) ever gotten into trouble at work because of his (her) drinking?" The F-SMAST and M-SMAST are both scored by counting the number of items endorsed. This results in total scores ranging from 0 to 13 for each parent. statistics reveal that the highest agreement between the Family History Research Diagnostic Criteria (Endicott, Andreason, & Spitzer, 1975) interview-based diagnoses of alcohol disorder and the SMAST occur with cut-off scores of 5 on the F-SMAST and 4 on the M-SMAST (Crews & Sher, 1992). Excellent intersibling reliability is demonstrated on both the F-SMAST and M-SMAST (Crews & Sher, 1992). The F-SMAST and the M-SMAST measure distinctly different events than the CAST. The SMASTs measure the presence of an alcohol use disorder in one's father and/or mother. The CAST, however, measures distress associated with the parental problem-drinking behavior.
The general purpose of the study was explained to potential participants. After providing written informed consent, the volunteers completed the questionnaire package during class time in the order listed above.
On average, the sample scored 17.3 (S.D.=8.8) on the ASI Total and 37.4 (S.D.=9.2) on the STAI-T Total. These scores are similar to those found in other samples of university students (e.g., Peterson and Spielberger, respectively). On average, the sample scored 1.09 (S.D.=0.74), 0.41 (S.D.=0.53), and 1.74 (S.D.=0.70) on the physical, psychological, and social concerns subscales of the ASI, respectively. These subscale scores are highly similar to those reported for previously tested samples of university students (e.g., Stewart et al., 1997). Based on responses to the PAQ-R, 41 (19.2%) reported a lifetime history of uncued panic attacks. This finding is consistent with rates of lifetime uncued panic self-reported in other university student samples. For example, a review by Norton, Cox, and Malan (1992) reports that an average of about 16% of nonclinical research participants acknowledge a history of uncued panic attacks on self-report measures.
Sample means on the CAST, F-SMAST, and M-SMAST were 4.0 (S.D.=6.7; cf. Yeatman, Bogart, Feer, & Sirridge, 1994), 2.1 (S.D.=3.0), and 0.5 (S.D.=1.3), respectively. Scores on these measures were highly positively skewed. A logarithmic transformation was applied to scores on the CAST, which resulted in skew values within acceptable limits (Ferguson, 1981). However, attempts to normalize SMAST scores through logarithmic and square-root transformations (Ferguson, 1981) were unsuccessful. Consequently, these latter measures were dichotomously scored (i.e., 0=nonalcoholic, 1=alcoholic) to indicate the absence/presence of an alcoholic parent.
Scores from the M-SMAST indicated that 4 (1.9%) participants had an alcoholic mother, while scores from the F-SMAST indicated that 24 (11.3%) participants had an alcoholic father. The SMAST measures identified 3 individuals as having both an alcoholic mother and an alcoholic father, 21 individuals having only an alcoholic father, and 1 individual having only an alcoholic mother. The numbers of individuals with alcoholic parents as identified by the SMASTs are similar to those found in previously tested university student samples (e.g., Sher, Walitzer, Wood, & Brent, 1991). The number of respondents in the present sample reporting the presence of an alcoholic mother (n=4) on the M-SMAST was insufficient to permit separate analyses of SMAST data by parental gender (cf. Sher et al., 1991). All subsequent analyses involving SMAST scores used dichotomous coding indicating the absence of parental alcoholism (scored as 0; n=188) or the presence of at least one alcoholic parent (scored as 1; n=25).
Correlations were calculated between the anxiety-related variables (i.e., ASI physical concerns, ASI psychological concerns, ASI social concerns, STAI-T, and PAQ-R) and the two measures of parental drinking behavior (log-transformed CAST scores and SMAST). Study variables were also correlated with participant gender. These correlations are reported in Table 1. The r2 values indicate that the SMAST and CAST share about 36% common variance, which is consistent with the notion that they are tapping overlapping, yet distinct, constructs. The ASI subscales were significantly intercorrelated with shared variance ranging from 12% (physical concerns with social concerns) to 26% (physical concerns with psychological concerns). General anxiety levels (STAI-T scores) were significantly correlated with all three ASI subscales. Uncued panic attacks (PAQ-R scores) were significantly correlated with ASI physical and psychological concerns and marginally correlated with ASI social concerns. The measures of general anxiety levels and uncued panic shared about 8% common variance. The CAST was significantly correlated with all the ASI variables, save social concerns, and with general anxiety levels (STAI-T scores). The CAST was also marginally correlated with uncued panic (PAQ-R scores). In contrast, the SMAST was correlated only with general anxiety levels (STAI-T scores). Finally, participant gender was correlated with several of the anxiety-related measures (ASI physical concerns, STAI-T, and PAQ-R), with women scoring higher than men in each case. This suggested the need to control for participant gender in all further analyses.
Table 1. Intercorrelation matrix
CAST=CAST (log transformed); ASPhys=ASI physical concerns subscale; ASPsych=ASI psychological concerns subscale; ASSoc=ASI social concerns subscale; PAQ-R=uncued panic history on PAQ-R (0=no uncued panic, 1=uncued panic); Gender (0=male, 1=female); SMAST= parental SMAST (0=no parental alcoholism, 1=one or more alcoholic parents).
We first conducted a set of hierarchical multiple regression analyses to determine whether distress due to parental problem drinking behavior (CAST scores) predicts any of the AS components in adult offspring over and above participant gender and parental alcoholism per se (SMAST scores). In each multiple regression, one of the three ASI subscales was the criterion variable. The predictors in the first step of each regression equation were dichotomously coded SMAST scores and participant gender. Log-transformed CAST scores were entered in the second step of each regression equation.
In the first regression, SMAST scores and gender together predicted a significant 2.8% of the variance in ASI physical concerns scores in the initial step [F(2,210)=3.07, P<.05]. When CAST scores were entered in the second step, they predicted a significant, albeit modest, additional 3.2% of the variance over and above participant gender and parental alcoholism per se [Finc(1,209)=7.13, P<.01]. In the final equation, female gender (=0.14, t=2.06, P<.05) and higher CAST scores (=.23, t=2.67, P<.01) were significant independent predictors at the univariate level.
In the second regression, SMAST scores and gender failed to predict ASI psychological concerns in the initial step [F(2,210)=1.07, n.s.]. When CAST scores were entered in the second step, they predicted a significant, albeit modest, additional 6.1% of the variance over and above participant gender and parental alcoholism [Finc(1,209)=13.75, P<.001]. In the final equation, the CAST (=0.31, t=3.71, P<.0005) was the only significant univariate predictor.
In the third regression, SMAST scores and gender failed to predict ASI social concerns in the initial step [F(2,210)=0.19, n.s.]. When CAST scores were entered in the second step, they failed to predict significant additional variance over and above participant gender and parental alcoholism [Finc(1,209)=3.14, n.s.]. In the final equation, none of the three predictor variables proved significant at the univariate level.
A series of regression analyses were used to examine the hypothesis that certain AS components would serve a "mediating" or intervening role in explaining associations between distress due to parental problem drinking and the development of anxiety disorder symptoms in the young adult offspring. A given variable is said to function as a "mediator" to the extent that it accounts for the relation between the predictor (independent variable) and the criterion (dependent variable). The approach recommended by Baron and Kenny (1986) to test a mediational hypothesis is to estimate a series of regression models. The first step involves demonstrating a relationship between the predictor variable and the potential mediator. In the second step, the predictor variable must affect the criterion variable. The third step involves regressing the criterion variable on the predictor and mediator variable simultaneously. For a mediational hypothesis to be supported, the mediator must affect the criterion variable in this third equation, and the relationship between the predictor variable and the criterion variable must be less in the third equation than in the second (Baron & Kenny, 1986). Complete mediation is indicated when a significant association between the predictor and criterion is reduced to nonsignificance after controlling for paths running through the potential mediator (Baron & Kenny, 1986). However, because complete mediation is rarely observed in psychological research, Baron and Kenny (1986) note that any degree of reduction in the predictor¯criterion relation, after controlling for the putative mediator, indicates partial mediation. This leaves researchers with the task of determining what degree of partial mediation is of interest given their purpose and theory. Kushner, Thuras, Abrams, Brekke, and Stritar (2001, this issue) have recommended that the relation (r2) between the predictor and the criterion should be reduced by at least 50% when the potential mediator is included, in order to conclude that meaningful mediation has occurred. This 50% reduction in r2 translates into about a 30% reduction in r. Given that we are working with similar variables to those examined by Kushner et al., we made an a priori choice of employing a similar criterion (i.e., a 30% reduction in r) for determining meaningful mediation in the present study.
Separate mediator analyses for the criterion variables of general anxiety and panic anxiety, respectively, were performed using log-transformed scores on the CAST as the predictor variable and each of the three subscales from the ASI (physical, psychological, and social concerns) as potential mediators. In all equations, dichotomously coded scores on the SMASTs and participant gender were entered simultaneously with CAST scores in order to control for the presence of parental alcoholism and participant gender on the outcome measures. We felt it important to control for the presence of parental alcoholism due to the substantial overlap between SMAST and CAST scores and to control for gender due to the associations of female gender with one of the potential mediators and both of the outcome variables (see Table 1).
We first examined mediational models with general anxiety as the criterion variable and AS physical and psychological concerns as the respective mediators. In the first equations, after controlling for SMAST scores and participant gender, the CAST was a significant predictor of two of the three ASI components. CAST scores predicted ASI physical concerns and ASI psychological concerns, but not ASI social concerns (see top section of Table 2). Thus, only ASI physical and psychological concerns could act as potential mediators. In the second equation, after controlling SMAST scores and participant gender, the CAST was a significant predictor of STAI-T (general anxiety) scores (see middle section of Table 2). In the third equations, both the predictor variable (CAST) and the potential mediators (ASI physical and psychological concerns, respectively) were entered together as predictors of STAI-T scores while continuing to control for SMAST scores and participant gender.
Table 2. Regression analyses testing conditions for mediation: (1) effect of CAST on mediators, (2) effect of CAST on anxiety disorder symptoms, and (3) effect of CAST on anxiety disorder symptoms after controlling for mediator
CAST=CAST (log transformed); PAQ-R=uncued panic history on PAQ-R (0=no uncued panic, 1=uncued panic). Standardized weight (B) and t value provided for multiple regression; odds ratios and Wald statistics (z) provided for logistic regression. In the third step, CAST and mediator were simultaneously regressed for each dependent variable.
In the equation involving ASI physical concerns, scores on both the CAST and ASI physical concerns were significant predictors of general anxiety symptoms (see bottom section (3a) of Table 2). Although the magnitude of the relationship between CAST and STAI-T scores was reduced once ASI physical concerns was entered into the equation (i.e., 's=0.37 vs. 0.30), the magnitude of the reduction was relatively small (a 19% reduction), suggesting that ASI physical concerns fails to substantially mediate the relation between distress due to parental drinking and the adult child's general anxiety levels.
In the equation involving ASI psychological concerns, scores on both the CAST and ASI psychological concerns were significant predictors of general anxiety symptoms (see bottom section (3b) of Table 2). The reduced magnitude of the significant relationship between CAST and STAI-T scores, once ASI psychological concerns was entered into the equation (i.e., 's=0.37 vs. 0.24¯¯a 35% reduction), suggests that ASI psychological concerns serves as a modest mediator between distress due to parental drinking and the adult child's general anxiety levels. The mediating role of ASI psychological concerns in explaining relations between CAST scores and general anxiety levels in the offspring is also illustrated in Fig. 1.
Fig. 1. Regression analyses depicting the mediating role of the adult child's levels of AS psychological concerns in explaining the relation between CAST-defined exposure to parental problem drinking and the adult child's general anxiety symptoms (STAI-T). Coefficients are standardized 's in a series of multiple regression analyses. Standardized 's in parentheses represent the relation between two variables after controlling for the influences of the third. All regressions control for the influences of parental alcoholism (dichotomously scored SMAST scores) and participant gender. CAST=log-transformed scores on the CAST (Jones, 1983). ASPsych=psychological concerns subscale of the ASI (Reiss et al., 1986). STAI-T (Spielberger et al., 1983). *P<.05; **P<.01; ***P<.005; ****P<.0005.
The mediational models with lifetime history of uncued panic attacks (PAQ-R) as the criterion variable and ASI physical concerns and psychological concerns as mediators were examined next. Since uncued panic was a categorical outcome variable, we used logistic regression in the second and third steps for this criterion variable as recommended by Tabachnick and Fidell (1989). As noted above, in the first equations, the CAST was a significant predictor of ASI physical and psychological (but not social) concerns, after controlling SMAST scores and participant gender (see top section of Table 2). Thus, again, only ASI physical and psychological concerns could serve as potential mediators. In the second equation, after controlling SMAST scores and participant gender, the CAST was a significant but modest predictor of uncued panic anxiety on the PAQ-R (see middle section of Table 2). In the third equations, ASI physical and psychological concerns, respectively, were each entered together with log-transformed CAST scores as predictors of uncued panic while continuing to control for SMAST scores and participant gender.
In the equation involving ASI physical concerns, physical concerns was a significant predictor of uncued panic, whereas transformed CAST scores were no longer a significant predictor (see bottom section (3c) of Table 2). Although the relationship between CAST and PAQ-R scores was reduced to nonsignificance once ASI physical concerns was entered into the equation, the reduction in magnitude of the relation was relatively small (i.e., odds RATIOS=2.5 vs. 2.0¯¯a 20% reduction). This suggests that ASI physical concerns failed to substantially mediate the relation between CAST and uncued panic.
In the equation involving ASI psychological concerns, psychological concerns was a significant predictor of uncued panic, whereas transformed CAST scores were no longer a significant predictor (see bottom section (3d) of Table 2). The reduced and no longer significant relationship between transformed CAST scores and PAQ-R uncued panic, once ASI psychological concerns was entered into the equation (i.e., odds RATIOS=2.5 vs. 1.7¯¯a 32% reduction), suggests that ASI psychological concerns serves as a modest mediator between distress due to parental drinking and the adult child's propensity to experience uncued panic.
The primary purpose of the present study was to investigate the relationship between childhood exposure to parental problem drinking behavior and levels of various AS components in the young adult offspring. Consistent with the hypotheses based on the earlier findings of Graves (1987), AS levels in young adulthood were related to retrospectively reported childhood exposure to parental problem drinking behaviors. We extended Graves' prior work by examining the impact of exposure to distressing parental drinking behavior on the child's AS level over and above parental alcoholism per se. We also extended his previous study by examining the relations of exposure to distressing parental drinking behavior on the various lower-order AS components. As hypothesized, the "psychological concerns" component was the dimension of AS that was most strongly associated with distress surrounding parental problem drinking behavior. This is consistent with the possibility that frequent exposure to uncontrolled and hostile behavior during parental drinking bouts can contribute to fears of loss of control in the offspring. This finding is also consistent with the results of Scher and Stein (1999) that parental hostile and rejecting behaviors, which can occur during intoxication, contributed to the prediction of AS psychological concerns in a nonclinical sample of young adults. To a lesser extent, we also found that the AS "physical concerns" component was predicted by distress surrounding parental problem drinking behavior on the CAST. This is consistent with the possibility that frequent exposure to parental hostility during drinking bouts might contribute to fears of arousal in the offspring.
The relation between AS psychological and physical concerns with CAST scores was significant even when controlling for the presence of parental alcoholism. This suggests that it is specifically the childhood exposure to distressing parental problem drinking behavior that is associated with the development of AS in the offspring rather than AS serving as a genetic marker for familial alcoholism. This pattern of findings is consistent with the argument that, if AS is an important risk factor for the development of alcoholism (see Stewart, Samoluk, & MacDonald, 1999), it appears to be a risk factor distinct from familial-genetic risk for alcoholism (see Conrod, Pihl, & Vassileva, 1998).
A second goal of this study was to investigate whether any of the three lower-order components of AS would mediate the relations between parental problem drinking behavior and general anxiety and panic outcomes in the young adult offspring. Using the frameworks suggested by Baron and Kenny (1986) and Kushner et al. (2001, this issue), AS psychological concerns were found to be a modest mediator in the relationship between parental problem drinking behavior and the adult child's general anxiety levels. Neither AS physical concerns nor AS social concerns served as substantial mediators of this relationship. In a similar series of mediator analyses, AS psychological concerns were found to be a modest mediator of the relationship between parental problem drinking and uncued panic attacks, whereas AS physical concerns and AS social concerns again played no substantial mediating role in this relationship.
In short, AS psychological concerns were found to consistently mediate the relations between parental problem drinking and anxiety disorder symptoms in the young adult offspring. This pattern is consistent with the possibility that childhood exposure to parental problem drinking sets the stage for the development of fear of loss of control and that these AS psychological concerns in turn set the stage for the development of anxiety disorder symptoms in young adulthood. This finding adds to the growing literature suggesting that acquired AS may play a modest mediating role in explaining relations between childhood learning experiences and anxiety-related symptoms in adulthood (e.g., Stewart and Watt). Given these promising initial findings, future research should investigate the role of AS as a potential mediator in explaining the intergenerational link between alcoholism and anxiety disorders (see Kushner et al., 2000).
Several possible study limitations should be noted. The impact of maternal and paternal alcoholism could not be considered separately in the present study due to the infrequent presence of maternal alcoholism in the sample (cf. Sher et al., 1991). This difficulty is not surprising as men are about six times more likely than women to develop alcoholism (Merikangas, 1990). Nonetheless, future research regarding the differing impacts of paternal and maternal alcoholism on AS levels in their offspring is warranted. It would also be interesting to examine potential gender differences in the impact of parental problem drinking on AS levels in the offspring given the consistent female-greater-than-male gender difference in AS levels (see review by Stewart & Baker, 1999).
Additionally, the present study did not include measures of general parental uncontrolled behaviors or any measure of uncontrolled behavior stemming from a specific problem other than with alcohol (e.g., parental personality disorder). This leaves open the possibility that exposure to uncontrolled behavior from any source (not just from parental problem drinking) would cause AS in the child (cf. Scher & Stein, 1999). We recommend that future research include measures of these additional constructs to determine whether CAST scores predict AS levels in the adult offspring over and above these alternative constructs.
A further limitation of the present study was its reliance on adult offspring's retrospective self-reports of their childhood experiences. Such reports can be substantially influenced by selective memory of past events and by current attitudes and experiences (Ehlers, 1993). Thus, replication of the present findings using alternative methodologies remains important. For example, retrospective reports obtained directly from the participants' parents and/or siblings could provide a worthy validation of the present findings (cf. Watt et al., 1998). However, the present reliance on retrospective reports may not be a serious limitation since retrospective measures that require individuals to report specific events or facts that they were sufficiently old and well placed to know about are likely to be reasonably accurate (Brewin, Andrews, & Gotlib, 1993). While problems registered by the CAST involved drinking behavior that occurred within the awareness of the subject, those registered by the SMASTs may have been significantly removed both spatially and temporally from the subject, making reliance on their retrospective reporting more potentially problematic. Nonetheless, the SMASTs have been shown to demonstrate excellent intersibling reliability in prior research (Sher & Descutner, 1986), attesting to their likely accuracy.
Finally, the correlational nature of the present findings precludes drawing conclusions that exposure to distressing parental problem drinking behavior causes elevated AS psychological or physical concerns in the offspring due to problems of directionality and possible third variable interpretations. To overcome the directionality problem, longitudinal studies need to be conducted with children who have been identified as having been exposed to parental drinking problems to determine if they develop higher AS levels relative to control children not exposed to such an environment. Such longitudinal research could also better evaluate the possible role of AS in mediating associations between parental problem drinking and the development of anxiety disorder symptoms in the offspring (e.g., Kessler et al., 1997) by providing the opportunity to track the emergence of mediating and outcome variables over time. Moreover, given that we did not obtain information on parents' AS levels or anxiety disorder histories, possible third variable interpretations of the observed relation between childhood exposure to parental problem drinking and AS in the offspring are possible. For example, elevated AS in the parent could both cause the parent to drink (see Stewart et al., 1999) and cause elevated AS in the child through genetic and/or learning mechanisms (see Buffett-Jerrott, Stewart, Watt, & Jang, 1999; Hale & Calamari, 1999). This would create an apparent association between parental problem drinking and elevated AS in the offspring. Future research on this issue must assess and control for such important third variables.
In summary, the present study replicated and extended Graves' (1987) finding of an association between childhood exposure to distressing parental problem drinking behavior on the CAST and elevated AS levels in the offspring. Graves' findings were extended in the present study by showing that the associations between exposure to parental problem drinking behavior was present even after controlling for parental alcoholism per se. His prior findings were also extended through our focus on the impact of exposure to distressing parental drinking behavior on the various lower-order components of AS. In a series of regressions in which gender and parental alcoholism were statistically controlled for, distress due to parental problem drinking behavior was significantly and positively associated with AS psychological and physical concerns (but not AS social concerns) of adult offspring. Further, AS psychological concerns were found to play a modest mediating role in explaining the relations between parental problem drinking and both general anxiety symptoms and uncued panic attacks in the young adult offspring. The findings thus contribute to a growing literature suggesting that acquired AS may help explain relations between childhood learning experiences and the development of psychopathological symptoms in young adulthood (cf. Stewart and Watt). The present study expands relevant childhood learning experiences involved in the development of AS, and ultimately anxiety disorder symptoms, to include frequent exposure to the uncontrolled and hostile behaviors that can accompany parental problem drinking.
The present study was funded by a grant from the Dalhousie University Research Development Fund for the Arts (RDFA) awarded to the second author. The first and third authors are now completing graduate degrees at the University of Manitoba, Department of Psychology. The authors would like to thank Ellen Rhyno for assistance in testing participants, Heather Lee Loughlin for technical assistance, and Dr. Bradley Frankland for statistical advice.
Adler, L.E., Freedman, R., Ross, R.G., Olincy, A. and Waldo, M.C., 1999. Elementary phenotypes in the neurobiological and genetic study of schizophrenia. Biological Psychiatry 46, pp. 8¯18.
American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders-4th edition (DSM-IV), American Psychiatric Association, Washington, DC.
Baron, R.M. and 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, pp. 1173¯1182.
Brewin, C.R., Andrews, B. and Gotlib, I.H., 1993. Psychopathology and early experience: a reappraisal of retrospective reports. Psychological Bulletin 113, pp. 82¯98.
Buffett-Jerrott, S., Stewart, S.H., Watt, M.C. and Jang, K.L., 1999. The relationship between anxiety sensitivity scores of parents and their adult children (Summary). Canadian Psychology 40 2a, p. 10.
Conrod, P.J., Pihl, R.O. and Vassileva, J., 1998. Differential sensitivity to alcohol reinforcement in groups of men at risk for distinct alcoholic syndromes. Alcoholism: Clinical and Experimental Research 22, pp. 585¯597.
Cox, B.J., Borger, S.C. and Enns, M.W., 1999. Anxiety sensitivity and emotional disorders: psychometric studies and their theoretical implications. In: Taylor, S., Editor, , 1999. Anxiety sensitivity: theory, research, and treatment of the fear of anxiety, Erlbaum, Mahwah, NJ, pp. 115¯148.
Cox, B. J., Norton, G. R., & Swinson, R. (1992). The Panic Attack Questionnaire, Revised. Unpublished questionnaire, Clarke Institute of Psychiatry, Toronto, Canada.
Crews, T.M. and Sher, K.J., 1992. Using adapted short MASTs for assessing parental alcoholism: reliability and validity. Alcoholism: Clinical and Experimental Research 16, pp. 576¯584.
Ehlers, A., 1993. Somatic symptoms and panic attacks: a retrospective study of learning experiences. Behaviour Research and Therapy 31, pp. 269¯278.
el Guebaly, N., Staley, D., Rockman, G. and Leckie, A., 1991. The adult children of alcoholics in a psychiatric population. American Journal of Drug and Alcohol Abuse 11, pp. 215¯226.
Endicott, J., Andreason, N. and Spitzer, R.L., 1975. Family history research diagnostic criteria, New York State Psychiatric Institute, Biometrics Research Department, New York.
Ferguson, G.A., 1981. Statistical analysis in psychology and education, McGraw-Hill, New York.
Graves, P. (1987). Adult children of alcoholics and control removal sensitivity: a preliminary study. Unpublished manuscript, Department of Psychology, University of Illinois at Chicago, Chicago, IL.
Hale, L.R. and Calamari, J.E., 1999. Parental symptomatology and child anxiety: the role of trait anxiety and anxiety sensitivity (Summary). In: Program and abstracts of the 19th National Conference of the Anxiety Disorders Association of America, Anxiety Disorders Association of America, Rockville, Maryland, USA, p. 87.
Iacono, W.G., 1998. Identifying psychophysiological risk for psychopathology: examples from substance abuse and schizophrenia research. Psychophysiology 35, pp. 621¯637.
International Diagnostic Systems, 1989. ASI annual update, International Diagnostic Systems, Worthington, OH.
Jones, J.W., 1983. The Children of Alcoholics Screening Test (CAST), Camelot Unlimited, Chicago, IL.
Karp, J., 1993. The interaction of alcohol expectancies, personality, and psychopathology among inpatient alcoholics (Summary). Dissertation Abstracts International 53, p. 4357-B.
Kessler, R.C., Davis, C.G. and Kendler, K.S., 1997. Childhood adversity and adult psychiatric disorder in the US National Comorbidity Survey. Psychological Medicine 27, pp. 1101¯1119.
Kushner, M.G., Abrams, K. and Borchardt, C., 2000. The relationship between anxiety disorders and alcohol use disorders: a review of major perspectives and findings. Clinical Psychology Review 20, pp. 149¯171.
Kushner, M.G., Thuras, P., Abrams, K., Brekke, M. and Stritar, L., 2001. Anxiety mediates the association between anxiety sensitivity and coping-related drinking motives in alcoholism treatment patients. Addictive Behaviors this issue.
Maller, R.G. and Reiss, S., 1992. Anxiety sensitivity in 1984 and panic attacks in 1987. Journal of Anxiety Disorders 6, pp. 241¯247.
Merikangas, K., 1990. The genetic epidemiology of alcoholism. Psychological Medicine 20, pp. 11¯22.
Norton, G.R., Cox, B.J. and Malan, J., 1992. Non-clinical panickers: a critical review. Clinical Psychology Review 12, pp. 121¯139.
Peterson, R.A. and Reiss, S., 1992. Anxiety sensitivity index manual (2nd ed.),, International Diagnostic Systems, Worthington, OH.
Reiss, S., 1991. Expectancy theory of fear, anxiety, and panic. Clinical Psychology Review 11, pp. 141¯153.
Reiss, S. and McNally, R.J., 1985. Expectancy model of fear. In: Reiss, S. and Bootzin, R.R., Editors, 1985. Theoretical issues in behavior therapy, Academic Press, New York, pp. 107¯121.
Reiss, S., Peterson, R.A., Gursky, D.M. and McNally, R.J., 1986. Anxiety sensitivity, anxiety frequency, and the prediction of fearfulness. Behaviour Research and Therapy 24, pp. 1¯8.
Scher, C.D. and Stein, M.B., 1999. Developmental antecedents of anxiety sensitivity (Summary). In: Program and abstracts of the 19th National Conference of the Anxiety Disorders Association of America, Anxiety Disorders Association of America, Rockville, Maryland, USA, pp. 86¯87.
Schmidt, N.B., Lerew, D.R. and 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, pp. 355¯364.
Selzer, M.L., Vinokur, A. and Van Rooijan, L., 1975. A self-administered Short Michigan Alcoholism Screening Test (SMAST). Quarterly Journal of Studies on Alcohol 36, pp. 117¯126.
Sher, K.J., 1991. Children of alcoholics: a critical appraisal of theory and research, University of Chicago Press, Chicago, IL.
Sher, K.J. and Descutner, C., 1986. Reports of paternal alcoholism: reliability across siblings. Addictive Behaviors 11, pp. 25¯30.
Sher, K.J., Walitzer, K.S., Wood, P.K. and Brent, E.E., 1991. Characteristics of children of alcoholics: putative risks factors, substance use and abuse, and psychopathology. Journal of Abnormal Psychology 100, pp. 428¯448.
Spielberger, C.D., Gorsuch, R.L., Lushene, R.E., Vagg, P.R. and Jacobs, G.A., 1983. Manual for the State-Trait Anxiety Inventory, Consulting Psychologists Press, Palo Alto, CA.
Stein, M.B., Jang, K.L. and Livesley, W.J., 1999. Heritability of anxiety sensitivity: a twin study. American Journal of Psychiatry 156, pp. 246¯251.
Stewart, S. H. & Baker, J. M. (1999). Gender differences in anxiety sensitivity. Anxiety Disorders Association of America Reporter, 10(3), 1, 17¯18.
Stewart, S.H., Samoluk, S.B. and MacDonald, A.B., 1999. Anxiety sensitivity and substance use and abuse. In: Taylor, S., Editor, , 1999. Anxiety sensitivity: theory, research, and treatment of the fear of anxiety, Erlbaum, Mahwah, NJ, pp. 287¯319.
Stewart, S.H., Taylor, S. and Baker, J.M., 1997. Gender differences in dimensions of anxiety sensitivity. Journal of Anxiety Disorders 11, pp. 179¯200.
Stewart, S.H., Taylor, S., Jang, K.L., Cox, B.J., Watt, M.C., Fedoroff, I.C. and Borger, S.C., 2001. Causal modeling of relations among learning history, anxiety sensitivity, and panic attacks. Behaviour Research and Therapy 39, pp. 443¯456.
Tabachnick, B.G. and Fidell, L.S., 1989. Using multivariate statistics (2nd ed.),, HarperCollins, New York.
Watt, M.C. and 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, pp. 107¯118.
Watt, M., Stewart, S.H. and Cox, B.J., 1998. A retrospective study of the origins of anxiety sensitivity. Behaviour Research and Therapy 36, pp. 505¯525.
Yeatman, F.R., Bogart, C.J., Geer, F.A. and Sirridge, S.T., 1994. Children of Alcoholics Screening Test: internal consistency, factor structure, and relationships to measures of family environment. Journal of Clinical Psychology 50, pp. 931¯936.
Zinbarg, R.E., Mohlman, J. and Hong, N.N., 1999. Dimensions of anxiety sensitivity. In: Taylor, S., Editor, , 1999. Anxiety sensitivity: theory, research, and treatment of the fear of anxiety, Erlbaum, Mahwah, NJ, pp. 83¯113.
Corresponding author. Fax: +1-902-494-6585; email: email@example.com
Volume 26, Issue 6
|9 of 11|