Filhos Pais Ansiosos

Behavior Problems in Children of Parents with Anxiety Disorders WENDY K. SILVERMAN, PH.D., JEROME A. CERNY, PH.D., WENDY B. NELLES, M.A., A ND ANNETTA E. BURKE, PH.D. Abstract. T he Ch ild Behavior Checklist was obtained on 42 children whose parent s were given DSM -Ill diagnoses of agoraphobia with panic attacks, panic disorder, generalized an xiety disorder. and a group of mixed phob ics. Although prelim inary, results suggest that avoida nce in agoraphobia ma y be a key variable associated with child maladjustment. Some possible mechanisms of familial transmission are discussed , with emphasis placed on modeling. the family structure. and the severity of disorder. J. Am. Acad. Child Adolcsc. Psychiatry, 1988, 27. 6: 779-784. Key Words : risk, behavior problems. offspring, parents with anxiety disorders. Evidence indicates that increased risk for child behavior problems is associated with parental psychopathology. Children of parents with anxiety disorders may constitute one such sample at risk. Weissman et al. (1984) reported that children ofdepressed patients with agoraphobia (AG) or panic disorder (PD) were more likely to display separation anxiety than ch ildren of depressed probands without anxiety disorders. Similarl y, while the morbidity risk for all anxiety disorders is 15% among the first degree relatives of controls, the morbidity risk is 32% among the first degree relatives of AG patients and 33% among the first degree relatives of indi viduals with PD (e.g., Harris et al., 1983). There is also a suggestion that genetic factors may be important in the development of anxiety disorders (e.g., Inouye . 1965; Rose et al., 1981; Torgersen, 1979, 1983; Young et al., 1971). Recentl y, data have appeared that focus directly on assessing the offspring of patients with anxiety disorders. Turner et al. (1987) evaluated 16 children (ages 7 to 12) of anxiet y disorders patients using a semistructured interview schedule , the Child Assessment Schedule (CAS) (Hodges et al., 1982), and two self-report instruments, the Fear Survey Schedule

Revised (FSSC-R) (Ollendick, 1983) and the State-Trait Anxiety Inventory for Children (STAIC) (Spielberger, 1973). The comparison control group consisted of children of normal parents (N = 13) who had responded to an announcement soliciting part icipants for the study , and a sample of normal school children (N = 16). Results indicated that the offspring of patients with anxiet y disorders met the criteria for a DSMlIJ childhood anxiety disorder more frequently than the other children. In particular, children of anxiety disorder parents were more than seven times as likely to be diagnosed as having an anxiety disorder compared to the children of normal controls and twice as likely to reach criteria for such a diagnosis than the offspring of dysthymic parents. On the CAS, the anxiety offspring had significantl y higher scores on factors usually associated with emotional distress and social adjustment than the children in the normal school children group. In terms of the questionnaire measures , although the FSSC-R showed no significant differences across groups, children of anxiety disorder parents reported the most fears. These children also scored highest on the STAIC state and trait scales compared to any other group. Sylvester et al. (1987) conducted a controlled study on children (aged 7 to 17) of parents with anxiety disorders (PD (N = 50), depression (N = 27), and normals (N = 48». The results revealed that children from families with PD or depression had poorer adjustment ratings than the children of normals. The child self-report measures of fearfulness and anxiety state and trait revealed that only relatively increased levels of severe fearfulness and anxiety trait differentiated children of pathological parents from those of normal controls. The depression measures differentiated children of PD parents from those of depressed parents but failed to differentiate children of PD probands from normal controls. Rosenbaum et al. (1988) investigated the role of"behavioral inhibition to the unfamiliar" as a predisposing characteristic in children of parents with PO and AG by blindly evaluating 56 children (aged 2 to 7) for behavioral inhibition. Behavioral indicators of "inhibition" include long latencies to interaction with unfamiliar adults, retreat from unfamiliar objects or people. cessation of play and vocalization, clinging to the mother, and crying. According to the authors, this temperamental quality of behavioral inhibition to the unfamiliar might be a predisposing characteristic in children at risk for PO and/or AG in later years. The results revealed that the rates of behavioral inhibition, as manifested by high latency to speak and a small number

Accepted June 6. 1988. Dr. Silverman is Director of the Child and Adolescent Fear and Anxiety Treatment Program at The Center for Stress and Anxiety Disorders. Psychology Department , The University at Albany-SUNY. Ms . Nelles is a doctoral candidate in the clinical psychology training program at The University at Albany-SUNY. Dr. Cerny is Professor of Psychology at Indiana Stat e University. Terre Hallie. IN. Dr. Burke recently received her Ph.D. in clinical psychology from the University at Albany. She is currently Clinical Coordinator f or the Astor House for Children. Rhinebeck. N Y. The authors thank Andre w Eisen. Wayne Fleisig, and Chris Kearny for assisting in the interviewing of subjects, and Bruce Dudek fo r consulting with us on the data analytic procedures. Special thanks are due to Dr. David H. Barlow f or his comme nts on an earlier draft of this manuscript and for providing us with the patient population necessary f or successf ul completion ofthis study. This study was supported by a Faculty Development Grant awarded to Dr. Silverman from the Research Foundation of the State University of New York. Dr. Cerny was on sabbatical leave at the University at Alban y. with partial support from a Faculty Development Grall!from the Indiana Stat e University Research Foundation. An elaboration of some of the issues raised in this article appears in Advances in Clinical Child Psychology, Vol. II. B. B. Lah ey & A. E. Kazdin, Plenum Press: New York. in press. Reprint requests to Dr. Silverman , Department of Psychology. University at Albany, State University of New York. Albany . NY /2222. 0890-8567/88/2706-0779$02.00/0 © 1988 by the American Academy of Child and Adolescent Psychiatry . 779

780

SILVERMAN ET AL.

of spontaneous comments, in children of probands with PD and AG, with or without comorbid major depressive disorder, were significantly higher than for the comparison group (i.e., children who had parents without PD and AG but had a family member in treatment at the same outpatient setting). The data further suggested a progression of increasing rates of inhibition from the comparison group without major depressive disorder, to major depressive disorder, to comorbid PD and AG and major depressive disorder, and to PD and AG. Interestingly, no significant differences emerged in the frequency of behavioral inhibition in the children of pro bands with primary major depressive disorder or major depressive disorder alone. Although longitudinal research is necessary to determine what might be specific indicators of risk for the anxiety disorders, the studies reviewed above suggest several variables that might have heuristic value . Nevertheless, in light of the suggestion that parental diagnosis may be less predictive of child maladjustment than other variables that interact or correlate with parental psychopathology (e.g., Campbell, 1984; Fisher et al., 1980), it is important that some of these other variables be explored in order to ascertain what it is about anxiety disorders that may place youngsters who live with an anxiety disorder parent at risk for psychological problems. This was the purpose of the study described below. Method

Subjects The sample consisted of 28 families that included a parent who had received a primary DSM-l/I diagnosis of agoraphobia with panic attacks (AG) (N = 14), generalized anxiety disorder (GAD) (N = 5), panic disorder (PD) (N = 5), and a group of mixed phobias (simple phobia, social phobia, and/ or obsessive-compulsive; N = 4). Information on the average severity of the diagnosis, average duration of the disorder, average number of panic attacks, and average severity rating of avoidance was also available for each patient. The families provided a total of 42 children who participated in the study: 22 children between the ages of 6 and II and 20 children between the ages of 12 and 16. Of these 42 children, there were 24 boys with a mean age of 10 years, 10 months and 18 girls with a mean age of 13 years, 1 month. There were nine sets of siblings included in this group of 42 children.

Measures and Procedures Patients completed the Child Behavior Checklist (CBCL) (Achenbach and Edelbrock, 1983), which assesses for a wide range of childhood behavior problems and has age and gender norms for clinic and non-clinic referred children. The children completed several self-report measures. including the FSSCR, the Revised Children's Manifest Anxiety Scale (RCMAS) (Reynolds and Richmond, 1978), and the Children's Depression Inventory (CDI). Child behavior problems were also assessed by a semistructured interview designed by the first author, the Anxiety Disorders Interview Schedule for Children, children and parent versions (the ADlS-C and ADIS-P) (Silverman & Nelles, 1988), to yield one overall diagnosis. In contrast to other

existing child interviews (e.g., the DICA, KIDDIE-SADS), the ADlS-C and ADIS-P are more specific for diagnosing childhood and adult anxiety disorders in children. The interviews also permit the clinician to rule out alternative diagnoses such as affective disorders and to provide quantifiable data concerning anxiety symptomatology, etiology, course, and a functional analysis of the disorder. The ADlS-C and ADIS-P have been found to be reliable tools for diagnosing behavior problems in children with a special focus on anxiety disorders (Silverman and Nelles, 1988). (A copy of the interview schedule is available from the first author upon request.) Specifically, agreement between two independent raters on both the child and the parent interviews for a primary diagnosis has been calculated using the kappa statistic (Heiss, 1971). The overall kappa coefficient has been found to be 0.84 for the ADIS-C, 0.83 for the ADIS-P, the 0.78 for the composite diagnosis (based upon combining the ADlS-C and ADlS-P data). These kappas indicate a high level of agreement for specific diagnostic categories. Results

Accuracy a/Child Behavior Problems as Reported by Patients Methods of identifying child behavior problems were examined first to determine their accuracy. Child behavior problems were identified from the CBCL if any of the individual CBCL scales or the total CBCL raw score was greater than the upper limit of the normal range reported by Achenbach and Edelbrock (1983) . The correlation between problems reported on the CBCL and problems identified during the clinical interview with the child was 0.67. An agreement of 74% was found between behavior problems reported by the parent in the parent interview and the behavior problems reported by the child in the child interview. A parent reported a current behavior problem that was not confirmed by a clinician when interviewing the child in just three cases, the sons of two AG parents and the daughter of one GAD parent. Thus, these data support the accuracy of the methods used in this study and suggest that anxiety disorder patients do not tend to overestimate problem behaviors in their offspring.

Parental Descriptive Data Table I presents descriptive data for the parents diagnosed with AG, PD, and GAD, as well as the group of mixed phobics. The identified patient/parent was the mother in all cases except three; the AG group, the GAD group, and the PD group each had one father who was the identified patient. In the AG group, none of the individuals had a concurrent primary diagnosis; two AGs had secondary diagnoses of simple phobia of heights and hypochondria. In the GAD group, one individual had a past episode of panic disorder; one also had two secondary diagnoses, social phobia and dysthymia. Of the five in the PD group, one had a concurrent primary diagnosis of dysthymia, one had a concurrent diagnosis of social phobia, and three had secondary diagnoses of simple phobia. The mixed phobic group consisted offour individuals, one with obsessive-compulsive disorder, one with social phobia and two with both obsessive-compulsive disorder and simple phobia. In addition, three of the four mixed phobics had secondary diagnoses of social phobia.

781

CHILDREN OF PARENTS WITH ANXIETY DISORDERS TABLE

1. Descriptive Datafor Parent Diagnostic Categories

Variable No. of families M:F ratio Average age Average no. of children Duration (yrs) Severity ( 1.17) Panic symptoms Panic frequency per week Avoidance

AG

PD

GAD

Mixed

14 1:13 36.92 (S.38)

S 1:4 40.2 (4.02)

S 1:4 36.0 (2.3S)

4 0:4 38.2S (2.22)

2.4 9.64 (6.4S) S.23 (0.00) 9.00 ( 1.9S) 4.09 (7.38) 12.71 (S.S4)

2.0 12.4 (16.60) S.OO (O.SS) 9.86 (2.41) 4.00 (2.00) 8.60 (3.21)

1.8 28.3 (13.9) 4.40 (1.29) 4.00 (S.OS) 0.40 (O.SS) 0.80 ( 1.79)

2.0 23.S· ( 11.7) S.SO 6.S0· (2.38) 3.S0 (4.04) 9.00' (4.97)'

Note: AG = agoraphobics with panic attacks. PD = panic disorder, GAD = generalized anxiety disorder, Mixed = mixed phobic group. Standard deviations are in parentheses. ·p
One-way analysis of variance indicated that both the GAD and the mixed phobic groups tended to have longer duration of disorder than the AG and PD groups (F[3, 24] = 6.36, p < 0.01). None of the groups was found to differ significantly, using a Games-Howell modification of the Tukey procedure, which is a stringent correction for heterogeneity of variance. The AG group had higher avoidance ratings than the mixed phobic, PD and GAD groups (F[3, 27] = 8.01, p < 0.01). A Games-Howell modification found the GAD group to be significantly lower than the other three groups. Both the AG and the PD groups reported more panic symptoms than either the mixed phobic or the GAD groups (F[3, 27] = 5.11, p < 0.0 I). None of the groups was found to differ significantly, using the Games-Howell modification. Although the AG and PD groups had almost identical mean scores on frequency of panic attacks, the range for panic episodes was three times greater for the AG group. Similar to previous reports (Barlow et aI., 1986), panic attacks were reported across all four diagnostic categories. Overall, these data confirm that the AG group has the highest rate of avoidance behavior and that the rate of panic attacks does not distinguish AG from PD. Behavior Problems Displayed by the Children

Whether the children of the anxiety disorder patients had more behavior problems on average than the children who comprised the CBCL normative sample was examined next. T-tests were conducted to compare the total CBCL raw scores of all the children in our sample with the appropriate age and gender norms from Achenbach and Edelbrock's (1983) nonclinical sample. When siblings fell into the same age-gender group, only the oldest child's scores were included. On the average, the CBCL total raw scores for our sample were higher than the normative sample for all four age-gender groups. The 6- to II-year-old girls and the 12- to 16-year-old girls both had significantly higher mean total behavior problem scores than did the respective nonclinical samples (see Table 2). Though not significant, the boys in our sample had con-

2. CBCL Total Mean Raw Scores for Offspring of Patients with Anxiety Disorders Compared with the CBCL Nonclinical Samples

TABLE

Group Girls 6-11 Mean SD N=9 Girls 12-16 Mean SD N=9 Boys 6-11 Mean N= 13 Boys 12-16 Mean SD N= II

Offspring

Nonclinical Sample"

39.S0 IS.60

19.90 14.20

3.10·

24.10 IO.S0

16.60 14.10

2.80·

29.90 IS.70

21.70

1.74

29.80 18.60

17.S0 IS.60

T

is.oo 1.98

Q: Data for the nonclinical samples are taken from Achenbach and Edelbrock, 1983. • P < O.OS.

sistently higher total behavior problem scores than the nonclinical samples. The number of children whose total CBCL raw scores would place them in the clinical range was examined next. Of the 42 children in our sample, five of 24 boys (two from the same family) and eight of 18 girls (two from the same family) had total CBCL raw scores that were in the clinical range. These data are displayed in Tables 3 and 4. Four boys and six girls were assigned diagnoses by the clinical interviewers but were not rated by their parents on the CBCL as having an identifiable behavior problem. Three of these four boys and three of these six girls had parents who were diagnosed as AG. The parents of the remaining four children were part of the mixed phobic group. These data again demonstrate that anxiety disorder patients, especially those with AG or other phobias, do not overestimate behavior problems in their children. Tables 3 and 4 reveal that 12 boys (50%) and 13 girls (72%) were identified as having a behavior problem based upon either or both parental reports on the CBCL or upon interviewing the child. With the exception of one boy who was diagnosed with oppositional disorder plus simple phobia, and one girl and one boy who were both diagnosed with attention deficit disorder, the clinical diagnoses derived from the ADIS-C were all anxiety related. To further examine the relationship between parent diagnosis and child maladjustments, the children were reclassified according to the diagnosis of the parent/patient. Seventeen of the 21 (81%) children with an AG parent, two of the seven (29%) children with a GAD parent, none of the six children with a PD parent, and six of the eight (75%) children with a mixed phobic parent received a clinical diagnosis and/or had one or more CBCL scale scores in the clinical range. The chisquare analysis of these data was significant (x~ = 16.40, p < O.OS) suggesting that the presence of phobic or avoidance behavior is highly associated with child behavior problems.

782

SILVERMAN ET AL.

TABLE

3. CBCL Scores and Interview Diagnoses for Boys of Parents

TABLE

4. CBCL Scores and Interview Diagnoses for Girls of Parents

with Anxiety Disorders

with Anxiety Disorders Subject"

Age

I,

9

2, 3, 4 5 6, 7 8,

8 7 II II

12 13, 14, 15. 16

6 13 12 16 15 15 17 15 14 16 6

17

II

18

6 7 12 8

9 10 II

19. 20~

21~ 226 236 24

II 9

12

Parent Ox " A+P A+P A+P A+P A+P A+P A+P A+P A+P A+P A+P A+P PO PO PO GAO GAO GAO GAO GAO SP SP OC OC

Total CBCL Score'

17 36

CBCL Scales in Clinical Ranged d. int

4

45 59

f, Sum c, e, int, ext

60

k, g. h, ext

27

g

19

10 19

20 18 31 15 24 18 45 53 25 18

h. j. int, ext a, b. c, f, i. int

Subject" Age None Sop.SP None None None None OPP. SP SP,OA Av, SP SA,OA Av, SP SP None None None None None None None Attdef OA SP None None

" Subjects having the same subscript are siblings. "A+P = agoraphobia with panic. PO = panic disorder. GAO = generalized anxiety disorder. SP = simple phobia. OC = obsessivecompulsive disorder. SoP = social phobia. OA = overanxious disorder, Opp = oppositional disorder. Attdef = attention deficit disorder, SA = separation anxiety disorder, Av = avoidance anxiety disorder. ' Scores in italics are above the upper limit of the nonclinical samples. ,J a = schizoid/anxious, b = depressed. c = uncommunicative, d = obsess/comp, e = somatic complaint, f = social withdrawal. g = hyperactive, h = aggressive, i = delinquent.j = immature, k = hostile withdrawal, int = internalizing, ext = externalizing.

I, 2,

10 II

Total CBCL Scales Parent in Clinical CBCL Ox " Range d Score ' A+P A+P

3 4 5

10 15 15

A+P A+P A+P

6 7 8

18 13

9

15

10

9

II

15 14 15

A+P A+P A+P A+P PO PO PO GAO GAO SoP SoP OC OC

12 13 14, 15, 16, 17, 18,

10

9

10 13 7 10

52 66

46 27 44

int , ext

a. b. c. d. into ext , Sum int, Sum e, f, int, Sum

17 60 60

14 23 28 26 51

g, h.

14 27 14 30 36

c, int

Child Ox " SP SP

OA OA SoP,SP SP,SoP SP,OA SP. OA, SoP Past episode PO None None None None None Attdef Av OC OC

" Subjects having the same subscript are siblings. " A+P = agoraphobia with panic, PO = panic disorder, GAO = generalized anxiety disorder, OC = obsessive-compulsive disorder, SoP = social phobia. SP = simple phobia, OA = overanxious disorder. Attdef = attention deficit disorder, Conduct = conduct disorder, Av = avoidant anxiety disorder. ' Scores in italics are above the upper limit of the nonclinical samples. d a = depressed. b = social withdrawal. c = sex problems, d aggressive, e = anxious/obsessive. f = somatic complaints. g = depressed/withdrawn, h = delinquent. int = internalizing. ext = externalizing.

=

(N = 3) families had more than one child identified as having

a behavior problem . Internalizing, externalizing. and total CBCL raw scores for each of the four diagnostic categories were also compared. Although not significant, the AG group tended to score higher than the other three groups for total behavior problems, and higher than the GAD and PD groups and equal to the mixed phobic group on both internalizing and externalizing scales. Only one significant difference emerged on the children 's questionnaire data. Children with an AG parent scored significantly higher on the FSSC-R than the children with a GAD parent (F[3 , 37] = 3.20, p < 0.05). The phobic and PD groups had scores that fell between these two groups. The number of families that had at least one child whose score had an identified behavior problem was also examined , since some of the children in this sample were siblings. Eleven of the 14 AG families. none of five PD families, one of five GAD families, and three of four mixed phobic families had one or more children who were identified as having a behavior problem that was significantly different (x~[3] = 6.59, p < 0.05). In addition, only the AG (N = 4) and the mixed phobic

Parental Avoidance as a Risk Factor

In addition to parental diagnosis, the above data suggest that parent diagnoses associated with higher levels of avoidance behavior (i.e., AG and mixed phobics) may present the greatest risk to the offspring. To examine this hypothesis, the children were classified according to the avoidance scores of their parents. regardless of parental diagnosis. Avoidance scores were obtained from the ADIS. the diagnostic interview administered to the patients (DiNardo et al., 1983). On the ADIS. the clinician rates each patient's avoidance in 23 commonly encountered situat ions on a 0 to 4 point scale ranging from no avoidance (enters without difficulty) to very severe avoidance (never enters a situation even with a safe person). The low avoidance group had avoidance scores between 0 and 9 whereas the high avoidance groups had scores between 10 and 20. The children of these patients were classified into either a low or high avoidance group according

CHILDREN OF PARENTS WITH ANXIETY DISORDERS

to a median split of their parents' avoidance scores. There were 19 children in the low avoidance group and 23 children in the high avoidance group. Nine of 19 (47%) children in the low avoidance group were assigned a clinical diagnosis and/or had one or more CBCL score in the clinical range; 15 of23 (65%) children in the high avoidance group received a clinical diagnosis and/or had one or more CBCL scale scores in the clinical range. Although the children in the high avoidance group tended to have higher internalizing, externalizing, and total CBCL raw scores than the children in the low avoidance group, these differences were not significant. On the children's questionnaire data, the high avoidance group had significantly higher trait anxiety scores than the low avoidance group (1[38] = -2.02, p < 0.050). No other fear or anxiety measure was significantly different between the two groups. Since the occurrence of panic attacks has also been identified as a key factor in the development of psychopathology in first degree relatives (Weissman et al., 1984), it seemed important to clarify the roles of panic attacks and avoidance behaviors in the increased risk associated with anxiety disorders. Thus, subjects were reclassified as offspring of AG patients with low panic frequency; offspring of AG patients with high panic frequency; offspring of PD patients (i.e., patients with panic attacks but no or very low rates of avoidance behaviors); and offspring of the mixed phobic group. The group of youngsters ofan AG parent with low panic frequency contained 12 children, the AG group with high panic frequency contained nine children, the PD group contained six children, and the mixed phobic group contained 25 children. The AG groups with low and high panic frequency had, respectively, 10of 12 (83%) and seven of nine (78%) children who had one or more behavior problem identified on the parental CBCL and/or the child interview. Eight of 15 children (53%) in the mixed phobic group were identified as having a behavior problem. None of the six children in the PD group, however, had an identified behavioral problem. Similar findings appeared when the data for families rather than for children were examined. Six of the seven low panic AG families and five of seven of the high panic AG families had children with identifiable behavior problems. Five of nine (65%) mixed phobic families had a child with an identifiable behavior problem, and none of the PD families had any children with identifiable behavior problems. Discussion Despite the relatively small sample sizes of this study, consistent findings emerged. Irrespective of panic attack frequency, the AG groups consistently had the most children with identifiable behavior problems, followed by the mixed phobic group and the PD group. Since both the low and high panic frequency AG groups had high rates of avoidance behavior but differed in their panic frequencies, it is likely that it is the occurrence of avoidance behavior rather than the rate of panic attacks that is primary in the development of child maladjustment. The relatively high rate of problem behaviors among the children of the mixed phobic group whose parents have, on average, rates of avoidance behavior

783

higher than PD patients but rates lower than AG patients, provides further support for this notion. Since the course of the childhood problems identified in this study is unclear, the children's developmental progress is being monitored. One year follow-up data from approximately one quarter of the sample reveals that the identified behavioral problems have persisted. If this trend continues, it will suggest that the identifiable problems may be early markers of greater susceptibility of anxiety related problems. It is premature to state definitively what it is about agoraphobic/avoidance behavior that may put children at risk, but a few speculations can be offered. As Rosenbaum et aI. (1988) noted, exposure to an AG parent may predispose a child to develop a more anxious stance by offering a model of caution and fearfulness. There have also been reports that the families of AG patients were overprotective of them as children (Roth, 1959; Terhune, 1949). Solyom and his colleagues ( 1974, 1976) have reported that mothers of AG patients score significantly higher than controls on measures of maternal control and concern than the overprotective mothers on whom the scale norms were based. This would suggest, therefore, that an AG parent may instill anxious attachment in the child. Clinical experience has also indicated that AG parents who avoid a large number ofdifferent situations frequently develop a rule system, either implicit or explicit, focusing on phobic symptomatology. These rules tend to be rigid and ritualistic, sustaining the parent's avoidance behavior. Many AG patients also deny their phobic problems to family members for fear of loss of self esteem and prestige within the family. Clinical observation has also shown that some patients insist that children remain home from school, so the patient has access to "a safe person." In light of the above, it is readily understandable how such a rigidly structured and seemingly irrational environment could contribute to behavior problems in the child. It has also been suggested that AG may represent the most extreme form of pathology among the anxiety disorders (e.g., Breier et al., 1986; Noyes et al., 1986). If this is true, it is possible that child maladjustment is associated with any parental mental disorder that is severe. In other words, the severity of the disorder, not the specific problem, is what is associated with child psychopathology (e.g., Campbell, 1984; Rutter, 1966). To more fully pursue this issue, the authors plan to include comparison groups of parents with different psychiatric disorders in future work. The above argument suggests that PD and AG are different points along the same continuum and that it is avoidance behavior that is the critical feature associated with the severity of the disorder. This then raises the question about the role of panic attacks as the critical variable associated with risk. In the authors' view, several important points weaken the suggestion that it is panic attacks as opposed to overt avoidance behavior that is the key element. One point is that panic attacks frequently occur in all the anxiety disorders (e.g., Barlow et al., 1986; Cerny et al., 1984). Indeed, in this sample, the number of panic attacks across the four diagnostic categories was not significantly different. Second, irrespective of panic attack frequency, the AG groups consistently had the most children with identifiable behavior problems. Certainly

784

SILVERMAN ET AL.

the relative contribution of parental panic attacks and avoidance behavior in predicting child maladjustment requires further research. The results from other risk studies (cf. Campbell, 1984), also underscore the importance of investigating the combined input of multiple risk factors, rather than examining these factors individually. This study is not without methodological limitations that should be noted , including the relatively small sample sizes studied and the absence of comparison groups (i.e., a normal comparison group and an other psychiatric disordered parent group). The absence of a normal comparison group is not a major problem, since the CBCL has normative data for both clinical and nonclinical samples that allowed the authors' to make comparisons. Statistical deviance may not necessarily imply pathology (Rutter et al., 1976); however, beyond using the normative data of the CBCL, it would have been useful if the study had used comparison groups from the local population. Also lacking in this study were ratings from additional "unbiased" sources, such as teachers or the other "normal" spouse. The use of multiple informants would have helped to generate a more complete, unb iased picture of the children's behavior across multiple settings . These data are being obtained in the authors' current work. In addition, these children need to be followed longitudinally to learn whether the identified problems are in fact early markers of anxiety. Finall y, one cannot determine from these data the nature of the familial factor (e.g., biological, psychological, etc.). Despite these limitations, the present study is a reasonable step in determining the nature of the familial influence in the anxiety disorders. References Achenbach, T. M. & Edelbrock, C. (1983), Manual f or the Child Behavior Checklist and Revised Child Behavior Profile. Queen City, Vt.: Queen City Printers. Barlow, D. H., DiNardo, P. A., Vermilyea, B. B., Vermilyea, J. & Blanchard, E. B. (1986), Cornorbidity and depression among the anxiety disorders. J. Nerv. Ment , Dis., 174:63-72. Breier, A., Charney, J. S. & Heninger, G. R. (1986), Agoraphobia with panic attacks. Arch . Gen. Psychiatry, 43: 1029-1036. Campbell, S. B. (1984), Research issues in clinical child psychology. In: Research Methods in Clinical Psychology, A. S. Bellack & M. Hersen, eds. New York : Pergamon. Cerny, J. A., Barlow, D. H. & Himadi , W. G . (1984), Issues in diagnosing anxiety disorders. Journal of Behavioral Assessm ent, 6:301-329. DiNardo, P. A., O'Brien, G. T ., Barlow, D. H., Waddel, M. T . & Blanchard, E. B. (1983), Reliability of DSM-III anxiety disorders using a new structured interview. Arch. Gen. Psychiatry, 22:10701078. Fisher, L., Kokes, R. F., Harder, D. W. & Jones, J. E. (1980), Child competence and psychiatric risk. J. Nerv. Ment . Dis., 168:353-355. Heiss, J. L., (1971) , Measuring nom inal scale agreement among man y raters . Psychol. Bull., 76:378-387. Harris, E. L., Noyes, R., Crowe , R. R. & Chaudhry, M. D. (1983), Family study ofagoraphobia. Arch. Gen. Psychiatry, 40:1061-1064.

Heatherington, E. M. (1979), Divorce: A child 's perspective . Am. Psychol., 34:851-859. Hodges, K., Kline, J., Stem, L., Cytryn, L. & McKnew, D. (1982), The development of a child assessment interview for research and clinical use. J. Abnorm. Child Psychol., 10:173-189. Inou ye, E. (1965). Similar and dissimilar manifestations of obsessivecompulsive neurosis in monozygotic twins. Am. J. Psychiatry, 121:1171-1175. Kovacs, M. & Beck, A. ( 1977), An empirical clinical approach towards a definit ion of childhood depress ion . In: Depression in childhood: Diagnosis. treatm ent and conceptual models, J. G. Schulterbrandt & R. Raskin , eds. New York: Raven. Noyes, R. Jr., Crowe, R. R., Harris, E. L., Hamra, B. J., McChesney , C. M. & Chaudhry, D. R. (1986), Relationship between panic disorder and agoraphobia. A rch. Gen. Psychiatry, 43:227-232. Ollendick, T. H. (1983), Reliability and validity of the revised fear survey schedule for children (FSSC-R). Behav. Res. and Th er., 21:395-399. Reynolds, C. R. & Richmond, B. O. (1978), What I think and feel. J. Abnorm . Psycho/., 6:271-280. Rose, R. J., Miller, J. Z., Pogue-Geile, M. F. et al. (1981), Twin family studies of common fears and phobias . In: T win research 3 Intelligen ce. personality and development, L. Gedda, P. Parisi & W. E. Nance , eds. New York : Alan R. Liss, pp. 169-174. Rosenbaum, J. F., Biederman, J., Gersten, M. et al. (1988), Behavioral inhibition in children of parents with panic disorder and agoraphobia. Arch. Gen. Psychiatry, 45:463-470. Roth, M. (1959), The phobic anxiety-depersonalization syndrome. Proceedings of the Royal Society of Medicine, 52:587-595. Rutter, M. (1966), Children of Si ck Parents. London: Oxford University Press. Rutter, M., Tizard, J., Yule, Woo Graham, P. & Whitmore, K. (1976), Research report: The Isle of Wright studies, 1964-1974. Psychol. Med., 6:313-332. Silverman, W. K. & Nelles, W. B. (1988), The Anxiety Disorders Interview Schedule for Ch ildren . J. Am. Acad. Child Adolesc. Psvchiatrv. 27:772-778. Solyom , L.:Beck, P; Solyorn , C. & Hugel, R. (1974), Some etiological factors in phobic neurosis. Canadian Psychiatric Association Journal, 19:69-78. Solyom, L., Silberfeld, M. & Solyom, C. (1976), Maternal overprotection in the etiology of agoraphobia. Canadian Psychiatric Association Journal, 21:109-113. Spielberger, C. D. (1973), Manualfor the state-trait anxiety inventory for children. Palo Alto, Calif.: Consulting Psychologists Press. Sylvester, c., Hyde, T. S. & Reichler, R. J. (1987), The diagnostic interview for children and personality inventory for children in studies of children at risk for anx iety disorders of depression. J. A m . Acad. Child Adolesc. Psychiatry, 26:668-675. Terhune, W. (1949), The phobic syndrome. Archives ofNeurol ogical Psychiatry, 62:162-172. Torgersen , S. (1979) , The nature and origin of common phobi c fears. Br. J. Psychiatry, 134:343-351 . Torgersen, S. (1983), Genetic factors in anxiet y disorders. Arch. Gen. Psychiatry, 40:1085-1089. Turner, S. M., Beidel, D. C. & Costello , A. (1987), Psychopathology in the offspring of an xiety disorder patients. J. Consult. Clin. Psych ol., 55:229-235. Weissman , M. M., Leckman, J. F., Merikangas, K. Roo Gammon, G . D. & Prusoff, B. A. (1984), Depression and anxiety disorders in parents and children. Arch. Gen. Psychiatry , 41:845-852. Young , J. P. Roo Fenton. G . W. & Lader, M.' H. (1971), The inheritance of neurotic traits . Br. J. Psychiatry, 134:343-351 .