Psychiatry
Research.
285
20,285-297
Elsevier
Searching Anxiety Anxious
for Evidence on the Validity of Generalized Disorder: Psychopathology in Children of Mothers
Naomi Breslau, Glenn C. Davis, and Kenneth Received
Prabucki
March 31, 1986; revised version received July 28, 19%6; accepted October 28. 1986.
Abstract. The diagnostic validity of generalized anxiety disorder (GAD) is tested by examining the relationship between GAD in mothers and children’s overanxious disorder (OAD), separation anxiety (SA), and anxious symptoms in 331 mother-child dyads from a geographically based probability sample. Data on the relationship between mothers’ major depressive disorder (MDD) and children’s depression are presented for comparison. The National Institute of Mental Health (NIMH) Diagnostic Interview Schedule (DIS) was used in mothers and the NIMH Diagnostic Interview Schedule for Children (DISC), in children. Children of mothers with GAD were not at increased risk for OAD, SA, or anxious symptoms. In contrast, MDD in mothers conferred a risk for OAD in younger children and of MDD in older children. Additionally, older children of depressed mothers exhibited significantly more depressive symptoms. The presence of diffuse anxiety in children of mothers with MDD may represent a nonspecific response pattern in psychiatrically vulnerable children. Like GAD in adults, these anxiety symptoms in children may constitute a prodromal manifestation of other disorders and transient responses to life stressors. Key Words. Generalized depression.
anxiety,
mothers,
offspring,
diagnostic
validity,
major
A comprehensive review of empirical evidence on the diagnostic validity of anxiety disorders concluded that whereas panic disorder has received consistent support as a distinct diagnosis, the validity of generalized anxiety disorder (GAD) remains in question (Breier et al., 1985). Although generalized, free-floating anxiety responds to anti-anxiety drugs, phenomenological, family, twin, and neurobiological studies failed to distinguish GAD as a distinct psychiatric disorder (Breier et al., 1985). Anxiety symptoms are common complaints during episodes of major depression and between panic attacks-clinical phenomena acknowledged in DSM-llfs text on the diagnosis of GAD (American Psychiatric Association, 1980). As Breier et al. (1985) point out, however, anxiety also characterizes most other psychiatric disorders and, instead of a disorder in its own right, might well be “a prodromal, incomplete, or residual manifestation of other psychiatric disorders” (p. 793).
Naomi Breslau, Ph.D., and Glenn C. Davis, M.D., are Associate Professors and Kenneth Prabucki, B.A., is Research Assistant, Department of Psychiatry, Case Western Reserve University, Cleveland, OH. Dr. Davis is also at the Cleveland Veterans Administration Medical Center. (Reprint requests to Dr. N. Breslau, Dept. of Psychiatry, Case Western Reserve University, Hanna Pavilion, 2040 Abington Rd.. Cleveland, OH 44106, USA.) 0165-1781/87/$03.50
@ 1987 Elsevier Science Publishers B.V.
286 Torgersen ( 1986) has argued that the evidence from family and twin studies speaks against a role for heredity in GAD and that “consequently, environmental factors have to be most important in the development of GAD” (p. 630). He failed to find, however, significantly more childhood separations in the histories of patients with GAD versus patients with panic disorder or agoraphobia with panic attacks, although the death of a parent before the patient reached age I6 was “almost statistically significantly” more frequent in patients with GAD (p < 0.10). In a previous report, we showed that the lifetime rate of GAD in a probability sample of women far exceeded recently estimated rates of other common psychiatric disorders. We showed, additionally, that the use of more stringent criteria, as recommended in a draft of the revised DSM-JJJ, defined a subset of GAD positives of whom a very high proportion (73%) met DSM-JJJ criteria for major depressive disorder (MDD) (Breslau and Davis, 19850). In a second report from the same study, we showed that episodes of GAD and MDD, in persons who met criteria for both diagnoses, occurred together rather than separately and that age of onset of the two diagnoses coincided in the majority of cases (Breslau and Davis, 19858). Evidence of this sort, we concluded, casts doubt on the validity of GAD as a distinct disorder. We suggested that the cluster of symptoms that define GAD in persons with MDD might characterize a subtype of MDD or might represent emotional distress associated with the experience of being psychiatrically ill. In this report, we extend further the examination of the diagnostic validity of GAD through another strategy. We present data on the relationship between GAD in our probability sample of women and symptoms of anxiety in their children. Previous reports documented that children of depressed parents were at an increased risk for MDD and depressive symptoms (Conners et al., 1979; Beardslee et al., 1983; Orvaschel, 1983; Weissman et al., 1984a, 19846) and that children of parents with agoraphobia were at an increased risk for school phobia (Berg, 1976). These reports indicate that reliable evidence about familial aggregation of psychiatric disorders can be found even among young offspring, and that above and beyond the overall excess in psychopathology among children of psychiatrically ill parents (Rutter, 1966), there tends to be a disorder-specific aggregation among the children of depressives and agoraphobics. Evidence that a disorder “breeds true” supports the validity of a diagnosis, irrespective of whether genetic or environmental interpretations are favored (Robins and Guze, 1970). There is little empirical evidence on the relationship between GAD in parents and anxiety disorders in children. Results from the Yale family study support neither a disorder-specific familial aggregation nor a significantly increased risk for other disorders in the offspring of depressed patients who also had GAD (Weissman et al., 1984~). However, inferences from that study must take into account several methodologic limitations. First, because all probands were MDD positives, the estimated risk in the children of probands with GAD as a concomitant diagnosis cannot be generalized to children of probands with GAD alone. Second, children’s diagnoses were not based on direct interviews with the children, but relied instead on other informants (primarily parents), a diagnostic procedure that might have biased the results, as the authors point out.
287
In this study, diagnostic assessment was based on direct interviews with the children and the epidemiological sampling design enabled us to compare children of mothers with GAD alone against children of mothers with MDD (with or without GAD) and with mothers who met criteria for neither GAD nor MDD. Evidence of an association between GAD in mothers and anxiety in children would support the diagnostic validity of GAD. Data on the relationship between MDD in mothers and children’s depression are reported for comparison
Methods Sample. Data were obtained from a geographically based probability sample of 357 families who served as controls in a study of childhood disability in Cleveland, Ohio. (See Breslau and Davis, 1986, for a description of the sample.) Mothers and children-one randomly selected child in families with two or more children-were interviewed. Data analysis presented here is on 33 I mother-child dyads on whom information necessary to generate the relevant DSM-III diagnoses was complete. Mean (i SD) age of mothers was 43 + 8 years, 80% were white, and mean years of schooling was 12.5. The children’s mean (* SD) age was 15.7 * 4.8, with a range from 8 to 23: 60% were 8- I7 years of age and 40% were 18-23. Female children slightly outnumbered males. Mothers and children were interviewed separately in their homes by trained lay interviewers. With few exceptions, the same lay interviewer conducted the mother’s and child’s interview. Although aware of respondents’ replies to individual items, the lay interviewers did not know the diagnoses of the mothers and the children, since diagnoses were generated at the data analysis stage by the application of computer algorithms. Moreover, the lay interviewers were unfamiliar with diagnostic criteria and unaware of the research questions addressed in this analysis. Mothers’
Assessment. The mothers’ interview schedule included sections from the National Institute of Mental Health (NIMH) Diagnostic Interview Schedule (DIS) that produce data for diagnosing DSM-III MDD and GAD. The DIS, designed for use in the Epidemiological Catchment Area Program (ECA), is a fully structured interview that can be used by trained lay interviewers to generate psychiatric diagnoses in DSM-III and other major classification systems. The validity of the DIS as administered by lay interviewers was tested in several patient and nonpatient populations, using a variety of approaches (Robins et al., 198 I, 1982; Hesselbrock et al., 1982; Helzer et al., 1985). The diagnosis of MDD fared well in the validation tests (Hesselbrock et al., 1982; Robins et al., 1982; Helzer et al., 1985). In the lay interviewer-psychiatrist comparisons, MDD’s kappa was 0.63, sensitivity was 80%, and specificity was 84% for the clinical samples (Robins et al., 1982), with a somewhat lower kappa and sensitivity, but higher specificity for the general population sample (Helzer et al., 1985). Information on the adequacy of the DIS diagnosis of GAD has not been previously reported, because the disorder was not covered in the NIMH ECA Program. The DIS section on GAD was developed for use in the second wave of the St. Louis ECA survey. The sequence of questions and the computer algorithms for making the diagnosis of GAD closely follow DSM-If/ rules. In a previous publication from this study, we reported that the 6 months’ test-retest kappa for GAD was 0.75 (Breslau and Davis, 1985~). Lay interviewer-psychiatrist kappa for the diagnosis of DSM-III GAD in a patient population was 0.67 (Breslau and Davis, in press). In this study, we replicated the field methods of the ECA surveys with respect to the DIS format, the interview situation, and interviewers’ training (see Eaton et al., 1984, for a discussion of these methods). Children’s
Assessment.
with the children
Data on children’s psychopathology are from direct interviews in which the NIMH Diagnostic Interview Schedule for Children (DISC)
288 was used.’ The DISC was designed to generate information necessary to yield DSM-III diagnoses in children (Costello et al., 1984; Edelbrock and Costello, 1984; Edelbrock et al., 1985). Although the DISC was developed for assessing psychiatric disorders in children, we used it on the entire sample of offspring who were 8 to 23 years of age. An instrument designed for diagnosing adults might have been more appropriate for those 18 to 23 years old, given that I8 years of age serves as a conventional boundary between childhood and adulthood in psychiatry. Because of the advantages of obtaining uniform data on the entire sample and the young age of those 18 and over, we preferred to use the DISC as the single diagnostic interview for this sample. Like the DIS, the DISC is a fully structured interview schedule, which specifies the exact wording and sequence of questions and provides a complete set of categories for classifying respondents’ replies (Costello et al., 1984). The highly structured format is intended to minimize clinical judgment in eliciting diagnostic information and recording responses. It is designed to be administered by lay interviewers who are trained to follow the interview schedule faithfully. Test-retest and interrater reliabilities on children’s interviews have been high. (See summary of available psychometric data in Orvaschel, 1986.) Test-retest reliability of children’s reports of symptom scales, constructed from diagnostically relevant DISC items, varied by age: reliability was generally higher in children 10 years of age and older (Edelbrock et al., 1985). In this study, we replicated the field methods used in the DISC’s validation studies, with respect to the training of lay interviewers, the administration of the instrument to children, and the application of computer algorithms that generated DSM-III diagnoses and symptom scales (Costello, 1985). Most of the children in this study were above the age at which reliability was found to be typically low (Edelbrock et al., l985).2 The analysis presented here is on DISC-generated DSM-I/I diagnoses of overanxious disorder, separation anxiety, and major depressive disorder (M DD). DSM-II/k criteria for overanxious disorder define a syndrome in which the predominant disturbance is generalized and persistent worry that is not focused on a specific situation or object. Thus, the diagnosis of overanxious disorder shares key features with GAD in adults, except for the differential emphasis placed on physical concomitants of anxiety in the two diagnoses. In overanxious disorder, symptoms of autonomic hyperactivity and motor tension are included among the seven symptom groups, but criteria can be met in the absence of any physical symptom if there are positive symptoms in four of the five other groups that comprise various aspects of anxious preoccupations. In GAD, the anxious mood must be accompanied by some configuration of physical symptoms. In separation anxiety, on the other hand, the excessive anxiety must be attributable to separation from major attachment objects. The definition of DSM-III MDD in children is virtually the same as in adults. We also analyzed data on DISC depression and anxiety symptom scales, constructed by Edelbrock et al. (1985), by adding items used to generate the diagnoses described above. A depression scale, with 35 items, comprised 4 subscales, each measuring a separate domain: affective (I 3 items), cognitive (6 items), vegetative (7 items), and suicidal (9 items). The scale measuring overanxious symptoms comprised 9 items used in the diagnosis of overanxious disorder, and the scale measuring separation anxiety comprised I2 items that constitute the diagnosis of that disorder. (See Appendix for a brief description of items.)
I. The Keith
DISC
was developed
Conners,
Ph.D.,
under
used in this research
was written
Western
Institute
Psychiatric
2. A parallel mothers childrenS
parent
form
as informants assessment.
contract
in conjunction
to NIMH
confound
by Barbara
the staff of the Division
and validated
and Clinic,
of the DISC,
would
with
under
Pittsburgh,
DISC-P,
contract
Herjanic,
to NIMH
PA (Costello
was administered
the results of this analysis,
M.D.,
of Biometry
Joaquim
Puig-Antich,
and Epidemiology.
by Anthony
Costello,
NIMH.
M.D..
M.D.,
and
The version
and colleagues
at
et al.. 1984). to mothers.
Because we were concerned
we relied on the children’s
that the use of
replies on the DISC
for the
289 Results Anxiety and Depressive Disorders in Mothers and Children. The lifetime rate of GAD in mothers was 44$!$ and of MDD, 17%. The lifetime rate of MDD in the mothers is similar to the lifetime rate found in the New Haven ECA survey for women in the same age range (see Breslau and Davis, 1985h, footnote number I). There is no comparable ECA information on tiAD. The rate of GAD in our probability sample appears inordinately high, an observation consistent with the view that GAD, as defined in DSM-III,is confounded with transient stress reactions and symptoms associated with other psychiatric conditions (Spitzer and Williams, 1984). More than 80% of MDD positives also met criteria for GAD. Among the 200 children 8-17 years of age, the I-year rate of overanxious disorder was 14(ro, the rate of separation anxiety was 7%, and of M DD. 6.50/u. Among the I3 I children 18-23 years of age, the rates of overanxious disorder and separation anxiety were lower-7% and l.S%, respectively-but the rate of MDD was higher--- 16%. There are few epidemiological data with which our estimates of the children’s rates of disorder can be compared. The rate of MDD in the younger age group is close to that reported recently for a geographically based epidemiological sample, in which a similar assessment procedure was used (Cohen et al., 1985). The rate in the older age group is higher than that reported for the 18-24 age range in the ECA project (Robins et al., 1984). Table 1 presents the numbers and rates of overanxious disorder, separation anxiety, and MDD in children of mothers who were positive for GAD only (i.e., negative for MDD) and in children of mothers negative for GAD and MDD. Data are presented separately for children 8-17 and 18-23 years of age. Rates of disorder in children of GAD positives were only slightly and not significantly different from rates in children of mothers negative for GAD and MDD. This was the case for the younger (8- 17) and the older ( 18-23) groups. Table 2 presents the numbers and rates of diagnoses in offspring of mothers positive and negative for MDD. Because most mothers with MDD also met criteria for GAD-24 of 30 mothers with children 8-17
Table 1. Rates of DSM-/I/ disorders in younger (8-17) and older (18-23) offspring of mothers with and without gene&z4 anxiety disorder (GAD) Child’s diagnosis (ages 8-17) Mother’s diagnosis GAD
only positives
Neither GAD
(n = 54)
nor MDD
(n =
116)
OAD
SA
MDD
6 (11%)
5 (9%)
6 (11%)
14 (12%)
5 (4%)
6(
(ages GAD
only positives
Neither GAD
(n = 47)
nor MDD
-
(n =
3 58)
4
( 6%) ( 7%)
5%)
18-23)
1 (2%)
6 (13%)
0 (0%)
8 (13%)
Abbreviations: OAD = overanxious disorder. SA = separation anxiety. MDD = major depressive disorder.
290
Table 2. Rates of DSM-//I disorders in younger (8-17) and older (18-23) offspring of mothers with and without major depressive disorder (MDD) Child’s diagnosis (ages 8-17) Mother’s
diagnosis
MDD
positives
MDD
negatives
MDD
positives
MDD
negatives
(n = (n =
OAD
30) 170)
SA
MDD
8 (27%)’
4 (13%)
1 ( 3%)
20 (12%)
10 ( 6%)
12 ( 7%)
(ages 18-23)
Abbreviations:
(n = (n =
26) 105)
2 ( 8%)
1 ( 4%)
7 ( 7%)
1 (1%)
8 (31%)’ 13 (12%)
OAD = overanxious disorder. SA = separation anxiety.
1.p < 0.05. and 23 of 26 mothers with children 18-23-we do not present data for the MDD-only groups. Note that the comparisons in this table are with children of MDD negatives, of whom almost half met criteria for GAD. (The rates in offspring of mothers with neither MDD nor GAD, presented in Table I, are similar to the rates in offspring of MDD negatives; comparisons with those subjects would, therefore, lead to similar conclusions.) Analysis of children’s rates of disorder according to mothers’diagnosis of MDD showed an increased risk for overanxious disorder in the younger subset and an increased risk for MDD in the older (Table 2). In each case, the increase in the rate of disorder in the offspring of depressed mothers vs. offspring of mothers who were not depressed was more than two-fold and statistically significant (p < 0.05). These results, taken together, indicate that the diagnosis of DSM-III GAD in mothers is unrelated to anxiety disorders (or depression) in the offspring, whereas the diagnosis of MDD in mothers confers a risk for overanxious disorder in children 8-17 and for MDD in children 18-23 years of age. and Depressive Symptoms in Children. The relationship between mothers’ psychiatric diagnoses and children’s symptoms of anxiety and depression was estimated in a series of multiple regression analyses in which the separate effects of age and sex of child were also assessed (see Table 3). Each of the four DISC depression subscales, the total depression scale, the overanxious symptom scale, and the separation anxiety symptom scale were regressed on child’s age (in years), sex (coded I if female, 0 if male), and mother’s diagnostic classification (MDD, GAD, neither), represented by two dummy variables, MDD (coded I if positive, 0 if negative) and GAD (coded I if positive, 0 if negative), with mothers negative for MDD and GAD serving as the reference group. Unstandardized partial regression coefficients are presented and standard errors of the coefficients are in parentheses. The unstandardized partial regression for the variable mother’s GAD is an estimate of the mean excess in number of symptoms in offspring associated with a diagnosis of GAD in mothers. (This is so, because the term MDD estimates and partials out the symptom differential associated with the diagnosis of MDD versus ho disorder.) By the same logic, the coefficient for the variable mother’s MDD represents the excess
Anxious
291 in number of symptoms associated with the diagnosis of MDD in mothers. Thus, each of the two coefficients estimates the unique association between mother’s specific disorder-MDD or GAD-and child’s symptomatology. The additive effect of MDD and GAD (i.e., the excess symptomatology in offspring of mothers with both disorders) is equal to the sum of the two coefficients. Put in other words, these multivariate analyses produce separate estimates of the excess symptomatology in children associated with mother’s diagnosis of MDD alone, GAD alone, and the two disorders together, as compared to neither. As can be seen in Table 3, offspring of mothers with GAD alone were indistinguishable from offspring of mothers with no disorder on any symptom scale. In contrast, children of mothers with MDD had statistically significant excesses in cognitive and vegetative symptoms. In each area, children of mothers with MDD reported approximately 1 symptom more than children of mothers with no disorder. On the total depressive symptom scale, the excess associated with MDD in the mothers was 3.760 symptoms. The results also show that the presence of GAD as a second diagnosis in mothers who met criteria for MDD adds a slight and statistically insignificant increment to their children’s symptomatology, given that the coefficient for GAD was not significant.
Table 3. Regressions of symptom scales on child’s age and sex and mother’s diagnosis (n = 331) Child’s symptom scales Affective
symptoms
Cognitive
symptoms
Vegetative Suicidal
symptoms symptoms
Child’s age B (SE)
Child’s sex B
(SE)
Mother’s MDD
Mother’s GAD
B
B
(SE)
(SE)
R*
0.223’
(0.045)
1.092’
(0.422)
1.159
(0.595)
0.299
(0.451)
0.11
0.026
(0.025)
0.355
(0.236)
0.793’
(0.333)
-0.054
(0.252)
0.03
0.164’
(0.037)
0.366
(0.354)
1.300’
(0.499)
0.192
(0.378)
0.09
0.047
(0.029)
-0.015
(0.274)
0.412
(0.386)
0.280
(0.292)
0.02
Total depressive symptoms Overanxious Separation
0.465’ symptoms anxiety
(0.105)
1.777
(0.997)
3.760’
(1.404)
0.648
(1.063)
0.10
-0.050
(0.071)
1 .012
(0.679)
1.159
(0.957)
-0.148
(0.724)
0.01
-0.576’
(0.074)
0.917
(0.701)
1.136
(0.987)
1.158
(0.747)
0.17
1. Coeffment exceeds twice its standard error.
Several other aspects of the results are worth noting. Two depression subscalesaffective and vegetative-and the total depression scale had significant positive regression slopes on child’s age. Separation anxiety symptoms declined with age, approximately one half a symptom with each year. Child’s sex was related to affective symptoms, with females exceeding males by approximately one symptom. To describe the relationship between mothers’ disorders and children’s symptomatology more clearly, we present in Tables 4 and 5 the means (SD) of symptom scales in the children, classified according to mothers’diagnosis. Because of the small number of mothers who were positive for MDD but not for GAD, we did not split the MDD group into those with and without GAD. Statistical significance of differences in group means was tested by a series of analyses of variance (ANOVAs), and
292 Table 4. Depressive and anxious symptoms in younger (8-77) offspring of mothers with MDD, GAD, or neither disorder Mother’s diagnosis MDD
GAD (n = 54)
(n = 30) Affective
symptoms
Cognitive
Suicidal
7.90
7.46
7.37
(2.72)
(4.15)
(3.97)
symptoms
Vegetative
3.00
2.61
2.32
(2.12)
(2.26)
(2.03)
symptoms
3.53
2.95
2.85
(3.48)
(2.88)
(3.09)
symptoms
Total depression Overanxious Separation
symptoms
symptoms anxiety
Total anxiety
symptoms
symptoms
Neither 116)
(n=
1.10
1.69
1.16
(1.88)
(2.91)
(2.20)
15.83
14.86
13.86
(7.40)
(9.62)
(8.93)
12.77
10.68
11.35
(6.64)
(6.35)
(6.08)
10.87
8.52
7.95
(8.62)
(7.95)
(6.88)
23.63
19.20
19.30
113.171
(11.56)
(10.41)
Note: Analyses of variance, ncne significantValues presented are means, with SD in parentheses. Abbreviations: MDD = major depressive disorder. GAD = generalized anxiety disorder.
Scheffe comparisons were performed when ANOVAs revealed differences significant at a < 0.05. A total anxiety symptom scale was added to this analysis by summing the two anxiety scales, overanxious and separation anxiety. Table 4 shows the results for children 8-17 and Table 5, for children 18-23. In the younger group, none of the comparisons yielded significant associations between mother’s disorder and child’s symptomatology. In the older group, differences across the three subsets of children were significant in the vegetative (F = 4.59 I, p < 0.02) and suicidal (F = 3.644, p < 0.03) areas and, in each, children of depressed mothers scored higher than children of mothers with GAD or with no disorder. The cumulative effect of these differences resulted in a total depression scale mean of 22.7 in children of depressed mothers, compared to 16.19 and 15.69 in children of mothers with GAD and no disorder, respectively (F = 5.904, p < 0.005) (L@= 2/ 128 in all ANOVAs). In the total depression and the vegetative scales, children of depressed mothers exceeded significantly children in either of the remaining groups, according to Scheffe paired comparisons.3 3. Among
mothers
ofyoungeroffspring.
and 3, respectively, children either 4.
the younger
We
were
of mothers
tested
depressive
with
with MDD
symptomatology
versus 1X-23) was included age contingency children
24 were with diagnosea
plus GAD
MDD
and with
versus MDD
and GAD
MDD
alone.
alone but found
and 6 with We tested
MDD
alone,
and of older olfspring.
the differences
no Ggnificant
differences
in the mean
23
score> I”
on any of-the x&s
for
or the older subset\.
the statistical
scale regressions,
both
significance in children.
in a series of multiple
the interaction of the depressive
of depressed
of the age speaficity A statistical
mothers.
of mother’s aggregation
interaction
regression
M DD
01 the term
analyses.
relationship between
In the total
and child’s age had significant
in mothers
and children
between
mothers’
MDD
maternal
depression
and children’s
and
age (X-17
depression
scale and in the vegetative
coefficients
(p <0.02).
or, conversely,
the specificity
supporting
the
01 the age effect to
293
Table 5. Depressive and anxious symptoms in dder (78-23) offspring of mothers with MDD, GAD, or neither disorder Mother’s diagnosis MDD (n = 26) Affective
symptoms
Cognitive
symptoms
Vegetative Suicidal
symptoms’
symptoms’
Total depression
symptoms2
Separation
symptoms anxiety
Total anxiety
symptoms
symptoms
Neither (n = 56)
10.52
8.66
8.38
(4.15)
(3.60)
(4.04)
3.44
2.29
2.43
(2.39)
(1.75)
(2.32)
5.85
3.83
3.53
(4.46)
(3.16)
(2.95)
2.59
1.29
1.07
(3.61)
(2.38)
(1.85)
22.70
16.19
15.69
(7.85)
(8.76)
(11.95) Overanxious
GAD (n = 47)
10.73
10.46
(5.40)
(4.66)
9.52 (6.70)
5.81
5.53
3.72
(5.15)
(5.84)
(3.73)
16.54
16.10
13.24
(8.89)
(8.79)
(9.09)
Note: Scheffb comparisons
yielded the following results: (1) For vegetative and total depression scales, MDD group was significantly different @ < 0.05) from GAD and from neither. (2) For suicidal scale, MDD group was significantly different (p < 0.05) from neither. Values presented are means, with SD in parentheses.
Abbreviations:
MDD = major depressive disorder. GAD = generalized anxiety disorder.
1. p < 0.03, analysis of vanance. 2. p < 0.005. analysis of vanance.
Note that the significant effect of age on depressive scales observed in the regression results (Table 3) is in fact confined to the subset of children of mothers with MDD. In children of mothers with GAD or with neither MDD nor GAD, differences in scores of depressive subscales between younger and older subsets were triviaL4 Younger offspring of depressed mothers, as well as younger and older offspring of mothers without MDD, manifested, on the average, similar levels of depressive symptomatology. The scores of subsets other than the older offspring of depressed mothers probably represent nonpathological levels of dysphoric mood and bodily complaints that do not increase with age. Thus, the deviation from this “normative” level of symptomatology, observed in older children of depressed mothers, may represent the emergence of pathological symptoms in vulnerable offspring.
Discussion The major findings of this analysis can be summarized as follows: (I) Children of mothers positive for DSM-I/J GAD were not at increased risk for DSM-III overanxious disorder, DSM-JJJ separation anxiety disorder, or symptoms of anxiety in general. (2) A diagnosis of MDD in the mothers conferred a risk of overanxious disorder in children %-I7 years of age and of MDD in children 18-23 years of age. Additionally, older offspring of mothers who met criteria for MDD showed
294
significantly more vegetative and suicidal symptoms and overall depressive symptomatology. In contrast with the results for MDD, we found no evidence of familial aggregation in connection with the diagnosis of GAD. Our findings on the risk for MDD and depressive symptoms in children of depressed mothers constitute an indirect but powerful check on our assessment methodology. Specifically, the failure to find an association between GAD in mothers and anxiety in children cannot be dismissed as due to the limitations of our assessment procedures, when using the same procedures we found, as others had previously, an association between mothers’ and children’s psychopathology in connection with depression. At first glance, it would appear that our evidence would have been more compelling had we been able to estimate rates in offspring of mothers positive for MDD but not for GAD. The paucity of MDD positives who were negative for GAD, we believe, is a consequence of the definitional problems in DSM-llfs GAD. A closer examination of the findings reveals, however, that the high rate of GAD in MDD positives does not limit our conclusions. First, our chief purpose was to examine the relationship between GAD in mothers and anxiety in children, and the comparisons bearing on this point are between children of mothers with GAD only versus mothers with neither GAD nor MDD. Second, our analysis indicates that the higher rates of disorder and the increases in symptomatology in children of mothers with MDD are not attributable to the presence of GAD as a second diagnosis: the unique contribution of maternal GAD to children’s symptomatology was not significant. Moreover, we also tested whether MDD and GAD in mothers had interaction effects on children’s symptomatology but found no evidence for such effects. (A significant interaction effect without a main effect for GAD would have suggested that GAD in mothers increases the children’s symptoms only when the mothers are also depressed. Such a finding, if observed, would not support the validity of GAD as a distinct diagnosis, but would instead support the validity of a subtype of MDD.) The lack of familial aggregation in GAD is consistent with our previous findings that GAD is confounded with MDD and that episodes of GAD and MDD, in persons with both disorders, coincide. These findings speak against the descriptive and construct validity of GAD as a psychiatric disorder. We did find in these data evidence in support of the diagnostic validity of MDD, evidence that is in accord with previously reported findings. Older children of mothers with MDD were at risk for MDD and showed, additionally, a significant excess in depressive symptoms. The age-specific risk we observed has not, however, been previously reported. In the Yale family study, which found a risk for depression in children of parents with MDD, the children were 6-18 years of age. In our study, the risk was statistically significant in children 2 18 but not in children X-17. The discrepancy in the results on children under 18 might be explained by a difference in the severity of MDD in the two studies. Clearly, the depressed patients who were the probands in the Yale study were more severely ill than the cases identified in our survey of the untreated (nonpatient) population. That a severity differential could account for the disparate results is suggested in findings from the Yale study that recurrent depression and familial loading increased the risk to children (Weissman et a!., l984h).
295
Older children in our sample did nof generally report more depressive symptoms than younger children. Instead, the observed increase in depressive symptoms among older offspring of mothers with MDD appears to signify the emergence (onset) of depressive disorder in children at high risk. As to the etiological nature of the risk, our findings, like previous findings of familial aggregation, are neutral. A high risk of depression in the offspring of depressed mothers may indicate genetic transmission or the influence of family environment (Conners et al., 1979; Weissman et al., 1984~). Note that the most marked excess in older offspring of depressed mothers was in the vegetative symptom cluster, a cluster that predominates in the clinical picture of endogenous, melancholic, and antidepressant-responsive subtypes of depression. The partial evidence of an increased risk of anxiety in the young offspring of mothers with MDD warrants comment. To recapitulate, we found among young offspring of depressed mothers an increased risk for overanxious disorder. Additionally, although not statistically significant, younger children of depressed mothers scored higher than other children on the anxiety symptom scales (see Table 4). If replicated in future research, the pattern of the combined results on anxiety and depression suggests that children at high risk for depression (because of depression in the mothers) show excessive anxiety when young but manifest symptoms of depressive disease, as well as higher rates of MDD, when they reach the age range in which the onset of depression typically occurs. Longitudinal studies of children of depressed parents are needed to test whether the presence of anxious symptoms identifies at an early age a subgroup of children who may be at especially high risk for depression. It may also be the case that the excessive anxiety in young children at high risk for MUD is not specific to MDD but, instead, represents a general response pattern in children who are psychiatrically vulnerable, whether genetically or environmentally. Put in other words, diffuse anxiety in children, as measured by the symptoms that define overanxious disorder, may not be different from GAD in adults. Both symptom configurations may constitute a prodromal manifestatio,r of a variety of psychiatric disorders. as well as transient responses to life stressors. Followup studies of children with these symptoms would clarify this question. Whether it is a transient response to stress or a more protracted distress associated with a variety of psychiatric disorders, according to the available evidence, GAD, defined as an anxious mood with its physiological concomitants, does not appear to constitute a discrete psychiatric disorder. Acknowledgment. This study was supported in part by NI H grant number H D-16821 and Research Scientist Development Award number K02 M H-00380 (Dr. Breslau) and Research Center Award MH-41684 from the National Institute of Mental Health. Rachel Gittelman, Ph.D., James Leckman, M.D., and Paul Ambrosini, M.D., made helpful comments.
References Psychiatric Association. DSM-III: Diugnostic and Statistical Manual of Mental 3rd ed. APA, Washington, DC (1980). Beardslee, W.R,., Bemporad, J., Keller, M.B., and Klerman, G.L. Chrldren of parents with major affective dtsorder: A review. American hufnaf of Psychiufry, 140, 825 (!983). American
Disorders.
296 Berg, I. School phobia in the children of agoraphobic women. British Journal qf Psychiatry, 128,86 ( 1976). Breier, A., Charney, D.S., and Heninger, G.R. The diagnostrc validity of anxrety drsorders and their relationship to depressive illness. American Journal of’ Psychiatry. 142, 787 ( 1985). Breslau, N., and Davis, G.C. DSM-III generalized anxiety disorder: An empirical investigation of more stringent criteria. Psychiatry Research. 15, 23 I ( 1985~). Breslau, N., and Davis, G.C. Further evidence on the doubtful validity of generalized anxiety disorder. Psychiatry Research, 16, I77 (I 9856). Breslau, N., and Davis, G.C. Chronic stress and major depression. Archives qf General Psychiutry. 43, 309 (I 986). Breslau, N., and Davis, G.C. Posttraumatic stress disorder: The etiologic specificity of wartime stressors. American Journal of’ Psychiatry (in press). Cohen, P., Velez, N.C., and Garcia, M. Epidemiology of childhood depression. Presented at the Annual Meeting of the American Academy of Child Psychiatry, San Antonio, TX, October 26 ( 1985). Conner% C.K., Himmelhoch, J., Goyette, C.H., Ulrich, R., and Neil, J.F. Children of parents with affective illness. Journal oj’ the American Academ.v of’ Child fs,~chiarr~~. 18, 600 ( 1979). Costello, A.J. Revised algorithms for DISC, October, 1983 (with minor revisions in 1985). Unpublished manuscript (1985). Costello, A.J., Edelbrock, C., Dulcan, M.K., Kalas, R., and Klaric. S.H. Development and testing of the NIMH Diagnostic Interview Schedule for Children in a clinic population: Final report (Contract #RFP-DB-81-0027). Center for Epidemiologic Studies, NIMH. Rockville. MD (1984). Eaton, W.W., Holzer, C.E. Ill, Von Korff, M.. Anthony, J.C., Helzer, J.E.. George. L.. Burnam, M.A., Boyd, J.H., Kessler, L.C., and Locke, B.Z. The design of the Epidemiologic Catchment Area surveys. Archives of’ General P.s~~c~hiutr,~. 41, 942 (1984). Edelbrock, C., and Costello, A.J. Structured psychiatric interviews for children and adolescents. In: Goldstein, G., and Hersen, M., eds. Handbook of Ps~~c~hologicalAssessment. Pergamon Press, New York (1984). Edelbrock. C., Costello, A.J., Dulcan, M.K., Kalas. R., and Conover, N.C. Age differences in the reliability of the psychiatric interview of the child. Child Developmenr. 56, 265 (1985). Helzer, J.E., Robins, L.N., McEvoy, L.T., Spitznagel. E.L., Stoltzman, R.K., Farmer, A., and Brockington, I.F. A comparison of clinical and Diagnostic Interview Schedule diagnoses. Archives of’ General P.y_ychialry, 42, 657 ( 1985). Hesselbrock, V., Stabenau, J., Hesselbrock, M., Mirkin. P.. and Meyer. R. A comparison of two interview schedules: The Schedule for Affective Disorders and Schizophrenia Lifetime and National Institute of Mental Health Diagnostic Interview Schedule. Archives o/‘Genera/ P.qxhiu/ry. 39, 674 ( 1982). Orvaschel, H. Parental depression and child psychopathology. In: Guze, S.B.. Earls, F.J., and Barrett, J. E.. eds. Childhood ~s~c,hol,otholo~.~, ontl Development. Raven Press. New York, p. 53 (1983). Orvaschel, H. Psychiatric interviews suitable for use in research with children and adolescents. P.~.vc,hol,harmac,ol~~~~Bullerin, 21, 737 ( 1986). Robins, E., and Guze, S.B. Establishment of diagnostic validity in psychiatric illness: Its application to schizophrenia. American Journal of P.sychio/ry. 126,983 (1970). Robins, L.N.. Heber. J.E.. Croughan, .I., and Ratcliff, K.S. National Institute of Mental Health Diagnostic Interview Schedule. Archives ~?/‘Genera/ f.~whiawv, 38, 381 (1981). Robins, L.N., Helzer, J.E., Ratcliffe, K.S.. and Seyfried, W. Validity of the Diagnostic Interview Schedule, Version II: I)SM-//I diagnoses. P.s~who/ogicd Medicine, 12, 855 (1982). Robins, I_.N., Helter, J.E.. Weissman, M.M., Orvaschel. H.. Gruenberg. E., Burke, J.D., and Regier, D.A. Lifetime prevalence of specific psychiatric disorders in three sites. Archives of’ General P.svchiurry. 41, 949 ( 1984). Rutter, M. Children of Sick Parents: An Environmental ontl P.s_vchiatric Stuc!c’. Oxford University Press, London (1966).
297 Spitzer, R.L., and Williams, J.B.W. Diagnostic issues in the DSM-III classification of the anxiety disorders. In: Grinspoon, L., ed. Psychiatry Updates. Vol. 111. American Psychiatric Press, Washington, DC, p. 395 (1984). Torgersen, S. Childhood and family
characteristics in panic and generalized anxiety disorders. American Journal of Psychiatry. 143,630 (1986). Weissman, M.M., Leckman, J.F., Merikangas, K.R., Gammon, G.D., and Prusoff, B.A. Depression and anxiety disorders in parents and children. Archives of General Psychiatry. 41, 845 ( 1984~). Weissman, M.M., Prusoff, B.A., Gammon, G.D., Merikangas, K.R., Leckman, J.F., and Kidd, K.K. Psychopathology in the children (ages 6-18) of depressed and normal parents. Journal of the American Academy of Child Psychiatry. 23,78 (I 9846).
Appendix. Description of symptom scales derived from the Diagnostic Interview Schedule for Children Symptom area 1. Affective
(13)
2. Cognitive
(6)
3. Vegetative 4. Suicidal
Content Sadness, anhedonia, worthlessness,
self-blame, irritability, crying
Boredom, loss of interest, confusion, indecision due to depression
(7)
Loss of appetite, sleep disturbance,
(9)
Hopelessness,
loss of energy, weight fluctuations
thoughts of death and dying, suicidal thoughts, plans, and
attempts 5. Overanxious
(9)
Unrealistic behavior
worry about the future, overconcern in the
past, excessive
complaints, self-consciousness, 6. Separation
anxiety
(12)
need
for
with competence reassurance,
or
somatic
tension, nervousness
Unrealistic worry about parents, school refusal, distress upon separation, impaired thinking and performance
when separated from parents