Symptoms Versus a Diagnosis of Depression: Differences in Psychosocial Functioning

Page created by Juanita Ruiz
 
CONTINUE READING
Journal of Consulting and Clinical Psychology                                                             Copyright 1995 by the American Psychological Association, Inc.
1995, Vol. 63, No. 1,90-100                                                                                                                      0022-006X/95/S3.00

                                    Symptoms Versus a Diagnosis of Depression:
                                      Differences in Psychosocial Functioning
                                 Ian H. Gotlib                                             Peter M. Lewinsohn and John R. Seeley
                           Northwestern University                                                  Oregon Research Institute

                              In studies of clinical depression, individuals who demonstrate elevated levels of symptoms but do
                              not meet interview-based diagnostic criteria are typically labeled as false positive and eliminated
                              from further consideration. However, the implicit assumption that false-positive participants differ
                              in important ways from true-positive (i.e., diagnosed) participants has not been tested systemati-
                              cally. This study compared the functioning of true-positive, false-positive, and true-negative adoles-
                              cents on clinical and psychosocial functioning. Although the false-positive participants manifested
                              higher levels of current and future psychopathology than did the true-negative participants, they did
                              not differ significantly from the true-positive participants on most of the measures of psychosocial
                              dysfunction. "False positive," therefore, is not a benign condition.

   A significant proportion of studies have used self-reported el-                  among adolescents (e.g., Garrison, Schluchter, Schoenbach, &
evated levels of depressive symptoms as a measure of depression                     Kaplan, 1989; Roberts, Lewinsohn, & Seeley, 1991; Schoen-
in both subclinical and clinical samples (cf. Coyne & Gotlib,                       bach, Kaplan, Grimson, & Wagner, 1982).
 1983; Gotlib, 1984; Vredenburg, Flett, & Krames, 1993). In-                           There is a related body of research examining the concor-
deed, individuals characterized by high levels of depressive                        dance of a symptom-based assessment of depression, typically
symptoms have been found in numerous investigations to ex-                          using the CES-D, and diagnosis-based assessments of depres-
hibit difficulties in both cognitive and psychosocial functioning                   sion, generally using such interviewer-based measures of de-
(see Gotlib, 1992, and Vredenburg et al., 1993, for reviews).                       pression as the Schedule for Affective Disorders and Schizo-
The two most frequently used instruments to assess individuals'                     phrenia (SADS; Endicott & Spitzer, 1978) and the Diagnostic
levels of depressive symptoms are the Beck Depression Inven-                        Interview Schedule (DIS; Robins, Helzer, Croughan, & Ratcliff,
tory (BDI; Beck, Ward, Mendelsohn, Mock, & Erbaugh, 1961)                            1981). Much of this research has been conducted with the goal
and the Center for Epidemiologic Studies Depression Scale                           of assessing the ability of symptom inventories to identify indi-
(CES-D; Radloff, 1977). The BDI is a 21 -item self-report mea-                      viduals who meet psychiatric diagnostic criteria for major de-
sure of depth or intensity of depression, with the total BDI score                  pressive disorder. In these studies, an interview-based psychiat-
representing a combination of the number of symptom catego-                         ric diagnosis of depression is typically used as the standard; in-
ries endorsed and the severity of the particular symptoms. This                     dividuals are considered to represent cases of depression only if
instrument was originally developed to assess the severity of de-                   they meet explicit interview-based criteria. To evaluate the abil-
pressive symptoms in psychiatric patients already diagnosed as                      ity of self-report symptom measures to detect cases of depres-
depressed.                                                                          sion, investigators have examined the specificity and sensitivity
   The CES-D is a 20-item questionnaire developed by research-                      of the measures. Sensitivity refers to the capability of an instru-
ers at the Center for Epidemiologic Studies at the National In-                     ment to accurately identify cases (i.e., true positives) as deter-
stitute of Mental Health to measure depressive symptoms                             mined by an independent and acceptably valid criterion
among adults in community surveys. Items were selected for                          (typically a diagnosis derived from a reliable structured psychi-
inclusion from previously validated depression scales to present                    atric interview). Specificity refers to the capability of the instru-
the major components of depressive symptomatology (Radloff,                         ment to accurately identify noncases (i.e., true negatives) by the
 1977). The CES-D has been used in numerous community sur-                          same criterion. Investigators have found the sensitivity of the
veys (e.g., Frerichs, Aneshensel, & Clark, 1981; Radloff, 1977;                     CES-D to range from 70% (Radloff, 1977) to 99% (Weissman
Roberts, 1980; Sayetta& Johnson, 1980). The CES-D has also                          et al., 1977) and its specificity to range from 56% (Weissman,
been used successfully to assess level of depressive symptoms                       Sholomskas, Pottenger, Prusoff, & Locke, 1977) to 94% (Boyd,
                                                                                    Weissman, Thompson, & Myers, 1982; see Roberts et al., 1991,
                                                                                    for similar figures in a large sample of adolescents).
  Ian H. Gotlib, Department of Psychology, Northwestern University;                    From these data, it appears that the CES-D is reasonably suc-
Peter M. Lewinsohn and John R. Seeley, Department of Psychology,                    cessful in detecting diagnosable depression. Attesting to this
Oregon Research Institute.                                                          success, a number of investigators have advocated the use of
  This research was partially supported by National Institute of Mental
Health Grant MII40501.
                                                                                    symptom-based self-report measures of depressive symptom-
  Correspondence concerning the article should be addressed to Ian                  atology, such as the CES-D, as a first-stage screening measure
H. Gotlib, Department of Psychology, 102 Swift Hall, Northwestern                   for diagnosable depression in community samples (e.g., Lewin-
University, Evanston, Illinois 60208-2710.                                          sohn & Teri, 1982; Shrout & Fleiss, 1981). By this procedure,
                                                                               90
SYMPTOMS VERSUS A DIAGNOSIS OF DEPRESSION                                                           91

only individuals who obtain high scores on the CES-D                     fact, psychosocial or clinical differences between individuals
(typically 16 and above; Weissman et al., 1977) are adminis-             who meet explicit psychiatric diagnostic criteria for depression
tered a subsequent structured psychiatric interview to deter-            and those who are labeled as false positives. In the absence of
mine diagnostic status.                                                  significant differences, the distinction between these two groups
    It is clear from this work that the critical referent is a psychi-   of individuals should become less important, and serious clini-
atric diagnosis of depression. Indeed, individuals who obtain            cal attention should be paid to the large group of individuals
high scores on self-report measures of depression (meeting cri-          with "only" elevated depressive symptoms.
teria for the first stage of the screening procedure) but who do            There were two related objectives for the present study. The
not meet psychiatric diagnostic criteria are labeled as false pos-       primary purpose was to identify psychosocial and clinical char-
itives and are either discarded as noise in these investigations or      acteristics that would differentiate those persons who both
examined to determine why they do not meet diagnostic cri-               scored high on a self-report measure of depressive symptom-
teria (e.g., Boyd et al., 1982; Breslau, 1985). An explicit objec-       atology and met the Diagnostic and Statistical Manual of Men-
tive of most of these studies, in fact, is to reduce the number of       tal Disorders (3rd ed., rev.; DSM-III-R American Psychiatric
false positives because they are regarded as less clinically im-         Association, 1987) diagnostic criteria for major depressive dis-
portant than the true-positive participants. Clearly, an implicit        order (MOD; true positives) from those persons who scored
assumption of this work is that individuals who both obtain              high on the symptom measure but did not meet diagnostic cri-
high scores on self-report measures of depressive symptoms and           teria for MDD (false positives). Second, we contrasted the
meet interviewer-rated psychiatric diagnostic criteria (true             group of false positives with a group of true-negative partici-
positives) differ in important ways from individuals who obtain          pants (i.e., both low CES-D and no diagnosis of MDD) with
similarly high scores on self-report measures of depressive              respect to a broad range of psychosocial characteristics to de-
symptoms but do not meet interviewer-rated criteria for a diag-          termine how false-positive participants differ from asymptom-
nosis of depression (false positives).                                   atic participants.
    It is important to note that the number of false-positive par-          We had four hypotheses, or expectations: (a) given that the
ticipants in these investigations is typically very large. Given         true-positive participants had to be above the threshold
that a higher proportion of the population is characterized by           (number of symptoms) to meet diagnostic criteria for MDD,
elevated depressive symptomatology than meets formal psychi-             we expected that they would score higher than the false-positive
atric diagnostic criteria for depression, particularly in samples        participants on the CES-D and on interview-based symptom
of adolescents, it is apparent that only a subset of those individ-      ratings of depression; (b) because the CES-D appears to be a
 uals who obtain scores above the cutoff on the CES-D will meet          relatively nonspecific measure of negative affect (e.g., Breslau,
diagnostic criteria for depressive disorder on clinical interview.        1985; Fendrich, Weissman, & Warner, 1990), we expected a
 For example, Boyd et al. (1982) found that only one third of a          high proportion of both the true-positive and the false-positive
 large community sample who obtained high scores on the CES-             participants to also meet criteria for psychiatric diagnoses other
 D received a psychiatric diagnosis of depression. Similarly,            than depression; (c) we expected that a higher proportion of the
 Breslau (1985) found that only 18% of high scorers on the CES-          false-positive than true-negative participants (none of whom
 D met clinician-rated diagnostic criteria for depressive disorder.      met criteria for a diagnosis of MDD) would meet diagnostic
 Finally, Roberts et al. (1991) found that only 10.3% of adoles-         criteria for depression and other psychiatric disorders at a 12-
cents who obtained high scores on the CES-D were diagnosed               month follow-up assessment; and (d) given that the threshold
as depressed on interview. Although all of these rates are supe-         for a DSM-III-R diagnosis of MDD is somewhat arbitrary, we
rior to the population base rate for clinical depression of be-          expected both the true-positive and false-positive participants
tween 2% and 4%, it is clear that the use of this type of screening      to demonstrate more problematic functioning than the true-
 measure yields many false positives. These individuals are typi-        negative participants on the psychosocial variables assessed in
cally eliminated from further consideration in clinical investi-         this investigation.
gations; indeed, given that they do not meet full diagnostic cri-
teria, in clinical practice these individuals would likely not be
considered as cases of depression.                                                                     Method
    Although false positives and true positives differ, by necessity,
with respect to diagnostic status, it is not clear that these two
                                                                         Participants and Procedure
groups of individuals actually differ significantly with respect            Participants were adolescents who were randomly selected in three
to psychosocial or clinical characteristics. Indeed, addressing a        cohorts from nine senior high schools representative of urban and rural
conceptually similar issue, Vredenburg et al. (1993) examined            districts in western Oregon. Sampling was proportional to size of the
the comparability of findings of research conducted with college         school, grade within school, and gender within grade. A total of 1,709
students identified by high scores on self-report measures of de-        adolescents completed the initial ( T l ) assessments (interview and
pression and investigations conducted with diagnosed clinically          questionnaires) between 1987 and 1989, with an overall participation
depressed patients. On the basis of their review, Vredenburg et          rate of 61% (additional details provided in Lewinsohn, Hops, Roberts,
                                                                         Seeley, & Andrews, 1993). The mean age of the Tl sample was 16.6
al. concluded that the results of studies examining the psycho-
                                                                         years (SD = 1.2), and slightly over half of the sample (52.9%) was fe-
logical functioning of self-reported "depressed" university stu-         male. The representativeness of the Tl sample was assessed using sev-
dents are generally similar to findings obtained with samples of         eral approaches; differences between the sample and the larger popula-
interviewer-rated clinically depressed patients. Given this con-         tion and between participants and the percentage who declined to par-
clusion, it becomes important to examine whether there are, in           ticipate were very small. No differences were found between the sample
92                                               I. GOTLIB, P. LEWINSOHN, AND J. SEELEY

 and 1980 census data on gender, ethnic status, or parental education,       parentheses): depression, consisting of major depressive disorder
 although our sample had a slightly higher proportion of two-parent fam-     (current = 2.6%, lifetime = 18.4%); anxiety, consisting of panic disor-
 ilies. Although decliners had a lower mean socioeconomic status (SES)       der, agoraphobia, social phobia, simple phobia, obsessive-compulsive
 level than did participants, both represented the middle class. Partici-    disorder, separation anxiety, and overanxious disorder (3.2%, 8.8%);
pants and decliners did not differ on gender of head of household, family    disruptive behavior, consisting of attention-deficit hyperactivity disor-
size, and number of parents in the household. Adolescents were paid for      der, conduct disorder, and oppositional disorder (1.8%, 7.3%); and sub-
their participation, and written informed consent was obtained from          stance use, consisting of substance abuse disorders and substance de-
both the participants and their legal guardians.                             pendence disorders (2.3%, 8.3%). A miscellaneous category, other, con-
    At the second assessment (T2), 1,507 participants (88.2%) returned       sisted of all other assessed disorders (1.2%, 7.8%).
for a re-administration of the interview and questionnaire (mean Tl-            Diagnostic interviewers were carefully selected, trained, and super-
T2 interval = 13.8 months, SD = 2.3). Biases that may have emerged           vised and all interviews were audio- or videotaped. For reliability
because of attrition in the T1-T2 panel sample were examined by com-         purposes, a second interviewer reviewed the recordings of approxi-
paring the adolescents who did not participate at T2 (n = 202) with          mately 12% of the interviews and made diagnoses. Interrater reliability
the 1,507 participants on demographic characteristics and measures of        was evaluated by the kappa statistic (Cohen, 1960). With three excep-
psychopathology. Small but statistically significant differences were        tions (diagnoses for lifetime eating disorders and current and lifetime
present. Attrition was associated with lower parental SES, F( 1, 1431)       anxiety disorders, «s = .66, .60, and .53, respectively), all Tl kappas
 = 11.6, p
SYMPTOMS VERSUS A DIAGNOSIS OF DEPRESSION                                                                  93

of the measures had been abbreviated (Andrews, Lewinsohn, Hops, &                 consciousness (9 items; Self-Consciousness Scale; Fenigstein, Scheier,
Roberts, 1993). Because a large number of interrelated variables had              & Buss, 1975).
been administered at Tl and T2, measures were reduced to a smaller                   Self-esteem. This construct assessed satisfaction with specific body
number of composite variables. Factorial structures reported by the               parts (3 items; Body Parts Satisfaction Scale; Berscheid, Walster, &
original authors of a measure that could be replicated using confirma-            Bohmstedt, 1973), general satisfaction with physical appearance (3
tory factor analyses were retained. The remaining variables were ra-              items; Physical Appearance Evaluation Subscale; Winstead & Cash,
tionally categorized into general clusters, which were submitted to prin-         1984), and self-esteem (3 items; Self-Esteem Scale; Rosenberg, 1965).
cipal components factor analysis with varimax rotation. Measures in                  Self-rated social competence. This construct assessed self-perceived
each factor with factor loadings >.40 were standardized and summed                social competence (5 items; Social Subscale of the Perceived Compe-
using unit weighting to create a composite score. If composite scores             tence Scale for Children; Harter, 1982; and 7 items from Lewinsohn,
were found to be strongly correlated (i.e., r > .50) and conceptually             Mischel, Chaplin, & Barton, 1980).
similar, the factors were combined into a single construct. Using these              Emotional reliance. This construct assessed the extent to which in-
procedures, the psychosocial measures were categorized into the 20                dividuals desire more support and approval from others, are anxious
constructs described later. Coefficient alphas for these constructs ranged        about being alone or abandoned, and are interpersonally sensitive (10
from .51 to .94, and all were significantly correlated with CES-D scores          items; Emotional Reliance Scale; Hirschfeld, Klerman, Chodoff, Kor-
(see Andrews et al., 1993, and Lewinsohn et al., 1994, for more detailed          chin, & Barrett, 1976).
information about these constructs). All variables were scored so that               Future goals: academic. This construct assessed estimated future
higher values indicate more problematic functioning.                              education (1 item), self-reported grade average last term, perceived ad-
   Stress: daily hassles. This construct assessed the frequency of oc-            equacy of scholastic performance (1 item), perceived ability to com-
currence of unpleasant social and nonsocial events in the past month              plete college (1 item), and the importance of future academic achieve-
(20 items; Unpleasant Events Schedule; Lewinsohn, Mermelstein, Al-                ments (5 items; adapted from the Importance Placed on Life Goals
exander, & MacPhillamy, 1985).                                                    scale; Bachman, Johnston, & O'Malley, 1985).
   Stress: major life events. This construct assessed the occurrence of              Future goals: family. This construct assessed the importance of fu-
 14 life negative life events to self or significant others (i.e., parent, sib-   ture goals related to marriage and family; for example, "Finding the
ling, other relative, or close friend) during the past year (selected from        right person to marry," "Having children" (5 items; adapted from
Schedule of Recent Experience; Holmes & Rahe, 1967; and Life Events               Bachman et al., 1985).
Schedule; Sandier, & Block, 1979). Because they represented symp-                    Future goals: occupational. This construct assessed the importance
toms of psychopathology, 3 of the 14 items ("got in a lot of arguments            of future income level and steady employment; for example, "Having
or fights," "had problems with drugs or alcohol," "tried to commit                lots of money," "Being able to find steady work" (3 items; adapted from
suicide") were not included as stressful events to self.                          Bachman et al., 1985).
   Other psychopathology: internalizing behavior problems. This con-                 Coping skills. This construct assessed the ways individuals cope
struct assessed tendency to worry (5 items; e.g., Maudsley Obsessional            with stressful situations. It consisted of 17 items originally selected from
Compulsive Inventory; Hodgson & Rachman, 1977); frequently recur-                 the Self-Control Scale (Rosenbaum, 1980), the Antidepressive Activity
rent hypomaniclike behavioral fluctuations (12 items; General Behav-              Questionnaire (Rippere, 1977), modified by Parker and Brown (1979),
ior Inventory; Depue & Klein, 1988), state anxiety (10 items; State-              and the Ways of Coping Questionnaire (Folkman & Lazarus, 1980).
Trait Anxiety Inventory; Spielberger, Gorsuch, & Lushene, 1970),                     Social support: family. This construct assessed enjoyable and aver-
quantity and nature of sleep (8 items), and hypochondriasis (8 items;             sive interactions with family members based on items from the Ap-
Pilowsky, 1967).                                                                  praisal of Parents subscale of the Conflict Behavior Questionnaire (11
   Other psychopathology: externalizing behavior problems. This con-              items; Prinz, Foster, Kent, & O'Leary, 1979), the Parent Attitude Re-
struct assessed externalizing problems as per the number of K-SADS                search Instrument (6 items; Schaefer, 1965), the Cohesion subscale of
symptoms at Tl for conduct disorder, oppositional disorder, and atten-            the Family Environment Scale (3 items; Moos, 1974), the Arizona So-
tion-deficit hyperactivity disorder; an unpublished scale assessing con-          cial Support Interview Schedule (Barrera, 1986), and the Social Com-
duct problems during the past week (six items); and a current diagnosis           petence Scale of the Youth Self-Report (2 items; Achenbach & Edel-
of disruptive behavior disorder, substance use disorder, or eating                brock, 1987).
disorder.                                                                            Social support: friends. This construct assessed the number of
   Suicidal ideation. This construct assessed current suicidal ideation           friends, frequency of interaction, and relationship quality based on
as per 3 of the K-SADS items along with a 4-item screener that assessed           items from the Social Competence Scale (2 items; Harter, 1982), the
suicidal ideation for the past week ("I thought about killing myself," "I         Social Competence Scales of the Youth Self-Report (3 items; Achen-
had thoughts of death," "I felt that my family and friends would be               bach & Edelbrock, 1987), the UCLA Loneliness Scale (8 items; Rus-
better off if I was dead," "I felt that I would kill myself if I knew a way").    sell, Peplau, & Cutrona, 1980), and the number of friends providing
   Depressotypic cognitions: pessimism. This construct assessed atti-             social support (Barrera, 1986).
tudes regarding self-reinforcement (10 items; Frequency of Self-Rein-                Interpersonal: conflict with parents. This construct assessed the
forcement Attitude Questionnaire; Heiby, 1982), likelihood of the oc-             number of parent-child conflictual issues during the past 2 weeks and
currence of future positive events (5 items; Subjective Probability Ques-         the average intensity of discussions regarding these issues (45 items;
tionnaire; Mufioz & Lewinsohn, 1976), endorsement of dysfunctional                Issues Checklist scale; Robin & Weiss, 1980).
attitudes (9 items; Dysfunctional Attitude Scale; Weissman & Beck,                   Interpersonal: attractiveness. This construct assessed physical at-
 1978), and perceived control over one's life (3 items; Pearlin &                 tractiveness and attractiveness as a potential friend and as a co-worker
Schooler, 1978).                                                                  as per interviewers' evaluation (17 items; Interpersonal Attraction Mea-
   Depressotypic cognitions: attributions. This construct assessed at-            sure; McCroskey & McCain, 1974).
tributional style along the internal-external, stable-unstable, and                  Physical health and illness. This construct assessed the number of
global-specific dimensions in which a negative events scale and a posi-           days spent in bed as a result of illness in the past year (1 item), the
tive events scale were derived (48 items; Kastan Attributional Style              number of visits to a physician in the past year (1 item), and the occur-
Questionnaire for Children; Kaslow, Tanenbaum, & Seligman, 1978).                 rence of 88 physical symptoms (e.g., broken bones, ulcers, double
   Self-consciousness. This construct assessed private and public self-           vision) during the past 12 months.
94                                           I. GOTLIB, P. LEWINSOHN, AND J. SEELEY

                             Results                                   were relatively low on anhedonia (12.7%), psychomotor agita-
                                                                       tion (16.3%), worthlessness (18.0%), and suicidal ideation
Depression                                                             (4.9%).
                                                                          Finally, we conducted a forward stepwise logistic regression
   Scores on the measures of psychopathology and the psycho-           analysis on the symptoms of depression in an attempt to differ-
 social variables for the true-positive, false-positive, and true-     entiate true-positive from false-positive participants. The re-
 negative participants are presented in Table 1. Our first hypoth-     sults of this analysis indicated that the true-positive and false-
esis predicted that the true-positive participants would obtain        positive participants were differentiated significantly by five
 higher scores than the false-positive participants on the CES-D       symptoms: anhedonia, weight change, sleep difficulties, indeci-
and on interview-based symptom ratings. Recall that we com-            siveness, and suicidal ideation, x 2 (5, N = 316) = 114.69, p <
puted three different scores for the CES-D: number of symp-            .001. With these five symptoms in the equation, 94.62% of the
toms endorsed, mean duration for the endorsed symptoms, and            true-positive and false-positive participants were classified
the more conventional total scores (symptoms times duration).          correctly.
To examine differences among the three groups, separate anal-
yses of variance (ANOVAs) were conducted on these three CES-           Other Psychopathology
D scores. The analysis for the total score yielded a significant
effect for group, F(2, 1695) = 1,410.49, p < .001. Subsequent             Our second hypothesis predicted that a significant proportion
Scheffe post hoc tests indicated that all three groups of partici-     of both the true-positive and the false-positive participants
pants differed significantly from each other (all ps < .05). It        would also meet criteria for psychiatric diagnoses other than
should be noted, however, that although the true-positive par-         depression. To examine this hypothesis, we first conducted sep-
ticipants obtained significantly higher scores on the CES-D than       arate chi-square analyses comparing the proportion of the true-
did the false-positive participants, this difference was less than     positive and false-positive participants who received any nonde-
4 points.                                                              pression psychiatric diagnoses with the relevant proportion of
   In order to examine responses to the CES-D more closely, we         true-negative participants. Both analyses yielded significant
also compared the three groups of participants with respect to         chi-squares: true positive-true negative, x 2 ( U N = 1,415) =
the number of symptoms they endorsed on this measure and the           94.21; false positive-true negative, x 2 ( 1, N = 1,665) = 27.07,
mean reported duration of the endorsed symptoms. The true-             both ps < .001. Thus, compared with the true-negative partici-
positive participants did not differ significantly from the false-     pants, a significantly higher proportion of false-positive and
positive participants with respect to either the number of symp-       true-positive participants received nondepression psychiatric
toms endorsed on the CES-D or the duration of their reported           diagnoses. To determine which specific diagnoses contributed
symptoms; as expected, both these groups obtained higher               to these significant results, we conducted chi-square analyses
scores on these measures than did the true-negative partici-           on each of the nondepression diagnoses for which there were a
pants, both Fs(2, 1695) > 325.00, bothps < .001.                       sufficient number of cases to analyze the data separately
   Our first hypothesis also predicted that the true-positive par-     (namely, substance abuse, anxiety, and disruptive behavior
ticipants would score higher than the false-positive participants      disorder). The true-positive participants were found to differ
on interview-based ratings of the symptoms of depression used          from the true-negative participants on all three diagnoses: sub-
by the DSM-IH-R as criteria for a diagnosis of major depres-           stance abuse, x 2 O, W = 1,414) = 8.81, p< .005; anxiety disor-
sion. All participants received a dichotomous rating of present-       der, x 2 ( 1, N = 1,414) = 54.71, p < .001; disruptive behavior
absent for each of the nine depression symptoms on the SADS.           disorder, x 2 ( l , N = 1,414)= 12.54,p< .001. The false-positive
We conducted two types of analyses on these data. First, we con-       participants differed from the true-negative participants only
ducted an ANOVA on the total number of symptoms exhibited              with respect to anxiety disorder, x 2 ( 1, N = 1,665) = 4.26, p <
by participants in the three groups. The results of this analysis      .05; these two groups of participants did not differ significantly
yielded a significant effect for group, F( 2, 1695) = 528.14,p<        with respect to diagnoses of either substance abuse or disruptive
.001. Scheffe post hoc tests indicated that the three groups of        behavior disorder, both x 2 (1, N = 1,666) < 3.11, ps > .05.
participants all differed significantly from each other (all ps <         Finally, we conducted a chi-square analysis directly compar-
.05). To examine symptom differences between the true- and             ing the proportions of false-positive and true-positive partici-
false-positive participants more closely, we conducted chi-            pants who received any nondepression psychiatric diagnosis.
square analyses on each of the symptoms, comparing the pro-            This analysis yielded a significant result, x 2 O> N = 316) =
portion of true-positive and false-positive participants who were      23.10, p < .001, reflecting the higher proportion of true than
exhibiting the symptom. Each of these analyses was significant,        false-positive participants with nondepression diagnoses.
all x 2 s( 1, N = 316) > 24.92, all ps < .001. For all symptoms,       Again, to examine which specific diagnoses were contributing
a greater proportion of the true-positive than the false-positive      to this result, chi-square analyses were conducted comparing
participants exhibited the symptom. We also compared the pro-          the proportions of true- and false-positive participants on diag-
portion of false-positive and true-negative participants who           noses of substance abuse, anxiety, and disruptive behavior dis-
were exhibiting each symptom. As expected, each of these anal-         order. These analyses yielded a significant effect only for anxiety,
yses was also significant, all x 2s( 1, N - 1,665) > 25.45, all ps <   X 2 ( 1, AT = 316) = 18.39, p < .001, reflecting the finding that a
.001, with a greater proportion of the false-positive than the         higher proportion of true-positive than false-positive partici-
true-negative participants exhibiting each symptom. It is im-          pants met criteria for a diagnosis of anxiety, the true- and false-
portant to note, however, that the false-positive participants         positive participants did not differ significantly with respect to
SYMPTOMS VERSUS A DIAGNOSIS OF DEPRESSION                                                       95

diagnoses of either substance abuse or disruptive behavior dis-           quent univariate ANOVAs yielded significant group effects for
order, both x 2 s( 1, N = 316) < 3.46, ps > .05. Thus, whereas a          all variables except Future Goals: Family, all Fs(2, 1695) >
diagnosis of anxiety disorder differentiated all three groups of          7.47, allps < .001. Subsequent Scheffe post hoc tests indicated
participants, diagnoses of substance abuse and disruptive be-             that with three exceptions, the true-negative subjects differed
havior disorder differentiated only the true-positive and true-           significantly (p < .05) from both the true-positive and false-
negative subjects.                                                        positive subjects. On Future Goals: Occupational, Interper-
                                                                          sonal: Conflict With Parents, and Interpersonal: Attractiveness,
Prediction of Subsequent Psychopathology                                  the true negative subjects differed only from the false-positive
                                                                          subjects.
   Our third hypothesis predicted that a higher proportion of                To examine more explicitly differences between the true-pos-
the false-positive than true-negative participants (none of               itive and false-positive subjects with respect to psychosocial
whom met criteria for a diagnosis of depression at the Time 1             functioning, a multivariate analysis of covariance was con-
assessment) would develop a psychiatric disorder during the 12-           ducted on the set of psychosocial variables using data only from
month period following the Time 1 assessment. As predicted, a             the true- and false-positive subjects and using CES-D scores as
chi-square analysis conducted on the proportions of true-nega-            a covariate. The results of this analysis yielded significant effect
tive and false-positive participants who met diagnostic criteria          for group, F(20, 294) = 3.67, p < .001. Subsequent univariate
for any psychiatric disorder during the 12-month follow-up pe-            analyses of covariance were conducted to examine which vari-
riod was significant, X 2 ( 1, N= 1,369) = 24.76, p< .001; 23.7%          ables were contributing significantly to this effect. The true-pos-
of the false-positive participants met criteria for a psychiatric         itive and false-positive subjects were found to differ significantly
diagnosis during the 12-month follow-up, compared with only               on both internalizing and externalizing psychopathology, Sui-
 11.1% of the true-negative participants. Subsequent chi-squares          cidal Ideation, and Physical Health and Illness, all Fs( 1,313) >
conducted on specific diagnostic categories indicated that a               3.86, all ps < .05. On all of these measures, the true-positive
higher proportion of false-positive than true-negative partici-           subjects obtained significantly more dysfunctional scores than
pants met diagnostic criteria during the 12-month follow-up for           did the false-positive subjects, even after controlling for CES-D
major depression, x 2 ( 1, N = 1,469) = 28.02, p < .001, and              scores.
substance abuse, x 2 ( 1, N = 1,442) = 4.81, p < .05; the two                Finally, we conducted a forward stepwise logistic regression
groups did not differ with respect to diagnoses of anxiety or             analysis on this subset of four psychosocial variables in an at-
disruptive behavior disorder (both x2s < 3.56, both ps > .05).            tempt to differentiate true-positive from false-positive subjects.
It should be noted that the base rates for these disorders overall        CES-D scores were entered in the first step to control for differ-
were very low. Finally, to examine whether the difference be-             ences in level of depressive symptoms. The four psychosocial
tween the false-positive and true-negative participants in the de-        variables were then entered in the second step. The results of
velopment of a psychiatric disorder was due to the higher pro-            this analysis indicated that true-positive and false-positive sub-
portion of participants with a previous history of depression in          jects were differentiated significantly only by the presence of
the false-positive than the true-negative group (29.7% vs.                suicidal behavior, \2(\, N = 316) = 30.56, p < .001, with an
 13.5%), a chi-square analysis examining the proportions of par-
                                                                          odds ratio of 1.63. In fact, with CES-D scores and suicidal be-
ticipants in these two groups who met diagnostic criteria for a
                                                                          havior in the equation, 90.51% of the participants were classi-
psychiatric disorder during the 12-month follow-up period was
                                                                          fied correctly.
conducted only for participants without a history of depression
and who received no psychiatric diagnosis at Tl. This analysis
y ielded a significant effect, x 2 ( l , A r = 1,155)= 14.49, p < .001;
within this subset of participants, 19.4% of the false-positive                                      Discussion
participants met criteria for a psychiatric diagnosis during the
 12-month follow-up, compared with only 9.1% of the true-neg-                The major purpose of this study was to identify clinical and
ative participants. It appears, therefore, that the higher rate of        psychosocial characteristics that would differentiate true-posi-
subsequent psychopathology among the false-positive partici-              tive depressed adolescents from false-positive adolescents. In de-
pants remains even after controlling for past history of                  fining these two groups, both the true-positive and the false-
depression.                                                               positive participants obtained elevated scores on the CES-D,
                                                                          but only the true-positive participants also met diagnostic cri-
Psychosocial Functioning                                                  teria for major depressive disorder. Given the high prevalence of
                                                                          adolescents who obtain high scores on the CES-D, it is critical
   Our final hypothesis predicted that the false-positive partici-        to examine differences between those who meet diagnostic cri-
pants would manifest many of the same psychosocial problems               teria for depression and those who do not. We predicted that
as the true-positive participants. We also predicted that the             although the true-positive participants would obtain higher
false-positive participants would demonstrate more problem-               scores than would the false-positive participants on the CES-D
atic psychosocial functioning than would the true-negative par-           and on interview-based symptom ratings of depression, the two
ticipants. To examine these hypotheses, we conducted a multi-             groups would not differ from each other with respect to reports
variate analysis of variance on the three groups of participants'         of difficulties in psychosocial functioning, although both these
scores on the psychosocial measures. This analysis yielded a sig-         groups were predicted to report greater dysfunction than the
nificant effect for group, F(40, 3354) = 27.67, p< .001. Subse-           true-negative participants.
96                                          I. GOTLIB, P. LEWINSOHN, AND J. SEELEY

Table
CES-D Group Differences on Measures of Psychopathology and Psychosocial Variables

                                                                                                                  Test statistic
                                                           True negative   False positive   True positive
                         Measure                            (N= 1,382)      (TV =283)         (TV =33)      F(2, 1695)           *2(2)

CES-D                                                                                                       1,410.49***
  M                                                            13.0.          34. l b          38.0C
  SD                                                           6.5             6.4              6.5
CES-D symptom (no.)                                                                                          544.57***
  M                                                            9.1.           17.2b            18.1b
  SD                                                           4.3             2.0              1.3
CES-D symptom duration                                                                                       325.72***
  M                                                             1.4.           2.0b             2.1b
  SD                                                           0.4             0.3              0.3
DSM-IH-R MDD symptom (no.)                                                                                   528.14***
  M                                                            0.4a            1.8b             6.8C
  SD                                                            1.0            2.1              1.2
DSM-IH-R MDD symptom (%)
  Depressed mood                                                5.6a          31.8b           lOO.Oc                         404.66**
  Anhedonia                                                     1.9.          12.7b            75.8C                         402.18**
  Weight change                                                 6.2a          20.8b            81.8C                         247.32*
  Sleep difficulties                                            8.6a          27.9b            84.8C                         225.14*
  Psychomotor agitation                                         2.2a          16.3b            63.6C                         296.14*
  Fatigue                                                      6.2a           20.8b            60.6C                         158.27*
  Worthlessness                                                 3.2a          18.0b            72.7C                         302.38*
  Indecisiveness                                               6.4a           25.8b            81.8C                         261.12*
  Suicidal ideation                                            0.9a            4.9b            51.5C                         342.97*
Other disorder prevalence
  Any nonaffective disorder                                    5.7a           14.5b            48.5C                           99.36***
  Substance use disorders                                      1.8.            2.8.,b           9.1b                            9.05*
  Anxiety disorders                                            2.4a            4.6b            24.2C                           52.20***
  Disruptive behavior disorders                                1.4.            2.8a,b           9.1b                           13.26**
Disorder 12-month incidence (%)
  Any disorder                                                11.1            23.7               —                            24.76"'***
  MDD                                                          6.0            15.9               —                            28.02"-***
  Substance use disorders                                      3.3a            6.3b            11. 5b                          8.65C'*
  Anxiety disorders                                            0.3a            1.3.            lO.Ob                          31.77"'***
  Disruptive behavior disorders                                0.6             0.4              0.0                            0.24e
Psychosocial variables
  Stress: Daily Hassles                                                                                      123.42***
     M                                                        -0.17.           0.73b            0.87b
     SD                                                        0.91            1.05             1.06
  Stress: Major Life Events                                                                                   26.70***
     M                                                        -0.09.           0.35b            0.47b
     SD                                                        0.94            1.18             1.05
  Other psychopathology: Internalizing behavior problems                                                     276.32***
     M                                                        -0.24.           0.96b            1.51C
     SD                                                        0.85            0.93             0.96
  Other psychopathology: Externalizing behavior problems                                                      47.93***
     M                                                        -0.11.           0.38b            0.96C
     SD                                                        0.89            1.16             1.51
  Suicidal ideation                                                                                          365.28***
     M                                                        -0.21.           0.62b            3.10C
     SD                                                        0.48            1.43             2.76
  Depressotypic conditions: Pessimism                                                                        186.90***
     M                                                        -0.20.           0.84b            1.20b
     SD                                                        0.89            0.97             1.09
  Depressotypic conditions: Attributions                                                                      91.79***
     M                                                        -0.15.           0.6 l b          0.93b
     SD                                                        0.94            1.03             0.95
  Self-Consciousness                                                                                          50.84***
     M                                                        -0.11.           0.49b            0.59b
     SD                                                        0.94            1.09             1.18
  Self-Esteem                                                                                                 70.00***
     M                                                        -0.13.           0.5 l b          1.00C
     SD                                                        0.94            1.04             1.07
  Self-Rated Social Competence                                                                                34.28***
     M                                                        -0.09.           0.37b            0.64b
     SD                                                        0.95            1.11             0.99
                                                                                                                         (table continues)
SYMPTOMS VERSUS A DIAGNOSIS OF DEPRESSION                                                                  97

Table 1 (continued)
                                                                                                                         Test statistic
                                                             True negative       False positive   True positive
                                                                                                                                          2
                           Measure                            (N= 1,382)           (N = 283)        (N=33)         F(2, 1695)         X       (2)

Emotional Reliance                                                                                                  120.31***
    M                                                           -0.17.               0.70b            1.01b
    SD                                                           0.92                0.99             0.99
  Future Goals: Academic                                                                                              15.48***
    M                                                           -0.06a               0.26b            0.40b
    SD                                                           0.98                1.00             1.25
  Future Goals: Family                                                                                                 1.97
    M                                                           -0.01a               0.03             0.32
    SD                                                           0.97                1.08             1.24
  Future Goals: Occupational                                                                                           7.47***
    M                                                             0.04a            -0.20b           -0.16.,b
    SD                                                            0.98               1.07             0.93
  Coping Skills                                                                                                     131.10***
    M                                                           -0.18.               0.7 l b          l.llb
    SD                                                           0.92                0.94             1.04
  Social Support: Family                                                                                             69.89***
    M                                                           -0.13.               0.56b            0.68b
    SD                                                           0.94                1.04             1.04
  Social Support: Friends                                                                                            47.99***
    M                                                           -0.11.               0.42b            0.86C
    SD                                                           0.93                1.11             1.30
  Interpersonal: Conflict With Parents                                                                               54.39***
    M                                                           -0.12.               0.53b            0.25a,b
    SD                                                           0.92                1.19             1.19
  Interpersonal: Attractiveness                                                                                       8.20***
    M                                                           -0.05a               0.1 7b           0.37a.b
    SD                                                            0.99               0.96             1.06
  Physical Health and Illness                                                                                        37.26***
    M                                                           -0.09a               0.3 l b          0.91C
    SD                                                           0.86                1.23             1.41

Note. Percentages or means with different subscripts differ significantly at p < .01. Dashes indicate data were not applicable. MOD = Major
Depressive Disorder; CES-D = Center for Epidemiologic Depression Studies—Depression Scale. DSM-HI-R = Diagnostic and Statistical Manual
of Mental Disorders (3rd ed., rev.).
a
  N = 1,369. b ;V= 1,469. C A ' = 1,467. d ]V= 1,452. ' N = 1,474.
*p
98                                           I. GOTLJB, P. LEWINSOHN, AND J. SEELEY

 symptoms, compared with 1.8 symptoms for the false-positive           lated more strongly to symptoms of depression and anxiety than
 participants. This finding reflects the fact that, to have received   to symptoms associated with other forms of psychopathology.
 a diagnosis of depression, the true-positive participants had to      Thus, although the CES-D is clearly not a measure that is spe-
 have reported at least five symptoms, whereas the false-positive      cific to depression, it is also not equally sensitive to all symp-
 participants must have reported fewer than five symptoms (or          toms of psychopathology.
 more symptoms but without depressed mood or required                     The final hypothesis concerning the examination of psycho-
duration). It is important to note that the false-positive and         pathology in this study involved the onset of psychopathology
true-positive participants differed not only with respect to the       over the year following the initial interview. Our prediction that
 mean number of symptoms reported, but also with respect to            the false-positive participants would be more likely to develop
the prevalence of each of the depressive symptoms. This finding        depression and other mental disorders than would the true-neg-
suggests that the difference between the true-positive and false-      ative participants was strongly supported. Indeed, the false-pos-
positive participants in the number of symptoms reported was           itive participants were at least twice as likely as the true-negative
not confined to a specific group of symptoms. Nevertheless, a          participants to develop a psychiatric disorder over the course of
discriminant function analysis indicated that the false-positive       the study. Clearly, being identified as a false-positive participant
and true-positive participants were differentiated significantly       is not benign. The results of this study indicate not only that
by anhedonia, weight change, sleep difficulties, indecisiveness,       these individuals may manifest a nonaffective psychiatric disor-
and suicidal ideation. Most of these particular symptoms ap-           der, but further, that they are at elevated risk for developing de-
pear to reflect a more melancholic or endogenous form of de-           pression, substance abuse, and anxiety disorder within a year
pression (e.g., Zimmerman & Spitzer, 1989). It appears, there-         after being identified as a "false positive."
fore, that the true-positive participants represent a somewhat
more severely depressed group of individuals than do the false-        Psychosocial Functioning
positive participants, particularly with respect to vegetative
symptoms.                                                                 The final goal of this study was to compare the psychosocial
   Despite these differences between the true-positive and false-      functioning of the true-positive, false-positive, and true-nega-
positive participants, it is important to remain cognizant of the      tive participants. We predicted that the false-positive partici-
fact that the false-positive participants reported significantly       pants would demonstrate greater difficulties in psychosocial
more DSM-III-R depression symptoms than did the true-neg-              functioning than would the true-negative participants. All of
ative participants. Indeed, the rates of these symptoms of de-         the measures of psychosocial functioning administered to par-
pression in the true-negative participants is relatively low. None     ticipants in this study were selected because of their hypothe-
of the symptom rates in this group, for example, exceeded 10%,         sized or demonstrated association with depression. Indeed, we
and half of them were under 6%. In contrast, the rates for the         included 20 different measures of psychosocial functioning.
false-positive participants were above 10% for eight of the nine       With only one exception (Future Goals: Family), the false-pos-
symptoms, and for five symptoms they were above 20%. Thus,             itive participants did demonstrate significantly (i.e., all ps <
although the false-positive participants reported fewer DSM-           .001) more problematic psychosocial functioning than did the
III-R symptoms of depression than did the true-positive par-           true-negative participants.
ticipants, they reported significantly more symptoms than did             This pattern of results is consistent with a large literature
the true-negative participants.                                        demonstrating that "subclinical" depressives (i.e., individuals
   The false-positive participants also differed from the true-        who endorse elevated levels of symptoms on self-report
negative participants with respect to the proportion of partici-       questionnaires) experience difficulties in psychosocial func-
pants in the group who met diagnostic criteria for DSM-III-R           tioning relative to their less "depressed" counterparts (cf. Bar-
disorders other than depression. Indeed, the proportion of false-      nett & Gotlib, 1988; Vredenburg et al. 1993). An important
positive participants who met psychiatric diagnostic criteria          question addressed in the present study, however, concerned po-
was almost three times higher than the rate within the group of        tential differences in psychosocial functioning between the true-
true-negative participants (14.5% vs. 5.7%). Other investiga-          positive and false-positive participants. Once initial differences
tors have reported that the CES-D appears to be sensitive to           between these two groups with respect to CES-D scores were
disorders other than depression and may reflect general distress       controlled, the true-positive and false-positive participants
rather than depression specifically (e.g., Breslau, 1985; Fen-         differed on only four of the 20 variables assessed in this study:
drich et al., 1990). The present findings are generally consistent     externalizing and internalizing behavior, suicidal behavior, and
with this position, but they also suggest qualifications. Specifi-     physical illness. Moreover, a discriminant function analysis in-
cally, in addition to depression, the only single diagnosis for        dicated that, again after initial CES-D differences were con-
which the false-positive participants differed from the true-neg-      trolled, the true-positive and false-positive participants were
ative participants was anxiety disorder (4.6% vs. 2.4%). This          differentiated only by suicidal behavior. Thus, consistent with
elevation on diagnoses of depression and anxiety is consistent         Vredenburg et al.'s contention, most of the psychosocial vari-
with the high rates of comorbidity of these two disorders that         ables did not significantly differentiate between the clinical and
have been reported in previous work (cf. Gotlib & Cane, 1989;          the subclinical depressed participants.
Maser & Cloninger, 1990). Indeed, the diagnosed cases (true               It is clear, therefore, that considering both psychopathology
positives) in this study also demonstrated a high rate of comor-       and psychosocial functioning, individuals who obtain elevated
bidity of depression and anxiety (cf. Rohde, Lewinsohn, &              scores on the CES-D but do not meet diagnostic criteria for
Seeley, 1991). It appears, therefore, that the CES-D may be re-        depression (i.e., the so-called false-positive subjects) demon-
SYMPTOMS VERSUS A DIAGNOSIS OF DEPRESSION                                                            99

strate problematic functioning. They are characterized by ele-               (1961). An inventory for measuring depression. Archives of General
vated levels of current psychopathology, they are at elevated risk           Psychiatry, 4, 561-511.
for experiencing psychopathology in the future, particularly de-           Berscheid, E., Walster, E., & Bohrnstedt, G. (1973, November). The
                                                                             happy American body: A survey report. Psychology Today, 7, 119-
pression and anxiety, and they report marked difficulties in psy-
                                                                             131.
chosocial functioning. Thus, even though false-positive partici-           Boyd, J. H., Weissman, M. M., Thompson, W. D., & Myers, J. K.
pants do not meet DSM-III-R diagnostic criteria for depres-                  (1982). Screening for depression in a community sample: Under-
sion, there is little question that we should not be clinically              standing the discrepancies between depression symptoms and diag-
indifferent about this group of individuals.                                 nostic skills. Archives of General Psychiatry, 39, 1195-1200.
   In closing, we note that there are certain limitations of this          Breslau, N. (1985). Depressive symptoms, major depression, and gen-
study that may qualify the obtained results. Most important,                 eralized anxiety: A comparison of self-reports on CES-D and results
perhaps, is that the present study was restricted to a large sam-            from diagnostic interviews. Psychiatry Research, 15, 219-229.
ple of adolescents. It is not clear, therefore, to what extent these       Cohen, J. A. (1960). A coefficient of agreement for nominal scales.
findings are generalizable to older people or to children, and it            Educational and Psychological Measurement, 20, 37-46.
is imperative that this issue be addressed in future investiga-            Coyne, J. C., & Gotlib, I. H. (1983). The role of cognition in depres-
tions. We also note that we used a rather stringent CES-D cutoff             sion: A critical appraisal. Psychological Bulletin, 94, 472-505.
in this study of 27, which represented the 90th percentile in this         Craig, T. J., & Van Natta, P. A. (1979). Influence of demographic char-
                                                                             acteristics on two measures of depressive symptoms. Archives of Gen-
sample. Two points are relevant here. First, this cutoff score is            eral Psychiatry, 36, 149-154.
higher than those typically used to identify probable cases of             Depue, R. A., & Klein, D. (1988). Identification of unipolar and bipo-
depression (cf. Frerichs et al., 1981; Radloff, 1977) and suggests           lar affective conditions by the General Behavior Inventory. In D.
that adolescents, in general, score higher on the CES-D than do              Dunner, E. E. Gershon, & J. Barrett (Eds.), Relatives at risk for men-
adults. Second, because of this relatively high cutoff score, there          tal disorder (pp. 257-282). New York: Raven Press.
were participants in the true-negative group who obtained CES-             Endicott, J., & Spitzer, R. L. (1978). A diagnostic interview: The
D scores of up to 26. Although as a group, the true-negative                 Schedule for Affective Disorders and Schizophrenia. Archives of Gen-
participants differed consistently from both the true-positive               eral Psychiatry, 35, 837-844.
and false-positive participants, we caution that the prevalence            Fendrich, M., Weissman, M. M., & Warner, V. (1990). Screening for
and total incidence rates of psychopathology rates in the group              depressive disorder in children and adolescents: Validating the Center
of true-negative participants may be slightly higher than one                for Epidemiologic Studies Depression Scale for children. American
would expect from a randomly selected nondepressed control                   Journal of Epidemiology, 131, 538-551.
                                                                           Fenigstein, A., Scheier, M. E, & Buss, A. H. (1975). Public and private
group. Despite these relatively minor limitations, we believe                self-consciousness: Assessment and theory. Journal of Consulting and
that the results of this study are compelling. It is apparent that           Clinical Psychology, 43, 522-527.
investigators could profitably turn their attention to an exami-           Folkman, S., & Lazarus, R. S. (1980). An analysis of coping in a mid-
nation of the psychosocial problems of the false positives, ado-             dle-aged community sample. Journal of Health and Social Behavior,
lescents who have elevated scores on self-report depression mea-             21,219-23,9.
sures but do not meet DSM-III-R criteria for a diagnosis of                Frerichs, R. R., Aneshensel, C., & Clark, V. A. (1981). Prevalence of
depression.                                                                  depression in Los Angeles County. American Journal of Epidemiol-
                                                                             ogy, 113,691-699.
                                                                           Garrison, C. Z., Schluchter, M. D., Schoenbach, V. J., & Kaplan, B. K.
                             References                                      (1989). Epidemiology of depressive symptoms in young adolescents.
Achenbach, T. M., & Edelbrock, C. S. (1987). Manual for the Youth            Journal of the American Academy of Child and Adolescent Psychiatry,
  Self-Report and Profile. Burlington: University of Vermont Depart-         28,343-351.
  ment of Psychiatry.                                                      Gotlib, I. H. (1984). Depression and general psychopathology in uni-
American Psychiatric Association. (1987). Diagnostic and statistical         versity students. Journal of Abnormal Psychology, 93, 19-30.
  manual ofmental disorders (3rd ed., rev.). Washington, DC: Author.       Gotlib, I. H. (1992). Interpersonal and cognitive aspects of depression.
Andrews, J. A., Lewinsohn, P. M., Hops, H., & Roberts, R. E. (1993).         Current Directions in Psychological Science, 1, 149-154.
  Psychometric properties of scales for the measurement of psychoso-       Gotlib, I. H., & Cane, D. B. (1989). Self-report assessment of depres-
  cial variables associated with depression in adolescence. Psychologi-      sion and anxiety. In P. C. Kendall & D. Watson (Eds.), Anxiety and
  cal Reports, 73, 1019-1046.                                                depression (pp. 131 -169). New York: Academic Press.
Angst, J., Merikangas, K., Scheidegger, P., & Wicki, W. (1990). Recur-     Harter, S. (1982). The Perceived Competence Scale for Children. Child
  rent brief depression: A new subtype of affective disorder. Journal of     Development, 53, 87-97.
  Affective Disorders, 19, 87-98.                                          Heiby, E. M. (1982). A self-reinforcement questionnaire. Behaviour
Bachman, J. G., Johnston, J., & O'Malley, P. M. (1985, May). Some            Research and Therapy, 20, 397-401.
  recent trends in the aspirations, concerns, and behaviors of American    Hirschfeld, R. M. A., Klerman, G. L., Chodoff, P., Korchin, S., & Bar-
  young people. Paper presented at the 40th Annual Conference of the         rett, J. (1976). Dependency—self-esteem—clinical depression.
  American Association for Public Opinion Research, McAfee, NJ.              Journal of the American Academy of'Psychoanalysis, 4, 373-388.
Barnett, P. A., & Gotlib, I. H. (1988). Psychosocial functioning and       Hodgson, R. J., & Rachman, S. (1977). Obsessional-compulsive com-
  depression: Distinguishing among antecedents, concomitants, and            plaints. Behaviour Research and Therapy, 15, 389-395.
  consequences. Psychological Bulletin, 104, 97-126.                       Holmes, T. H., & Rahe, R. H. (1967). The social readjustment rating
Barrera, M., Jr. (1986). Distinctions between social support concepts,       scale. Journal of Psychosomatic Research, 11, 213-218.
  measures, and models. American Journal of Community Psychology,          Kaslow, N. J., Tanenbaum, R. L., & Seligman, M. E. P. (1978). Kastan-R:
  14, 413-445.                                                               A Children's Attributional Style Questionnaire, Unpublished manu-
Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J.             script, University of Pennsylvania.
100                                             I. GOTLIB, P. LEWINSOHN, AND J. SEELEY

Lewinsohn, P. M., Hops, H., Roberts, R. E., Seeley, J. R., & Andrews,          National Institute of Mental Health Diagnost-c Interview Schedule:
   J. A. (1993). Adolescent psychopathology: I. Prevalence and inci-           Its history, characteristics and validity. Archives of General Psychia-
   dence of depression and other DSM-III-R disorders in high school            try, 38, 381-389.
   students. Journal of Abnormal Psychology, 102, 133-144.                  Rohde, P., Lewinsohn, P. M., & Seeley, J. R. (1991). Comorbidity of
Lewinsohn, P. M., Mermelstein, R. M., Alexander, C, & MacPhillamy,             unipolar depression: II. Comorbidity with other mental disorders in
   D. (1985). The Unpleasant Events Schedule: A scale for the mea-             adolescents and adults. Journal of Abnormal Psychology, 100, 214-
   surement ofaversive events. Journal of Clinical Psychology, 41, 483-        222.
   498.                                                                     Rosenbaum, M. (1980). A schedule for assessing self-control behaviors:
Lewinsohn, P. M., Mischel, W., Chaplin, W., & Barton, R. (1980). So-           Preliminary findings. Behavior Therapy, 11, 109-121.
   cial competence and depression: The role of illusory self-perceptions.   Rosenberg, M. (1965). Society and the adolescent self-image.
   Journal of Abnormal Psychology, 89, 203-212.                                Princeton, NJ: Princeton University Press.
Lewinsohn, P. M., Roberts, R. E., Seeley, J. R., Rohde, P., Gotlib,         Russell, D., Peplau, L. A., &Cutrona, C. E. (1980). The Revised UCLA
   I. H., & Hops, H. (1994). Adolescent depression: II. Psychosocial risk      Loneliness Scale: Concurrent and discriminant validity evidence.
   factors. Journal of Abnormal Psychology, 103, 302-315.                      Journal of Personality and Social Psychology, 39, 472-480.
Lewinsohn, P. M., & Teri, L. (1982). Selection of depressed and non-        Sandier, I. N., & Block, M. (1979). Life stress and maladaption of chil-
   depressed subjects on the basis of self-report data. Journal of Con-        dren. American Journal of Community Psychology, 7, 425-439.
   sulting and Clinical Psychology. 50, 590-591.                            Sayetta, R. B., & Johnson, D. P. (1980). Basic data on depressive symp-
Maser, J. D., & Cloninger, C. R. (1990). Comorbidity of mood and an-           tomatology, United States, 1974-1975: Data from the National
   xiety disorders. Washington, DC: American Psychiatric Press.                Health Survey (Series II, No. 216, DHEW Publication No. 80-1666).
McCroskey, J. C., & McCain, T. A. (1974). The measurement of inter-            Washington, DC: National Center for Health Statistics.
   personal attraction. Speech Monographs, 41, 261-266.                     Schaefer, E. S. (1965). Children's reports of parental behavior: An in-
Moos, R. H. (1974). Family environment scale and preliminary man-              ventory. Child Development, 36, 413-424.
   ual, Palo Alto, CA: Consulting Psychologists Press.                      Schoenbach, V. J., Kaplan, B. H., Grimson, R. C., & Wagner, E. H.
Munoz, R. F., & Lewinsohn, P. M. (1976). The subjective probability            (1982). Use of a symptom scale to study the prevalence of a depres-
   questionnaire. Unpublished manuscript, University of Oregon.                sive syndrome in young adolescents. American Journal of Epidemiol-
Orvaschel, H.. Puig-Antich, J., Chambers, W. J., Tabrizi, M. A., &             ogy, 116, 791-800.
   Johnson, R. (1982). Retrospective assessment of prepubertal major        Shapiro, R., & Keller, M. (1979). Longitudinal Interval Follow-Up
   depression with the Kiddie-SADS-E. Journal of the American Acad-            Evaluation (LIFE). Unpublished manuscript, Massachusetts Gen-
   emy of Child Psychiatry, 21, 392-397.                                       eral Hospital, Boston.
Parker, G. B., & Brown, L. B. (1979). Repertoires of response to poten-     Shrout, P. E., & Fleiss, J. L. (1981). Reliability and case detection. In J.
   tial precipitants of depression. Australian and New Zealand Journal         Wing, P. Bebbington, & L. N. Robins (Eds.), What is a case? The
   of Psychiatry, 13, 327-333.                                                problem of definition in psychiatric community surveys (pp. 117-
Pearlin, L. I., & Schooler, C. (1978). The structure of coping. Journal        128). London: Grant Mclntyre.
                                                                            Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual
   of Health and Social Behavior, 19,2-2\.
                                                                              for the State- Trail Anxiety Inventory. Palo Alto, CA: Consulting Psy-
Pilowsky, I. (1967). Dimensions of hypochondriasis. British Journal of
                                                                               chologists Press.
   Psychiatry, 113, 89-93.
                                                                            Vredenburg, K., Flett, G. L., & Krames, L. (1993). Analogue versus
Prinz, R. J., Foster, S., Kent, R. N., &O'Leary, K. D. (1979). Multivar-       clinical depression: A critical reappraisal. Psychological Bulletin,
   iate assessment of conflict in distressed and nondistressed mother-         113,327-344.
   adolescent dyads. Journal of Applied Behavior Analysis, 12,691-700.      Weissman, A. N., & Beck, A. T. (1978, November). Development and
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale         validation of the Dysfunctional Attitude Scale. Paper presented at the
   for research in the general population. Applied Psychological Mea-          annual meeting of the Association for the Advancement of Behavior
   surement, 1, 385-401.                                                       Therapy, Chicago.
Rippere, V. (1977). What's the thing to do when you're feeling de-          Weissman, M. M., Sholomskas, D., Pottenger, M., Prusoff, B. A., &
   pressed? A pilot study. Behaviour Research and Therapy, 15, 185-            Locke, B. Z. (1977). Assessing depressive symptoms in five psychiat-
    191.                                                                       ric populations: A validation study. American Journal of Epidemiol-
Roberts, R. E. (1980). Prevalence of psychological distress among Mex-         ogy, 706,203-214.
   ican-Americans. Journal of Health and Social Behavior, 21, 134-145.      Winstead, B. A., & Cash, T. F. (1984, November). Reliability and va-
Roberts, R. E., Lewinsohn, P. M., & Seeley, J. R. (1991). Screening for        lidity of the Body Self-Relationship Questionnaire: A new measure
   adolescent depression: A comparison of depression scales. Journal of        of body image. Paper presented at the meeting of the Southeastern
   the American Academy of Child and Adolescent Psychiatry, 30, 58-            Psychological Association, New Orleans.
   66.                                                                      Zimmerman, M., & Spitzer, R. L. (1989). Melancholia: From DSM-
Robin, A. L., & Weiss, J. G. (1980). Criterion-related validity of behav-      III to DSM-III-R. American Journal of Psychiatry, 146, 20-28.
   ioral and self-report measures of problem-solving communication
   skills in distressed and nondistressed parent-adolescent dyads. Be-                                            Received January 24, 1994
   havioral Assessment, 2, 339-352.                                                                            Revision received June 2, 1994
Robins, L. N., Helzer, J. E., Croughan, J., & Ratcliff, K. S. (1981).                                                  Accepted June 8, 1994 •
You can also read