Elsevier

Journal of Affective Disorders

Volume 45, Issue 3, 1 September 1997, Pages 127-134
Journal of Affective Disorders

Research report
Relationship of Beck Depression Inventory factors to depression among adolescents

https://doi.org/10.1016/S0165-0327(97)00045-1Get rights and content

Abstract

Beck Depression Inventory (BDI) scores of 328 adolescents referred to a depression clinic were factor analyzed to test the discriminant validity of each factor. Three of the four factors (Negative Self Attitude, Performance Difficulty, and Somatic Symptoms) discriminated depressed adolescents from those with a behavior disorder or no diagnosis; the Negative Self Attitude and Performance Difficulty factors also discriminated depressed from anxious adolescents. The fourth factor, Physical Worry, failed to discriminate diagnostic groups. Diagnostic efficiency statistics are reported for both the BDI and for items comprising the 13-item BDI Short Form. Results indicate the BDI is a valid screening tool for adolescent depression in a clinical setting, regardless of the presence of comorbid conditions.

Introduction

The Beck Depression Inventory (BDI; Beck et al., 1979) has been increasingly used to assess depressive severity among adolescents. Although the BDI is not a diagnostic tool, its psychometric properties are similar to or better than other self-rating scales in adolescent samples (e.g. Roberts et al., 1991).

The BDI assesses cognitive, behavioral, affective, and somatic dimensions of depression. Yet, its factor structure in adolescent samples has varied from two to seven factor solutions (Byrne et al., 1993, Hill et al., 1986, Larsson and Melin, 1990, Shek, 1990, Teri, 1982). Furthermore, no study has analyzed whether resultant factors distinguish depressed from nondepressed adolescents.

The BDI total score has adequate sensitivity and specificity in adolescent samples (Ambrosini et al., 1991, Marton et al., 1991, Strober et al., 1981), but has been criticized for having low positive predictive power in a community sample of adolescents (Roberts et al., 1991).

One potential reason for such a high false positive rate in community samples is that the BDI, or at least some of its factors or items, may assess a more general construct than clinical depression. The frequent comorbidity of depression and anxiety symptoms raises questions as to whether such internalizing problems are best conceptualized as one construct (e.g. negative affectivity; Watson and Clark, 1984). BDI items assessing weight loss and somatic preoccupation, for example, have failed to distinguish generalized anxiety disorder from depression among adults (Steer et al., 1986). Similarly, certain somatic symptoms are related more to overanxious disorder than to depression in preadolescent children (Hodges and Craighead, 1990) on another self-rating scale (Children's Depression Inventory; Kovacs, 1983), further suggesting that somatic symptoms are not specific to depression.

The aims of the present study are (a) to examine the factor structure of the BDI in a large clinical sample of adolescents; and (b) to test the discriminant validity of each factor to differentiate diagnostic groups. Also, the presence of comorbid disorders will be examined for their potential impact on the ability of the BDI to distinguish depressed from nondepressed adolescents. Prior research with clinical samples of adolescents has generally had modest sample sizes that precluded the examination of the BDI's discriminant validity among comorbid groups.

Section snippets

Sample

The sample consisted of 328 patients (136 males, 192 females; 308 outpatients, 20 inpatients; age 11 years, 10 months, to 19 years, 11 months) referred to a child and adolescent depression clinic at one of two university medical centers. Patients were excluded if they were physically ill, mentally retarded, or had symptoms clearly suggestive of anorexia/bulimia nervosa or autism.

The Cleveland and Philadelphia samples were similarly matched on major demographic variables (see Table 1). The only

Demographic characteristics

Gender was not significantly different across categories (chi-square7d.f.=13.53, p=.06), although 16 of 22 adolescents diagnosed with disruptive behavior disorder only were boys. Ethnicity was not equally distributed across categories (chi-square14d.f.=28.25, p=.01); most notably, all 13 adolescents meeting criteria for comorbid depressive and disruptive behavior disorders were European–American. One-way ANOVAs were conducted to test for association between age and SES with diagnostic category;

Discussion

A 4-factor solution to the BDI explained 52.8% of the variance. This factor structure was similar to that noted by Beck and Lester (1973), who summarized early factor analytic studies of adults as consistently containing Negative Attitude, Performance Impairment, and Somatic Disturbance factors. The present factor structure was also highly similar to that of a large community sample of adolescents (Byrne et al., 1993). The two largest factors in the present study, Negative Self Attitude and

Acknowledgements

The authors gratefully acknowledge the statistical consultation of Mitch Berman and comments of Katherine Kitzmann on an earlier draft of this manuscript. This work was funded in part by NIMH grant MH41684. The BDI was used with the permission of Dr. Aaron T. Beck, Philadelphia, PA.

References (26)

  • A.T Beck et al.

    Components of depression in attempted suicides

    Journal of Psychology

    (1973)
  • Beck, A.T., Rush, A.J., Shaw, B.F., and Emery, G. (1979). Cognitive therapy of depression. New York:...
  • E.U Brady et al.

    Comorbidity of anxiety and depression in children and adolescents

    Psychological Bulletin

    (1992)
  • Cited by (113)

    • Facebook addiction associated with internet activity, depression and behavioral factors among university students of Bangladesh: A cross-sectional study

      2020, Children and Youth Services Review
      Citation Excerpt :

      A score of 0–4 was considered as normal, 5–13 as borderline clinical depression, 14–20 as moderate depression, and 21–63 as severe depression. The scoring was similar to a study done in China (Chen et al., 2013) and Bangladesh (Mostafa Alim et al., 2020) and the cutoff score for depression was 14 (Bennett et al., 1997; Birmaher et al., 1996; Chen et al., 2013; Mostafa Alim et al., 2020). The internal consistency of this scale was high (Cronbach's alpha was 0.83).

    View all citing articles on Scopus
    View full text