Elsevier

Psychiatry Research

Volume 200, Issues 2–3, 30 December 2012, Pages 843-848
Psychiatry Research

The Beck Depression Inventory-II in adolescent mental health patients: Cut-off scores for detecting depression and rating severity

https://doi.org/10.1016/j.psychres.2012.05.011Get rights and content

Abstract

The Beck Depression Inventory—Second Edition (BDI-II) was developed as an indicator of the presence and severity of depression in psychiatric patients from age 13. Its cut-off scores were derived from an adult sample and differentiate four categories of severity but contain no screening cut-off score for classifying patients as depressed vs. nondepressed. We aimed to determine this screening cut-off score and to examine the utility of the severity cut-off scores for adolescents in mental health care. 88 adolescent psychiatric patients (13–16 years, 58% female) completed the German BDI-II. A structured diagnostic interview served as the reference standard for computing receiver operating characteristic (ROC) curves and identified 24 adolescents (27%) as depressed. ROC analysis of depressed vs. nondepressed patients yielded an area under the curve (AUC) value of 0.93. The optimal screening cut-off score according to Youden's Index was ≥23, where sensitivity was 0.88 and specificity was 0.92. The severity cut-off scores yielded satisfying sensitivity (≥0.89) and specificity (≥0.72) for mild and moderate but not for severe depression. Our findings indicate that the BDI-II can be recommended for screening for depressive disorders in adolescent mental health patients. However, the currently used severity cut-off scores may be suboptimal for this population.

Introduction

Depressive symptoms, such as low mood or diminished pleasure, affect almost 30% of all adolescents (Roberts et al., 1995). A considerable proportion of them experience clinically significant depressive disorders: Life-time prevalence of Major Depressive or Dysthymic Disorder by the end of adolescence was recently estimated 11.7% in a representative population sample (Merikangas et al., 2010). Beyond current symptoms, depressive disorders often persist in adulthood and are associated with a higher risk of suicidal behavior (Jonsson et al., 2011). Consequently, early identification and treatment are essential.

However, it can pose a substantial challenge for clinicians to differentiate a depressive disorder from subclinical depressive symptoms. To assess the adolescents' own perspective, self-report questionnaires are useful tools that relate the results of an individual patient to normative data. In addition to measuring depression on a dimensional scale, most of these self-report questionnaires provide cut-off points or other simple algorithms to classify scores into different groups and thus facilitate interpretation. A dichotomy, the distinction between probably “depressed” and “nondepressed” individuals, is necessary for screening purposes, that is, for a preliminary decision if a patient is depressed before or as an adjunct to making a formal diagnosis. A set of cut-off scores can be calculated for more than two levels of depression severity, for example, by comparing the results of patients who were diagnosed with mild, moderate and severe depressive episodes as defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV TR, American Psychiatric Association, 2000). In an early stage of the diagnostic process, this categorization can aid making an elaborate formal diagnostic classification and is the basis for measuring change from one category to another over time.

One instrument that is widely acknowledged for measuring depression severity and that provides such cut-off scores is the revised Beck Depression Inventory (BDI-II, Beck et al., 1996). Its reliability and validity have been confirmed in various studies conducted on adults worldwide (e.g., Hiroe et al., 2005; reviewed by Hautzinger et al. (2006)). Since the questionnaire was designed for individuals from the age of 13, its reliability and its convergent, discriminant and factorial validity have also been approved in adolescent samples (e.g., Byrne et al., 2004, Osman et al., 2004).

Studies investigating the criterion validity of the BDI-II in adolescent samples either compare clinical and non-clinical groups (e.g., Osman et al., 2008) or patients from clinical samples who were and were not diagnosed with a depressive disorder. In two inpatient samples, accuracy of the BDI-II was high (AUC=0.92) in reference to a focused interview guide (Kumar et al., 2002) and moderate (AUC=0.78) in reference to clinical diagnosis (Krefetz et al., 2002). Optimal cut-off scores for the differentiation between patients with and without a depressive disorder were ≥21 and ≥24, respectively. However, no study has reported such a cut-off score for screening adolescents who enter mental health care although the BDI-II and other self-rating scales have been recommended for screening adolescent psychiatric patients early in the diagnostic process (Birmaher et al., 2007).

Concerning the severity cut-off scores, Beck et al. (1996) determined cut-off scores in a sample of adult and adolescent outpatients, which was part of a larger sample with a mean age of 37.20 years. The percentage of adolescents and the mean age of the subsample are not reported. The authors compared patients with mild, moderate, or severe Major Depression according to a structured clinical interview (n=57) and nondepressed patients (n=44), excluding patients with Dysthymic Disorder and others who showed depressive symptoms, but did not meet DSM-III-R criteria of Major Depression at the time of the interview (n=26). Using receiver operating characteristic (ROC) curves, each group was compared to less severely affected patients (e.g. moderately depressed patients to non- and mildly-depressed) to identify cut-off scores for rating depression severity with an optimal sensitivity (i.e., to correctly classify all patients at or above the cut-off score as more severely depressed). Exact figures for the sensitivity and specificity are not reported, although these cut-off scores constitute the only interpretation guideline for the BDI-II in the manual. As none of these analyses compared all patients who were diagnosed with depressive disorders to all non- or subclinically depressed patients, a screening cut-off score for the differentiation of patients with and without a depressive disorder cannot be derived.

The authors of the German BDI-II version (Hautzinger et al., 2006) also recommend the severity cut-off scores reported by Beck et al. (1996). The properties of this version, however, have been examined only once among adolescent psychiatric patients (Besier et al., 2008). In this study with 111 in- and outpatients aged 15–18 years, patients with a clinical diagnosis of a depressive disorder according to ICD-10 criteria (Remschmidt et al., 2006) attained a significantly higher mean BDI-II sum score than nondepressed patients (our calculation: Cohen's d=0.86). As the authors considered the mean BDI-II sum scores comparable to the results in adult samples, they concluded that the cut-off scores provided in the manual could be transferred to adolescent patients without investigating them further. The only study that did evaluate the severity cut-off scores in adolescents used the Turkish BDI-II version in a mixed clinical and non-clinical sample (Uslu et al., 2008). The cut-off score for mild depression (≥10) was found to be four points lower than that proposed by Beck et al. (1996) and three points lower than that found for Turkish adults (Kapci et al., 2008), indicating that age should be regarded when deciding on a cut-off score. Though, it should be noted that the authors did not conduct several ROC analyses for different levels of severity but derived the severity cut-off scores from one global ROC curve.

Therefore, the aim of this study on the German version of the BDI-II (Hautzinger et al., 2006) was twofold:

  • 1.

    Screening: To determine a cut-off score that differentiates depressed from nondepressed adolescent mental health patients with the best possible accuracy.

  • 2.

    Severity rating: To examine the accuracy of the cut-off-scores for depression severity suggested by Beck et al. (1996) in adolescent mental health patients.

Section snippets

Participants

Patients were eligible if they (a) were aged 13–16 years, (b) attended five institutions specialised on child and adolescent psychiatry, psychotherapy and/or psychosomatic medicine in Munich for the diagnosis or treatment of any mental disorder, (c) had sufficient German language skills and (d) sufficient cognitive abilities to fill out the questionnaire and participate in the diagnostic interview.

The study team informed 141 adolescents and families about the study between May 2010 and March

Diagnostic interview results

At the time of the diagnostic interview, four adolescents met the DSM-IV-TR criteria of Dysthymic Disorder (4%) and 20 adolescents met the criteria of Major Depression (23%). Nine adolescents reported a mild, five reported a moderate, and six reported a severe episode. In sum, 24 of 88 adolescents (27%) were diagnosed with a current depressive disorder. Nine of these 24 cases were outpatients, four were inpatients and 11 were waiting for a place in inpatient care.

Of the patients who did not

Detection of depressive disorders

The BDI-II can accurately differentiate adolescent mental health patients with and without a depressive disorder. The AUC value in our study corresponds to the high value found by Kumar et al. (2002). The optimal cut-off score in our heterogeneous sample of mental health users (≥23) is almost equivalent to the scores identified in homogeneous inpatient samples using the original BDI-II version (≥24, Krefetz et al., 2002; ≥21, Kumar et al., 2002). This finding speaks for the utility of our

Acknowledgments

This project was funded by the “Förderprogramm für Forschung und Lehre (FöFoLe)” at the Ludwig-Maximilians-Universität München. We are grateful to all adolescents and families who participated in our study and would like to thank Julia Grzimek, Helga Chiara Schlenz and the teams of the cooperating institutions for their assistance in conducting this study.

References (23)

  • J. Cohen

    Statistical Power Analysis for the Behavioral Sciences

    (1988)
  • Cited by (49)

    • High Beck Depression Inventory 21 scores in adolescents without depression are associated with negative self-image and immature defense style

      2018, Psychiatry Research
      Citation Excerpt :

      For diagnostics in adolescents the gold standard is the semi-structured clinical interview called The Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime version (K-SADS-PL) (Kaufman et al., 1997; Ambrosini, 2000). Studies that compare the results of BDI and clinical diagnostics usually aim at defining the psychometrics of BDI (Kumar et al., 2002; Osman et al., 2008; Dolle et al., 2012). Our aim, in contrast, is to investigate what psychological and background factors explain why the self-report and the diagnostic appraisal may diverge.

    View all citing articles on Scopus
    View full text