Self-report vs. biological markers in assessment of childhood depression

Journal of Affectwe Elsevier

Dmrders,

339

15 (1988) 339-345

JAD 00597

Self-report vs. biological markers in assessment of childhood depression Mary A. Fristad

I, Elizabeth B. Weller I, Ronald A. Weller ‘, Marijo Teare ’ and Sheldon H. Preskorn 3

’ The Ohio State University Department of Psychiatry and Neuroscience Program, 473 W. 12th Avenue, Columbus, OH 43210, U.S.A., ’ University of LouwiNe, Louisudle, KY, U.S.A. and -’ Unioersity of Kansas, School of Medicine and VA Hospital, Wichita, KS, U.S.A. (Accepted

1 June 1988)

Summary

Sensitivity, specificity, and diagnostic confidence of the Children’s Depression Inventory (CDI) and the dexamethasone suppression test (DST) in assessing childhood depression were determined. Parent and child forms of the CD1 and plasma cortisol levels at 8 a.m. and 4 p.m. were used. Three groups of prepubertal children were sampled: (1) 63 depressed inpatients, (2) 14 non-depressed inpatient psychiatric controls, and (3) 21 normal community sample controls. Results suggest that scores 2 15 on the CD1 obtained from either parent or child provide a good screening instrument (sensitivity = 89%). If the DST is administered to all children with elevated CD1 scores, diagnostic confidence is 97.5%. Cautions and recommendations for use of the CD1 and DST are made.

Key words:

Childhood

depression;

Biological

markers;

Introduction

As depression in children has been more frequently diagnosed, self-report inventories and biological markers have been used increasingly (Kazdin, 1981; Kovacs, 1981; Kazdin and Petti, 1982; Weller et al., 1984a). The Children’s Depression Inventory (CDI, Kovacs, 1983, 1985) has

Address for correspondence: M.A. Fristad, Ph.D., The Ohio State University Department of Psychiatry and Neuroscience Program, 473 W. 12th Avenue, Columbus, OH 43210, U.S.A. 0165-0327/88/$03.50

0 1988 Elsevier Science Publishers

Self-report

been one of the most commonly used and researched self-report instruments. The most studied biological marker of depression in prepubertal children has been the dexamethasone suppression test (DST, Weller et al., 1985). The CD1 is a 27-item self-report questionnaire completed by the child. It is patterned after the Beck Depression Inventory (BDI, Beck et al., 1961), and, like the BDI, is heavily weighted with items assessing cognitive aspects of depression. For example, there is only one question each regarding sleep disturbance, appetite disturbance and fatigue, and none on psychomotor agitation/ retardation. However, there are six questions each

B.V. (Biomedical

Division)

340

on loss of interest and worthlessness/guilt. Test-retest reliability (Kazdin et al., 1983) concurrent validity (Hodges et al., 1983; Strauss et al., 1984; Asarnow and Carlson, 1985) and discriminant validity (Saylor et al., 1984) have been demonstrated for the CDI. Norms for children of different ages and sexes have been established (Finch et al., 1985), and its factor structure has been reported (Hodges et al., 1983). A parent form of the instrument has also been developed (CDI-P, Weller and Weller, 1979). Its reliability and validity in a non-clinical population (Wierzbicki, 1988) and its sensitivity and specificity in a clinical population (Fristad et al., 1986) have been reported. It has been used effectively in treatment studies of depressed children as a measure of improvement (Preskorn et al., 1982). The DST has been used to assess hypothalamic-pituitary-adrenal (HPA) axis function in depressed patients (Carroll, 1984). It detects failure of dexamethasone, a potent synthetic corticosteroid, to inhibit cortisol secretion following administration of dexamethasone. There have been several recent reviews of the DST in adults (Arana et al., 1985; Carroll, 1985; APA Task Force, 1987) and children (Weller and Weller, 1988). These reviews indicate that: (1) the sensitivity of the DST ranges from 40% to 78% in adult patients with psychotic affective disorders, melancholia, and mixed manic-depressive states. In children, sensitivity ranges from 14% to 82%; (2) the specificity of the DST in adults ranges from less than 70% to 96%, with the average in studies of psychiatric controls being around 80%. In children and adolescents, specificity ranges from 53% to 100%; (3) while the DST has limitations which must be considered in its use, it can provide useful information when properly performed and interpreted. Carroll (1985) presents a thoughtful review of the controversity and confusion surrounding the use of the DST in research and clinical practice. The DST has also been used in follow-up studies of adults (Greden et al., 1980) and children (Weller et al., 1986). Although the CD1 and the DST have both been used to assess depression in prepubertal children, to date no published study has (to our knowledge) compared the sensitivity and specificity of these two assessment techniques. The purpose of this

study was to compare the sensitivity and specificity of the DST and CD1 utilizing three groups of prepubertal children: depressed inpatients, nondepressed inpatient psychiatric controls, and normal community-sample controls. Method Subjects

Three groups of prepubertal children aged 6-12 were studied: (1) 63 consecutively hospitalized depressed patients. These children met DSM-III criteria for a major depressive episode and had been depressed for at least 30 days prior to hospitalization. All were moderately to severely depressed; (2) 14 consecutively hospitalized psychiatric patients who were not clinically depressed and did not meet DSM-III criteria for any depressive disorder. Their diagnoses included conduct disorder, attention deficit disorder, and oppositional disorder; (3) 21 non-hospitalized normal volunteers from the community without past or present evidence of psychiatric illness. Children with mental retardation, organic brain syndrome, seizure disorder, endocrine disorders, other serious medical illnesses, anorexia, malnutrition, obesity, or who were taking medication known to affect the DST were excluded from the study. Informed assent was obtained from all subjects and informed consent from their parents prior to their participation in the study. Procedure

The Diagnostic Interview for Children and Adolescents (DICA), a structured diagnostic interview, was used to establish criteria-based DSM-III diagnoses (Herjanic and Reich, 1982). It was given to the child in a face-to-face interview by either a clinical research assistant or a nurse clinician trained in its administration. Next, the Children’s Depression Inventory-Child form (CDI-C, Kovacs, 1983, 1985) was administered. Parents were separately administered parent forms of each instrument (DICA-P and CDI-P) by the research assistant or nurse clinician. The DICA and CD1 were administered as part of the routine diagnostic evaluation for the depressed and psychiatric control children.

341

Normal control children and their parents completed these instruments as part of a larger study in which they participated. Because it was necessary to screen normal controls with the DICA, it was administered first for all subjects. The CD1 employs a forced-choice format: each of the 27 items receives a score of 0, 1, or 2. Total scores can range from 0 to 54. A score of 2 15 had previously been determined to be an appropriate cut-off for major depressive disorder in a child inpatient sample (Fristad et al., 1986). In this study, a score of 2 15 from either the parent or child report was considered positive for depression. The DST was performed according to the following protocol. A 0.5-mg dose of dexamethasone was administered at 11 p.m. Post-dexamethasone cortisol levels were measured the next day at 8 a.m. and 4 p.m. (A ‘true suppressor’ would have suppressed cortisol levels at both points in time while an ‘early escaper’ would have non-suppression only at 4 p.m., Preskorn et al., 1987.) The 11 p.m. sampling time sometimes used in adults was not used with these children both because obtaining 11 p.m. blood samples for this group was not deemed practical and because normal controls were being used for comparison. Plasma cortisol levels were determined by radioimmunoassay, utilizing a commercial laboratory assay kit (RIA Kit, Serono Laboratories, Braintree, MA). For purposes of this study, the DST was considered positive if either the 8 a.m. or 4 p.m. post-dexamethasone cortisol level was 2 5 pg/dl. Analyses The potential impact of age and sex on CD1 scores and DST values was determined using chianalyses square and analysis of variance (ANOVA). Then, ANOVAs and Scheffe’s multiple comparison method were used to test for differences in CD1 scores and DST values between diagnostic groups. Next, sensitivity, specificity and diagnostic confidence were calculated. This was done for the individual tests (i.e., the CDI-C, CDI-P, DST-8, DST-4) as well as combinations of CDI-Either (CDI-E: the CDI-C or CDI-P is positive), CDI-Both (CDI-B: both the CDI-C and CDI-P are positive), DST-Either (DST-E: the DST-8 or DST-4 is positive), DST-Both (DST-B:

both the DST-8 and DST-4 are positive). Finally, the distribution pattern of CD1 and DST results was determined. SAS, a statistical software package, was used for all analyses (SAS Institute, 1985). Results Demographics The age and sex of the subjects appear in Table 1. Mean age is the same for all three subject groups. There were more girls in the normal control group than in the two psychiatric groups (x2 = 14.77, P < 0.001). However, sex of the child had no significant effects on CD1 scores or DST values (based on two-way ANOVAs with sex and diagnosis as the independent variables). CDI and DST values Results of DST testing and CD1 scores for the three groups, including the CDI-C, CDI-P, DST-8, DST4 appear in Table 2. As variances were unequal between groups, a log transformation was performed (with scores of 0 on the CD1 being replaced with 0.0001) to bring the variances closer to equality. The groups differed significantly on all four measures (CDI-C, F= 21.76, P < 0.0001; CDI-P, F = 22.50, P < 0.0001; DST-8, F = 13.94, P < 0.0001; DST-4, F = 13.37, P < 0.0001). Pairwise comparisons to determine precisely which groups differed from one another were conducted, with results shown in Table 2. For both the CDI-C

TABLE

1

MEAN

AGE AND

Variable

SEX DISTRIBUTION

OF SUBJECTS

Gr0llp Depressed inpatients (n = 63)

Psychiatric inpatient controls (n =14)

Normal controls (n = 21)

9.7i 1.8 6-13

9.2 f 2.0 6-12

9.311.6 7-12

70 30

71 29

29 71 a

Age Mean + SD Range Sex 96 Boys % Girls * Significantly P < 0.005).

more girls in normal

control

group (x2 = 11.93,

342 TABLE

2

CHILDREN’S DEPRESSION INVENTORY SCORES AND PLASMA CORTISOL LEVELS FOR DEPRESSED, PSYCHIATRIC CONTROL, AND NORMAL CONTROL CHILDREN Variable

CDI-C Mean Range CDI-P Mean Range DST-8 Mean Range DST-4 Mean Range

Group Depressed children (n =63)

Psychiatric inpatient controls (n =14)

Normal controls (n = 21)

* SD

16.3 f 9.6 a l-48

10.9 f 6.4 * O-25

3.2 + 3.1 o-14

k SD

19.6 * 1.1 a 3-41

17.3 * 9.9 il 2-31

4.1 * 3.9 o-14

+ SD

7.2 f 8.2 h 0.5-33.3

1.1*1.1 0.5-4.5

1.5k2.3 0.5-11.1

k SD

6.7 + 5.5 h 0.5-27.2

2.4i1.6 0.5-6.4

3.0 + 3.7 0.5-18.4

a Significantly higher than normal controls (P < 0.05). h Significantly higher than psychiatric controls (P -C0.05) and normal controls (P < 0.05).

and CDI-P, scores were highest for the depressed group and lowest for the normal controls. Psychiatric controls were intermediate in both cases. Scores for the depressed and inpatient control groups did not differ significantly, but both were significantly higher than those for the normal control group. On the DST (both the 8 a.m. and 4 p.m. readings), the depressed group had significantly higher cortisol levels than both the psychiatric and normal control groups. The two control groups did not differ significantly from one another. Classification with the CDI and DST The percent of children classified as depressed according to the CDI-C, CDI-P, CDI-E, CDI-B, the DST-8, DST-4, DST-E, DST-B, or the CDI-E and/or the DST-E (i.e., results from CDI-E and/or DST-E are positive) appear in Table 3. Specificity for individual control groups can be determined by subtracting the percent of children classified as depressed (which appears in Table 3) from 100. Diagnostic confidence (i.e., the ratio of

‘true positives’ to overall number of positives) is also provided in Table 3. Sensitivity (the percentage of depressed children correctly identified as depressed) was highest for the ‘CD1 and/or DST’ category (97%). Specificity (the percentage of non-depressed children correctly identified as non-depressed) is best for the DST-B (100%). Diagnostic confidence is also best for the DST-B (100%). It should be noted that figures for diagnostic confidence provided in this study are based on unequal sample sizes. Had sample sizes been equal (n = 63) for all three groups (and results assumed to be the same as in this study), estimated diagnostic confidence would change, as follows: CDI-C, 67%; CDI-P, 57%; CDI-E, 55%; CDI-B. 81%; DST-8, 90%; DST-4, 81%; DST-E, 79%; DST-B, 100%; CD1 and/or DST, 54%. However, if the purpose of administering the DST is to clarify clinical situations, an equal likelihood of the patient in question having a depressive disorder, some other psychiatric disorder, or no psychiatric disorder would not be expected. Thus, the diagnostic confidence figures in Table 3, which were calculated with fewer non-depressed than depressed subjects, probably reflects how the DST

TABLE

3

PERCENT ON AND AND DST

CD1 Child Parent Either ’ Both DST 8 a.m. 4p.m. Either ’ Both CD1 and/ or DST’ a b ’ d

OF CHILDREN WITH POSITIVE DIAGNOSTIC CONFIDENCE OF

Normal controls (n=Zl)

RESULTS THE CD1

Depressed children (n=63)

Psychiatric controls (n =14)

Diagnostic confidence

44 75 89 35

21 57 71 10

0 0 0 0

90 85 85 96

45 54 67 32

0 7 7 0

5 5 10 0

97 94 93 100

91

71

10

84

Number of true positives/total number of positive Either parent and/or child scores 2 15. Either 8 a.m. or 4 p.m. cortisol levels > 5 pg/dl. Either CDI-E or DST-E.

results.

a

343 TABLE

4

DST AND CHIATRIC CHILDREN

CD1 RESULTS AMONG DEPRESSED, PSYCONTROL AND NORMAL CONTROL

because they had negative DST results. Thus, diagnostic confidence and specificity were increased at the expense of lower sensitivity. Discussion

Test results a

DST + /CD1 + DST+/CDIDST - /CD1 + DST-/CDI-

Group Depressed (n = 63)

Psychiatric control (n=14)

Normal control (n=27)

62% 5% 29% 5%

7% 0% 64% 29%

0% 10% 0% 90%

a Results are for combined DST-4.

CDI-C

and CDI-P

and DST-8 and

would be used in clinical practice, i.e., administered only when a diagnosis of affective disorder is strongly considered. Distribution pattern of CDI and DST results The distribution of DST and CD1 (CDI-C and CDI-P) results among the three groups appears in Table 4. A majority of depressed children had positive DST and CD1 results (62%). A sizeable portion had negative DSTs and positive CDIs (29%). Few children had positive DSTs and negative CDIs (5%) or both CD1 and DST results negative (5%). Nearly two-thirds of the psychiatric control children had positive CDIs and negative DSTs (64%). Most other psychiatric controls had negative results on both tests (29%). Only 7% had positive results on both tests. None had a positive DST with a negative CDI. Among the normal controls, 90% had negative results on both tests. The remaining 10% had positive DSTs and negative CDIs. If the CD1 was used as a ‘screening test’ and the DST used as a ‘confirmatory test’ on all children with elevated CD1 scores (from either parent or child form), diagnostic confidence was increased. In this sample of 67 children with elevated CD1 scores, 40 had elevated cortisols and of these, 39 had a clinical diagnosis of depression. The diagnostic confidence was 97.5%. However, false-negatives occurred in both steps of this procedure. In this sample, 10% of depressed children had negative CDIs and would have been missed. Another 29% of depressed children were missed

This study examined the sensitivity, specificity, diagnostic confidence, and distribution of results for the CD1 and DST in 63 inpatient depressed, 14 inpatient control, and 21 normal control children. Scores from both parent and child forms of the CD1 were significantly higher for depressed child psychiatry inpatients and non-depressed child psychiatry inpatients than for normal control children. These findings suggest that a CD1 score of 2 15 is a good indicator of psychopathology. Unfortunately, among inpatients, it did not discriminate between depressed and non-depressed children. It is unclear if the order effects of administering the structured interview first changed the responses on the CDI. In our experience, it was felt to be unlikely. The DST discriminated depressed inpatients from psychiatric inpatient controls and from normal controls. This was true for both 8 a.m. and 4 p.m. post-dexamethasone cortisol levels. As regards the distribution of test results, a majority (64%) of psychiatric control children had both a positive CD1 and a negative DST. Most (90%) normal control children had both tests negative. If these tests are considered in terms of ‘screening’ and ‘confirmatory test’ instruments, then their use could be complementary. The CD1 is quick and inexpensive, and could easily be administered to patients in whom depression is suspected. If the parent or child CD1 score is elevated, the DST. a more expensive and ‘demanding’ test. would be administered. If this paradigm had been used with these subjects, diagnostic confidence would have been improved to 97.5% although sensitivity would have been 62%. Use of the two tests in such a complementary fashion is similar to the use of tests in other areas of medicine. By reducing the number of DSTs performed, it would also minimize laboratory expenses of the DST. Conclusions

not

In conclusion, it appears that the CD1 should be used to diagnose childhood depression.

344

Specificity among this inpatient psychiatric control group was 29%. Nor should child and parent CD1 scores be used interchangeably since parents of non-depressed psychiatrically disturbed children were more than twice as likely as their children to report scores indicating depression. The CD1 may be used as a general screening test for depression, at least among psychiatrically hospitalized children, especially if both parent and child forms are used. Sensitivity was 89% in this study. Once factors that invalidate the DST are considered (Carroll, 1985) and accurate standardization of the laboratory determined (Carroll, 1985) it can be used to support a suspected clinical diagnosis. The DST should not, however, be used as a general screening test (Baldessarini et al., 1983). In this study. diagnostic confidence was 97.5% when the DST was used only on children with elevated CD1 scores. Additional research will be required to clarify further the predictive value of both the CD1 and DST with regard to treatment response and longterm outcome. References APA Task Force on Laboratory Tests in Psychiatry (1987) The dexamethasone suppression test: an overview of its current status in psychiatry. Am. J. Psychiatry 144, 1253-1262. Arana, G.W., Baldessarini, R.J. and Ornsteen, M. (1985) The dexamethasone suppression test for diagnosis and prognosis in psychiatry. Arch. Gen. Psychiatry 42, 1193-1204. Asarnow. J.R. and Carlson, G.A. (1985) Depression Self-Rating Scale: utility with child psychiatry inpatients. J. Consult. Clin. Psychol. 53, 491-499. Baldessarini, R.J., Finkelstein, S. and Arana, G.W. (1983) The predictive power of diagnostic tests and the effect of prevalence of illness. Arch. Gen. Psychiatry 40, 569-573. Beck, A.T.. Ward, C.H., Mendelson, M. et al. (1961) An inventory for measuring depression. Arch. Gen. Psychiatry 4. 561-571. Carroll. B.J. (1984) Dexamethasone suppression test. In: R.C.W. Hall and T.P. Beresford (Eds.), Handbook of Psychiatric Diagnostic Procedures. Spectrum, New York, NY. Carroll, B.J. (1985) Dexamethasone suppression test: a review of contemporary confusion. J. Clin. Psychiatry 46, 13-24. Carroll, B.J., Feinberg, M., Greden, J.F. et al. (1981) A specific laboratory test for the diagnosis of melancholia: standardization, validation. and clinical utility. Arch. Gen. Psychiatry 38, 15-22. Finch, A.J., Saylor, C.F. and Edwards, G.L. (1985) Children’s Depression Inventory: sex and grade norms for normal children. J. Consult. Clin. Psychol. 53, 424-425.

Fristad. M.A., Weller, E.B., Weller, R.A. and Teare. M. (1986) Children’s Depression Inventory: psychometric properties of parent and child forms. Presented at the American Academy of Child and Adolescent Psychiatry Meeting, Los Angeles, CA. Greden. J.F., Albala. A.A., Haskett, R.F. et al. (1980) Normalization of dexamethasone suppression test: a laboratory index of recovery from endogenous depression. Biol. Psychiatry 15, 449-458. Herjanic, B. and Reich, W. (1982) Development of a structured psychiatric interview for children: agreement between child and parent on individual symptoms. J. Abnorm. Child Psychol. 10, 3077324. Hodges. K.D., Siegel, J.S., Mullins, L. and Griffin, N. (1983) Factor analysis of the Children’s Depression Inventory. Psychol. Rep. 53, 7599763. Kazdin, A.E. (1981) Assessment techniques for childhood depression J. Am. Acad. Child Psychiatry 20. 358-375. Kazdin, A.E. and Petti. T.A. (1982) Self-report and interview measures of childhood and adolescent depression. J. Child Psychol. Psychiatry 23, 437-451. Kazdin, A.E., French, N.H., Unis. A.S. and Esveldt-Dawson, K. (1983) Assessment of childhood depression: correspondence of child and parent ratings. J. Am. Acad. Child Psychiatry 22. 157-164. Kovacs, M. (1981) Rating scales to assess depression in children. Acta Paedopsychiatr. 46, 3055315. Kovacs, M. (1983) The Children’s Depression Inventory. A self-rated depression scale for school-aged youngsters. Unpublished manuscript, University of Pittsburgh School of Medicine (Available from Dr. M. Kovacs. WPIC. 3817 O’Hara St., Pittsburgh, PA 15213). Kovacs, M. (1985) The Children’s Depression Inventory (CDI). Psychopharmacol. Bull. 21, 995-998. Poznanski, E.O., Carroll. B.J., Banegas, M.D. et al. (1982) The dexamethasone suppression test in prepubertal depressed children. Am. J. Psychiatry 139, 321-324. Preskorn, S.H., Weller, E.B. and Weller, R.A. (1982) Depression in children: relationship between plasma imipramine levels and response. J. Clin. Psychiatry 43, 450-453. Preskorn, S.H., Weller, E.B., Hughes, C.W., Weller, R.A. and Bolte, K. (1987) Depression in prepubertal children: dexamethasone non-suppression predicts differential response to imipramine vs. placebo. Psychopharmacol. Bull. 23, 1288133. SAS Institute (1985) SAS User’s Guide: Statistics, Version 5 edn. Sas Institute Inc., Cary, NC, 956 pp. Saylor, C.F., Finch, A.J., Spirito, A. and Bennett, B. (1984) The Children’s Depression Inventory: a systematic evaluation of psychometric properties. J. Consult. Clin. Psychol. 52, 9555967. Strauss, C.C., Forehand, R., Frame, C. and Smith. K. (1984) Characteristics of children with extreme scores on the Children’s Depression Inventory. J. Clin. Child Psychol. 13. 227-231. Targum, S., Chastek, C. and Sullivan, A. (1981) Dexamethasone suppression test in prepubertal conduct disorder. Psychiatr. Res. 5, 107-108.

345 Weller, E.B. and Weller, R.A. (1979) The Children’s Depression Inventory-Parent Form. Unpublished document, The University of Kansas Medical Center, Kansas City, KS. Weller, E.B. and Weller, R.A. (1988) Neuroendocrine aspects of depression in children and adolescents. Psychiatr. Clin. N. Am. (in press). Weller, E.B., Weller, R.A. and Fristad, M.A. (1984a) Historical and theoretical perspectives on childhood depression. In: E.B. Weller and R.A. Weller (Eds.), Current Perspectives on Major Depressive Disorders in Children. American Psychiatric Press, Washington, DC, pp. l-18. Weller, E.B., Weller, R.A., Fristad, M.A. et al. (1984b) The dexamethasone suppression test in hospitalized prepubertal children. Am. J. Psychiatry 141, 290-291.

Weller. E.B., Weller. R.A.. Fristad, M.A.. Preskom, S.H. and Teare, M. (1985) The dexamethasone suppression test in prepubertal depressed children. J. Clin. Psychiatry 46. 511-513. Weller, E.B., Weller, R.A., Fristad, M.A., Cantwell, M.L. and Preskorn, S.H. (1986) Dexamethasone suppression test and clinical outcome in prepubertal depressed children. Am. J. Psychiatry 143, 146991470. Wierzbicki, P. (1988) Reliability and validity of the children’s depression inventory in a nonclinical population. J. Clin. Psychol. (in press).