assignments
Eriana15
CDI_predictive_of_MDD_article_April2018.pdf
Hispanic Journal of Behavioral Sciences 32(2) 232 –258
© The Author(s) 2010 Reprints and permission: http://www. sagepub.com/journalsPermissions.nav
DOI: 10.1177/0739986310361919 http://hjbs.sagepub.com
A Study of the Predictive Validity of the Children’s Depression Inventory for Major Depression Disorder in Puerto Rican Adolescents
Carmen L. Rivera-Medina,1 Guillermo Bernal,1 Jeannette Rosselló,1 and Eduardo Cumba-Aviles1
Abstract
This study aims to evaluate the predictive validity of the Children’s Depression Inventory items for major depression disorder (MDD) in an outpatient clinic sample of Puerto Rican adolescents. The sample consisted of 130 adolescents, 13 to 18 years old. The five most frequent symptoms of the Children’s Depression Inventory that best predict the presence of MDD were “I worry about others’ aches and pains,” “I don’t have any friends,” “I have to push myself to do my schoolwork,” “I have trouble sleeping every night,” and “I do very badly in subjects I used to be good in.” Results demonstrated that the symptoms that best predict MDD in Puerto Rican adolescents are not necessarily the ones commonly described as characteristic of the disorder.
Keywords
Hispanic, depression, adolescents, psychometrics, inventory
1University of Puerto Rico, San Juan, Puerto Rico
Corresponding Author: Carmen L. Rivera-Medina, Institute for Psychological Research, University of Puerto Rico, Río Piedras Campus, PO Box 23174, San Juan, Puerto Rico Email: [email protected]
Rivera-Medina et al. 233
Identifying and treating major depression disorder (MDD) in adolescents represent a challenge for mental health professionals. The negative effects of MDD symptoms affect various levels of adolescent functionality. Also, the effects related to the severity of MDD itself or to its combination with other disorders often make diagnosis difficult. In most cases, comorbid disorders develop as a complication of depression and persist even after the depression has remitted (Kovacs, Palauskas, Gatosnis, & Richards, 1988). The risk of recurrent depression in adolescents is substantial, with 12% having a subse- quent depressive episode within 1 year and 33% within 4 years (Lewinsohn, Clarke, Seeley, & Rohde, 1994). Also, the presence of depressive disorders in adolescents suggests a high probability of recurring episodes in adulthood (Harrington, Fudge, Rutter, Picklels, & Hill, 1991; Kandel & Davies, 1986).
It is well known that MDD and dysthymic disorder (DD) are associated with significantly higher rates of suicidal behavior (Bennett, 1994) than non- depressive disorders (Kovacs, Goldston, & Gatsonis, 1993). A study of hospitalized adolescents with MDD found that 61% of them reported suicidal ideation (Ryan et al., 1987). However, only 28% of suicidal adolescents had received psychological or emotional counseling (Pirkis et al., 2003). Consid- ering that suicide is the third leading cause of death for adolescents aged 15 to 19 years (Martin, Smith, Matthews, & Ventura, 1999), the American Acad- emy of Pediatrics (2000) has urged pediatricians to become involved in efforts to prevent adolescent suicide by being aware of the symptoms of depression and other presuicidal behavior. However, the challenge of identi- fying depressive symptoms in adolescents is not only relevant to pediatricians but also to all health professionals and friends and family. Precise identifica- tion of depressive symptoms in clinical and nonclinical contexts is important; thus, there is a need for brief scales of symptoms that accurately allow professionals to predict the probability of MDD in diverse contexts.
Challenges for Health Professionals There are several problems faced by clinicians in their attempts to identify depressive symptoms in adolescents. First, the prevalence of MDD in adoles- cents is thought to be higher than what has been reported, since the disorder frequently passes unnoticed (Lewinsohn, Rohde, & Seeley, 1998). This could be a consequence of the difficulty that parents and other significant persons have in recognizing the disorder. Also, since children with externalizing or disorganized behavior tend to call more attention than those with internalizing behavior, depression in the latter often goes unnoticed (Hammen & Rudolph, 1996). However, this explanation may be reasonable when working with adolescents
234 Hispanic Journal of Behavioral Sciences 32(2)
from a European American background, but when working with groups of diverse cultural backgrounds, the problem of recognizing the disorder could be more complicated.
The literature points to ethnic disparities in depression among adults and recently to ethnic differences among adolescents (Crockett, Randball, Shen, Russell, & Driscoll, 2005). Not only have higher rates of depressive symp- toms been observed in Latinos compared with Anglos (Knight, Virdin, Ocampo, & Roosa, 1994; Roberts & Chen, 1995; Roberts & Sobahn, 1992), but the possibility of different symptomatic expressions of depression among different cultural groups has been raised as well, bringing into question issues of measurement equivalence. For example, the use of screening measures developed in the United States or in other countries is widely accepted among Latino/a professionals in general and Puerto Rican mental health profession- als in particular (Rivera, Bernal, & Rosselló, 2005). Yet instruments that do not have measurement equivalence across diverse ethnic groups could result in misleading results and measures with different meaning and validity across groups (Vandenberg & Lance, 2000).
The construct of depression may differ across cultural groups, with each group conceptualizing it differently and using different symptoms to identify it (Marsella & Yamada, 2007), therefore affecting the measurement equiva- lence of the instruments (Crockett et al., 2005). If the items used to measure depression are poor indicators of depression in one group, the MDD in that group will be underestimated. It is, therefore, necessary to incorporate criteria that are sensitive to cultural differences (Abdel-Kahalec & Soliman, 1999). According to the American Psychiatric Association’s (APA) Diagnostic and Statistical Manual of Mental Disorders, Text Revision (DSM-IV-TR; APA, 2000), the available evidence suggests that the symptoms and the develop- ment of mental disorders may be influenced by ethnic and cultural factors, recognizing the importance of assessing differences in symptoms that may be present in diverse cultures.
The understanding of childhood and adolescent depression has also been hampered in part by the lack of well-established instrumentation for case assessment in these populations (Roberts, Lewinsohn, & Seeley, 1991). For example, consider one context in which one would expect to easily identify depressive symptoms: clinical research. In this scenario, case assessment has historically been accomplished by using a combination of two approaches: clinical and nonclinical. Nonclinical measures, usually in the form of brief symptoms checklists, are used as a first step because they provide a fast and economical method of partitioning the study population into groups presumed to be “well” and “ill.” The clinical interview approach is a second alternative
Rivera-Medina et al. 235
that uses more comprehensive techniques to establish a psychiatric diagnosis. However, little is known about the precision and validity of brief depressive symptom scales as first-step screening instruments. Although several instru- ments have been developed to measure depressive symptoms, there is little evidence as to the agreement between the scales and diagnostic procedures when this two-stage identification process is used in studies with children and adolescents. If this is the case in the clinical research context, where measurement and evaluation procedures tend to be structured, one wonders what could be happening in mental health service delivery settings that generally do not employ rigorous screening procedures.
Challenge in Evaluating Predictive Validity Another challenge is selecting the proper method for the evaluation of predictive validity. Many instruments used as screening scales include the symptoms that characterize a disorder, thus permitting the evaluation of the diagnostic efficiency of the symptoms on the scale. Faraone, Biederman, Sprich-Buckminster, Chen, and Tsuang (1993) define diagnostic efficiency as the degree to which the individual criteria or symptoms correctly discrimi- nate cases from noncases as evaluated by an actuarial prediction (conditional probability). This method has been useful for the development of diagnostic algorithms and to demonstrate the divergent validity for psychiatric disorders in children. One of the typical methods to evaluate diagnostic efficiency is using the sensitivity (SEN) and specificity (SPE) indexes. SEN is used to evaluate the predictive validity of the symptom and the probability of having the symptom given the presence of a disorder (known as the true positives; Mausner & Kramer, 1985). SPE is the probability of not having the symptom given the absence of a disorder.
However, while SEN and SPE are essential for the evaluation of the effi- ciency of a symptom, those indexes in and of themselves do not provide sufficient or precise enough information (Laurent, Landau, & Stark, 1993; Widiger, Hurt, Frances, Clarkin, & Gilmore, 1984). Having the highest SEN for a selected symptom does not necessarily mean that this symptom will give you the higher probability of identifying the presence of a disorder. Actually, this probability, known as the positive predictive power (PPP), may, in some cases, decrease while the SEN increases. A similar tendency may be observed with the SPE index and the probability of the absence of the disorder given the absence of a symptom, known as the negative predictive power (NPP). Therefore, the PPP and the NPP must not be overshadowed by the SEN and SPE indexes. Assessing the degree to which individual
236 Hispanic Journal of Behavioral Sciences 32(2)
symptoms predict the disorder can provide an estimate of the probability that the person has for meeting the criteria for a disorder, even in the absence of an instrument that measures the full criteria for the disorder.
There are studies on the SEN and SPE of instruments that measure depres- sive symptomatology related to MDD but only consider the total score of the instrument (Ambrosini, Metz, Bianchi, Rabinovich, & Undie, 1991; Barrera & Garrison-Jones, 1998; Bird, Gould, Rubio-Stipec, Staghezza, & Canino, 1991; Kashani, Sherman, Parker, & Reid, 1990; Kovacs, 1983; Nurcombe et al., 1989; Strober, Green, & Carlson, 1981). Other diagnostic studies have focused on identifying the symptoms that significantly differentiate children or adolescents referred to treatment from those not referred or from those with other diagnoses (Pellham, Gnagy, Greenslade, & Milich, 1992) but not on evaluating their predictive validity with the disorder.
MDD may be manifested in several ways (DSM IV; APA, 1994). For some adolescents, symptoms may be similar to those in adults, with symptoms such as depressed mood most of the day, crying spells, discouragement, irritability, a sense of emptiness and meaninglessness, negative expectations of self and the environment, low self-esteem, isolation, a feeling of helpless- ness, markedly diminished interest or pleasure in most activities, significant weight loss or weight gain, insomnia or hypersomnia, fatigue or loss of energy, feelings of worthlessness, and diminished ability to think or concen- trate. It is, however, more common for an adolescent with serious depression to exhibit psychosomatic symptoms or behavioral problems.
Lucas et al. (2001) evaluated a series of brief diagnosis-specific scales to identify subjects who have a high probability of meeting diagnostic criteria for MDD. They identified, as gate items for MDD in youth, issues such as depressive mood, loss of interest/boredom, and hopelessness. Roberts et al. (1991) found that the symptoms that best predict MDD in an adolescent sample, using the Beck Depression Inventory (BDI) as a screening measure were, in hierarchical order: “feel sad,” “blame myself,” “cry,” and “no appe- tite.” With the same sample, but using the Center for Epidemiologic Studies Depression Scale (CES-D), they found that the best predictors were “felt depressed,” “poor appetite,” “felt sad,” and “could not get going.” Laurent et al. (1993) reported that “feeling that no one loves you,” anhedonia, exces- sive guiltiness, and depressed mood best serve as inclusion criteria for depression. Although not with the purpose of assessing the predictive validity of the items, Rubio-Stipec et al. (1996) factor analyzed the Diagnostic Inter- view Schedule for Children (DISC-2.3) stem items and derived an 18-item scale for measuring depression that included the following: sad, grouchy, anhedonia, loss of interest, anorexia, hypersomnia, talkative, poor academics,
Rivera-Medina et al. 237
tired, loss of energy, feelings of guilt, worthlessness, tearful, attention and concentration problems, and the perception that things go wrong. Yet among the few studies that have assessed the predictive validity of the symptoms for MDD and have taken into consideration Spanish-speaking or Latino partici- pants, none have evaluated the Children’s Depression Inventory (CDI) items for screening purposes (Bird et al., 1988; Lucas et al., 2001; Roberts et al., 1991; Rubio-Stipec et al., 1996).
Predictive Validity of the CDI Items The CDI (Kovacs, 1992) is consistently cited in the literature as a good instrument to identify depressive symptoms in children and adolescents and screen for MDD (Chan, 1997; Fristad, Emery, & Beck, 1997; Nurcombe et al., 1989). It is not only one of the most frequently used instrument to identify depressive symptoms (Compas, Ey, & Grant, 1993; Fristad et al., 1997; Pon- terotto, Pace, & Kavan, 1989; Sitarenios & Kovacs, 1999) but also one of the few symptom scales that has items corresponding to each of the DSM-III-R (APA, 1987) criteria for the MDD. These criteria have remained nearly iden- tical in the various editions of diagnostic manuals: the DSM-III, DSM-III-R, DSM-IV, and DSM-IV-TR. Although the CDI is a good indicator of self- reported distress in children, some studies have demonstrated that it does not have adequate SEN and SPE as a screening measure of depression (Fristad, Weller, Weller, Teare, & Preskorn, 1991; Nelson, Politano, Finch, Wendel, & Mayhall, 1987; Saylor, Finch, Spirito, & Bennett, 1984; Weiss et al., 1991). Other studies have concentrated on assessing the variables related to depressive symptoms (Alatorre-Alva & De los Reyes, 1999; Del Barrio, Moreno-Rosset, López-Martínez, & Olmedo, 1997; Llabre & Hadi, 1997; Puura et al., 1998). Still, other investigators (Mitchell, McCauley, Burke, & Moss, 1988; Ryan et al., 1987) only report the frequency of the symptoms in the diagnosis of MDD in samples with children and adolescents. In addition, the majority of the studies with the CDI in other cultures or ethnic groups outside the United States have been limited to the assessment of the levels of depressive symptoms in adolescents reporting comparisons across groups or settings (Charman & Pervova, 1996; Hayward, Gotlib, Schraedley, & Litt, 1999; Siegel, Aneshensel, Taub, Cantwell, & Driscoll, 1998).
In Puerto Rico, the CDI was translated and culturally adapted to a Spanish version considering the semantic, contextual, and technical equivalence to the English version (Bravo, 2003; Rosselló, Guisasola, Ralat, Martínez, & Nieves, 1992). However, the predictive validity of the items was not evalu- ated. Depression was found to be the disorder with the fourth highest
238 Hispanic Journal of Behavioral Sciences 32(2)
prevalence in children and adolescents in Puerto Rico (Canino et al., 2004). Therefore, it is critical to identify the symptoms that have the diagnostic efficiency necessary for the identification of the disorder in its early stages. The objective of this study was to evaluate the extent to which the CDI items predict DSM-IV MDD in a Puerto Rican clinical sample of adolescents. Specifically, the question is whether or not subsets of the items of the CDI Spanish version (Rosselló et al., 1992) would work well (have a good diagnostic efficiency) as a brief screener for MDD.
Method Participants
The sample consisted of 130 Puerto Rican (Spanish-speaking) adolescents originally screened for participation in a controlled randomized clinical trial to assess the efficacy of two treatments for depression in adolescents (Rosselló, Bernal, & Rivera-Medina, 2008). Participants eligible for this study were between 13 and 18 years old, met regular DSM-IV criteria (APA, 1994) for MDD and/or DD, and were deemed by a clinical interviewer to be impaired. Also eligible were those who did not meet the DSM-IV criteria for MDD but who obtained a score of 13 or higher on the CDI, the cutoff point suggested by Kovacs (1983) to identify mild to moderate depressive symptoms. Since the sample was part of a clinical trial, the following conditions resulted in exclusion from the study: serious imminent suicide risk, psychosis, mental retardation, hyperaggression, current regimen of psychopharmacological medication or psychotherapy, involvement in legal proceedings, bipolar or conduct disorder, or drug use. Adolescents were referred from schools in the metropolitan San Juan area. A total of 322 referrals were received; 130 of these met criteria for inclusion in the present study. Of the participants, 70 (54%) were female, and 60 (46%) were male. All were between 13 and 18 years old, with a mean age of 15. Their mean grade level was 9.28, with 50.8% of the participants attending public schools and 49.2% attending private schools.
Instruments The Children’s Depression Inventory (CDI), Spanish version. The Spanish version
of the CDI is a 27-item self-rated symptom-oriented scale translated to Spanish and culturally validated for Puerto Rican adolescents by Rosselló et al. (1992). In its original English version, this instrument was developed, based on the BDI, to distinguish youth with a psychiatric diagnosis of MDD
Rivera-Medina et al. 239
or DD from those with other psychiatric conditions or “normal” school chil- dren (Kovacs, 1992). Each item is scored according to one of the following alternatives: 0 = absence of the symptom, 1 = moderate symptom, 2 = severe symptom. In this study, a symptom was considered present if the adolescent rated the item with the most severe score. Kovacs (1992) reported a reliabil- ity coefficient of .86 for the scale and found it to be a valid measuring device when compared with other instruments. Data obtained using the CDI with different Puerto Rican samples suggest that it has a good internal consistency with a values above .83 (Bernal, Rosselló, & Martínez, 1997; Rosselló et al., 1992). For the present study, a reliability coefficient of .82 was observed.
The Diagnostic Interview Schedule for Children, version 2.3 (DISC-2.3). The DISC-2.3 is a comprehensive, highly structured diagnostic research instru- ment that assesses the most common diagnoses among children and adolescents through parent and child interviews. The English version of the DISC-2.3 has been shown to generate reliable and valid diagnoses for most diagnostic categories (Schwab-Stone et al., 1996); similar findings have been reported for the Spanish version of the instrument (Bravo, Woodbury- Fariña, Canino, & Rubio-Stipec, 1993; Ribera et al., 1996). The Spanish version of the DISC-2.3 was used to provide a structured format for the interview and reduce informant variance in the interviews with adolescents and their parents. However, the diagnoses of MDD were not computer derived from the DISC-2.3 algorithms. Only the section that evaluates affective disorders was used. A symptoms checklist based on DSM-IV cri- teria for MDD was completed for each participant. The diagnoses for MDD were made by a clinician using the information obtained by the DISC-2.3 and the symptoms checklist.
Procedure for Diagnosis Doctoral candidates in clinical psychology were the interviewers performing the clinical evaluations. A PhD clinical psychologist supervised all evalua- tions. DSM III-R diagnoses were made by PhD candidates using the information obtained in the Spanish version of the DISC-2.3 and a symptom checklist based on DSM-IV (APA, 1994) completed for each participant. Therefore, the algorithms of the DISC-2.3 were not used to arrive at a DSM-IV diagnosis. The interviewers were aware of the principal aim of the clinical trial but not of the aims of the present study. According to the infor- mation obtained, 62.5% of the adolescents received a diagnosis of MDD, 5.0% received a diagnosis of dysthymia, and 32.5% did not comply with the criteria for either MDD or DD.
240 Hispanic Journal of Behavioral Sciences 32(2)
Analysis of Data
Receiver operating characteristic curve analysis is another statistical approach that appears in the literature to evaluate the predictive validity of the items or the instruments in general. However, this analysis only considers the SEN and SPE of the items to create the area under the curve not providing enough information to evaluate the probability of MDD given the presence of the item. Therefore, predictive validity indexes were used to evaluate the predic- tive validity of the CDI items for MDD in this sample.
Once the diagnosis of MDD was established, the following information was calculated to obtain the predictive validity indexes:
1. True positives (TP): had MDD and scored 2 (“severe symptoms”) on the items.
2. True negatives (TN): scored negative for MDD and scored 0 or 1 on the items.
3. False positives (FP): scored 2 on the symptom but do not comply with the DSM-IV criteria for MDD.
4. False negatives (FN): scored 0 or 1 but met the DSM-IV criteria for MDD.
With the above information, the following indexes were calculated to assess predictive validity of the symptoms:
1. Sensitivity (SEN): the probability of having the symptom given the presence of the disorder; expressed as the percentage of cases scoring positively to the symptom from those diagnosed with MDD (SEN = TP/(TP + FN)).
2. Specificity (SPE): the probability of not having the symptom given the absence of the disorder; expressed as the percentage of cases scoring negatively to the item from those not diagnosed with MDD (SPE = TN/(TN + FP)).
3. Positive predictive power (PPP): the probability of having MDD given the presence of the symptom; expressed as the percentage of cases with MDD of those who scored 2 on the symptom (PPP = TP/(TP + FP)).
4. Negative predictive power (NPP): the probability of not having MDD given the absence of the symptom; expressed as the percent- age of true negative from those who scored 0 or 1 on the symptom (NPP = TN/(TN + FN)).
Rivera-Medina et al. 241
5. Total predictive power (TPP): the overall capability of the item to accurately identify either cases or noncases (TPP = TP + TN/(TP + FP + FN +TN)).
6. Diagnostic efficiency index (d): indicates the distance between the ideal point and the point where the symptom is actually localized. Lower values mean less distance between the two points, and therefore are better indexes (d = √(1-SEN)2 + (1-PPP)2).
7. Prevalence of symptoms (Q): (TP+FP/(TP+FP+FN+TN)).
The literature tends to present only the results for the indexes (usually the SEN and SPE indexes in some articles and the PPP and the NPP in others), providing no information on how the sample is distributed in terms of the disorder (TP, TN, FP, and FN). We consider this omitted information to be essential for the evaluation process of the items. As can be seen from the analysis of the results, the indexes themselves did not provide sufficient information to decide whether a given item is superior to another. For the purposes of this report, obtaining a balance between the indexes as well as between the FP and FN rates was important when considering the diagnostic efficiency and predictive validity of the symptoms. Prevalence of the symp- toms was also considered, since the PPP and the NPP indexes may be affected by their prevalence in such a way that a symptom with high SEN and SPE but extremely high prevalence may obtain a low predictive power and have no diagnostic utility (Landau, Milich, & Widiger, 1991; Widiger et al., 1984). Finally, since the main purpose of the study was to assess the predic- tive validity of the items for MDD, special attention were given to the PPP, NPP, PPT, and d indexes.
Results The evaluation of the psychometric characteristics of the CDI items revealed that, in general, the probability of identifying the presence of the symptoms given the presence of MDD, SENs were quite low for all the items, fluctuat- ing from .00 to .39 for the complete sample. However, the indexes obtained for the probability of MDD given the presence of the symptoms, PPP, evi- dence their diagnostic utility. Table 1 presents the CDI items with their predictive values, diagnostic efficiency index, and percentage distribution of cases for MDD. In hierarchical order, the symptom that obtained the best diagnostic efficiency was “I worry about others’ aches and pains all the time,” with a diagnostic efficiency (d) index of .73 and a TPP index of .50. The probability of having MDD given the presence of this symptom, PPP, was
242
T a b
le 1
. Pr
ed ic
ti ve
V al
ue s,
D ia
gn o st
ic E
ffi ci
en cy
, a nd
P er
ce nt
ag e
D is
tr ib
ut io
n o f th
e M
D D
fo r
th e
To p
10 C
D I It
em s
fo r
th e
C o m
pl et
e Sa
m pl
e o f Pu
er to
R ic
an A
do le
sc en
ts F
ro m
a n
O ut
pa ti en
t C
lin ic
C D
I It
em s
Q (
% )
SE N
SP
E PP
P N
PP
T PP
d
T P
(% )
FP (
% )
FN (
% )
T N
( %
)
I w
o rr
y ab
o ut
o th
er s’
a ch
es a
nd p
ai ns
a ll
th e
ti m
e 33
.3
6 .7
1 .6
5 .4
3 .5
0 .7
3 21
.7
11 .6
38
.0
28 .7
I ha
ve t
o p
us h
m ys
el f al
l t he
t im
e to
d o m
y sc
ho o lw
o rk
40
.3
9 .6
0 .5
9 .4
0 .4
7 .7
4 23
.3
16 .3
36
.4
24 .0
I do
n o t
ha ve
a ny
f ri
en ds
26
.3
0 .7
9 .6
8 .4
4 .5
0 .7
7 17
.7
8. 5
41 .5
32
.3 I ha
ve t
ro ub
le s
le ep
in g
ev er
y ni
gh t
22
.2 6
.8 5
.7 1
.4 4
.5 0
.7 9
15 .5
6.
2 44
.2
34 .1
T hi
ng s
bo th
er m
e al
l t he
t im
e 20
.2
3 .8
5 .6
9 .4
3 .4
8 .8
3 13
.8
6. 2
45 .4
34
.6 I fe
el li
ke c
ry in
g ev
er y
da y
19
.2 2
.8 5
.6 8
.4 3
.4 8
.8 4
13 .1
6.
2 46
.2
34 .6
I ca
n ne
ve r
be a
s go
o d
as o
th er
k id
s 23
.2
5 .7
9 .6
3 .4
2 .4
7 .8
4 14
.6
8. 5
44 .6
32
.3 I am
s ad
a ll
th e
ti m
e 18
.2
1 .8
7 .7
0 .4
3 .4
8 .8
5 12
.3
5. 4
46 .9
35
.4 N
o th
in g
w ill
e ve
r w
o rk
fo r
m e
15
.1 8
.9 0
.7 4
.4 3
.4 7
.8 6
10 .9
3.
9 48
.8
36 .4
M o st
d ay
s I do
n o t
fe el
li ke
e at
in g/
I ea
t al
l t he
t im
e 21
.2
1 .7
9 .5
9 .4
1 .4
5 .8
9 12
.3
8. 5
46 .9
32
.3
N o te
: M D
D =
m aj
o r
de pr
es si
o n
di so
rd er
; C D
I =
C hi
ld re
n’ s
D ep
re ss
io n
In ve
nt o ry
; Q =
p re
va le
nc e
o f sy
m pt
o m
s; SE
N =
s en
si ti vi
ty ; S
PE =
s pe
ci fic
it y;
PP
P =
p o si
ti ve
p re
di ct
iv e
po w
er ; N
PP =
n eg
at iv
e pr
ed ic
ti ve
p o w
er ; T
PP =
t o ta
l p re
di ct
iv e
po w
er ; d
= d
ia gn
o st
ic e
ffi ci
en cy
in de
x; T
P =
t ru
e po
si ti ve
s; FP
= f al
se p
o si
ti ve
s; FN
= f al
se n
eg at
iv es
; T N
= t
ru e
ne ga
ti ve
s.
Rivera-Medina et al. 243
found to be .65, and the probability of not having MDD given the absence of the symptom, NPP, was .43. The second-best symptom to predict MDD was “I have to push myself all the time to do my schoolwork,” with a d index of .74. Its capacity to identify those who do have MDD or those that do not have MDD, TPP, was .47. This symptom obtained a PPP index of .59 and an NPP of .40. The third symptom, “I do not have any friends,” reflected a d index of .77. This symptom also showed a TPP index of .50, and the probability of having MDD given the presence of the symptom was .68, while the probabil- ity of not having MDD given the absence of the symptom was identified to be .44. The other two symptoms with relatively high diagnostic efficiency indexes were “I have trouble sleeping every night” (d = .79) and “Things bother me all the time” (d = .83). Their capacities to identify those who have and those who do not have MDD, TPP, were found to be .50 and .48, respec- tively. Their probability indexes of having MDD given the presence of the symptoms, PPP, were .71 and .69, respectively. Finally, their respective PPN indexes where .44 and .43.
On the other hand, the presence of thoughts such as “Nobody really loves me,” “I am bad all the time,” “I hate myself,” “I cannot make up my mind about things,” and “Nothing is fun at all” obtained very high PPP indexes, but their prevalence rates were between 1% and 7%, thus denoting a limited diagnostic utility. We could also notice that the above symptoms also obtained higher percentages of TP and lower percentages of FN. It is worth mentioning that although some of the symptoms that are usually described in the literature as more characteristic of MDD were not included in the first 5 positions for the complete sample, they did fall within the first 10 symptoms with better diagnostic efficiency indexes. They were, in hierarchical order: “I feel like crying every day,” “I can never be as good as other kids,” “I am sad all the time,” “Nothing will ever work for me,” and “Most days I do not feel like eating.”
The diagnostic efficiency of the symptoms was also evaluated to assess gender differences. The results show that the first five symptoms with better diagnostic efficiency indexes for the complete sample were very similar to those obtained by gender. However, in females, two of the symptoms were found to differ, while in the case for males, the symptoms were basically the same but in a different order. Specifically for females, the symptom with the highest diagnostic efficiency was “I do not have any friends,” with a d index of .60. Its capacity to correctly identify those with or without MDD was found to be .47, the probability of having MDD given the pres- ence of the symptom was .69, and the probability of not having the disorder given its absence was .34.
244 Hispanic Journal of Behavioral Sciences 32(2)
The symptom that revealed the second highest diagnostic efficiency for females was “I have to push myself all the time to do my schoolwork” (see Table 2), with d = .65. This symptom obtained the highest TPP index, at .50. The probability of having MDD given the presence of the symptom was .73, while the probability of not having the disorder given the absence of the symptom was .36. It is worth mentioning that this symptom obtained the highest prevalence in the female sample, with 40% of the females indicating having the symptom. The third most indicative symptom for females was “I worry about others’ aches and pains all the time,” with a d index of .66. Its total predictive power was .49, the probability of having MDD given the presence of the symptom was .75, and the probability of not having the dis- order given its absence was .36. The other two symptoms with better diagnostic efficiency indexes were “I feel like crying every day” (d = .76) and “Nothing will ever work for me” (d = .80). Their capacities to identify those with or without MDD were .43 for both. The probability of having MDD given the presence of the symptoms was .65 and .83, respectively, and the probability of not having MDD given the absence of the symptoms was .32 and .35, respectively. In addition, symptoms like “I can never be as good as other kids,” “I have trouble sleeping every night,” “Things bother me all the time,” “I am sad all the time,” and “I am tired all the time” obtained diagnos- tic efficiency indexes between .80 to .83, placing them in the first 10 positions for MDD. This is confirmed by the probability of having MDD given the presence of symptoms, which fluctuates between .63 and .69.
The items “Nobody really loves me,” “I never do what I am told,” and “I am bad all the time” obtained PPP indexes of 1.0 each, in spite of their low prevalence. This is important since it means that whenever they are present in females the disorder is present. Also it is observed that, for females, the percentage of false negatives is higher than males, at least when symptoms are considered to be present if rated as severe (score of 2). Therefore, the probabilities obtained for not having MDD given the absence of the symptoms NPP were 20% lower in comparison with the male participants.
The results obtained for male participants show that the symptom with the highest diagnostic efficiency was “I have trouble sleeping every night,” with a d index of .74 (see Table 3). It also obtained the highest TPP (.59) and the probability of MDD given its presence was found to be .75. The probability of not having MDD given the absence of the symptoms was .55.
The symptom with the second highest level of diagnostic efficiency for the male sample was “I do very badly in subjects I used to be good in,” with a d index of .78. The total predictive power of this symptom was .55, the probability of having MDD given the presence of the symptom was .59, and
245
T a b
le 2
. Pr
ed ic
ti ve
V al
ue s,
D ia
gn o st
ic E
ffi ci
en cy
, a nd
P er
ce nt
ag e
D is
tr ib
ut io
n o f th
e M
D D
p er
C D
I It
em s
fo r
th e
Su bs
am pl
e o f Fe
m al
e A
do le
sc en
ts
C D
I It
em s
Q (
% )
SE N
SP
E PP
P N
PP
T PP
d
T P
(% )
FP (
% )
FN (
% )
T N
( %
)
I am
s ad
a ll
th e
ti m
e 27
.2
6 .7
0 .6
3 .3
1 .4
0 0.
83
17 .1
10
.0
50 .0
22
.9 N
o th
in g w
il l e v
e r
w o
rk f
o r
m e
1 7
.2 1
.9 1
.8 3
.3 5
.4 3
0 .8
0
1 4 .5
2 .9
5 3 .6
2 9 .9
I do
e ve
ry th
in g
w ro
ng
3 .0
2 .9
6 .5
0 .3
1 .3
2 1.
09
1. 4
1. 4
66 .7
30
.4 N
o th
in g
is f un
a t
al l
6 .0
6 .9
6 .7
5 .3
3 .3
6 0.
97
4. 3
1. 4
62 .9
31
.4 I am
b ad
a ll
th e
ti m
e 4
.0 6
1. 00
1.
00
.3 4
.3 7
0. 94
4.
3 0
62 .9
32
.9 I am
s ur
e th
at t
er ri
bl e
th in
gs w
ill h
ap pe
n to
m e
7 .0
6 .9
1 .6
0 .3
1 .3
3 1.
02
4. 3
2. 9
63 .8
29
.0 I ha
te m
ys el
f 7
.0 9
.9 6
.8 0
.3 4
.3 7
0. 94
5.
7 1.
4 61
.4
31 .4
A ll
ba d
th in
gs a
re m
y fa
ul t
13
.1 3
.8 7
.6 7
.3 3
.3 7
0. 93
8.
6 4.
3 58
.6
28 .6
I w
an t
to k
ill m
ys el
f 6
.0 4
.9 1
.5 0
.3 2
.3 3
1. 08
2.
9 2.
9 64
.3
30 .0
I fe
e l li ke
c ry
in g e
ve ry
d ay
3 3
.3 2
.6 5
.6 5
.3 2
.4 3
0 .7
6
2 1 .4
1 1 .4
4 5 .7
2 1 .4
T hi
ng s
bo th
er m
e al
l t he
t im
e 27
.2
6 .7
0 .6
3 .3
1 .4
0 0.
83
17 .1
10
.0
50 .0
22
.9 I do
n o t
w an
t to
b e
w it h
pe o pl
e at
a ll
9 .0
6 .8
7 .5
0 .3
1 .3
3 1.
06
4. 3
4. 3
62 .9
28
.6 I ca
nn o t
m ak
e up
m y
m in
d ab
o ut
t hi
ng s
9 .0
9 .9
1 .6
7 .3
3 .3
6 0.
97
5. 7
2. 9
61 .4
30
.0 I lo
o k
ug ly
20
.1
9 .7
7 .6
4 .3
1 .3
8 0.
85
13 .0
7.
2 55
.1
24 .6
I h
a v e t
o p
u sh
m ys
e lf
a ll t
h e t
im e t
o d
o
3 7
.4 0
.7 0
.7 3
.3 6
.5 0
0 .6
5
2 7 .1
1 0 .0
4 0 .0
2 2 .9
m
y sc
h o
o lw
o rk
I ha
ve t
ro ub
le s
le ep
in g
ev er
y ni
gh t/
23
.2
3 .7
8 .6
9 .3
3 .4
1 0.
82
15 .7
7.
1 51
.4
25 .7
sl
ee p
al l t
he t
im e
I am
t ir
ed a
ll th
e ti m
e 27
.2
6 .7
0 .6
3 .3
1 .4
0 0.
83
17 .1
10
.0
50 .0
22
.9 M
o st
d ay
s I d
o n
o t
fe el
li ke
e at
in g/
I e at
a ll
th e
ti m
e 23
.2
1 .7
4 .6
3 .3
2 .3
9 0.
87
14 .3
8.
6 52
.9
24 .3
I w
o rr
y a b
o u
t o
th e rs
’ a ch
e s
a n
d p
a in
s 3 5
.3 8
.7 3
.7 5
.3 6
.4 9
0 .6
6
2 6 .1
8 .7
4 2 .0
2 3 .2
a ll t
h e t
im e
I fe
el a
lo ne
a ll
th e
ti m
e 19
.1
7 .7
8 .6
2 .3
2 .3
7 0.
91
11 .4
7.
1 55
.7
25 .7
I ne
ve r
ha ve
f un
a t
sc ho
o l
14
.1 3
.8 3
.6 0
.3 2
.3 6
0. 95
8.
6 5.
7 58
.6
27 .1
(c on
tin ue
d)
T a b
le 2
. ( co
n ti
n u
e d
)
C D
I It
em s
Q (
% )
SE N
SP
E PP
P N
PP
T PP
d
T P
(% )
FP (
% )
FN (
% )
T N
( %
)
I d
o n
o t
h av
e a
n y
fr ie
n d
s 3 7
.3 8
.6 5
.6 9
.3 4
.4 7
0 .6
0
2 5 .7
1 1 .4
4 1 .4
2 1 .4
I do
v er
y ba
dl y
in s
ub je
ct s
I us
ed t
o b
e go
o d
in
22
.2 2
.7 8
.6 7
.3 3
.4 1
0. 85
14
.5
7. 2
52 .2
26
.1 I ca
n ne
ve r
be a
s go
o d
as o
th er
k id
s 29
.2
8 .7
0 .6
5 .3
2 .4
1 0.
80
18 .6
10
.0
48 .6
22
.9 N
o bo
dy r
ea lly
lo ve
s m
e 1
.0 2
1. 00
1.
00
.3 2
.3 3
0. 98
1.
4 0
66 .7
31
.9 I ne
ve r
do w
ha t
I am
t o ld
3
.0 4
1. 00
1.
00
.3 3
.3 5
0. 96
2.
9 0
65 .2
31
.9 I ge
t in
to f ig
ht s
al l t
he t
im e
3 .0
0 .9
1 .0
0 .3
1 .3
0 1.
41
0 2.
9 67
.1
30 .0
N o te
: M D
D =
m aj
o r
de pr
es si
o n
di so
rd er
; C D
I =
C hi
ld re
n’ s
D ep
re ss
io n
In ve
nt o ry
; Q =
p re
va le
nc e
o f sy
m pt
o m
s; SE
N =
s en
si ti vi
ty ; S
PE =
s pe
ci fic
it y;
PP
P =
p o si
ti ve
p re
di ct
iv e
po w
er ; N
PP =
n eg
at iv
e pr
ed ic
ti ve
p o w
er ; T
PP =
t o ta
l p re
di ct
iv e
po w
er , d
= d
ia gn
o st
ic e
ffi ci
en cy
in de
x; T
P =
t ru
e po
si ti ve
s; FP
= f al
se p
o si
ti ve
s; FN
= f al
se n
eg at
iv es
; T N
= t
ru e
ne ga
ti ve
s. R
o w
s in
b o ld
d en
o te
it em
s th
at o
bt ai
ne d
th e
be st
d ia
gn o st
ic e
ffi ci
en cy
w it h
hi gh
v al
ue s
o n
d an
d T
PP in
di ce
s.
246
247
T a b
le 3
. Pr
ed ic
ti ve
V al
ue s,
D ia
gn o st
ic E
ffi ci
en cy
, a nd
P er
ce nt
ag e
D is
tr ib
ut io
n o f th
e M
D D
p er
C D
I It
em s
fo r
th e
Su bs
am pl
e o f M
al e
A do
le sc
en ts
C D
I It
em s
Q (
% )
SE N
SP
E PP
P N
PP
T PP
d
T P
(% )
FP (
% )
FN (
% )
T N
( %
)
I am
s ad
a ll
th e
ti m
e 7
.1 3
1. 00
1.
00
.5 4
.5 7
0. 87
6.
7 0
43 .3
50
.0 N
o th
in g
w ill
e ve
r w
o rk
fo r
m e
12
.1 3
0. 90
.5
7 .5
1 .5
2 0.
96
6. 7
5. 0
43 .3
45
.0 I do
e ve
ry th
in g
w ro
ng
0 .0
0 1.
00
.0 0
.5 0
.5 0
—
0 0
50 .0
50
.0 N
o th
in g
is f un
a t
al l
0 .0
0 1.
00
.0 0
.5 0
.5 0
—
0 0
50 .0
50
.0 I am
b ad
a ll
th e
ti m
e 7
.1 0
0. 97
.7
5 .5
2 .5
3 0.
93
5. 0
1. 7
45 .0
48
.3 I am
s ur
e th
at t
er ri
bl e
th in
gs w
ill h
ap pe
n to
m e
3 .0
0 0.
93
.0 0
.4 8
.4 7
1. 41
0
3. 3
50 .0
46
.7 I ha
te m
ys el
f 2
.0 3
1. 00
1.
00
.5 1
.5 2
0. 97
1.
7 0
48 .3
50
.0 A
ll ba
d th
in gs
a re
m y
fa ul
t 5
.0 7
0. 97
.6
7 .5
1 .5
2 0.
99
3. 3
1. 7
46 .7
48
.3 I w
an t
to k
ill m
ys el
f 5
.0 7
0. 97
.6
7 .5
1 .5
2 0.
99
3. 3
1. 7
46 .7
48
.3 I fe
el li
ke c
ry in
g ev
er y
da y
3 .0
7 1.
00 1.
00
.5 2
.5 3
0. 93
3.
3 0
46 .7
50
.0 T
h in
g s
b o
th e r
m e a
ll t
h e t
im e
1 2
.2 0
0 .9
7
.8 6
.5 5
.5 8
0 .8
1
1 0 .0
1 .7
4 0 .0
4 8 .3
I do
n o t
w an
t to
b e
w it h
pe o pl
e at
a ll
3 .0
0 0.
93
.0 0
.4 8
.4 7
1. 41
0
3. 3
50 .0
46
.7 I ca
nn o t
m ak
e up
m y
m in
d ab
o ut
t hi
ng s
5 .1
0 1.
00 1.
00
.5 3
.5 5
0. 92
5.
0 0
45 .0
50
.0 I lo
o k
ug ly
13
.1
7 0.
90
.6 3
.5 2
.5 3
0. 91
8.
3 5.
0 41
.7
45 .0
I h
av e t
o p
u sh
m ys
e lf
a ll t
h e t
im e
4 2
.3 7
0 .5
2
.4 4
.4 4
.4 4
0 .8
4
1 8 .6
2 3 .7
3 2 .2
2 5 .4
to
d o
m y
sc h
o o
lw o
rk I
h av
e t
ro u
b le
s le
e p
in g e
ve ry
n ig
h t/
2 0
.3 0
0 .9
0
.7 5
.5 5
.5 9
0 .7
4
1 5 .3
5 .1
3 5 .6
4 4 .1
sl
e e p
a ll t
h e t
im e
I am
t ir
ed a
ll th
e ti m
e 3
.0 3
0. 97
.5
0 .5
0 .5
0 1.
09
1. 7
1. 7
48 .3
48
.3 M
o st
d ay
s I do
n o t
fe el
li ke
e at
in g/
I ea
t al
l t he
t im
e 18
.2
0 0.
83
.5 5
.5 1
.5 2
0. 92
10
.0
8. 3
40 .0
41
.7 I
w o
rr y
a b
o u
t o
th e rs
’ a ch
e s
a n
d p
a in
s a ll
3 2
.3 3
0 .7
0
.5 3
.5 1
.5 2
0 .8
2
1 6 .7
1 5 .0
3 3 .3
3 5 .0
th
e t
im e
I fe
el a
lo ne
a ll
th e
ti m
e 8
.1 0
0. 93
.6
0 .5
1 .5
2 0.
98
5. 0
3. 3
45 .0
46
.7 I ne
ve r
ha ve
f un
a t
sc ho
o l
18
.1 7
0. 80
.4
6 .4
9 .4
8 0.
99
8. 3
10 .0
41
.7
40 .0
(c on
tin ue
d)
T a b
le 3
. (c
o n
ti n
u e d
)
C D
I It
em s
Q (
% )
SE N
SP
E PP
P N
PP
T PP
d
T P
(% )
FP (
% )
FN (
% )
T N
( %
)
I do
n o t
ha ve
a ny
f ri
en ds
13
.1
7 0.
90
.6 3
.5 2
.5 3
0. 91
8.
3 5.
0 41
.7
45 .0
I d
o v
e ry
b ad
ly in
s u
b je
ct s
I u
se d
t o
b e g
o o
d in
2 8
.3 3
0 .7
7
.5 9
.5 4
.5 5
0 .7
8
1 6 .7
1 1 .7
3 3 .3
3 8 .3
I ca
n ne
ve r
be a
s go
o d
as o
th er
k id
s 17
.2
0 0.
87
.6 0
.5 2
.5 3
0. 89
10
.0
6. 7
40 .0
43
.3 N
o bo
dy r
ea lly
lo ve
s m
e 0
.0 0
1. 00
.0
0 .5
0 .5
0 —
0
0 50
.0
50 .0
I ne
ve r
do w
ha t
I am
t o ld
7
.0 7
0. 93
.5
0 .5
0 .5
0 1.
06
3. 3
3. 3
46 .7
46
.7 I ge
t in
to f ig
ht s
al l t
he t
im e
2 .0
0 0.
97
.0 0
.4 9
.4 8
1. 41
0
1. 7
50 .0
48
.3
N o te
: M D
D =
m aj
o r
de pr
es si
o n
di so
rd er
; C D
I =
C hi
ld re
n’ s
D ep
re ss
io n
In ve
nt o ry
; Q =
p re
va le
nc e
o f sy
m pt
o m
s; SE
N =
s en
si ti vi
ty ; S
PE =
s pe
ci fic
it y;
PP
P =
p o si
ti ve
p re
di ct
iv e
po w
er ; N
PP =
n eg
at iv
e pr
ed ic
ti ve
p o w
er ; T
PP =
t o ta
l p re
di ct
iv e
po w
er , d
= d
ia gn
o st
ic e
ffi ci
en cy
in de
x; T
P =
t ru
e po
si ti ve
s; FP
= f al
se p
o si
ti ve
s; FN
= f al
se n
eg at
iv es
; T N
= t
ru e
ne ga
ti ve
s. R
o w
s in
b o ld
d en
o te
it em
s th
at o
bt ai
ne d
th e
be st
d ia
gn o st
ic e
ffi ci
en cy
w it h
hi gh
v al
ue s
o n
d an
d T
PP in
di ce
s.
248
Rivera-Medina et al. 249
the probability of not having the disorder given its absence was .54. The third most indicative symptom reported by males was “Things bother me all the time,” with a d index of .81. Its capacity to correctly identify those with or without MDD was found to be .58, the probability of MDD given the pres- ence of the symptom was .86, and the probability of not having the disorder given its absence was .55. The other two symptoms with better/relatively high diagnostic efficiency indexes were “I worry about others’ aches and pains all the time” (d = .82) and “I have to push myself all the time to do my schoolwork” (d = .84). Their capacities to identify those with or without MDD are .52 and .44, respectively. The probability of having MDD given the pres- ence of the symptoms was .53 and .44, respectively, and the probability of not having MDD given the absence of the symptoms was .51 and .44, respec- tively. The next 5 symptoms to complete the top 10 items for males were “I am sad all the time,” “I can never be as good as other kids,” “I am ugly,” “I do not have any friends,” and “I cannot make up my mind about things.”
The symptoms “I hate myself” and “I fell like crying every day,” in spite of their low prevalence, obtained a PPP index of 1.00, indicating the presence of MDD each time one of these symptoms was present. In the male sample, there were some symptoms for which PPP indexes were so low or nonexis- tent that they could be categorized as exclusion criteria or indicators of another disorder. These symptoms were “I do everything wrong,” “Nothing is fun at all,” “I am sure terrible things will happen to me,” “I do not want to be with people,” “Nobody really loves me,” and “I get into fights all the time.” This phenomenon was true for females in the case of the symptom “I get into fights all the time.”
Discussion The results support the assertion in the DSM-IV-TR (APA, 2000) that MDD may be manifested differently in different cultures. It also provides evidence that the symptoms most characteristic of MDD may not necessarily be the most efficient indicators of its presence. In the Puerto Rican sample of ado- lescents from an outpatient clinic, the five symptoms with the highest predictive validity for MDD differ from the ones reported by Roberts et al. (1991), who found that the symptoms with the highest predictive validity in their Anglo European sample for the BDI were “feel sad,” “blame myself,” “cry,” and “no appetite,” while “felt depressed,” “poor appetite,” “felt sad,” and “could not get going” were found to be the best predictors from the CES-D. The findings in this study also differ from those reported in other studies that evaluate the frequency of the symptoms of MDD that obtained
250 Hispanic Journal of Behavioral Sciences 32(2)
the highest percentages in a sample of children from the United States (Mitchell et al., 1988; Ryan et al., 1987). The results also contrast those found by Laurent et al. (1993), whose evaluations of the conditional probabilities of the CDI items for MDD showed that the items with the highest predictive validity were feeling unloved, anhedonia, excessive guilt, and depressed mood. Also in contrast to our findings, Laurent et al. (1993) found that “aca- demic capacities” and “feeling worried about others” turned out to be better predictors for exclusion criteria.
Our results are similar to those reported by Crockett et al. (2005). Although not evaluating the predictive validity of symptoms for MDD, these investiga- tors found that a four-factor structure established for the CES-D scale was not supported for Puerto Rican and Cuban youth. Puerto Ricans in this study showed a different set of factors in which negative affect and somatic symp- toms loaded on one factor, suggesting a co-occurrence of these symptoms and a blurring between affective and somatic symptoms. The symptoms that appeared on this factor were “could not shake the blues,” “felt depressed,” “thought life a failure,” “felt tearful,” “felt lonely,” “frequent crying,” “felt sad,” “bothered by things,” “didn’t feel like eating,” “trouble sleeping,” and “talked less than usual.” They also found that individual CES-D items did not operate in the same way, especially for Cuban and Puerto Rican youth, con- cluding that cross-ethnic comparisons involving Latino youth could potentially yield inaccurate results. Crockett et al. (2005) suggest that it is possible that Cubans and Puerto Ricans have somewhat different concepts of depression than Anglo Americans, leading them to experience or report dif- ferent symptoms patterns. These authors point out the potential risk of using a screening measure that does not have measurement equivalence across ethnic groups, especially Latino groups. Use of scales without adequate metric equivalence could result in a misclassification of Puerto Rican and Cuban youth, who may not share the same depression symptomatology as Anglo European and Mexican American youth.
In terms of the analysis by gender and the comparison with the results for the complete sample, for both females and males, the symptoms presented as the most accurate predictors of MDD were basically the same, differing only in hierarchical order. However, when evaluating the symptoms separately for each subsample, especially when considering the first 10 rather than simply the first five, the differences by gender are more evident, thus denoting that a differential manifestation of the disorder. For example, the male subsample showed five anhedonia symptoms, while females showed three. Males showed three symptoms denoting concern about interpersonal relationship, whereas females showed only one. The females had more negative moods
Rivera-Medina et al. 251
than males, and even when both showed just one symptom related to nega- tive self-esteem, this symptom denoted hopelessness for females and negative physical image for males.
These findings are significant since they evidence not only that the depres- sive symptoms may differ cross-culturally but that they vary based on gender as well. An important implication of our finding is the possibility of assign- ing a priority value to symptoms according to their diagnostic efficiency. The diagnostic criteria in the DSM-III-R are given equal weight within a particu- lar category, in spite of the growing evidence suggesting that symptoms vary considerably in their respective diagnostic efficiency (Laurent et al., 1993). Unfortunately, this tendency has not changed for the latest version of the DSM-IV-TR (2000), leading to questioning the practicability of the Univer- salist framework that has governed nosology to date (Alegría & McGuire, 2003). The social, cultural, and contextual factors mediate the relationship between disorder and symptoms endorsement for psychiatric disorders. Therefore, not considering sociocultural factors in the diagnostic process may have different responses to a diagnostic battery because the pattern of responses to the screener items may vary as a function of their sociodemo- graphic characteristics and not because of the presence or absence of the disorder itself. It is important to acknowledge that, given the inclusion and exclusion criteria for the study, our sample does not represent the most severe conditions of MDD. The participants in the study were not evaluated for comorbid conditions, and in fact, having some previous diagnosis of a comor- bid condition, such as conduct disorder, was an exclusion criterion. In addition, the participants were excluded if serious imminent suicide risk was present, therefore affecting the prevalence of the items related with suicide ideation in the CDI. However, this does not diminish the validity of our find- ings. The conditional probability analysis and the fact that the symptoms found as better predictors of MDD could be different in different cultures or in different severity levels of the disorder. These findings suggest the need for further analysis with similar populations that take into consideration the methodological limitations of our work.
Finally, considering the results of this study, it seems reasonable to evalu- ate the possibility of an item reduction of the CDI where items could be placed in hierarchical order. A scale reduction with the top 10 items with the highest diagnostic efficiency could possibly be used as a first-stage screening measure rather than a smaller scale with the five highest items. This strategy takes into account that the top five items for the complete sample were more similar to the top five items found in the male subsample, in which case the
252 Hispanic Journal of Behavioral Sciences 32(2)
scale would be gender biased. The top 10 would include six items that were identified as having a good diagnostic efficiency for the complete sample, two items found only for the females, and two items found only for the males (see Table 4). The shorter scale suggested would allow screening of more adolescents in a short amount of time and identification of predictive symp- toms for MDD in the early stages of the condition. The shorter version would also enable the easy and less time-consuming evaluation of depression symp- tom in nonclinical contexts (schools, hospitals, community centers), thus making it possible to reach more adolescents and work with them on a pre- ventive level or to detect potential cases of MDD and intervene at earlier stages. A strategy based on item analyses of symptoms allows for a useful way to inform diagnostic criteria based on the conditional probability of dis- orders given the presence of symptoms.
Acknowledgments
The authors are grateful to Dr. Rafael Ramírez for his statistical advice, Amy Leah Fontenot and Patricia Pujals for their editorial recommendations, Amarilys Galloza for her data analysis contribution, and the research and administrative staff of the Institute for Psychological Research at the University of Puerto Rico, Río Piedras Campus.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the authorship and/or publication of this article.
Table 4. Shorter Version of the CDI Suggested for Screening
CDI Items
I worry about others’ aches and pains all the time I have to push myself all the time to do my schoolwork I do not have any friends I have trouble sleeping every night/sleep all the time Things bother me all the time I can never be as good as other kids I feel like crying every day I do very badly in subjects I used to be good in Nothing will ever work for me I look ugly
Note: CDI = Children’s Depression Inventory.
Rivera-Medina et al. 253
Funding
The authors disclosed receipt of the following financial support for the research and/ or authorship of this article:
This research was supported by NIH Research Grant 5R24-MH-49368-12 funded by the National Institute of Mental Health and by the Division of Mental Disorders, Behavioral Research & AIDS to Guillermo Bernal. The content is solely the respon- sibility of the authors and does not represent the official views of the NIMH or the National Institute of Health. The research also received support from the Institutional Funds for Research from the Dean of Graduate Studies and Research at the University of Puerto Rico, Rio Piedras Campus.
References
Abdel-Kahalec, A. M., & Soliman, H. H. (1999). A cross cultural evaluation of depression in children in Egypt, Kuwait and the United States. Psychological Reports, 85, 973-980.
Alatorre-Alva, S., & De los Reyes, R. (1999). Psychosocial stress, internalized symp- toms, and the academic achievement of Hispanic adolescents. Journal of Adoles- cents Research, 14, 343-358.
Alegría, M., & McGuire, T. (2003). Rethinking a universal framework in the psychi- atric symptoms-disorder relationship. Journal of Health and Social Behavior, 44, 257-273.
Ambrosini, P. J., Metz, C., Bianchi, M., Rabinovich, H., & Undie, A. (1991). Concur- rent validity and psychometric properties of the Beck Depression Inventory in outpatient adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 30, 51-57.
American Academy of Pediatrics. (2000). Suicide and suicide attempts in adolescent. Pediatrics, 105, 871-874.
American Psychiatric Association. (1987). Diagnostic and statistical manual of men- tal disorders (3rd ed., revised). Washington, DC: Author.
American Psychiatric Association. (1994). Diagnostic and statistical manual of men- tal disorders (4th ed.). Washington, DC: Author.
American Psychiatric Association. (2000). Diagnostic and statistical manual of men- tal disorders (text revision). Washington, DC: Author.
Barrera, M., & Garrison-Jones, C. (1998). Properties of Beck Depression Inventory as a screening instrument for adolescent depression. Journal of Abnormal Child Psychology, 16, 263-273.
Bennett, D. S. (1994). Depression among children with chronic medical problems: A meta analysis. Pediatric Psychology, 19, 149-169.
254 Hispanic Journal of Behavioral Sciences 32(2)
Bernal, G., Rosselló, J., & Martínez, A. (1997). El Inventario de depresion para niños y niñas: Propiedades psicometricas en dos muestras puertorriqueñas. Psicologia Contemporanea, (The Child Depression Inventory: Psychometric properties in two Puerto Rican Samples) 4(1), 12-23.
Bird, H. R., Canino, G., Rubio-Stipec, M., Gould, M. S., Ribera, J., Sessman, M., et al. (1988). Estimate of the prevalence of childhood maladjustments in a community survey in Puerto Rico. Archives of General Psychiatry, 45, 1120-1126.
Bird, H. R., Gould, M. S., Rubio-Stipec, M., Staghezza, B. M., & Canino, G. (1991). Screening for childhood psychopathology in the community using the child behav- ior checklist. Journal of the American Academy of Child & Adolescent Psychiatry, 30, 116-122.
Bravo, M. (2003). Instrument development, cultural adaptations for ethnic minority research. In G. Bernal, J. E. Trimble, A. K. Burlew, & F. T. Leong (Eds.), Handbook of racial and ethnic minority psychology (pp. 220-236). Thousand Oaks, CA: Sage.
Bravo, M., Woodbury-Fariña, M., Canino, G., & Rubio-Stipec, M. (1993). The Spanish translation and cultural adaptation of the Diagnostic Interview Schedule for Chil- dren (DISC) in Puerto Rico. Culture, Medicine and Psychiatry, 17, 329-344.
Canino, G., Shrout, P., Rubio-Stipec, M., Bird, H. R., Bravo, M., Ramírez, R., et al. (2004). DSM-IV rates of child and adolescent disorder in Puerto Rico: Prevalence, correlates, service use and the effects of impairment. Archives of General Psychiatry, 61, 85-93.
Chan, D. W. (1997). Depressive symptoms and perceived competence among Chinese secondary school students in Hong Kong. Journal of Youth and Adolescence, 26, 303-319.
Charman, T., & Pervova, I. (1996). Self-reported depressed mood in Russian and U.K. schoolchildren. A research note. Journal of Child Psychology and Psychiatry and Allied Disciplines, 37, 879-883.
Compas, B. E., Ey, S., & Grant, K. E. (1993). Taxonomy, assessment, and diagnosis of depression during adolescence. Psychological Bulletin, 114, 323-344.
Crockett, L., Randball, B., Shen, Y. L., Russell, S., & Driscoll, A. (2005). Measure- ment equivalence of the Center for Epidemiological Studies Depression Scales for Latino and Anglo adolescents: A national study. Journal of Consulting and Clinical Psychology, 73, 47-58.
Del Barrio, V., Moreno-Rosset, C., López-Martínez, R., & Olmedo, M. (1997). Anxiety, depression and personality structure. Personality and Individual Differences, 23, 327-335.
Faraone, S. V., Biederman, J., Sprich-Buckminster, S., Chen, W., & Tsuang, M. T. (1993). Efficiency of diagnostic criteria for attention deficit disorder: Toward an empirical approach to designing and validating diagnostic algorithms. Journal of the American Academy of Child & Adolescent Psychiatry, 32, 166-174.
Rivera-Medina et al. 255
Fristad, M. A., Emery, B. L., & Beck, S. J. (1997). Use and abuse of the Children’s Depression Inventory (CDI). Annals of Clinical Psychiatry, 65, 699-702.
Fristad, M. A., Weller, R. A., Weller, E. B., Teare, M., & Preskorn, S. H. (1991). Com- parison of the parent and child version of the Children’s Depression Inventory (CDI). Annals of Clinical Psychiatry, 3, 341-346.
Hammen, C., & Rudolph, K. D. (1996). Childhood depression. In E. J. Mash & R. A. Barkley (Eds.), Child psychopathology (pp.153-195). New York: Guilford.
Harrington, R., Fudge, H., Rutter, M., Picklels, A., & Hill, J. (1991). Adult outcomes of childhood and adolescent depression: II. Archives of General Psychiatry, 43, 255-262.
Hayward, C., Gotlib, I. H., Schraedley, P. K., & Litt, I. F. (1999). Ethnic differences in the association between pubertal status and symptoms of depression in adolescent girls. Journal of Adolescents Health, 25, 143-149.
Kandel, D., & Davies, M. (1986). Adult sequel of adolescent depressive symptoms. Archives of General Psychiatry, 43, 255-262.
Kashani, J. H., Sherman, D. D., Parker, D. R., & Reid, J. C. (1990). Utility of the Beck Depression Inventory with clinic-referred adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 29, 278-282.
Knight, G., Virdin, L., Ocampo, K., & Roosa, M. (1994). An examination of the cross-ethnic equivalence of measures of negative life events and mental health and mental health among Hispanic and Anglo American children. American Journal of Community Psychology, 22, 767-783.
Kovacs, M. (1983). Children’s Depression Inventory (CDI). Pittsburgh, PA: Univer- sity of Pittsburgh School of Medicine.
Kovacs, M. (1992). Children’s Depression Inventory (CDI) manual. New York: Multi-Health Systems.
Kovacs, M., Goldston, D., & Gatsonis, C. (1993). Suicidal behaviors and childhood- onset depressive disorder: A longitudinal investigation. Journal of the American Academy of Child & Adolescent Psychiatry, 32, 8-20.
Kovacs, M., Palauskas, S., Gatosnis, C., & Richards, C. (1988). Depressive disorder in childhood: III. A longitudinal study of comorbidity with and risk for conduct disorders. Journal of Affective Disorders, 15, 205-217.
Landau, S., Milich, R., & Widiger, T. A. (1991). Conditional probabilities of child interview symptoms in the diagnosis of attention deficit disorder. Journal of Child Psychology and Psychiatry, 32, 501-513.
Laurent, J., Landau, S., & Stark, K. D. (1993). Conditional probabilities in the diag- nosis of depressive and anxiety disorders in children. School Psychology Review, 22, 98-114.
Lewinsohn, P. M., Clarke, G. N., Seeley, J. R., & Rohde, P. (1994). Major depression in community adolescents: Age at onset, episode duration and time to recurrence.
256 Hispanic Journal of Behavioral Sciences 32(2)
Journal of the American Academy of Child & Adolescent Psychiatry, 33, 809-818. Lewinsohn, P. M., Rohde, P., & Seeley, J. R. (1998). Major depressive disorder in older
adolescents: Prevalence, risk factors and clinical implications. Clinical Psychology Review, 18, 765-794.
Lucas, C. P., Zhang, H., Fisher, P. W., Shaffer, D., Regier, D. A., Narrow, W. E., et al. (2001). The DISC Predictive Scales (DPS): Efficiently screening for diagnoses. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 443-449.
Llabre, M. M., & Hadi, F. (1997). Social support and psychological distress in Kuwaiti boys and girls exposed to the Gulf crisis. Journal of Child and Clinical Psychology, 26, 247-255.
Marsella, A. J., & Yamada, A. M. (2007). Culture and psychopathology: Foundations, issues and directions. In S. Kitayama & D. Cohen (Eds.), Handbook of cultural psychology (pp. 797-820). New York: Guilford.
Martin, J. A., Smith, B. L., Matthews, T. J., & Ventura, S. J. (1999). Births and deaths: Preliminary data for 1998. National Vital Statistic Reports, 47(25), 1-48.
Mausner, J., & Kramer, S. (1985). Epidemiology: An introductory text. Philadelphia: Saunders.
Mitchell, J., McCauley, E., Burke, P. M., & Moss, S. J. (1988). Phenomenology of depression in children and adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 27, 12-20.
Nelson, W. M., Politano, P. M., Finch, A. J., Wendel, N., & Mayhall, C. (1987). Children’s Depression Inventory: Normative data and utility with emotionally disturbed children. Journal of the American Academy of Child & Adolescent Psy- chiatry, 26, 43-48.
Nurcombe, B., Seifer, R., Scioli, A., Tramontana, M., Lexington, W., & Beauchesne, H. (1989). Is major depressive disorder in adolescence a distinct diagnostic entity? Journal of the American Academy of Child & Adolescent Psychiatry, 28, 333-342.
Pellham, W. E., Gnagy, E. M., Greenslade, K. E., & Milich, R. (1992). Teacher rat- ings of DSM-III-R symptoms for the disruptive behaviors disorders. Journal of the American Academy of Child & Adolescent Psychiatry, 31, 210-218.
Pirkis, J. E., Irwin, C. E., Brindis, C. D., Sawyer, M. G., Friestad, C., Biehl, M., et al. (2003). Receipt of psychological or emotional counseling by suicidal adolescents. Pediatrics, 111, 388-393.
Ponterotto, J. G., Pace, T. M., & Kavan, M. G. (1989). A counselor’s guide to the assessment of depression. Journal of Counseling and Development, 67, 301-309.
Puura, K., Almqvist, F., Tamminen, T., Piha, J., Kumpulainen, K., Räsänen, E., et al. (1998). Children with symptoms of depression: What do the adults see? Journal of Child Psychology and Psychiatry, 39, 577-585.
Ribera, J., Canino, G., Rubio-Stipec, M., Bravo, M., Bauermeister, J. J., Alegría, M., et al. (1996). The Diagnostic Interview Schedule for Children (DISC-2.1) in Spanish:
Rivera-Medina et al. 257
Reliability in a Hispanic population. Journal of Child Psychology and Psychiatry and Allied Disciplines, 37, 195-204.
Rivera, C., Bernal, G., & Rosselló, J. (2005). The Children Depression Inventory (CDI) and the Beck Depression Inventory (BDI): Their validity as screening measures for major depression in a group of Puerto Rican adolescents. International Journal of Clinical and Health Psychology, 5, 485-498.
Rosselló, Bernal and Rivera-Medina (2008). “Randomized trial of CBT and IPT in indi- vidual and group format for depression in Puerto Rican adolescents”. Cultural Diver- sity and Ethnic Minority Psychology, 14, 3, 234-245.
Roberts, R., & Chen, Y. (1995). Depressive symptoms and suicidal ideation among Mexican origin and Anglo adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 34, 81-90.
Roberts, R., Lewinsohn, P. M., & Seeley, J. R. (1991). Screening for adolescent depression: A comparison of depression scales. Journal of the American Academy of Child & Adolescent Psychiatry, 30, 58-66.
Roberts, R., & Sobahn, M. (1992). Symptoms of depression in adolescence: A compari- son of Anglo, African and Hispanic Americans. Journal of Youth and Adolescence, 21, 639-651.
Rosselló, J., Guisasola, E., Ralat, S., Martínez, S., & Nieves, A. (1992). La evaluación de la depresión en un grupo de jóvenes puertorriqueños (The evaluation of depres- sion in a group of Puerto Rican youth). Revista Puertorriqueña de Psicología, 8, 155-162.
Rubio-Stipec, M., Shrout, P., Canino, G., Bird, H. R., Jensen, P., Dulcan, M. K., et al. (1996). Empirically defined symptom scales using the DISC 2.3. Journal of Abnor- mal Child Psychology, 24, 67-83.
Ryan, N. D., Puig-Antich, J., Ambrosini, P. J., Rabinovich, H., Robinson, D., Nelson, B., et al. (1987). The clinical picture of major depression in children and adoles- cents. Archives of General Psychiatry, 44, 854-861.
Saylor, C. F., Finch, A. J., Spirito, A., & Bennett, B. (1984). The Children’s Depres- sion Inventory: A systematic evaluation of psychometric properties. Journal of Consulting and Clinical Psychology, 52, 955-967.
Schwab-Stone, M., Shaffer, D., Dulcan, M. K., Jensen, P., Fisher, P. W., Bird, H. R., et al. (1996). Criterion validity of the NIMH Diagnostic Interview Schedule for Children Version 2.3 (DISC-2.3). Journal of the American Academy of Child & Adolescent Psychiatry, 35, 878-888.
Siegel, J. M., Aneshensel, C. S., Taub, B., Cantwell, D. P., & Driscoll, A. K. (1998). Ado- lescent depressed mood in a multiethnic sample. Journal of Youth and Adolescence, 4, 413-427.
Sitarenios, G., & Kovacs, M. (1999). Use of the Children’s Depression Inventory. In M. E. Maruish (Ed.), The use of psychological testing for treatment plan- ning and outcomes assessment (2nd ed., pp. 267-298). Mahwah, NJ: Lawrence Erlbaum.
258 Hispanic Journal of Behavioral Sciences 32(2)
Strober, M., Green, J., & Carlson, G. (1981). Utility of the Beck Depression Inventory with psychiatrically hospitalized adolescents. Journal of Consulting and Clinical Psychology, 49, 482-483.
Vandenberg, R. J., & Lance, C. E. (2000). A review and synthesis of the measurement invariance literature: Suggestions, practices and recommendations for organiza- tional research. Organizational Research Methods, 3, 4-69.
Weiss, B., Weisz, J. R., Politano, P. M., Carey, M., Nelson, W. M., & Finch, A. J. (1991). Developmental differences in the factor structure of the Children’s Depres- sion Inventory. Psychological Assessment, 3, 38-45.
Widiger, T. A., Hurt, S. W., Frances, A., Clarkin, J. F., & Gilmore, M. (1984). Diag- nostic efficiency and DSM-III. Archives of General Psychiatry, 41, 1005-1012.
Bios
Carmen L. Rivera-Medina received her doctorate in clinical psychology from Uni- versity of Puerto Rico, Río Piedras (UPR-RP). She is an assistant research scientist at the Institute for Psychological Research at UPR-RP and a part-time professor in the Department of Psychology, also at the UPR-RP. Her research and interests are the psy- chometric properties of the instruments and interventions used in Latinos in the mental health field. Specifically, for the past few years, she has focused on the study of the metric equivalence of instruments used to evaluate depressive symptoms in Latino ado- lescents and adults and the manifestation of major depression disorder in both groups.
Guillermo Bernal received his doctorate from the University of Massachusetts at Amherst. He is a professor of psychology in the Department of Psychology, University of Puerto Rico, Río Piedras, and is the director of the Institute of Psychological Research. His research, training, and practice interests are mental health treatments, interventions, and services responsive to ethnocultural groups, particularly adolescent depression, family processes, and dissemination of effective interventions. He is also interested in program development and evaluation for research training and mentoring.
Jeannette Rosselló received her doctorate from New York University. She was professor of psychology at the University of Puerto Rico, Río Piedras (UPR-RP), until her retire- ment in 2008. She is now in private practice and is affiliated with Institute for Psychological Research at UPR-RP. Her interests are in the treatment of depression in adolescents.
Eduardo Cumba-Aviles received his doctorate in clinical psychology from University of Puerto Rico, Río Piedras (UPR-RP). He is an assistant research scientist at the Insti- tute for Psychological Research at UPR-RP. His interests are in the manifestation of attention deficit disorder, depression and comorbid conditions as well as the interven- tions and the psychometric properties of the instruments to evaluate and treat those conditions.
