Contribution to the Validation of the Kutcher Adolescent Depression Scale ( KADS-6 ) in a Portuguese Population Contribuição para a Validação da Kutcher Adolescent Depression Scale ( KADS-6 ) para a População Portuguesa

The KADS-6 is a self-report assessment instrument known for its ease application in assessing depression in young people. This study aimed to contribute to the validation of the Portuguese version of this tool and analyze its psychometric characteristics in comparison with other self-report instruments for depression in adolescents, in Portugal. Two samples were collected, a non-clinical group of in school youth (n=773; 52.4% male) and a clinical sample (n=134 youth; 44% male). Comparisons also used the Beck Depression Inventory (BDI-II) and the Children’s Depression Inventory (CDI). Results: The factor analysis revealed the unidimensionality of the measure. The KADS-6 demonstrated good internal consistency (Cronbach’s alpha) in both clinical (.74) and school (.80) samples. Its convergent validity with the BDI-II and CDI was r=.44; r=.61, respectively in the clinical sample and r=.60; r=.57 in the school sample. The KADS-6, a short and time effi cient instrument, showed good psychometric characteristics in terms of internal consistency and convergent validity in comparison with the BDI-II and the CDI. Given its ease of use and scoring, the KADS-6 could be considered for use in both school and clinical settings when addressing adolescent depression.

Major depression is a common and impairing mental disorder, and globally is a leading source of disability, accounting for almost 11% of the global burden of disease and the third leading source of disability worldwide (Saluja et al., 2004;World Health Organization [WHO] & World Bank [WB], 2011).Its onset is commonly in the adolescent years and it leads to substantial short and long term morbidity including social, vocational/academic and interpersonal diffi culties, increases risk for other mental disorders, substance abuse and a variety of other illnesses (such as diabetes and cardiovascular disease) and is a key risk factor for suicide (Bertolote, Fleischmann, de Leo, & Wasserman, 2003;Goodwin, Kroenke, Hoven, & Spitzer, 2003;Gusmão, Xavier, Heitor, Bento, & Almeida, 2005;Kessler, Avenevoli, & Ries Merikangas, 2001).A number of safe and effective treatments for depression in young people exist but many depressed youth are still neither identifi ed, properly diagnosed nor receive appropriate care (Kessler et al., 2001;Leaf et al., 1996).It is thus essential that substantive attempts be made to identify and provide best evidence based treatments to young people with depression.
Identifi cation of depression in young people in school settings can be a diffi cult challenge, and a number of self-report tools have been developed to assist in that task (Cheung et al., 2010;Gray, Dubin-Rhodin, Weller, & Weller, 2009).While their psychometric properties vary somewhat their utility in both population and clinical settings is related to numerous factors including but not limited to: acceptable sensitivity and specifi city; ease of administration; acceptability to adolescents; ease of use; brevity of use; ease of scoring; availability in different languages; demonstrated applicability in different countries.
The KADS was developed to try to address most of these issues, to provide a brief, user-friendly, clinically useful and valid tool for assessment of adolescent depression.These characteristics enhance the probablilty of patient and clinician acceptance.This is the reason why the validation of this instrument in Portugal is important.We here report on the psychometric properties of the Kutcher Adolescent Depression Scale (KADS), a recommended population and clinical screening tool for depression in this age group (Jensen et al., 2007(Jensen et al., /2012) ) applied in a population of Portuguese adolescents.

The Kutcher Adolescent Depression Scale
The KADS was developed by adolescent mental health specialists, pediatricians and health researchers to assist in the identifi cation and clinical care of Depression in young people aged 12 to 22 (LeBlanc, Almudevar, Brooks, & Kutcher, 2002).The initial studies of KADS were conducted on population samples of students attending secondary school, in healthcare clinics and of adolescent subjects in clinical research projects related to the diagnosis and treatment of Major Depressive Disorder.
There are three different versions of the KADS: a 6-item, an 11-item, and a 16-item version.While all three versions of the KADS can be used, extensive experience with the scale has demonstrated that the KADS-6 is the version most commonly used in everyday clinical settings, both to screen for possible depression and to monitor the outcome of interventions in youth diagnosed with and treated for depression (personal communication: SK).The KADS-6 is designed for use in institutional settings (such as schools or primary health care units), and can be used as a screening tool to identify adolescents at risk for depression by health care providers (e.g.public health nurses, primary care physicians) and educators in the school setting (e.g.school psychologists, social workers, counselors, etc.) to help assess young people who are at risk for or have been identifi ed as possibly presenting a depressive disorder.KADS is a self-report measure intended to be completed independently by the youth after having been introduced to the scale by the health or human services provider using the KADS.The KADS is reading accessible at the sixth grade level, thus making it accessible to young people during the period in which depression is commonly diagnosed (DSM V).The KADS-6 is much briefer than other available adolescent depression scales and has been recommended for use in independent expert reviews, including the GLAD-PC Guidelines (U.S. and Canada; Jensen et al., 2007Jensen et al., /2012)).The KADS-6 has been translated into several different languages and is used globally (Kutcher & Chehil, 2007a).It is freely available at: www.teenmentalhealth.org.
The six items of the KADS-6 are rated using a fourpoint Likert scale (0-Hardly ever, to 3-All of the time).A total item's summed score of six or more is consistent with a diagnosis of Major Depressive Disorder and should trigger a more comprehensive assessment of the mental health status of the youth completing the scale.The last item of the KADS-6 addresses suicide risk.Youth scoring one or higher on this item should undergo a more detailed suicide risk assessment (LeBlanc et al., 2002).Further, the construction of the KADS-6 (with the use of the phrase "Hardly Ever" as point zero) allows the clinician to probe for suicidal ideation if there is suspicion of increased risk even if the respondent has not endorsed a positive score on this item.
In the validation study of the scale, LeBlanc et al. ( 2002) compared the validity of the diagnosis by the full 16-item instrument, and shorter versions (6 and 11-items), against a score higher than 15 with the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) and the DSM-IV TR diagnostic category of Major Depressive Episode (MDE) defi ned by clinical evaluation using the evaluation semi-structured interview instrument, the Mini International Neuropsychiatric Interview (MINI, version 4.4;Lecrubier et al., 1997).The sample consisted of 1712 adolescent students from the 7th to the 12th grade.Firstly, participants completed the BDI and 309 had scores above 15.Of these 309 were invited to undergo further evaluation, 161 accepted, and completed the KADS, the BDI (for a second time) and a diagnostic interview using the MINI for diagnosis of MDE.A receiver operating characteristic (ROC) analysis was performed to determine which KADS items most accurately identifi ed participants with MDE.The ROC curve analysis showed that the KADS-6 was as effective as the BDI.Used with a cutoff of six, the KADS-6 had rates of sensitivity and specifi city of 92% and 71%, respectively, a positive predictive value (PPV) of 10% for the total sample and 26% for the clinical sample and a negative predictive value (NPV) of 100% for the total sample and 99% for the clinical sample, values not reached by other assessment tools.
In a multicenter study (U.S. and Canada) Brooks, Krulewick and Kutcher (2003) obtained an internal consistency Cronbach's α of .80 for the KADS-6, .84 for the KADS-11 and .82for the KADS-16, demonstrating good psychometric properties of the KADS-6 in diagnosis and treatment outcome capabilities in the clinical setting.Thus, the KADS-6 has been validated in both population and clinical samples, making it a useful tool that can be applied across the continuum of care: from case identifi cation to measurement of the outcomes of interventions.

Other Self-Report Scales of Depression in Adolescents in the Portuguese Population
There are a number of other self-report scales addressing depression in adolescents.The Beck Depression Inventory (BDI-II) is a 21-item tool often used in screening for depression (Beck, 1973;Beck, Steer, & Brown, 1996;Gray et al., 2009;Oliveira-Brochado & Oliveira-Brochado, 2008;Osman, Barrios, Gutierrez, Williams, & Bailey, 2008).Its scores are often expressed as ranges apparently related to degrees of depression severity (mild, moderate, severe) but the utility and validity of this differentiation has not been substantively demonstrated.A validation study of the BDI-II with an adolescent Portuguese population (n=775) was performed by Martins (2000) who obtained a Cronbach alpha of .89, a value suggestive of good internal consistency but its length makes its clinical utility in adolescents challenging.The CDI (Kovacs, 1985) is a selfassessment tool targeting youth ages 7 -17 years, consisting of 27 items, divided into 5 factors: negative mood, interpersonal problems, ineffectiveness, anhedonia, and low self-esteem (Pullen, Modrcin-McCarthy, & Graf, 2000).The total score is calculated by summing all items and can range from 0 to 54, 15 being the threshold value (Kovacs, 1985).The CDI has been used in various studies and ethnic populations (DeRosa & Logsdon, 2006).A psychometric study of the CDI in the Portuguese population has been carried out by Cardoso, Rodrigues and Vilar (2004), but concerns about its poor discriminant validity, high rate of false-negatives and its relatively narrow age range suggest weak psychometrics and limited applicability (Cole & Martin, 2005;Kovacs, 1985;Myers & Winters, 2002) while its length makes its utility in adolescents challenging.

Aims of the Study
The main objective of this study was to analyze the psychometric characteristics of the Portuguese version of the KADS-6 and compare it with other measures of depression in adolescents previously validated in Portugal in this group (BDI-II and CDI).In this way we intended to verify the unidimensionality of KADS-6 and analyze its internal consistency.
We hypothesised that the KADS-6 would: (a) demonstrate an internal consistency similar to the BDI-II and CDI; (b) be reasonably correlated with the BDI-II and CDI.

Participants
The study design is a cross-sectional survey.Two convenience samples were collected.A fi rst sample of 773 non-clinical adolescents in a school setting in the West-Central Region of Portugal, 44.7% male, mean age 15.9 years (range: 12 to 18 years; SD: 1.37 years), and a second, clinical sample of 134 adolescents in three youth oriented clinical settings in the northern region of Portugal, 52.6% male, mean age 15.07 years (range: 12 to 18 years; SD 1.37 years).A data collection tool was designed specifically for this study and included questions pertaining to demographic, social and school related items.
The BDI-II and CDI, and the KADS-6, are all available in Portuguese.The KADS-6 has been previously translated by RG who is a Portuguese native speaking, and backtranslated by an English native speaking translator and the fi nal Portuguese version was adjusted for mental health constructs by RG and SK, mental health professionals with extensive clinical practice experience (Kutcher & Chehil, 2007b;Appendix).

Procedures
A pre-study survey, addressing the feasibility of this approach was conducted with four teenagers, two males and two females.They were asked to read the protocol in order to critically determine the time required to complete it and to raise questions about its implementation.No substantive concerns were identifi ed in this survey and accordingly the protocol as described above was implemented.
Following authorization and ethics review by the appropriate institutions, the study took place between April and June 2011.Informed consent was obtained from all participants as well as their legal guardians.Instructions were the same in both groups.The instruments were applied in the following order: KADS-6, BDI-II and CDI, with each application taking on average 20 minutes to complete.In the non-clinical sample, the completion of the survey took place in the school classroom, under the supervision of one of the authors (SD).
Data collected were entered in a database, and statistical procedures were performed using the Statistical Package for Social Sciences (SPSS) version 17.0 for Windows.
A factor analysis of the KADS-6 items was performed with varimax rotation, in order to test the unidirectionality of the instrument.This analysis was performed with the aim of comparing results of the Portuguese study to studies conducted in the English language.Reliability was calculated according to classical test theory (Kline, 2005).To examine convergent validity, a Pearson correlation between the three depression measures was performed.

Ecological Validity and Acceptability of the Portuguese KADS-6
The ecological validity of the KADS-6 Portuguese version is reasonable.The overall response rate was 100%, no questionnaires had to be discarded from analysis, and the applicability was good with 897 respondents (98.9%) fully completing the scale.Field researcher reports noted that in both groups, students preferred the use of the KADS-6 and that it was completed in less time and with greater ease than other the BDI-II or the CDI.

General Data on the Psychometrics of the KADS-6 Portuguese Version
A factor analysis of the items of the KADS-6 was performed with varimax rotation.Data is appropriate to execute this analysis, with KMO= .839and good sphericity with a signifi cance test of Bartlett .000.The varimax results obtained support the unidimensionality of the measure (eigenvalue 2.948), with an explained variance of 49.14%.
In order to study the KADS-6 validity and reliability and compare it with the other measures of depression, internal consistency was assessed.Table 1 presents the results obtained for Cronbach's α, the average inter-item correlation and the range of item-total correlation for the clinical, non-clinical and total sample.In the clinical sample, all dimensions show adequate internal consistency, with Cronbach α values ranging from .75 (Self-esteem) to .91 (depression measured by the CDI).In the non-clinical sample, all dimensions show adequate internal consistency, with Cronbach α values ranging from .69 (Self-esteem) to .90 (depression measured by the BDI).In the total sample, all dimensions show adequate internal The symptomatic response categories of "much of the time", "most of the time" and at "all of the time" are scored considerably higher in the clinical compared to the school (non-clinical) sample (evaluated with Qui-squared with p < .001, in each group).
In the clinical sample, depression measured by the KADS-6 average scores are the double those in the nonclinical sample.In the other two scales, scores are much higher in the clinical sample, but the percentage differences are not as marked as in the KADS-6.The CDI is remarkable for the small differences between clinical and non-clinical samples (see Table 3).

Group Differences
In order to investigate the existence of differences in scores between academic cycles, an analysis of variance, one-way ANOVA for three independent samples was used.There were statistically signifi cant differences between the academic cycles, in the clinical sample, for depression measured by KADS-6 with F (2; 130) = 5.335, p = .006.To verify in which cycles the differences were most pronounced, a post hoc analysis, using the Tukey method was performed showing that students endorsed more depression during the second cycle of the 4 th year.
In order to investigate differences between clinical and non-clinical samples an analysis using Student t test for independent samples was performed.Differences were found between clinical and non-clinical participants with, t (153) = 6.747, p <.001 (M = 4.62, SD = 3.94 in clinical sample, M = 2.23, SD = 2.59 in non-clinical sample).These results confi rmed that participants in the clinical sample exhibit more depressive symptoms than participants in the non-clinical sample.

Correlations
In the clinical sample, moderate positive and statistically signifi cant correlations were found between total KADS-6 and total BDI and CDI scores, with correlation values of (r = .44,p <.001) and (r = .61,p <.001), respectively.
In the non-clinical sample, moderate positive and statistically signifi cant correlations between the KADS-6 and total BDI and CDI were found, with correlation values of (r = .60,p <.001) and (r = .57,p <.001), respectively.In relation to both samples considered together, moderate positive and statistically signifi cant correlations between the KADS-6 and total BDI and CDI scores are found, with correlation values of (r = .60,p <.001) and (r = .61,p <.001), respectively.

Discussion
We studied the psychometric characteristics of the KADS-6, in a mixed sample of Portuguese adolescents including youth in both clinical and non-clinical (school) settings, comparing it with other self-report measures of assessment of depression in adolescents.Our intention was to verify the unidimensionality of KADS-6, analyze the internal consistency of the KADS-6 compared with the BDI and CDI and assesses the convergent validity of the KADS-6.
Our results show the KADS-6 to be a unidimensional measure with internal consistency as good as the other scales evaluated.This fi nding is consistent with other reports regarding the KADS-6 from North American samples (Brooks et al., 2003).Furthermore, the KADS-6 demonstrated better discrimination between clinical and non-clinical samples than the other two scales, suggesting it may be a better discriminant between common but not diagnostically signifi cant depressive symptoms in this population.
Students in higher grades showed more depressive symptoms than those in lower grades, consistent with the known epidemiology of the onset of depression in this age group (Arnarson & Craighead, 2011;Harrington, 2001;Pullen et al., 2000) and fi eld note responses indicated greater youth comfort with, utility and brevity of use for the KADS-6 compared to the other options.
Our second hypothesis was that the KADS-6 scores would have a positive and signifi cant correlation with BDI-II and CDI scores.In congruence with previous studies (Brooks et al., 2003;LeBlanc et al., 2002), our results confi rmed this hypothesis.The results show that there is convergent validity with total BDI and CDI scores in both clinical and non-clinical samples.However, this correlation is not very high.This may be because both other scales assess more symptom possibilities, many of which are commonly found in non-depressed youth and thus perhaps showing less discriminant value for depression.The poorer differentiation between clinical and non-clinical samples of the BDI-II and CDI compared to the KADS-6 supports this possibility.
Our study had a number of limitations.First, we did not compare any of the scales studied to the gold standard of a structured clinical interview to determine caseness.Second, although our total sample included both rural and urban settings it was not necessarily representative of the national population.We also did not conduct evaluations nor comparisons of temporal stability in this study.
Despite these limitations we consider that our fi ndings support the use of the KADS-6 in the diagnostic assessment of depression in young people in Portugal.This is consistent with studies reported from North America and Poland (Brooks et al., 2003;Mojs, Warchol-Biederman, & Samborski, 2012).It demonstrates good psychometric properties, is easy to understand and its brevity and adolescent friendly language make it useful in both school and clinic settings.It better differentiates between clinic and non-clinic samples than either the BDI or the CDI.It is much shorter and takes less time to complete than either the BDI or the CDI and was the subject of many fewer queries from respondents during the application of the scales in both clinical and non-clinical settings.T he use of a youth friendly six-item self-report scale with established specifi city and sensitivity for depression may be preferred to longer and less diagnostically focused alternatives.
Future studies in Portugal could involve validation comparisons of these instruments against a diagnostic gold standard, such as the MINI or other semi-structured interview, temporal stability, as well as the scale's application across a wider age group and in more geographic locations.

Signifi cant Outcomes
• KADS-6 presents a high internal consistency.
• KADS-6 effi ciently differentiates between clinic and non-clinic samples.• There is convergent validity with total BDI and CDI scores in both clinical and non-clinical samples.• KADS-6 can be advantageously applied in Portuguese cultural settings to foster recognition of depression in youth.

Table 1
Fidelity Measures of Depression

Table 2
Response Rates for Each Item

Table 3
Average, Standard Deviation and Ranges of Scores

Table 4
Differences between Academic Cycles