Confirmatory Factor Analysis and Internal Consistency of the Suicidal Ideation Scale of the Center for Epidemiological Studies Depression Among Colombian Adolescents

Abstract

The current study tested the one-dimensionality of the Center for Epidemiological Studies Depression Scale for Suicidal Ideation (CES-D-SI) and its reliability among high school students in Santa Marta, Colombia. A total of 1,462 tenth and eleventh-grade students from public and private schools completed the CES-D-SI. The students were between 13 and 17 years (M = 16.0, SD = 0.8), 60.3% were female, and 55.3% were in the tenth grade. A confirmatory factor analysis (CFA) was conducted, and the chi-square, root mean square error of approximation (RMSEA), comparative fit index (CFI), Tucker-Lewis index (TLI), and standardized root mean square residual (SRMR) statistics were calculated. The internal consistency of the dimension was calculated using Cronbach’s alpha and McDonald’s omega coefficients. The results of the CFA were as follows: chi-square = 26.51, df =2, p = .001; RMSEA = .092; 90% confidence interval = .063–.124; CFI = .983; TLI = .950; and SRMR = .019. Cronbach’s alpha coefficient was .75, and McDonald’s omega coefficient was .77. As conclusions, the scale exhibited a unidimensional structure similar to that in other studies and presented good reliability values. Hence, its use is recommended to screen suicidal ideation in school-aged adolescents in Santa Marta, Colombia.

Keywords

Suicidal ideation adolescent students reliability and validity validation studies

Introduction

Suicide is a significant public health problem, and the suicide rate is considered an indicator of mental health. Worldwide, approximately 800,000 people die by suicide each year, particularly in low- and middle-income countries (World Health Organization, 2014). Globally is the second cause of death in adolescents and young between 15 and 29 years (World Health Organization, 2014), while is the second leading cause of death among adolescents in the United States (Centers for Disease Control and Prevention, 2017), and the third cause of death adolescents in Latin America and Caribbean (Panamerican Health Organization, 2014). Suicide is often preceded by thoughts about dying or the presence of suicidal ideation (SI); 4% of adolescents with suicidal thinking will attempt suicide (Nock et al., 2013).

From a syndemic perspective, the epidemiology of suicide is linked to structural determinants that configure vulnerability and emotional suffering (Campo-Arias & Suárez-Colorado, 2019). In the case of Colombia, like other countries in Latin America and the Caribbean, socioeconomic inequality, unsatisfied basic needs, and a history of violence due to armed conflict or derived from organized crime are pieces of evidence of complex social problems (Ortiz-Hernandez et al., 2007). The impact of the various adverse events propitiates mental health effects, particularly in adolescents, given the early exposure and potential to manifest in other stages of development (Brockie et al., 2015).

The most recent National Survey of Mental Health in Colombia determined that 6.6% of adolescents reported SI and 2.5% reported having attempted suicide (Ministerio de Salud-Colciencias, 2015), while 10.4% of people die by suicide (Christiansen & Jensen, 2007). Various Colombian studies of high and middle school students have revealed SI prevalence rates of 13% in Santa Marta (Ceballos-Ospino et al., 2015), 17.9% in Pasto (Villalobos-Galvis, 2009), 24% in Bogotá (Pineda-Roa, 2019), and 45% in Manizales (Amezquita et al., 2008).

The multifactorial nature of suicide in adolescents is evident in different studies that show an association with conflict, emotional loss, or other significant adverse experiences (Paul, 2018). Although, factors associated in adolescents may be related to the psychosocial and biological changes of the transition period, difficulties in regulating impulsivity, low tolerance for frustration, low coping capacity, mental disorders, consumption of psychoactive substances, violence, bullying, family conflicts, academic or social pressures, among others (Chan et al., 2020). In summary, suicide during adolescence is a complex problem associated with proximal, intermediate, and distal risk factors; it is an event that can be prevented, especially if SI is identified early (World Health Organization, 2014).

The early tracking of risk factors for suicide, such as SI, can contribute to the early care of the adolescent and consequently improve the prognosis regarding the severity given by intentionality, ambivalence, or rigidity about the idea, persistence, or intermittency of suicidal thoughts (Suárez et al., 2016). Likewise, early detection of SI could reduce the impact emotionally overload the family, friends, community, and providers of medical and psychological services (Séguin et al., 2014); it is difficult to estimate an individual’s immediate risk of suicide based on information from the family, school, and primary care (Lenz et al., 2019). Consequently, it is necessary to identify valid reliable screening measures to advance the understanding of suicide. While in the global context, the number of instruments available to quantify SI is broad, and several of them exhibit acceptable psychometric properties (Pianowski et al., 2015), in Colombia, there are limited valid reliable instruments available for the adolescent population in schools (Campo-Arias et al., 2019; Pineda-Roa et al., 2018; Villalobos-Galvis, 2010).

The Center for Epidemiological Studies Depression Scale for Suicidal Ideation (CES-D-SI) was derived from the Center for Epidemiologic Studies Depression Scale, CES-D (Radloff, 1977). The CES-D is used to evaluate depression; however, the instrument also explores thoughts about dying as symptoms related to depression (Roberts & Chen, 1995). Accordingly, the CES-D-SI is frequently used in Latin America, and some studies report its application among university students (González-Forteza et al., 1988; Rosales et al., 2012), among women diagnosed with eating disorders (Unikel et al., 2004), and among secondary school students (Roberts & Chen, 1995; Rosales-Pérez et al., 2016; Villalobos-Galvis, 2010).

The CES-D-SI exhibited high internal consistency when administered to 2,614 adolescents. Specifically, Cronbach’s alpha was .88 when the instrument was administered to 924 adolescents in the United States and .92, when administered to 1,354 adolescents in the Mexican population (Roberts & Chen, 1995). In another study, an exploratory factor analysis (EFA) of the scale administered to a sample of 476 Mexican adolescents reported a one-dimensional structure that explained 63.4% of the variance. In contrast, the single factor revealed an internal consistency of .76, Cronbach’s alpha (Unikel et al., 2004). Additionally, when administered to 363 Colombian adolescents, the CES-D-SI demonstrated reliability based on Cronbach’s alpha of .86 and McDonald’s omega of .87. In addition, a confirmatory factor analysis (CFA) found one dimension that explained 71.2% of the variance with good indicators for goodness of fit, i.e., root mean square error of approximation (RMSEA) = .00, 90% confidence interval (90%CI) = .00–.09, comparative fit index (CFI) = 1.0, Tucker-Lewis index (TLI) = 1.0, and standardized root mean square residual (SRMR) = .01 (Pineda-Roa et al., 2018).

Nonetheless, psychometric studies that apply CES-D-SI to adolescents are scarce. Among Colombian adolescents, the CES-D-SI performance is based only on a group of adolescents from a municipality of few inhabitants with rural characteristics (Pineda-Roa et al., 2018). Hence, it is necessary to study the instrument's psychometric performance in different contexts, given that its performance varies according to the characteristics of the population (Keszei et al., 2010). Similarly, few studies corroborate the CES-D-SI structure using a CFA (Pineda-Roa et al., 2018). To date, research has focused on the estimation of internal consistency (Roberts & Chen, 1995) and EFA (Unikel et al., 2004).

Evaluating SI can cause anxiety among health professionals due to underestimating SI or the overestimating of the imminent risk of suicide. Consequently, adequate care of people at the highest risk is not provided (Bryan & Rudd, 2006). In response, valid and reliable assessment measures will allow professionals to recognize better and assess the degree of risk for suicide. Furthermore, such measures are considered useful instruments because individuals reveal more significant SI levels in self-report measurements than clinical interviews administered by general practitioners (Johnson et al., 1999).

The present study's objective was to test the one-dimensionality and reliability of the CES-D-SI among high school students in Santa Marta, Colombia.

Method

Design

A validation study was designed, and the research was endorsed by the ethics committee of a Colombian university. The study followed the ethical recommendations for research on human subjects following the Declaration of Helsinki (World Medical Association, 2008). The participants' parents or guardians signed informed consent forms, and the sampled students agreed to participate.

Participants

The sample was calculated with an expected SI prevalence rate of 50%, the alpha error of 5%, and an accepted margin of error for the prevalence of 2%. Sampling was probabilistic and multistage, with a clustering and sample replacement for eventual losses of 20%. The schools were selected randomly through Epi-Info, taking into account a population of 10,810 school adolescents from tenth to eleventh grade and a sample of 1,948 students. The response rate was 75% for a final sample of 1,462 tenth and eleventh-grade students between 13 and 17 years (M = 16.0, SD = 0.8) in accredited private schools. Of the 1,462 participants, 60.3% were female, and 39.7% were male; 55.3% were in the tenth grade, and 54.7% were in the eleventh grade. In Colombia, economic strata are classified according to the level of income from stratum one to six. This investigation found 27.1% belonged to stratum one; 22.5% were in stratum two; 28% were in stratum three; 9.3% were divided among strata four, five, and six; and 13% did not report their stratum of residence.

Instrument

The students completed the CES-D-SI (Radloff, 1977). This instrument evaluates the symptoms related to SI in the last seven days and consists of four response alternatives. The response alternatives are 0 days, 1–2 days, 3–4 days, and 5-7 days, allowing for a total score between 0 and 12. The CES-D-SI has demonstrated adequate psychometric performance among Colombian adolescents (Pineda-Roa et al., 2018), young Americans, and Latin Americans (Roberts & Chen, 1995). Below are the four items on the CES-D-SI:

Have I felt that I couldn’t go on?

Have I had thoughts about dying?

Have I felt that my family would be better off if I were dead?

Have I thought about killing myself?

Procedure

The field team explained the objectives and purpose of the research to the educational institutions' directors, and upon receiving authorization, the consent forms were distributed to the parents or legal guardians of the sampled students. Once the students agreed to participate, the instruments were administered collectively in the classroom under the supervision of one member of the field team. Participation was anonymous, and general reports were made for the schools. Data collection was conducted between September and October 2018.

Statistical analysis

Using EFA, the Kaiser-Meyer-Olkin index (KMO) and Bartlett’s test for sphericity were calculated. A CFA was performed to corroborate the dimensionality of the CES-D-SI using the maximum likelihood method, with the assumption of finding a one-dimensional structure due to the limited number of items on the scale. The goodness of fit was assessed using CFA, and the coefficients for the RMSEA and 90% CI, CFI, TLI, and SRMR were calculated. Concerning the chi-square, it is expected that the probability is greater than .05, while the RMSEA and SRMR are approximately .06, and the CFI and TLI exhibit values greater than .89. The internal consistency was calculated using Cronbach’s alpha coefficients and McDonald’s omega coefficients. The analysis was performed using the statistical program STATA 13.0.

Results

The results of the EFA revealed that the KMO index = .765 and the Bartlett’s chi-square test = 1,460.31, while additional analysis found that the df = 6, p < .001. The CES-D-SI was determined to be a single-dimension instrument with eigenvalues of 2.37, which explained 59.1% of the total variance (Table 1).

Table 1.

Commonalities and loadings for the CES-D-SI among Colombian high school students.

Item	Commonality	Loading
1	.364	.603
2	.395	.629
3	.545	.738
4	.529	.728

The goodness of fit indices for the one-dimensional model of the CES-D-SI were chi-square = 26.51, df = 2, p = .001, RMSEA = .092 (90%CI .063–.124), CFI = .983, TLI = .950 and SRMR = .019. The Cronbach’s alpha coefficient was .75, and the McDonald’s omega coefficient was .77.

Discussion

The present study corroborates the one-dimensionality using CFA and the high reliability (Cronbach’s alpha and McDonald’s omega) of the CES-D-SI among high school students of Santa Marta, Colombia.

The CFA revealed that the CES-D-SI instrument in this adolescent sample was one-dimensional and explained 59.1% of the total variance. These results are similar to a study of Mexican adolescents that also showed a one-dimensional structure; however, only an EFA was performed (Unikel et al., 2004). Among the Colombian adolescents, the EFA resulted in the extraction of one dimension that explained 71.2% of the variance. The CFA showed an adjusted model with chi-square = 690.1, df = 6, p < .001, RMSEA = .00 (90%CI .00–.09), CFI = 1.0, TLI = 1.0, and SRMR = .01 (Pineda-Roa et al., 2018).

The CFA allows testing hypotheses against previous evidence of a single scale's dimensional structure and overcomes the deficiencies of an EFA that does not allow an indicator that only measures a specific factor to be identified. This is achieved when the chi-square goodness of fit indicators exhibit a probability greater than 5%, and RMSEA and SRMR less than .06, and a CFI and TLI greater than .89 (Hu & Bentler, 1999).

In the present study, Cronbach’s alpha and McDonald’s omega were calculated as internal consistency indicators. The coefficients were high, with a Cronbach’s alpha of .75 and a McDonald’s omega of .77. These findings are similar to those observed in Mexican adolescents, for whom the CES-D-SI alpha exhibited a Cronbach value of .76 (Unikel et al., 2004), and in Colombian adolescents, for whom Cronbach’s alpha was .86 and McDonald’s omega was .87 (Pineda-Roa et al., 2018). The adequate reliability for Cronbach’s alpha and McDonald’s omega is maintained in Colombian adolescents from different cultural contexts, similar to studies of Mexican and American adolescents (Roberts & Chen, 1995). McDonald's omega complements the calculation of internal consistency reliability using Cronbach’s alpha because the latter is a better indicator of internal consistency when the items are not tau-equivalent (McDonald, 1970). The equivalence standard is a principle that must be met for a precise estimation of the internal consistency using Cronbach’s alpha.

The systematic and rigorous evaluation of SI and other suicidal behaviors require valid and reliable measures, especially in vulnerable populations. In this study, the participating adolescents were from a middle-income country but were lower and lower-middle-class residents. According to various studies, as suicidal behaviors are associated with economic and social inequalities (Arenas et al., 2016), preventive care is particularly relevant to selective higher risk groups (World Health Organization, 2014).

As this study was conducted using a representative random sample of adolescents in tenth and eleventh grades at accredited private schools in Santa Marta, Colombia, the results are presented with the lowest degree of bias in the population. However, it is limited in that it does not include educational institutions in rural areas or other academic secondary educational institutions to corroborate its structure.

It is concluded that similar to other studies, the CES-D-SI scale exhibits a unidimensional structure and presents good reliability values. Thus, its use is recommended to screen SI in adolescent students from Santa Marta, Colombia.

Footnotes

Article Notes

References

Amezquita

González

Zuluaga

(2008). [Prevalence of depression and suicidal thoughts in students from 8th to 11th grade in eight public high schools in Manizales]. Hacia la Promoción de la Salud, 13(1), 143–153.

Arenas

Gómez-Restrepo

Rondón

(2016). [Suicidal behaviour and associated factors in Colombia. Results from the 2015 National Mental Health Survey]. Revista Colombiana de Psiquiatría, 45(1), 68–75. https://doi.org/10.1016/j.rcp.2016.03.006

Brockie

Dana-Sacco

Wallen

Wilcox

Campbell

(2015). The relationship of adverse childhood experiences to PTSD, depression, poly-drug use and suicide attempt in reservation-based Native American adolescents and young adults. American Journal of Community Psychology, 55(3-4), 411–421. https://dx-doi-org.web.bisu.edu.cn/10.1007/s10464-015-9721-3

Bryan

Rudd

(2006). Advances in the assessment of suicide risk. Clinical Psychology, 62(2), 185–200. https://doi.org/10.1002/jclp.20222

Campo-Arias

Suárez-Colorado

(2019). [Is the suicide a preventable event?] Revista Universidad Industrial de Santander, 51(3), 197–199. https://doi.org/10.18273/revsal.v51n3-2019002

Campo-Arias

Zuñiga-Díaz

Mercado-Marín

García-Tovar

(2019). [Factors and internal consistency analysis of the Okasha’s Suicidality Scale among adolescents]. Revista Cubana de Salud Pública, 45(1), 1–8.

Ceballos-Ospino

Suárez-Colorado

Suescún-Arregocés

Gamarra-Vega

González

Sotelo-Manjarres

(2015). [Suicidal ideation, depression and self-esteem in teenagers school in Santa Marta]. Duazary, 12(1), 15–22. https://doi.org/10.21676/2389783X.1394

Centers for Disease Control and Prevention. (2017). Youth risk behavior surveillance system. YRBS Questionnaire content -1991-2017. https://www.cdc.gov/healthyyouth/data/yrbs/pdf/2017/2017_standard_YRBS_item_rationale.pdf

Chan

Young

Hwan

Lee

Woon

Kim

(2020). Identification of influencing factors for suicidal ideation and suicide attempts among adolescents: 11-year national data analysis for 788,411 participants. Psychiatry, 291(1), 113228. https://doi.org/10.1016/j.psychres.2020.113228

10.

Christiansen

Jensen

(2007). Risk of repetition of suicide attempt, suicide or all deaths after an episode of attempted suicide: A register-based survival analysis. Australian and New Zealand Journal of Psychiatry, 41(3), 257–265. https://doi.org/10.1080/00048670601172749

11.

Cortina

J. M.

(1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78, 98–104. https://doi.org/10.1037/0021-9010.78.1.98

12.

Cronbach

(1951). Coefficient alpha and the internal structure of the test. Psychometrika, 16(3), 297–334. https://doi.org/10.1007/BF02310555

13.

González-Forteza

García

Medina-Mora

Sánchez

(1988). [Psychosocial indicator predictors of suicidal ideation in two generations of university students]. Salud Mental, 21(3), 1–9.

14.

Bentler

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

15.

Johnson

Lall

Bongar

Nordlund

(1999). The role of objective personality inventories in suicide risk assessment: An evaluation and proposal. Suicide Life-Threatening Behaviors, 29(2), 165–185. https://doi.org/10.1111/j.1943-278X.1999.tb01054.x

16.

Keszei

Novak

Streiner

(2010). Introduction to health measurement scales. Journal of Psychosomatic Research, 68(4), 319–323. https://doi.org/10.1016/j.jpsychores.2010.01.006

17.

Lenz

Röther

Bouna-Pyrrou

Mühle

Tektas

Kornhuber

(2019). The androgen model of suicide completion. Progress in Neurobiology, 172, 84–103. https://doi.org/10.1016/j.pneurobio.2018.06.003

18.

McDonald

R. P.

(1970). Theoretical foundations of principal factor analysis and alpha factor analysis. British Journal of Mathematics and Statistical Psychology, 23(1), 1–21. https://doi.org/10.1111/j.2044-8317.1970.tb00432.x

19.

Ministerio de Salud-Colciencias. (2015). [National Mental Health Survey 2015].

20.

Nock

Green

Hwang

McLaughlin

Sampson

Zaslavsky

Kessler

(2013). Prevalence, correlates, and treatment of lifetime suicidal behavior among adolescents: Results from the National Comorbidity Survey Replication Adolescent Supplement. JAMA Psychiatry, 70, 300–310. https://doi.org/10.1001/2013.jamapsychiatry.55

21.

Ortiz-Hernandez

Lopez-Moreno

Borges

(2007). Desigualdad socioeconómica y salud mental: revisión de la literatura latinoamericana [Socioeconomic inequality and mental health: a Latin American literature review]. Cadernos Saude Publica, 23(6), 1255–1272. https://doi.org/10.1590/S0102-311X2007000600002

22.

Panamerican Health Organization. (2014). Suicide mortality in the Americas: Regional report.

23.

Paul

(2018). Proximally-occurring life events and the first transition from suicidal ideation to suicide attempt in adolescents. Journal of Affective Disorders, 241, 499–504. https://doi.org/10.1016/j.jad.2018.08.059

24.

Pianowski

de Oliveira

Baptista

(2015). [Suicidal ideation and behavior: psychological assessment measures]. Psicología Desde el Caribe, 32(1), 81–120. https://doi.org/10.14482/psdc.32.1.5872

25.

Pineda-Roa

(2019). [Risk factors for suicidal ideation in a sample of Colombian adolescents and young adults who self-identify as homosexuals]. Revista Colombiana de Psiquiatría, 48(1), 2–9. https://doi.org/10.1016/j.rcp.2017.06.001

26.

Pineda-Roa

Martínez-Chía

Á.

Corredor-González

Herazo

Campo-Arias

(2018). [Psychometric findings of the suicidal ideation scale of the center of epidemiological studies in school enrolled adolescents in Samaca, Boyaca, Colombia]. Biosalud, 17(2), 47–55. https://doi.org/10.17151/biosa.2018.17.2.4

27.

Radloff

(1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1(3), 385–401. https://doi.org/10.1177/014662167700100306

28.

Roberts

Chen

(1995). Depressive symptoms and suicidal ideation among Mexican-origin and Anglo adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 34(1), 81–90. https://doi.org/10.1097/00004583-199501000-00018

29.

Rosales

Córdova

Ramos

(2012). [Suicidal ideation in Mexican students: a model of multiple relation with variables of personal identification]. Revista Iberoamericana de Psicología de la Salud, 22(1), 63–74. https://doi.org/10.25009/pys.v22i1.558

30.

Rosales-Pérez

Córdova-Osnaya

Cortés-Granados

(2016). [Reliability and validity of Roberts’ suicidal scale]. Journal of Health and Social Behaviors, 7(2), 31–41. https://doi.org/10.5460/jbhsi.v7.2.44302

31.

Séguin

Bordeleau

Drouin

Castelli-Dransart

Giasson

(2014). Professionals’ reactions following a patient's suicide: Review and future investigation. Archives of Suicide Research, 18(4), 340–362. https://doi.org/10.1080/13811118.2013.833151

32.

Suárez

Restrepo

Caballero

(2016). [Suicidal ideation and its relationship to emotional intelligence in Colombian college students]. Revista de la Universidad Industrial de Santander, 48(4), 410–478. https://dx-doi-org.web.bisu.edu.cn/10.18273/revsal.v48n4-2016005

33.

Unikel

Bojórquez

Carreño

(2004). [Validation of a brief questionnaire to measure the risk of abnormal eating behaviors]. Salud Pública de México, 46(6), 509–515.

34.

Villalobos-Galvis

(2009). [Suicidal behavior in high school and college students in San Juan de Pasto, Colombia]. Salud Mental, 32(2), 165–171.

35.

Villalobos-Galvis

(2010). Validity and reliability of the Positive and Negative Suicidal Ideation Inventory in Colombian students. Universitas Psychologica, 9(2), 509–520.

36.

World Health Organization. (2014). Preventing suicide: A global imperative.

37.

World Medical Association. (2008). Declaration of Helsinki.