Abstract
This study examined familial factors unique to Latino adolescents within a cognitive vulnerability stress framework, in which negative patterns of thinking contribute to the development of depressive symptoms following stressful life events. Participants were Latino adolescents (N = 90), 12 to 18 years old. Almost half (~48%) reported elevated depressive symptoms. The final regression model explained 67% of the prevalence of depressive symptoms; stressful life events (including family conflict) and a ruminative response style emerged as unique predictors. Additional sociocultural factors contributing to the model’s explanatory power were familism, family cohesion, and perceived discrimination. Rumination and stressful life events should be considered targets for depression prevention and early intervention programs with Latino youth. Such programs would benefit from cultural modifications that encourage family cohesion and a sense of familism.
Depressive disorders, a leading global health problem (Mathers, Boerma, & Ma Fat, 2008), are a significant burden for individuals, families, and society. The majority of depressive illnesses in the United States originate during adolescence (Kessler et al., 2005), when rates of depression increase sixfold (Merikangas et al., 2010), and Latino adolescents consistently report the highest levels of depressive symptoms and disorders in comparison with their non-Hispanic White, African American, and Asian American counterparts (Saluja et al., 2004; Stein et al., 2010). Suicide, depression’s worst outcome, is the third leading cause of death of youths aged 10 to 24 years, and Latina adolescents attempt suicide significantly more often than their non-Hispanic peers (Eaton et al., 2012). Yet little research has evaluated why Latino adolescents seem to be at greatest risk for depressive symptoms.
Currently in the United States, 20% of youths under 18 years are Latino; by 2050, they may number 40% (Ortman & Guarneri, 2009). Latino adolescents’ levels of depressive symptoms suggest a growing need for depression prevention and treatment programs sensitive to their specific cultural and psychosocial needs. National and international agencies have called for increased efforts to prevent depression but cite a need for research on cultural factors that may influence the development of depressive symptoms to ensure that prevention and treatment programs will be culturally relevant (Merry et al., 2011; National Research Council and Institute of Medicine, 2009). In this study, we therefore examine the impact of familial factors, cognitive vulnerabilities, and stressful life events on depressive symptoms in Latino adolescents.
Background and Theoretical Framework
Many depression treatment and prevention programs are based on the psychological theory that individuals are at high risk for depressive symptoms if they tend to process stressful life events through overly negative thought processes, or cognitive vulnerabilities. Three primary cognitive vulnerabilities have been found to contribute to the development of depressive symptoms in adolescents: (a) dysfunctional attitudes, (b) negative inferential style, and (c) ruminative response style. Dysfunctional attitudes are cognitive distortions regarding achievement, interpersonal factors, and/or intrapersonal factors; they lead to overly negative appraisals of self, world, and the future (Beck, 1987). Negative inferential style includes three types of negative inferences that individuals make when confronted with a stressful life event: inferences about (a) the event’s cause, (b) its consequences, and (c) characteristics of one’s self, given the event’s occurrence (Abramson, Metalsky, & Alloy, 1989). Negative inferential style encompasses both helplessness expectancies (the expectation that highly desired outcomes will not occur) and negative outcome expectancies (the expectation that additional negative events/outcomes will occur). Ruminative response style is a way of thinking in which individuals attend to their negative emotional states but do not act to relieve their symptoms or improve their situation (Nolen-Hoeksema, Wisco, & Lyubomirsky, 2008).
Within a cognitive vulnerability stress framework, dysfunctional attitudes and negative inferential style provide an individual’s negative thought content following a stressful life event, and a ruminative response style prevents that individual from moving beyond negative thoughts and feelings, which leads in turn to further deterioration in mental health (Hankin & Abramson, 2001). Stressful life events are necessary but not sufficient for depressive symptom development. In one study with rural young adolescents, stressful life events in the domains of family, friends, school, and romantic relationships were independently associated with depressive symptoms, but the cumulative effect of these stressors was associated with depressive symptoms most highly (Young, 2012). That study also documented a relationship between stressful events within the family and socioeconomic disadvantage. Ethnic minorities are overrepresented in lower socioeconomic classes, so it is plausible that Latino adolescents may experience increased stressors related to both socioeconomic disadvantage and stressful family environments.
Additional experiences of stressors may derive from ethnic minority status. Perceived discrimination can increase risk for psychological maladjustment during adolescence, particularly for Hispanic males (Zeiders, Umaña-Taylor, & Derlan, 2013). Experiences of discrimination are not simply stressful; they may also influence an adolescent’s views of his or her own cultural context and traditions, as well as interactions within the family. Ethnic identity may buffer the effects of perceived discrimination (Umaña-Taylor & Updegraff, 2007); ethnic identity has been associated with positive mental health (Phinney, Cantu, & Kurtz, 1997; Umaña-Taylor, 2004).
Cognitive vulnerability stress theories have guided many investigations of depressive symptom development in adolescents (Abela & Hankin, 2011; Young & Dietrich, 2015) but not of depression in Latino youth. Most research on predictors of depressive symptoms in Latino youth has focused on cultural and ecological contexts (Céspedes & Huey, 2008; Lorenzo-Blanco, Unger, Baezconde-Garbanati, Ritt-Olson, & Soto, 2012; Potochnick & Perreira, 2010). It is therefore important to determine whether cognitive vulnerability stress theories can be applied to Latino youth and to examine the contributions of familial factors to their depressive symptoms. Zayas, Lester, Cabassa, and Fortuna’s (2005) conceptual model of Latina adolescent suicide can be used to guide the selection of relevant familial variables associated with depressive symptoms in Latino adolescents.
In Zayas et al.’s (2005) model, the family’s sociocultural environment has three components: (a) culture and cultural traditions, (b) adolescent development, and (c) family functioning. Adolescence is difficult for many youths, but Latino adolescents may face unique challenges owing to dissonance between acculturating within the dominant culture and maintaining their own cultural identity and family closeness. Through acculturation, individuals change their behavior and attitudes to align with the host culture (Barona & Miller, 1994), but Latino adolescents’ development within American culture can conflict with their cultural traditions, increasing their risk for stressful events and depressive symptoms. In a study of 175 Mexican American families with adolescents, family conflict (i.e., stressful events) mediated the relationship between acculturation and depressive symptoms (Gonzales, Deardorff, Formoso, Barr, & Barrera, 2006).
The family is central to Latin American culture (Sabogal, Marín, Otero-Sabogal, Marín, & Perez-Stable, 1987). Researchers have generally found familism, familial interconnectedness and subjugation of the self to the family, to be protective against depressive symptoms (Keeler, Siegel, & Alvaro, 2014). This strong link between familial relations and depressive symptoms exists among Latinos regardless of country of birth or acculturation (Perez & Cruess, 2011), but it can be both positive and negative. According to Zayas et al. (2005), a traditional Hispanic family structure characterized by rigid, authoritarian parenting may prevent flexible responses to adolescents’ developmental needs, leading to family conflict and disruptions in family cohesion. Family dysfunction (family conflict and family cohesion) has been linked to poor psychosocial outcomes, including depressive symptoms (Céspedes & Huey, 2008).
Method
Design, Setting, and Participants
A cross-sectional descriptive correlational design was used in this study. Latino adolescents 12 to 18 years old were recruited through a rural pediatric primary health care clinic in the southern United States. The clinic is the only provider of pediatric services in its area, serving predominantly Latino, low-income, medically underserved children and families. The community lacks many public services, its transportation options are limited, and it has a high rate of transience.
Procedure
Flyers introducing the study were posted in the clinic; parents and children interested in participating provided their contact information on cards that were then kept in a locked box at the clinic. The study team contacted potential participants in order to explain the study. Team members met the participants and their parent(s)/legal guardian(s) at their homes or in a private room at the clinic. Following informed consent and adolescent assent, each parent/guardian completed a demographic questionnaire, and the adolescent participant completed standardized measures to assess study variables. Institutional review board (IRB) approval was obtained in advance.
Measures
The 70-item Adolescent Life Events Questionnaire (Hankin & Abramson, 2002) is a self-report checklist of negative life events typical among adolescents. The events fall into four domains: (a) family and parents, (b) relationships, (c) school, and (d) friends and social activities. Adolescents read each event and indicate with yes or no whether the event has happened to them in the last 3 months (yes = 1, no = 0). Scores are summed for a total from 0 to 70.
The Dysfunctional Attitudes Scale (Weissman & Beck, 1978) assesses pervasive negative attitudes and beliefs regarding self, world, and the future. The original scale contains two parallel 40-item forms; we used a short version with nine items found to have an internal consistency reliability higher than that of the full scale (Andrews, Lewinsohn, Hops, & Roberts, 1993). Higher scores indicate more dysfunctional attitudes. Internal consistency in an adolescent sample has been reported at 0.70 (Young, LaMontagne, Dietrich, & Wells, 2012).
The modified Adolescent Cognitive Style Questionnaire (ACSQ; Alloy et al., 2012) assesses negative inferential style by measuring negative inferences for cause, consequence, and self across three domains: (a) achievement, (b) interpersonal, and (c) appearance. The original ACSQ demonstrated good internal consistency and test-retest reliability (Hankin & Abramson, 2002). Cronbach’s alpha for the modified ACSQ is .94 (Alloy et al., 2012).
The Ruminative Response Style Subscale (Nolen-Hoeksema & Morrow, 1991) consists of 22 items describing responses to depressed mood that focus on self, symptoms, and possible consequences and causes of mood. Participants indicate how much each item applies to what they generally do when they are feeling down or sad. Cronbach’s alpha for this subscale with an adolescent sample was .95 (Young et al., 2012).
The 12-item Short Acculturation Scale for Hispanic Youth (Barona & Miller, 1994), a self-report measure, assesses cultural behaviors related to family and social contexts. Items are scored on a Likert-type 5-point scale; specific items are reworded for comprehension by adolescents. Cronbach’s alpha with Hispanic adolescents has been reported at .94 (Barona & Miller, 1994).
A revised version of the original Perceived Discrimination Scale (Whitbeck, Hoyt, McMorris, Chen, & Stubben, 2001) was used to assess perceived discrimination. The 10 items in the revised scale focus on Hispanic/Latino rather than Native American culture (Umaña-Taylor, Updegraff, & Gonzales-Backen, 2011). Participants’ responses indicate how often a discriminating event has happened to them. This scale has demonstrated acceptable internal consistency of .80 (Umaña-Taylor et al., 2011).
The 14-item Multi-Group Measure of Ethnic Identity (Phinney, 1992) assesses three aspects of ethnic identification (positive ethnic attitudes and sense of belonging, ethnic identity achievement, and ethnic behaviors/practices). It has been found valid and reliable for ethnically diverse adolescents (Roberts et al., 1999).
The Familism Scale (Steidel & Contreras, 2003) measures behavioral familism; its 18 items are related to support, honor, subjugation of self to family, and interconnectedness. Participants select how much they agree with each item on a 10-point scale. Items are summed; higher scores indicate lower familism. Internal consistency for this scale with a Hispanic adolescent sample was .87 (Baumann, Kuhlberg, & Zayas, 2012).
Family Conflict, a nine-item subscale of the Family Environment Scale (Moos & Moos, 2009), evaluates expressed anger and conflict within the family. Responses are summed; higher total scores indicate greater family conflict. This measure has been validated in an adolescent sample, with internal consistency reported as .72 (Boyd, Gullone, Needleman, & Burt, 1997).
Family Cohesion, also a nine-item subscale of the Family Environment Scale (Moos & Moos, 2009), assesses commitment, help, and support of family members for one another. Higher scores represent higher levels of family cohesion. Internal consistency has been reported at .78, with test-retest reliability of .86 after 2 months (Moos & Moos, 2009).
The Center for Epidemiologic Studies–Depression Scale for Children (CES-DC; Weissman, Orvaschel, & Padian, 1980) is a 20-item self-report modified version of the adult CES-D that assesses frequency of depressive symptoms over the past week. Scores are summed; higher scores indicate greater frequency of depressive symptoms. The internal consistency in an adolescent sample has been reported as .92 (Young et al., 2012).
Data Analysis
IBM SPSS Version 21 was used to analyze the data. Descriptive statistics were calculated first, and data distributions were inspected for selection of appropriate measures of central tendency. Variables lacking normal distributions were rank transformed prior to bivariate and multivariate analyses. Pearson’s correlation was used for bivariate associations between demographic information, contextual factors, stressful life events, cognitive vulnerabilities, and depressive symptoms. Hierarchical multiple linear regression was then used to test multivariate associations among study variables to determine the contributions of each independent variable to depressive symptoms.
Results
Sample
A total of 123 adolescents 12 to 18 years of age completed the questionnaires. Those who indicated Hispanic/Latino ethnic heritage and completed all measures (N = 90) were included in the analyses. Participants were asked to write down their ethnic group; the majority indicated Hispanic (n = 41, 45.5%) or Mexican/Mexican American (n = 29, 32%). The majority were 16 years or younger (n = 74, 82%); 56% (n = 50) were female. Parents were usually married or living with a partner (n = 74, 82%). Family socioeconomic status (SES) was low: 92% (n = 80) were in the two lowest SES groups, and only 15% (n = 13) were educated beyond high school.
Descriptive Statistics
The majority of variable distributions were skewed, so medians, 25th and 75th interquartile ranges, and minimum and maximum values for all standardized measures are presented in Table 1. Overall, numbers of stressors were relatively low, but within the family domain, several items were endorsed frequently: ~48% (n = 43) reported an argument with a close family member; ~47% (n = 42) having to do unwanted chores. Additional stressors included fighting with parents over personal goals, desires, or choice of friends (~39%, n = 35); having to care for siblings (~38%, n = 34); being unable to please parents (~31%, n = 28); and parental criticism/yelling about not doing well in school (~32%, n = 29). Depressive symptoms were elevated, with ~48% (n = 43) at or above the cutoff score of 16. Levels of perceived discrimination were fairly low, with a median of 12 and the max score 23 falling just above the middle range (i.e., 10-40 total).
Descriptive Statistics of Measures (N = 90).
Note. IQR = 25th and 75th interquartile range; ALEQ = Adolescent Life Events Questionnaire; CES-DC = Center for Epidemiological Studies–Depression Scale for Children; DAS = Dysfunctional Attitudes Scale; ACSQ = Adolescent Cognitive Style Questionnaire; RRS = Ruminative Response Scale; SASHY = Short Acculturation Scale for Hispanic Youth; PDS = Perceived Discrimination Scale; MEIM = Multi-group Measure of Ethnic Identity; FAM = Familism Scale; FCON = Family Conflict Scale; FCOH = Family Cohesion Scale.
Bivariate Associations
Table 2 presents correlations for all study variables. No gender differences were detected. One statistically significant result emerged for age: Younger participants reported more dysfunctional attitudes (r = −.348, p = .001). Lower SES was associated with lower levels of ethnic identity (r = −.227, p = .036) and higher levels of acculturation (r = .256, p = .018). Stressful life events were inversely related to family cohesion (r = −.266, p = .011) and ethnic identity (r = −.295, p = .005): Higher levels of stressors were associated with lower levels of family cohesion and ethnic identity. High levels of reported stressors were also strongly related to higher levels of rumination (r = .587, p < .001) and depressive symptoms (r = .615, p < .001).
Correlations Between Key Study Variables (N = 90).
Note. Values in each cell are r (p value). SES = socioeconomic status; ALEQ = Adolescent Life Events Questionnaire; DAS = Dysfunctional Attitudes Scale; ACSQ = Adolescent Cognitive Style Questionnaire; RRS = Ruminative Response Scale; SASHY = Short Acculturation Scale for Hispanic Youth; PDS = Perceived Discrimination Scale; MEIM = Multi-group Ethnic Identity Measure; FAM = Familism Scale; FCOH = Family Cohesion Scale; FCON = Family Conflict Scale; CES-DC = Center for Epidemiological Studies–Depression Scale for Children. Gender: Female = 0, Male = 1.
Correlation is Point-Biserial. All other correlations are Pearson’s correlations.
Of the cognitive vulnerabilities, dysfunctional attitudes was the only variable not significantly associated with depressive symptoms (r = .175, p = .103). Although overall perceived discrimination levels were low, those who reported higher levels of discrimination also reported higher levels of depressive symptoms (r = .313, p = .003). Lower levels of family cohesion (r = −.299, p = .004) and familism (r = .217, p = .04) were associated with higher levels of depressive symptoms. Participants with high levels of family conflict also had high levels of depressive symptoms (r = .340, p = .001).
Multivariate Associations
Hierarchical linear regression was used to examine the unique contribution of each independent variable to depressive symptoms. Variables in the final model were those that demonstrated a statistically significant association with depressive symptoms in the bivariate analyses (see Table 3). At Step 1, the familial variables of familism and family cohesion were added. These accounted for ~11% of the variability in depressive symptoms; the multivariate association was statistically significant (multiple R = .33, p = .008, adjusted R2 = .086). Family cohesion contributed uniquely as a protective factor (β = −.256, p = .018). In Step 2, the stress variables (stressful life events, perceived discrimination, and family conflict) were added to the model, yielding a statistically significant increase in the ability to explain depressive symptoms, which went from 11% to 47%. The resulting multiple correlation was statistically significant (multiple R = .686, p < .001, adjusted R2 = .439). Stressful life events (β = .487, p < .001), perceived discrimination (β = .215, p = .016), and family conflict (β = .260, p = .012) all contributed uniquely to depressive symptoms.
Hierarchical Multiple Linear Regression Examining the Contribution of Familial Factors, Stress, and Cognitive Vulnerabilities to Depressive Symptoms (N = 90).
Note. Multiple R = .820, p < .001; R2 = .673 (Adjusted R2 = .645). FAM = Familism Scale; FCOH = Family Cohesion Scale; ALEQ = Adolescent Life Events Questionnaire; PDS = Perceived Discrimination Scale; FCON = Family Conflict Scale; ACSQ = Adolescent Cognitive Style Questionnaire; RRS = Ruminative Response Scale.
Finally, Step 3 of the hierarchical analyses included the cognitive vulnerabilities, resulting in another statistically significant increase in the ability to explain the variability in depressive symptoms (from 47% in Step 2 to a final shared variability of 67%). The resultant overall multiple correlation with depressive symptoms was .82, statistically significant (p < .001). Given that the adjusted R2 was .645 (~65% shared variance), it could be expected that this finding would not change dramatically if replicated. After controlling for all study variables, rumination (β = .575, p < .001), family conflict (β = .163, p = .05), and stressful life events (β = .206, p = .014) were all unique contributors to depressive symptoms.
Discussion
In this study of familial factors, cognitive vulnerabilities, and stressful life events related to depressive symptoms in Latino adolescents, the final regression model explained 67% of the variance in depressive symptoms. Three variables were unique contributors: rumination, family conflict, and stressful life events. These findings support a cognitive vulnerability stress framework in which stress (negative emotional arousal produced by the experience of stressors) and the tendency to passively and repetitively focus (to ruminate) on negative emotional states contributes to elevated depressive symptomatology. They support the use of the cognitive vulnerability stress model to guide depression prevention efforts for Latino youth, and they provide insight into the impact of familial factors on the prevalence of depressive symptoms.
Almost half (48%) of the participants reported elevated depressive symptoms. Although participants were recruited as they accessed care at a medical clinic, data were collected at times and places convenient for them and their families. This procedure should have minimized the effect of acute illness on depressive symptoms, but it did not address the impact of chronic illnesses. Of course many participants were simply accessing preventative care (e.g., vaccinations and sports physicals), but future studies would benefit from examining presence of chronic illness and its relationship to depressive symptoms. Nevertheless, the high prevalence of depressive symptoms in this sample is similar to findings from the Youth Risk Behavior Surveillance (Kann et al., 2014) for Latina adolescents, 48% of whom reported cessation of their typical activities due to feelings of sadness or hopelessness (although only 25% of males in that study did so). Surprisingly, in the present study, there were no gender differences in prevalence of depressive symptoms or likelihood of being in the high-risk group (i.e., score ≥16; males = 46.2%, females = 50%). One potential explanation may be suggested by Hankin et al. (1998), who identified gender differences in depressive symptoms as only beginning to emerge between the ages of 13 and 15 years. The majority of participants in this study were younger than 16 years old, so it may be that the participants were still too young to show that difference.
Elevated depressive symptoms were associated with higher levels of perceived discrimination (r = .313, p = .003) and family conflict (r = .340, p = .001) and lower levels of familism (r = .217, p = .04) and family cohesion (r = −.299, p = .004). These findings are supported by previous research documenting the deleterious effect of discrimination on adolescents’ mental health. In a sample of immigrant Latino adolescents, Potochnick and Perreira (2010) found, after controlling for multiple stressors, that perceived discrimination was significantly associated with increased likelihood of depressive symptoms. As conceptualized by Zayas et al. (2005), family functioning was also associated with depressive symptoms; conflictive family environments were associated with higher levels of depressive symptoms, whereas those reporting cohesive family environments reported fewer depressive symptoms. The majority of participants (87.5%) reporting high levels of conflict also reported low levels of family cohesion, suggesting that family-level depression prevention interventions should focus on reducing family conflict and strengthening family cohesion.
The remaining familial factors of acculturation (r = .104, p = .320) and ethnic identity (r = −.159, p = .135) were not significantly associated with depressive symptoms. Perhaps acculturative stress, as opposed to acculturation level, is more highly related to depressive symptoms. Acculturative stress has been found to mediate the relationship between perceived discrimination and psychological distress (Torres, Driscoll, & Voell, 2012). Schwartz, Unger, Zamboanga, and Szapocznik (2010) have called for a broader conceptualization of the acculturation process to improve our understanding of acculturation and its effect on health outcomes. Future studies should consider including the measurement of acculturative stress, as well as dimensions of adolescent-parent interactions (e.g., communication patterns, mutuality) to more fully describe the relationship between Latino adolescent acculturation and depressive symptoms. Interestingly, higher levels of acculturation were associated with less ethnic identity and familism. Given the protective nature of familism, interventions for Latino youth should perhaps include components that tap into the family as a source of support.
Although familial variables contributed to the overall explanatory power of the final regression model, rumination, stressful life events, and family conflict were the only unique predictors of depressive symptom prevalence. Rumination alone explained 33% of the prevalence of depressive symptoms in this sample. Research has demonstrated the detrimental effects of rumination on various aspects of mental health, including depression and anxiety (Michl, McLaughlin, Shepherd, & Nolen-Hoeksema, 2013; Young & Dietrich, 2015). Our findings too demonstrate the deleterious effects of rumination on depressive symptoms in Latino adolescents. Stressors were also directly linked to depressive symptoms, so perhaps depression prevention and early intervention programs should address stress reduction in general as well as specific strategies for reducing ruminative thinking.
Surprisingly little intervention research has addressed stress management with adolescents. Recently, Rew, Johnson, and Young (2014) have identified that teaching cognitive skills to cope with stress improved perceived stress levels and decreased psychological distress. Contemplative practices such as transcendental meditation, mindfulness, and yoga have shown promising results with adolescents (Barnes, Treiber, & Johnson, 2004; Tan & Martin, 2015; White, 2012). Although Rew et al. noted that research needs to develop and test adolescent stress reduction interventions further, the positive effects of teaching cognitive skills for adaptive coping seems promising for the improvement of adolescent mental health.
Limitations
The present study has certain limitations. First, the sample was fairly small. But the robustness of the results points to their validity. Future studies should include larger samples of Latino youth from diverse geographical regions to enhance generalizability. Second, the data consisted only of adolescent self-reports. Future studies would benefit from parental and teacher reports, as well as other data collection methods (e.g., personal interviews). Third, the internal consistency of the Family Conflict subscale (Moos & Moos, 2009) was quite low. Fortunately, this limitation was minimized by the measurement of family stress with the Adolescent Life Events Questionnaire (Hankin & Abramson, 2002). Finally, country of origin and time in the United States were not assessed. Alegria et al. (2008) have cautioned against aggregating Latinos into a single group, because evidence has shown different rates of psychiatric and substance use disorders within Latino subgroups. Future studies with larger groups of Latino adolescents would benefit from a more thorough evaluation of their ethnic heritage and family immigration history.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported in part by a grant from Sigma Theta Tau International, Honor Society of Nursing, and the Council for the Advancement of Nursing Science. Editorial support with manuscript development was provided by the Cain Center for Nursing Research and the Center for Transdisciplinary Collaborative Research in Self-management Science (P30, NR015335) at The University of Texas at Austin School of Nursing.