Robin H. Miyamoto

Prediction of Major Depression and Dysthymia From CES-D Scores Among Ethnic Minority Adolescents

OBJECTIVE The Native Hawaiian Mental Health Research Development Program is an epidemiological longitudinal study of adolescents residing in Hawaii. This article examines the utility of the Center for Epidemiologic Studies-Depression Scale (CES-D) for predicting DSM-III-R diagnoses of major depression (MD) and dysthymic disorder (DD) and investigates whether prediction differs by gender and ethnicity. METHOD Diagnostic Interview Schedule for Children interviews were conducted with 556 adolescents randomly selected from among more than 7,000 students who had completed the CES-D. RESULTS Six-month prevalence rates were as follows: MD = 8.5%, DD = 4.7%, either (MDDD) = 9.9%. Prevalence rates were significantly higher among females, but after CES-D scores were accounted for, gender no longer predicted depression in most analyses. When a cutoff score of 16 was used, classification accuracy was lower for Native Hawaiians than non-Hawaiians. However, after group differences in gender and grade level were accounted for, the predictive validity of the CES-D did not differ by ethnicity. CES-D factor 1 scores identified MD, DD, and MDDD about as well as the total score or all three factors together. CONCLUSIONS These results support the validity of the CES-D for screening for depression among adolescents of Native Hawaiian and other minority backgrounds.

Cultural identification and attempted suicide in Native Hawaiian adolescents.

OBJECTIVES To determine rates of lifetime suicide attempts in a community sample of Native Hawaiian adolescents and determine the contribution of Hawaiian cultural affiliation, socioeconomic status, and psychiatric symptoms as risk factors for suicide. METHOD High school students were surveyed in the state of Hawaii for lifetime suicide attempts, Hawaiian cultural affiliation, socioeconomic status, and symptoms of depression, substance abuse, aggression, and anxiety. Multiple logistic regressions were used on 3,094 subjects to develop prediction models for lifetime suicide attempts. RESULTS Native Hawaiian adolescents had significantly higher rates of suicide attempts (12.9%) than other adolescents in Hawaii (9.6%). Hawaiian cultural affiliation rather than ethnicity was uniquely predictive of suicide attempts. Logistic regression indicated that depression, substance abuse, grade level, Hawaiian cultural affiliation, and main wage earners education best predicted suicide attempts in Native Hawaiian adolescents, while depression, substance abuse, and aggression predicted suicide attempts in non-Hawaiians. CONCLUSIONS Native Hawaiian adolescents have higher rates of attempted suicide than non-Hawaiian adolescents. Strong Hawaiian cultural affiliation rather than ethnicity is a risk factor for attempted suicide.

Cumulative effect of family environment on psychiatric symptomatology among multiethnic adolescents.

PURPOSE To examine the influence of family adversity and support on levels of psychiatric symptomatology in Hawaiian and non-Hawaiian adolescents. METHOD More than 4000 students from five high schools in Hawaii completed a survey during the 1992-1993 school year about their family environment and mental health. The response rate was approximately 60%. Logistic regression analyses were performed and responses for Hawaiians and non-Hawaiians were compared. RESULTS Hawaiian adolescents experienced significantly more adversity than their non-Hawaiian counterparts. The cumulative effect of family adversity had a greater effect on psychiatric symptomatology than any single indicator. Family support reduced the risk for internalizing symptoms, particularly for Hawaiian adolescents. The influence of family support was less clear for externalizing symptoms, increasing the risk for some adolescents and decreasing the risk for others. CONCLUSION We identified strong associations between family adversity and levels of psychiatric symptomatology. We found that Hawaiian adolescents are at increased risk for psychiatric symptomatology, given the presence of family adversity and the effect of reduced family support. However, risk was also substantial for non-Hawaiians. Clinicians need to assess the family environment routinely and implement family-oriented interventions.

Structural Equation Modeling of Group Differences in CES-D Ratings of Native Hawaiian and Non-Hawaiian High School Students.

This report uses different forms of latent variable structural equation models (SEMs) of group differences in self-reported Center for Epidemiologic Studies Depression (CES-D) inventory results among a sample of Hawaiian and non-Hawaiian high school students.The models examined are organized progressively by increasing complexity, including mean differences models in the ANOVA tradition, mean differences combined with path analysis models, extensions to confirmatory factor analysis models, and common factor differences in multiple group models.Within each type of analysis, SEMs are defined and fitted to real CES-D data, and broad issues of group differences and invariance are considered.The substantive results indicate a high degree of invariance between Hawaiian and non-Hawaiian students within the same high schools on self-ratings of depression from the CES-D scale.These analyses illustrate theoretical and practical issues about common features and relationships among standard analysis of variance approaches and contemporary structural equation methodology.

Prediction of anxiety disorders using the State-Trait Anxiety Inventory for multiethnic adolescents

The purpose of this study was to determine the validity of the State-Trait Anxiety Inventory (STAI) in predicting DSM-III-R anxiety disorders based on the Diagnostic Interview Schedule for Children (DISC, Version 2.3) and using Asian/Pacific Islander adolescents. An overall prevalence rate of 9.19% for generalized anxiety disorder, overanxious disorder, or social phobia was consistent with past studies. As hypothesized, STAI negatively worded (i.e., Factor 2) items were better predictors than positively stated (i.e., Factor 1) items. The STAI State mean was a better predictor of concurrent DISC anxiety disorders as compared to STAI State Factors I or 2. In contrast, the STAI Trait Factor 2 (negatively worded) composite was the best predictor for nonconcurrent DISC anxiety disorders as compared to STAI Trait Factor 1 or the overall STAI Trait subscale. Satisfactory predictive-validity values were obtained when using the STAI State mean and Trait Factor 2 composite. Implications of these findings are discussed, including using the STAI as a screening measure for ethnically diverse adolescents.

A conceptual model of cultural predictors of anxiety among Japanese American and part-Japanese American adolescents.

A model integrating Japanese ethnicity, cultural identity, and anxiety was developed and assessed in Japanese American and part-Japanese American high school seniors (N = 141). Using measures from the Hawaiian High Schools Health Survey, the model incorporated the State-Trait Anxiety Inventory, the Major Life Events Scale, and the Japanese Culture Scale (JCS). Japanese American adolescents scored higher on the JCS and reported fewer anxiety symptoms than part-Japanese American adolescents. Predictors for anxiety were being Japanese American versus part-Japanese American, income, and culturally intensified events. A significant interaction of behavior by self-identification was obtained. The model had good overall fit, suggesting that cultural identity formation may contribute to anxiety experienced particularly by adolescents of mixed heritage.

Variation in self-esteem among adolescents in an Asian/Pacific-Islander sample

Abstract Variation in global self-esteem among adolescents differentiated in terms of age, gender, Hawaiian/part-Hawaiian and non-Hawaiian ethnicity, self-reported grades (SRGs) and socioeconomic status (SES) was examined in the present study. Being older, male or non-Hawaiian and higher SRGs or SES were hypothesized to be associated with higher global self-esteem. Six hundred and ninety-six Hawaiian/part-Hawaiian and non-Hawaiian high school students provided responses to demographic items, the 10-item Rosenberg Self-Esteem Scale [RSES; Rosenberg, M. (1965). Society and the adolescent self-image . Princeton, NJ: Princeton University Press], and items assessing SRGs and SES. Adolescent males reported significantly higher self-esteem than adolescent females. Adolescents differing in SRGs and SES also varied significantly from one another in self-esteem, with the general pattern indicating higher SRGs and SES to be associated with higher self-esteem. The significant age-by-SRGs interaction effect suggested the consequence of academic success throughout the high school years. Findings regarding the present Asian/Pacific-Islander adolescent sample corresponded with that found for other adolescent samples of different ethnic compositions, which provided support for the generalizability of variation in global self-esteem and its noted correlates across various socio-cultural contexts.

Path models linking correlates of self-esteem in a multi-ethnic adolescent sample

Abstract Path analysis was performed to delineate the interrelations between state anxiety, aggression, substance abuse, depression, gender, ethnicity, family support, and self-esteem in a sample of 684 Hawaiian/part-Hawaiian and non-Hawaiian (e.g. Caucasian, Filipino, Hispanic, Japanese, or Mixed/two or more ethnicities) adolescents. The theorized model proposed that self-esteem was determined by family support and gender, and that family support, in turn, was predicted by state anxiety, aggression, substance abuse, depression, and ethnicity. The derivation of this model was based on previously documented findings of the associations between self-esteem and indices of adolescent adjustment, theorized multi-directionality of these interrelations, non-existence of an exhaustive theory accounting for the predictive links between these variables, and centrality of the ohana or family system and its integral network of support within the Native-Hawaiian culture. Goodness-of-fit indices indicated a poor fit between the theorized model and data. Subsequent model modifications were based on succeeding path analytical results, and conjointly encompassed an effort to find a model that fit the data well. Changes to the theorized model included eliminating three paths (i.e. from aggression and ethnicity to family support and from gender to self-esteem), and adding three paths (i.e. from depression, state anxiety, and aggression to self-esteem). The retained “final” Model 6 fit the data very well, and delineated the prediction of self-esteem by depression, state anxiety, aggression, and family support, which was, in turn, determined by depression, state anxiety, and substance use.

Prediction of DISC substance abuse and dependency for ethnically diverse adolescents

PURPOSE This study examines the validity of selected items from the Substance Abuse Subtle Screening Inventory-Adolescent (SASSI-A) version in predicting Diagnostic Interview Schedule for Children (DISC version 2.3) Substance abuse and dependency (SA/D) for Native Hawaiian (i.e., indigenous people of the Hawaiian Islands) and non-Hawaiian adolescents (youth without any Native Hawaiian indigenous ancestry). METHODS 542 students were randomly selected from the larger sample to participate in the DISC administration. Demographic information, SASSI-A scores, and DISC diagnoses were obtained for each student. Univariate and multiple logistic regressions were performed in the prediction of DISC SA/D. RESULTS SASSI-A Factor 1, consisting of three items measuring substance use, was found to have the best utility, accounting for 18.1% of the variance, in predicting DISC SA/D. IMPLICATIONS These results support selected SASSI-A items in screening for SA/D for Native Hawaiian and non-Hawaiian adolescents in Hawaii as compared to other community-based screening instruments for other populations.

Equivalencies Regarding the Measurement and Constructs of Self-Esteem and Major Life Events in an Asian-Pacific Islander Sampler.

Construct, scalar, and functional measurement equivalencies of the Rosenberg Self-Esteem Scale (RSES) and Major Life Events checklist (MLE) and the constructs assessed were investigated across groups differentiated on Hawaiian/part-Hawaiian and non-Hawaiian (e.g., Caucasian, Filipino, Hispanic, Japanese, and mixed/2 or more) ethnicity and gender. Initial results from maximum likelihood factoring with promax rotation showed that RSES negatively worded Item 5 loaded with the positively worded Items 1, 2, 4, 6, and 7 on 1 of 2 factors for Hawaiian/part-Hawaiian female participants. Similarly, negatively worded Item 8 and the same positively worded items comprised 1 of 2 factors for non-Hawaiian male participants. For the other 2 Ethnicity x Gender groups, factors were respectively comprised of the 5 positively and 5 negatively worded RSES items. Construct equivalence or simple (2-factor) structure underlying the RSES was indicated across the 4 groups after Items 5 and 8 were excluded from a subsequent factoring procedure. Simple structure showed that Factor 1 comprised the positively worded Items 1, 2, 4, 6, and 7, and the remaining negatively worded Items 3, 9, and 10 loaded on Factor 2. Scalar equivalence of the self-esteem and major life events measures was supported by the statistical nonsignificance of the Major Life Events x Ethnicity x Gender interaction effect in multiple regression models. The consistency in the absolute size and direction of the intercorrelations between overall self-esteem, self-esteem Factors 1 and 2, and major life events variables indicated the functional equivalence of respective measures and constructs assessed. Measurement equivalency findings concerning the RSES and MLE, the constructs measured, and their utility versus caution against their use in multiethnic studies were discussed.