You are on page 1of 30

U.S.

Cultural Involvement and its Association with Co-Occurring Substance Abuse and Sexual Risk Behaviors Among Youth in the Dominican Republic: A Latent Class Regression Analysis
Elin P. Cabrera-Nguyen, MSW Juan B. Pea, Ph.D., LCSW
Society for Prevention Research | June 3, 2011

Suggested citation
!

Cabrera-Nguyen, E. P., & Pea, J.B. (2011, June). US cultural involvement and its relationship to co-occurring substance abuse and sexual risk behaviors among youth in the Dominican Republic. Paper presented at the annual conference of the Society for Prevention Research, Washington, D.C.

Keywords: Acculturation, Hispanic populations, immigration, HIV risk behavior, substance abuse, latent class analysis, immigrant paradox

Acknowledgements
! This study is the result of a collaborative effort among

US and DR-based institutions and individuals including:


DR Ministry of Education ! Autonomous University of Santo Domingo ! Fundacin Familia Sana ! Center for Latino Family Research, Washington University in St. Louis
!

! This research was supported by NIMH Grant No. R03-

MH085203, PI: Juan B. Pea, Ph.D., LCSW

Purpose of Study
! Explore substance abuse and sexual risk behavior

profiles among adolescents in the Dominican Republic


! Examine how US cultural involvement relates to risk

behavior profiles among DR youth


! Situate findings within existing theory to inform future

prevention efforts with DR youth and US Latino/a adolescents.

Why study substance abuse and sexual risk behavior patterns among DR youth?
!

Substance abuse and sexual risk behaviors often co-occur among adolescents, placing them at increased risk for HIV Incidence of HIV among youth due to co-occurring substance abuse and sexual risk behaviors is a public health concern in the US and much of the world. Cooccuring substance abuse and risky sex among youth in DR requires further study. US Latino/a adolescents have higher substance abuse and sexual risk behavior rates compared to other ethnic groups & their foreign counterpartsHIV/AIDS is a leading cause of death Dominicans are one of the fastest-growing yet least studied Latino/a groups in the US, and nearly 30% are below 18 years of age

The Hispanic Immigrant Paradox


! Evidence suggests that a US nativity and being more

Americanized increases Latino/a adolescents risk for substance use, sexual risk behaviors, and a range of negative outcomes compared to their less Americanized foreign-born counterparts
!

Pea and colleagues (2008) classified hypotheses about the Latino immigrant paradox into 3 broad categories:
! Protective Culture Model ! Intergenerational Acculturation Conflict Model ! Resilient Immigrant Model

Methods

Sample
!

We used a novel approach to control for selection bias as an explanation for the Latino immigrant paradox:
!

Examined how US cultural involvement related to risk behaviors among youth in the DRoutside of the US context. Possibly via mechanisms related to cultural globalization.

Strata consisted of the DRs 18 national educational regions PSUs were public high schools (n = 80) Sample of youths (N = 8,446) weighted to adjust for grade, gender, and educational region.

! !

Measures
! U.S. cultural involvement (USCI): ! Non-Hispanic domain of Bidimensional Acculturation Scale (BAS) ! Two items from our study questionnaire ! Substance use and sexual risk behavior ! Items from 2009 Youth Risk Behavior Survey (YRBS) ! Demographic covariates

Two Latent Factors from BAS


! English Proficiency: 6 items, good EFA model fit

(CFI = .99, TLI = .98). Factor loadings ranged from .74 to .90
! Use of US Electronic Media and Language:

6 items, good EFA model fit (CFI = .96, TLI = .95). Factor loadings ranged from .67 to .86

USCI Indicators
! Time lived in the US: ! never (reference group)
! < 1 year ! > 1 year

! Five or more best friends lived in US at least one month

Demographic covariates
! Urban residence ! Parental education
! Parent(s) did not complete high school (reference) ! At least 1 parent completed high school ! At least 1 parent completed college

! Dual parent household ! Gender ! Age

Sample Characteristics
Variable Gender Female Male Region Urban Non-Urban Family structure Dual parent household Non dual parent household Parental Education No parent completed high school A parent completed high school A parent completed college Lived in US Never Less than 1 year More than 1 year At least 5 friends lived in US No Yes Age Weighted Percentage 57.0 43.0 67.7 32.3 52.0 48.0 45.8 30.9 23.3 90.9 6.2 2.9 82.4 17.6 Weighted Mean (SD) 16 (1.5) Unweighted Frequency 5205 3241 5718 2728 4391 4055 3834 2601 1979 7642 510 231 6914 1457

Latent Class Indicators


! Items from 2009 Youth Risk Behavior Survey (YRBS) ! Binge drinking, past 30 days
! Lifetime marijuana use ! Lifetime inhalant use ! Lifetime number of sex partners ! No condom use during last sexual intercourse

Statistical Analysis
! Determined baseline latent class model ! Single-step multinomial logistic regression of 4 USCI

predictors on latent classes


! Single-step multinomial logistic regression of USCI

predictors on latent classes adjusted for demographic covariates with direct effects

Results

Latent Class Extraction

Three Class Solution


! Bayesian Information Criterion (BIC) & Sample-Size

Adjusted BIC (SS-BIC) increased with 4th class, only Akaike Information Criterion (AIC) decreased
! Estimator = MLR ! Checked model identification, possible local maxima: ! 1000 sets of random starting values ! replication using Latent GOLD

Class Labels & Latent Class Assignment


!

Three distinct classes (Entropy = 0.90):


! ! !

Low risk for all behaviors (LR) Binge drinkers with risky sex behavior, no drugs (BD-RS) High risk for all behaviors (HR)

Average latent class probabilities for most likely latent class membership (row) by latent class (column) BD-RS BD-RS (0.29) HR LR (0.05) (0.66) 0.92 0.17 0.01 HR 0.05 0.80 0.01 LR 0.03 0.03 0.98

Conditional response probabilities for baseline latent class model

Indicator

Conditional Response Probabilities BD-RS Class (0.29)a High Risk Class (0.05)a 17.90 25.20 19.90 14.60 4.90 8.30 9.00 73.80 26.20 74.50 11.50 14.00 6.40 9.50 9.00 7.00 12.80 11.80 43.40 31.40 Low Risk Class (0.66)a 82.10 12.00 3.10 1.40 0.50 0.40 0.40 99.80 0.20 93.60 4.60 1.80 97.70 0.00 0.00 0.50 0.50 0.10 1.30 0.60

Binge drinking, past 30 days 0 days 1 day 2 days 3 to 5 days 6 to 9 days 10 to 19 days 20 days or more Any lifetime marijuana use no yes Lifetime inhalant use never 1 or 2 times 3 or more times Number of lifetime sex partners 0 1 2 3 4 5 6 or more No condom last time had sex
a

62.90 23.70 6.50 3.30 1.90 1.10 0.60 99.80 0.20 94.50 4.20 1.30 0 54.00 21.30 11.40 4.60 2.60 6.00 38.90

Percentage in each class based on estimated model

Conditional Response Probabilities

Conditional Response Probabilities

Multinomial Logistic Regression Results of Odds Ratios for Latent Class Assignment
Model 1: OR (95% CI) BD-RS vs. LR US Media, English English Proficiency 5+ friends lived US Lived in US < 1 yr1 Lived in US > 1 yr1 Urban residence Dual parent home Parenthigh school2 Parentcollege2 Age Female
1. Reference is never lived in US 2. Reference is no parent(s) finished high school * p< .05 **p<.01 p<.001

Model 2: OR (95% CI) BD-RS vs. LR 1.05 (0.99, 1.10) 0.98 (0.95, 1.00) 1.10 (0.92, 1.33) 1.38 (1.11, 1.71)* 1.51 (1.11, 2.06)* 0.91 (0.77, 1.08) 0.75 (0.83, 0.68)*** 1.17 (1.03, 1.32)* 0.81 (0.69, 0.95)* 1.84 (1.70, 1.99)*** 0.24 (0.21, 0.28)*** HR vs. LR 1.29 (1.08, 1.56)* 0.89 (0.82, 0.96)* 4.12 (2.74, 6.19)*** 6.68 (3.81, 11.70)*** 8.48 (4.08, 17.63)*** 1.04 (0.66, 1.64) 0.75 (0.52, 1.08) 1.35 (0.96, 1.90) 1.09 (0.75, 1.56) 2.30 (1.77, 2.98)*** 0.13 (0.08, 0.21)***

HR vs. LR 1.33 (1.12, 1.59)** 0.84 (0.78, 0.91)*** 4.51 (3.06, 6.65)*** 7.21 (4.46, 11.64)*** 9.87 (5.29, 18.42)***

1.05 (0.99, 1.12) 0.94 (0.91, 0.97)*** 1.10 (0.93, 1.30) 1.41 (1.15, 1.72)** 1.79 (1.36,2.35)***

Continuous latent factors in bold

Latent Class Proportions Across Models


Baseline BD-RS HR LR Entropy 0.29 0.05 0.66 0.90 Model 1 0.30 0.04 0.66 0.90 Model 2 0.31 0.05 0.64 0.91

Latent class regression model with latent and manifest covariates (and direct effects)
Binge Drinking Pot use Inhalant use Multiple sex partners Condom use

C
Age Gender 5 friends lived in US Lived in US < 1 year Lived in US > 1 year Urban Dual parent home A parent finished high school A parent finished college US Media & Language Use1 English Proficiency2

1. Standardized factor loadings range from .63 to .85, p < .001. Variance= 2.40 2. Standardized factor loadings range from .75 to .91, p<.001. Variance = 8.15.

Relationship of USCI to Substance Abuse & Sexual Risk Behavior Classes


! Odds of being HR versus LR were greater by a factor of

4.12 for youths with at least 5 friends who had lived in the US compared to youths with < 5 friends who had lived in the US.
! One unit increase in Use of U.S. Electronic Media and

English was associated with a 29% increase in the odds of assignment to HR versus the LR.
! Even after inclusion of direct effects of gender on

latent class indicators, the odds of being HR vs LR were 87% greater for males versus females.

Conclusions
!

Robust relationship of USCI with LC assignment consistent with acculturation related theories for the immigrant paradox while controlling for selection bias This finding is consistent with parallel analysis that found USCI substantially increases suicide behavior among youth in the DR. Findings suggest more research is needed to understand the mechanisms by which US culture increases risk for Latino/a youth as well as the salubrious aspects of Latino/a culturesboth in the US and abroad Results highlight need for culturally congruent prevention efforts targeting Latino/a youth in the US and abroad that are tailored to gender

Limitations
! ! ! ! !

Latent class analysis creates potential for reification of classes Cross-sectional design prevents causal inference Limited sociodemographic covariates Computational burden is severe Our novel methodological approach does not account for reverse selection; however, Use of US Electronic Media and English predicted HR class assignment independent of time lived in USan effect we would not necessarily expect if reverse selection explained assignment to HR class.

References
1. Connell CM, Gilreath TD, Hansen NB. A multiprocess latent class analysis of the co-occurrence of substance use and sexual risk behavior among adolescents. J Stud Alcohol Drugs. 2009;70(6):943-951. 2. Eaton DK, Kann L, Kinchen S, et al. Youth risk behavior surveillance -- United States, 2007. MMWR SURVEILLANCE SUMM. 2008;57(-4):1-131. 3. Flores G, Brotanek J. The healthy immigrant effect: A greater understanding might help us improve the health of all children. Arch Pediatr Adolesc Med. 2005;159:295-297. 4. Hair EC, Park MJ, Ling TJ, Moore KA. Risky behaviors in late adolescence: Co-occurrence, predictors, and consequences. Journal of Adolescent Health. 2009;45(3):253-261. 5. Langer LM, Tubman JG. Risky sexual behavior among substance-abusing adolescents: Psychosocial and contextual factors. Am J Orthopsychiatry. 1997;67(2):315-322. 6. Marn G, Gamba R. A new measurement of acculturation for Hispanics: The Bidimensional Acculturation Scale for Hispanics (BAS). Hispanic journal of behavioral sciences. 1996;18(3):297-316. 7. Pea J, Wyman P, Brown C, et al. Immigration generation status and its association with suicide attempts, substance use, and depressive symptoms among Latino adolescents in the USA. Prevention science. 2008;9(4):299-310. 10.1007/s11121-008-0105-x. 8. Prado G, Schwartz SJ, Pattatucci-Aragon A, et al. The prevention of HIV transmission in Hispanic adolescents Drug Alcohol Depend. 2006;84 Suppl 1:S43-53. 10.1016/j.drugalcdep.2006.05.006.

9. Torres LR, Pea JB, Westhoff WW, Zayas LH. A cross-national comparison of adolescent alcohol and drug use behaviors: U.S. Hispanics and youth in the Dominican Republic. J Drug Iss. 2008;38(1):149-170. 10. Westhoff WW, McDermott RJ, Holcomb DR. HIV risk behaviors: A comparison of U.S. Hispanic and Dominican Republic youth AIDS Educ Prev. 1996;8(2):106-114. 11. World Health Organization. Alcohol Use and Sexual Risk Behaviour: A Cross-Cultural Study in Eight Countries. Geneva: WHO Press; 2005.

You might also like