Hispanic adolescents, the largest and fastest growing minority group in the US, often report higher rates of HIV risk behaviors, relative to their non-Hispanic white counterparts [6]. For example, compared to their non-Hispanic white counterparts, Hispanic adolescents are more likely to report not having used a condom during their last sexual intercourse [7] and twice as likely to get a sexually transmitted infection. Moreover, Hispanics have the highest prevalence rate of teenage pregnancy among all major ethnic groups [8, 9]. In addition to these sexual risk behaviors, Hispanic adolescents are also at increased risk for substance use. In fact, Hispanic 8th and 10th grade youth report the highest use of both licit and illicit substances among all major ethnic groups, with the exception of amphetamines [10]. This is especially disconcerting given that substance abuse and HIV/AIDS are intertwined epidemics with 40 % of US. AIDS deaths related to drug abuse [11]. For example, 22 % of those who are sexually active report engaging in sexual activity while high or drunk, increasing their risk for contracting sexually transmitted infections including HIV [1]. In addition, early onset of substance use has been associated with future adult drug dependence, which in turn increases the likelihood of HIV infection [6].

Delinquent youth are also at increased risk for engaging in HIV risk behaviors when compared to the general adolescent population. For example, delinquent youth report a higher prevalence of being sexually active and unprotected sex, which in turn places them at greater risk for HIV infection [12]. When coupled with substance use, delinquent youth report 5 or more HIV risk behaviors including unprotected sex [12]. Hispanic delinquent youth are also at greater risk for engaging in HIV risk behaviors. Hispanic delinquent youth, for example, report higher rates of substance use when compared to African Americans, but similar rates to non-Hispanic whites [13], which in turn places them at greater risk for HIV infection [6].

Acculturation is traditionally defined as a unidimensional process and the degree to which a person moving from one culture to another adopts the new culture, typically measured through language preference [14, 15]. More recently, however, researchers have conceptualized acculturation as a bi-dimensional process that includes both the adoption of host-country culture (e.g., customs, values, language) and the retention of country-of-origin culture. The extent to which Hispanics’ level of orientation is toward American and Hispanic culture refers to Americanism and Hispanicism, respectively [16]. One of the more widely utilized bi-dimensional models of acculturation is Berry’s [17] acculturation framework. According to Berry, there are four typologies of acculturation: (1) marginalized (i.e., low Americanism and low Hispanicism); (2) separated (i.e., low Americanism, high Hispanicism); (3) assimilated (i.e., high Americanism, low Hispanicism); and (4) integrated (i.e., high Americanism, high Hispanicism). Within Berry’s model, the integrated typology represents biculturalism, and this group is postulated to be a low-risk group because of high orientation towards both American and Hispanic cultures [17]. Research has shown that rates of HIV risk behaviors (including early sex initiation and substance use) in Hispanic youth vary as a function of acculturation level [18, 19]. Specifically, Hispanic youth with higher levels of Americanism have shown higher rates of HIV risk behaviors, relative to youth with lower Americanism levels [14, 18, 20]. A limitation of these studies, however, is that they examined acculturation as a unidimensional model. That is, these studies have only looked at how acculturation (i.e., Americanism) is related to HIV risk behaviors. In fact, relatively few studies have examined differences in problem behaviors by utilizing a bi-dimensional model of acculturation [20]. In one study, utilizing a bi-dimensional model of acculturation, Sullivan et al. [20] found that integrated adolescents who both maintain country-of-origin culture practices (i.e., Hispanicism) and adopt host-culture practices (i.e., Americanism) reported the highest levels of parental involvement, positive parenting, and family support. In contrast, assimilated adolescents who adopt hostculture practices and do not retain country-of-origin culture practices reported the greatest levels of aggressive behavior.

Many Hispanic cultures emphasize the importance of family and interconnected relationships [21]. Previous studies have highlighted the mechanism by which family functioning, defined as family cohesion, parental involvement, and family communication, may mediate the effects of acculturation on HIV risk behaviors, including substance use [15, 22–26]. For example, Martinez [27] found that family functioning mediated the effects of acculturation (defined as Americanism) on substance use. Again, Martinez’s [27] article was limited in that the measure of acculturation was unidimensional. Whether, and the extent to which, family functioning mediates acculturation typology (i.e., using a bi-dimensional acculturation model) on HIV risk behaviors (including early sex initiation and unprotected sex), however, is not well understood. To the best of our knowledge, there are no published studies examining whether HIV risk behaviors (including unprotected sex, early sex initiation, and substance use) vary by acculturation typology and whether family functioning mediates these effects.

The purpose of this study was to examine whether HIV risk behaviors vary as a function of acculturation typology and whether, and to what extent, family functioning mediates the relationship between acculturation typology and HIV risk behaviors in a sample of Hispanic delinquent youth. Acculturation typology is hypothesized to be related to HIV risk behaviors. Specifically, we hypothesized that marginalized and assimilated adolescents will report higher rates of HIV risk behaviors when compared to integrated and separated youth. Furthermore, family functioning is hypothesized to mediate the effects of acculturation typology on HIV risk behaviors.

Participants were recruited from August 2009 to February 2010 from either the Juvenile Justice System (8.9 %) or from public schools in Miami-Dade County (91.1 %). Youth were eligible to enroll in this study if they: (a) were female or male adolescents of Hispanic immigrant origin, defined by at least one parent born in a Spanish-speaking country of the Americas, (b) were between the ages of 12–17, and (c) had an adult primary caregiver who was willing to participate in the study. Additionally, youth had to meet the study’s definition of delinquency, which was defined as having committed one or more offenses or as having committed at least one ‘‘Level III Behavior Problem’’ described by Miami Dade County Public Schools as assault/threat against a non-staff member, breaking and entering/burglary, fighting (serious), hazing, possession or use of alcohol and/or controlled substances, possession of simulated weapons, trespassing, and vandalism. Of the 446 potential participants, 310 were contacted and screened. Of those, 2421 were assented (parents also consented) and were enrolled. The current study is a secondary analysis of this data.

Participants were 235 Hispanic adolescents (152 boys, 83 girls) with a mean age of 14.70 years (SD = 1.40). Sixty-four percent of adolescents were born in the US and immigrant adolescents were predominantly from Cuba (25.0 %) and Honduras (15.5 %). The majority of adolescents had resided in the US for over 10 years (67.2 %), 23.0 % reported living in the United States between 3 and 10 years, and 9.8 % less than 3 years. About 60 % of participants came from families with a total yearly income of less than $20,000 (Table 1).

A survey assessment was conducted in the language preference of the adolescent (i.e. English or Spanish) using an audio computer-assisted self-interviewing (CASI) methodology. Each item on the questionnaire (along with scale response) was read to the adolescent through the use of headphones. The adolescent then indicated her or his response on a laptop computer, moving on to the next question after their response was entered. Families were compensated $60 for their participation in the assessment.

Tests of the study hypotheses proceeded in three steps. First, a measurement model [32] was estimated to discern the feasibility of collapsing three subscales of family functioning (i.e., parental involvement, family cohesion, and parent-adolescent communication) into a single latent variable. Second, we estimated a structural equation model (Fig. 1) to test whether acculturation typology was related to HIV risk behaviors. One structural equation model was tested for each of the HIV risk behaviors. Adolescent’s age, gender, nativity status (US born or foreign born) and household income were controlled for in the models as covariates for outcomes. Given that within Berry’s [17] acculturation typology, integrated typology refers to biculturalism and is hypothesized to be low-risk, the integrated group was used as the reference group. Therefore, each acculturation typology was compared to the integrated group. Third, we tested whether family functioning mediated any effects of acculturation typology on the HIV risk behaviors. Using Mplus version 6.12 [33], MacKinnon’s asymmetric distribution of products test in each of the structural equation models (i.e., one test for each of the HIV risk behavior outcomes) was utilized, which creates asymmetric confidence intervals for the mediated effect. Mediation is assumed when the confidence interval for the product does not include 0. Prior to testing the study hypotheses using the three steps described above, demographic characteristics such as age, family income, nativity status, and family functioning were compared across acculturation typology.

Analysis yielded four typologies of acculturation: (1) Marginalized (n = 6; 2.6 % of the study sample); (2) Separated (n = 9; 3.8 % of the sample); (3) Assimilated (108; 46.0 % of the sample), and; (4) Integrated (112; 47.7 % of the sample).

Analysis of variance and Chi-square tests were conducted to examine significant differences in age, gender, nativity status and years in US by acculturation typology (Table 1). Significant differences in nativity status and years residing in the US by acculturation typology emerged. Specifically, separated adolescents were more likely to be foreign-born than adolescents in any of the other acculturation typologies (Table 1). Family functioning variables, including parental involvement and family cohesion, differed across acculturation groups (Table 1), suggesting a relationship between acculturation typology and family characteristics.

A confirmatory factor analysis was performed within a structural equation modeling framework to confirm that all three subscales of family functioning, including parental involvement, family cohesion, and family communication, significantly loaded onto a single latent variable of family functioning (all p’s < 0.05). The standardized factor loadings were 0.61, 0.86, and 0.81 for parental involvement, family cohesion, and family communication, respectively. Because this model was saturated (with no degrees of freedom), fit statistics are not reported.

All of the structural equation models (i.e., across all of the outcomes) provided an adequate fit (comparative fit index, or CFIs, ranged from 0.962 to 0.978 and root mean square errors of approximation, RMSEAs, ranged from 0.026 to 0.034 for all outcomes). Path coefficients in structural equation modeling statistics can be thought of as analogous to regression coefficients in simple or multiple regression, where regression coefficients represent the change of outcome with 1 unit of change in a predictor variable while holding all covariates constant. The results for each of the hypothesized relationships are reported below.