South African adult caregivers as “protective shields”: Serving as a buffer between stressful neighborhood conditions and youth risk behaviors

Abstract

Low-income youths in KwaZulu-Natal, South Africa, face elevated risks to their well-being from exposure to neighborhood conditions correlated with engaging in risky behaviors. These risks can be mitigated through adult caregivers who serve as protective shields, buffering adverse conditions. However, this protective role is dependent on the caregivers’ mental health and well-being. This secondary analysis uses baseline data from 475 child-caregiver dyads in an HIV-prevention program to examine the mediating effects of caregiver mental health on the relationship between neighborhood conditions and child risk-behaviors. Multivariate analyses identify the direct and indirect effects of neighborhood stressors and caregiver mental health on child risk-behavior. Findings suggest that caregivers mitigate the impact of neighborhood conditions on their children, but caregivers’ mental health is directly affected by neighborhood conditions. Therefore, caregivers’ mental health and well-being must be considered key elements in developing youth risk-behavior interventions.

1 |. INTRODUCTION

Poverty-impacted youths in KwaZulu-Natal (KZN), South Africa, are surrounded by serious threats such as community violence and the highest prevalence of HIV worldwide (Govender, Reardon, Quinlan, & George, 2014; Sawyer-Kurian, Browne, Carney, Petersen, & Wechsberg, 2011). Living in extreme poverty has elevated these youth’s risk for exposure to neighborhood violence, substance use, and instability. These conditions have the potential to not only exponentially increase risk of HIV (Parchment, Small, Osuji, McKay, & Bhana, 2016), but also encourage youth to perpetrate violence and engage in criminal activity. However, some of these community “toxins” can be mitigated by the presence of protective caregivers. Adult caregivers can serve as protective shields buffering their children’s exposure to stressors associated with living in dangerous communities.

The term adult protective shields refers to adults who buffer children against potential emotional and/or physical dangers, and actively engage in efforts to preempt conflict, reduce disorder, and prevent violence (McBride & Bell, 2011). Adult caregivers can mitigate the influence of adverse neighborhood conditions on child outcomes (Bhana et al., 2004; McKay et al., 2014; Parchment et al., 2016; Resnick, 2000). However, the caregivers’ own mental health and well-being are integral to their ability to provide a supportive environment that fosters the positive development of their children. To fill the protective shield role in KZN, caregivers must protect both their children and themselves from the constant threats of community violence.

There is significant U.S.-based research regarding the effect of neighborhood conditions on youth outcomes, and the buffering roles of caregivers (Forehand & Jones, 2003; Forehand, Lafko, Parent, & Burt, 2014; Jones, Forehand, O’Connell, Armistead, & Brody, 2005; Osofsky, 1995; Leventhal & Brooks-Gunn, 2000). Less is known regarding caregivers as buffering protective shields for at-risk youth, within the specific context of South Africa (Darling & Steinberg, 1993; Lochman, 2000). This study explores the direct and indirect effects of neighborhood conditions on child-risk behaviors for South African youth. The paper examines whether caregivers can serve as buffers between neighborhood conditions and child risk behaviors.

2 |. BACKGROUND

2.1 |. Adverse neighborhood circumstances and child well-being

Neighborhood conditions are significantly correlated with child behaviors (Jocson & McLoyd, 2015; Kessler et al., 2014). Given that 61% of South African youth between 15 and 24 years old live in low-income households (Statistics South Africa, 2014), many youth are likely to live in poverty-impacted communities where they have increased exposure to violence. Researchers have documented a strong relationship between poverty-impacted neighborhoods and social problems, including violence, crime, adolescent delinquency, and social/physical disorder (Côté-Lussier, Jackson, Kestens, Henderson, & Barnett, 2015; Krishnakumar, Narine, Roopnarine, & Logie, 2014; Sampson, Morenoff, & Gannon-Rowley, 2002). Several studies have also shown a correlation between living in violent communities and adverse youth outcomes, such as school disengagement (Borofsky, Kellerman, Baucom, Oliver, & Margolin, 2013; Schwartz, Kelly, Mali, & Duong, 2016), academic failure (Lepore & Kliewer, 2013; McCoy, Roy, & Sirkman, 2013; Schwartz & Gorman, 2003; Sharkey, Schwartz, Ellen, & Lacoe, 2014 ), mental health disorders (Cecil, Viding, Barker, Guiney, & McCrory, 2014; Ten Have, de Graaf, van Weeghel, & van Dorsselaer, 2014; Turner, Shattuck, Hamby, & Finkelhor, 2013), and high-risk behavioral health (Aisenberg & Herrenkohl, 2008; Cooley-Quille, Boyd, Frantz, & Walsh, 2001; Gorman-Smith & Tolan, 1998; Osofsky, Wewers, Hann, & Fick, 1993; Voisin, Patel, Hong, Takahashi, & Gaylord-Harden, 2016). This relationship has been explained at the community level as the result of concentrated poverty (Morenoff, Sampson, & Raudenbush, 2001; Sampson, 2001), weak social ties, low collective efficacy, and low social control (Sampson, Jeffrey, & Gannon-Rowley, 2002).

The social disorganization theory posits that communities with high rates of poverty, residential mobility/instability, and ethnic heterogeneity are less socially cohesive and less likely to effectively exert informal social control; therefore, these communities are more likely to experience social and physical disorder (Bruinsma, Lemoine, Biancale, & Vales, 2010; Pattillo, 2000). Conventional institutions of social control, including family, schools, churches, and volunteer organizations, tend to be among the weaker community influences and unable to regulate the behavior of the youth residing in these neighborhoods. Limited social cohesion and control also impede parents from actively engaging in supporting and controlling adolescent risk behaviors (Akers, Muhammad, & Corbie-Smith, 2011; Browning, Soller, & Jackson, 2015; Chuang, Ennett, Bauman, & Foshee, 2009; Kerrigan, Witt, Glass, Chung, & Ellen, 2006; Sampson, Morenoff, & Earls, 1999; Sampson, Raudenbush, & Earls, 1997). Consequently, this social disorganization is associated with high incidence of youth problem behavior (Aneshensel & Sucoff, 1996; Hill et al., 2000; Leventhal & Brooks-Gunn, 2000). Social disorganization in neighborhoods results in lower access to resources, such as health care and retail stores, as well as poor quality of available resources, such as public spaces or recreational facilities. Moreover, lack of access to needed resources becomes an additional stressor because of the extra effort required to meet daily needs (Cutrona, Wallace, & Wesner, 2006; Sampson et al., 2002). Consequently, an impoverished, unsafe neighborhood can influence parents’ mental and behavioral health, which in turn, can affect child outcomes through parenting. In addition, it can affect access to resources that, when available, can reduce youth vulnerability to sexually transmitted infections, including HIV/AIDS (Maas, Patch, Christian, & Coplan, 2007; Szwarcwald, Bastos, Fonseca, Esteves, & Andrade, 2000).

Adverse neighborhoods are often areas of concentrated poverty with poor-quality resources and little or no integration with more prosperous families. As a result, a wide range of social problems will appear, including threats to mental health. Moreover, despite homogenous characteristics of a population such as shared language and culture, little is known about social cohesion and control in lower-middle income countries with homogenous populations, such as KZN, South Africa, where people are predominantly isiZulu speakers from the Zulu tribe (Statistics South Africa, 2011).

2.2 |. Child risk taking behavior in South Africa

Youth who experience adverse conditions that interfere with their emotional, social, and physical development face a higher risk of experiencing behavioral challenges (Cortina, Sodha, Fazel, & Ramchandani, 2012; Grantham-McGregor et al., 2007, 2007a; Hackett, Hackett, Bhakta, & Gowers, 2000; Rudolph & Flynn, 2007; Rutter, 1999; Shaw, Vondra, Hommerding, Keenan, & Dunn, 1994; Thabet & Vostanis, 1998). In South Africa, approximately 1 in 5 youth experience some form of mental or behavioral health challenges (Kleintjes, Lund, & Flisher, 2010). Children who exhibit early externalizing behaviors are more likely to have increased rates of alcohol-related problems, illicit drug use (Fergusson, Horwood, & Lynskey, 1995), and increased odds of engaging in risky behaviors such as sexual risk-taking (Fergusson, Horwood, & Ridder, 2005; Fontaine et al., 2008; Parkes et al., 2014; Timmermans, van Lier, & Koot, 2008; Wu, Witkiewitz, McMahon, & Dodge, 2010). In addition, youth who exhibit destructive childhood behavioral patterns (e.g., lying, cheating, yelling, public disobedience, and physical violence; Frick, 2012) are more likely to have an elevated risk for experimenting with illicit substances as they enter adolescence and adulthood (Hopfer et al., 2013).

2.3. |. Caregivers as protective shields for youth

Caregivers have been shown to be a shield for their children, protecting them from community stressors that can adversely affect their children’s development (Barakat, 2008; Dutra et al., 2000). Therefore, the quality of the relationship between the caregiver and the child plays an important role in promoting youth prosocial behavior, especially among HIV-positive adolescents (Forehand et al., 2000; Mellins et al., 2008; Wallander & Varni, 1998).

Adult protective shields can include parents, grandparents, caregivers, and community members who act as mentors to protect youth from potentially engaging in risky behaviors (Bell & McBride, 2010; McBride & Bell, 2011). Research has shown that decreased behavioral risks for youth was associated with the positive behavior of the youth’s caregivers; that is, caregivers who supervised and monitored child behavior while providing the child with moderate to high levels of acceptance, affection, and warmth were associated with low risk for the youth engaging in negative, risky behaviors (Bell, Bhana, McKay, & Petersen, 2007).

Studies have found that the quality of the caregiver-child relationship is an essential component in the social and emotional development of the child (Parchment et al., 2016; Walker, Wachs, Grantham-McGregor et al., 2011; Walker, Wachs, Meeks et al., 2007). The quality of the caregiver-child relationship is especially at risk for families in South Africa impacted by HIV/AIDS because these families are at increased risk of poverty (Gillies, Tolley, & Wolstenholme, 1996; Lachman, Cluver, Boyes, Kuo, & Casale, 2014), lack of social support (Casale & Wild, 2012; Lachman et al., 2014), parental depression (Kuo, Operario, & Cluver, 2012; Lachman et al., 2014), and decreased child prosocial behaviors (Lachman et al., 2014; Sipsma, Eloff, Makin, & Forsyth, 2013)—all of which are risk factors with the potential to compromise the quality of parental care.

2.4 |. Caregiver mental health and its impact on parenting

In South Africa, the combination of historical social stressors such as apartheid and contemporary social stressors such as poverty and HIV/AIDS has negative effects on the well-being of caregivers and their capacity to parent, which can disrupt a young child’s healthy development (Belsky, 1984; Lachman et al., 2014; Parchment et al., 2016). When caregivers are chronically stressed, they are more likely to be socially isolated and to experience increased mental health and relationship problems (Belsky, 1993; Conger, Conger, & Martin, 2010; Lachman et al., 2014).

Caregiver distress negatively affects their capacity to serve as protective shields, which leads to diminished parent-child relationships (e.g., more distant, less warm) and adoption of authoritarian parenting styles and harsh parenting practices (Belsky, 1993; Conger et al., 2010; Lachman et al., 2014). The weakening of the protective shield is accompanied by a significant level of parental disempowerment, poor parent–child communication, punitive parenting practices, and poor parental monitoring, all of which increase the likelihood of youth engaging in risk behaviors (Bell et al., 2008).

3 |. THEORETICAL FRAMEWORK

The ecological mediational model (Felner, Brand, DuBois, & Adan, 1995; Felner & DeVries, 2013 ) provides a helpful framework for elucidating how conditions such as poverty and socioeconomic disadvantage can affect child outcomes. Specifically, this model allows for examining the impact of poverty on proximal experiences that impact children and youth through the direct effects of socioeconomic disadvantage. In this study, the ecological mediational model was used to understand how poverty and its associated disadvantages (e.g., neighborhood conditions) impact child/adolescent engagement in risk behavior. While socioeconomic disadvantage may influence child risk behavior directly, its impact may also be mediated through more proximal influences (e.g., caregiver mental health). Disentangling these relationships can help inform policies and interventions that aim to address child risk behaviors.

4 |. METHODS

This study involved a secondary analysis of baseline data from the Collaborative HIV Prevention and Adolescent Mental Health Program in South Africa (CHAMPSA; Bell et al., 2008; Bhana, McKay, Mellins, Petersen, & Bell, 2010). CHAMPSA is based on the CHAMP program that was originally developed as a U.S.-based family intervention to strengthen parent–child relationships among caregivers and preadolescents. CHAMP was previously implemented in New York, Chicago, as well as Trinidad and Tobago, and then adapted for South Africa. A key concept of the intervention is the caregiver’s role as an adult protective shield, buffering negative community factors, and thereby, helping to reduce the risk of child HIV infection (Petersen, Mason, Bhana, Bell, & McKay, 2006).

4.1 |. Sample characteristics

The CHAMPSA data used in this study were collected from 475 child-caregiver dyads participating in an HIV prevention program at four sites outside of Durban, South Africa. To be eligible for study inclusion, children had to be enrolled in school and living with an adult caregiver who was able to provide consent. Children were also required to provide assent (Petersen et al., 2006).

No youth were excluded from this analysis (n = 475). The children’s age ranged from 9–13 years old (M = 10.97; SD = .85), and 60% of the sample population was female. The majority of the children were in Grades 4 and 5. Sixty seven percent of children lived with their mother, the remaining reported living with other relatives (father, aunt/uncle, grandparent). Caregiver reports indicated an average household size of three persons. All households were low income, with 58% of caregivers unemployed and 51% receiving disability grants. Most caregivers had completed primary school, and all participants spoke isiZulu as their primary language (Table 1).

TABLE 1.

Caregiver and child demographics

Child Characteristics, n = 475	%
Lives with mother	67
Child gender
Boy	40.2
Girl	59.6
Child age (yr)
9	1.3
10	32.8
11	34.5
12	30.3
13	1.1
Child grade
2	1.9
3	9.5
4	24.8
5	35.6
6	18.7
7	9.3
Caregiver characteristics n = 475
Education	%
Never attended high school	19
Primary school	46.4
High school	34.4
Tertiary school	0.2
Gender
Male	52.5
Female	47.5
Employed
Yes	41.4
No	58.6
Disability grants
Yes	48.8
Pension
Yes	48.8

4.2 |. Measures

Seven measures were used to collect data from caregivers.

4.2.1 |. General health questionnaire

The general health questionnaire is a 12-item, short-form self-administered adult mental health screener. It captures the respondent’s mental health regarding symptoms of anxiety, depression, and social dysfunction (Goldberg & Hillier, 1979). A total score of the GHQ was obtained, and items were reversed scored with a higher score indicating higher GHQ scores (4 = better than usual, 1 = much worse than usual). Reliability scores from a comparable sample of South African adults yielded an alpha coefficient of 0.85 (Pernice, Trlin, Henderson, & North, 2009).

4.2.2 |. Global indicator of well being

The global indicator of well being is a single item question asked of the adult caregivers to assess how they feel about their life as a whole. Responses are given on 7-point scale: terrible ( = 1), unhappy ( = 2), mostly dissatisfied ( = 3), about equally satisfied and dissatisfied ( = 4), mostly satisfied ( = 5), pleased ( = 6), and delighted ( = 7).

4.2.3 |. Child behavior checklist (CBCL)

The Child Behavior Profile (Achenbach, 1978) is a 37-item measure designed to be filled out by parents/caregivers to assess their child’s social competence and behavior problems. Caregivers score items on a three-step response scale, with 1 indicating that the item is not true of the child, 2 indicating that the item is sometimes true of the child, and 3 indicating the item is very true of the child (Achenbach, 1978). At baseline, the Child Behavior Profile had a Cronbach’s alpha of 0.89.

4.2.4 |. Neighborhood social control

Neighborhood social control (Sampson et al., 1997) is a 6-item measure that assesses the experience of informal social control within the respondent’s community. Caregivers were asked whether their neighbors could be counted on to intervene in the following situations: (a) children were not in school and hanging out on the street; (b) children showing disrespect to an adult; (c) children fighting in the street; (d) boys harassing children walking on the street; (e) an adult male known to be selling drugs from his house; and (f) an adult male known to be abusing girls. The measure captures responses on a 5-point scale, ranging from very unlikely (1) to very likely (5), with higher scores indicating better informal social control. At baseline, the Neighborhood Social Control measure had a Cronbach’s alpha of 0.87.

4.2.5 |. Neighborhood social cohesion

A neighborhood’s level of social cohesion can influence community members’ sense of connectedness and promote the adoption of health-related behaviors (Sampson et al., 1997). The Neighborhood Social Cohesion Scale (NSC) is a 13-item measure that assesses a caregiver’s sense of community connectedness. Response options range from strongly disagree (1) to strongly agree (4), with a higher score representing greater cohesion. At baseline, the NSC Cronbach’s alpha was 0.85.

4.2.6 |. Neighborhood disorganization

To measure the level of community disorganization and violence experienced by the caregiver, the CHAMPSA study used the Neighborhood Disorganization Scale (Sampson et al., 1997). This 8-item scale measures specific problems that might arise in the caregiver’s community such as (a) litter or garbage on the pavements and streets; (b) graffiti on buildings and walls; (c) shacks erected on vacant lots; (d) drinking in public; (e) robbery and burglary; (f) selling drugs and alcohol from houses; (g) taking over other people’s property; and (h) fighting in the street. Responses are captured with a 3-point scale ranging from often ( = 1) to never ( = 3), with a Cronbach’s alpha of 0.76 at baseline.

4.3 |. Data analysis

Descriptive analyses were conducted to provide estimates of caregiver and child demographics (Table 1). Mediation modeling was conducted. Guidelines from Baron and Kenny (1986) were implemented, followed by a Sobel’s test to examine the strength of the indirect effect of neighborhood conditions on child risk behaviors and “confirm the mediation effect”. Multivariate linear regression analysis examined the role of caregiver mental health as a mediator between neighborhood conditions and child risk behaviors (Figure 1, Table 2).

FIGURE 1.

Model representing the direct and indirect effect of neighborhood conditions on child risk behaviors

TABLE 2.

Mediation analysis: Neighborhood conditions and child risk behaviors mediated by caregiver mental health

Path	β (unstandard)	SE	β (standard)	Significance
c	0.086	0.028	0.138	0.003**
a	0.328	0.051	0.285	0.000**
b	0.078	0.025	0.142	0.002*
c’ (m)	0.061	0.026	0.113	0.018*

Abbreviation: SE, standard error.

^*p < .05.

^**p < .001.

As seen in Figure 1, three linear regression models were used to test the effect of the mediator (caregiver mental health, M). First, the independent variable (neighborhood conditions, X) was regressed on the dependent variable (child risk behavior, Y) with a regression coefficient. Second, the independent variable (neighborhood conditions, X) was regressed on the mediation variable (caregiver mental health, M) with a regression coefficient a. Third, caregiver mental health (M) was regressed onto the dependent variable (child risk behavior, Y), with a regression coefficient b. Fourth, both the independent variable (X) and the mediation variable (M) were regressed on the dependent variable (Y), with regression coefficients c′ (X on Y) and b (M on Y). In addition, a test by Sobel (1982) was conducted to check the strength of the indirect effect on child risk behavior on neighborhoods, and thereby confirm the mediation effect. Further, the significance of the indirect effect was tested using macros developed by Preacher and Hayes (2008) for mediation analyses and bootstrapping procedures.

5 |. RESULTS

Mediation analyses were used to measure the effect of neighborhood conditions on child risk behaviors with caregiver health as a mediator. The analysis showed a statistically significant indirect effect of neighborhood conditions on child risk behaviors. Caregiver general health partially mediated the effect of neighborhood conditions on child risk behaviors. The conditions established by Baron and Kenny (1986) were met and produced the partial mediation model when the strength of the relationship between neighborhood conditions and child risk behaviors remained significant, but decreased (from b = 0.086; p = .003 to b = 0.066; p = .027) when the mediator and caregiver mental health was introduced. This partial mediation was confirmed by the indirect effect described by Sobel (1982); Table 3). Unstandardized indirect effects were computed for each of the 10,000 bootstrapped samples, and the 95% confidence interval (CI) was computed. The total effect of neighborhood conditions on child risk behaviors was 0.086 (p < .01), whereas the direct effect was neighborhood conditions on child risk behaviors was .066 (p < .05). The bootstrapped unstandardized indirect effect was 0.020, and the 95% bias corrected bootstrap CI [.0036, 0.0420]. Therefore, the difference between total and indirect effects of neighborhood conditions on child risk behaviors is different from zero. Better neighborhood conditions are associated with better caregiver health, which in turn, are associated with lower child risk behaviors (Preacher & Hayes, 2008). Thus, the indirect effect was statistically significant (Table 2).

TABLE 3.

Sobel’s test for indirect effect

Indirect effect (axb)	0.026
Sobel test statistic	2.807	.009(SE)
Two-tailed prob	.005	p-Value

Abbreviation: SE, standard error.

Specifically, mediation analysis showed significant indirect effects of neighborhood conditions on child risk behaviors ((0.33)(0.08) = 0.03). The significance was tested using bootstrapping procedures. Caregiver mental health partially mediated the effect of neighborhood conditions on child risk behaviors. Improved neighborhood conditions were associated with better caregiver emotional health (β = 0.33; SE = .05; p < .05), which was associated with lower child risk behaviors (β = 0.08; SE = .03; p < .001). After controlling for the mediator, neighborhood conditions remained significantly associated with child risk behaviors, consistent with partial mediation (β = 0.06; SE = .03; p < .05). Multiple regression revealed that neighborhood conditions had statistically significant associations with caregiver emotional health (β = 0.33; SE = .05; p < .001) and caregiver overall wellbeing (β = 0.86; SE = .02; p < .001; Table 4).

TABLE 4.

Multiple regression of neighborhood conditions on caregiver mental health and overall wellbeing

	GHQ				GIWB
	B (Unst)	SE	β Std	p	B (Unst)	SE	β Std	p
Neighborhood conditions	0.328	0.051	0.285	.000**	0.856	0.0167	0.23	.000**

Abbreviatons: GHQ, general health questionnaire; GIWB, global indicator of well being; SE, standard error.

^**p < .001.

6 |. DISCUSSION

The data used in this study were drawn from a National Institute of Mental Health-funded longitudinal study of youth and their caregivers living in communities in KZN, South Africa, with exceedingly high HIV seroprevalence. There is a growing consensus that the youth living in poverty face greater vulnerability to societal toxins that increase their engagement in risky behaviors (Brook, Morojele, Zhang, & Brook, 2006; Parchment, 2018; Romer et al., 1999; Yoshikawa, Aber, & Beardslee, 2012).

There is significant U.S.-based research regarding the effect of community environment on youth outcomes, and the buffering roles of caregivers (Forehand & Jones, 2003; Jones et al., 2005; Osofsky, 1995; Leventhal & Brooks-Gunn, 2000). In understanding parenting and child well-being, particularly in the South African context, examination of the structural influences such as neighborhood conditions, can provide a better picture of the various forces impacting families. Consistent with the literature (Yoshikawa et al., 2012), our findings show that neighborhood functioning has a direct impact on youth behavior. Neighborhoods with increased chaotic conditions (e.g., lack of cohesion and social control, greater disorganization) have been linked to children being more likely to engage in risky behaviors, including theft, violence, and truancy (Aisenberg & Herrenkohl, 2008; Yoshikawa et al., 2012). However, in this study, this relationship was partially mediated by caregiver mental health. In other words, caregiver mental health mediated the relationship between youth and their communities, suggesting that caregivers can shield children from the risks associated with neighborhood conditions when they are in good emotional health. The presence of healthy caregivers can mitigate youth’s exposure to community stressors, and therefore, the likelihood that youth will engage in negative behaviors.

The bond between a child and their primary caregiver mediates their understanding and conceptualization of the world around them (DeBellis et al., 1999; McKay et al., 2010; Spannring, 2012). This helps the child feel safe and secure, which improves his functional development. When the caregiver has a strong support system, it also improves their capability to parent effectively (Casale et al., 2015; Lindsey et al., 2008, 2012; Ward, Gould, Kelly, & Mauff, 2015). This remains salient even if they reside in areas that are impacted by poverty (McKay et al., 2014; Morris et al., 2017).

Consistent with the existing literature on the social disorganization theory (Leventhal & Brooks-Gunn, 2003; Ross, 2000; Silver, Mulvey, & Swanson, 2002), this study found that neighborhood conditions impacted adult mental health. As these conditions improved, caregivers reported better emotional health and an improved sense of overall well-being. Therefore, while attempting to protect their children, these adults also had to navigate through the effects of neighborhood conditions on their own mental health and well-being.

Caregivers’ mental health is integral to their ability to provide a supportive environment that fosters optimal child development and helps guide their children in avoiding risk taking. Hence, fostering the adult protective shield for youth living in poverty and HIV-affected communities can reduce the likelihood of children engaging in risky behaviors and aid in the prevention of HIV infection. These findings related to the impact of caregiver well-being have care implications as well as implications for global HIV intervention research because the development of HIV interventions will benefit from addressing the mental health needs of both children and caregivers.

South Africa continues to have the highest HIV rate worldwide, and KZN remains at the forefront of HIV transmission (http://aidsinfo.unaids.org; UNAIDS, 2011). Initiatives aimed at reducing new infections and eradicating AIDS (South African National AIDS Council, 2011; UNAIDS, 2011) should give attention to both caregivers and youth. Supporting caregiver health beyond pregnancy can have a positive ripple effect because caregiver mental health and overall well-being are significantly associated with the behavioral health of their children.

The study findings suggest that programs and services in these contexts should target both children and caregivers. Behavioral change in families is unlikely to be sustained when only caregivers or children are the sole target of interventions and initiatives. While creating opportunities for children that will reduce their likelihood of engaging in risky behaviors, programs should simultaneously seek to bolster and support caregiver mental health as a means of more effectively addressing risk behaviors among children living in poverty-impacted communities with low social cohesion and high disorganization. Given the negative impact of neighborhood circumstances on both children and families, efforts should also continue to target adverse community conditions. Fostering the adult protective shield (caregiver) for youth living in HIV-impacted low-income communities has implications for global HIV intervention research and care because the presence of positive adults can reduce the likelihood of children engaging in risky behaviors and aid in the prevention of HIV transmission.

This study has certain limitations that must be considered. The sample is not generalizable to represent children and families throughout KZN. This study examined the health of caregivers in the home, but does not represent the caregiver’s engagement with the child. Additionally, the original data set may have omitted variables that could reveal other influences on child outcomes. This cross-sectional data cannot conclusively demonstrate causality. Hence, future studies should use longitudinal data to establish causality. Moreover, future research should also include measures to assess the child-caregiver relationship and the level of the caregiver’s involvement with the child. Despite these limitations, the study results from youth in neighborhoods heavily impacted by HIV in South Africa adds to the existing body of research that show the complex interchange between neighborhood conditions, caregiver mental health and overall well-being, and child risk behaviors. It contributes to building knowledge regarding these relationships for families in South Africa. Future longitudinal research should address the needs of caregivers when investigating youth outcome. In addition, future research should include measures to assess the child-caregiver relationship and the level of the caregiver’s involvement with the child.