Abstract
Black South Africans are disproportionately affected by HIV compared with White counterparts. In their unique social context, South African families affected by HIV are vulnerable to adverse psychosocial effects. U.S.-based and emerging South African research suggests mothers living with HIV may experience compromised parenting. In the United States, mother-child relationship quality has been associated with internalizing (anxiety, depression) and externalizing (delinquency, acting out) child behaviors. This study adds to South African research with emphasis on the role of the mother-child relationship among HIV-affected South Africans from multiple communities. Structural equation modeling examined relationships between maternal health and child adjustment, operating through mother-child relationship. The best-fitting model suggested maternal health influences youth externalizing behaviors through the mother-child relationship, but that maternal health is directly related to child internalizing problems. Findings support and extend previous results. Further research would benefit from investigating ways the unique South African context influences these variables and their interactions.
Keywords
HIV affects many families in South Africa, with approximately 7.1 million South Africans living with the virus, 56% of whom were on antiretroviral treatments and, thus, better able to continue to care for their children (UNAIDS, 2016). Given the context of poverty, stigma, and violence, South African families affected by HIV are particularly susceptible to adverse psychosocial outcomes (Richter et al., 2009; Wingood et al., 2008). Despite the prevalence of maternal HIV infection in South Africa and the current research elucidating the impact of HIV infection on U.S. children, empirical examination of the factors impacting South African children in families affected by HIV is sparse. The current study adds to the literature with a focus on the role of the mother-child relationship in HIV-affected families in terms of child internalizing (anxiety, depression) and externalizing (delinquency, acting out) problems across multiple (rural, urban) South African contexts. This study further extends the existing South African literature with an examination of the impact of mothers’ HIV-related physical and psychological symptoms, as opposed to the presence or absence of an HIV diagnosis.
Scholars ranging from Bronfenbrenner (1989) to Bowlby (1988), Baumrind (1989), and beyond elucidate the importance of parenting to child development. A considerable body of literature demonstrates the relevant empirical connections in typical families (e.g., Brody, Dorsey, Forehand, & Armistead, 2002; Steinberg, Lamborn, Dornbusch, & Darling, 1992), as well as families experiencing various stressors. For example, Rutter and Quinton (1984) pointed to parenting, and particularly the parent-child relationship, as a pathway through which parental depression leads to children’s negative outcomes, and numerous studies provide support for this pathway (for a review, see Cummings & Davies, 1994). Parenting has also been implicated in the relation between divorce or marital conflict and child functioning (for a review, see Kelly & Emery, 2003). Two aspects of parenting, parent-child relationship quality and parental monitoring of child behavior, have been consistently related to child adjustment (Lamborn, Dornbusch, & Steinberg, 1996). The current study attends to one of these aspects, parent-child relationship quality.
Relative to research focused on the aforementioned stressors, the literature addressing the impact of parental illness on families and children is limited, and the initial research, particularly that focused on parental HIV, was conducted in the United States (Chi et al., 2015). In 1995, Armistead and colleagues (Armistead, Klein, & Forehand, 1995), informed by developmental scientists (e.g., Cumming & Davies, 1994; Montemayor, 1983), reviewed the literature focused on families affected by parental illness and offered a conceptual framework for understanding relations between parental illness and child outcomes. This framework pointed to the importance of examining dimensions of parental illness (e.g., degree of incapacitation) and the role of family variables (e.g., parenting) as mechanisms in the parental illness-child functioning relation. Since that time and coinciding with the ever-growing numbers of mothers living with HIV (MLH), considerable U.S.-based research indicates that maternal HIV infection compromises parenting, and that compromised parenting affects youth outcomes (e.g., Murphy, Marelich, Armistead, Herbeck, & Payne, 2010; for review, see Rochat, Netsi, Redinger, & Stein, 2017). Specifically, U.S. research links maternal HIV status to mother-child relationship quality (e.g., Reyland, Higgins-D’Alessandro, & McMahon, 2002), and poor mother-child relationship quality is associated with internalizing (anxiety, depression, overall well-being) and externalizing (child behavior) problems in children in the United States (Murphy, Armistead, Marelich, & Herbeck, 2015). Children who do not view their parents as supportive may develop feelings of inadequacy and insecurity, which puts them at risk for developing internalizing problems (McCarty & McMahon, 2003). In addition, children who feel less close to their parents may be less likely to internalize parental standards of behavior, and, thus, exhibit externalizing behavior problems (Nunes, Faraco, Vieira, & Rubin, 2013). Furthermore, a systematic review of the global literature suggests that having a trusting relationship with a parent may serve as a protective factor for children who have a caretaker diagnosed with HIV (Chi & Li, 2013). Given the epidemiology of HIV/AIDS, attention has turned to families affected by HIV in other countries, and particularly South Africa, which bears a significant proportion of the illness burden.
The emerging literature in South Africa suggests an association between caregiver HIV and a variety of child outcomes. In a sample of young South African children (6-10 years old), Sipsma and colleagues (2013) found those children whose mothers were HIV-positive and symptomatic were more likely to exhibit externalizing (acting out) behaviors than children whose mothers were HIV-negative. In a similar sample, findings suggested maternal depression was related to parenting stress and parent/child dysfunction, and maternal coping, stress, and parent/child dysfunction related to child behavior, providing support for parenting as a mediator between maternal depression and child outcomes (Allen et al., 2014). One systematic literature review demonstrates that maternal mental health symptoms can disrupt the parent/child relationship; specifically, this review suggests that MLH experiencing psychosis may also evidence difficulty parenting, which may impact child well-being and attachment (Spies, Sterkenburg, Schuengel, & van Rensburg, 2014). Lachman, Cluver, Boyes, Kuo, and Casale (2014) also pointed to the role of parenting in the parental illness-child outcomes relations in an examination of families with older children (ages 10-17). These researchers found an association between parental AIDS-related illness and parenting. In comparison with families with non-AIDS-ill caregivers, families with AIDS-ill caregivers exhibited less positive parenting. The current study expands upon the recent South African literature, examining the intersections of HIV-related physical and mental health concerns, parenting, and child outcomes.
Parenting in the context of HIV infection must consider challenges unique to living with HIV/AIDS in South Africa. Although South Africa has abolished the apartheid system, its legacy remains. Apartheid-based policies (e.g., forced migrant labor) and economically driven urbanization have changed Indigenous African family structures from extended to nuclear families, which is exacerbated by AIDS-related loss of generations within families (Dunn & Parry-Williams, 2008; Skinner & Davids, 2006). In some South African communities, parenting is no longer a community-based activity, and, primarily, immediate family members are entrusted to care for children (Peterson, Bhana, & McKay, 2005). Thus, many of the protective resources previously available to youth are absent, potentially enhancing the impact of mothers’ parenting on child functioning.
The focus of the current study, Black South African women, on average, are the poorest, least educated, and most economically marginalized and disadvantaged of all groups in South Africa (Hunter, 2007). Many are consistently exposed to multiple stressors, such as poverty, crime, and poor physical health in general (Bray, Gooskens, Kahn, Moses, & Seekings, 2010; Vorster et al., 2000), which may exacerbate the negative impact of their HIV on their children (Chi & Li, 2013). Furthermore, Black South African women are most often the primary, or sole, caregivers of children and, as a result, their children often face cumulative and multiple stressors above and beyond those related to maternal HIV infection and poor maternal health (Sherr et al., 2014). Research also suggests that MLH may be affected by HIV-specific risk factors, such as stigma and abuse (Cluver & Gardner, 2007; Richter et al., 2009; Wingood et al., 2008). The combination of limited support from family members and the physical and psychological sequelae of HIV, despite recent advances in care, can render day-to-day parenting and other tasks particularly challenging (for a review, see Spies et al., 2014).
A nuanced examination of the paths through which maternal HIV is related to child outcomes, such as depression, anxiety, and acting out (e.g., rule-breaking behavior), requires consideration of mothers’ physical and psychological symptoms. HIV-related physical incapacitation may be particularly salient for MLH in South Africa, given the challenges associated with treatment access (UNAIDS, 2016), and the other illnesses often accompanying HIV in this context (e.g., tuberculosis). In addition, higher levels of anxiety have been observed in MLH in South Africa, compared with those not living with the virus (for review, see Brandt, 2009), and research elsewhere demonstrates a significant relationship between HIV status and depressive symptoms (for review, see Brandt, 2009). Given HIV’s physical and psychological sequelae, we examined relations between physical and psychological symptoms among MLH and the mother-child relationship.
For several reasons, mothers with early adolescents (youth aged 11-16) were the focus of this study. Given considerable developmental differences between children of various ages, it is important to focus on children who fall within a relatively constrained age range. Moreover, significant changes in children’s behavior often occur during the transition to adolescence, and family stress (e.g., maternal HIV infection) may have differential effects on pre-adolescent children as compared with adolescents. For example, adolescents, particularly females, may be more likely to experience symptoms of depression and anxiety in response to family stress, compared with pre-adolescent children (Compas et al., 1994; Thastum et al., 2009).
We examined the relationship between maternal health and child adjustment, as mediated by the mother-child relationship. In this sample of South African MLH with 11- to 16-year-old children, we hypothesized that poorer maternal health (as measured by physical symptoms, depression, and anxiety) would be associated with poorer mother-child relationships and anticipated that poorer relationships would be related to more child internalizing and externalizing symptoms. In particular, we anticipated that mothers’ health would be related to child internalizing and externalizing problems indirectly through its association with the mother-child relationship.
Method
Participants
The participants in this study were part of a larger sample that included 256 women, only some of whom were living with HIV. In particular, 107 participants self-reported a positive HIV status. Of these 107, three individuals reported the presence of another non-HIV-related terminal illness (e.g., cancer) and were excluded from the current analyses. Participants reporting medical conditions such as hypertension and tuberculosis were not excluded. Therefore, a total of 104 participants comprise the current sample. These MLH had at least one child between the ages of 11 and 16 who resided with the mother. If the mother had more than one child between 11 and 16 years old, the mother reported on the oldest child for this study’s outcome measures. The mothers’ mean age was 35.71 (SD = 6.09), and children had a mean age of 13.27 (SD = 1.73). In total, 51% of the children were female.
Measures
There was a dearth of culturally salient measures developed with or for Black South African families for the constructs explored in this study. To maximize the cultural salience of the measures, several steps were taken. First, focus groups were conducted with individuals demographically similar to the research participants to operationalize the study constructs. Second, meetings were held with key informants in South Africa to discuss and consider each construct and identify proposed measures. These measures were reviewed and modified by the key informants for cultural competence and comprehensibility at a subsequent meeting. Third, the revised measures, as well as newly created ones, were piloted with women who were representative of the research population of interest. The piloting also resulted in changes to the measures. Given the extent to which all measures were modified, where appropriate, measures were subjected to factor analyses, and Cronbach’s alphas were obtained.
In addition, for all measures to be available in three commonly spoken languages by the target population (i.e., English, Afrikaans, and Sotho), measures were translated into Afrikaans and Sotho and subsequently back-translated. In the back-translation technique (Brislin, 1970), the measure is translated from its source language (English) to the target language (Afrikaans and Sotho), and then translated back to the source language. The two versions are then compared to assess for equivalency. Two independent translators were used for both segments of the translation.
Demographic information
The Household Economic and Social Status Index (HESSI: O. A. Barbarin & Khomo, 1997), created in South Africa, was used to provide demographic and health information for both mothers and children. This measure is composed of 21 items and assesses information such as maternal age, marital status, and educational level, as well as child age, gender, and health status. The HESSI was modified from its original format for the purposes of this study by adding items related to HIV.
Assessment of HIV status
Women’s HIV status was established through their self-report during the interview. We relied on women’s self-report for several reasons. During focus groups and piloting, the investigators for the larger study found that women willingly disclosed their HIV status to interviewers. Moreover, asking questions related to HIV at the end of the interview, once rapport had been established, created a context that facilitated the disclosure of HIV status. Furthermore, funding for this investigation did not support HIV-testing. Of the total sample recruited in the parent study (N = 256), 107 women (42%) disclosed a positive HIV status.
Maternal health
A latent variable signifying overall maternal health was created through the observed variables of the mother’s report of her physical and psychological symptoms, that is, depression and anxiety. Scales measuring these constructs are described below. Despite the small sample size, reliance on this latent construct is particularly important given our adaptation of individual measures. Using this latent variable, rather than three separate observed variables, allows appropriate power and greater confidence in the construct’s validity for the current analyses.
Physical symptoms
Participants reported the physical symptoms they experienced within the previous year using the Physical Symptoms Inventory (PSI: Wahler, 1969). The PSI is a self-report measure of physical symptomatology, demonstrates adequate internal consistency and test-retest reliability (Wahler, 1973), and has been used in research with HIV+ women (Family Health Project Research Group, 1998). This measure was modified to improve its assessment of symptoms associated with HIV infection (e.g., vaginal discharge) by deleting seven items and adding four items resulting in a total of 37 items (Family Health Project Research Group, 1998). Women were asked to rate how often each physical symptom bothered them using a 5-point scale: 1 = never; 2 = a few times a year, 3 = about once a month, 4 = about once a week, 5 = nearly every day. Factor analysis indicated that 11 of the 37 items did not load at levels of 0.40 and above, thus, 26 items were retained for the final measure. The total score for the measure is the sum of the 26 items, which could range from 26 to 130. The coefficient alpha for the current sample was .90.
Maternal anxiety
The Institute for Personality and Ability Testing Anxiety Scale (IPAT: Cattell & Scheier, 1963) was used to assess symptoms of anxiety. The IPAT demonstrates good internal consistency in previous samples (Cattell & Scheier, 1963) and has been used to assess psychological functioning in South African women (Mfusi & Mahabeer, 2000). In this study, all 40 items were subjected to a confirmatory factor analysis. Results indicated that the items did not load together onto one factor, with the majority of items not loading at the appropriate 0.40 level. Thus, an exploratory factor analysis was conducted, and results indicated a single 11-item factor. A summary score was created by summing all 11 items. The possible range of scores on this measure was 0 to 33, with higher scores indicating higher levels of anxiety. The final coefficient alpha for this measure was .73.
Maternal depression
The Center for Epidemiologic Studies Depression Scale (CES-D: Radloff, 1977) was used to assess symptoms of depression among MLH. This measure consists of 20 items that address the presence and frequency of symptoms of depression during the past week. The CES-D was developed by the National Institute of Mental Health to screen for clinical depression in community samples, and this measure is suggested for use with medically ill populations (Derogatis, Fleming, Sudler, & DellaPietra, 1995). It has been used previously with Black South Africans (Hughes, Jelsma, Maclean, Darder, & Tinise, 2004). Each item was rated on a scale of 1 (rarely) to 4 (most of the time), and the total score for the final measure could range from 16 to 64. Four of the 20 items were not included in analyses, one because of its conceptual overlap with HIV infection, and three others as they did not load at 0.40 or above during factor analysis. The coefficient alpha for the sample was .89.
Mother-child relationship
The mother-child relationship was assessed using the Interaction Behavior Questionnaire (IBQ: Prinz, Foster, Kent, & O’Leary, 1979). This measure consists of 19 items that assess communication and conflict behavior between a mother and her child. Mothers are requested to think back over the past few weeks at home and respond to the various items by answering yes or no. Examples of items on this measure include, “Most of the time, your child likes to talk to you”; “Your child doesn’t listen to what you say”; and “Your child is easy to get along with.” The IBQ demonstrates good internal consistency and discriminant validity in U.S. samples (Prinz et al., 1979; Robin & Weiss, 1980) and has been used in research with HIV-positive African American women in the United States (Kotchick et al., 1997) and in a South African sample (Armistead et al., 2014). In this study, all 19 items were subjected to a confirmatory factor analysis. A total of 12 of the items loaded at 0.40 and above and were summed in order to create a total score for the measure. Total scores could range from 0 to 12, with higher scores indicating a better mother-child relationship. The coefficient alpha for this sample was .74.
Child internalizing and externalizing symptoms
The Child Behavior Checklist (CBCL: Achenbach & Edelbrock, 1991) is a parent-report 113-item questionnaire that measures behavioral problems of children aged 4 to 18 over the preceding 6 months. This measure provides a total behavior problem score, two broad-band dimensions (internalizing and externalizing), and several narrow-band dimensions. The internalizing scale is the sum of symptoms from the anxious-depressed, shy-withdrawn, and somatic complaints factors, and the externalizing scale from the delinquent and aggressive behavior factors. Each item is rated on a 3-point scale, 0 = not true, 1 = sometimes true, 2 = very often/often true. Extensive reliability and validity data are provided with U.S. samples (Achenbach & Edelbrock, 1991), and the CBCL was used in a large longitudinal study of Black South African children (D. Barbarin & Richter, 2007). Although this measure has been used with Black South Africans, it was translated for the purposes of this study, and, as such, factor analyses were conducted for the internalizing and externalizing scales separately.
For the internalizing scale, the somatic complaints items were eliminated for analyses, as they did not load adequately, and they overlap conceptually with symptoms associated with living in resource-poor areas. In addition, five of the items on the internalizing scale were eliminated because of translation errors during the back-translation process. The remaining 18 internalizing items were subjected to a confirmatory factor analysis. A total of 13 items loaded at 0.40 and above. A total score for the internalizing scale was created by summing these items. The total possible score for the internalizing scale ranges from 0 to 26, and the coefficient alpha for this sample was .77.
For the externalizing scale, a confirmatory factor analysis was conducted with 32 of the 33 externalizing items. One item (i.e., sets fires) was excluded given feedback received from focus groups and piloting. Of the 32 items, 23 items loaded at 0.40 and above. A total raw score for the externalizing scale was created by summing these items. The total possible score for the externalizing scale ranges from 0 to 46, and the coefficient alpha for this sample was .87.
Procedure
Study procedures were approved by the Institutional Review/Ethics Boards of the United States and South Africa-based universities housing this research. Participants were recruited from three communities in South Africa: Hammanskraal, a rural community; Mamelodi, an urban community; and Atteridgeville, a suburban community. Recruitment strategies included outreach conducted by community agencies, advertisement on a local radio station, and word-of-mouth. Women who arrived to participate in the study were screened for eligibility by an interviewer. Eligibility included having a child between 11 and 16 who resided with the mother, and the ability to participate in either English, Afrikaans, or Sotho.
Women were interviewed by a trained interviewer in a private location. Interviews were conducted at satellite campuses of the University of Pretoria, in a hospital, or at the AIDS Training Information and Counseling Center, depending upon which location was most convenient for participants. Each of the five interviewers, three of whom were doctoral students from the United States and two of whom were Master’s students from the University of Pretoria, had previous experience working with people living with HIV and were trained on measure administration, management of participant distress, and informed consent procedures. Interviews and informed consent were conducted in English, Afrikaans, or Sotho, depending on the language preference of the participant.
Subsequent to informed consent, all measures were administered verbally, and cue cards were used because of low literacy rates. The cue cards consisted of pictorial representations of Likert-type scale response choices such as Never, Sometimes, Often, and Always as well as the response choices in English, Afrikaans, and Sotho. These cards allowed participants to choose their responses by pointing if they preferred. During the interview, participants were offered a healthy snack that included fruit, juice, and biscuits. At the end of the interview, participants were provided with a small gift (hand lotion) and 70 Rand (about US$10) for their time and effort.
Results
Analyses were conducted in Mplus 7.0 using a robust maximum likelihood estimator. Missing data were determined to be missing at random, and, therefore, these few cases (N = 4) were dropped, as dropping cases missing at random would not significantly impact the results.
The current study investigated the role of mother-child relationship quality in the relationship between maternal health and child internalizing and externalizing symptoms. The proposed model was analyzed with structural equation modeling, using two-step modeling (Anderson & Gerbing, 1988). In this two-step approach, a measurement model is first estimated to test for goodness of fit of the latent constructs. The latent covariance structure is fully saturated in that all correlations between constructs are included in the model. In the second step, this fully saturated latent covariance model is then compared with structural models in which only hypothesized paths in between constructs are estimated.
The measurement model included one maternal health latent variable with three observed indicators (maternal anxiety, maternal depression, and maternal physical symptoms), as well as measured variables of mother-child relationship quality, child internalizing symptoms, and child externalizing symptoms. Model fit was evaluated based on the chi square test of model fit, root mean squared error of approximation (RMSEA), standardized root mean squared residual (SRMR), and comparative fit index (CFI), and all indices provided evidence for good model fit, χ2 = 5.66, df = 6, p = .46; RMSEA = 0.00, 90% confidence interval (CI) = [0.00, 0.12], SRMR = 0.02; CFI = 1.00. Each indicator loaded onto the maternal health factor significantly with a factor loading of 0.62 or greater (Maternal Depression = 0.76, Maternal Anxiety = 0.74, Physical Symptoms = 0.62). Correlations between variables in the measurement model are presented in Table 1. Correlations were in the expected direction. The strongest correlation was between mother-child relationship quality and child externalizing symptoms. A model was also run that included child age and child gender; neither was correlated with any study variables, so they were excluded from the analyses. Furthermore, no significant correlations between age and any outcome variables were found, nor did the pattern of correlations differ when examined by child age groups (i.e., 11-13, 14-16). Given these findings and the limited sample size, structural models were not run separately for the two age groups.
Descriptive Statistics and Bivariate Correlations Among All Study Variables.
p = .05. **p < .01.
In the second step, a structural model was tested to examine the hypothesized relations among the variables. This model was examined for comparable fit against the measurement model.
A fully saturated structural model (Figure 1) was specified with all covariance paths between the study’s constructs, estimated as in the original measurement model, the only difference being that directional paths instead of correlations were estimated from maternal health to mother-child relationship quality, child internalizing symptoms, and child externalizing symptoms, and from mother-child relationship quality to child internalizing symptoms and child externalizing symptoms. Statistically, the model fit was identical to that of the measurement model, χ2 = 5.66, df = 6, p = .46; RMSEA = 0.00, 90% CI = [0.00, 0.12], SRMR = 0.02; CFI = 1.00.

Fully saturated structural model.
Consistent with hypotheses, maternal health was significantly related to mother-child relationship quality, such that worse maternal health (i.e., more symptoms of depression, anxiety, and physical symptoms) was associated with poorer mother-child relationship quality. Mother-child relationship quality was significantly negatively related to child externalizing symptoms, suggesting that higher quality of mother-child relationship predicted fewer child externalizing symptoms.
This initial structural model was compared with the hypothesized indirect-only model in which the direct effects of maternal health on child internalizing and externalizing symptoms were removed. Its relative model fit was compared with the fully saturated modeling using a Satorra-Bentler scaled chi-square difference test. This method, which is preferred when using MLR estimation, divides the chi-square value by scaling correction in order to estimate more accurately chi-square in non-normally distributed samples (Satorra & Bentler, 2001). In comparing the initial model that included direct paths between all study variables (Figure 1) to an indirect model, a Satorra-Bentler scaled chi-square difference test revealed that the indirect model was significantly worse-fitting, χ2M2–M1(2) = 6.36, p < .05.
Additional models (in which other theory-driven paths were estimated) were compared with the model in Figure 1 1 ; additional Satorra-Bentler scale chi-square difference tests revealed that these models were also significantly worse-fitting than the fully saturated model. Therefore, the best-fitting model was a fully saturated one involving paths specifying both direct and indirect effects of maternal health on child internalizing and externalizing symptoms by mother-child relationship (Figure 1). This model suggests that the maternal health–child externalizing symptoms relationship operates indirectly through mother-child relationship quality. Indirect effects of maternal health on child internalizing and externalizing symptoms were estimated using a bootstrapping technique in Mplus that allows for asymmetrical confidence intervals (MacKinnon, Lockwood, & Williams, 2004). There was an indirect effect of maternal health on child externalizing problems, β = 0.22, SE = .07, p = .001, 95% CI = [0.10, 0.37]. There was no indirect effect of maternal health on child internalizing problems though, β = 0.07, SE = .05, p = .17, 95% CI = [–0.03, 0.18], suggesting that maternal health is directly, but not indirectly, associated with child internalizing problems.
Discussion
The purpose of the current study was to examine the extent to which the quality of the mother-child relationship explains the effect of maternal health on child adjustment among a group of Black South African mothers living with HIV. The current study largely confirms the study’s hypotheses and replicates previous findings, though results differed by outcome variable. As expected (e.g., Murphy et al., 2010; for review of similar international findings, see Sherr et al., 2014; Sipsma et al., 2013), maternal health was associated with mother-child relationship quality, which in turn was associated with children’s externalizing symptoms. This expands previous South African literature documenting the association between AIDS-ill caregivers and less positive parenting (Lachman et al., 2014). However, for internalizing problems, the relationship with maternal health symptoms was direct, rather than operating through the mother-child relationship quality. Mothers with more health symptoms had children with more internalizing symptoms, which is also consistent with previous research (e.g., Forehand et al., 2002), but the path between mother-child relationship and internalizing symptoms was not significant in the best-fitting model. This latter finding is inconsistent with prior research with HIV-affected families, which has generally demonstrated relations between mother-child relationship quality and child internalizing problems (e.g., Forehand et al., 2002). A number of factors may explain this inconsistency. The most obvious is the research setting, with the bulk of prior research occurring in the United States (e.g., Muze, 2013), whereas this sample is situated in South Africa. In addition, the children in the current study are older than those enrolled in prior research, and the bulk of the existing literature relies on HIV status, as opposed to maternal physical and psychological symptoms, as an indicator of health (e.g., McDonnell, Gielen, O’Campo, & Burke, 2005; for review, see Rochat et al., 2017). Furthermore, although specific research may not have been conducted on this particular result, it is possible that older, compared with younger, children may be better able to observe mothers’ symptoms and respond with anxiety or depression, resulting in this direct link between the two constructs.
U.S. research has found that positive maternal HIV status affects child outcomes negatively. In the context of South African mothers living with HIV, we found evidence to suggest that maternal mental and physical symptoms, perhaps a richer measure of illness-related compromise compared with HIV status, are associated with increased child internalizing and externalizing behaviors. Furthermore, the present study identified a malleable mechanism through which maternal health affects child adjustment, particularly externalizing behaviors: quality of the mother-child relationship, corroborating previous findings on maternal health, parenting, and child outcomes among U.S. parents living with HIV. Taken together, this research provides cross-cultural evidence to suggest the importance of parenting to understanding how HIV affects the health of parents and their children. As parenting is a malleable factor, researchers can develop interventions, or modify existing parenting interventions (e.g., Parent-Child Interaction Therapy) to mitigate the negative impact of parental HIV on parenting abilities and child adjustment.
Family-based HIV prevention interventions that aim to decrease HIV infection among South African youth by targeting parenting quality have found that program participants report better parent-child relationship quality (Armistead et al., 2014), in comparison with control participants. Although Armistead and colleagues did not exclusively recruit parents living with HIV, results demonstrate the effectiveness of a parenting intervention on changing the quality of the parent-child relationship in South Africa. Eloff and colleagues (2014) examined the effects of an intervention for HIV-positive mothers with children ages 6 to 10 years old. The program, focused on parent-child communication and parenting, significantly reduced children’s externalizing behaviors and improved communication. This study shows promise but targets younger children. Future studies should focus on testing whether interventions aiming to develop and improve parenting skills among South African parents living with HIV with older children could reduce adolescent problem behaviors and promote positive health outcomes.
Although the current study makes a significant contribution to understanding the relationship between maternal health and child adjustment among Black South African families affected by maternal HIV, the study is not without limitations. Given the study only enrolled Black South African mothers, generalizability to parents differing in race or gender should not be assumed. Only parental report of mother-child relationship and adolescent adjustment was used in the present study. Including adolescent report in future studies will enhance the validity of the measurement of these constructs. In addition, the internalizing scale used in the present study does not assess posttraumatic symptoms. The inclusion of posttraumatic symptoms is important as many South African adolescents experience high rates of community violence and violent crimes (e.g., Seedat, van Nood, Vythilingum, Stein, & Kaminer, 2000). Future studies, therefore, need to assess posttraumatic symptoms and exposure to violence to gain a more comprehensive understanding of how maternal health affects child mental health within South African families. Furthermore, the present study conducted focus groups to assess the cultural validity of the study constructs and the measurements of these constructs. Addressing cultural relevance of construct and cultural validity of measurement are critical to maximizing research’s ecological validity. Nonetheless, several of this study’s measures have not been validated with the target population.
To deepen our understanding of how Black South African mothers living with HIV parent, future research needs to consider broader contextual influences, such as poverty and social inequality, and how these influences translate into risk factors such as exposure to violence. More importantly, research should examine factors promoting resiliency across the ecology that can help these mothers and their children thrive in the face of adversity. For example, sense of community and collective efficacy have been identified as protective among populations who have a communal cultural orientation (Braun-Lewensohn & Sagy, 2014). Parenting interventions designed for Black South African families may benefit from supporting the development of these collective resources in addition to building capacity for adaptive parenting skills. In particular, clinicians may be able to use information gained from this and future research to understand more fully the ways in which Black South African families may be impacted uniquely by HIV, particularly in the context of potential PTSD symptoms. This information may aid in targeted interventions focused on the specific stressors experienced by this population.
Footnotes
Acknowledgements
The authors acknowledge the support of the women who participated in this research.
Authors’ Note
Jenelle Shanley is now affiliated with Pennsylvania State University, State College, PA, USA.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors acknowledge the financial support provided by the National Institute of Child Health and Human Development and the National Institute of Mental Health (R03HD046371).
