Menu
Menu
Menu
Menu
Background The intergenerational effects of HIV require long-term investigation. We compared developmental outcomes of different generations impacted by HIV – children of mothers not living with HIV, the ‘second generation’ (ie, with recently infected mothers) and the ‘third generation’ (ie, children of perinatally infected mothers). Methods A cross-sectional community sample of N=1015 young mothers (12-25 years) and their first children (2-68 months, 48.2% female), from South Africa’s Eastern Cape Province. 71.3% (n=724) of children were born to mothers not living with HIV; 2.7% (n=27; 1 living with HIV) were third-generation and 26.0% (n=264; 11 living with HIV) second-generation children. Child scores on the Mullen Scales of Early Learning (MSEL), the WHO Ten Questions Screen for Disability and maternal demographics were compared between groups using χ 2 tests and univariate approach, analysis of variance analysis. Hierarchical linear regressions investigated predictive effects of familial HIV infection patterns on child MSEL composite scores, controlling for demographic and family environment variables. Results Second-generation children performed poorer on gross (M=47.0, SD=13.1) and fine motor functioning (M=41.4, SD=15.2) and the MSEL composite score (M=90.6, SD=23.0) than children with non-infected mothers (gross motor: M=50.4, SD=12.3; fine motor: M=44.4, SD=14.1; composite score: M=94.1, SD=20.7). The third generation performed at similar levels to non-exposed children (gross motor: M=52.4, SD=16.1; fine motor: M=44.3, SD=16.1, composite score: M=94.7, SD=22.2), though analyses were underpowered for definite conclusions. Hierarchical regression analyses suggest marginal predictive effects of being second-generation child compared with having a mother not living with HIV (B=-3.3, 95% CI=-6.8 to 0.1) on MSEL total scores, and non-significant predictive effects of being a third-generation child (B=1.1, 5% CI=-7.5 to 9.7) when controlling for covariates. No group differences were found for disability rates (26.9% third generation, 27.7% second generation, 26.2% non-exposed; χ 2 =0.02, p=0.90). Conclusion Recently infected mothers and their children may struggle due to the disruptiveness of new HIV diagnoses and incomplete access to care/support during pregnancy and early motherhood. Long-standing familial HIV infection may facilitate care pathways and coping, explaining similar cognitive development among not exposed and third-generation children. Targeted intervention and fast-tracking into services may improve maternal mental health and socioeconomic support.
Data used within these analyses originate from the ‘Helping Empower Youth Brought up in Adversity with their Babies and Young children’ study. The study was conducted in rural and periurban health districts of South Africa’s Eastern Cape and aimed at investigating the effects of adolescent motherhood, as well as intergenerational effects of HIV exposure. A total of 1046 adolescent and young adult mothers (10–25 years) with at least one living child were interviewed between March 2018 and July 2019.23 The required sample size was estimated based on expected effect sizes for key outcomes. Participants were partially recruited from a previous study (n=159: any young mothers included), as well as through six parallel sampling strategies developed in cooperation with local experts and an adolescent mother advisory group to ensure representativeness (n=887: only adolescent mothers). This comprised sampling through 73 known health facilities within the districts, 43 secondary schools, 9 maternity units and referrals by service providers, social workers and adolescent mothers themselves. A total of 95%–98% of eligible mothers from each recruitment channel were successfully enrolled into the study. The current analyses were limited to data on first child of the young mother only (10–25 years), and children aged 68 months or younger who were within the normed range for the Mullen Scales (n=31 excluded based on this age restriction), limiting the sample to n=1015. We also performed analyses including adolescent mothers (age at pregnancy ≤19, n=972) only for the key outcomes (see online supplemental appendix 1), as these mothers had been the main recruitment focus for the study. bmjopen-2021-058340supp001.pdf The study team were advised on recruitment methods by adolescent mothers, whose suggestions were included in the study protocol. Furthermore, the team has been working with two Teen Advisory groups in the Eastern and Western Cape of South Africa, who were involved in piloting the study. Feedback was incorporated to improve relevance and acceptability of the questionnaire and the research procedure. Finally, through the engagement of community leaders, it was ensured that recruitment strategies were effective, sensitive to participant’s circumstances and minimised stigma risks. The dataset analysed combined four data sources. First, all adolescent mother participants completed a detailed study questionnaire relating to sociodemographic characteristics, sexual and reproductive health, physical and mental health, relationships and social support. Second, they completed an adolescent parent questionnaire, which collected data on maternal and child health, child development, the father of the child and maternal factors including social support, the parenting experience and violence exposure. Third, cognitive assessments of the children were performed by a trained administrator, using the Mullen Scales of Early Learning. Finally, details from the child’s medical records (Road to Health Booklet) were also included in the database. All participants provided written consent, and interviews were conducted in the language of their choice (ie, English or isiXhosa). Regardless of participation, all adolescents received a small ‘snack pack’ containing a snack and juice during the interview, and a small ‘thank you pack’, with personal products such as toothpaste and a toothbrush. Items included in these packs were selected by the adolescent advisory group as preferable and appropriate. New mothers also received a ‘baby pack’, the contents of which were also chosen by an adolescent advisory group, and included nappies and baby cream. Mode of maternal HIV acquisition (perinatal vs recent) was assessed through an algorithm, given that the study was community-based and not linked to clinical testing data. Accordingly, it was derived using a logic tree based on clinical and fieldwork experiences.17 The algorithm allowed for categorisation according to self-report, age of ART initiation, and parental death information (Tolmay, Saal et al., in preparation). For the current analyses, we compared the following three groups: children of the third generation (mother perinatally infected), children of the second generation (mother recently infected) and children of mothers not living with HIV (see figure 1). Child HIV status was not taken into account in these classifications, since absolute numbers were low (n=12 based on maternal self-reported data), with only one of these children being in the third generation. Thus, we were underpowered to study these children separately. We decided however to retain them in the analyses since they are relevant members of the second and third generations, respectively. Flow chart of classification by mode of maternal HIV acquisition. HEY BABY, Helping Empower Youth Brought up in Adversity with their Babies and Young children. Child developmental outcomes were assessed using the Mullen Scales of Early Development (MSEL).24 The MSEL is normed for children aged 0–68 months (USA) and assesses child performance across five domains: gross motor (only for age <39 months), fine motor, visual reception, receptive and expressive language (score range: 20–80). A composite score (score range: 49–155) can be derived, and the scales have been validated for use in sub-Saharan Africa.23 25 26 Child disability status was assessed using the WHO Ten Questions Screen for Disability.27 This measure can be applied to detect common disabilities (hearing, visual, physical, speech, mental and epilepsy) in children. A score indicating any disability (‘yes’ across any of the 10 items) was derived for the current analyses. Adolescent mothers were compared on various variables, including demographic factors (maternal age, maternal and paternal age at pregnancy), child feeding method used (formula, breastfeeding, combined, other), maternal school progression (self-report: repeated at least one school grade), poverty (number of the eight socially perceived necessities for children the family had access to,28 household access to any government cash transfers child support, foster child, pension, disability or care dependency grant, measured via South Africa Census item,29 food security number of days there was not enough food for the household in the past 7 days,30 full ART adherence over the past 7 days (Patient Medication Adherence Questionnaire,31 any HIV clinic appointments missed in the past year, extent of HIV-related stigma Adolescents Living with HIV Stigma Scale,32 depressive symptoms in the past 2 weeks Child Depression Inventory-Short Form,33 anxiety symptoms in the past month Revised Children’s Manifest Anxiety Scale,34 PTSD symptoms in the past month Child PTSD Checklist-Short Form,35 suicidality symptoms during the past month Mini International Psychiatric Interview for Children and Adolescents- Suicidality and Self-Harm Subscale,36 extent of community violence exposure in the past year item from the Child Exposure to Community Violence Checklist,37 intimate partner violence exposure in the past month (Juvenile Victimisation Questionnaire,38 current parenting stress (Parental Stress Scale39) and amount of social support received (adjusted version of the Medical Outcomes Study Social Support Survey,40 previously used in the South African context.41 All analyses were conducted using STATA V.16 SE. First, child and adolescent mother descriptive information for the three groups of interest (children of the third generation, children of the second generation and HIV-unexposed children) was provided. Next, child developmental outcomes and maternal factors surrounding socioeconomic status, HIV-related variables (only HIV-affected mothers), mental health and social environment were compared between groups, usingχ2 tests and univariate analyses of variance as appropriate. For continuous variables, Tukey’s range tests were used to conduct post hoc group comparisons. Effects at p<0.05 were considered relevant. Finally, a hierarchical regression model was run, with the child composite score on the Mullen scales as a key outcome. In a first step, child familial HIV exposure (third generation, second generation, HIV-unexposed) was added as a predictor. In a second step, we included relevant control variables that were found to be associated with the predictor or outcome variables in univariate analyses or were identified as relevant within the literature to see whether child familial HIV exposure would remain a relevant predictor.
