Adolescent Motherhood and HIV in South Africa: Examining Prevalence of Common Mental Disorder

The mental health of adolescents (10–19 years) remains an overlooked global health issue, particularly within the context of syndemic conditions such as HIV and pregnancy. Rates of pregnancy and HIV among adolescents within South Africa are some of the highest in the world. Experiencing pregnancy and living with HIV during adolescence have both been found to be associated with poor mental health within separate explorations. Yet, examinations of mental health among adolescents living with HIV who have experienced pregnancy/parenthood remain absent from the literature. As such, there exists no evidence-based policy or programming relating to mental health for this group. These analyses aim to identify the prevalence of probable common mental disorder among adolescent mothers and, among adolescents experiencing the syndemic of motherhood and HIV. Analyses utilise data from interviews undertaken with 723 female adolescents drawn from a prospective longitudinal cohort study of adolescents living with HIV (n = 1059) and a comparison group of adolescents without HIV (n = 467) undertaken within the Eastern Cape Province, South Africa. Detailed study questionnaires included validated and study specific measures relating to HIV, adolescent motherhood, and mental health. Four self-reported measures of mental health (depressive, anxiety, posttraumatic stress, and suicidality symptomology) were used to explore the concept of likely common mental disorder and mental health comorbidities (experiencing two or more common mental disorders concurrently). Chi-square tests (Fisher’s exact test, where appropriate) and Kruskal Wallis tests were used to assess differences in sample characteristics (inclusive of mental health status) according to HIV status and motherhood status. Logistic regression models were used to explore the cross-sectional associations between combined motherhood and HIV status and, likely common mental disorder/mental health comorbidities. 70.5% of participants were living with HIV and 15.2% were mothers. 8.4% were mothers living with HIV. A tenth (10.9%) of the sample were classified as reporting a probable common mental disorder and 2.8% as experiencing likely mental health comorbidities. Three core findings emerge: (1) poor mental health was elevated among adolescent mothers compared to never pregnant adolescents (measures of likely common mental disorder, mental health comorbidities, depressive, anxiety and suicidality symptoms), (2) prevalence of probable common mental disorder was highest among mothers living with HIV (23.0%) compared to other groups (Range:8.5–12.8%; Χ2 = 12.54, p = 0.006) and, (3) prevalence of probable mental health comorbidities was higher among mothers, regardless of HIV status (HIV & motherhood = 8.2%, No HIV & motherhood = 8.2%, Χ2 = 14.5, p = 0.002). Results identify higher mental health burden among adolescent mothers compared to never-pregnant adolescents, an increased prevalence of mental health burden among adolescent mothers living with HIV compared to other groups, and an elevated prevalence of mental health comorbidities among adolescent mothers irrespective of HIV status. These findings address a critical evidence gap, highlighting the commonality of mental health burden within the context of adolescent motherhood and HIV within South Africa as well as the urgent need for support and further research to ensure effective evidence-based programming is made available for this group. Existing antenatal, postnatal, and HIV care may provide an opportunity for mental health screening, monitoring, and referral. Data utilised within these analyses are drawn from a large prospective longitudinal cohort study of adolescents in the Eastern Cape province of South Africa (n = 1526). One thousand and fifty-nine adolescents living with HIV were recruited to the study utilising records from 53 public health facilities providing antiretroviral therapy to adolescents within the province. Sampling was undertaken in two stages: (1) public health facilities were identified through the national Department of Health register and, (2) all adolescents on public health facility records that had initiated treatment in the previous 3 years were approached inclusive of those disengaged from care. Adolescents were followed up utilising community tracing methods to ensure the inclusion of both those engaged and disengaged with HIV services. At baseline, 90.1% of the eligible sample identified through clinical records were interviewed. The comparison group (n = 467) were age-matched, and selected from the same environments, co-residing with or near adolescents living with HIV study participants also completed interviews. These participants self-reported that they were not living with HIV and had not initiated antiretroviral therapy and had not experienced possible opportunistic infections, nor had history of familial HIV/AIDS, thus classified as not living with HIV for the purpose of these analyses. Baseline data collection was undertaken between February 2014 and September 2015. Follow-up data was collected between November 2015 and February 2017. The cohort had a 95.3% retention rate at follow-up (n = 1454). All adolescents and caregivers (if adolescents were < 18 years of age) provided informed consent. To protect participants confidentially interviews were conducted in a private space chosen by the participant. Given the risk of stigma relating to HIV within the communities from which data were gathered, to further protect confidentiality of participants the study was presented as a broad study focusing on youth access to health and social care services. All participants completed interviews consisting of a detailed study questionnaire consisting of validated scales and study specific questions focused on sociodemographic characteristics, health (inclusive of mental health screening measures), relationships, communities, schooling, and managements of HIV (if appropriate). Study questionnaires were developed in conjunction with adolescents (pre-piloted with a teen advisory group local to South Africa prior to data collection), as well as local, national, regional, and global organisations. Participants completed questionnaires in their language of choice (isiXhosa or English), and data was translated and back translated if appropriate. Analyses present data relating to adolescent motherhood (previous pregnancy between 10 and 19 years of age). Only data for female adolescents’ ≤ 20 years of age were included within these analyses. Females who were 20 years of age were retained in the sample if they were within 40 weeks (average length of a full-term pregnancy) of their twentieth birthday to ensure that any reporting of pregnancy was between 10 and 19 years of age. Dates of birth and dates of interview were used to ascertain eligibility. Adolescent motherhood status was ascertained from participant self-report. Given the potentially differing mental health profiles of current adolescent mother and adolescents who had experienced elective abortion, miscarriage or still birth, adolescents who reported a pregnancy but did not report a live birth were excluded from these analyses. In total, 723 adolescents were included within analyses (see Fig. 1). Participant selection criteria These analyses utilise cross-sectional data. Both baseline and follow-up data were used to corroborate reported adolescent pregnancy within the sample, all other measures were obtained within the follow-up round of data collection (2015–2017; total sample n = 1454; see Fig. 1). Sociodemographic characteristics were routinely gathered during baseline data collection and corroborated during follow-up data collection. Sociodemographic characteristics included age, biological sex, dwelling location, housing, school enrolment, orphanhood status and cash grant receipt via self-report measures. Adolescent motherhood (previous pregnancy and live birth between 10 and 19 years) was identified using two self-report measures [having had a previous pregnancy resulting in a live birth and having one or more biological child(ren)]. Adolescents were classified as experiencing motherhood if they scored on any one of the measures of motherhood (above). Adolescents who had experienced miscarriage, abortion, stillbirth, or the outcome of the pregnancy was unknown were excluded from analyses (see Fig. 1). Whether the pregnancy was planned and/or wanted (most recent pregnancy), age of last pregnancy (years), the number of previous pregnancies and the number of children were obtained from participant self-report. HIV status was ascertained through clinical notes and corroborated by participant or caregiver self-report on a case-by-case basis. All participants within the sample responded to all items of the four validated mental health scales used within these analyses. Cut-off scores were utilised to ascertain prevalence estimates of probable disorder. Two measures of overall mental health status are presented within analyses; (1) Any common mental disorder (CMD); If a participant scored above the cut-off on any of the mental health measures within the study (see below) they were classified as experiencing poor mental health. (2) Any mental health comorbidities (MHCs); If a participant scored above the cut-off on two or more of the mental health measures within the study (see below) they were classified as having mental health comorbidities. Depressive symptomology was measured using the 10-item Child Depression Inventory short form (CDI-S) [43, 44]. Items consisted of a series of statements (i.e. ‘I like myself’, ‘I don’t like myself’ ‘I hate myself’) for which adolescents had to choose the statement that reflected how they felt. Items were scored 0–2 (0 = absent symptoms, 1 = mild symptoms and, 2 = definitive symptoms) [43, 44]. Scores of ≥ 3 (based on mild/definitive symptoms due to the small number of participants reporting definitive symptoms) [45] were used to indicate symptomology consistent with a positive screen for probable depression (binary; yes/no). The CDI-S been previously found to be highly correlated (r = 0.89) with the broader 27-item CDI scale as such scores were prorated based on the suggested inclusive cut-off within the within the full scale [43, 46, 47]. A cut-off ≥ 3 has been previously been used in high income contexts [48]. The CDI (from which the CDI-S is derived) has generally been found to have strong psychometric properties and is well validated in the sub-Saharan African region (CDI-S: α = 0.66 in the current sample) [49–52]. The CDI-S has previously been used with adolescent populations in South Africa. n adolescents [47, 53–56]. Anxiety symptomology was measured using an abbreviated version (14 items) of the Children’s Manifest Anxiety Scale—Revised (RCMAS) [57, 58]. Adolescents identified if statements (i.e. “I worry a lot of the time”) were true for them. The 14 items were scored as “yes” (1) or “no” (0); indicative of experience consistent with anxiety symptomology and not, respectively. Scores ≥ 10 were used to indicate symptomology consistent with a positive screen for anxiety [57, 58]. The RCMAS has been validated within sub-Sharan Africa and shows good internal consistency among HIV-affected children and adolescents (α = 0.85 in the sample) [59]. Posttraumatic stress symptomology was measured using a 19-item version of the Child PTSD checklist [60, 61]. Adolescents were asked to think about something upsetting or frightening that has happened in their lives and identify how often they felt a series of items (i.e. “Do you get upset when you think about what happened?’” Items were scored 0–3 (0 “not at all”, 1 “some of the time”, 2 “most of the time”, 3 “all of the time”). Items within the checklist represent four domains relating to posttraumatic stress disorder (re-experience, avoidance, hyperarousal and, dysphoria), based on a four factor model proposed by Boyes et al. [61] for South African Adolescents [60]. Participants were classified as experiencing symptomology consistent with a probable positive screen for posttraumatic stress disorder if they scored on items across all four domains with affirmative responses (i.e. “most of the time”/”all of the time”) in the following frequencies: re-experience >  = 1, avoidance >  = 1, hyperarousal >  = 2, dysphoria >  = 2 [61]. Classifications were used to determine the presence of posttraumatic stress based on the DSM-5 criteria and were prorated based on the full Child PTSD scale [60, 62, 63]. The Child PTSD checklist showed good internal consistency within the sample (α = 0.84), has been widely used among adolescents and youth with South Africa [64, 65] and, the 19-item scale has been validated within the South African context [61]. Suicidality/self-harm was measured using the five-item Mini International Neuropsychiatric Interview (MINI-KID; scored 0–5) [66]. The MINI-KID used the following questions to identify suicidal symptoms: “In the past month did you: wish you were dead?” “Want to hurt yourself?” “Think about killing yourself?” “Think of a way to kill yourself?” “Attempt suicide?” Participants responded “yes” (1) or “no” (0). Participants were classified as reporting suicidal symptoms if they scored on any item on the MINI-Kid [66]. Globally, the MINI-KID has been extensively validated, demonstrates good internal consistency (α = 0.89 in the current sample), and good test–retest reliability [66–68]. All analyses were undertaken using STATA v.15. [69] Chi-square tests (Fisher’s exact test, where appropriate) and Kruskal Wallis tests were used to explore sample characteristics (inclusive of mental health status) according to motherhood status. Prevalence and associations of likely common mental disorder with adolescent motherhood, HIV status (including the two factors combined) were described descriptively and assessed using chi-square tests. Finally, logistic regression models were used to explore the cross-sectional associations between motherhood and HIV status (inclusive of interaction effects) and common mental disorder. Interaction effects of motherhood and maternal HIV status were assessed by introducing interaction terms into the multivariable models. Adjusted odds ratios from the models including interaction terms were used to develop forest plots as a visual representation of the odds of experiencing common mental disorder among adolescent mothers who are living with HIV. Confounding factors were included in multivariable regression models if they were identified as being relevant factors within the literature of interest or found to be associated (p =  < 0.2) [70, 71] with either, or both, the predictor and outcome variables.