How do new crises impact HIV risk behaviour – exploring HIV risk behaviour according to COVID-19-related orphanhood status in South Africa?

ABSTRACT

The COVID-19 pandemic resulted in high death rates globally, and over 10.5 million children lost a parent or primary caregiver. Because HIV-related orphanhood has been associated with elevated HIV risk, we sought to examine HIV risk in children affected by COVID-19 orphanhood. Four hundred and twenty-one children and adolescents were interviewed, measuring seven HIV risk behaviours: condom use, age-disparate sex, transactional sex, multiple partners, sex associated with drugs/alcohol, mental health and social risks. Approximately 50% (211/421) experienced orphanhood due to COVID-19, 4.8% (20/421) reported living in an HIV-affected household, and 48.2% (203/421) did not know the HIV status of their household. The mean age of the sample was 12.7 years (SD:2.30), of whom 1.2% (5/421) were living with HIV. Eighty percent (337/421) reported at least one HIV risk behaviour. HIV sexual risk behaviours were more common among children living in HIV-affected households compared to those not living in HIV-affected households and those with unknown household status (35.0% vs. 13.6% vs.10.8%, X2 = 9.25, p = 0.01). Children living in HIV-affected households had poorer mental health and elevated substance use (70.0% vs. 48.5%, X2 = 6.21, p = 0.05; 35.0% vs. 19.9%, X2 = 4.02, p = 0.1306, respectively). HIV-affected households may require specific interventions to support the health and well-being of children and adolescents.

KEYWORDS:

• Children
• adolescents
• COVID-19 pandemic
• HIV risk behaviours
• orphanhood

Background

To date, the COVID-19 pandemic has resulted in over 6.9 million deaths globally (World Health Organization, 2023), however, excess mortality figures suggest that between 14.83 million (Msemburi et al., 2023) and 18.2 million (Wang et al., 2022) lost their lives due to COVID-19. Most of these deaths have occurred among adults, and much of the emergent research, policy, and programming has centred on adults. However, a secondary, devastating consequence of the COVID-19 pandemic has been the number of children globally who have experienced orphanhood (Hillis et al., 2022). Such loss is also concentrated among groups who are already vulnerable (Unwin et al., 2022).

UNICEF defines orphanhood as the death of one or both parents (Joint United Nations Programme on HIV/AIDS, 2009). However, children and adolescents are often cared for by a network extending beyond their biological parents – particularly within contexts where multigenerational households are common. As such, it is important to consider caregivers in the definition of orphanhood, especially as grandparents were most vulnerable to COVID-19 (Jensen et al., 2021). Approximately 10.5 million children and adolescents experienced the death of a parent or a caregiver during the first 28 months of the pandemic (Hillis et al., 2021), yet there is limited understanding and discussion of their specific experiences and needs.

Children who have experienced orphanhood have been found to have greater risk of mental health issues (Ntuli et al., 2020), exposure to abuse (Cluver et al., 2011), and poverty (Cluver et al., 2011; Gentz et al., 2018). Specific support and interventions are required to avert negative psychosocial or developmental sequelae (Thomas, 2010). Data from earlier pandemics, including HIV and Ebola, as well as from crisis situations including conflict and displacement, has shown that orphanhood may negatively shape children’s trajectories (Berens & Nelson, 2015). Children who experience orphanhood may experience practical and logistical changes to their lives in the immediate aftermath of a parent/caregiver death, as well as the longer term (Alvis et al., 2023). These include loss of caregiving support (Dereje & Jibat, 2015; Neville et al., 2022), disruption of routines (Lachman et al., 2014), institutionalised care (Sherr et al., 2017) and shifts in supervision and protection of the child (Yendork & Somhlaba, 2015). They may also face economic ramifications, such as loss of income, diversion of resources to manage illness, as well as cope with legal or cultural expectations after death (Gentz et al., 2018).

In settings already characterised by persistent social, economic, and health-related risks, children experiencing orphanhood may face elevated risks to their own health and wellbeing (Cluver et al., 2014). Children experiencing orphanhood are more likely to experience economic vulnerability, and school disruption (Ardington & Leibbrandt, 2010; Thomas, 2010). Children and adolescents experiencing a lack of basic needs such as food, shelter, and school are particularly vulnerable to poor health and wellbeing, and these needs may drive HIV risk behaviours (Cluver et al., 2011, p. 6) such as early sexual debut (Mkandawire et al., 2013), multiple sexual partners (Neville et al., 2022), and transactional sex (Cluver et al., 2011).

In the case of the COVID-19 pandemic, negative outcomes linked to orphanhood may have been further amplified with accompanying social and economic policies designed to curb the impact of the virus: sudden, dramatic increases in parental and caregiver morbidity and mortality occurred alongside stringent lockdowns, schooling disruption, and service provision interruption (Adegboye et al., 2021). These extended responses to crises precipitated and occurred alongside loss of jobs and livelihoods, food insecurity and elevated levels of poverty (Kollamparambil & Oyenubi, 2021), as well as the closure of support avenues. Furthermore, the COVID-19 pandemic did not occur in a vacuum but may have affected families already facing other crises such as HIV.

In South Africa in particular, the consequences of the COVID-19 pandemic, including orphanhood, also echoed events from the HIV epidemic, a generation earlier. At the height of South Africa’s HIV epidemic in 2006, approximately 282,904 people died from HIV/AIDS (Statista, 2023). Coupled with delays in access to antiretroviral therapy and poor government response at this time of peak incidence, an estimated 720,000 (UNAIDS, 2022) children and adolescents in South Africa have experienced parent and/or caregiver loss. While South Africa’s HIV epidemic has improved substantially in the past two decades, it is still home to the greatest number of individuals living with HIV globally – 7.8 million (Jardim et al., 2022). South Africa also reported the greatest number of reported COVID-19 cases in Africa (Abdool Karim & Baxter, 2022).

Despite key differences in the viruses themselves, the twin pandemics of HIV and COVID-19 share similar social and economic drivers, and it is likely that lessons from the HIV epidemic may transfer to subsequent pandemic and caregiver loss situations, including COVID-19. For instance, COVID-19 orphanhood may well be layered onto pre-existing social systems where children already experience a multitude of social challenges, including being affected by HIV. The notion of cumulative orphanhood or loss is also a possibility due to the devastating impact of HIV and tuberculosis in South Africa (Mahtab & Coetzee, 2017).

The prospect of this layering of pandemics – and the lack of evidence on the impact of COVID-19 orphanhood on children and adolescents – warrants further attention. Given that evidence from the HIV pandemic indicates that AIDS-related orphanhood is associated with elevated HIV risk amongst children and adolescents (Ndlovu et al., 2023; Operario et al., 2011), this study therefore set out to understand the risk factors associated with COVID-19 for HIV acquisition, and to explore if this was magnified among those experiencing the double burden of COVID-19 and HIV in the family. The research questions were: (1) What is the HIV risk among children that lost caregivers due to COVID-19? and (2) what are the HIV risk behaviours among children and adolescents living in households affected by HIV?

Methods

Participants and procedure

We recruited a cross-sectional cohort of children and adolescents (9–18 years) living in the Cape Town metropolitan area (Khayelitsha), South Africa which is a densely populated and very low-income environment. Children who had experienced orphanhood (death of one or both biological parents), or caregiver loss (person who looks after the child or adolescent most) due to COVID-19 (n = 211), were included, as well as a comparison group who did not experience loss (n = 210). Children who experienced orphanhood from other causes were not included in this study. Data from 421 children and adolescents are included in these analyses.

Khayelitsha is South Africa’s third largest township with an estimated population size of 750,000. Residents of Khayelitsha experience high rates of poverty, with most residents living in informal housing with shared water sources and insufficient sanitation (Super, 2015). There is a high unemployment rate within the township (approximately 50%) (City of Cape Town, 2013) and high levels of crime (Freeman & McDonald, 2015). Over 100,000 people died as a result of COVID-19 in South Africa (WHO, 2023).

Participants were interviewed between July 2022 and April 2023, chosen for the period immediately after the peak waves of COVID-19 infection-related deaths (2020–2021). Sampling strategies included 4 strategies: (1) recruitment from schools; (2) non-government organisations; (3) door-to-door enquiries; and (4) referrals from local organisations. All data collection tools were piloted with children and adolescents prior to the commencement of the study. All participants completed a questionnaire collecting data on sociodemographic characteristics, health, and wellbeing (experience of COVID-19, HIV and mental health), relationships, support, schooling and education, behavioural risk, violence, stigma and bullying. Trained data collectors administered questionnaires using electronic tablets, and participants could complete the questionnaire in their language of choice (isiXhosa or English). For a detailed overview of study procedures, see Steventon Roberts et al. (2023).

Ethical considerations

Detailed consideration was given to the ethical conduct in the study given the sensitive nature of the topic under study. All participants provided written informed voluntary consent to participate in the study, and written consent was obtained from the primary caregiver or guardian of all children and adolescents under age 18. Ethical approval was obtained from Stellenbosch University’s HREC (N22/04/040). All measures were translated to isiXhosa and back translated. Participants could choose their language for response. All interviews were scheduled by appointment at a dedicated University facility in the local neighbourhood, with trained data collectors, a social worker on site for referrals and the provision of refreshments and transport. All participants were invited to be accompanied by a responsible adult. A token reimbursement was provided to cover time, inconvenience and any expenses by a local shopping voucher. Consent and reminder of freedom to withdraw were scheduled at the start and end of each interview. A referral protocol was associated with the study to ensure that linked service providers were available from a comprehensive referral directory to serve as onward referral avenues if the data collectors deemed this necessary. Training was provided to the data collectors to identify referral triggers based on questionnaire responses or observed need. The referral agencies included service providers with expertise in bereavement, mental health support, violence support, bullying support and issues of social concern such as food insecurity. Appointments and transport to referral agencies were included in the study protocol. Data collector supervision was carried out throughout the data collection phase and a qualified social worker was stationed at the data collection centre throughout the data collection phase.

Measures

COVID-19 associated orphanhood and caregiver loss were assessed at study recruitment via participant and caregiver reports. Orphanhood and caregiver loss were also confirmed by participants in a single item in the study questionnaire. Sociodemographic characteristics included child age, biological sex and basic demographic information including housing status and cash grant receipt. These measures were assessed using items from the South African Census (SSA, 2001). Personal and household HIV status was measured by study-specific self-report measures documenting whether the participant had ever been tested for HIV, and if anyone in their household was living with HIV. Any positive answers to these two questions were marked as “affected by HIV”.

HIV risk behaviour was assessed utilising seven self-report measures focusing on sexual risk behaviour, mental health, and social risk behaviour. Five sexual risk measures were drawn from the National Survey of Risk Behaviour among South Africans (Pettifor et al., 2005), including early sexual debut (age at first sex <16 years); multiple sexual partners in the previous 12 months; contraception risk in the last month (lack of contraception use within sexual activity); lifetime engagement in transactional sex (receiving monies or gifts in exchange for sex); engaging in sex under the influence of drugs or alcohol. Three aspects were used to measure mental health. Depressive symptoms were measured by the Patient Health Questionnaire version 9 (PHQ-9) (Johnson et al., 2002). It includes 9 items and is validated as a diagnostic tool. Anxiety was assessed by the Generalised Anxiety Disorder – 7 measure (Spitzer et al., 2006) which is a 7-item scale again with a validated cut off score to indicate levels of concern. Suicidality symptoms were evaluated by the Mini-Kid (Sheehan et al., 2004). Scoring above the cut off for any of the three measures could indicate a poor mental health score. Social risk behaviour and substance use were measured using the 17-item rule-breaking sub-scale of the Child Behaviour Checklist (CBCL) (Achenbach, 2000). Each item was scored 0–3 (0 = never engaged with rule-breaking behaviour, 3 = definitely engaged with rule-breaking behaviour).

Items were clustered to develop a composite risk score (0–7) and a binary variable indicating any HIV risk behaviour (scoring on any of the seven items). These seven items were also clustered to explore any sexual risk behaviour (scoring “yes” on any of the sexual risk measures above), any poor mental health (see above), and any social risk behaviour (scoring above 0 on any item of the CBCL rule-breaking sub-scale). Social risk behaviour was also explored according to total social behaviour risk score (0–51; total score from the 17 items) and number of social risk behaviours (how many items on the rule-breaking sub-scale participants scored above 1 [scored 0–3]).

A note was kept if any participant was referred to any support service. This was recorded as a binary (yes/no) on the outcome data.

Statistical analyses

Analyses examined the cross-sectional relationship between HIV risk behaviour and orphanhood status, and HIV risk behaviour and household HIV status. Sample characteristics were described and explored according to COVID-19 orphanhood/caregiver loss status using t-tests, Wilcoxon rank-sum tests, and chi-square tests, as appropriate. Likewise, HIV risk behaviours were explored according to COVID-19 orphanhood/caregiver loss status using t-test and chi-square tests. Kruskal Wallis H Tests for nonparametric analysis of variance (ANOVA) and chi-square tests were used to examine HIV risk behaviours according to household HIV status, and Dunn’s post hoc tests were undertaken to further explore differences between groups. HIV risk behaviours were further examined according to orphanhood and household HIV status using exploratory linear and logistic regression models. Covariates were included in multivariable regression models if they were found to be associated with both or either the predictor and outcome variables or were identified as relevant in the literature. Due to the small number of participants known to be living in a household affected by HIV, only selected covariates were included within regression models exploring risk behaviour according to household HIV status. All data analyses were undertaken using Stata v.18.

Results

Sample characteristics

Approximately fifty percent of the sample were classified as experiencing orphanhood or caregiver loss due to COVID-19. The mean age of the sample was 12.7 years (SD: 2.3). Forty percent (40.4%) of the sample lived in informal housing. The majority of the sample lived in a household that had access to government cash grants (90%). A small proportion of the sample were known to be living with HIV (1.2%) and 4.8% % lived in household with someone residing therein, known to be living with HIV. Almost one-fifth of the sample reported having tested for HIV (21.1% %). Sample demographic characteristics did not differ according to orphanhood status (see Table 1).

Table 1. Sample characteristics according to COVID-19 related orphanhood and caregiver loss status.

	Total sample (n = 421)	Orphanhood or caregiver death resulting from COVID-19 (n = 211)	No orphanhood or caregiver death (n = 210)	t, X2, p-value
Age (years)	12.7 (SD:2.3)	12.7 (SD:2.4)	12.7 (SD: 2.2)	−0.24, 0.81
Biological sex (Female)	208 (49.4%)	110 (52.1%)	98 (46.7%)	1.26, 0.26
Living in informal housing	170 (40.4%)	82 (38.9%)	88 (41.9%)	0.41, 0.52
Accessing cash grants	377 (90.0%)	185 (87.7%)	192 (91.4%)	1.58, 0.21
Living with HIV	5 (1.2%)	3 (1.4%)	2 (0.95%)	0.08, 0.78
Ever tested for HIV				0.66, 0.72
Yes	89 (21.1%)	48 (22.8%)	41 (19.5%)
No	328 (77.9%)	161 (76.3%)	167 (79.5%)
Declined to answer	4 (0.95%)	2 (0.95%)	2 (0.95%)
Household HIV	20 (4.8%)	12 (5.7%)	8 (3.8%)	11.4, 0.03

HIV risk behaviour and COVID-19 related orphanhood/caregiver loss

Within the total sample, HIV risk score was found to be 1.61 (SD: 1.34, range 0 −7). Most of the sample (80.1%) reported at least one HIV risk behaviour. Sexual risk behaviour was reported by 13.3% of the sample. Over half of the sample reported poor mental health (54.4%), and 71.3% reported at least one social risk behaviour. The average social risk score was 3.03 (SD: 4.04). Social risk scores were higher among children and adolescents experiencing orphanhood when compared to those not experiencing orphanhood, although this difference was not statistically significant (3.44 vs. 2.62, t = −0.80, p = 0.42). All other HIV risk behaviours were similar according to orphanhood status, (see Table 2). Table 3 presents a series of regression models exploring associations between orphanhood status and HIV risk behaviours. Experiencing orphanhood remained a risk factor for increased social risk scores (B = 0.84 [95%CI: 0.08, 1.59 ], p = 0.03) after controlling for covariates. Male biological sex and increased age in years were both found to be risk factors for HIV risk behaviours (see Table 3).

Table 2. HIV risk behaviour among children and adolescents according to COVID-19 related orphanhood and caregiver loss status.

	Total sample (n = 421)	Orphanhood or Caregiver death resulting from COVID-19 (n = 211)	No orphanhood or caregiver death (n = 210)	W, X2, p-value
HIV risk score (0–7)	1.61 (SD:1.34)	1.62 (SD:1.35)	1.60 (SD:1.33)	−0.09, 0.93 (Wilcoxon Rank Sum Test)
Any HIV risk behaviour	337 (80.1%)	169 (80.1%)	168 (80.0%)	0.001, 0.98
Number of HIV risk behaviours				1.14, 0.77
0	84 (20.0%)	42 (19.9%)	42 (20.0%)
1	125 (29.7%)	64 (30.3%)	61 (29.1%)
2	158 (37.5%)	75 (35.6%)	83 (39.5%)
3+	54 (12.8%)	30 (14.2%)	24 (11.4%)
Any sexual risk behaviour	56 (13.3%)	31 (14.7%)	25 (11.9%)	0.74, 0.39
Any poor mental health	229 (54.4%)	118 (55.9%)	111 (52.9%)	0.40, 0.53
Any social risk behaviour	300 (71.3%)	146 (69.2%)	154 (73.3%)	0.88, 0.35
Social risk score (0–51)	3.03 (SD:4.04)	3.44 (SD:4.77)	2.62 (SD:3.11)	−0.80, 0.42 (Wilcoxon Rank Sum Test)
Number of social risk behaviours				3.07, 0.38
0	121 (28.7%)	65 (30.8%)	56 (26.7%)
1	88 (20.9%)	37 (17.5%)	51 (24.3%)
2	52 (12.4%)	27 (12.8%)	25 (11.9%)
3+	160 (38.0%)	82 (38.9%)	78 (37.1%)

Table 3. Linear and logistic regression models exploring the association between HIV risk behaviours and COVID-19 related orphanhood.

	HIV risk score (0–7)		Any HIV risk behaviour		Any sexual risk behaviour		Any poor mental health		Any social risk behaviour		Social risk score (0–51)
	B (95% CI)	P	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	OR (95% CI)	p	B (95% CI)	P
Model 1.
Orphanhood	0.026 (−0.23, 0.28)	0.85	1.01 (0.62–1.62)	0.98	1.27 (0.72–2.24)	0.40	1.13 (0.77–1.66)	0.52	0.82 (0.54–1.25)	0.35	0.82 (0.05, 1.59)	0.04
Model 2.
Orphanhood	0.025 (−0.20, 0.25)	0.83	1.02 (0.62–1.69)	0.94	1.1 (0.58–2.34)	0.67	1.13 (0.76–1.67)	0.56	0.81 (0.52–1.26)	0.35	0.84 (0.08, 1.59)	0.03
Biological sex (female)	−0.35 (−0.58, −0.13)	0.002	0.57 (0.34–0.94)	0.03	0.44 (0.22–0.91)	0.03	0.803 (0.54–1.19)	0.28	0.66 (0.42–1.03)	0.07	−1.08 (−1.83, −0.33)	0.005
Age (years)	0.27 (0.22, 0.32)	<0.001	1.38 (1.22–1.55)	<0.001	2.60 (2.03–3.32)	<0.001	1.21 (1.11–1.32)	<0.001	1.32 (1.20–1.47)	<0.001	0.37 (0.21, 0.53)	<0.001
Informal housing	−0.05 (−0.28, 0.18)	0.68	0.77 (0.46–1.27)	0.30	1.20 (0.60–2.41)	0.61	1.03 (0.69–1.54)	0.89	0.81 (0.52–1.27)	0.37	0.13 (−0.64, 0.89)	0.75
No receipt of cash grants	0.10 (−0.27, 0.47)	0.60	1.40 (0.58–3.40)	0.45	1.44 (0.48–4.28)	0.51	1.56 (0.80–3.02)	0.19	1.12 (0.54–2.33)	0.75	0.78 (−0.45, 2.01)	0.24

HIV risk behaviour and household HIV status

Table 4 presents the frequencies and averages of HIV risk behaviours according to household HIV status (household known to be living with HIV, no HIV in household, and unknown HIV status). HIV risk behaviours were found to be elevated among children and adolescents living in households affected by HIV compared to children living in households who are not affected by HIV, and those who did not know the HIV status of their household (HIV risk score: 2.60 vs. 1.50 vs. 1.62, respectively FH = 7.20, p = 0.03; any HIV risk behaviour: 90.0% vs. 74.8% vs. 84.2%, F = 6.95, p = 0.03; any sexual risk behaviour: 35.0% vs.13.6% vs.10.8%, F = 9.25, p = 0.001). Children and adolescents living in households known to be affected by HIV reported a higher prevalence of any poor mental health (70.0% vs. 48.5% vs. 58.6%, F = 6.21, p = 0.05). Prevalence of social risk behaviour was found to be similar regardless of household HIV status (80.0% vs. 67.7% vs. 73.9%, respectively, chi-squared = 2.67, p = 0.26; see Table 4). Table 5 compares three groups, COVID orphanhood plus background HIV (n = 12) COVID orphanhood only (n = 199), No COVID orphanhood/no HIV background (n = 203). This table identifies a significantly higher HIV risk score for those layered with both COVID orphanhood and HIV in the background. Similarly, this group has a significantly higher level of any sexual risk behaviour (41.7 vs 13.1% and 11.4% f = 9.14 p = .01).

Table 4. HIV risk behaviour according to child and adolescent household HIV status.

	Household affected by HIV
	Household known to be living with HIV (n = 20)	No (n = 198)	Don’t know (n = 203)	H, X2, p-value
HIV risk score (0–7)	2.60 (SD:1.88)	1.50 (SD:1.32)^a	1.62 (SD:1.25)^a	7.20, 0.03 (Kruskall Wallis)
Any HIV risk behaviour	18 (90.0%)	148 (74.8%)	171 (84.2%)	6.95, 0.03
Number of HIV risk behaviours				16.50, 0.01
0	2 (10.0%)	50 (25.3%)	32 (15.8%)
1	4 (20.0%)	54 (27.3%)	67 (33.0%)
2	7 (35.0%)	69 (34.9%)	82 (40.4%)
3+	7 (35.0%)	25 (12.6%)	22 (10.8%)
Any sexual risk behaviour	7 (35.0%)	27 (13.6%)	22 (10.8%)	9.25, 0.01
Any poor mental health	14 (70.0%)	96 (48.5%)	119 (58.6%)	6.21, 0.05
Any social risk behaviour	16 (80.0%)	134 (67.7%)	150 (73.9%)	2.67, 0.26
Social risk score (0–51)	3.55 (SD:3.24)	2.73 (SD:3.54)	3.27 (SD:4.54)	2.81, 0.25 (Kruskall Wallis)
Number of social risk behaviours				15.3, 0.02
0	4 (20.0%)	64 (32.3%)	53 (26.1%)
1	3 (15.0%)	41 (20.7%)	44 (21.7%)
2	0 (0.00%)	17 (8.6%)	35 (17.2%)
3+	13 (65.0%)	76 (38.4%)	71 (35.0%)

Note: ^aStatistically different from group known to be living in a household affected by HIV (p = <0.05).

Table 5. HIV risk behaviour according to combined child and adolescent orphanhood and household HIV status.

	COVID-19 + HIV (n = 12)	COVID-19 Only (n = 199)	No COVID-19 (n = 203)	H, X2, p-value
HIV risk score (0–7)	2.83 (1.95)	1.55 (1.28)^a	1.57 (1.30)^a	5.57, 0.06 (Kruskall Wallis)
Any HIV risk behaviour	11 (91.7%)	158 (79.4%)	161 (79.7%)	1.07, 0.59
Number of HIV risk behaviours				10.70, 0.10
0	1 (8.3%)	41 (20.6%)	41 (20.3%)
1	2 (16.7%)	62 (31.2%)	59 (29.2%)
2	4 (33.3%)	71 (35.7%)	80 (39.6%)
3+	5 (41.7%)	25 (12.6%)	22 (10.9%)
Any sexual risk behaviour	5 (41.7%)	26 (13.1%)	23 (11.4%)	9.14, 0.01
Any poor mental health	8 (66.7%)	110 (55.3%)	105 (52.0%)	1.24, 0.54
Any social risk behaviour	10 (83.3%)	136 (68.3%)	148 (73.3%)	2.08, 0.35
Social risk score (0–51)	4.08 (3.70)	3.40 (4.84)	2.61 (3.14)	2.31, 0.32 (Kruskall Wallis)
Number of social risk behaviours				8.53, 0.20
0	2 (16.7%)	63 (31.7%)	54 (26.7%)
1	2 (16.7%)	35 (17.6%)	50 (24.8%)
2	0 (0.0%)	27 (13.6%)	25 (12.4%)
3+	8 (66.7%)	74 (37.2%)	73 (36.1%)

N.B. HIV only group excluded (n = 8).

Table 6 presents a series of regression models exploring the association between household HIV status and HIV risk behaviours. Within multivariate models, living in a household affected by HIV was found to be associated with increased HIV risk score (B = 1.1 [95%CI: 0.49, 1.71 ], p < 0.001) and an increased risk of sexual risk behaviour (OR: 3.31 [95%CI: 1.21–9.06], p = 0.02). A trend was also identified in relation to living in a household affected by HIV being associated with any poor mental health (OR: 2.44 [95%CI: 0.90–6.61], p = 0.08; see Table 6).

Table 6. Regression models exploring in the association between HIV risk behaviours and household HIV status.

	HIV risk score (0–7)		Any HIV risk behaviour		Any sexual risk behaviour		Any poor mental health		Any social risk behaviour		Social risk score (0–51)
	B (95% CI)	p	OR (95% CI)	p	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	p	B (95% CI)	P
Model 1.
Household not living with HIV	Ref.		Ref.		Ref.		Ref.		Ref.		Ref.
Unknown household HIV status	0.12 (−0.14, 0.37)	0.38	1.81 (1.10–2.96)	0.02	0.77 (0.42–1.40)	0.39	1.51 (1.01–2.23)	0.04	1.35 (0.88–2.08)	0.17	0.54 (−0.26, 1.33)	0.20
Known household HIV status (living with HIV)	1.10 (0.49, 1.71)	<0.001	3.04 (0.68–13.57)	0.15	3.41 (1.25–9.31)	0.02	2.48 (0.92–6.71)	0.07	1.91 (0.61–5.95)	0.26	0.82 (−1.05, 2.68)	0.39
Model 2.
Household not living with HIV	Ref.		Ref.		Ref.		Ref.		Ref.		Ref.
Unknown household HIV status	0.13 (−0.13, 0.39)	0.33	1.91 (1.15–3.17)	0.01	0.77 (0.42–1.41)	0.39	1.52 (1.02–2.26)	0.04	1.40 (0.90–2.18)	0.13	0.53 (−0.26, 1.33)	0.20
Known household HIV status (living with HIV)	1.07 (0.46, 1.67)	0.001	2.94 (0.65–13.24)	0.16	3.31 (1.21–9.06)	0.02	2.44 (0.90–6.61)	0.08	1.85 (0.59–5.80)	0.29	0.71 (−1.14, 2.57)	0.45
Biological sex (female)	−0.32 (−0.57, −0.07)	0.013	0.60 (0.37–0.98)	0.04	0.69 (0.39–1.24)	0.22	0.83 (0.56–1.22)	0.35	0.68 (0.44–1.04)	0.08	−1.01 (−1.78, −0.24)	0.011
Informal housing	−0.14 (−0.40, 0.12)	0.30	0.66 (0.40–1.09)	0.10	0.99 (0.55–1.77)	0.97	0.93 (0.63–1.39)	0.73	0.74 (0.48–1.15)	0.18	−0.06 (−0.86, 0.73)	0.87

Notes: HIV risk behaviours = Early sexual debut, multiple sexual partners, inconsistent use of contraception, transactional sex, sex under the influence of drugs or alcohol, any poor mental health, any social risk behaviour (CBCL).

Any sexual risk behaviours = Early sexual debut, multiple sexual partners, inconsistent use of contraception, transactional sex, sex under the influence of drugs or alcohol.

Any poor mental health = Scoring above cut-off on depression, anxiety, suicidality symptomology scales.

Any social risk behaviour = scoring on any item of 17 item rule breaking subscale of the child behaviour checklist. Each item scored 0–3 (never engaged with behaviour – definitely have).

Within the study, the number of referrals to relevant services was elevated among children and adolescents living in households affected by HIV compared to children and adolescents not living in households affected by HIV, and those who did not know the HIV status of their household (50% vs. 21.7% vs. 37%, respectively, X²= 14.8, p = 0.001; see Table 7).

Table 7. Referral receipt according to HIV household status.

	Yes (n = 20)	No (n = 198)	Don’t know (n = 203)	X2, p-value
Received referral	10 (50.0%)	43 (21.7%)	75 (37.0%)	14.8, 0.001

Discussion

This study was set up to examine any HIV – related risk behaviour which may be associated with COVID-19-related orphanhood given that HIV-related orphanhood has been shown to link to elevated HIV risk behaviour in youth. Globally, to the best of our knowledge, this is the first study to explore characteristics relating to HIV in the context of COVID-19-associated orphanhood and caregiver loss. We found that HIV risk behaviours were elevated for both children and adolescents that had experienced orphanhood and caregiver loss due to COVID-19 as well as the ones that did not experience loss. However, social risks were significantly higher in the children and adolescents that experienced orphanhood and caregiver loss. In terms of risk behaviours and HIV-affected homes, children living in HIV-affected households had a greater prevalence of HIV sexual risk behaviours than those living in non-HIV-affected households or those with unknown household status. Furthermore, when compared to their non-HIV-affected counterparts, children living in HIV-affected households exhibited a higher burden of potential mental health challenges and substance abuse. Lastly, children and adolescents living in households affected by HIV were more likely to be referred for additional services (such as counselling, bereavement, food insecurity) in comparison to children not living in household affected by HIV and those with unknown household status.

Emerging data from this study, conducted in a high-density, peri-urban area outside of Cape Town, highlights several overlapping challenges facing children and adolescents – both those who experienced COVID-19-related caregiver loss and those who did not. Building on well-established evidence from HIV literature showing how parental loss is associated with elevated risk behaviours (Ndlovu et al., 2023; Operario et al., 2011) our data suggests that COVID-19-associated loss may follow some similar patterns, but was not conclusive. Importantly, when such loss is layered onto pre-existing HIV considerations, the risks are significantly higher. Previous research has demonstrated that children experiencing orphanhood have greater odds of risk behaviour in comparison to their non-orphaned peers (Lee et al., 2023; Operario et al., 2011). Although behavioural risk was present among all groups in our analyses, for those experiencing compound exposures (COVID 19 plus HIV in the family) the risks were significantly higher.

While poverty and socioeconomic vulnerability are well-understood as key drivers of HIV risk behaviour (Cluver et al., 2022; Leung Soo et al., 2023) and also increase the likelihood of morbidity and mortality in pandemics such as HIV and COVID-19 (Hussey et al., 2021), it is critical to consider the important role of mental health, as well as an array of social and economic protections, in driving these risks for children and adolescents. Based on the findings of this study, children and adolescent living in households affected by HIV had poorer mental health (depressive, anxiety and suicidality symptoms) and were more likely to be referred for additional support services in comparison to those not affected by HIV and those with unknown HIV status. This finding is consistent with prior research indicating the impact of HIV on the family system, particularly the mental health of children and adolescents (Gamarel et al., 2017; Sherr et al., 2014). Family-based approaches are needed to provide holistic and contextually responsive services (Gamarel et al., 2017). As poor mental health is in itself a driver of HIV-associated risk, attention to mental health would be vital.

For children experiencing multiple adversities, including COVID-19 bereavement, service provision needs to be better attuned to the complex, embedded challenges they may be navigating. A more complex model of pre-existing vulnerabilities and compound exposures is needed to inform both preparation and responsive interventions. Children and adolescents who have experienced orphanhood and caregiver loss should be active participants in the process of identifying solutions and ways to best support them (UNICEF, 2021). In low resource settings provision of food and mental health promotion will be useful services. Given the sensitivity and high level needs within such groups, particular attention needs to be paid to ethics, consent and referral for such groups so that insights can be gathered and they can inform the provision of service need for the future.

Data should be interpreted in the context of study limitations. First, data are cross-sectional and therefore the direction of causality, and causality overall, cannot be determined. Future explorations of longitudinal data are required further examine the causal relationship between orphanhood and risk behaviour. Second, data are drawn from self-reported responses of children and adolescents, as such the impacts of bias should remain a consideration and alternative observation perspectives may be useful. This is particularly relevant in relation to potentially stigmatised issues e.g., HIV status of individuals in the households, children may not have or report accurate information. As such, there is potential for such measures to be underreported. Third, only a small proportion of the sample reported that they lived in a household affected by HIV. As such, these analyses are seemingly underpowered. Analyses were also underpowered to explore differences according to combined COVID-19-associated orphanhood/caregiver loss status, and household HIV status. Despite this, significant differences were found and future studies on this particular group may be needed to provide more detailed insight to the extent and nature of the double burden. Nevertheless, these analyses highlight the potentially layered impacts of both the COVID-19 pandemic and the HIV epidemic in South Africa.

Conclusions

This research adds valuable contribution by providing insight into the needs and experiences of children that have experienced orphanhood and caregiver loss in the COVID-19 era, in a context where HIV/AIDS is deeply embedded. When there is a layering of pandemics, the most vulnerable children and adolescents suffer the most. More research is required to better understand ways to provide holistic, and cost-effective interventions, especially in limited resource settings.

Data sharing and accessibility

Prospective users, policymakers/government agencies/researchers (internal/external) will be required to contact the study team to discuss and plan the use of data. Research data will be available on request subject to participant consent and having completed all necessary documentation. All data requests should be sent to the Principal Investigators.

Acknowledgements

The authors are grateful to the families who participated within the study, the tireless data collection team, and our partner organisations who supported the research process.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by UKRI Global Challenges Research Fund (GCRF), Oak Foundation/GCRF “Accelerating Violence Prevention in Africa” [grant number OFIL-20-057], Accelerating Achievement for Africa's Adolescents (Accelerate) Hub [grant number ES/S008101/1], Wellspring Philanthropic Fund [grant number 16204], and UNICEF ESARO.