Prevention of Mother-to-Child HIV Transmission (PMTCT) programs in sub-Saharan Africa have been extremely successful in reducing the numbers of children infected with HIV [1]. In South Africa with its extensive use of effective PMTCT regimens new HIV infections in infants in the future should be rare [2, 3]. However, in high HIV prevalence areas, at least 40% of pregnant women may be HIV-infected, and as a consequence large numbers of HIV-uninfected children are being cared for by at least one HIV-infected parent [4]. HIV-infected mothers face the dilemma of when, and how, to disclose their own HIV status to their children [5–7].

Research, mainly from the United States of America (USA) and Europe, has shown that maternal HIV disclosure has benefits for mothers and children [8–10], including improvements in maternal mental health [11–14] and the quality of relationship between mother and child [9, 10], increased preparation for the child’s future in terms of custody and care planning for emergencies [8, 15, 16], and improvements in children’s emotional and social functioning [16, 17]. However, disclosure rates remain low with between 30-45% of mothers reporting HIV disclosure to their children in both observational and intervention studies [5–7, 18, 19]. Only a few studies have explored paternal HIV disclosure [20–23], and most intervention research has focused on mothers disclosing to their children. While the reasons for this are likely pragmatic since the majority of HIV-affected children live with, and are cared for, primarily by their mothers, this exclusion of fathers (whether biological or social) is increasingly critiqued in the literature on family interventions in the context of HIV [24].

There is limited research relating to maternal HIV disclosure in the African context [7], and few studies globally have targeted mothers with primary school-aged children [9, 25–27]. Only one recent intervention study from the United States (US) evaluated a maternal HIV disclosure intervention (the TRACK program) targeted at primary school-aged children [28], demonstrating that disclosure interventions in pre-adolescence are not only feasible, but also effective among mothers who had no intention to disclose to their children at the start of the study. TRACK was a randomized controlled trial of an intervention which included three home visits, telephone support and educational materials. Intervention mothers were over four times more likely to disclose, with a disclosure rate of 33% in the intervention compared to 7.3% in the control arm.

More generally in the literature, mothers frequently express the desire to disclose their HIV status to their children themselves, but report feeling unsure about how to approach the issue [18, 19, 29, 30]; are worried about addressing matters concerning sex and sexuality [15, 31, 32]; and have concerns that disclosure may cause emotional difficulties for their child [18, 19, 29]. A significant amount of international research has been conducted on the consequences of a lack of maternal HIV disclosure among 6–18 year olds [33]. While few mothers report disclosing to their children in the primary school years many are forced to disclose before adolescence following HIV-related illnesses or hospitalisations [21], which impacts negatively on children’s emotional and behaviour outcomes in case–control studies [5–7]. In most disclosure research to date mothers emphasize the need for assistance in planning and preparing towards disclosure [5, 19, 29, 31, 32], an important consideration in the design of disclosure interventions for Southern Africa.

In considering approaches to maternal HIV disclosure in the Southern African context, unique challenges exist. Although the largest population of HIV-infected parents lives in Africa, very little research has examined maternal HIV disclosure to children in this region. As access to HIV treatment increases, so does exposure to HIV at a family level. For example, by January 2010, 3% of the population in the research area had initiated ART, and as a result 40% of the population shared household or living arrangements with people who had either started ART or were enrolled into pre-ART care [34]. Therefore, children living in communities where ART is being scaled up are more likely to be socially exposed to HIV at younger ages. In general, increases in access to treatment also tend to reduce stigma within communities and families, allowing for more preventive activities at a household and family level [35].

Concurrently, there is growing evidence of the risks facing pre-adolescent and adolescent children growing up in HIV-affected families in Southern Africa [36–38]. Therefore, disclosure interventions that target younger children in these contexts may have the benefit of helping not only with parental disclosure but also of promoting more generalised preventive interventions including improved quality of parent–child relationships, health promotion and custody planning. Given that South Africa has the largest antiretroviral treatment (ART) programme in the world there are unique opportunities to explore psychological interventions, using family approaches, to develop culturally-appropriate disclosure interventions. These interventions should draw on existing evidence, and be feasible for integration into standard of care psychosocial support for HIV-infected parents, if found to be effective. We describe here the development and piloting of a complex psychological intervention (the ‘Amagugu’ ^a Intervention) to support age-appropriate maternal HIV disclosure to primary school-aged children, aged 6–10 years, to be delivered by lay counsellors or community health care workers in poorly resourced settings.

All 24 mothers approached agreed to participate, and all completed the intervention. Most women required only one session of each of the visits, however two women repeated Visit 2 (where the mother relates her life and HIV stories) on the suggestion of the counsellor who felt they needed more counselling to work through personal issues before engaging in the intervention training. One mother repeated Visit 4 (where the mother practises the intervention tools she will use with her child), because she requested further opportunities to practise with the counsellor to build her confidence. Seventeen women were literate in both English and Zulu, six were literate in Zulu only and one participant was illiterate. The illiterate participant expressed no reservations about proceeding with the intervention and elicited assistance from a literate family member to assist with reading where required.

Data relating to self-reported health status of mothers is shown in Table 2. The mean age of mothers was 33.5 (median 32; range 24–46) years, most women were living healthily with CD4 counts above 300 (mean 339; median 361; range 1–747) while 9 women had initiated ART (current eligibility for ART in South Africa is a CD4 count of <350 cells/ml). Of the 24 women, 20 reported being unemployed; six received a regular remittance; 23 received a government child support grant; and 11 reported that the child’s father was financially contributing towards the study child's care. Most (19) women reported being unmarried but in a relationship with a long-term partner. In 18 cases the father of the study child was still alive, and eight of the mothers were still in a relationship and living with him. A further two were living with new partners while 14 were living on their own with their children and/or other family members. Fifteen of the 24 women had disclosed their HIV status to their current partner, 12 to a parent and a sibling and 10 to adult friends outside the household. Only three had not disclosed to any other adult, all three of whom were living with a new partner who was unaware of their HIV status. None of these three women had initiated ART and their last CD4 counts were 568, 380 and 200 cells/ml respectively.

Five mothers indicated that they had either ‘partially or fully disclosed’ to their children prior to this intervention, mostly driven by pragmatic reasons following an illness and a need to ensure child care. In all five cases the mothers had fully disclosed (i.e. used the words “HIV”) to their older children (10–18 years), but had told their younger children (<10 years) that they had an illness which required them to go to hospital. While HIV was spoken about with older children and other family members, and it was possible the younger children were aware of, or had heard about HIV, in the family, in these five cases the mothers were unsure what the younger child knew and understood about HIV at baseline. As a result, all five of these mothers wished to participate in this intervention to disclose fully to the younger index child, as they felt their previous partial disclosures around illness and hospitalisations had been inadequate. These mothers were considered to have a clear intention to disclose at the start of the intervention. Of the remaining 19 women, three had disclosed fully to an older child but had made no disclosure to the younger index child while 16 had not disclosed to any of their children regardless of their age. None of these 19 mothers expressed any intention to disclose to the index child prior to enrolling in the study.

The most salient concerns about disclosure prior to the disclosure intervention were that the child was too young (22/24), might suffer emotionally afterwards (21/24), would disclose to others (21/24) or that the mother felt unable to answer difficult questions relating to the source of the HIV infection or death (19/24).

All 24 mothers chose to disclose something about their illness to their children during the intervention, but levels of HIV disclosure differed. Almost half the mothers (11) disclosed fully using the words “HIV” and indicating to the child that they were infected with HIV, while the remaining 13 disclosed partially, explaining that they had a virus, and using the tools to explain what the virus did in their bodies and how it can be controlled, without specifically using the words “HIV”. Among the 19 mothers who had not disclosed previously, and had no intention to disclose at the start of the intervention, 8/19 fully disclosed using the words ‘HIV’ while 11/19 chose to undertake partial disclosure. In two cases indicated by *in Table 3 (Family Study Numbers 20 and 21), the mother disclosed using the words "Virus" during the disclosure activity, but the child subsequently began asking questions about “AIDS” in the case of Family 20 and directly asked for the virus name in the case of Family 21. In Family 21 the mother did not name “HIV” while in Family 20 the mother shifted to discussing HIV openly. In both these cases the maternal disclosure is listed as partial as this most closely reflects the mothers’ intention and action during the intervention disclosure. However, in these two cases it is clear that for one child (Family 20) significant knowledge of HIV may have preceded the intervention, and for the other (Family 21) when the mother was presented with an option to disclose more fully she continued with partial disclosure. It is noteworthy that of the 5/11 mothers who stated they had disclosed fully to older children prior to the intervention and intended to disclose fully to younger children as part of the intervention, only 2/5 subsequently disclosed fully and reported their children were calm (Family 8 and 15); the remaining 3/5 chose to disclose partially. Unlike in Family 20, in Families 23 and 24 the children’s questions following partial disclosure did not indicate any knowledge of HIV or willingness to raise questions about HIV in response to the partial disclosure.

Most mothers described the reaction of their children to the disclosure intervention as calm, accepting and confident (14), eight reported that their child was initially confused and in need of further explanation, and only two reported fear or emotional distress as the initial reac-tion to disclosure. Table 3 summarises children’s reactions, commonly asked questions after disclosure and mothers’ experiences of undertaking disclosure using the intervention materials. A few mothers (6/24) reported that there were aspects of disclosure that they did not enjoy, including: feeling frustrated that the child did not understand initially, feeling disappointed that the child had not asked more questions, feeling concerned that knowing about her HIV status was hurtful for the child and caused emotional distress, finding it difficult to talk openly about HIV, and feeling nervous about using the HIV Body Map and saying the words “HIV” out loud. Two mothers reported not enjoying the Family Life Line exercise (Session 4, Table 1), the first because it brought up difficult issues around step-siblings in the household who were raised by the current partner, and the other because it raised difficult issues about bereavement and people who had died which she felt ill-prepared for. The most salient challenge raised by mothers in the post-disclosure interview was finding a quiet space in the home to undertake disclosure, with many mothers reporting having to deal with interruptions by younger children who wanted to play with the intervention materials.

While family members were encouraged to participate in the intervention, only a few (3/24) helped the mother with practice and preparation, two of whom subsequently participated in the disclosure event with the mother. However, more family members became involved in health promotion and custody planning activities following the disclosure being completed by the mother. Four mothers reported challenges in taking their children to clinic as they were employed: in two cases the father of the child completed the clinic component of the intervention, and in the remaining two an older sibling accompanied the child to the clinic.

This pilot study with 24 families showed high levels of acceptability and encouraging results in terms of effectiveness as defined by mothers undertaking full or partial disclosure. Almost half the mothers who had never previously disclosed to a child in the family, disclosed to a child aged 6–10 years making a full disclosure using the words ‘HIV’ as part of this intervention. Participation and disclosure rates suggest the intervention package is culturally acceptable and there appear to be no major barriers to intervention support using this family-centred model. Thus, the intervention shows promise for supporting maternal HIV disclosure in rural HIV endemic settings in Southern Africa.

Our data suggest that decisions around disclosure are complex, and the choice to make a full or partial disclosure may be dependent on several factors. The small sample size in this pilot prevents any meaningful analysis of factors associated with the choice to make a full or partial disclosure, for example child age or gender.

It is possible that the younger age of most children in this sample may have influenced mothers’ preference for partial disclosure, but it is also possible that the preference for partial disclosure was influenced by other factors such as the mothers’ personal readiness or household factors not captured in the pilot. Further research is required to better understand these decisions and how disclosure intentions of mothers prior to the intervention influence whether or how much they disclose to their children.

While these findings are preliminary and should be interpreted with caution, they are encouraging. The intervention is innovative in that it targets a young age group in a high HIV prevalence area. To our knowledge this is the first intervention targeting primary school-age children (6–10 years) from a resource-limited setting, with the intervention messages focusing not only on HIV disclosure itself but also on health promotion for pre-adolescent children. As in the TRACK program [28], we focus on the ‘relationship context’ and improving parenting skills [68], which have been associated with decreased child problem behaviours [28, 53, 55]. However, our intervention is somewhat more intensive, involving six as opposed to three home visits, is more directive and supplemented in each session with illustrated materials and games to help mothers facilitate disclosure and to assist lay counsellors to focus on the topic. This increased intensity, structure and the strength of the counselling relationship may explain the positive disclosure rates. It is not appropriate to compare these results to the TRACK program which reported post intervention disclosure rates of 33% in the intervention group and 7% in the control given that the TRACK program only enrolled mothers with no intention to disclose. However, in light of generally low disclosure rates around the world, achieving close to fifty percent full HIV disclosure in this pilot study, using a locally adapted, similar approach to other international studies shows promise for future work in this area. Further research is required to test whether these findings are replicable and the intervention feasible at a larger scale. A larger demonstration project which aims to enrol 300 families is currently underway.

While the intervention was highly directive and structured, feedback from mothers and counsellors suggests that an important intervention component was the focus on counselling, and training the mother to achieve competence to intervene with her own children as opposed to study counsellors directly intervening with children. This was highly acceptable to both counsellors and mothers and likely accounts for reported improvements in maternal confidence to deal with HIV and health-related matters in the family.

Importantly, this intervention was conducted by lay counsellors, who were able to learn to use the materials and deliver the intervention with this pilot group of mothers. This is particularly important for later scale-up given the increasing time constraints on professional health staff within HIV programmes. Task shifting is already taking place in many areas of health care in sub-Saharan Africa, with lay counsellors or CHWs taking on roles in infant feeding counselling, HIV counselling, ART adherence, and nutrition counselling [69, 70]. Lay counsellors and CHWs have shown ability to conduct psycho-social interventions at the primary health care level, including HIV counselling, testing, and training to initiate HIV treatment [71] and to deliver simplified psychological interventions in poorly resourced settings [72, 73]. Further, the South African National Department of Health plans to include community health care workers as key players in the new primary health care teams, which will focus particularly on improving maternal and child health outcomes [61].

However, little intervention research has focused on the development of techniques, activities and guidelines to train lay counsellors or CHWs to deliver psychologically complex interventions in Southern Africa. To address this, our intervention approach considered several common factors known to impact on therapeutic outcomes [74]. The intervention specifically focused on integrating structured counsellor activities to increase counsellor empathy, positive unconditional regard, and to facilitate encouragement and acknowledgment of achievements by the mother. Such factors are known as ‘therapeutic variables’, and have been shown to account for at least a third of positive outcomes in psychotherapy [75] and social work [76]. It is possible that a similar intervention approach could be applied to a variety of topic areas addressing challenges that HIV-positive parents and families face.

Besides ‘therapeutic variables’ we considered ‘extra therapeutic variables’ known to contribute to positive outcomes in therapeutic relationships, including increasing maternal support in the home and the health care system [75]. The intervention successfully draws on and adapts established therapeutic techniques and processes (from psychotherapy, family and cognitive behavioural therapy) into a structured intervention which facilitates the enactment of therapeu-tic techniques by a lay counsellor. Counsellors found the materials user-friendly, and quality assurance data suggests that the materials assisted in ensuring a high level of intervention fidelity, which has been shown to be a challenge in family HIV interventions [35].

In line with the international literature [7, 33], the qualitative data suggests that mothers had significant distress around disclosure of their own HIV status to their children and felt unable to approach the topic without intervention support. In particular, the qualitative data from participating mothers suggests that these investments in material design and culturally-sensitive illustrations have proved critical to reinforcing maternal self-efficacy. Investing in high quality, age-appropriate and attractively designed materials communicates value and respect, reinforces positive attributes and inspires confidence in HIV disclosure, helping the mother and counsellor to organise their thoughts around the disclosure process. Children’s questions were fairly predictable and the counsellor and mother had worked through these questions and some possible ways to answer them in Visit 4.

In summary this intervention illustrates that complex therapeutic processes can be transformed into practical, structured, step-by-step activities that encourage both experiential learning and reflective practice, and bring about health behaviour change at a familial level. There are several limitations to this study, most importantly it is a small pilot study, and the sample size does not allow for wide generalisation of the findings. Furthermore, given the convenience sampling strategy and lack of a control group it is possible that the results are influenced by selection bias. Further research is required to test these results in a larger sample and a controlled study, to determine the factors that influence the degree of maternal disclosure, and the impact this disclosure has for mothers, children and families over time.

^aAmagugu is a Zulu term meaning “Treasures” to capture the concept that children are our future and are precious.

^bFor information on play for communication see http://www.dlalanathi.org.za and http://www.uthandoproject.org.