Abstract
Objective:
We aimed to: (1) follow parents and guardians through the process of pediatric HIV disclosure to understand how often pre-disclosure worries are realized; and (2) estimate the effects of disclosure on child, caregiver, and family well-being.
Design:
We conducted a 12-month prospective cohort study in Zimbabwe with 123 primary caregivers of children ages 9 to 15 years who were HIV positive but did not know their serostatus at baseline. By the end of the study period 65 caregivers reported that their child learned his or her HIV-positive status.
Main Outcome Measures:
We used three waves of data to compare caregivers’ pre-disclosure worries to post-disclosure reports and to characterize associations between disclosure and well-being of the child (Strengths and Difficulties Questionnaire), caregiver (Patient Health Questionnaire-9), and family (Family Relationship Quality) over time.
Results:
Caregivers’ pre-disclosure worries and fears about how their child would react to disclosure of their HIV status largely went unrealized. Furthermore, we did not find strong evidence of clinically-important increases in problems on average following disclosure.
Conclusion:
Findings support the call to identify supportive intervention strategies that address caregiver fears at the beginning of the disclosure process.
Keywords: HIV, disclosure, mental health, adolescent, Africa, parents
Introduction
Substantial progress in provisioning antiretroviral treatments for children with HIV has decreased childhood mortality rates (Fatti et al., 2011; Hazra et al., 2010). As a result, many children who acquired HIV perinatally—predominantly in Sub Saharan Africa—are surviving into adolescence. While much work remains to fully scale up ART coverage (Burrage et al., 2018; Bygrave et al., 2012), another challenge has emerged as these children have grown: many are partially or completely unaware of their diagnosis or the implications of being HIV-positive. This has particular significance for HIV prevention as HIV-positive adolescents reach sexual debut and begin to consider questions about future goals for relationships.
Estimated pediatric HIV disclosure rates range from 2% to 75% in the published literature, which likely reflects true variation across contexts, in addition to methodological differences in study design that might cloud our understanding (Finnegan et al., 2019). Whatever the true level, we know that without full knowledge of their own HIV status, young people cannot directly benefit from interventions intended to keep them healthy and to prevent them from contributing to ongoing transmission of the virus (Cluver et al., 2015; Langhaug et al., 2020).
Healthcare providers can play a role in creating a supportive environment for pediatric HIV disclosure, and there has been some progress in implementing multi-level interventions (Willis et al., 2018), but parents and guardians typically decide what is shared, when, and how (Mburu et al., 2014; Vreeman et al., 2013). There are numerous factors that influence a caregiver’s decision to withhold information about HIV status from their child, including doubt about their child’s ability to understand, concern about stigma, and the desire to protect their children from possible emotional distress related to knowing that they have a serious and potentially life-threatening illness (Claflin & Barbarin, 1991; Young et al., 2003; Vaz et al., 2008; Vreeman et al., 2013). Additionally, given the nature of perinatal infection, pediatric disclosure often forces caregivers to disclose their own positive status, or that of the child’s biological mother, leading to fear of rejection and difficult discussions about past or future parental morbidity and mortality (Ostrom Delaney, Serovich, & Lim, 2008).
These fears may loom large in the mind of caregivers, but are their worries eventually realized? Is disclosure actually as painful and harmful for their children as many caregivers anticipate? The literature comparing disclosed to non-disclosed children and adolescents suggests indirectly that the answer is no. Studies comparing disclosed adolescents with non-disclosed show very few negative psychological consequences, with some even suggesting psychological and quality of life benefits (Bulali, Kibusi, & Mpondo, 2018). Even in a study with mixed results, showing some persistence of negative emotions, almost all youth expressed being glad that they knew their status (Vaz et al., 2008). The literature further associates disclosure with other positive medical outcomes, such as better medication adherence (Haberer et al., 2011; Cluver et al., 2015). However, most cross-sectional studies assess children, adolescents, and their caregivers only after the disclosure process has taken place and do not prospectively connect parents’ fears pre-disclosure to parent perceptions of outcomes afterwards.
In addition to considering child-level outcomes, caregivers’ own emotional well-being is important to consider during the decision-making process and post-disclosure. Studies including caregiver experiences have been few and small but have reported both positive and negative feelings post-disclosure. Qualitative studies with caregivers in the Democratic Republic of the Congo highlighted the diversity of emotional experiences, from happiness and relief to ongoing sadness and worry (Vaz et al., 2008); some also reported less frustration and conflict related to adherence (Fetzer et al., 2011). Less attention has been paid, however, to measuring the actual mental health symptoms of caregivers deciding about disclosure, whose experiences could parallel the experiences of PLWH deciding whether to self-disclose. In their review, Evangeli and Wroe (2017) describe perceived threats influencing disclosure and present a theoretical model of HIV disclosure anxiety. The model incorporates both standard cognitive elements of health behavior theories alongside anxiety-related theory to propose that negative emotions, especially about anticipated interpersonal risks, can be intense and powerful influences on disclosure (Evangeli & Wroe, 2017). For PLWH, these fears about interpersonal risks can be well-founded, as discrimination is associated with internalized stigma (Fazeli et al., 2017), and social support may not have universal benefits (Rzeszutek & Gruszczyńska, 2018).
Fears related to interpersonal consequences are certainly consistent with caregivers’ fears about their child talking with others after their status is disclosed (Moodley et al., 2006), rendering it likely that they experience negative affect during the decision-making process, sometimes preventing them from disclosing at all. However, for those who decide to disclose their child’s status, the next question is whether they continue to experience negative affect following disclosure, especially if their worries about negative child outcomes do not come true.
Current Study
We conducted a prospective cohort study in Zimbabwe that can help fill this gap. Over the course of 12 months, we identified and followed a prospective cohort of primary caregivers of children between the ages of 9 and 15 years who were HIV positive but did not know it (Finnegan et al., 2019). We surveyed them three times (baseline, 6-months, 12-months), tracking whose children had learned anything about their HIV and whose children remained unaware. In each survey round we asked them about the well-being of their families, their children, and themselves. For those caregivers who had disclosed since the previous round, we inquired about the experience and their child’s reaction. For those caregivers whose children remained undisclosed, we asked about their own worries related to disclosure.
Our aims in this secondary analysis were to compare caregivers’ pre-disclosure worries to their post-disclosure reports, and to characterize associations between disclosure and well-being of the child, caregiver, and family over time. This was an unregistered exploratory analysis. In the process of analyzing data for the original study—whose aim was to describe caregivers’ readiness to disclose to their children—we observed that a substantial proportion of caregivers reported worries about the consequences of disclosure that kept them from taking action. This led us to design this secondary analysis to examine how children, caregivers, and families fared after the children learned their status and to compare them to those who remained undisclosed. This analysis tested the hypotheses, based on the prior literature described, that most caregiver worries would not come to pass and that disclosure would not lead to worse outcomes.
Materials and Methods
Setting and Participants
As described in Finnegan et al. (2019), we randomly selected 400 primary caregivers of children ages 9 to 15 years living with HIV who were enrolled in antiretroviral therapy (ART) or pre-ART (i.e., before initiation of ART treatment) across 21 HIV clinics in Bikita and Zaka districts in Masvingo Province, Zimbabwe and invited them to participate in a study about pediatric disclosure. These parents and guardians were recruited by patient advocates called Community HIV & AIDS Support Agents (CHASAs); inclusion criteria included being a primary caregiver of a child aged 9 to 15 years who was HIV positive. In total, 372/400 parents and guardians (henceforth, caregivers) completed our cross-sectional baseline survey (93.0%), and 123/372 of these caregivers formed the prospective cohort because they reported that their child was unaware of his or her HIV status (33.1%). The current study analyzes data from 111 caregivers in this prospective cohort who participated in all three waves of data collection. As shown in Appendix Table A1, attrition over all three waves was 9.8% and mostly non-systematic on observables. See Finnegan et al. (2019) for additional details about the setting, population, and sample.
Procedures
Surveys were administered in the local language, Shona, by trained Zimbabwean enumerators who used Android tablets running the ODK Collect software (Brunette et al., 2013) to view survey questions and record participants’ responses. Individual surveys were conducted in a private setting at the HIV clinic where the caregiver filled the child’s prescriptions. CHASAs invited them to return to the clinic for follow-up surveys at 6- and 12-months after the baseline wave.
Measures
We administered the same survey instrument in each wave (see Finnegan et al., 2019 for additional details). Our selection of measures was not guided by our research aims in this exploratory secondary analysis. We adapted existing items or created new ones that we thought might have predictive value in an analysis of caregiver readiness to disclose, the aim of the original study. As described in Finnegan et al. (2019), we developed our survey battery through a literature review, focus group discussions with 17 HIV-positive adolescents, 18 parents of HIV-positive children, and health workers at 23 clinics using free-listing and card sorting activities, and cognitive interviewing with a separate group of caregivers.
Caregiver and Child Demographics
We collected demographic characteristics of caregivers, including their age, gender, level of education, marital status, HIV status, a measure of household wealth, and relationship to the child. We also collected demographic information about their child living with HIV (reference child), including gender and age. We based the demographic questionnaire on relevant modules from the 2015 Zimbabwe DHS questionnaire (Zimbabwe National Statistics Agency & ICF International, 2016).
Pediatric HIV Disclosure
Child Knowledge of HIV Status.
We asked caregivers what their child knows about his or her HIV status, including whether the child knows she or he has an illness called HIV. By design, every child of a caregiver in this prospective cohort did not know his or her status at baseline. If the caregiver reported that the child knew their status information at the 6- or 12-month follow-up, we asked the caregiver to recall in which month the child learned this information.
Caregiver Pre-Disclosure Worries About Disclosure.
During the baseline survey and each follow-up survey in which a caregiver reported that the child remained unaware of his or her HIV status, we assessed caregivers’ worries about disclosing their child’s HIV status to the child. We assessed four specific worries with statements that began, “I worry that telling my child about their HIV status will”: (a) “lead [him/her] to reject me” (anger or rejection); (b) “lead [him/her] to ask more tough questions” (ask tough questions); (c) “will make [him/her] too sad” (too sad); and (d) “make [him/her] stop taking medication” (stop medication). Caregivers responded to each statement on a 4-point scale: not at all worried, not very worried, somewhat worried, or very worried. We classified a caregiver as “worried” (1/0) if they responded that they were “somewhat” or “very” worried.
Child Post-Disclosure Reactions to Disclosure.
For caregivers who reported in the 6- or 12-month follow-up survey that their child had learned about his or her HIV-positive status since the last survey, we asked them to report on the disclosure experience and how their child had reacted. This included parallel questions to the statements used to assess caregivers’ pre-disclosure worries: (a) “Did your child get angry?”; (b) “Did your child ask questions?”; (c) “Did your child get sad?”; and (d) “Has [their adherence to medication] gotten worse, better, or stayed the same?” Caregivers responded to questions (a-c) on a 4-point scale: not at all, not very much, somewhat, a lot. We classified each potential worry as realized (1/0) if the caregiver said “somewhat” or “a lot.” For question (d) about medication adherence, we classified the potential worry about lower adherence as actualized if the caregiver reported that the child’s adherence had gotten worse.
Child, Caregiver, and Family Well-Being
Child Social-Emotional Problems.
Caregivers completed the 25-item Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997) to assess a child’s emotional and behavioral well-being across five domains: conduct, emotional problems, peer problems, hyperactivity, and prosocial behavior. Caregivers rated their child on a 3-point scale ranging from “not true” (0) to “certainly true” (2) on items such as, “Has many worries or often seems worried” and “Picked on or bullied by other children.” We averaged responses to the 20 items from the conduct, emotional problems, peer problems, and hyperactivity subscales to create the Total Difficulties score. The average scale score could range from 0 to 2 with higher scores representing more problems. We observed Cronbach’s alpha values of 0.75, 0.79, 0.81 in Rounds 1 to 3.
Caregiver Depression.
We used the 9-item Patient Health Questionnaire to measure caregiver depression symptom severity (PHQ-9; Kroenke et al., 2001). Caregivers were asked to report how often they experienced any of the problems queried in the PHQ-9 (e.g., feeling tired, down, poor appetite) over the last 2 weeks. Responses could range from “not at all” (0) to “nearly every day” (3). We constructed a scale score by averaging answers to all items (possible range 0 to 3); higher scores indicate greater symptom severity. We observed Cronbach’s alpha values of 0.83, 0.83, 0.88 in Rounds 1 to 3.
Family Relationship Quality.
We developed eight items to assess family relationship quality from the perspective of caregivers. Caregivers used a 4-point scale ranging from “never” (3) to “often” (0) to answer questions such as, “How often do people in your family listen to each other at home?” and “How often do you feel that there is joy (or love or peace) in your family?” We constructed a scale score by averaging answers to all items (range 0 to 3); higher scores indicate more problems in a family’s relationship. We observed Cronbach’s alpha values of 0.71, 0.75, 0.65 in Rounds 1 to 3.
Statistical Analyses
We report the results of two primary analyses conducted with R version 4.0.2 (R Core Team, 2016).
Are Caregiver Worries About Disclosure Realized?
The first analysis involved the subset of 65 caregivers who reported that their child learned his or her HIV-positive status at some point during the study period. We calculated the proportion of caregivers who expressed a particular worry pre-disclosure and compared it to the proportion of caregivers who reported this outcome post-disclosure. For caregivers who reported that the disclosure took place between the baseline and 6-month follow-up survey, we used the baseline data for the pre-disclosure assessment and the 6-month survey for the post-disclosure assessment. For caregivers who reported that the disclosure took place between the 6- and 12-month follow-up surveys, we used the 6-month survey for the pre-disclosure assessment and the 12-month survey for the post-disclosure assessment.
What is the Impact of Disclosing on Child, Caregiver, and Family Well-Being Outcomes?
In the second analysis, we used data from all 111 caregivers to estimate the impact of disclosing on child well-being, caregiver well-being, and family relationship quality over time through a quasi-experimental difference-in-differences analysis. We fit separate Bayesian multilevel regressions of the well-being outcomes on an indicator of disclosure status at the 12-month follow-up survey (0 non-disclosed; 1 disclosed), an indicator of whether the outcome was observed before (0) or after disclosure (1), dummy variables indicating if the observation came from the 6-month or 12-month follow-up survey, and several baseline covariates (caregiver education status, marital status, and HIV serostatus). We developed the causal directed acyclic graph presented in Figure A1 to identify this minimally sufficient covariate adjustment set for estimating the effect of disclosure on well-being outcomes. The models also included a random effect for person to account for the repeated measurement (baseline, 6-month follow-up, 12-month follow-up). We fit these models in a Bayesian setup with default priors using the {brms} package, using Markov-Chain Monte Carlo estimation to sample from the posterior distributions (version 2.13.5; Bürkner, 2018). In the interest of transparency and completeness, we also present the results of a multiverse analysis that implements different analytical decisions.
Ethical Review
Institutional review boards at OMITTED joined the Joint Parirenyatwa Hospital and College of Health Sciences Research Committee and the Medical Research Council of Zimbabwe in reviewing and approving the study protocol. All study participants provided written informed consent.
Results
Sample Characteristics
Of the 123 eligible caregivers in the prospective cohort, 111 completed all three waves of data collection (90%). Table A1 in the Appendix compares the found and unfound subgroups. Baseline characteristics of the complete data sample are reported in Table 1. The majority of caregivers were female with an average age of 49.4 years (range 18 to 86 years). Most caregivers were married with households categorized in the two poorest wealth quintiles relative to national data. 3 out of 5 caregivers surveyed were not biological parents, with grandparents making up 35% of the sample. The sample of reference children was balanced on gender, and the average child was 11.7 years old (range 9 to 15 years). Caregivers of children who learned their HIV status during the study follow-up period were slightly younger on average and more educated, but otherwise very similar to the caregivers of children who did not learn their HIV status.
Table 1.
Baseline characteristics of sample by final disclosure status at 12-months
| Child Learned HIV Status |
|||
|---|---|---|---|
| Characteristic | Overall (N=111) | No (N=46) | Yes (N=65) |
| Caregivers | |||
| Mean age (SD) | 49.4 (14.7) | 52.1 (16.0) | 47.5 (13.5) |
| Female | 100 (90%) | 40 (87%) | 60 (92%) |
| Biological caregiver | 44 (40%) | 19 (41%) | 25 (38%) |
| HIV positive | 66 (59%) | 27 (59%) | 39 (60%) |
| Completed primary | 64 (58%) | 21 (46%) | 43 (66%) |
| Married | 68 (61%) | 28 (61%) | 40 (62%) |
| Poorest 2 wealth quintiles | 73 (66%) | 29 (63%) | 44 (68%) |
| Mean PHQ-9 score (SD) | 0.4 (0.5) | 0.4 (0.4) | 0.4 (0.5) |
| Mean Family Relationship Quality score (SD) | 0.6 (0.4) | 0.6 (0.4) | 0.6 (0.4) |
| Reference Children | |||
| Mean age (SD) | 11.7 (1.8) | 11.7 (1.8) | 11.8 (1.7) |
| Female | 57 (51%) | 24 (52%) | 33 (51%) |
| SDQ Total Difficulties score (SD) | 0.3 (0.3) | 0.3 (0.2) | 0.3 (0.3) |
Pediatric HIV Disclosure
According to the prospective study design, all 111 caregivers in the prospective cohort reported in the baseline survey that their child did not know that they were HIV-positive at the start of the study. Sixty-five of these caregivers (59%) reported that their child learned their status over the course of the 12-month follow-up period. Among this group of disclosed children, the mean time to disclosure was 5.6 months (SD=3.6). Table 1 presents baseline characteristics of caregivers and reference children by ultimate disclosure status.
Are Caregiver Worries About Disclosure Realized?
Figure 1 compares expected outcomes to actual outcomes measured before and after disclosure among the subset of 65 caregivers who reported that their child learned his or her HIV-positive status during the study period. This figure shows that caregivers overestimated their child’s negative reactions to learning their HIV-positive status. For instance, before disclosing their child’s HIV-status to the child, 54% of caregivers worried that disclosure would make the child too sad. After disclosure, however, only 7% of caregivers reported that this fear was realized. Similarly, caregiver worries about their children reacting in anger (28% expected, 7% realized), asking hard questions about how they were infected (49% expected, 25% realized), and stopping taking their medication (22% expected, 2% realized) were shown to have been overestimated. See Table A2 in the Appendix for item response proportions on the original categorical metric.
Figure 1. The disclosure expectation-reality gap.
Percent of caregivers who worried about each outcome prior to disclosure compared to actual percentage of caregivers who reported each outcome following disclosure
Descriptive Summary of Well-Being Outcomes
The plots in Figure 2 present raw data points, distributions, and boxplot summaries of child, caregiver, and family well-being outcomes in the original metrics of the scales. While all children were unaware of their status at baseline, some children learned their status prior to the last follow-up survey. This is indicated by color, with teal representing the group of children whose caregivers eventually disclosed.
Figure 2. Child, caregiver, and family well-being by eventual disclosure status.
These plots present raw data points, distributions, and summaries in the original metrics of the scales (constructed as average scores, not total scores). All children were unaware of their status at baseline. Some children (Teal) learned their status prior to the follow-up survey. On average, children and caregivers at baseline had low scores on all outcomes regardless of event status.
On average, caregivers reported few problems at baseline. The child SDQ baseline mean was 0.33 (SD=0.26) on a scale of 0 to 2, where 0.85 represents a commonly used clinical cutoff (equivalent to a cutoff of 17 on the 0 to 40 scale). The mean caregiver PHQ-9 score was 0.43 (SD=0.46), which is well below the clinical cutoff of 1.11 on the 0 to 3 scale (equivalent to a cutoff of 10 on the 0 to 27 scale). Family relationship problems were low on average at baseline, with a mean of 0.61 (SD=0.37) on a scale of 0 to 3. Table A3 reports summary statistics for all three rounds.
What is the Impact of Disclosing on Child, Caregiver, and Family Well-Being Outcomes?
We answered this question with a combined analysis that examined within-subjects differences (i.e., comparing pre-post disclosure) and between-groups differences (comparing disclosed to non-disclosed). Figure 3 displays the results of three separate regressions that estimate the effect of disclosure on child, caregiver, and family well-being (full model with numerical summaries are presented in Table A4).
Figure 3. Is disclosure associated with worse outcomes for children, caregivers, and families? Estimates from Bayesian multilevel regressions.
Markov chain Monte Carlo draws from the posterior distribution of model parameters. Fill shows 89% credible intervals, solid vertical lines represent medians, and dashed vertical lines show ROPE bounds.
The solid vertical line in the center of the distribution represents the estimate of the median difference between disclosed and non-disclosed children, adjusted for survey round and several covariates. The yellow filled area surrounding this line spans the 89% credible interval. The estimate and the interval are similar to Frequentist regression coefficients and confidence intervals, but the credible interval has a more intuitive interpretation—given the observed data, the effect has a 89% probability of falling within this range.
The dashed vertical lines represent the lower and upper bounds of the regions of practical equivalence, or ROPE (Kruschke, 2014). To the extent that the yellow credible interval falls within these dotted lines, we conclude that the effect of disclosing on the outcome is close enough to zero to be practically (clinically) insignificant.
Overall, there is not strong evidence to suggest that children and families fared worse on average following disclosure. In Model 1, where the dependent variable is children’s social-emotional problems (SDQ Total Difficulties score), the median of the posterior of the coefficient is centered on zero (0.002), and 98% of the posterior falls inside the ROPE (96%, no practical effect) or below the lower ROPE bound (2%, reduced problems). Thus, it appears very unlikely (2% probability) from caregivers’ reports that child problems increased following disclosure (and any increase would be from a very low baseline). In Model 2, where the dependent variable is caregiver symptoms of depression (PHQ-9 score), the median estimate of the effect is 0.07, and 97% of the posterior falls inside the ROPE (no practical effect). Therefore, it is also very unlikely (3% probability) that caregiver distress increased following disclosure (and here too, any increase would be from a very low baseline). Finally, in Model 3, where the dependent variable is family relationship quality, the median of the posterior of the coefficient is centered within the ROPE at −0.03, and 100% of the posterior falls within the ROPE (no practical effect). There is no evidence that family relationship quality worsened on average following disclosure.
Figure A2 in the Appendix presents the same results along with a multiverse analysis (Steegen et al., 2016) that estimates the effects with different subsets of data: disclosure between the baseline and the 6-month follow-up (Rounds 1 and 2), disclosure between the 6-month follow-up and the 12-month follow-up (Rounds 2 and 3), disclosure between the baseline and the 12-month follow-up (Rounds 1 and 3), and disclosure over all rounds (Rounds 1, 2, and 3). The results presented above are based on all three rounds of data and have the most precise estimates. The multiverse analysis largely supports the main analysis. It is only in the Round 2-3 window that we see any evidence that child and caregivers experience slightly more problems on average after disclosure.
Discussion
In this prospective cohort study, we find little support for one of the main reasons that caregivers hesitate to disclose a child’s positive serostatus to the child: worries that the child will not be able to cope with the news. As a follow-up to a population-level cross-sectional survey in rural Zimbabwe (Finnegan et al., 2019), we identified caregivers of children who did not know their status and surveyed them two more times over the next 12 months. Our results suggest that caregivers’ worries and fears about how their child would react largely went unrealized in this sample. Furthermore, we did not find strong evidence that child, caregiver, or family well-being suffered on average after disclosure.
Our study makes two main contributions to the literature. First, using data collected prospectively and at multiple time points, we find support for previous (mostly) cross-sectional work suggesting that disclosure seldom leads to negative outcomes (Domek, 2010). In US-based research, disclosed children and adolescents have not shown poorer mental health or quality of life (Mellins et al., 2002; Butler et al., 2009), and the limited evidence from Sub Saharan Africa largely echoes this. In a cross-sectional sample of youth in Zambia, non-disclosed children and adolescents exhibited slightly poorer emotional well-being (Menon et al., 2007), and a hospital-based study in Tanzania documented an association between disclosure and better quality of life (Bulali, Kibusi, & Mpondo, 2018). Multiple studies in Sub Saharan Africa also have documented treatment-related benefits of disclosure and communication about HIV, including improved medication adherence and retention in care (Bulali, Kibusi, & Mpondo, 2018; Kajubi et al., 2014; Haberer et al., 2011; Cluver et al., 2015). Second, because disclosure may also have impacts beyond the level of the child, we expanded the scope of inquiry. We asked caregivers about their own mental health and their family’s relationship quality in each survey round (i.e., before and after disclosure), also finding no evidence of negative consequences on these levels. This contributes quantitative findings to existing qualitative findings on caregivers’ emotional experiences. While emotional reactions seem to vary (Fetzer et al., 2011; Vaz et al., 2008), our results suggest that, overall, caregivers are unlikely to experience depression symptoms.
Our study also makes a contribution to practice. Developmentally-appropriate pediatric HIV disclosure is strongly encouraged across contexts, with guidance promoting a developmentally-informed process (Lesch et al., 2007; Gerson et al., 2001; Aderomilehin et al., 2016), but disclosure rates remain low in many places (Finnegan et al., 2019). Parents and guardians continue to worry about the negative outcomes, and this is a barrier to timely and complete disclosure. Findings like ours—directly from caregivers’ prospective experiences—might reassure caregivers, especially in Sub-Saharan Africa, who are afraid to begin the process.
Limitations and Future Directions
Results and implications of this study should be considered in the context of several important limitations. First, the effect estimates reported herein come from quasi-experimental analysis of observational data. As such, the analysis is subject to selection bias as the decision to disclose was not random. To mitigate this risk, we recruited a prospective cohort of non-disclosed children and collected several rounds of data that we could use in a difference-in-differences analysis. Under the assumption that unobserved differences between disclosed and non-disclosed children remain fixed over time, it is possible to make causal inferences from this type of observational data. While not a perfect test of the research question on the impact of disclosure, this study builds on a literature that consists of largely cross-sectional studies that have even greater risk of bias.
Second, we relied on caregiver self-report for both the assessment of disclosure status as well as child well-being. As a prospective study, we did not want to run the risk of accidental disclosure by having our study team interview the children directly. While we believe this was the best decision, the results depend on caregivers being reliable reporters. It is not clear to us whether potential misclassification of disclosure status would be more likely to be systematic error or random error. Third, in the analysis of whether caregiver worries are realized post-disclosure, we had to rely on single items that assessed each worry as this was an unplanned analysis when we developed the study. Therefore, we limited our analysis to comparisons of proportions. Fourth, we only assessed depression for caregiver mental health, and only after disclosure; this limits our ability to examine disclosure-related anxiety documented in the literature (Evangeli & Wroe, 2017). Fifth, while suitable for the main analysis, the prospective sample was too small to investigate differential impacts by child age.
Lastly, it is not clear to what extent the results can generalize beyond rural Zimbabwe. We constructed the sample from an initial population-based cross-sectional survey that was representative of children on ART in two rural districts with high HIV prevalence rates. This sampling approach distinguishes the work from much of the literature on pediatric HIV disclosure that has typically relied on convenience samples (Finnegan et al., 2019), but it does not make a clear case for wide generalizability beyond the characteristics of the target population.
Future studies—especially mixed methods studies and intervention trials—should aim to understand the ways in which the details of the disclosure process can minimize caregiver fear at the beginning of the process and then promote the most positive psychosocial wellbeing outcomes for children and adolescents, their caregivers, and their families. This focus on caregiver fear could add to intervention strategies already showing promise (Beima-Sofie et al., 2017; Willis et al., 2018) and push our knowledge further on core components. What are the most effective strategies for helping caregivers decide to disclose? What is the ideal pace and sequence of disclosure; how can we measure and consider readiness on the parts of the caregiver and child? Furthermore, are there coping or family intervention strategies (beyond HIV education) that can bolster positive disclosure outcomes? How could external interventions provide support to parents and caregivers engaging with their children in the disclosure process? How can the unique contributions of the caregiver and provider be optimized? And importantly, how do these associations differ across cultures and contexts?
Conclusions
This analysis provides an important data point for caregivers weighing the decision to disclose, and it can inform practitioners tasked with creating and implementing interventions to support caregivers and children who are engaged in the process of disclosure.
Acknowledgments
Study funded by NICHD (5R21HD076695).
Appendix
Table A1.
Attrition from the Analysis Sample by Baseline Characteristics
| Characteristic | Analysis Sample (N=111) | Lost to Follow-up (N=12) |
|---|---|---|
| Caregivers | ||
| Mean age (SD) | 49.4 (14.7) | 46.3 (13.2) |
| Female | 100 (90%) | 10 (83%) |
| Biological caregiver | 44 (40%) | 4 (33%) |
| HIV positive | 66 (59%) | 3 (25%) |
| Completed primary | 64 (58%) | 8 (67%) |
| Married | 68 (61%) | 8 (67%) |
| Poorest 2 wealth quintiles | 73 (66%) | 8 (67%) |
| Mean PHQ-9 score (SD) | 0.4 (0.5) | 0.6 (0.4) |
| Mean Family Relationship Quality score (SD) | 0.6 (0.4) | 0.7 (0.6) |
| Reference Children | ||
| Mean age (SD) | 11.7 (1.8) | 11.4 (2.2) |
| Female | 57 (51%) | 7 (58%) |
| SDQ Total Difficulties score (SD) | 0.3 (0.3) | 0.3 (0.2) |
Table A2.
Caregiver worries about child prior to disclosure compared to post-disclosure reports of child’s reaction
| Caregiver Worries / Reality | Expected (Pre-disclosure) | Actual (Post-disclosure) |
|---|---|---|
| Too sad | ||
| Not at all | 19 (29.2%) | 47 (82.5%) |
| Not very | 11 (16.9%) | 6 (10.5%) |
| Somewhat | 15 (23.1%) | 2 (3.5%) |
| Very / A lot | 20 (30.8%) | 2 (3.5%) |
|
| ||
| Ask tough questions | ||
| Not at all | 21 (32.3%) | 30 (49.2%) |
| Not very | 12 (18.5%) | 16 (26.2%) |
| Somewhat | 12 (18.5%) | 13 (21.3%) |
| Very / A lot | 20 (30.8%) | 2 (3.3%) |
|
| ||
| Anger or rejection | ||
| Not at all | 24 (36.9%) | 50 (86.2%) |
| Not very | 23 (35.4%) | 4 (6.9%) |
| Somewhat | 12 (18.5%) | 2 (3.4%) |
| Very / A lot | 6 (9.2%) | 2 (3.4%) |
|
| ||
| Stop medication | ||
| Not at all / No | 31 (47.7%) | 62 (98.4%) |
| Not very | 20 (30.8%) | - |
| Somewhat | 5 (7.7%) | - |
| Very / Yes | 9 (13.8%) | 1 (1.6%) |
Note. Among caregivers of children who disclosed to their children during the study period.
Table A3.
Descriptive summary of well-being outcomes, N=111
| Overall | Non-Disclosed | Disclosed | ||||
|---|---|---|---|---|---|---|
|
|
||||||
| Outcomes | Mean | SD | Mean | SD | Mean | SD |
| SDQ, Round 1 | 0.33 | 0.26 | 0.31 | 0.21 | 0.34 | 0.29 |
| SDQ, Round 2 | 0.36 | 0.29 | 0.37 | 0.29 | 0.33 | 0.27 |
| SDQ, Round 3 | 0.29 | 0.28 | 0.29 | 0.29 | 0.28 | 0.28 |
| PHQ-9, Round 1 | 0.43 | 0.46 | 0.41 | 0.45 | 0.44 | 0.47 |
| PHQ-9, Round 2 | 0.43 | 0.41 | 0.40 | 0.40 | 0.49 | 0.43 |
| PHQ-9, Round 3 | 0.35 | 0.47 | 0.32 | 0.47 | 0.37 | 0.48 |
| Family Relationship Quality, Round 1 | 0.61 | 0.37 | 0.62 | 0.35 | 0.61 | 0.38 |
| Family Relationship Quality, Round 2 | 0.60 | 0.45 | 0.62 | 0.43 | 0.58 | 0.48 |
| Family Relationship Quality, Round 3 | 0.56 | 0.39 | 0.48 | 0.35 | 0.61 | 0.40 |
Table A4.
Model summaries
| Child SDQ | Caregiver PHQ-9 | Family Relationships | ||||
|---|---|---|---|---|---|---|
| Predictors | Estimates | CI (95%) | Estimates | CI (95%) | Estimates | CI (95%) |
| Intercept | 0.41 | 0.31 – 0.50 | 0.53 | 0.37 – 0.69 | 0.49 | 0.34 – 0.64 |
| Disclosed | −0.03 | −0.11 – 0.06 | 0.03 | −0.11 – 0.19 | 0.11 | −0.03 – 0.23 |
| Round 2 | 0.03 | −0.05 – 0.10 | −0.02 | −0.14 – 0.09 | 0 | −0.10 – 0.11 |
| Round 3 | −0.04 | −0.13 – 0.04 | −0.12 | −0.26 – 0.02 | −0.04 | −0.16 – 0.09 |
| Post (DiD) | 0 | −0.09 – 0.10 | 0.07 | −0.09 – 0.23 | −0.03 | −0.17 – 0.10 |
| Completed Primary | 0 | −0.08 – 0.07 | −0.13 | −0.25 – −0.00 | 0 | −0.11 – 0.12 |
| Married | −0.07 | −0.14 – 0.00 | −0.08 | −0.20 – 0.05 | 0.01 | −0.11 – 0.12 |
| HIV Positive | −0.03 | −0.10 – 0.04 | −0.01 | −0.13 – 0.11 | 0.1 | −0.01 – 0.20 |
| Random Effects | ||||||
| σ2 | 0.06 | 0.16 | 0.13 | |||
| τ00 | 0.01 id | 0.05 id | 0.04 id | |||
| ICC | 0.17 | 0.23 | 0.22 | |||
| N | 111 id | 111 id | 111 id | |||
| Observations | ||||||
| Observations | 333 | 333 | 333 | |||
| Marginal R2 / Conditional R2 | 0.049 / 0.208 | 0.061 / 0.265 | 0.050 / 0.254 | |||
Figure A1.
Causal directed acyclic graph to inform the statistical adjustment for the models estimating the effect of disclosure (exposure) on child well-being (outcome). Green represents ancestors of exposure. Blue represents ancestors of outcomes. Red represents ancestors of exposure and outcome (and minimal adjustment set).
Figure A2. Is disclosure associated with worse outcomes for children, caregivers, and families? Estimates from Bayesian multilevel regressions.
Markov chain Monte Carlo draws from the posterior distribution of model parameters. Fill shows 89% credible intervals, and dashed vertical lines show ROPE bounds.
Footnotes
Declaration of Interest
The authors have no conflicts of interest to disclose.
Data Availability
Data and code available at OMITTED
References
- Aderomilehin O, Hanciles-Amu A, & Ozoya OO (2016). Perspectives and practice of HIV disclosure to children and adolescents by health-care providers and caregivers in Sub-Saharan Africa: a systematic review. Frontiers in Public Health, 4, 166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Beima-Sofie KM, Brandt L, Hamunime N, Shepard M, Uusiku J, John-Stewart GC, & O’Malley G (2017). Pediatric HIV disclosure intervention improves knowledge and clinical outcomes in HIV-infected children in Namibia. Journal of Acquired Immune Deficiency Syndromes, 75(1), 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Brunette W, Sundt M, Dell N, Chaudhri R, Breit N, & Borriello G (2013). Open Data Kit 2.0: Expanding and refining information services for developing regions. [Google Scholar]
- Bulali RE, Kibusi SM, & Mpondo BC (2018). Factors associated with HIV status disclosure and its effect on treatment adherence and quality of life among children 6–17 years on antiretroviral therapy in southern highlands zone, Tanzania: unmatched case control study. International Journal of Pediatrics, 8058291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bürkner P-C (2018). Advanced Bayesian multilevel modeling with the R package brms. The R Journal, 10(1), 395–411. 10.32614/RJ-2018-017 [DOI] [Google Scholar]
- Burrage A, Patel M, Mirkovic K, Dziuban E, Teferi W, Broyles L, & Rivadeneira E (2018). Trends in antiretroviral therapy eligibility and coverage among children aged< 15 years with HIV infection—20 PEPFAR-supported sub-Saharan African countries, 2012–2016. Morbidity and Mortality Weekly Report, 67(19), 552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Butler AM, Williams PL, Howland LC, Storm D, Hutton N, & Seage GR (2009). Impact of disclosure of HIV infection on health-related quality of life among children and adolescents with HIV infection. Pediatrics, 123(3), 935–943. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bygrave H, Mtangirwa J, Ncube K, Ford N, Kranzer K, & Munyaradzi D (2012). Antiretroviral therapy outcomes among adolescents and youth in rural Zimbabwe. PLOS One, 7(12), e52856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Claflin CJ, & Barbarin OA (1991). Does “telling” less protect more? Relationships among age, information disclosure, and what children with cancer see and feel. Journal of Pediatric Psychology, 16(2), 169–191. [DOI] [PubMed] [Google Scholar]
- Cluver LD, Hodes RJ, Toska E, Kidia KK, Orkin FM, Sherr L, & Meinck F (2015). ‘HIV is like a tsotsi. ARVs are your guns’: associations between HIV-disclosure and adherence to antiretroviral treatment among adolescents in South Africa. AIDS, 29, S57–S65. [DOI] [PubMed] [Google Scholar]
- Domek GJ (2010). Debunking common barriers to pediatric HIV disclosure. Journal of Tropical Pediatrics, 56(6), 440–442. [DOI] [PubMed] [Google Scholar]
- Evangeli M, & Wroe AL (2017). HIV disclosure anxiety: A systematic review and theoretical synthesis. AIDS and Behavior, 21(1), 1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fatti G, Bock P, Eley B, Mothibi E, & Grimwood A (2011). Temporal trends in baseline characteristics and treatment outcomes of children starting antiretroviral treatment: An analysis in four provinces in south africa, 2004–2009. JAIDS Journal of Acquired Immune Deficiency Syndromes, 58(3), e60–e67. [DOI] [PubMed] [Google Scholar]
- Fazeli PL, Turan JM, Budhwani H, Smith W, Raper JL, Mugavero MJ, & Turan B (2017). Moment-to-moment within-person associations between acts of discrimination and internalized stigma in people living with HIV: An experience sampling study. Stigma and Health, 2(3), 216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fetzer BC, Mupenda B, Lusiama J, Kitetele F, Golin C, & Behets F (2011). Barriers to and facilitators of adherence to pediatric antiretroviral therapy in a sub-Saharan setting: insights from a qualitative study. AIDS Patient Care and STDs, 25(10), 611–621. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Finnegan A, Langhaug L, Schenk K, Puffer ES, Rusakaniko S, Choi Y, Mahaso S, & Green EP (2019). The prevalence and process of pediatric HIV disclosure: A population-based prospective cohort study in zimbabwe. PloS One, 14(5), e0215659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gerson AC, Joyner M, Fosarelli P, Butz A, Wissow L, Lee S, … & Hutton N. (2001). Disclosure of HIV diagnosis to children: when, where, why, and how. Journal of Pediatric Health Care, 15(4), 161–167. [DOI] [PubMed] [Google Scholar]
- Goodman R (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38(5), 581–586. [DOI] [PubMed] [Google Scholar]
- Haberer JE, Cook A, Walker AS, Ngambi M, Ferrier A, Mulenga V, … & Bangsberg DR (2011). Excellent adherence to antiretrovirals in HIV+ Zambian children is compromised by disrupted routine, HIV nondisclosure, and paradoxical income effects. PLoS One, 6(4), e18505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hazra R, Siberry GK, & Mofenson LM (2010). Growing up with HIV: Children, adolescents, and young adults with perinatally acquired HIV infection. Annual Review of Medicine, 61, 169–185. [DOI] [PubMed] [Google Scholar]
- Kajubi P, Whyte S, Muhumuza S, Kyaddondo D, & Katahoire AR (2014). Communication between HIV-infected children and their caregivers about HIV medicines: a cross-sectional study in Jinja district, Uganda. Journal of the International AIDS Society, 17(1), 19012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kroenke K, Spitzer RL, & Williams JB (2001). The PHQ-9. Journal of General Internal Medicine, 16(9), 606–613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kruschke J (2014). Doing Bayesian Data Analysis: A Tutorial with T, Jags, and Stan. Academic Press. [Google Scholar]
- Langhaug L, Finnegan A, Schenk K, Puffer ES, Rusakaniko S, & Green EP (2020). Caregiver self-efficacy to talk about sex predicts conversations about HIV transmission risk with perinatally infected young people in Zimbabwe. AIDS Care, 32(12), 1524–1528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lesch AIDS, Swartz L, Kagee A, Moodley K, Kafaar Z, Myer L, & Cotton M (2007). Paediatric HIV/AIDS disclosure: towards a developmental and process-oriented approach. AIDS Care, 19(6), 811–816. [DOI] [PubMed] [Google Scholar]
- Mburu G, Hodgson I, Kalibala S, Haamujompa C, Cataldo F, Lowenthal ED, & Ross D (2014). Adolescent HIV disclosure in Zambia: barriers, facilitators and outcomes. Journal of the International AIDS Society, 17(1), 18866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mellins CA, Brackis-Cott E, Dolezal C, Richards A, Nicholas SW, & Abrams EJ (2002). Patterns of HIV status disclosure to perinatally HIV-infected children and subsequent mental health outcomes. Clinical Child Psychology and Psychiatry, 7(1), 101–114. [Google Scholar]
- Menon A, Glazebrook C, Campain N, & Ngoma M (2007). Mental health and disclosure of HIV status in Zambian adolescents with HIV infection: implications for peer-support programs. JAIDS Journal of Acquired Immune Deficiency Syndromes, 46(3), 349–354. [DOI] [PubMed] [Google Scholar]
- Moodley K, Myer L, Michaels D, & Cotton M (2006). Paediatric HIV disclosure in South Africa-caregivers’ perspectives on discussing HIV with infected children. South African Medical Journal, 96(3), 201–204. [PubMed] [Google Scholar]
- Ostrom Delaney R, Serovich JM, & Lim JY (2008). Reasons for and against maternal HIV disclosure to children and perceived child reaction. AIDS Care, 20(7), 876–880. [DOI] [PMC free article] [PubMed] [Google Scholar]
- R Core Team. (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/ [Google Scholar]
- Rzeszutek M, & Gruszczyńska E (2018). Paradoxical effect of social support among people living with HIV: A diary study investigating the buffering role of relationship status. Journal of Psychosomatic Research, 109, 25–31. [DOI] [PubMed] [Google Scholar]
- Steegen S, Tuerlinckx F, Gelman A, & Vanpaemel W (2016). Increasing transparency through a multiverse analysis. Perspectives on Psychological Science, 11(5), 702–712. [DOI] [PubMed] [Google Scholar]
- Vaz L, Corneli A, Dulyx J, Rennie S, Omba S, Kitetele F, … & Behets F. (2008). The process of HIV status disclosure to HIV-positive youth in Kinshasa, Democratic Republic of the Congo. AIDS Care, 20(7), 842–852. [DOI] [PubMed] [Google Scholar]
- Vreeman RC, Gramelspacher AM, Gisore PO, Scanlon ML, & Nyandiko WM (2013). Disclosure of HIV status to children in resource-limited settings: a systematic review. Journal of the International AIDS Society, 16(1), 18466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Willis N, Napei T, Armstrong A, Jackson H, Apollo T, Mushavi A, … & Cowan FM (2018). Zvandiri—bringing a differentiated service delivery program to scale for children, adolescents, and young people in Zimbabwe. JAIDS Journal of Acquired Immune Deficiency Syndromes, 78, S115–S123. [DOI] [PubMed] [Google Scholar]
- Young B, Dixon-Woods M, Windridge KC, & Heney D (2003). Managing communication with young people who have a potentially life threatening chronic illness: Qualitative study of patients and parents. BMJ, 326(7384), 305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zimbabwe National Statistics Agency, & ICF International. (2016). Zimbabwe Demographic and Health Survey 2015: Final Report. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data and code available at OMITTED