A Mixed-Methods Longitudinal Investigation of Mothers’ Disclosure of HIV to Their Children

Abstract

Mothers living with HIV (MLH) must navigate disclosing their serostatus to their children, but the longitudinal impact on families remains unknown. This study examined HIV disclosure, parenting, parenting stress, and child adjustment among 174 MLH-child dyads (aged 6–14; 35% Latinx; 57% Black/African American). Quantitative data were collected over four waves spanning 15 months. Qualitative data were collected with 14 families in which disclosure had occurred. Latent change score modeling revealed that disclosure led to improvements in parenting stress, communication, and relationship quality. Disclosure did not predict child adjustment. Qualitative themes contextualized these findings, revealing stability and improvements in family functioning. MLH should be supported in disclosing their serostatus to their children to minimize parenting stress and bolster parenting skills.

An estimated 36.9 million people are currently living with HIV globally, and women are recognized as an at-risk group for HIV infection (UNAIDS, 2018). Although the incidence of HIV is on a downtrend in the United States, African American and Latina women remain 18 times and four times, respectively, more likely to be HIV positive than White women (CDC, 2018). Many women living with HIV are primary caregivers of one or more children. With increased access to antiretroviral therapy (ART), the face of the HIV epidemic has shifted dramatically, and the life expectancy for people living with HIV is approaching that of noninfected adults (Samji et al., 2013; Siddiqi, Hall, Hu, & Song, 2016). Mothers living with HIV (MLH) are now living long enough to raise their children, and yet the illness can still have a profound impact on families. For example, many MLH experience stress related to the difficult decision of whether to disclose their HIV status to their children. Living with HIV also poses additional challenges, such as elevated parenting stress for mothers, disruptions in parenting quality, and heightened risk of child emotional and behavioral difficulties (Lachman, Cluver, Boyes, Kuo, & Casale, 2014; Rochat, Netsi, Netsi, Redinger, & Stein, 2017). Although researchers have demonstrated cross-sectional associations between HIV disclosure and family processes (Armistead et al., 2018; Murphy, Marelich, Armistead, Herbeck, & Payne, 2010; Qiao et al., 2015), few studies have examined these associations longitudinally or using mixed methods.

MLH are faced with the difficult decision of whether, and when, to disclose their HIV serostatus to their children. For many MLH, this is a primary challenge of living with HIV (Murphy, 2008; Qiao, Li, & Stanton, 2013b; Rochat, Netsi, et al., 2017). The disclosure process model (Chaudoir, Fisher, & Simoni, 2011) provides a framework for understanding the antecedents, mechanisms, and correlates of HIV disclosure to other individuals. The model asserts three interrelated aspects of disclosure: decision making, disclosure event, and outcomes. Decision making drives the disclosure event, which in turn drives individual, relational, and social outcomes. Outcomes may inform future disclosure decision making to other individuals through a feedback loop. Although initially developed to describe disclosure in any relationship, the disclosure process model has also been applied to MLH and their children in studies examining motivations and consequences of disclosure (or nondisclosure) for families (Gachanja & Burkholder, 2016; Qiao, Li, & Stanton, 2013a).

Some MLH are reluctant to disclose their HIV status due to concerns that the child is too young to understand, that the child will disclose to others, that the child will be angry or resentful, or that she will be unable to answer the child’s HIV-related questions in a developmentally appropriate manner (Murphy, 2008; Palin et al., 2009; Qiao et al., 2013b; Rochat, Mitchell, et al., 2017; Yang et al., 2016). On the other hand, MLH have cited the stress of keeping a secret, the possibility of receiving support from adult children, and the desire to educate children about HIV as motivations for disclosing to children. Research examining the outcomes of disclosure for MLH and their children is somewhat mixed. Although most studies demonstrate benefits of disclosure (and, conversely, negative effects of nondisclosure; Qiao et al., 2013a), others reveal short-term maladaptive consequences (Palin et al., 2009; Yang et al., 2016), and still others show few or no effects (Mellins et al., 2008; Zhao et al., 2015). Outcomes of disclosure vary based on the child’s age at the time of disclosure, whether the disclosure was planned, whether children were asked to keep the disclosure a secret, and methodological differences across studies (Qiao et al., 2013a). Younger children tend to benefit more from disclosure, and outcomes are more positive when disclosure is planned and when children can speak with designated safe people about the illness.

In general, when children are younger and information is conveyed in a developmentally appropriate manner, HIV disclosure is related to positive outcomes for children and families, such as less aggression and better self-esteem for children, more social support for mothers, and stronger parent–child bonds and communication (Lee & Rotheram-Borus, 2002; Murphy, Steers, & Dello Stritto, 2001; Qiao et al., 2013a). Murphy, Marelich, and Hoffman (2002) demonstrated that HIV disclosure predicted decreases in child depressive symptoms over a period of 1 year. Furthermore, several studies have demonstrated a negative impact of nondisclosure on MLH’s physical and psychological functioning as well as children’s adjustment (Murphy et al., 2001). In a 5-year longitudinal study, maternal HIV disclosure was associated with reductions in adolescent problem behaviors, whereas nondisclosure predicted increases in problem behavior (Lee & Rotheram-Borus, 2002). Additionally, children whose mothers have not disclosed may still experience stress related to the limited information they have about their mothers’ health. For example, in a study of primarily nondisclosing MLH, 92% of mothers reported that their children knew they were taking medication, and 39% of these families reported that the children seemed worried about their mothers’ medication (Murphy et al., 2001).

However, despite its potential benefits, disclosing HIV to children may carry some risks. One study conducted among South African MLH revealed higher rates of child externalizing problems among children whose mothers had disclosed versus those who had not (Palin et al., 2009). This study found no effect of disclosure on internalizing symptoms. Importantly, researchers pointed out that high levels of HIV-related stigma and limited access to HIV care were important contextual factors influencing disclosure for this sample. In another study among families in rural China, children’s knowledge of parental HIV status was cross sectionally associated with worse social and emotional functioning, which the researchers attributed in part to potential misunderstanding of HIV-related facts (Xu, Wu, Rou, Duan, & Wang, 2010). Qiao et al. (2013a) posited that disclosure outcomes tend to carry a greater risk of negative outcomes for youth and families when disclosures occur spontaneously or in a developmentally inappropriate manner. This hypothesis has been supported by more recent qualitative evidence suggesting that children who experience unintended or unplanned disclosure report worse outcomes and more negative attitudes toward disclosure (Li et al., 2016).

Additional studies have not found significant long-term impacts of disclosure on child adjustment. For example, a qualitative study revealed that following MLH’s disclosure of HIV to their children, many children reacted with an initial increase in anxiety, but this reaction subsided over time as children adjusted to the news (Murphy, Roberts, & Hoffman, 2005). A study of early adolescents revealed no significant differences in internalizing or externalizing problems between youth who knew about the mothers’ HIV status and youth who did not (Mellins et al., 2008). Notably, this study relied on mothers’ report of whether youth were aware of their status and did not assess how or by whom the disclosure occurred. Overall, there is some conflicting evidence regarding the potential benefits and drawbacks of disclosure. Although most studies indicate positive outcomes for MLH’s individual functioning, parent–child relationships, and children’s adjustment, the documented benefits vary based on several factors. It remains unclear how HIV disclosure contributes to family processes—such as parenting, parenting stress, and child adjustment—across time.

An important consideration in understanding the impact of HIV disclosure on children and families is the child’s developmental stage. The developmental period between ages 6 and 14 (the age range of the current sample) comprises middle childhood and early adolescence. Middle childhood is characterized by an expansion of social relationships beyond the family, significant cognitive changes that allow children to reflect on their own abilities, and increased social comparison with same-age peers (Eccles, 1999). Children in this stage take on an increased sense of responsibility, which in the context of HIV disclosure may prepare children to learn of their parents’ illness but may also lead to parentification of the child. In early adolescence, youth are developing a more autonomous sense of self and experience accompanying biological, social, cognitive, and psychological changes. Although early adolescents continue to rely on family members for support, they increasingly turn to their peers for guidance and may receive misinformation or stigmatized messages regarding HIV. HIV disclosure and its effects unfold within the context of the many changes in development and family relationships across this period.

Based on typical timing of diagnosis for MLH, families’ needs and the impact of HIV on families may shift across time (Rochat, Netsi, et al., 2017). Many women are diagnosed in the perinatal period (Momplaisir et al., 2015). During the early years of managing the illness and parenting a young child, there is a focus on supporting MLH in adjusting to the diagnosis and emphasizing ART adherence. As children enter into middle and later childhood and MLH have been on ART treatment for several years, periods of illness are more likely, increasing the need to support MLH around Parent–Child Communication (PCC) and HIV disclosure. As children transition to adolescence, MLH must navigate the challenge of educating their youth about sexual health and HIV risk. For the youth in the current sample, who are in middle childhood and early adolescence, considerations about MLH’s ability to disclose their status, communicate about HIV and related issues, and educate their adolescents about sexual health are critical. These developmental changes also highlight the need for longitudinal studies examining change in child and family functioning across time and the over-time effects of disclosure.

Current Study

Existing knowledge about HIV disclosure and its impact on families is limited by several methodological factors, including cross-sectional designs, binary measurement of disclosure, and reliance on one reporter (i.e., parent or child; Qiao et al., 2013a). Theories of disclosure suggest that disclosure occurs as a process that unfolds over time (Qiao et al., 2013a). To more closely reflect this conceptualization, this study examined disclosure ordinally as nondisclosure, partial disclosure, and full disclosure. This measurement approach affords the opportunity to capture gradual disclosure over time. Qualitative interviews were also used to explore disclosure as a long-term process, rather than one discrete event. To further address limitations of the existing literature, a longitudinal design was implemented and data were collected from both children’s and mothers’ perspectives. Consistent with recommendations in the literature (Loeber, Green, & Lahey, 1990), children reported on their internalizing symptoms and mothers reported on externalizing behaviors. Additionally, a mixed-methods approach was utilized to triangulate quantitative and qualitative data and provide a richer view of the longitudinal associations between HIV disclosure and family outcomes.

Guided by the disclosure process model (Chaudoir et al., 2011), the purpose of this mixed-methods study was to investigate HIV disclosure and its interplay with other family processes over time. The first aim of this study was to quantitatively examine HIV disclosure as a longitudinal predictor of latent change in parenting practices, parenting stress, and child adjustment among a sample of MLH and their 6- to 14-year-old children. Using latent change score (LCS) modeling, we hypothesized that HIV disclosure would be beneficial for parenting, parenting stress, and child adjustment outcomes over time. Based on previously demonstrated cross-sectional associations between HIV disclosure and family processes, these hypotheses were largely confirmatory, though the effects have not been established longitudinally. The second aim of this study was to qualitatively explore the process of HIV disclosure and its perceived impact on family functioning, child adjustment, parenting practices, and parenting stress. No specific hypotheses were made for this exploratory qualitative aim.

Method

Design and Sample

This study used a sequential explanatory mixed-methods design (Ivankova, Creswell, & Stick, 2006), in which the qualitative portion of the study was intended to elaborate, enrich, and deepen the quantitative findings. The quantitative data for this study were drawn from a larger longitudinal NIMH-funded study testing the efficacy of an HIV disclosure intervention called Teaching Raising and Communicating with Kids (TRACK; Schulte et al., under review). Mother–child dyads were eligible for the study if the mother had a confirmed HIV/AIDS diagnosis and was the primary caregiver of a well child (not HIV+) between the ages of 6 and 14 who lived with her and was unaware of her HIV serostatus. Families must have been English or Spanish speaking. Mothers could be any female primary caregiver (e.g., biological mother, grandmother, or other female caregiver), but the terms “mother” and “MLH” are used to refer to all female caregivers in the sample. All caregivers in the sample identified as female. A total of 174 MLH and their 6- to 14-year-old children (N = 348) enrolled in the study across two sites: in California (Los Angeles and San Diego) and Georgia (Atlanta). Of the 174 dyads who enrolled, 100% completed baseline assessments, 95% completed 3-month follow-up, 93% completed 9-month follow-up, and 79% completed 15-month follow-up. After baseline, MLH were randomly assigned to receive the three-session individual intervention or to a wait-list control group. More detail about the TRACK intervention can be found in Schulte et al. (under revision) and Murphy, Armistead, Marelich, Payne, and Herbeck (2011). For MLH in the intervention condition, the intervention was delivered in three 90-min, one-on-one sessions, with the first session typically held within 1 week of the baseline assessment. MLH in the waitlist control group were invited to participate in a group-based, one-session format of the intervention after completing all quantitative follow-up assessments. Therefore, all families had the opportunity to participate in the intervention, either in an individual format between baseline and 3-month follow-up, or in a group format after completing 15-month follow-up. Demographic characteristics of the sample are presented in Table 1. The sample was racially, ethnically, and socioeconomically diverse, with one-third (35%) identifying as Hispanic/Latinx; over half (58%) identifying as Black/African American; majority single parents (80%); nearly one-third (30%) employed; monthly household income ranging from $0 to $6,000; and a wide range of educational attainment levels. As anticipated, the ethnic breakdown of the sample reflected the demographic impact of the HIV epidemic in California and Georgia (CDC, 2014).

Table 1.

Demographic Characteristics of Sample

	Overall sample			Qualitative subsample
Variable	M (SD) or %	Min	Max	M (SD) or %	Min	Max
Child age	9.65 (2.49)	6	14	12.50 (1.91)	9	16
Mother age	39.24 (7.90)	24	58	37.21 (8.75)	26	53
Child race
Black/African American	57.7%			42.9%
White	32%			42.9%
Multiracial	8%			7.1%
American Indian/Alaska Native	1.1%			0%
Asian	0.6%			0%
Native Hawaiian/Pacific Islander	0.6%			0%
Child ethnicity
Latino/Hispanic	35.4%			50%
Non-Hispanic	64.6%			50%
Mother relationship status
Never married	44%			71.4%
Married	20%			14.3%
Separated	14.9%			7.1%
Divorced	13.1%			7.1%
Widowed	8%			0%
Mother education
≤ 8th grade	16.6%			21.3%
Some high school	28%			14.4%
High school/GED	22.3%			0%
Vocational school	5.7%			0%
Some college	22.9%			28.6%
≥ College degree	4.6%			0%
Mother employment
Employed in the last 30 days	30.3%			42.9%
Not employed in the last 30 days	69.7%			57.1%
Monthly household income	$1,367.91 (931.41)	$0	$6,000	$1,431.71 (818.89)	$333	$3,200

Note. For the qualitative subsample, child age represents the age at the time of the qualitative interview; mother age represents age at baseline.

Qualitative interviews were conducted at both sites with a subsample of 14 MLH and 13 children who completed the study and in which full disclosure had occurred (n = 27). One child whose mother participated in a qualitative interview could not be located for an interview. Dyads were recruited using a purposive sampling strategy (Miles & Huberman, 1994), in which families are selected in a nonprobabilistic manner to capture a wide range of experiences. Of the 14 families who were interviewed, 12 were in the intervention condition and two were in the waitlist control group. The 12 mothers in the intervention condition had received the three-session, individual intervention between baseline and 3-month follow-up, and the two mothers in the waitlist control group had participated in a one session, group-based format of the intervention after completing all quantitative assessments (baseline and 3-, 9-, and 15-month follow-ups), but prior to completing qualitative interviews. Eight families were recruited from the California site, and six were recruited from the Georgia site. Half (n = 7) of target children were girls. The average age of children was 12.50 (SD = 1.91), with ages ranging from 9 to 16. Additional demographic characteristics for the qualitative subsample are displayed in Table 1. Demographic characteristics were similar to those of the overall sample, although there was a higher proportion of mothers who were never married. The qualitative sample also included a higher proportion of families who identified as Latinx/White, likely due to the higher recruitment for qualitative participants at the California site.

Quantitative Data Collection

Procedures

All procedures were approved by university Institutional Review Boards at both sites. Participants were recruited through partnerships with a range of community agencies. Informed consent and assent were obtained from mothers and children, respectively. Mother and child assessments were conducted by two separate interviewers using Computer Assisted Personal Interview software. All child assessments were conducted in English. Mother assessments at the Georgia site were also conducted in English, but those at the California site were conducted in English or Spanish, according to the mother’s preference. Mothers were paid $60 in cash for each assessment, and children were given $30 in gift cards. Quantitative data were collected at baseline and at 3-, 9-, and 15-month follow-ups.

Measures

In order to empirically validate the use of scale scores, a series of one-factor confirmatory factor analyses (CFAs) were used to establish the measurement models for each construct and evaluate longitudinal measurement invariance. For all scales, a single-factor model imposing scalar longitudinal measurement invariance adequately represented the data.

HIV Disclosure.

MLH reported whether they had disclosed their HIV serostatus to the target child using a series of quantitative (e.g., yes/no) and qualitative open-ended items. The first item is, “Have you or anyone else disclosed to [Child] that you are HIV-positive?” Response options include “Yes, just me,” “Yes, someone else,” “Yes, me and someone else,” “No, no one” and “Maybe.” Based on the response to this item, participants were asked follow-up probes to assess possible partial or full disclosure. Each participant’s responses were coded on an ordinal scale with 0 = no disclosure; 1 = partial disclosure, and 2 = full disclosure based on a set of coding criteria established by the research team. Criteria for full disclosure are met only if the mother reports that she has used the term “HIV” or “AIDS” in disclosing her illness to her child. Criteria for partial disclosure are met if the mother has told the child that she has an illness or sickness but has not disclosed the name of the illness. Nondisclosure status indicates that the mother either has not disclosed anything about her illness or has disclosed that she has an illness other than HIV which she does not actually have (e.g., cancer, diabetes, etc.).

Parenting practices.

MLH and children reported on two aspects of parenting: parent-child communication and relationship quality.

Parent–child communication scale.

Children completed the PCC Scale (Loeber, Farrington, Stouthamer-Loeber, & Van Kammen, 1998), a 10-item measure with a Likert-type scale ranging from 1 = almost never to 5 = almost always. Sample items include, “Do you think that you can tell your mother how you really feel about things?” and “Is your mother a good listener?” Previous studies conducted among a similar sample to that in this study have also utilized the PCC and demonstrated adequate psychometric properties (Murphy, Armistead, Payne, Marelich, & Herbeck, 2017). Items were averaged to create a scale score that could range from 1 to 5. Higher scores indicate more communication. Cronbach’s alpha for this sample was .68, which is comparable to that found in previous studies with demographically similar samples (McCarty & McMahon, 2003; Murphy et al., 2011).

Conflict behavior questionnaire.

Children completed the Conflict Behavior Questionnaire, short form (CBQ; Prinz, Foster, Kent, & O’Leary, 1979), which assesses relationship quality between parents and children. The measure includes 20 items assessed on a dichotomous True/False scale. A sample item is, “My mom seems to be always complaining about me.” Certain items are reverse scored such that higher scores indicate better relationship quality. This measure has previously been used extensively with a range of diverse populations, including families affected by HIV/AIDS (Goodrum, Armistead, Tully, Cook, & Skinner, 2017; Tompkins & Wyatt, 2008). Items were averaged to create a scale score, with possible scores between 1 and 2. Higher scores indicate stronger parent–child relationship quality. Cronbach’s alpha for this sample was .86.

Parenting stress index.

The Parenting Stress Index, short form (PSI; Abidin, 1990) is a 36-item scale completed by MLH. It includes three domains: Parental Distress (PSI-PD), Parent–Child Dysfunctional Interaction (PSI-DI), and Difficult Child (PSI-DC). The PD domain refers to parents’ perceptions of their functioning as a parent and distress within the parental role (e.g., “I feel trapped by my responsibilities as a parent”). Parent–Child DI assesses the extent to which parents find interactions with the child to be difficult and nonreinforcing (e.g., “My child rarely does things for me that make me feel good.”) Finally, the DC subscale refers to the parent’s perception of the child’s temperament and demandingness (e.g., “My child seems to cry or fuss more often than most children.”). This measure has previously been used with samples of MLH and their children (Murphy et al., 2010; Silver, Bauman, Camacho, & Hudis, 2003). Based on an initial CFA of the PSI-PD, one item was removed from the scale: “Having a child has caused more problems than I expected in my relationship with my spouse.” This item was removed both for theoretical (i.e., cultural context) and empirical reasons. For each subscale, items were averaged to create a scale score, with higher scores indicating more parenting stress. Possible scores ranged from 1 to 5. Internal consistency for each subscale was adequate in this sample (Cronbach’s alpha for PSI-PD = .82, PSI-DI = .82, PSI-DC = .85).

Child psychosocial adjustment.

Families reported on two aspects of child psychosocial functioning, including depressive symptoms (child-reported) and aggressive behavior (mother-reported).

Children’s depression inventory.

The Children’s Depression Inventory (CDI; Kovacs, 1985) assesses child-reported depressive symptoms using 27 items that each consist of three statements scored on a scale from 1 to 3 (e.g., 1 = I am sad once in a while; 2 = I am sad many times; 3 = I am sad all the time). Some items are reverse scored such that higher total scores indicate more depressive symptoms. The measure has good psychometric properties (Kovacs, 1985) and has previously been used among samples of MLH and their children (Murphy et al., 2011). Items were summed to create a scale score, as is commonly done with this instrument (Kouros, Quasem, & Garber, 2013). Possible scores could range from 27 to 81, with higher scores indicating more depressive symptoms. Cronbach’s alpha was .83 in this sample.

Child behavior checklist.

MLH reported on children’s aggressive behavior using the Child Behavior Checklist (CBCL; Achenbach, 1991). The CBCL aggression subscale consists of 17 items with a 3-point response scale from 1 = not true to 3 = very true or often true. Sample items include “Argues a lot” and “Gets in many fights.” The CBCL demonstrates good reliability and validity (Achenbach, 1991) and has been extensively utilized among diverse samples (Tompkins & Wyatt, 2008). As in previous studies (e.g., Kouros et al., 2013), the items were averaged to create a scale score with higher scores indicating more aggressive behavior. Scores could range from 1 to 3. The measure demonstrated adequate reliability (α = .86).

Demographic variables.

Mothers reported on demographic variables, including the target child’s age and gender, race, ethnicity, household income, and parental education and employment status. Study site and group assignment were also considered as potential covariates. Demographic characteristics of the sample are presented in Table 1.

Quantitative Data Analysis

Quantitative analyses were conducted in Mplus 8 (Muthén & Muthén, 2017) using LCS modeling. Univariate LCS models were used to assess longitudinal outcomes for each construct. HIV disclosure was then examined as a time-varying predictor of each outcome (child adjustment, parenting, and parenting stress) in separate LCS models. LCS modeling offers some benefits over traditional latent growth modeling, including greater flexibility in the functional form of the change process and in the understanding of time-dependent effects of covariates and predictors on the change process (Ferrer & McArdle, 2003). Analyses were conducted using full information maximum likelihood (ML) estimation under the missing at random assumption. Final models were re-estimated using a Bayesian approach as a sensitivity analysis to evaluate the robustness of the results to the large sample asymptotic assumptions of ML. Although the analyses were potentially underpowered to detect small effect sizes given the sample size (n = 174 dyads), the amount of information was bolstered by the use of repeated measures across four timepoints.

An unconditional univariate LCS model was estimated for each variable to examine longitudinal change within constructs. First, a constant change model, specifying linear change with no proportional change parameter, was evaluated. The constant change model assumes a linear change process within person. Second, the constant change model was relaxed by adding a proportional change parameter to the constant change model, yielding an exponential growth trajectory (Grimm, Ram, & Estabrook, 2016). Each pair of constant change and proportional change models was compared using a nested model likelihood ratio test. Homoscedasticity of residual variances was tested by imposing equality constraints and assessing decrements in model fit. A latent basis model was also evaluated for each outcome. Model fit indices and likelihood ratio tests for all univariate LCS models are displayed in Table S2.

To test HIV disclosure as a predictor of changes in child functioning, parenting, and parenting stress, disclosure status was included as a time-varying predictor in each univariate LCS model. In each model, LCS for the outcome variable were regressed on the ordinal HIV disclosure status variable at the corresponding timepoint. The assumptions of equal intervals (e.g., between non and partial disclosure, and between partial and full disclosure) and linearity for disclosure effects in relation to the outcomes were empirically supported in all models using a nested model comparison approach. Group assignment was included as a covariate in all conditional LCS models.

Qualitative Data Collection

Procedures

Qualitative data were collected from a subsample of 14 mothers and 13 children (n = 27) who had completed the study. Only families in which full HIV disclosure had occurred were recruited for qualitative interviews. During the screening and consent processes, mothers were asked to confirm whether they had disclosed their HIV status to their children, and mothers were informed of and consented to the fact that HIV would be discussed during their children’s interview if they chose to participate. Individual in-depth qualitative interviews were conducted separately with mothers and children by the first author and a graduate research assistant. Qualitative interviews primarily centered on HIV disclosure and its impact on child and family functioning. Qualitative interviews also probed children’s emotional and behavioral adjustment, parenting practices, children’s reactions to parenting behaviors, and family’s perceptions of the transactional interplay among family processes. As with the quantitative assessments, qualitative interviews were conducted in English or Spanish at the family’s home, the research office, or the recruitment site. Interviews lasted 75 min for mothers and 60 min for children. MLH were paid $60 in cash, and children were given a $30 gift certificate.

Qualitative Data Analysis

Qualitative interviews were transcribed verbatim by two trained graduate research assistants. Spanish interviews were transcribed, translated to English, and back-translated to Spanish to ensure accuracy, according to the method specified by Brislin (1970). After being transcribed, the qualitative interviews were analyzed using a directed qualitative content analysis approach (Hsieh & Shannon, 2005). This approach involves using preestablished categories or themes based on existing theory as an initial guide to group the data. Interviews are coded using the pre-established categories and new codes are generated as needed. QSR International Pty Ltd. (2015) NVivo (Version 11), https://www.qsrinternational.com/nvivo-qualitative-data-analysis-software/home software was used to conduct these analyses. Qualitative coding was conducted by a team of three coders, with each interview coded by the lead coder (first author) and one additional coder. Meetings were held to reach a consensus on the codes for each interview, and previously coded interviews were reviewed and re-coded to ensure consistency in the codebook.

Results

Quantitative Findings

Descriptive Findings and Bivariate Correlations

Descriptive statistics are presented in Table 2, and bivariate correlations are found in Table S1. Group assignment was included as a covariate in all conditional LCS models. Due to statistical power constraints, demographic variables that were not independently associated with disclosure and each outcome variable (e.g., parenting, parenting stress, child adjustment) at concurrent or adjacent timepoints were excluded from the conditional LCS models examining associations between disclosure and outcomes. Therefore, demographic variables, including race, ethnicity, child age, child gender, and study site were not included as covariates in the LCS models due to statistical power considerations and the pattern of bivariate correlations.

Table 2.

Descriptive Statistics for All Study Variables Unconditional LCS Models

	Baseline			3-month			9-month			15-month
Variable	M (SD)	Min	Max	M (SD)	Min	Max	M (SD)	Min	Max	M (SD)	Min	Max
CDI	35.93 (6.95)	27.00	58.00	34.45 (7.09)	27.00	61.00	33.11 (5.66)	27.00	57.00	33.11 (6.29)	27.00	70.00
CBCL-Agg	1.48 (0.35)	10.00	20.71	1.44 (0.36)	10.00	20.71	1.45 (0.36)	10.00	20.71	1.44 (0.35)	10.00	20.76
PCC	3.66 (0.73)	10.50	50.00	3.81 (0.72)	10.70	50.00	3.82 (0.69)	20.00	50.00	3.94 (0.69)	10.50	50.00
CBQ	1.83 (0.19)	10.05	20.00	1.84 (0.20)	10.05	20.00	1.86 (0.17)	10.30	20.00	1.88 (0.16)	10.05	20.00
PSI-PD	2.48 (0.77)	10.00	40.55	2.25 (0.72)	10.00	40.18	2.18 (0.75)	10.00	40.30	2.21 (0.84)	10.00	40.82
PSI-DI	2.02 (0.73)	10.00	40.60	1.91 (0.64)	10.00	30.40	1.90 (0.68)	10.00	40.00	1.83 (0.70)	10.00	40.20
PSI-DC	2.29 (0.79)	10.00	40.92	2.25 (0.79)	10.00	40.17	2.15 (0.77)	10.00	40.50	2.20 (0.76)	10.00	40.08

Note. CDI = Children’s Depression Inventory; CBCL-Agg = Child Behavior Checklist—Aggression Subscale; PCC = Parent–Child Communication Scale; CBQ = Conflict Behavior Questionnaire; PSI-PD = Parenting Stress Index—Parental Distress; PSI-DI = Parenting Stress Index—Dysfunctional Interaction; PSI-DC = Parenting Stress Index—Difficult Child.

Unconditional LCS Models

For child depression (CDI), the retained model was a constant (linear) change model with homoscedastic residual variances. On average, depressive symptoms decreased significantly across time (M_slope = −1.03, p < .001). For aggression (CBCL), a constant change model with homoscedastic residual variances was retained. The slope was nonsignificant (M_slope = −.02, p = .08); aggression did not change significantly across time. For PCC, the retained model was a constant change model with homoscedastic residual variances. PCC increased significantly across time (M_slope = .09, p < .001). A constant change model with homoscedastic residual variances was retained for parent–child relationship quality (CBQ). The slope of the CBQ was significant and positive (M_slope = .02, p < .001); children reported improvements in relationship quality across time. Based on inspection of plots and results of Bayesian estimation, each parenting stress LCS model was specified as a latent basis model by removing the slope factor and freely estimating the variances and means for each LCS. This specification addressed convergence issues and allowed for the longitudinal trajectory of parenting stress to be freely estimated, without imposing a fixed shape (e.g., linear).

Conditional LCS Models With Disclosure

Model fit indices and parameter estimates for all conditional LCS models with disclosure are displayed in Table 3. HIV disclosure was examined as a time-varying predictor of parenting practices in a separate model for each parenting outcome (PCC; and relationship quality, CBQ). Both models demonstrated good fit to the data. When accounting for group assignment, HIV disclosure was a significant predictor of changes in parenting for both PCC and CBQ. In the PCC model, disclosure status at 9-month follow-up predicted significantly steeper increases in PCC during the same timeframe, β = .65, p = .02. Children whose mothers had partially or fully disclosed by 9-month follow-up were more likely to report improvements in communication between 3- and 9-month follow-up, compared with those whose mothers had not disclosed. However, disclosure was unrelated to changes in PCC at other timepoints. In the CBQ model, disclosure status at 3-month follow-up was related to steeper increases in parent–child relationship quality, β = .73, p < .001. In families in which mothers had fully or partially disclosed by 9-month follow-up, children reported more improvements in their relationship with their mother between 3- and 9-month follow-up, compared with families in which disclosure had not occurred. HIV disclosure was not related to changes in relationship quality at other timepoints.

Table 3.

Model Fit Indices and Parameter Estimates for Latent Change Score Models with Disclosure as Time-Varying Predictor

	CFI	RMSEA	χ 2	df	P	IntY with SlpY	IntY on Disc1	ΔY2 on Disc2	ΔY3 on Disc3	ΔY4 on Disc4
XDISC, YPCC	1.00	0.00	16.79	20	0.67	–0.02 (0.02)	–0.24 (0.15)	0.10 (0.06)	–0.06 (0.07)	–0.05 (0.07)
						–0.39 (.2l)†	–0.l5 (0.l0)	0.65 (0.29) *	–0.45 (0.45)	–0.43 (0.53)
XDISC, YCBQ-C	1.00	0.00	13.79	20	0.84	0.00 (0.00)*	–0.07 (0.04)†	0.04 (0.02)*	–0.02 (0.01)	–0.02 (0.02)
						–0.80 (0.28)**	–0.l6 (0.09)†	0.73 (0.l9) **	–0.64 (0.42)	–0.58 (–.51)
XDISC, YPSI-PD	0.96	0.07	23.94	16	0.09	—	0.27 (0.17)	–0.16 (0.08)*	0.00 (0.07)	0.02 (0.08
						—	0.l7 (0.ll)	–0.0.4l (0.28)†	0.02 (0.48)	0.05 (0.25)
XDISC, YPSI-DI	0.94	0.08	28.46	16	0.03	—	0.07 (0.15)	–0.15 (0.07)*	0.10 (0.06)†	0.06 (0.06)
						—	0.05 (0.ll)	–0.57 (0.23)*	0.45 (0.26) †	0.26 (0.28)
XDISC, YPSI-DC	0.97	0.07	25.90	16	0.06	—	0.27 (0.17)	–0.15 (0.07)*	0.01 (0.06)	0.10 (0.06)†
						—	0.l6 (0.l0)	–0.56 (0.24)*	0.04 (0.29)	0.40 (0.23) †
XDISC, YCDI	0.91	0.08	31.39	19	0.04	—	0.05 (1.46)	–0.22 (0.72)	0.66 (0.75)	0.42 (0.77)
						—	—	—	—	—
XDISC, YCBCL-AGG	0.99	0.03	22.83	20	0.30	0.00 (0.00)	0.06 (0.07)	–0.02 (0.03)	0.02 (0.03)	0.04 (0.03)
						–0.24 (0.l6)	0.07 (0.09)	–0.26 (0.3l)	0.29 (0.32)	0.44 (0.25)

Note. Unstandardized coefficients displayed, with standardized coefficients beneath in italics. All models included group assignment as a covariate. Dashed lines (—) indicate parameters that were not specified or could not be estimated due to model convergence issues. DISC = HIV Disclosure; CDI = Children’s Depression Inventory; CBCL–Agg = Child Behavior Checklist—Aggression Subscale; PCC = Parent–Child Communication Scale; CBQ = Conflict Behavior Questionnaire; PSI–PD = Parenting Stress Index—Parental Distress Subscale; PSI–DI = Parenting Stress Index—Dysfunctional Interaction; PSI–DC = Parenting Stress Index—Difficult Child.

^†p < .10.

^*p < .05.

^**p < .01

Next, HIV disclosure was examined as a time-varying predictor of each subscale of parenting stress. All three models demonstrated good fit to the data. Across all three models, HIV disclosure between baseline and 3-month follow-up predicted either steeper reductions (for PD, β = –.51, p = .07, and DI, β = –.57, p = .01) or more gradual increases (for DC, β = –.15, p = .04) in parenting stress during the same timeframe, even when accounting for group assignment. Mothers who fully or partially disclosed between baseline and 3-month follow-up reported greater reductions in PSI-PD and PSI-DI scores, and fewer increases in PSI-DC scores.

Finally, to examine HIV disclosure as a longitudinal predictor of child outcomes, HIV disclosure status at each timepoint was included as a time-varying predictor in the LCS model for depression (CDI) and aggression (CBCL) separately. The model fit for both models was good. However, HIV disclosure status was not significantly related to changes in child depression or aggression when accounting for group assignment. Across all LCS models, results from Bayesian estimation were consistent with the ML estimation findings.

Qualitative Findings

Consistent with a sequential explanatory design, the second aim of this study was to qualitatively explore HIV disclosure and its perceived connections with family processes from the perspective of MLH and their children, and utilize this data to provide context and potential explanations for the quantitative findings. Three overarching themes were identified: children’s reactions to HIV disclosure, mothers’ experiences of HIV disclosure, and family changes after disclosure. Participant ID numbers are displayed parenthetically after each quote; the first digit reflects the study site (1 = California; 2 = Georgia), and the last letter reflects mothers (M) or children (C).

Theme 1: Children’s Reactions to HIV Disclosure

Children and parents reported a wide array of negative, positive, and neutral emotional and behavioral responses to HIV disclosure, ranging from relief and happiness, to feeling “normal,” to sadness and worry. Overall, most families reported that children adjusted to the news quickly and easily, though some did experience (usually temporary) negative emotional reactions. The most common response described by both mothers and children was that the child felt normal or neutral about the disclosure. Several children described their emotional reactions as “normal,” “neutral,” and “calm.” For example, a 14-year-old girl shared,

She tells me the whole thing, goes down the whole rabbit hole. I think she was confused as to why I was just like, “Ok!” and she’s like, “What?” [laughs] I’m like “Ok!” ‘Cause like, it didn’t bother me. Same mother I’ve always had. You’re the same person before you told me this, same person after you told me. I’m not just gonna be like [gasps], “Oh no!” You’re the same person. I don’t really mind. Doesn’t bother me in any way. (1067C)

Another child shared that she felt happier after learning of her mother’s HIV status:

I think I kept being the same child, smiling child. But no, it didn’t change none of me. Like, no. Actually it made me more happy because I kind of understood why and I kind of understood why I went to these places [HIV-related camps], and so, yeah, so I’m like, okay now I understand. (1023C)

Some mothers shared that they did not perceive significant changes in their children’s behavior after the disclosure:

I don’t notice like he was impacted. Like that it was like a change he would make with me. Like, I don’t know that he would be different with me. That he would have said, “Ay my mom, poor her” or “She is going to die.” No, no, he took it—he took it not too strongly, I believe. Mhm, he understood that it was a normal illness. (1053M; translated from Spanish)

One mother shared that her children responded more calmly than she had expected:

They were more open to me telling them than I thought they would be. And they was like, ‘Ma, you’re okay, you know you’re going to fight through it.’ . . . You know, I thought, you know, someone was going to cry. And you know they did, you know, do a little crying, but it wasn’t as bad as I thought. (2044M)

A few families described positive changes related to the child’s attitude toward people living with HIV and/or openness to discussing HIV-related issues:

I guess kind of opening up a little bit more, like, “Oh ok. This is a normal thing. Mom has it, and she’s normal.” I guess it gave me another reason not to alienate people who are like that. Cause I have my own living example. They’re just normal people, who are just a little bit different. (1067C)

Some children did experience negative emotional reactions, including sadness, worry, and anger. For example, one child described feeling sad and worried:

It was pretty like sad knowing that like it was incurable and she was gonna have to live with that, but like other than that it was more like just worrying. But like, is she gonna be ok? Is, like, anything gonna happen to her? Is she fine? Does she need anything? (1032C)

For a teenage boy who had existing difficulties with anxiety, finding out about his mother’s HIV status exacerbated his symptoms temporarily: “I guess in the beginning it made me feel a little more anxious, but with the therapy that kinda went—went away” (1042C). He went on to describe that despite the initial adjustment period, his anxiety overall improved after learning of his mother’s HIV status, and he felt a sense of relief after knowing about his mother’s health.

Several children expressed fears that their mother would pass away. One adolescent boy, whose mother experienced co-occurring health issues in addition to her HIV status, said,

I was worried that she was like—she would like pass away and stuff. . . . it does [worry me] because, like, I know she still has it, you know? But like, not that much because, like, she’s healthy right now. (1025C)

HIV-related deaths of other family members also affected how children received the news. For example, one mother perceived that her son’s experience losing another parent to HIV exacerbated his negative emotional response to learning of her status. Consistent with his mother’s report, the adolescent shared that he felt afraid that his mother would pass away: “I felt like scared because like usually by the disease you usually die quicker. And I’m scared that 1 day I’m just going to wake up and she’s not responding” (1049C). He further explained that he had not communicated this fear directly to his mother, and that he also felt afraid of losing other family members who are living with HIV. Another child expressed feeling upset that his mother had not disclosed her status earlier, out of fear that she could pass away:

It’s not just she didn’t tell us; it’s, like, the fact that she could die from it, yeah, the disease that she has. It would’ve been better for us to know ‘cause she could’ve died at any time from having HIV and we wouldn’t have known how she died. (2039C)

Many children expressed that they felt a combination of positive and negative emotions, such as a 9-year-old boy who shared that he felt sad and happy when he learned of his mother’s HIV status: “I kind of feel sad that my mom has that and has to take medication or else, you know . . . [I feel] kinda sad, but I still feel happy because she’s still alive” (1053C). A 13-year-old girl described feeling sad and angry about her mother’s illness:

I felt, um, sad and angry at the same time . . . Like, ‘cause the way she told us that she was, um, HIV—like how she got HIV positive, I felt very angry and stuff like that. Since she said she had gotten it from our dads, and I felt angry towards them. . . . [I felt sad because of] like all the stuff that I had heard about HIV and stuff because I thought she was gonna like die from it and things like that. (2044C)

Her mother, describing her child’s adjustment to the disclosure, stated that after an initial period of sadness lasting about 30 min, her daughter became calm. Other participants also shared that the child experienced an initial reaction of sadness or worry, but that they adjusted relatively quickly. For example, a 13-year-old girl and her mother both agreed that the child’s initial sadness gave way to feeling calm and neutral. The mother shared,

At the moment I told her, I saw her being really sad and I saw her reaction like being really sad, like “What’s going to happen? Are you going to die like the boy [in the movie]?” “No.” But then she changed when I started explaining it to her more, and she’s like “Okay.” (1023M; translated from Spanish)

Similarly, the daughter described that she adjusted to the news relatively easily:

I wasn’t mad with her, ‘cause some kids would be, but I was like, I accepted her because she is my mom, and people say it’s bad because it goes—it passes through or something, like it spreads, but I’ve been—I’ve been seeing in time and nothing happens. She seems like a normal parent. [. . .] I wasn’t surprised, I was alright with it, like okay. I wasn’t mad, I wasn’t mad at her, I wasn’t sad; I was just, like, calm, like alright.

Because she told me that as long as she drinks the medicine everything will be alright. (1023C)

Another child shared that she “was just sad for 1 day” (2066C). Her mother’s perspective provided a slight contrast, as she perceived the child to be uninterested and calm from the beginning: “She wasn’t even interested. She was like, ‘Mmm.’ . . . She didn’t say much” (2066M). In some cases, children remained largely silent during the disclosure conversation, making it difficult for mothers to gauge their children’s emotional reactions. As one mother put it, “I ain’t get no feedback” (2076M).

Although there was consistency between mothers and children’s accounts in several families, some families had slightly or widely different perspectives on the child’s adjustment to the disclosure. Generally, where discrepancies were observed, mothers reported more calm and neutral reactions compared to children, who reported more internalized emotions (e.g., sadness, worry) related to the disclosure. For example, the grandmother of a 10-year-old boy reported that the disclosure “didn’t phase” her grandson (1082M). She described a 1- to 2-week period in which he was more “clingy,” but shared that he adjusted quickly after that. She also noted that she was unsure if the child paid full attention during their conversations about HIV. In contrast, the child shared that it took him about a month to adjust to the news. He said, “[The rest of my family] had 10 years to adjust to it. I only had, like, only a few” (1082C). The child stated that at the time of the disclosure, he was about 6 years old and did not know what HIV was.

Theme 2: Mothers’ Experiences of HIV Disclosure

Nearly all mothers perceived that the disclosure went well overall. One mother described thinking that she handled the disclosure appropriately, including communicating on the child’s developmental level:

I feel that I did it in a way, um, um, appropriate. I feel that I did it in a way—on a level with her capacity to understand of her age—her level of age. I feel that I did it in, um, uh, um, um, very, very private with her. In that it was her dad, me, and her. And I feel that it was the right time to tell her too. (1032M; translated from Spanish)

Several mothers echoed similar sentiments that they were pleased with how the disclosure conversation went, and that they felt a sense of relief after disclosing their HIV status to their child. Mothers shared that they felt more comfortable speaking openly about their status and about sex, and that they felt like a burden had been lifted after the disclosure.

I don’t feel like I’m holding as much in as a secret, you know, like, I don’t have to dance and prance around and tiptoe around my conversations when I’m talking about it, so I can freely talk about condoms and sex and HIV like it’s just normal conversation in the house. So it doesn’t seem like abnormal or uncommon to them; it’s part of their lives. (2019M)

Actually it helped me more than I felt that it helped them. It was like more of a release . . . After that, I felt like I could really tell anybody; it doesn’t bother me, ‘cause if I can tell my kids, I can tell anybody. [. . .] Since I’ve told them, it’s been more calm, like I don’t feel like a weight on my shoulder. (2044M)

Other positive aspects of the disclosure described by caregivers included that mothers provided reassurance and emotional support to the child, kept an appropriate and calm emotional tone, were open and honest and offered clear information, all of which are addressed in the study intervention. Mothers also shared that children listened and paid attention during the discussion. When asked what could have gone better during the disclosure, many mothers could not identify an aspect of the conversation that they would have changed. These mothers felt that the discussion proceeded as smoothly as could be expected, often exceeding mothers’ expectations.

I did explain to her the word HIV, but I feel like, I think the program really prepared me to talk about it more and focus and talk to her more directly. Yes, so I don’t think I would change anything, everything would remain the same. (1023M; translated from Spanish)

Other parents identified aspects of the disclosure process that they would have liked to improve, such as being able to tell the child more details about their health and HIV infection, having a better awareness of the child’s existing understanding of sexual health, regretting that the child’s father was not present for the disclosure, and wishing the child had asked more questions.

Theme 3: Family Changes After HIV Disclosure

Participants described a number of changes, primarily positive, in their family functioning, including with regard to parent–child relationship quality, communication, children’s behaviors, and mothers’ stress level. Although many families endorsed at least some type of positive change, the vast majority also expressed that most aspects of their family relationships remained the same. Several children shared that their family relationships remained stable.

Well, it didn’t change nothing. The family’s still the same. My mom is still the same. It didn’t change much. Our family, we talk normally. I guess it’s just a normal family. Even though my mom has that disease, it’s not going to change nothing. (1023C)

Mothers shared similar perspectives that there have not been major changes in their parenting or family relationships since the disclosure. In addition to endorsing stability in their family functioning, most families also shared that the mothers’ disclosure of her HIV status served to strengthen the parent–child relationship and mothers’ parenting quality. One mother described feeling a stronger bond with all of her children following the disclosure:

I think our bond has gotten a little stronger. And, um, we talk like every day about—sometimes it’s just random stuff, but it’s like it drew them closer to me, which I love. (2044M)

Some families attributed their strengthened relationship with the trust that the mother communicated by sharing personal information about her health with the child.

It helped our relationship because she told us her biggest, biggest, biggest secret. So now I feel more close to my mom because of what she told us. (2076C)

If I had to guess, it’s gotta make you feel pretty damn good to think somebody would trust you with something so important, even at a young age, like, ‘you trust me enough?’ Like, so it’s—it opens up something, like, ‘You picked me,’ you know what I mean? [. . .] So maybe you do love me. (1067M)

Mothers believed that the quality of their parenting improved after disclosing their HIV status. Some shared that the disclosure increased their openness and communication with their children, that they began to give their children more freedom as the communication increased, and that they began sharing more information about sexual health with their children.

So I think that my parenting has changed in the fact that, you know, just taking accountability and realizing that, you know, if I don’t want my children to go through this, then I need to, you know, equip them with the right amount of knowledge and information so that they can make the healthiest and best decision for themselves. (2019M)

Similarly, in the context of how the mothers’ parenting had changed, one mother stated the following.

Our conversations, and that we spend more time together. We talk more. I mean, it’s a work in progress. She’s 12. She’s at that age where they don’t want to talk to they parents. They don’t want—they want to be rebelling. And they hold everything in. But like I can tell when something is wrong with her because—but my main goal is to make sure every day I ask her, “How is your day? How is your day go?” Instead of saying, “I had a bad day.” Instead I put my feelings on the side . . . Instead of me saying how was my day and complaining about myself, I ask her . . . That’s what I learned to going to [the program] parenting class and stuff like that. How to have open, honest, open conversations and try to put their feelings and how they day was. It will make them come to you even more. (2083M)

From children’s perspectives, communication in the parent–child relationship also improved. Children noted that they talk with their parents and other family members more.

We’ll talk about more things, ‘cause she doesn’t, you know, now that—since you’ve reacted good about something that wasn’t good, she knows that she could tell you a lot more about other subjects. So it’s kind of like more open, like not that much worry going on. Like more reassurance and stuff. And you feel more comfortable.

A common theme regarding changes in the parent–child relationship after HIV disclosure was that many children began offering increased support to their mothers, sometimes leading to parentification of the child. Children expressed concern for their mother’s well-being, tried to help take care of their mother’s health, or started helping more with household tasks such as chores. Several families shared that the child became more obedient, helpful, and responsible. The majority of children began reminding their mothers to take their medication. Some children provided emotional support to their mothers by offering reassurance or trying to cheer them up. For example, an adolescent boy shared that after learning of his mother’s HIV status, he worked to be “respectful to her, like listening to her, and not like trying to make her mad and stuff” (1025C). In some cases, the support offered by children to their parents was related to children’s increased anxiety regarding their mother’s health. For example, one child shared that she checked in with her mother frequently about taking her medication, which contributed to feelings of anxiety and worry.

Ever since I was little, I would kind of like, have her in the back of my mind. And I would be like, “So are you ok?” and I would be like, “So did you take your pill? You forgot yesterday. I saw you forgot.” So like, try and whenever I see that she would forget, I would, like, after then, be more persistent in asking her. And it kind of just, it just was mainly ongoing worry which led also to my, like, nervousness and stuff, and it added to that. But like, I wouldn’t pay attention to that. I would pay more attention to my mom. (1032C)

In another family, a mother said that her son began reminding her to take her medication out of a fear that she would pass away otherwise:

And so he started making a change. A change—he made a change—a change that, “Mom, your medications,” now every night, every, every night he’s like, “Mom your medications, la la”. He thought that I—well, I think that he thought that if I didn’t take my medication, then I would die, that’s what I think. So he would tell me “Mami.” He lasted a while where he treated me well. He would ask, “Mom do you feel okay, do you feel okay?” (1025M; translated from Spanish)

Discussion

MLH make the decision to disclose in the context of complex and dynamic family processes as children develop and undergo significant social, cognitive, and psychological changes. Quantitative results of this study revealed that HIV disclosure was associated with over-time benefits for parenting quality and parenting stress, but not child adjustment. Qualitative themes mirrored quantitative findings and revealed several perceived benefits of disclosure for parent–child relationships and maternal stress, as well as stability in child and family functioning following disclosure. These findings were demonstrated among a racially, ethnically, socioeconomically, and geographically diverse sample of MLH, with characteristics largely reflecting the demographic breakdown of the HIV epidemic in California and Georgia (CDC, 2014). The sample was approximately one-third Hispanic/Latinx, over 50% Black/African American, primarily single mothers (80%), with a wide range of educational and household income levels.

Consistent with study hypotheses, children whose mothers disclosed or partially disclosed their status reported greater improvements in PCC and relationship quality, compared with families in which disclosure did not occur. This finding is consistent with previous cross-sectional research and suggests that HIV disclosure may serve to enhance the parent–child relationship across time. Children may feel more connected to their mothers because of being entrusted with personal health information. Qualitative interviews supported this notion. Several families stated that their relationships improved following the disclosure, and some families specifically discussed a bolstered sense of trust after the conversation. This increase in trust may be particularly important for youth in middle childhood and early adolescence as they begin to take on a heightened sense of responsibility and feel more prepared for knowledge of family stressors such as maternal HIV (Eccles, 1999). Taken together, the quantitative and qualitative findings suggest that HIV disclosure serves to enhance open communication and closer bonds within the relationship.

In addition to enhancing children’s perceptions of parenting quality, results indicated that HIV disclosure is beneficial for parenting stress among MLH. Mothers who fully or partially disclosed had greater decreases (or smaller increases) in parenting stress compared with those who did not disclose. This finding is consistent with hypotheses and is likely underpinned by the fact that nondisclosure is a significant source of stress for mothers (Murphy, 2008). Similarly, qualitative interviews revealed that most mothers felt a sense of relief after sharing their HIV status with their children. These findings indicate that disclosing HIV status is favorable for mothers’ stress levels and conversely, nondisclosure leads to elevated stress. The benefits of disclosure for parenting stress may operate by relieving mothers’ burden of secrecy, allowing them to be more open with their health information and thereby increasing medical adherence. Mothers’ parenting stress may also be reduced because children take on more household responsibility as they develop a sense of autonomy (Eccles, 1999); this change was noted by several families in qualitative interviews. At the same time, this increased sense of responsibility and autonomy may also have negative consequences by leading to parentification of the child, as referenced by a few families qualitatively.

Although LCS modeling revealed that HIV disclosure predicted improvements in parenting and parenting stress, significant effects were not detected between HIV disclosure and child adjustment. This was a surprising finding given cross-sectional data indicating some benefits of disclosure for child outcomes (Murphy et al., 2001) but was somewhat consistent with other studies demonstrating null effects of disclosure on child functioning (Mellins et al., 2008). Several possible explanations may account for this unexpected finding. One possibility, particularly given the effect sizes observed for the conditional LCS model for child depression, is that the effect does exist but we were unable to detect it due to empirical limitations. For example, it is possible that significant covariance between HIV disclosure and child adjustment across time was not observed because of the high temporal stability within each variable and/or lack of sensitivity in the measurement tools to detect change across time. With limited variability in measures across timepoints, it is difficult to detect significant covariance effects. It is also possible that the time period in this study (15 months) does not capture the longer-term effects of HIV disclosure on families. Given the benefits of disclosure on parent–child relationship quality, as well as the efficacy of the disclosure intervention in promoting child outcomes (Schulte et al., under revision), it is likely that with a longer follow-up window, more benefits in child adjustment would be observed—perhaps indirectly via changes in parenting. Alternatively, it may be that the spacing of the follow-up assessments (i.e., 3 months or 6 months apart) did not adequately capture fluctuations in child adjustment between each timepoint. It is possible that patterns of disclosure change, rather than disclosure status at each timepoint, would be more predictive of child adjustment over time. However, patterns of change could not be modeled in this study due to collinearity and lack of sufficient variability in patterns of change across time. Based on previous studies (Li et al., 2016; Qiao et al., 2013b), it is also likely that the quality of the disclosure, including whether it was well-planned or spontaneous, may be more important for child adjustment than disclosure status.

An alternative explanation to this pattern of findings is that HIV disclosure in this population may not have a significant impact on child adjustment, and other family and contextual factors may play a more important role. Notably, while the TRACK-II outcome study (Schulte et al., under review) demonstrated an impact of the HIV disclosure intervention on child outcomes, this study did not examine the effect of the intervention but rather the longitudinal impact of disclosure itself, finding no significant effects on child adjustment within the study timeframe. Qualitative findings emphasized the overall stability in functioning following disclosure; the most common reaction to the disclosure was that very little changed for children’s adjustment. Given the changing social ecology of the HIV epidemic, youth may now be less affected by news about their mothers’ HIV status, compared to those sampled in previous decades when HIV was associated with a higher mortality rate. Also, developmentally, youth are experiencing many changes during middle childhood and early adolescence, including a broadening of social relationships beyond the family. It is possible that other changes, not measured in this study, supersede HIV disclosure in influencing child adjustment.

Furthermore, a notable difference between MLH in this sample and those in previous samples demonstrating benefits of disclosure is that the mothers in this sample were specifically recruited into a randomized controlled trial because they had not shared their HIV status with their child at baseline. There are likely meaningful differences between mothers who choose to disclose and those who choose not to disclose. For example, a qualitative study among MLH who had already disclosed their serostatus revealed that, despite being socioeconomically similar to the current sample, the mothers felt they had adequate access to resources and could adeptly navigate government assistance programs (Tarantino, 2015). In contrast, MLH in this sample expressed high levels of financial and economic stress (e.g., 70% unemployed). Additionally, because these MLH have delayed disclosure, children may experience more stress related to their growing awareness of the unrest in the home. These distinctions may indicate that disclosure is more beneficial for children of MLH who decided to disclose without the aid of an intervention program.

Limitations and Directions for Future Research

This study represents the first longitudinal, mixed-methods investigation of HIV disclosure and its interplay with family processes among MLH. The findings of this study should be taken in the context of the study’s limitations. First, participants of this study were recruited into a randomized controlled trial targeting HIV disclosure; at the time of enrollment, all the mothers had chosen not to disclose to their children. Therefore, the findings of this study may not generalize to MLH who choose to disclose without the aid of an intervention. Second, patterns of changes in disclosure status across time could not be included in the LCS models due to insufficient variability in change patterns. The quality of the disclosure conversation, such as whether it was planned, was not quantitatively measured in this study. Future research with larger samples should examine how patterns of change in disclosure and quality of disclosure conversations relate to family outcomes. Third, no covariates beyond group assignment were included in the conditional LCS models due to statistical power constraints and patterns of bivariate correlations; future research with larger samples should examine the role of age, race/ethnicity, socioeconomic factors, and other demographic variables in the longitudinal relations among HIV disclosure and family processes. Finally, due to the timing of data collection and the lower base rate of disclosure among MLH not receiving the intervention, all mothers in the qualitative subsample had received the intervention, either in an individual format (intervention condition) or a group format (waitlist control condition). Thus, qualitative themes are likely influenced by the content of the intervention, which includes parenting skills as well as cognitive and behavioral content related to the disclosure process.

Conclusions

Despite its limitations, this study represents an important step in illuminating the impact of HIV disclosure on families. This study was the first to examine HIV disclosure as a longitudinal predictor of latent change among family processes. The integration of quantitative and qualitative findings from both mothers’ and children’s perspectives strengthens the existing literature by providing a richer and more contextualized view of HIV disclosure and family processes in this population. Results highlight the beneficial role of HIV disclosure, particularly in spurring reductions in parenting stress and improvements in children’s perceptions of the parent–child relationship. The parent study (Schulte et al., under revision) demonstrated the efficacy of the TRACK intervention, which focuses on developmental considerations, positive parenting and family practices (e.g., routines, PCC skills), disclosure self-efficacy, perceived barriers to disclosure, and behavioral strategies for the disclosure process. Findings from this study lend additional support to the overall benefits of promoting maternal HIV disclosure in the context of an intervention such as TRACK. Disclosure interventions can be further strengthened by providing increased support to youth as they navigate learning of their mother’s illness during an already vulnerable developmental period. MLH and their children should be supported through the disclosure process to reduce parenting stress, bolster parenting skills, and promote positive outcomes for families as they adjust to this transition.

Supplementary Material

sm1-2

Table S1. Bivariate Correlations Among All Study Variables at Baseline

Table S2. Model Fit Indices for Unconditional Latent Change Score Models