We conducted semi-structured interviews between March 2004 and March 2005 with a sample of HIV-infected parents from the HIV Cost and Services Utilization Study (HCSUS) about their memory of disclosure. HCSUS is a national probability sample of people ≥18-years-old with known HIV infection who made ≥1 visit to a medical provider in the contiguous United States during January–February 1996 [22]. HCSUS participants were eligible for this follow-up study if they participated in HCSUS’ wave 3 and a 1997–1998 follow-up survey [23], had a child ≤23 years on 3/1/04, and had seen ≥1 of their children in the past month when they were contacted for participation.

To maximize sample diversity on key variables, we conducted a random stratified sampling of the 975 eligible participants using four strata: parent gender, parent race/ethnicity (African-American, Hispanic/Latino, White/Other), age of youngest child (<13, 13–17), and level of contact with child (lives with children, does not live with any children). We over-sampled those who participated in the 1997–1998 follow-up survey and selected all families with a child <18. The result of these procedures was a sample of 486 parents. We attempted contact with all sampled parents, located 146 using tracking information from previous surveys and a commercial address correction service, and discovered that 88 parents were deceased or ineligible. Among the 58 remaining parents, 33 agreed to participate along with 27 children (9–17-years-old), 19 adult children (≥18 years old), and 15 caregivers. Children were eligible to be interviewed if they knew about their parents’ HIV status and lived with or had seen their parents in the past month. We followed procedures developed with the RAND and UCLA Institutional Review Boards to obtain informed consent from respondents. One of the authors, an expert in qualitative methodology, trained seven interviewers to conduct semi-structured interviews, with an emphasis on how to balance following an interview protocol with using various exploratory probing techniques, and how to interact with children of HIV positive parents. All of the interviewers had prior masters training in psychology, counseling, or social work and experience conducting interviews on sensitive topics. Parents consented for interviews with children <18 who knew about their infection, and these children gave their assent for study participation. Two nieces were interviewed as adult children because they lived with and considered target parents to be parental figures.

The parents were 48% African American, 3% American Indian, 21% Latino, and 27% White. Most (73%) were mothers, and the mean age of parents was 44 years (SD = 7; Range = 30-62). Almost half (48%) reported annual household incomes of $10,000 or less, a quarter reported annual household incomes of $10,001 to $25,000, and 12% reported annual household incomes greater than $25,000. Over 40% had less than a high school education, 36% had a high school diploma or GED, 18% reported some college, and only 3% had at least a 4-year college degree. Two-thirds (67%) of parents lived with their spouse or partner.

Of the 46 children in the sample, 59% were 9–17 years old (“children”) and 41% were 18 years of age or older (“adult children”). The majority of children were male (63%), and the majority of adult children were female (58%). Children’s mean age was 14 years (SD = 2; range = 9–17), and adult children’s mean age was 22 years (SD = 4; range = 18–30). Caregivers were the infected parent’s spouse or partner (73%; 6 females and 5 males), mother (13%), and friend (13%). Caregivers’ mean age was 46 years (SD = 11; range = 35–79).

We asked parents to recall everything that happened when they disclosed their HIV status to their children, in addition to other topics related to being an HIV positive parent [24, 25]. Interviewers asked parents follow-up questions and prompted them to remember and describe others’ involvement, their decision to disclose, what they had said, and advice they would give other parents. Interviewers first asked children “ice-breaker” questions about their parents’ health in general. Once children began to discuss their parents’ infection voluntarily, interviewers asked children to recall the disclosure event, their thoughts during disclosure, and disclosure to siblings. These procedures had been used previously with children of HIV-infected parents and found to be comprehensible and to elicit useful information [19]. Interviewers asked caregivers to recount their involvement in the disclosure. Interviews were audio-recorded and lasted 60–90 min for adults and 30–60 min for children.

We analyzed the interview data using two approaches. First, we conducted an across-interview team-based thematic analysis of interview transcripts, managing texts with ATLAS/ti [26]. A project team member identified themes relevant to the disclosure process from the scientific literature and from a close reading of a sample of interview transcripts, and developed a codebook based on standard inductive and deductive techniques [27]. The first two authors trained two coders with extensive experience coding this data set to follow the inclusion and exclusion criteria and to identify exemplary and atypical examples of each theme. These coders independently read all transcripts and marked appropriate text relating to these themes. The lead author identified subthemes within each topical area by sorting all instances of marked texts into piles based on similarities [28, 29]. Each pile (i.e., subtheme) was labeled and added to the project codebook, which included sample quotations that exemplified each theme (see Tables 1, 2, 3, and 4 for exemplary quotations). To ensure that our codebook served as a reliable instrument for identifying and describing subthemes, two research team members sorted the original marked texts into subthemes [26, 30]. We assessed intra-coder reliability with Cohen’s kappa and found high agreement among coders (kappa range = 0.75–1.0; mean = 0.89) (Tables 1, 2, 3, and 4) [31]. The only code with a kappa below 0.80 was the code with by far the largest number of quotes (fear, with 141 quotes compared to 12–40 quotes for other codes) and required the most amount of interpretation of indirect descriptions of fearfulness.Table 1

Preparation for disclosure

Our second analytic approach was to treat each household as a case and to conduct a within-household analysis of the descriptions of the disclosure process. We read through each interview conducted for each household, wrote a summary of the household experiences on the themes identified in the across-interview analysis, and identified similarities and/or dissimilarities in the descriptions of the disclosure experience from the multiple perspectives in the household. Two coders rated each household as either “yes” or “no” for the following codes: parent mentioned a positive disclosure experience (kappa = 0.78); parent mentioned a negative disclosure experience (kappa = 0.80); child(ren) mentioned a positive disclosure experience (kappa = 0.74); child(ren) mentioned a negative disclosure experience (kappa = 0.85). We then compared these codes within households to assess the extent of intra-familial agreement regarding positive and negative aspects of the disclosure process.

Two central themes emerged in the across-interview analysis: preparation for disclosure and perceived outcomes of disclosure. We divided the preparation theme into two sub-themes: planned and unplanned disclosure. We divided the perceived outcomes theme into perceived positive reactions, perceived negative reactions, and perceived absence/mitigation of negative reactions based on how the respondents characterized their experiences.

Family members (parents, children, caregivers) we interviewed often characterized the disclosure experience positively and/or negatively. Their reactions were often mixed, with both positive and negative aspects. Only a few households contained no mention of any positive result, and only one household had no negative characterizations of the disclosure experience. In addition to these positive and negative reactions, some families described disclosure events that were not positive, but were not as bad as parents anticipated.

Table 5 summarizes the distribution of perceived positive and negative experiences within the 33 households for parents and children and indicates the level of agreement between children and parents within households. In only 11 households, parents and children completely matched in their description of the disclosure experience as positive and/or negative. Parents and children more frequently disagreed about whether the disclosure process was negative or positive. In 11 households, children described the disclosure experience negatively but parents did not. For example, in one household, the children talked about the lasting emotional toll that disclosure had on them. One 18-year-old son, who was around 6 years old when he found out his mom was HIV positive, discussed the ongoing worries he has had about his mom’s death. He said, “Well, I always thought that…one day I would wake up and see her dead…you know you have to cherish every moment.” The 24-year-old daughter, who was 12 when she found out, described the enduring disruption in her life after disclosure: “I always felt extremely unstable…and always…never secure.” On the other hand, their mother did not describe any enduring effects that disclosure had on her children. She described the immediate crying that her son did when she told him about being HIV positive but, on the whole, her description of the disclosure experience was primarily positive. When asked if she was happy that she disclosed to her children, she said, “I’m glad because I’m relieved ‘cause it was like keeping a big secret…from them…it’s better for them to know because then they know exactly what’s happening.”Table 5

Comparison of overall household perceived experiences of disclosure (N = 33, average kappa = 0.79)

On the other hand, there were 8 households in which children described the experience as positive but parents did not describe it positively. For example, in a one household with an unintentional disclosure, the HIV positive father described disclosure in exclusively negative terms. He discussed not wanting to disclose to his children in order to protect them: “I didn’t want to tell them about it when they’re too young because of ignorance about their friends’ parents.” He said that he worried that, once his children’s friends’ parents found out, they would not let their children play with the children of someone who had HIV. On the other hand, his sons did not describe the effects of disclosure as being particularly negative. In fact, they described how disclosure helped them improve their relationship with their father. One of his sons described being happy to know about his father’s illness (“It’s always good to know how someone you love is doing”) and also described his motivation to help his father (“I guess just to try to make him feel better”).

Our sample was chosen to maximize the potential variation in types of families (with younger, older, and adult children) in order to explore and describe the range of possible experiences and uncover factors that have not been adequately addressed in previous research. The wide range of ages of children we interviewed aided our goal of uncovering the range of experiences by mixing in older children, who can articulate their experiences better than younger children but might experience memory bias, with younger children, who can recount recent disclosure experiences but can not describe long-term effects. Nevertheless, the study’s high attrition/refusal-rate, non-random sampling strategy and exploratory, qualitative data collection limit the generalizability of these findings. We cannot generalize to families of parents who have never disclosed to their children or determine whether respondents’ descriptions of disclosure are biased by denial, rationalization, or a desire to provide socially desirable answers. Also, we cannot determine if disclosure experiences characterized by respondents as positive or negative were actually positive or negative for them in the long or short term. For example, children who described their increased household responsibilities positively may experience problems in the long term. Some recent studies of children with parents suffering from chronic illness have shown a negative effect of age-inappropriate caretaking responsibilities on children, while other studies show long term positive effects [35, 36].

Although a qualitative study is by design generally not used for hypothesis testing, these findings can be valuable in generating hypotheses for studies using methods better suited for hypothesis testing. For example, future studies can explore the relationship between unintended disclosure and child coping outcomes and test if unintended disclosure is more likely to result in negative outcomes than planned disclosure. Additional studies are also necessary to test how much the age of the child and the characteristics of the relationship between the child and parent before disclosure (e.g., how frequently they had contact) affect the outcome of disclosure. Additional research can help understand the intra-familial dynamics in communicating about parental HIV infection to determine which type of dynamic best promotes coping or to test if greater agreement between parents and children about the effects of disclosure is related to lower stress. These associations are only suggested by the findings of this study and they will have to be tested in additional studies with appropriate research designs. For example, a prospective research design that followed households over time starting before disclosure and continuing for many years after disclosure could measure the effect of child age and parent–child relationship characteristics at time of disclosure. Such a design would also allow for a better understanding of the effect of age and parent–child relationship on retrospective accounts of disclosure.