|Home | About | Journals | Submit | Contact Us | Français|
Our objective was to describe the association that childcare burden, household composition, and health care utilization have with adherence to highly active antiretroviral therapy (HAART) among women in the United States. The primary outcome was 95% or more adherence to HAART evaluated at 10,916 semiannual visits between October 1998 and March 2006 among 1419 HIV-infected participants enrolled in the Women's Interagency HIV Study. HAART adherence levels of 95% or more were reported at 76% of the semiannual visits. At only 4% of the person-visits did women report either quite a bit or extreme difficulty in caring for child; at 52% of the person-visits women reported at least one child 18 years of age or older living in the household. We found a one-unit increase in the difficulty in caring for children (childcare burden was assessed on a 5-point scale: not difficult  to extremely difficult ) was associated with a 6% decreased odds of 95% or more HAART adherence (adjusted odds ratio [OR]=0.94; p=0.07). Each additional child 18 years of age or less living in the household was associated with an 8% decreased odds of 95% or more adherence (adjusted OR=0.92, p=0.03). Both the number and type of adult living in the household, as well as health care utilization were not associated with HAART adherence. Greater child care burden and number of children 18 years old or younger living in household were both inversely associated with HAART adherence. Assessing patients' difficulties in caring for children and household composition are important factors to consider when addressing adherence to HAART.
Until the mid-1990s, acute and end-of-life care were the primary health models used for managing opportunistic infections and cancers resulting from HIV infection. Since the introduction of highly active antiretroviral therapy (HAART) in 1995, HIV morbidity and mortality have declined dramatically.1–4 A critical component of the effectiveness of HAART is patient adherence. While recent publications have demonstrated high rates of viral suppression with HAART adherence levels as low as 54% among individuals treated with non-nucleoside reverse transcriptase inhibitor (NNRTI)-based HAART regimens, and as low as 24% among individuals treated with boosted protease inhibitor (PI)-based HAART, others have demonstrated that HAART adherence rates of at least 95% are associated with an increased likelihood of viral suppression.5–9
Researchers have examined how family stressors impact the burden of physical illness and psychological stress among HIV-infected women.10–12 However, less research has examined how life stressors affect adherence to HIV treatment regimens.13,14 Previous research has demonstrated that both social support and a stable relationship improve medication adherence and that living alone is associated with decreased ability to maintain viral suppression.15–17 Previously, we along with others have found that HAART adherence was inversely associated with caring for children.18,19 However, our previous research did not examine the impact that additional stressors such as child care burden, household composition and healthcare utilization had with adherence to HAART. This analysis helps clarify these issues and provide robustness to our previous analysis.
In order to further understand the role these family situations and healthcare utilization play on adherence we examine them in this analysis, using data from the Women's Interagency HIV Study (WIHS).
A total of 2791 HIV-infected WIHS women were enrolled in either 1994–1995 or 2001–2002 from six United States sites: New York (Bronx and Brooklyn sites), Chicago, San Francisco, Los Angeles, and Washington, D.C. Every 6 months, participants complete a physical examination and interviewer-administered questionnaire, which collects data on sociodemographic variables, health care utilization, HIV disease characteristics, and antiretroviral therapy (ARV) use. Institutional review boards at each site approved study protocols and consent forms, and each participant provided written informed consent.
At each semiannual visit, participants are shown photo-medication cards of all available ARV medications and asked about ARV medication used since their previous visit. The WIHS uses a standard definition of HAART, adapted from the Department of Health and Human Services/Kaiser Panel guidelines.20 All non-HAART combination therapy regimens are classified as combination therapy; use of a single nucleoside reverse transcriptase inhibitor (NRTI), protease inhibitor (PI), or non-nucleoside reverse transcriptase inhibitor (NNRTI) is classified as monotherapy.
Beginning in October 1998, participants were asked at each semiannual visit to indicate how often they had taken their ARV medications as prescribed in the previous 6 months. Participants categorized their level of adherence into one of five categories: 100% of the time, 95%–99% of the time, 75%–94% of the time, less than 75% of the time, and have not taken any of prescribed medications. For analysis, participants were categorized dichotomously by whether they reported taking antiretroviral medications as prescribed 95% or more of the time.18,21 Only visits at which a participant reported using a HAART regimen since their last visit were included in analyses; all visits at which either no therapy, monotherapy, or non-HAART combination therapy was reported were excluded.
For child care burden, participants were first asked whether they had any living children less than 18 years of age and if so whether or not they knew the health status of their children. If both of these questions were answered affirmatively the participant then rated the difficulty in caring for their children during the past 6 months (assessed on a 5-point scale: 1=“not at all”; 2=“a little bit”; 3=“some/moderately”; 4=“quite a bit”; 5=“a lot/extremely”). For study visits in which a participant reported that they had no living children less than 18 years of age or that they did not know the health status of their children, a rating of “not at all” was assigned to minimize the occurrence of missing data. In the primary analyses, child care burden were treated as a categorical exposure with the odds of 95% or greater HAART adherence in categories 2 (a little bit) through 5 (a lot/extremely difficult) each compared to the odds of 95% or greater HAART adherence among those in category 1 (not at all difficult). To check for a dose-response relationship between child care burden and adherence to HAART, the analyses were repeated with child care burden treated as a discrete value from 1 to 5 in which we estimated how the odds of 95% or greater HAART adherence changed for each additional unit change in child care burden.
Participants were asked how many people they lived with and the relationship (i.e., husband or male sex partner, other adult relatives, other adult nonrelatives) of all persons reported living in the household. Participants were also queried on the number of individuals with whom they currently lived that were 18 years of age or younger. Only study visits in which the participant reported living in a house or apartment (94% of all person-visits) were included in analyses. In the primary analyses, the number of adults (0, 1, and ≥2) and children (0, 1, 2, ≥3) who lived in the household were treated as categorical variables. To check for dose-response relationships the analyses were repeated with these primary exposures treated as discrete, with two or more adults coded as 2 and three or more children coded as 3 in order to avoid the influence of any outlying values. The change in the odds of 95% or greater HAART adherence for each additional adult (or child) in the household was estimated.
The WIHS collects data on child care burden at even-numbered study visits and data on household composition at odd-numbered visits. Since HAART adherence is ascertained at each study visit in the WIHS and we wanted to include both child care burden and household composition exposures in the same analysis we carried forward the child care burden data to the next odd-numbered study visit (e.g., data from visit 12 were carried forward to visit 13) as well as the household composition data to the next even-numbered study visit (e.g., data from visit 13 were carried forward to visit 14) to allow both primary exposures to be defined at the same visit. In carrying forward or imputing “missing” child care burden data in we are assuming that had the WIHS collected data on child care burden at all study visits the data from a given even-numbered visit it would be the same as the data collected at the following odd-numbered visit.
Health care utilization data were collected at every visit. Specifically, participants were asked if they had seen a health care provider (e.g., any doctor, nurse practitioner, or physician assistant) since the previous visit, had a social services agency assist with food, meals on wheels, food pantries, or arranged to have groceries delivered to home since the previous visit, had been seen by a social worker or a case manager to help obtain services since the previous visit.
The unit of analysis was a study visit with adherence levels defined at the current visit and the exposure defined at the preceding visit to ensure that the exposure preceded the outcome. The odds of 95% or greater adherence were compared between different primary exposure groups over time using logistic regression models with generalized estimating equations to account for the statistical dependence incurred by repeated measures of adherence over time on the same individual.22
In addition to the sequence of the univariate analyses of each primary exposure and adherence, two types of adjusted analyses were conducted to describe the association that child care burden, household composition, and health care utilization had with HAART adherence. One type of analyses consisted of three domain-specific models (domains: child care burden, household consumption, health care utilization) with each model including the primary exposures from a particular domain and possible confounders (listed below). The second type consisted of one model that contained the participant rating of the difficulty in caring for her children, the number of adults in household, the number of children 18 years of age or younger in household, and whether or not husband or male sex partner lived in household, along with possible confounders (listed below).
All adjusted logistic regression models included the following variables as possible confounders: study site, race/ethnicity (African American, Hispanic, white, and other), enrollment period (1994–1995, 2001–2002), education (high school graduate or higher, less than high school graduate) and time-varying values of age (per 10 years), income (<$6000, $6001–12,000, $12,001–18,000, $18,001–30,000, >$30,000), marijuana/hash use, cocaine, crack-cocaine, or heroin use (all self-report since last visit), depression (Center for Epidemiologic Studies Depression Scale [CESD]≥16, CESD<16), health-related quality-of-life summary score (per 10 points) (modified version of SF-36), health insurance status (presence or absence), CD4 cell count (per 100cells/mm3) and whether criteria for clinical AIDS were met.
Of the 2791 HIV-infected WIHS participants, 2199 (79%) had a visit between October 1998 and March 2006, and 1626 (74%) reported HAART use and had adherence data available for at least two visits. One hundred eighty-eight (12%) participants who reported being pregnant during at least one visit were excluded from all analyses since adherence levels have been shown to vary greatly during and around pregnancy.23–25 Data from an additional 19 participants were excluded because all primary exposure data were missing immediately prior to each of the visits with adherence data, resulting in a final study population of 1419 women who contributed a total of 10,916 on-HAART study visits to the analysis.
Table 1 provides descriptive statistics on the study population of 1,419 HIV-infected WIHS participants at the first semiannual visit with adherence data. The majority of our population identified themselves as being African American or having Hispanic ethnicity. The median age was 39.2 years, 14% reported a family income greater than $30,000 per year, and 86% reported currently having health insurance. Thirty-seven percent had at least one clinical AIDS diagnosis, while 45% were depressed (CESD≥16). Seventeen percent reported marijuana/hash usage and 13% use of cocaine, crack, or heroine.
Table 2 shows that women reporting “a little bit” (adjusted OR=0.83; p=0.05) or “some/moderate difficulty” (adjusted OR=0.77; p=0.06) in caring for children were less likely to have ≥95% adherence to HAART than women who claimed that caring for children was not at all difficult. When this exposure was treated as continuous, the adjusted odds of ≥95% adherence to HAART decreased 6% (OR=0.94; p=0.07) for each one-unit increase in the difficulty in caring for children.
Table 2 shows that women who reported living in the household with one adult were more likely to report 95% or more adherence to HAART than women who reported living with no adults in both univariate (p=0.14) and multivariate analyses (p=0.10) although the differences did not reach conventional statistical significance. This association was not present when the number of adults was two or more. In the model treating the number of adults as a continuous exposure, the unadjusted and adjusted odds ratios were 0.95 (p=0.20) and 0.98 (p=0.87), respectively (data not shown).
Conversely, in the adjusted analyses, the presence of children 18 years of age or younger living in the household resulted in decreased adherence (OR=0.93, OR=0.78, OR=0.84) among those women reporting 1, 2, or 3 or more children, respectively, compared to those women reporting no children). In analyses treating the number of children as a continuous exposure, the number of children 18 years of age or younger who living in the household were negatively associated with 95% or greater adherence to HAART (adjusted odds of ≥95% adherence decreased by 7% [p=0.03] for each additional child). After adjusting for other household composition exposures and confounders those reporting either a husband or male sexual partner were 20% less likely (OR=0.80; p=0.09) to report 95% or greater adherence to HAART. Participants who reported living with an adult relative or non-relative were just as likely to report 95% or greater adherence to HAART as those living without an adult relative.
Women who reported receiving assistance from social services were significantly less likely than women who reported not receiving assistance to have 95% or greater adherence in a univariate analysis (OR=0.84) but the effect was attenuated after adjustment for other characteristics (Table 2). Women who reported being seen by a social worker or case manager and those seeing their health care provider since the previous visit had similar HAART adherence levels as women who did not report seeing a social worker or physician.
Our final adjusted analysis (last column of Table 2) contained the participant rating of the difficulty in caring for children in the past 6 months, the number of adults living in the household, number of children 18 years of age or younger living in the household, whether or not husband or male sexual partner lived in the household, along with other confounders. The odds ratios and confidence intervals for these primary exposures were very similar to those reported in the domain-specific analyses. Treating each primary exposure as continuous, each additional child 18 or younger was associated with a 7% decrease (p=0.04) in the odds of 95% or greater adherence, each additional adult living in household was associated with 5% decrease (p=0.28), and each one-unit increase in the difficulty of caring for children was associated with a 4% (p=0.25) decrease (results not shown).
Among those who reported one child 18 years of age or younger living in the household (Fig. 1A), the percentage of women who reported no difficulty in caring for children who also reported less than 95% HAART adherence ranged from 75% to 82% (average=78%); the percentage of women who reported a little bit of difficulty in caring for children who also reported 95% or greater HAART adherence ranged from 62% to 84% (average=73%); and the percentage of women who reported at least some/moderate difficulty in caring for children who also reported less than 95% HAART adherence ranged from 64% to 85% (average=72%).
Among those who reported two children 18 years of age or younger living in the household (Fig. 1B), the percentage of women who reported no difficulty in caring for children who also reported less than 95% HAART adherence ranged from 69% to 81% (average=75%); the percentage of women who reported a little bit of difficulty in caring for children who also reported 95% or greater HAART adherence ranged from 51% to 77% (average=68%); and the percentage of women who reported at least some/moderate difficulty in caring for children who also reported less than 95% HAART adherence ranged from 53% to 83% (average=67%).
Among those who reported at least three children 18 years of age or younger living in the household (panel C the Figure), the percentage of women who reported no difficulty in caring for children who also reported greater than 95% HAART adherence ranged from 69% to 76% (average=73%); the percentage of women who reported a little bit of difficulty in caring for children who also reported 95% or greater HAART adherence ranged from 63% to 81% (average=71%); and the percentage of women who reported at least some/moderate difficulty in caring for children who also reported greater than 95% HAART adherence ranged from 45% to 73% (average=62%).
In summary, in each strata of children (i.e., in each panel of Fig. 1) those with more child care burden are less likely to have greater than 95% HAART adherence on average. Furthermore, among those reporting no difficulty or at least some/moderate severe difficulty in care for children, the proportion with 95% or greater HAART adherence decreases as the number of children increases (i.e., moving from Fig. 1A to 1C).
We have previously shown that adherence to HAART is inversely associated with the number of children living in the household.18 The strengths of this analysis compared to our previous findings are that we now adjusted for and examined different stressors that may affect a women's adherence. Specifically, we have now adjusted for the household composition, child care burden, health care utilization, and social services. It appears that women with HIV and children under 18 years old living in their household experience increased burdens compared to HIV-infected women without children, which subsequently results in poorer adherence to HAART.
Previous studies have shown life stressors impact HIV compliance.26,27 This has resulted in attempts to implement stress management interventions to improve HIV adherence.28,29 We believed that stress was likely the reason we have previously found mothers have lower adherence. However, other authors have suggested that not stress but a women's predilection to place their child's needs above their own results in lower adherence.19,30 Our findings are consistent with this latter theory, as previously explained the best indicator for adherence was the presence of children not other stressors or household members. There appears to be something about caring for children beyond the specific stress that results in lower adherence rates. Based on our previous research we hypothesized that a mothers' psychosocial well-being, having a significant other, additional housemates, receiving social services, and access to community based resources would result in higher adherence rates. However, these a priori hypotheses were not supported by our analyses. There is therefore something specifically about children that seems to be related to adherence, not simply the number of household members since neither the number nor type of adult living in the household was associated with adherence.
Our study has several important limitations. WIHS primarily collects data on the women themselves and information about the children they report living with them is limited as is more robust information about the specific details of the social services women receive. Thus, data on the age, HIV, and general health status of the children, if the participants have disclosed their HIV status to household members and the specific reasons women were receiving social services are lacking. This is all potentially important information, as for example, mothers have been reported to approach their HIV status differently according to the age of their children.31 Also, since adherence was self-reported, there is a possibility of misclassification whereby HIV-infected participants may tend to over-report adherence. There is no reason to believe that the rate of misclassification would be any higher among women without children than among women with children (i.e., differential misclassification) and any nondifferential misclassification would bias our estimates of association towards the null.32
Additionally, previous WIHS research has shown self-reported adherence to be consistent with objective measures, such as CD4 count, HIV viral load and self report of physical functioning.33 It is also possible that participants with less stress participate more in study visits, however our overall rates of retention are greater than 90%.
In two different analyses, this one being more robust, we have shown that the more children a women lives with, the worse her adherence to HAART is likely to be. The importance of family-centered care on children's health has long been recognized and examined on multiple levels.34 Conversely, the role the child plays in the health of parents has not been as readily explored and we have shown that this role may potentially be very important. Further research into this relationship should help elucidate this issue, alert patients and health care providers to potential problems and result in approaches to address decreased adherence due to child care burden.
Data in this manuscript were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, New York (Howard Minkoff); Washington D.C. Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Stephen Gange). The WIHS is funded by the National Institute of Allergy and Infectious Diseases with supplemental funding from the National Cancer Institute, the National Institute on Drug Abuse (UO1-AI-35004, UO1-AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, and UO1-AI-42590). Funding is also provided by the National Institute of Child Health and Human Development (UO1-HD-32632) and the National Center for Research Resources (MO1-RR-00071, MO1-RR-00079, MO1-RR-00083).
No competing financial interests exist.