Self-reported adherence was high and comparable to other studies in Brazil [8, 36, 51]. We found that women were negatively associated with adherence to ARVs in this setting. While SES factors have not been consistently associated with adherence, we found a high asset index to be significantly associated with adherence. In addition social support was found to be significantly associated with the outcome, which is consistent with research among PLWH in other settings. Sociodemographic factors have generally not been consistently associated with adherence; we only found number of children to be significant which will be discussed in relation to gender. These factors and their relationships, especially pertaining to the history of the epidemic in Brazil and particularities of the research setting, will be discussed.
Women are especially vulnerable to HIV and negative HIV-related outcomes such as biological factors due to toxicity and social factors due to gender inequality [52–54]. A recent study examined factors associated with ARV discontinuation or modification and found women had a higher hazard for short-term toxicity . In addition, women were found to have more severe side-effects than men, leading to higher rates of ARV discontinuation among women. Furthermore, there is a “feminization” of the epidemic with the male to female ratio decreasing from 5.9:1 in 1989 to 1.4:1 in 2006, highlighting the importance of increasing focus of research and interventions on women in Brazil . Women might also experience challenges in accessing ARVs due to lack of knowledge, stigma and discrimination [57, 58]. The history of activism in the Brazilian AIDS community started within the gay community and focused on MSM early on in the epidemic . Women have not had the same amount of support from NGOs and most of the attention in Brazil to date among women has been on prevention efforts, which has also been the case globally [60, 61].
We found having one child was associated with being adherent, versus not having any children, but the same did not apply for those having two or more children. A study in the US found that the number of children living at home is associated with nonadherence, and caretaking of children is often the responsibility of women . Upon further analysis, adjusting for additional life stressors, the study still found the presence of children and not other stressors associated with nonadherence . It appears that women place others’ needs, especially their children’s, ahead of their own [62, 63]. While having at least one child in our study is associated with adherence, it might be a proxy for something else, such as stability in their life, but more than one child may create an extra burden. The role of caretaking and number of children warrants further research with regard to their role on adherence to ARVs among women in Brazil.
The pattern of the Brazilian HIV epidemic seems to initially affect higher socioeconomic levels and then disseminate into poorer populations [1, 64]. This can be seen in the difference between the southeastern region where the epidemic began and is now among the middle and lower class versus the northeastern region where the epidemic is more recent and is concentrated in higher socioeconomic populations [1, 65]. Even within our lower income population, respondents with higher levels of assets were significantly more likely to be adherent. This signifies the need to understand how wealth plays a role in adherence even in the context of free health services and free medication. Socioeconomic factors often serve as a proxy for other barriers to adherence such as transportation costs, lack of time, difficulty obtaining health information, and discrimination . Poverty increases vulnerability to HIV infection and can also make it harder to cope with a complex chronic infection .
Social support is a multidimensional construct that has been defined and measured in various ways, focusing on sources of support and satisfaction with the support received . Our measure was heavily focused on the provision of tangible material support such as help with obtaining medicines, performing chores, or going to the doctor. This is key to understanding the kinds of support important in addressing adherence. Future research should examine potential differences in the kinds of support women receive versus men since women are often the caretakers and carry a larger burden of household responsibilities, while men might have more economic burdens. Again, since the epidemic has reached women more recently, their needs have not received the same attention as other populations. They may not have the same extra-familial support networks to go to for support as MSM have within AIDS activist organizations within a setting such as Rio de Janeiro. In addition, fear of stigma may inhibit accessing support from their networks [57, 68, 69].
There was a linear trend between social support and assets, where those with more assets also had more social support. It is possible that participants with higher levels of assets also have family and friends with higher levels of assets; better enabling them to obtain the necessary support when needed. It is possible that those who suffer material deprivation also do not have the ability to access the necessary kinds of support from their social surroundings due to the lack of available resources. This can be potentially explained through the components of social capital which can be seen as horizontal or vertical, where linkages with people who are of similar status or background is called bonding social capital (horizontal) and people who are of a different background or identity is called bridging social capital (vertical) . Further research in the realm of social capital, where resources available to an individual come from being a part of social networks, could help in understanding the relationship between assets and social support to develop care and support interventions among PLWH in lower income settings.
Anxiety and depression were not significant in the final model but the high prevalence among PLWH in this setting warrants further research. A recent study in Brazil using the same depression and anxiety measure but at a higher cutoff point to determine severe cases of depression and anxiety, found a high prevalence of anxiety (35.8%) and depressive (21.8%) symptoms among patients before initiating ARVS, which increased (51.5% and 40.6%, respectively) in the 175 day follow-up time period . In addition, they found severe anxiety was significantly associated with non-adherence but not severe depression. We found similar anxiety levels in our study but a much greater prevalence of depressive symptoms (70%). This may be due to the longer average treatment duration among our study population since depressive symptoms increased in the aforementioned study over time.
Adherence is a dynamic behavior but our analysis is only cross-sectional; therefore we are only capturing one moment in time hence we cannot establish temporality in our associations. While self-report tends to overestimate adherence levels, it is reliable and tends to indicate average adherence . The AGTC questionnaire has been tested in multiple settings and in Brazil as well. In addition, the self-report of adherence in this study was significantly correlated with the viral load data from their medical records. We attempted to decrease social desirability bias with a statement commenting on the difficulty of adherence before asking our questions. We also tried to limit recall bias by asking about adherence in the previous four days. The findings from this study come from regular clinics, where most Brazilians access care across the different regions, and not referral centers, where most of the studies in Brazil are conducted. Therefore these findings are more generalizable than other studies in Brazil but since Rio de Janeiro is the second largest city in Brazil, these findings should not be generalized to either poorer areas inside Brazil or low-income countries.