Findings from our study suggest that infant feeding patterns among women with HIV have changed, following changes in infant feeding guidelines. A greater proportion of postpartum women with HIV, than reported previously in the South African literature, reported exclusive breastfeeding, and the majority of pregnant women with HIV reported an intention to breastfeed [6]. HIV infection was, however, still a significant factor associated with breastfeeding intentions. Pregnant women who expressed an intention to exclusively breastfeed, compared with other feeding methods, had 3.6 times increased adjusted odds of not having HIV. For all groups of women in the study, healthcare facilities were reported to be the main source of information on infant feeding, and influenced feeding intentions and practices. There were however important knowledge gaps on safe infant feeding practices among both pregnant and postpartum women. Women with HIV had higher knowledge scores than women without HIV, particularly for infant feeding in the context of HIV infection. The perception expressed by most women in the study was that it was not easy to exclusively breastfeed for six months, and that cultural factors and influence from elders in families prevented most mothers from exclusively breastfeeding. There was also a perception in about a third of the participants; those women who elect to formula feed will be perceived as having HIV. This perception on the association of formula feeding with having HIV has been reported previously, and was common when infant formula was available for free in South Africa, for women with HIV who elected to formula feed [22, 23]. Despite this perception, a number of pregnant women with HIV in our study reported an intension to formula feed because of fears of infecting their infants.
We found high rates of reported exclusive breastfeeding among women with and without HIV. In a study done in Malawi, Kafulafula et al. [20] found higher parity, previous positive experience with breastfeeding and positive beliefs about exclusive breastfeeding all associated with a greater intention to exclusively breastfeed among women with HIV. In our study, similar to findings from the Malawian study, prior breastfeeding experience was associated with an intention to exclusively breastfeed. Although not significant, higher knowledge scores on safe infant feeding practices were also linked to a greater probability of an intention for breastfeed. Having HIV infection was associated with decreased adjusted odds of an intention to breastfeed, with a proportion of women with HIV reporting an intention to formula feed because of fears of infecting their infants. Similar to other studies, higher levels of maternal education and discussing infant feeding intentions with others were associated with decreased adjusted odds of an intention to exclusively breastfeed [8, 20]. Zulliger et al. [8] found that women with HIV, with lower levels of education, were more likely to express an intention to breastfeed. Women with higher levels of education are likely to be employed and hence be able to afford infant formula [20]. Disclosure of HIV status has been associated with decreased intended duration of breastfeeding among women with HIV, perhaps because women fear disapproval of extended breastfeeding with the risks of HIV transmission, but this was not significant in our study [20].
Consistent with our study findings, greater knowledge of exclusive breastfeeding has been associated with increased likelihood of exclusive breastfeeding, while maternal employment and infant age have been associated with a decreased practice of exclusive breastfeeding [19, 22]. Studies have shown that women who are employed are more likely not to breastfeed or wean early, and that this is related to the need to return to work after giving birth, the demands of being employed and also the practicalities of breastfeeding while at work [8, 24, 25]. Younger infants are likely to be exclusively breastfed, and this could be related to the perception that breast milk only is inadequate for the older infant [19]. Contrary to the finding on intention to breastfeed, HIV infection was positively associated with exclusive breastfeeding among postpartum women, but the association was weak. A South African cohort study also found that women with HIV were more likely to exclusively breastfeed than women without HIV [7]. This could be a reflection of the misconception that only women with HIV need to practice exclusive breastfeeding, to prevent HIV transmission, and is optional for women without HIV [26]. In most sub-Saharan countries, including South Africa, mixed feeding is the norm and adherence to exclusive breastfeeding raises suspicions of HIV infection, and to avoid the stigma, women with and without HIV may elect to ‘mix feed’ [26–28]. This is somewhat a contradiction given that, as discussed earlier, there is a perception that formula feeding is also associated with having HIV [22, 23]. Availability of male partner support is one of the factors found to mitigate the stigma and encourage exclusive breastfeeding in women with and without HIV [24]. Consistent with this finding, we found not being married and not cohabiting associated with decreased adjusted odds of exclusive breastfeeding. Several other factors contribute to suboptimal infant feeding practices, and include previous experience with infant feeding; perceived inadequacy of exclusive breastfeeding; family and cultural influences; and a need to return to work [8, 24, 27, 29]. Poor knowledge in mothers, as well as inconsistent and inadequate infant feeding counselling by healthcare workers also contribute to unsafe infant feeding practices [8, 24, 27, 29].
It is well-established that mixed feeding is associated with a high risk of MTCT in the absence of either infant or maternal ARV prophylaxis [1]. While maternal ART is protective of breast feeding transmission, adherence to treatment throughout the breastfeeding period is essential for prevention of HIV transmission [30]. In a study done in Zambia, where women received continuous ART starting in pregnancy and throughout breastfeeding, the majority of cases of MTCT occurred in infants older than six months, and adherence to maternal ART was suboptimal in all the cases [30].
Healthcare workers remain the main and important source of information on infant feeding, as we found in our study [27, 31, 32]. The messages need to be consistent as rapidly changing guidelines on infant feeding in the context of HIV infection have led to confusion among both healthcare providers and women accessing care [27, 31, 33]. Data from an established Option B+ PMTCT programme in Malawi however highlighted that most infant feeding counselling still occurs during pregnancy and delivery, with minimal counselling during the postpartum period [32]. Our finding of better knowledge among women with HIV is consistent with findings from other studies as there has been a greater emphasis on counselling and educating women with HIV on safe infant feeding practices and exclusive breastfeeding [26]. A perception has been created that women without HIV do not have to adhere to safe feeding practices, and that exclusive breastfeeding is a recommendation only for women with HIV, for prevention of HIV transmission [26]. This is of concern, as poor infant feeding practices are associated with increased infant morbidity and mortality, independent of HIV infection [24, 34]. Hence safe infant feeding messaging needs to target both women with and without HIV, and continue throughout pregnancy and the postpartum period.
Reported disclosure rates, especially to partners, were high among pregnant and postpartum women with HIV in our study. Disclosure of HIV status to partners, and to family members in those who live in extended families, is associated with adherence to PMTCT interventions, including safe infant feeding practices [27, 33]. Cultural factors and influence from elders have been identified as important factors that impact negatively on breastfeeding practices, consistent with perceptions expressed by the majority of participants in our study [35]. Cultural barriers to exclusive breastfeeding need to be addressed and the interventions must be specific to the local context [35].
There are several limitations to our study. It would have been more desirable to do a longitudinal study to compare intentions and actual postpartum practices in the same group of women, to determine how intentions translate to practice. The study design used was the most efficient use of available resources, and there are several challenges with mobility and loss to follow-up among postpartum women [28]. There could also have been a social desirability bias in reporting of infant feeding intentions and practices, and because the study was quantitative, we did not explore in detail the motivation behind the intentions and practices. Also, undertaking the interviews in the healthcare facilities might have been perceived as a test of the women’s uptake of information given in the facilities rather than their true intentions and practices. Social desirability could have also influenced perceptions expressed by the study participants. The other limitation is that the questionnaire was not translated into the local vernacular languages, but steps were undertaken to minimise any measurement bias by training the interviewers and standardising the translation of technical terms. While the questionnaire used in our survey was not formally evaluated for validity and reliability, similar questions to assess knowledge and perceptions have been used in several infant feeding studies done sub-Saharan Africa, as discussed in the methods section. The HIV prevalence among participants in our study was higher than that reported in the PMTCT programme data and in national surveys, and hence there may have been an overrepresentation of women with HIV. It is likely that women with HIV visit healthcare facilities more frequently for follow-up and to access treatment.