A qualitative investigation into knowledge, beliefs, and practices surrounding mastitis in sub-Saharan Africa: what implications for vertical transmission of HIV?

In sub-Saharan Africa (SSA), where breastfeeding constitutes the most common infant feeding practice [1–5], postnatal transmission of human immunodeficiency virus (HIV) through breastfeeding represents at least 24% (and possibly as much as 42%) of overall mother-to-child transmission (MTCT) of HIV [6]. The infants' risk of getting infected through breastfeeding appears to be highest in the first weeks of life and to be strongly associated with HIV viral load in breast milk [7]. In turn, recent studies have demonstrated that the presence of mastitis, an inflammatory process in the breast, is associated with an increase in HIV viral load in breast milk [8–12]. Therefore, mastitis is considered to be an important independent risk factor increasing HIV vertical transmission [13].

The literature suggests that mastitis represents just one of several pathologic states from which breastfeeding mothers may suffer. Medical problems linked to breastfeeding comprise a continuous spectrum of pathologic states ranging from breast engorgement due to reduced milk flow (stasis) to clinical mastitis (an inflammatory process in the breast producing localized tenderness, redness, and heat, together with systematic reactions of fever, malaise) [14], which may extend to breast abscess [13]. Sub-clinical mastitis has been identified by several researchers as a stage of the disease in which women do not complain about any subjective signs or symptoms except reduced milk flow but have a particular biochemical composition of breast milk [13]. In such cases, laboratory analysis of expressed breast milk revealed an elevated breast milk leukocyte count, an increase in the concentration of pro-inflammatory cytokines, and an increase in sodium or the sodium-potassium-ratio [13, 15–17].

The literature further suggests that the occurrence of mastitis and/or breast inflammations in general is common both in resource-rich and in resource-poor settings. Clinical mastitis was diagnosed in 10% of American women during the first three months of lactation [18] and in 17% of Australian women during the first six months of lactation [19]. In a recent survey in Turkey, 80 (71%) out of 112 lactating women reported having suffered from breast problems (engorged breast, tenderness, and pain) in the two months postpartum [20].

Accurate information on the overall prevalence of breast inflammations and in particular of mastitis in SSA is not available. The most reliable estimates are derived from measurements among HIV-infected women enrolled in prevention of mother-to-child transmission (PMTCT) programs. In Kenya, 11% of HIV-infected women were diagnosed with mastitis and 12% with breast abscess [11]. A study in Malawi indicated that approximately 27% of HIV-infected women had experienced at least one episode of sub-clinical mastitis, defined as elevated breast milk leukocyte count, in the first year postpartum [16]. In Tanzania, sub-clinical mastitis, defined as elevated milk sodium-to-potassium-ratio, was diagnosed among 13% and 11% of women respectively one month and three months postpartum [15].

Even if the natural history and the clinical importance of sub-clinical mastitis remain unclear [15], these findings yield important implications for the design of PMTCT programs. In SSA, exclusive breastfeeding (EBF) followed by early weaning has been identified as the preferred infant feeding practice [21]. Given the poor hygienic conditions and the price of milk substitutes in fact, EBF followed by early weaning is generally considered to bear greater benefits than costs [22–27]. Thus, given the increased risk of HIV transmission associated with the presence of breast inflammations and mastitis, prophylaxis, early diagnosis, and treatment of such conditions have been proposed as an additional PMTCT strategy in resource-limited settings [28].

In spite of the role played by mastitis in the transmission of HIV and in spite of the fact that the first studies conducted in SSA have indicated the prevalence to be quite high among HIV-infected women [11, 16], public health scientists have channelled little efforts towards understanding how women frame health problems related to breastfeeding and how they deal with them. With the exception of one ethnographic study by Alfieri and Taverne [29], little is known on local illness concepts and health care seeking behaviour in relation to breastfeeding problems in SSA. This study, conducted exclusively on two ethnic groups in West Africa, the Mossi and the Bobo Madare, revealed that women recognise as relevant to their everyday life health problems associated with breastfeeding. Women differentiate relatively simple problems linked solely to reduced milk production from more severe problems described as real breast pathologies. The study also revealed that women distinguish between natural and unnatural causes of disease and that they alternatively use traditional and modern medical remedies.

A comprehensive understanding of local illness concepts and health care seeking behaviour relevant to public health practice is still clearly missing, although essential to design effective, yet culturally sensitive PMTCT programs. This paper draws from the results of a qualitative study which preceded and informed the introduction of a PMTCT program in rural Burkina Faso. We aimed at: (a) documenting local illness concepts and the indigenous nosography of breast pathologies during lactation; (b) documenting women's health seeking behaviour; and (c) describing available remedies for prophylaxis and treatment.

The field work was conducted in the Nouna Health District (located approximately 300 km from the capital Ouagadougou) between December 2002 and March 2003 within the framework of a larger anthropological study exploring local beliefs and practices related to breastfeeding. The methodology has been described in detail elsewhere [30].

In brief, we collected data using both focus group discussions (FGD) and in-depth individual interviews with women who were purposively selected on the basis of their experience with breastfeeding [31]. Given our focus on mastitis and HIV, we included women who reported having experienced lactating problems at least once in their lifetime. Women to be interviewed were identified at informal women's gatherings by JH with the assistance of a key informant. In addition, we conducted in-depth individual interviews with local guérisseurs, i.e. indigenous health practitioners ranging from herbalists to diviner mediums. We continued data collection until we reached saturation and redundancy [32].

JH conducted all interviews personally with the assistance of two translators, one working in Djoula and one in Bwamu. Before proceeding with the interview, JH explained the purpose and relevance of the study and sought the women's explicit verbal consent. Information was solicited through a series of semi-structured open-ended questions. JH and TB developed the interview guides for both the individual interviews and the FGD. The interview guides touched on different aspects of the culture and practice of breastfeeding. With specific reference to the themes addressed in this publication, the interview guide explored how women and guérisseurs construct and define what constitutes a lactating problem and what prevention and treatment options they resort to in case of need.

All material was recorded, transcribed, and translated into French by trained translators. We carried out the analysis on the French text, translating into English only the material which appears in our publications. We analysed the data inductively. We applied analyst triangulation as two independent researchers, JH and MDA, read the material separately and only compared and converged findings at a later stage [31]. Afterwards, we discussed the interpretation and the policy relevance of the findings among all authors. The systematic comparison of findings across data sources, women and guérisseurs, and between the individual interviews and the FGD provided an additional source of triangulation [31].

The study was approved by the Ethics committee of the Faculty of Medicine of the University of Heidelberg, Heidelberg, Germany, and by the Nouna Ethics Committee, Nouna, Burkina Faso.

We interviewed 38 women, as the result of 32 individual interviews and 2 FGD, and 5 guérisseurs. In addition, JH was invited to attend a meeting held among guérisseurs and had the opportunity to pose additional questions on such occasion. The respondents, whose age ranged from 17 to 80, were representative of all local major ethnic groups in the area: Marka, Bwaba, Mossi, Peuhl, and Samo. The vast majority of women were uneducated, were married, and as a source of income, they engaged in small-scale commercial activities.

The presentation of the findings is organized in three sections. For each reported verbatim quotation, we indicate the respondent's age and ethnicity. We explicitly indicate when quotations report the guérisseurs' speech. In addition, we have included a case study, Salima's story [see additional file 1] with the aim of allowing the reader to gain a better understanding of how a typical woman in Nouna perceives and defines her lactating problem and how she is likely reach a decision regarding health care seeking. We wish to point out that Salima is a fictional name used to protect the identity of the woman originally reporting the story. Given that the woman was illiterate, consent to use the information she shared with us was obtained verbally.

This study provides an overview of local illness concepts and current health care seeking behaviour in relation to health problems associated with breastfeeding in rural Burkina Faso. Our aim has been that of informing the design of mastitis control measures. Controlling mastitis is in fact desirable both in its own right and in the light of its dangerous potential to increase HIV transmission among breastfed infants [8–13, 28].

The first element to deserve attention is that women recognise health problems associated with breastfeeding to be highly prevalent and to constitute a threat to the well-being of both mothers and infants. In addition, although women use a variety of local expressions to define the illness, their recognition of the pathological states associated with a health problem related to breastfeeding is very much in line with biomedical definitions. They identify the same set of states ranging from breast engorgement due to reduced milk flow (statis) to swelling of the mammary glands accompanied by an inflammation (mastitis) and at times developing into a painful breast abscess (infection) recognised by biomedicine. Like women elsewhere in the world, women in the Nouna Health District just do not recognise the existence of sub-clinical mastitis as this, as explained extensively in the introduction, has been defined only in terms of changes, e.g. in breast milk leukocyte numbers and/or in the concentration of pro-inflammatory cytokines, detected through laboratory analysis in the absence of any subjective recognition or symptoms of disease [13, 15–17].

In the light of these findings, the dearth of information on the prevalence of mastitis in SSA appears to be almost paradoxical and points at the existence of a large gap between what communities perceive as an important health problem and what clinical and public health specialists have identified as one. Only a handful of studies have attempted to measure the prevalence of mastitis in SSA, generating estimates that range from 10 to 30% [11, 15, 16]. In order to plan adequate public health interventions to address the problem both within and beyond the framework of PMTCT programs, precise estimates of prevalence based on larger population samples are urgently needed.

Furthermore, the fact that the disease is widely recognised and perceived to constitute an important threat to health indicates that there is a perceived need for treatment and that such need could be translated into actual demand should the necessary conditions, i.e. the provision of culturally sensitive low-cost easily-accessible health services, come into place [33–35]. Likewise, the overlap between the local identification of symptoms and the biomedical definition of mastitis offers health workers an initial "common ground" against which to set educational campaigns and public health efforts aimed at controlling the disease.

What may constitute a challenge to the effective implementation of programs aimed at controlling mastitis is the fact that women, supported by the opinion of local guérisseurs, frequently define their breast problems as the product of sorcery or of a supernatural source. Given that one's health care seeking behaviour is inevitably shaped by one's understanding of a disease [36], it does not appear surprising that once women perceive their illness to be "unnatural", they prefer to be treated by a healer. Our findings to this regard are in line with previous findings from SSA, concerning both specific problems related to the breast [29] and women's health care seeking behaviour more in general [37–39].

Bridging a link between biomedical services and traditional healers may serve as an instrument to reach women irrespective of their recognised aetiology of disease. Bridging the link between biomedical services and healers would in fact allow health providers to intervene at the point where traditional medicine is no longer sufficient to treat the disease and still to do so in the respect of local beliefs and practices. The fact that traditional healers also recognise a set of pathological states resembling those defined as relevant by biomedicine coupled with their explicit wish to collaborate with health professionals represent encouraging elements for the set up of a closer collaboration between the two health sectors. Collaborations between healers and biomedical professionals have already been widely implemented in the field of HIV prevention and care [39–43], and there is no reason to believe that they should not be successful also to control mastitis.

The collaboration between traditional healers and biomedical professionals is also essential to secure a community-based intervention to control mastitis. Given that in SSA many women do not come into contact with modern health facilities during their pregnancy [44, 45] and that many more women do not test for HIV even in the presence of a PMTCT program [46, 47], community-based rather than hospital-based interventions represent a more adequate means of achieving mastitis control and potentially reducing HIV transmission even among women who are unaware of their HIV status. The need for community-based rather than hospital-based interventions is further motivated by the fact that, as reported both by our findings and by prior literature [48], women often do not make their own decisions regarding what care to seek, but act under the influence of older family members and community leaders. Thus, controlling mastitis is only possible if the entire community is involved in the process of prevention and care.

While continued lactation has been observed both in our study area and elsewhere [29, 49, 50] to be the preferred strategy to outset the development of further breast complications, recent research has identified its dangerous potential to lead to an increase in HIV transmission [8–12]. It has therefore been suggested that HIV positive mothers with breast inflammations should be discouraged from feeding their infants from the affected breast during the period of inflammation [51]. Although recent evidence suggests that EBF may be preferable to substitute feeding even among HIV positive women [21], when suffering from a breast inflammation, HIV positive women should be provided with safe and affordable milk substitutes at least for the time required to treat their condition. Alternatively, due to the fact that the safe and affordable provision of milk substitutes is not always feasible in SSA [21], women could be advised to continue breastfeeding only with the unaffected breast. These strategies, however, have the potential to reach only women who have been identified as HIV positive. Thus, in the light of what stated earlier regarding the possibility that many HIV positive women remain undetected even in contexts where PMTCT programs are available, it is of extreme importance that above all research urgently focuses on developing specific interventions to reduce the incidence of breast inflammations at the community level and subsequently the risk of HIV transmission among breastfed infants.

In particular, specific efforts should be channelled towards assessing the impact of traditional practices, such as the ecrasement, and herbal remedies, such as the application of karité butter, on the prevention of severe breast problems and on HIV viral load in breast milk. To our knowledge in fact, there is no information available on the effect that traditional practices and African herbal remedies have on mastitis prevention and on viral load in breast milk. If research should show that traditional practices and remedies are effective in draining the breast and lowering HIV concentration in breast milk (thereby reducing the risk of HIV transmission), then such practices and remedies could be promoted within the framework of PMTCT programs. Alternatively, if research should show the opposite effect, educational campaigns would need to actively discourage the adoption of such practices and remedies. This is of great importance given that the traditional practices and remedies reported in our study have also been observed elsewhere [29], suggesting the existence of commonalities in approaches to prevent and treat mastitis across SSA.

Our study has shown that health problems related to breastfeeding are perceived to be very common in rural Burkina Faso. The spectrum of pathologic states described by the respondents resembles biomedical notions of disease in spite of the fact that the local aetiology differs substantially from the biomedical one. Further studies are needed both to provide an adequate epidemiological picture of mastitis in SSA and to explore the effect of local remedies and practices on the development of severe breast inflammations and on HIV viral load in breast milk. Our study suggests that prevention and care of breast problems related and leading to mastitis should be integrated into PMTCT programs, reaching into the community beyond the boundary of hospital-based interventions and establishing a partnership with traditional healers.