Introduction

Top Abstract Introduction Subjects and methods Results Discussion References

Increasingly medical research is showing the health benefits of breast feeding for both mothers and babies.^1-3 The number of women who start breast feeding in Britain, however, has changed little since 1980, ^{4 5} despite many health promotion initiatives.^6-8

Previous research into decision making about infant feeding has predominantly used surveys to look at the sociodemographic variables and attitudes associated with breast feeding, ^{4 9-12} and negative associations have been found consistently with young mothers, low social class, and stopping full time education at an early age.

The underlying assumption of these studies is a rational model of decision making whereby women weigh up the pros and the cons of breast feeding and decide accordingly. There is little robust evidence, however, that education increases rates of breast feeding, and most women who choose formula feeding "know" that breast feeding would be better for their babies.⁹ The limitations of a health promotion model and the need for more sociocultural models for understanding how women make decisions about infant feeding is discussed by Maclean in her overview of the literature.¹³ She highlights the lack of qualitative research in a topic where surveys predominate.

This qualitative research arose from involvement of one of the authors (PH) in commissioning maternity services, where attempts were being made to increase local rates of breast feeding. We set out to explore how first time mothers from an inner city area decide whether to breast or bottle feed by focusing on their knowledge and previous exposure to infant feeding and its relation to their decision.

	Subjects and methods

Top Abstract Introduction Subjects and methods Results Discussion References

Twenty one white women of lower social class and with low educational level who were having their first baby and were living in a deprived inner London health authority were selected for investigation as they belong to a group known to have low rates of breast feeding. They were initially approached about the study before antenatal booking by general practitioners and midwives known to the researcher (PH). The aim was to interview the women as soon as possible after confirmation of pregnancy to ascertain their views before professional input about infant feeding. If they agreed they were contacted by PH who introduced herself as a researcher, not a doctor, and explained that the research was about choices women make when they are looking after their first baby. This enabled infant feeding to be discussed in the wider context of pregnancy and family life according to the woman's own priorities. The infant feeding agenda was declared later in the interview. Ethics committee approval was obtained.

Recruitment
Contrary to expectations women initially recruited were older and intended to breast feed, so purposeful sampling¹⁴ was used to target teenage women who intended to formula feed to ensure that all viewpoints were represented. Eight women recruited knew that PH was a doctor but were not her patients. A discussion of the influence of her role as a researcher and a general practitioner on both the recruitment and the interview data has been reported elsewhere.¹⁵ Recruitment ceased when theoretical saturation had been reached.¹⁶ Twenty one women were interviewed antenatally, of whom 19 were reinterviewed 6-10 weeks after the birth. Two women had moved away.

Interviewing
Data were collected with a topic guide developed during four pilot interviews rather than with a structured questionnaire to enable respondents to tell their stories in their own way. Women chose the time and place of interview (all except three took place at home) and whether to be interviewed alone or with another person of their choice (nine partners, three mothers, one father, and two sisters). Interviews were tape recorded, fully transcribed, and field notes of reflexive observations were recorded in a research diary.

Data collection and analysis
Data collection and analysis proceeded in an iterative manner in accordance with grounded theory,¹⁷ allowing concepts to be confirmed, rejected, or modified as the study progressed. The framework method of data analysis¹⁸ was applied systematically both within and across cases with categories and themes identified by reading the transcripts. A coding index was developed and applied to each transcript with Microsoft Word computer software. The language used by women when they discussed their intentions about infant feeding was examined in detail using the principles of discourse analysis.¹⁹

Validation and trustworthiness
Respondent validation was used to check whether the data analysis and interpretation truly represented the women's views. The 19 women remaining in the study were sent a synopsis of their individual case analysis, together with a summary of the key research findings. Confirmatory feedback was received from 11 women, with two letters being returned undelivered. The emerging analysis was cross checked with data obtained from different sources (individuals, couples, and the focus group interviews). Both authors were involved in reading and analysing transcripts. This paper focuses on the views expressed by the index women unless otherwise stated. A more detailed account of the research methodology and findings is available.²⁰

	Results

Top Abstract Introduction Subjects and methods Results Discussion References

Within the intended sample of white women of lower social class and educational level we had a broad spectrum of personal and demographic characteristics (table 1). Two key concepts arose from discourse analysis of the antenatal interviews: a woman's confidence in her ability to breast feed and her commitment to a particular method of feeding. This enabled women to be classified into five distinct groups according to antenatal feeding intention (see box). Antenatal and postnatal matrices of the coded themes were then created for each feeding intention group. There was no strong association between the father's antenatal views about breast feeding and the mother's feeding intention and actual behaviour. A discussion of the role of partner, family, and friends in the decision making process is available elsewhere.²⁰

Feeding outcome was classified according to the definitions of Labbok and Krasovec²¹ into exclusive breast feeding, high, medium, and low partial breast feeding, token breast feeding, and formula feeding. Initiation of breast feeding was defined as the baby having been put to the breast even if only once. Feeding outcome was recorded for the first 24 hours, 24-72 hours, 1 week, 2 weeks, 4 weeks, and 6 weeks.

The relation between the woman's feeding intention group and feeding outcome is shown in table 2. Breast feeding was initiated by all women in the committed and probable breastfeeding groups (groups 1 and 2) but by only two of the six women in the possible breastfeeding group (group 3). One woman in the probable formula feeding group (group 4) tried a token breast feed and no women in the committed formula group (group 5) initiated breast feeding. At 6 weeks all four women in the committed breastfeeding group (group 1) were still giving some breast feeds compared with three of the six women in the probable breastfeeding group (group 2) and one of the six in the possible breastfeeding group (group 3).

Exposure to breast feeding and influence on feeding intention and behaviour
A strongly held view was that breast feeding is seldom seen or talked about. Women's history of their exposure to breast feeding was strongly associated with their antenatal commitment and confidence in their own ability to breast feed and the initiation of breast feeding. It was less strongly associated with duration and exclusivity of breast feeding. Previous exposure to breast feeding was either a positive or a negative influence on women's decision whether to breast feed, depending on the context in which it occurred. Crucial factors determining women's reactions were the nature of their relationship to the breastfeeding woman, the presence of other people and their reaction, the frequency of exposure, the perceived appropriateness of the setting, and their own level of body confidence. Many women in this study revealed negative feelings about their own bodies and low levels of body confidence. Seeing other women breast feed could either improve their body confidence or reinforce these negative feelings. When breast feeding was witnessed as part of normal everyday life by both the woman and her family and friends she was more confident in her own ability to breast feed and committed to her decision. If breast feeding had been seen only infrequently and other people present had made negative comments her reaction was less positive. This is illustrated by the quotations in the box.

Women's conversation as a reflection of their level of exposure
All women in this study were aware that breast feeding is best for health. There were differences, however, in the ownership of this knowledge. There were striking differences between feeding intention groups in the language they used when they were discussing their knowledge and beliefs. The personal pronouns chosen by women reflect the level of ownership of a viewpoint, and this was associated with their exposure to breast feeding. The following quotations illustrate this. Penny in group 2 said, "I just feel that it would be better for the baby in the long runit's just how I would imagine that I would deal with being a mother." Penny had positive recollections of seeing her cousin breast feed and owned her beliefs about the benefits of breast feeding by using the first person singular. Vicky in group 3 said, "It protects them against catching things I suppose, like virusesthe only thing is, you don't really know how much they are getting though, do you?" Vicky remembered seeing breast feeding once on a bus as a child and being "gob smacked." She made a factual statement in the third person singular about the protective effects of breast feeding but "I suppose" suggests uncertainty of ownership. She then voiced a common concern about breast feeding by using the second person "you." This suggests collective rather than individual ownership. Hayley in group 5 said, "They say that it helps their immune system and all that. Someone did try and tell me the other day that it makes them more intelligent." Hayley reported a negative reaction to the one occasion when she witnessed breast feeding. She distanced herself from the views of others about the benefits of breast feeding by using third person pronouns "they" and "someone." She does not own these statements.

	Discussion

Top Abstract Introduction Subjects and methods Results Discussion References

The finding that women of lower socioeconomic status who perceive exposure to breast feeding in a positive way may be more likely to initiate breast feeding provides a new perspective on decision making about infant feeding. With trends towards smaller family size, more women working, and increasing geographical separation of families, opportunities for exposure to breast feeding may be decreasing. Existing research looking at a possible association between exposure and the decision to breast feed has not looked in detail at the context or the frequency of exposure. ^{9 10 22}

Reference to the work of Hastrup was useful in developing the findings of this study into a model for decision making about infant feeding.²³ Hastrup describes two different kinds of knowledge required for performing a skillembodied knowledge and cognitive knowledge: "Words may store cognitive knowledge but have a much lesser capacity for storing other kinds of experience, which are embodied and stored in the social-habit memory."

Possession of embodied knowledge may also explain the consistent research findings that previous feeding behaviour is the strongest predictor of future feeding behaviour and that having family and friends who breast feed is associated with deciding to breast feed.⁴ The relative importance attributed to theoretical or cognitive knowledge and embodied knowledge is likely to vary between women. A hypothesis arising from this study is that for women from lower socioeconomic groups who learn skills through apprenticeship, embodied knowledge gained through exposure to breast feeding may be more influential than theoretical knowledge. In contrast, women with higher educational qualifications are more familiar with learning and making decisions on the basis of theoretical knowledge. Further research could evaluate interventions to increase women's embodied knowledge of breast feeding and the possible role of television or video as a proxy for real life exposure.

Implications for clinical practice and policy
Most clinicians ask women antenatally how they are planning to feed their baby. The emphasis is on providing factual information about the health benefits of breast feeding. Attention to women's discourse when recounting their knowledge and experience of breast feeding will help clinicians to assess a woman's confidence, commitment, and ownership of her knowledge. Women hoping to breast feed but with little exposure to breast feeding may benefit from an antenatal apprenticeship with a known breastfeeding mother.