Changes in diet and eating behaviour are an essential part of the stereotypical image of pregnancy, but surprisingly little research has concentrated on what women actually eat and why. Pregnant women are typically depicted as being plagued by strange and irresistible cravings as well as having aversions to certain foods. Nausea, familiarly, though inaccurately, known as 'morning sickness', is seen to be characteristic of pregnancy and in popular culture is regularly depicted as the earliest somatic symptom.
The regularity of the reporting of somatic symptoms across the world suggests that these symptoms and dietary change in pregnancy are driven by physiological and endocrinal factors. Certainly research in this area routinely assumes that these are the only drivers but it is likely that other, psychological factors may be as important. In the case of eating behaviour, a combination of dietary beliefs, an association of symptoms with diet in the past, and past dietary behaviour may be used to guide behaviour and interpret experience.
There are many traditional beliefs about what and how women should and should not eat during pregnancy, some of which appear to be common across cultures, for example that women should increase their food intake at least in the early stages, as summarised in the phrase `eating for two'. Other dietary beliefs seem to be very culturally specific and derive from belief systems relating to the body and the development of the foetus, for example pica ± the craving for and eating of non-food substances such as earth and clay, as Walker et al. (1997) investigated in South Africa.
Although the adoption of stereotypical beliefs may limit women's choices, it also sanctions behaviour that is otherwise not regarded as acceptable in young women, for example satisfying `cravings' allows high calorie eating patterns. Many young women restrict their calorific intake in pursuit of the current ideal feminine body shape in the developed world and concern is often expressed in the popular press in the developed world about children and young girls as young as seven years old restricting their food intake. It has been estimated that on any given day approximately 45 per cent of American women are on a diet. Eating disorders, principally anorexia and bulimia nervosa, are largely affictions of women (Andersen, 1995) and women of all ages express dissatisfaction with their body (Stevens and Tiggemann, 1998). Estimates of the prevalence of eating disorders in women of childbearing age have been found to be between 1 and 2 per cent (Fairburn and Beglin, 1990).
Further pressures on pregnant women come from external sources. As David-Floyd (1994) points out, the pregnant body can be seen as inappropriate. We see how pregnant celebrities are currently usually depicted in the media as remaining slim during pregnancy and rapidly regaining their pre-pregnancy shape. Therefore, for many women, pregnancy, with its accompanying change in body size and shape, may be seen as a personal challenge.
To add to these pressures, pregnant women are often the target of food scares in the media. In some instances this is because a link has been posited, by epidemiologists or basic scientists, between particular foodstuffs and foetal wellbeing (for example, there were reports in 2002 on the possible risks of drinking too much coffee and the dangers of mercury in tuna fish). In other cases, targeting arises because pregnant women are generally regarded as a vulnerable group, alongside older people and the very young. So if a foodstuff is discovered, or thought, to pose some health risk, then vulnerable groups are advised to avoid it. This was the case in the UK when there were reports on the risks of Salmonella in chicken eggs, which originally appeared in the 1980s and reoccurred in the late 1990s. How women respond to these scares is less frequently reported.
And it is not just the potential risk of poor foetal outcome; the diet of women during pregnancy has a significant impact on their long-term health. The most rapid rise in obesity and overweight in women occurs during the peak childbearing years (Department of Health, 2002) and obesity is a major factor in ante- and perinatal maternal deaths (Lewis and Drife, 2004). Importantly, for long-term health, 14-20 per cent of women are 5kg or more heavier 6±18 months post partum, compared to their prepregnancy weight (Keppel and Taffel, 1993; Ohlin and Rossner, 1990). As has been regularly documented, obesity and overweight are increasingly important health problems and are associated with a number of diseases including hypertension, type II diabetes, cardiovascular disease and some types of cancer (NIH, 1998).
Despite the known impact of diet on the health of women, it has taken the results of long-term studies of its impact on the health of offspring into adulthood to prompt the interest of mainstream medical researchers in maternal nutrition during pregnancy, outside underdeveloped countries where even basic nutrition is problematic. Poor maternal nutrition has long been linked to foetal and child ill-health. This effect is due not only to insufficient energy intake overall but also to the incorrect balance of food types and nutrients, leading to restricted intrauterine growth, low birthweight, prematurity and other perinatal morbidity (Kramer, 1993). More recent research suggests that several diseases of later life also originate from impaired intrauterine growth and development, leading to permanent effects on structure, physiology and metabolism (Godfrey and Barker, 2000; Mathews et al., 1999).
This is known as the 'foetal origins' or Barker hypothesis, named after David Barker who studied the records of 16,000 men and women born in Hertfordshire, England from 1911 to 1930 and whose records can be traced to the present day. The birth records on which these studies were based came to light as a result of the Medical Research Council's systematic search of the archives and records offices of Britain. The Hertfordshire records were maintained by health visitors and include measurements of growth in infancy as well as birthweight. Death rates from coronary heart disease fell two-fold between those at the lower and upper ends of the birthweight distribution. Barker concluded: The fetal origins hypothesis states that fetal under nutrition in middle to late gestation, which leads to disproportionate fetal growth, programmes later coronary heart disease' (Barker, 1995: 171).
Similar results have been reported in other European countries, India and the US. More recently, excessive maternal weight gain has also been related to perinatal problems in babies (Kabiru and Raynor, 2004) and to childhood obesity (Whitaker, 2004). Higher levels of obesity and of infant mortality and morbidity (associated with poor maternal nutrition) are seen in more disadvantaged groups in the UK (Department of Health, 2002; Macfarlane et al., 2000). This work prompted an ongoing large-scale survey of the lifestyle and dietary behaviour of 20- to 34-year-old women in Southampton in southern England. Three thousand of the 12,500 women surveyed became pregnant during the course of the study, and their dietary behaviour is being closely monitored.
Such surveys and monitoring research will add considerably to our knowledge of what women eat during pregnancy and how their diet changes. However, we still know little about what prompts women to change their diets during pregnancy and what external pressures, personal beliefs and habits underlie the dietary choices they make: for example, whether women who eat healthily prior to pregnancy make more changes than those who do not. In this chapter we consider the research on various aspects of dietary behaviour during pregnancy and re¯ect on research perspectives. On the one hand these perspectives take pregnancy out of the context of women's lives and, except in the extreme case of eating disorders, disregard previous eating behaviours. On the other hand they fail to take account of the influence of women's culturally embedded beliefs about pregnancy as a different and specific physical experience.
Dietary beliefs and dietary change
There seems to be general agreement among those with expertise in nutrition and women themselves that diet should change during pregnancy.
At the very least, the extra demands on the body call for increased calorie consumption of about an extra 200 calories a day. Beyond this consensus, however, there seems to be wide variation about what exactly is an appropriate diet during pregnancy, with competing information from the media, health professionals and pregnancy manuals and from family and friends. Beliefs about changing one's diet during pregnancy may be associated with the wellbeing of the mother, with the wellbeing of the baby or with a desirable weight gain. Such beliefs may be rooted in the woman's own past eating behaviour, in antenatal health education or may have been transmitted from generation to generation within a particular culture or subculture.
One of the first questions we should ask is whether women do deliberately change their diet during pregnancy for either their own or their child's wellbeing. The answer, from our own and others' work, suggests that they do, and that the changes seem rather more motivated by concern for their child than themselves.
Two early studies of US women looked at how they reported changing their diet (Norman and Adams, 1970; Orr and Simmons, 1979). In Norman and Adams' (1970) study, approximately two-thirds of the women reported adjusting their diet. Such adjustments included adding, reducing or eliminating foods. Greater intakes of dairy products together with fruit and vegetables have generally been reported as usual dietary additions.
High sugar foods such as desserts, chocolates and biscuits were the items most commonly reported to be reduced or eliminated, as were foods with a high salt or fat content. Orr and Simmons (1979) found that most of the women they studied believed diet to be important for both mother and baby, though a substantial number did not recognise its importance for mothers. However, they did report that they were prompted to change their diet on the basis of advice from health professionals, who may have placed more explicit emphasis on change.
Most studies rely on women's reports of how they change their diets rather than measuring actual food intake. In a study we carried out we examined the eating patterns of a demographically mixed sample of 102 women during their first or second pregnancy by exploring specific changes that they made to their diet, as well as how somatic symptoms associated with pregnancy, such as nausea, affect food choice, and how dietary beliefs influenced women's food choice (Pattison and Bhagrath, 2003, 2004).
We found that 79 per cent of women reported that they should increase consumption of certain foods and 82 per cent reported trying to avoid certain foods. The foods increased were fruit, vegetables and dairy products whereas the foods avoided were foods high in sugar and fat and those that health professionals and other advisors had suggested were dangerous, such as soft cheeses. However, when we measured the actual frequency of consumption, no significant difference was found between when women last consumed the food they felt they should increase or avoid and their current reported intake, suggesting that other factors are at play.
In a study in the US, Pope et al. (1997) studied dietary changes in pregnant adolescents. Their results indicated that the pregnant girls' diets were more nutrient dense than a matched sample of non-pregnant girls. Since becoming pregnant, a majority reported that they had increased the amount of food eaten, specifically milk/dairy products, vegetables, fresh fruit/unsweetened juices, breads/cereals and chocolate. Health professionals' influence was cited for increased intake of vitamin supplements and milk, but not for changes in food intake. The major motivations for increasing food intake during pregnancy seemed to be food cravings, increased appetite, improved taste of food and concern for the baby.
So there is evidence that women report changing their diets in such a way as to increase their calorific intake, and specifically increasing certain foodstuffs and reducing intake of others. However, what they actually eat is not simply motivated by dietary advice from midwives or nutritionists. One interpretation of our own findings is that the women in our sample knew what foods their midwives would recommend them to eat, but that somatic symptoms such as nausea, or other beliefs about diet, affected their food choices as well as presumably personal preferences. Traditional beliefs may significantly influence dietary patterns and many are not consistent with recommended guidelines for nutrition during pregnancy. Examples of these include eating for two, not mixing certain foods, taking vitamins to overcome an inadequate diet and eating only a few selected foods.
The impact these traditional beliefs have on dietary behaviour in developed countries may be limited because of increasing access to resources, for example formal education, the internet and pregnancy magazines as well as positive media attention promoting healthy eating and regular contact with health professionals, which would subsequently encourage a different attitude towards diet to be established. In our own work in a UK population, belief in traditional eating patterns varied with educational level so that more highly educated women were less likely to endorse such beliefs and less likely to report suffering cravings (Pattison and Bhagrath, 2003). However, in this sample educational level was confounded with socioeconomic status, as it is in many studies.
In a sample of 6,125 non-pregnant women from the Southampton study, mentioned above, Robinson et al. (2004) examined the in¯uence of socio-demographic and anthropometric factors on the quality of the diets of young women in the UK. They found that educational attainment was the most important factor related to the quality of the diet consumed. In all, 55 per cent of women with no educational qualifications had scores in the lowest quarter of the distribution, compared to only 3 per cent of those who had a degree. Smoking, watching television, lack of strenuous exercise and living with children were also associated with lower diet scores. After taking these factors into account, no other factor including social class, the deprivation score of the neighbourhood or receipt of benefits added more than 1 per cent to the variance in the diet score. The significance of these findings is that they suggest that poor diets in general in this group are not simply a result of the level of deprivation, but reflect a more general pattern of health behaviour that is linked to poor access to information sources through education.
Some support for this thesis comes from our study (Pattison and Bhagrath, 2004) where women who reported making changes to their diet were also more likely to have made additional changes to their lifestyle. Although there was no variation on alcohol intake (all women who previously drank alcohol reported cutting down or abstaining from alcohol consumption during pregnancy), more educated and younger women were more likely to have attended antenatal classes and changed their exercise levels. In our study, women who increased exercise and women who decreased exercise were classified together as having made a change.
In considering how women respond to pregnancy we should not forget that people's belief systems are complex and they can simultaneously hold beliefs which are conflicting and contradictory. A study carried out by Carruth and Skinner (1991) found that a substantial proportion of clients of the 1,771 practitioners they surveyed had beliefs about physiological needs during pregnancy, practices related to a healthy baby and alcohol and caffeine consumption that were not significantly different from those endorsed by the American Dietetic Association.
However, they also held beliefs, particularly about cravings, which showed strong regional differences, and which represent traditional views not supported by dieticians (e.g. eating for two, eating only a few selected foods, restricting salt intake, taking vitamins to overcome an inadequate diet and deciding that pregnancy is a good time to lose weight). This study was performed in the US. However, few similar studies have been done elsewhere to assess whether similar beliefs exist and if so to what extent. Nevertheless, as we discuss below, advice given by midwives and in publications for pregnant women is often vague, recommending a 'healthy diet' and being open to interpretation within the woman's own belief system. Many traditional beliefs about diet in pregnancy revolve around cravings, aversions and somatic symptoms of pregnancy, particularly nausea and vomiting, and we will now consider these in more detail.
Cravings, aversions and somatic symptoms
Many women report cravings and aversions towards particular foods during pregnancy; the reported occurrence in the literature ranges from 66 to 85 per cent. Cravings and aversions are undoubtedly at least partially interrelated with beliefs as the behaviour of consuming or avoiding particular foods during pregnancy may be directly related to cultural or social values. For example, there is a strong belief system within certain cultures to support pica, which is the consumption of non-food substances such as clay and earth. Food cravings may also be experienced as a somatic symptom though these are also likely to be influenced by cultural beliefs (Bayley et al., 2002).
The medical model of pregnancy suggests that all experience of pregnancy is related to physiological and endocrinal change, thus much early research on cravings and aversions assumed that the root of these desires is a mechanism to protect the foetus. Therefore, cravings are seen as a way of making up for dietary inadequacies and aversions, and nausea and vomiting are seen as a way of protecting the foetus from noxious substances.
Traditional beliefs about food restrictions have also been investigated in this way. Fessler (2002), for example, suggests that maternal immunosuppression, which is necessary for tolerance of the foetus, results in vulnerability to pathogens. Symptoms could be a 'behavioural prophylaxis' against infection, with nausea and aversions leading to the avoidance of foods likely to carry pathogens, and cravings leading to foods which boost the immune system. A similar conclusion is reached in a review by Flaxman and Sherman (2000) of morning sickness and pregnancy outcome. This was particularly assumed in the case of pica, the most extreme and unusual of cravings. These assumptions are also found in the explanations women themselves give for what they are experiencing.
Several studies carried out in the US by Carruth and Skinner on pregnant adolescents identified beliefs which gave a 'physiological basis' for cravings. For example 'I should give in to my cravings or I will harm my baby' and 'foods that make me feel sick must be bad for my baby' (Pope et al. 1992).
Several other studies which have looked at the impact of pica on pregnancy outcome appear to refute the dietary deficiency theory. In certain societies pica is common. Luoba et al. (2004) found that 378 of the 827 women they studied in western Kenya were eating earth. Horner et al. (1991), in a review of pica in the US, showed that the prevalence of pica among pregnant women from poor, rural and predominantly black areas declined between the 1950s and the 1970s but then remained constant.
They conclude that the evidence suggests that pica during pregnancy is associated with anaemia and with maternal and perinatal mortality. Lopez et al. (2004) found a prevalence of 23 to 44 per cent in Latin America. Rainville (1998) investigated the association of pica with two adverse pregnancy outcomes: low birthweight and preterm birth in a group of women from Texas, US. This study found a wide range and a high prevalence of pica if it was more broadly defined than usual; normally pica is used to refer to the craving for and practice of eating soil, clay or dirt.
In particular, the pica sample comprised those eating: ice, 53.7 per cent of their sample; ice and freezer frost, 14.6 per cent; other substances such as baking soda, baking powder, cornstarch, laundry starch, baby powder, clay or dirt, 8.2 per cent. Those reporting no pica as defined in this way only amounted to 23.5 per cent of the sample. Women in all three pica groups had lower iron levels at delivery but there were no differences in mean birthweight. In the UK, pica is rarer; our study (Pattison and Bhagrath, 2003) found only three women who experienced craving for non-food substances, all of whom came from non-European ethnic groups and none of whom actually ate the substances they craved.
So there is little evidence that pica attenuates dietary deficiencies, though this may be the belief of women who practice it (Ukaonu et al., 2003); in fact it probably increases them. A meta-analysis of pica research found that ethnicity was the most important predictive variable (Simpson et al., 2000). Geissler et al. (1999) showed a strong associated between pica and anaemia and iron depletion in women from Kenya. The women themselves described soil-eating as a predominantly female practice with strong relations to fertility and reproduction. They made associations between soil-eating, the condition of the blood and certain bodily states. The beliefs women held about eating soil re¯ect both a kind of dietary deficiency thesis and the protection against illness thesis explored below.
Geissler et al. emphasise the importance of social and cultural contexts for how women interpret the experience of pregnancy. They conclude that pica is not simply a behavioural response to physiological need but rather that it is a rich cultural practice. Most western cultures regard pica as deviant and repulsive; Lopez et al. (2004) describe pica as a 'disorder'. Its practice is therefore secret and hidden and Henry and Kwong (2003) argue that pica is stigmatised in American society because of the meaning of dirt in that culture. However, they also argue that the consumption of vitamins and dietary supplements constitutes a similar type of behaviour, done for similar reasons, albeit that it is regarded differently in health terms.
In contrast to pica, nausea is experienced by pregnant women of many cultures. In studies in the developed world, the majority of women report experiencing some nausea. A cross-cultural analysis by Flaxman and Sherman (2000) revealed 20 `traditional' societies in which morning sickness has been observed and seven in which it has never been observed.
As we discuss below, there is evidence that nausea affects food choice and is related to food aversions. However, the theory or belief that nausea and vomiting in pregnancy protect women from ingesting certain vegetables or foods that cause congenital abnormalities and other adverse outcomes of pregnancy is questionable. There have been a number of studies exploring the links between nausea, dietary intake and pregnancy outcome in terms of miscarriage or birthweight. Several of these have found no significant association between them (Brown et al., 1997; Hook, 1978; Walker et al., 1985; Wijwardene et al., 1994) but Lee et al. (2004) found an association between even mild morning sickness and birthweight, and concluded that this was because it reduces dietary diversity and nutrient intakes. A study carried out in the US suggested that the women with the most extreme condition (hyperemesis gravidarum) had babies of lower gestational age and had longer antenatal hospital stays (Paauw et al., 2005).
In our own work (Pattison and Bhagrath, 2003, 2004), nausea and vomiting were the most common symptoms affecting food choice; most women responded by avoiding altogether foods they associated with nausea. Reasons that were cited for aversions in a study among Saudi women were smell (9.4 per cent), vomiting (28 per cent), diarrhoea (2.5 per cent), undesirable effect on foetus (7.8 per cent) and heartburn (18.7 per cent) (Al-Kanhal and Bani, 1995).
Dietary aversions usually occur earlier in pregnancy than do cravings and are frequently reported as being more severe. The most common aversions in US samples appear to be towards alcohol, coffee, meat and foods which have a distinct flavour or smell, for example spicy foods or Italian foods (Hook, 1978; Pope et al., 1992). Pope et al. (1997) found that many of the adolescents they studied (66 per cent) experienced aversions during pregnancy towards previously liked foods. The most common aversions were to meats, eggs and pizza and led to decreased consumption of these foods.
In our study too (Pattison and Bhagrath, 2003), 72 per cent of women developed aversions to food. The most commonly reported aversions were to meat (20 per cent) and spicy foods (20 per cent), though a small number (3 per cent) had developed an aversion to fruit and vegetables. Aversions were usually linked to nausea, with the smell or taste of these foods inducing nausea and/or being associated with an incidence of vomiting.
This pattern of aversion suggests that rather than being a specific characteristic of pregnancy, aversions could re¯ect a way in which women respond generally to foods that they associate with nausea. It is well known that people generally can develop aversions to foods through a process of associative learning. Whether or not the food was the cause of the nausea, the coincidental association of a bout of nausea or vomiting with a food is enough to create an aversion. In other words, nausea is created by hormonal changes during pregnancy but women interpret this symptom in the same way they would at other times and develop a taste aversion.
Data to support this come from a study by Bayley et al. (2002) who studied the temporal association between the first occurrences of nausea, vomiting, food cravings and food aversions during pregnancy. Of the women in their sample, nausea and vomiting were reported by 80 per cent and 56 per cent respectively, and food cravings and aversions by 61 per cent and 54 per cent respectively. Cravings and aversions were not related. There was a significant positive correlation between week of onset of nausea and of aversions. In 60 per cent of women reporting both nausea and food aversions the first occurrence of each happened in the same week of pregnancy. No such association was found for cravings.
In the developed world, while pica is very uncommon, other cravings and aversions are common and rather prosaic. Pope et al. (1997) found that their US sample most frequently reported cravings for: sweets, especially chocolate; fruit and fruit juices; fast foods; pickles; ice cream; and pizza. Adolescents craving sweets during pregnancy consumed more sugar than those who did not crave sweets. Cravings generally resulted in increased intake, and aversions led to decreased food consumption. In our study (Pattison and Bhagrath, 2003), 62 per cent of women reported cravings.
The most popular food craved was chocolate (32 per cent) and other foods craved were generally high carbohydrate and/or high fat foods, that is, bread, pasta, ice cream, chips, fruit, meat and what was generically termed 'McDonalds' (5 per cent of the sample). As in the study reported earlier (Pope et al., 1997), the women with cravings had increased their intake of these foods, with 91 per cent having consumed the food they craved in the 24 hours before they were interviewed.
It is clear then that cravings can have a significant role in diet during pregnancy as they may increase total intake of food or change the proportion of foods eaten. However, cravings are not exclusive to pregnancy. They are frequently reported in the general population and typically tend to involve foods high in sugar and/or fat, such as chocolate (Yanovski, 2003). So, can cravings in pregnancy be regarded as an extension of a normal experience?
There are two relevant theories as to why cravings develop and why they endure (Cepeda-Benito and Gleaves, 2001). The first suggests that substances in the food supply a dietary imbalance. This imbalance may be caused in various ways, for example by dieting or by a nutritional deficiency. This is the theory that most closely links to the dietary deficiency hypothesis outlined above. So the increased need for calories in pregnancy, for example, would cause cravings for high calorie foods. The second type of craving theory is that of 'incentive hypothesis' of craving. This suggests that cravings are a result of learning what foods produce feelings of wellbeing.
This theory suggests that people have cravings for these particular foods because they have learned that the consumption of particular foods leads them to feel good. In psychological learning theory terms, they have learned to associate the food with positive reinforcement. This reinforcement can either take the form of physiological or psychological reinforcement (Wise, 1988).
The incentive hypothesis is supported by research into chocolate craving. In both the UK and the US, chocolate is widely reported to be the most commonly craved food. Michener and Rozin (1994) refuted the suggestion that this is because of the psycho-pharmacologically active substances in chocolate (e.g. caffeine), as they found that capsules containing the same
substances did not reduce cravings. It seems most likely that chocolate tastes and smells good to people. Rogers and Smit (2000) concluded that chocolate is simply a common example of the kind of food which people tend to associate with pleasant taste, smell and texture, that is, one that is high in fat and sugar.
Hill and Heaton-Brown (1994) looked at food cravings in healthy, non-binge-eating women. They found that the most frequently craved food was chocolate (high fat, high carbohydrate), with cravings for savoury foods, such as pizza, being much less frequently observed. In contrast to the accounts given by pregnant women, the food cravings reported by these women were seen as positive, pleasant, hunger-reducing, mood-improving experiences rather than reflecting any biological need. So despite differences in the beliefs that pregnant and non-pregnant women have for their cravings, the cravings themselves are for similar types of food. Furthermore, Crystal et al. (1999) found a significant association between experiencing cravings and aversions prior to pregnancy and experiencing cravings and aversions during pregnancy.
A number of more general studies suggest that women's diet during pregnancy is strongly influenced by their tastes and eating habits before pregnancy. Mathews and Neil (1998) studied 774 women in the early stages of pregnancy and found that their dietary intake was very similar to that of non-pregnant women and accordingly they were short of some nutrients thought to be important for foetal health. Perhaps the most striking results in this regard come from a qualitative study of the diets of pregnant teenagers for the Maternity Alliance and the Food Commission in the UK (Burchett and Seeley, 2003). They gave detailed accounts of the reasons why they did not eat foods that they regarded as healthy, and the most common reason, given by nearly half of the teenagers, was dislike of that foodstuff. Cost was also a factor for a fifth of them and a number also said that the foods were unfamiliar or not offered in their homes. Other reasons for avoiding healthy foods were the effort required to buy them and cook them.
In summary, most of the research on aversions and cravings in pregnancy has stemmed from the assumption that the dietary behaviour of pregnant women is a direct result of pregnancy. So aversions and cravings are assumed to result from biological processes which protect women from infection and restore dietary deficiencies. Although there may be some merit in this approach, it ignores the lifetime of experience that women have had with food, particularly in relation to cravings. So is this a time when women feel less restrained in their eating?
Restrained and unrestrained eating
Unlike diet in pregnancy, the concept of dietary restraint has been widely studied by psychologists. Dietary restraint refers to the tendency to restrict food intake, usually in order to lose weight, or to maintain slimness. It is a volitional but stable behaviour. Herman and Polivy (1983) developed the 'boundary' model of eating behaviour, which suggests that two physiological boundaries determine when people start and stop eating: hunger and satiation. However, restrained eaters have another self-imposed boundary, which overrides the other boundaries - the diet boundary, that is, the amount of food (or calories) that restrained eaters believe they should consume. This diet boundary overrides the normal hunger and satiation boundaries. Dietary restraint is common in women in western cultures as evidenced by the high proportion of women who report dieting at any one time. It is beyond the scope of this book to give a detailed account of the impact of pregnancy on severe eating disorders. Here we will look at the evidence that what might be termed 'normal' dieting behaviour before pregnancy has an impact on what and how much women eat during pregnancy.
Pregnancy might be a time when social pressures for slimness could be expected to be relaxed, thus resulting in reduced weight concern despite an increase in body size. Women may therefore be less restrained in terms of what and how much they choose to eat, causing weight gain to be higher. On the other hand, restrained eaters may remain subject to the cultural pressure to be slim and continue or even increase their dieting behaviour. Similarly, restrained eaters may be happy with their pregnancy shape, as it is something apart from their normal experience, or restrained eaters may see the weight and size gained in pregnancy as distasteful. The evidence on both these issues is contradictory.
Davies and Wardle (1994) evaluated body image, body satisfaction and dieting behaviour in pregnancy, expecting women to feel less social pressure to be slim. Pregnant women certainly had a lower 'drive for thinness', had lower body dissatisfaction and rated themselves as less overweight than non-pregnant comparisons. However, they showed similar preference for size of figure to non-pregnant women.
These findings suggest that pregnancy is a time of relaxation in concerns about weight, but that this change is temporary and does not override women's general beliefs about their ideal weight and body shape. Davies and Wardle's findings chime with our study (Pattison and Bhagrath, 2003). We did not measure dietary restraint directly; however, the women we interviewed were significantly more likely to be satisfied with their pre-pregnancy shape than current shape. And those who were more satisfied with their pre-pregnancy shape were more confident they could regain it. This suggests that the women who had experience of successful weight control before pregnancy were confident in their ability to exercise such control again.
Clark and Ogden (1999) investigated the role of dietary restraint in mediating changes in eating behaviour and weight concern in pregnancy. They also compared pregnant and non-pregnant women. The pregnant women reported eating more, showed lower levels of dietary restraint and were less dissatisfied with their body shape than the non-pregnant group. They also showed higher eating self-efficacy, that is, the belief that one can control one's own eating. The pregnant women rated themselves as less restrained in their eating behaviour than they had been immediately before their pregnancy and nearly half reported eating more.
Clark and Ogden also found that the previously restrained eaters, when pregnant, rated themselves as significantly less hungry and having greater eating self-efficacy than the non-pregnant restrained eaters. They were comparable in these regards to non-restrained eaters. The results showed no effect of restrained eating on weight change. Clark and Ogden concluded that for women who normally restrain their eating, pregnancy both legitimises an increased food intake and removes previous intentions to eat less.
But other studies contradict these findings. For instance, Conway et al. (1999) studied dietary intake and weight gain during pregnancy in relation to dietary restraint in a longitudinal study of women from early to late pregnancy. In their study, current dietary restraint was measured (i.e. restraint employed during pregnancy). They found that restrained eaters were less likely to experience weight gains within the recommended range for their pre-pregnancy body mass index (a ratio of height to weight). This went either way such that some gained more weight and some less weight than recommended.
DiPietro et al. (2003) studied pregnant women's weight-related attitudes and behaviours in relation to several psychological and social characteristics. This was not a longitudinal study, rather women's attitudes about weight gain were assessed once at 36 weeks of pregnancy Several variables had been assessed prior to this, namely anxiety, depression, social support, emotionality and perceived stress (pregnancy-specific and non-specific). Twenty-one per cent of the women were restricting their food intake in some way during pregnancy. The women who reported more restrictive behaviours were more anxious, depressed, angry, stressed and felt less uplifted about their pregnancies in general. Those women who were more positive about their bodies during pregnancy felt better about their pregnancies in general. They also were less depressed and felt less angry. On the other hand, women who were self-conscious about their pregnancy weight gain felt more hassled by their pregnancies and felt greater anger, though they also reported more support from their partners.
Women's feelings about their weight gain were not related to their body mass index before their pregnancy. The authors noted that negative attitudes about weight gain existed among women who gained weight within the recommended ranges. All this suggests that women's attitudes to weight gain during pregnancy are related to their general feelings about their pregnancy and psychological health rather than to their general feelings about their weight and their eating habits during pregnancy. A number of other studies have also found that women with a history of dieting are less satisfied with their bodies during pregnancy than those who do not normally diet (Abraham et al., 1994; Fairburn and Welch, 1990; Wood Baker et al., 1999).
So why do different studies have contradictory findings on the influence of women's dietary restraint before pregnancy? One obvious difference between studies is whether they involve women who restrained their eating before pregnancy (e.g. Clark and Ogden, 1999) or refer only to women who restrained their eating during pregnancy (e.g. Conway et al., 1999). These may well represent different groups of women, or the latter may be a subset of the former. However, other reasons for contradictory findings may lie in more recent theories of dietary restraint.
Recent work has established that dietary restraint itself is not a unitary phenomenon and can be applied in different ways. Joachim Westenhoefer proposes that there are two types of restraint: glexible and rigid. These two styles may lead to different strategies for dietary change during pregnancy.
Flexible restraint involves adaptation to the current circumstances, so while food intake is carefully controlled overall, if large amounts of food, or high calorie foods, are eaten on one occasion, this is compensated for by eating less on a later occasion. Rigid restraint on the other hand is an 'all or nothing' approach. Rigidly restrained eaters tend to diet frequently, but if they do eat foods that they feel they should avoid, then they do not compensate by eating less. These are the classic type of restrainers classified by Herman and Polivy (1983) as exhibiting the 'what the hell' effect. One implication of this for diet during pregnancy is that rigidly restrained women, once they have veered away from a weight control diet, may be expected to give up weight control entirely. The main reasons why rigid restrainers may stop restraining what they eat are the lack of social pressure to be slim and the sanction of eating forbidden foods because of cravings.
Herman and Mack (1975) discovered that an important characteristic of restrained eaters is that they can be induced to eat more than non-restrained eaters if they first consume a 'preload' - usually a sweet high calorie drink. However, Westenhoefer et al. (1994) found that flexible restrained eaters ate less following eating the preload than did rigid restrained eaters. Presumably this mimics their normal eating patterns. So flexible eaters make up for eating a high calorie food by eating less or low calorie foods, whereas once rigid eaters breach their 'diet boundary' they do not seem able to control their eating. It is noteworthy that most craved foods during pregnancy have high sugar content and are high in calories. If rigidly restrained eaters eat craved foods one would predict that this would act like a preload, and they would not compensate for it. Flexible restraint is associated with the absence of overeating more generally and low levels of depression and anxiety (Smith et al., 1999). If the participants in different studies of eating during pregnancy involve different types of restrained eaters, or a mixture of the two, they should find different patterns of restraint and different levels of weight control. Unfortunately, studies of dietary change in pregnancy have not provided
conclusive evidence on this yet.
Advice, recommendations and food scares
During the last century the majority of medical authorities recommended that weight gain during pregnancy should not exceed 9.1kg, primarily to prevent the development of maternal toxaemia, foetal macrosomia and caesarean deliveries. These recommendations increased to 11.4kg in the 1970s because it was felt that insuf®cient weight gain could contribute to premature births and to low birthweight babies born at the expected date. However, in 1990, an influential report from the Institute of Medicine in the US (U.S. Institute of Medicine, 1990) recommended weight gain ranges of 11.4-15.9kg with the primary goals of improving infant birthweight and ensuring the best outcome for the mother.
These weight gain recommendations vary according to the pre-pregnancy weight to height ratio as measured by body mass index (BMI). However, a significant number of normal weight women and an even greater proportion of overweight women in the US exceed these guidelines (Abrams et al., 2000). In fact, published studies suggest that only 30-40 per cent of women have weight gains within the Institute of Medicine's recommended ranges, with some gaining less weight than recommended but most gaining more weight than the guidelines suggest they should (International Federation of Gynaecology and Obstetrics, 1993).
In countries such as the US and UK, midwives and other health professionals see it as part of their role to offer advice on diet and weight gain, so why is this advice apparently not acted on? Is it so difficult to follow? As we have discussed above, there are various factors which influence dietary behaviour which may lead to weight gain above or below guidelines, such as dietary beliefs, cravings and aversions.
However, the nature of the advice that women receive and their interpretation of that advice may also influence behaviour. As we also discuss in relation to physical activity in the next chapter, advice given by midwives and publications for pregnant women is often vague, recommending a 'healthy diet'. Here, as in the general population, if health education messages do not fit lay health models, they are less likely to be taken up (Ikeda, 1999; Lupton and Chapman, 1995). In other words, the form and content of the advice, the language used and directions for how to act on the advice have to be understood and integrated into what the woman knows and believes. For example, American adolescents interviewed by Skinner et al. (1996) said they would prefer to watch a video with a 'talking baby' or teenage actresses presenting the information than read a lea¯et or book. They also wanted more information about food than nutrients.
It should also be remembered that health professionals are not the only sources of advice; women have access, to varying degrees, to information from family, friends, magazines, books, television and other media and increasingly to the internet. For example, Lewallen (2004) found that family members were a common source of advice for low-income pregnant women in the US, and in our study of a varied group of women in the UK (Pattison and Bhagrath, 2003), less highly educated women and women from minority ethnic groups were less likely to use books, magazines and the internet. These variations are important because the type and content of advice from different sources vary and may conflict.
The majority of women in Norman and Adams' (1970) study reported that they had made changes in their diet because of dietary advice from health professionals. Orr and Simmons (1979) assessed patients' satisfaction with dietary advice received and found that the majority of patients expressed satisfaction with the amount of information received. A study by Cogswell et al. (1999) revealed that reported advice during pregnancy is strongly associated with actual weight gain. However, about half of the women in their study reported having received no advice, or inappropriate advice from healthcare professionals about weight gain during pregnancy: Overweight women were more likely to report having received advice to gain weight greater than the recommended amount during pregnancy. What these studies have in common is that the reported behaviour fits in with the reported advice. Thus, women have created a narrative which is internally consistent, sanctioning behaviour by providing an account of official advice.
In our study (Pattison and Bhagrath, 2003) 30 per cent reported having received no advice from their midwife or general practitioner. The majority of women who remembered receiving advice said they would have liked more than simply being advised to 'eat healthily' and explanations of why certain foods should be avoided. Women who were more highly educated and expecting their first child were most likely to seek out alternative sources of information, particularly books, magazines and the internet. Often, nutritional advice is given in antenatal clinics, however not all women actually attend these clinics and the women who do are usually found to be of higher than average socioeconomic, educational and occupational status, characteristics which are also found to be associated with already better than average nutritional knowledge and dietary practices (Fowles, 2002). This implies that populations that are more in need of additional advice and information are less likely to receive it.
Midwives in the UK no longer specify optimum levels of weight gain for most women, and for several years women were not weighed. Fowles (2002) found that most women had inadequate general nutritional knowledge and therefore, hardly surprisingly, their dietary intake did not meet all the nutritional requirements of pregnancy. Women are usually encouraged to improve their diet during pregnancy but information on how to improve diet is vague. Most advice mentions fresh fruit and vegetables or eating a 'balanced diet'. However, this kind of advice, to simply eat 'more healthily' throughout pregnancy, is not sufficient if women do not have the knowledge for it to act as a prompt to particular behaviours. Furthermore, as we have discussed above, traditional beliefs about what constitutes a healthy diet during pregnancy are likely to be at odds with current nutritional theories.
The vagueness of advice on positively improving diet during pregnancy is in stark contrast to advice on what should be avoided. Often starting as food scares in newspapers, or on television and radio news programmes, advice about avoiding hazardous foodstuffs is often extremely specific. As we said in the introduction to this chapter, pregnant women often find themselves the focus of food scares. They may be a specific focus of information because a link has been made between a food and foetal or, more rarely, maternal health. They may also be targeted because they are perceived as vulnerable to health hazards. Women are more vulnerable, of course, during pregnancy because of their suppressed immune system (necessary so their body does not reject the foreign tissue of their baby). However, in this instance the person perceived to be vulnerable is more likely to be the baby; the targeting of the mother stems from their custodianship of their baby's health.
During the period of our study (Pattison and Bhagrath, 2003) there were two main food scares directed at pregnant women. One concerned coffee, which was linked to stillbirth and early infant death in an epidemiological study published in the British Medical Journal (Wisborg et al., 2003). This finding was taken up by various newspapers and other media in the UK. The second concerned tuna fish, and followed on from previous studies on the mercury content of oily sea fish such as marlin and shark. These other fish do not form a major part of British women's diets.
However, when it was found that tuna may also contain high levels of mercury, this information was quickly spread in the media and incorporated into The Food Standard Agency guidelines. Of the women we interviewed only about a third had heard either or both the coffee and tuna stories. However, 72 per cent of those who had heard responded by eliminating or drastically reducing their intake of the food, but the others did not change their consumption at all. This emphasises one of the harmful effects of food scares. While reducing coffee intake is unlikely to harm women, it may make them feel uncomfortable. However, tuna is generally regarded as a healthy food, so the elimination of it is not likely to improve women's diets.
In a survey commissioned by SMA (a baby milk producer) in 2003, 558 mothers with children aged between 12 months and two years in the UK and the Republic of Ireland were questioned about their diet during pregnancy and what they believed about foods that constituted a healthy diet. The results showed that while they were aware of food scares, they did not always know or understand the research findings which formed their basis. Some foods were regarded as unsafe through a generalisation from another food. So 60 per cent of women believed that cottage cheese, which is a safe, low-fat source of protein, was unsafe because they failed to make a distinction between this and soft cheeses which may carry listeria.
However, in other cases women failed to generalise from one food to others which were similarly hazardous. For example, most avoided or reduced their intake of coffee, because of the risk of caffeine, yet 70 per cent believed that diet cola drinks, which also contain caffeine, were safe. In one of our studies we also found that women reduced their intake of coffee in an attempt to avoid caffeine, but increased their intake of other caffeine-containing drinks such as tea and cola drinks (Gross and Pattison, 1995).
Research that underpins dietary advice is often presented in a way which makes it very difficult to interpret. Take the following extract from the Babyworld website:
"Research published in 1999 suggested that high doses of vitamin C and vitamin E may help reduce the incidence of pre-eclampsia in women at high risk of developing the illness. Although this seems encouraging news, most experts remain unconvinced. First, the study was very small (only 160 women completed the study) so the results may not be accurate (a larger trial is being planned). Secondly, there is some doubt over the safety of the massive doses required of the two vitamins". (Hulme Hunter, 2005)
Women who are concerned about pre-eclampsia are advised to talk this study over with their obstetrician. However, the study is unattributed and is so heavily criticised that it would be difficult to imagine any woman feeling comfortable raising these findings if she does not have knowledge of scientific procedures, or access to medical journals to look up the study. It seems that an attempt not to blind readers with science has led to an oversimplified version, which will only have the effect of making women feel worried.
However, attempting to produce all the caveats and exceptions to advice given is also confusing and likely to make readers worried. Take, for example, the following extract from the BBC website:
"Research indicates that mothers who eat fish once a week are less likely to give birth prematurely. Oily fish eaten in pregnancy also helps with children's eyesight. However, when you're pregnant have no more than two portions of oily fish a week. Oily fish includes fresh tuna (not canned tuna, which does not count as oily fish), mackerel, sardines and trout. Avoid eating shark, swordfish and marlin and limit the amount of tuna to no more than two tuna steaks a week (weighing about 140g cooked or 170g raw) or four medium-size cans of tuna a week (with a drained weight of about 140g per can). This is because of the levels of mercury in these fish. At high levels, mercury can harm a baby's developing nervous system." (Welford, 2005).
Again the research is unattributed, and even undated, making it very difficult to trace, and there is not enough information to evaluate it. What constitutes oily fish here is unclear; the passage seems to suggest both that women should and should not eat tuna, fresh or canned. In an attempt to be accurate and all encompassing, the advice becomes controlling.
While information from research that is incorporated into professional leaflets and websites may be balanced, much of what appears in the media is not. For example, the research paper referred to above on coffee actually indicated that this was not really a problem for women who were drinking less than eight cups of coffee a day (Wisborg et al., 2003). Similarly, a later report by Bech et al. (2005) suggested that the risk of foetal death was only significantly higher if women drank more than four cups of coffee a day. However people tend to classify things as either safe or unsafe, so the media portrays foods in this way and the likelihood is that, if women act on food scares at all, they will avoid the apparently hazardous foods completely. The distinction between safe and unsafe foods also tends to vary across cultures and be embedded in more general eating habits. So people from European countries tend to regard wine as safe in moderation, whereas it is definitely on the list of things to avoid completely in the US, even though the research evidence on which advice is based is the same.
A final aspect of food scares to consider is that they nearly always come too late for pregnant women to act on them. Finding out that tuna contains mercury when you are several weeks into pregnancy, and you have already consumed large quantities of this formerly healthy food, is only likely to induce guilt and anxiety. Neither of these emotions are likely to increase the health of women or their babies. The BUPA website even gives a list of foods that women should have avoided before pregnancy:
There are also certain foods that women should avoid pre-pregnancy.
These include:
a. liver and large quantities of vitamin A in supplements,
b. unpasteurized dairy products,
c. raw eggs,
d. pâtés,
e. soft cheese.
(BUPA, 2005)
There is little evidence to support this draconian advice and since so many pregnancies are not planned with the precision required by this, many women will not have been able to act on it anyway.
Concluding remarks
Research on diet and dietary change during pregnancy is unusual in several respects. One important characteristic is the amount of research which has been carried out in countries other than those in the developed world. While little of this work could be said to be cross-cultural, it does at least give us some insight into how pregnancy is experienced by the women outside the mainstream focus. The differences and similarities between women of different cultures are illuminating in that they show how important it is to consider the context and cultural underpinnings of women's lives.
We have reiterated several times in this chapter that women's eating behaviour during pregnancy is studied out of the context of their everyday lives and history. In particular, little account is taken of dietary restraint and dieting behaviour before pregnancy. Yet at the same time the exception to this is a fascination with the dietary habits of what to most researchers is 'the other', notably pica.
A further unusual feature of research on diet is the direct impact that research has on sanctioning women's behaviour during pregnancy. Epidemiological studies which show some association between what women have eaten during pregnancy and subsequent pregnancy outcomes make almost daily appearances in the media. The risks associated with food types are amplified through newspapers, magazines and television, and, perhaps most pervasively, through the internet. Often these studies are later refuted, dealing as they often do with statistically very small increases in risk. However, few women in the developed world can be unaware of the food scares and risk messages directed at them. Yet, what use they make of this information, or the effect of receiving risk messages, often too late to act, on psychological health still goes largely unexplored.
By Harriet Gross and Helen Pattison in "Sanctioning Pregnancy - A Psychological Perspective On The Paradoxes and Culture of Research",Routledge, UK, 2007, excerpts pp.75-94. Digitized, adapted and illustrated to be posted by Leopoldo Costa.