5. When does a woman's diet affect the outcome of her pregnancy?

OPTIMUM BIRTHWEIGHT AND MATERNAL NUTRITION

If a group of women is found to have babies in the optimum birthweight range it may be assumed that their diet was compatible with this happy outcome of pregnancy. The birthweight range associated with the lowest perinatal mortalities may be called the optimum birthweight range. This range is associated with the lowest incidence of birth defects as illustrated in Figure 5.1, which shows the incidence of perinatal deaths attributed to congenital anomalies for England and Wales. In the official statistics of several countries this optimum birthweight range with lowest mortalities is 3,500 to 4,500g, including the USA (US National Center for Health Statistics, 1972); Germany (Beck et al., 1978); Scotland (Scottish Health Service, 1977); and Sweden (Sveriges Officiella Statistik, 1979). The optimum birthweight range in the official statistics of England and Wales is 3,500 to 4,OOOg, but figures for the range 4,000 to 4,500g are not published. Moreover infant mortalities in all countries are also very highly correlated negatively with birthweight, the mortality increasing only slowly at birthweights below 3,500g, more rapidly below 3,000g and more rapidly still below 2,500g. Figure 5.2 shows death-rates under 12 months of age by birthweight for England and Wales.

The diet eaten by 165 pregnant women whose babies were born into the to optimum birthweight range was studied in a population of London women. Their nutrient intakes are set out in Table 5.1 with the diet of 28 mothers whose babies proved to be of low birthweight. The table is based on a study id of diet in early pregnancy by Doyle et al. (1989b) of the Nuffield Institute of Comparative Medicine. The mothers were all patients at the Salvation Army Mothers' Hospital, Hackney, and in social class composition were approximately but not precisely representative of the British population, except that mothers of Asian origin were not included. Each mother kept a diary of food and drink for one week after interview and instruction. The diet was recorded in all cases around the end of the first trimester of pregnancy. The diet of the reference mothers in Table 5.1 is called the "reference diet" and it is implied that it is a safe or desirable diet because shown to be compatible with a satisfactory birth outcome. The reference mothers are seen to have eaten more of all nutrients than the mothers of the low birthweight babies. For example the reference mothers consumed 74.5g/day average of protein while the mothers of the low birthweight babies consumed only 62.8g/day. Table 5.1 throws doubt on any suggestion that a protein intake below 75g/day is acceptable for women wishing to conceive or during pregnancy. This finding is in line with the old, but classical, papers of Bertha Burke and colleagues of the Harvard School of Public Health, published in the early 1 940s, which showed associations of maternal protein consumption with birthweight and with birth length, illustrated in Figures 5.3 and 5.4. A low protein intake depresses gonadotrophin secretion, and for this reason alone the reduction of protein

TABLE 5.1

MEAN DAILY NUTRIENT INTAKES OF MOTHERS OF LOW BIRTHWEIGHT BABIES LESS THAN 2,500g AND OF REFERENCE MOTHERS OF BABIES IN THE OPTIMUM BIRTHWEIGHT RANGE 3,500 TO 4,500g; 193 LONDON MOTHERS

		n = 28	n = 165
birthweight range		2,500g	3,5OO-4,500g
energy	kcal	1,642	1,974
	MJ	6.87	8.26
protein	g	62.8	74.5
vitamins
thiamin (B1)	mg	1.0	1.2
riboflavin (B2)	mg	1.5	2.0
niacin	mg	12.3	16.1
pyridoxine (B6)	mg	1.2	1.5
folic acid	mcg	161	201
pantothenic acid	mg	3.7	4.4
minerals
calcium	mg	761	953
iron	mg	9.4	12.9
magnesium	mg	209	283
phosphorus	mg	1,039	1,316
potassium	mg	2,493	2,993
sodium	mg	2,026	2,688
zinc	mg	8.16	10.2

Source: Doyle et al., 1990.

intake below 75g/day could be damaging to pregnancy outcome for some women. However it was seen in the last chapter that maternal protein intake begins to affect gonadotrophin secretion and pregnancy outcome already be fore pregnancy, during ovulatory maturation. The maternal protein intakes recorded by Burke and in the Hackney study during pregnancy were probably similar to the intakes of the women's habitual diets and to their intakes around the time of conception. Large trials of protein supplementation during the last two trimesters of pregnancy have never produced increases in birth dimensions in any way comparable to the differences in dimensions associated with differences in diet in the Hackney study (Rush, 1986). The extensively documented Women Infant and Child Supplemental Feeding Program (WIC) in the USA showed little effect on birthweight or other pregnancy outcome (Rush et al., 1988), except in one set of circumstances when mothers who had received extra cereals, eggs and fruit juices during the subsequent interpregnancy interval and during the early months of the next pregnancy. The babies of the second pregnancy then had a significantly higher birthweight, birth length and a reduced risk of low birthweight (Caan et at., 1987).

Maternal protein intake is invariably correlated with the intake of other nutrients and the associations in Figures 5.3 and 5.4. may have been wholly or partly caused by these other nutrients (Hurley, 1979). The Hackney study had a total of some 100,000 items of information and analysed not only the association of food and nutrient intake with birthweight but also with newborn length, head circumference, weight of placenta and length of gestation.

B VITAMINS AND BIRTH DIMENSIONS

Birth dimensions were found to be more significantly associated with maternal consumption of some B vitamins and with some minerals, notably magnesium, than with protein or calorie consumption. B vitamins are discussed first. Of all the constituents of maternal diet in the Hackney study B vitamins were most closely correlated with birth dimensions. Figure 5.5 shows the association of low birthweight with maternal intake of 4 B vitamins.

The maternal intakes of thiamin and niacin were most closely correlated with birthweight, birth length and head circumference, with maternal intakes of riboflavin and pyridoxine next in significance. The hypothalamic pituitary-gonadal axis, as discussed in Chapter Four, is highly sensitive in other mammals to the intake of the B vitamins. Low intakes depressed gonadotrophin secretion and thereby follicular and embryonic development before there is any direct effect on germ or embryonic cells. The low intakes of B vitamins of the mothers at highest risk of a low birthweight baby, seen in Figure 5.5, may well have been associated with depressed levels of pituitary ho..rmones which would increase the risk of low birthweight by slowing down ovulatory maturation before conception. In the Hackney study protein intake was also correlated with birth dimen sions but at a lower level of significance than thiamin or niacin intakes and the correlations of maternal protein intakes with birth length and head circumference were only just significant (p = 0.012 & p = 0.019). This does not indicate that protein is unimportant for birth outcome but would be explained if there were many more Hackney women with B vitamin intakes than protein intakes low enough to depress pituitary hormone levels.

Studies of diets even in the most prosperous countries show that important minorities of women have daily intakes of B vitamins below recommended levels and have enzyme saturation levels involving "moderate" or "high" risk. Thus the Swiss "Basle Study" of 6,400 employed adults found 22.8 per cent of women with depressed enzyme saturation for thiamin, 7.5 per cent for riboflavin, 25.9 per cent for pyridoxine (Ritzel, 1975). Another study of 200 pregnant women attending the Basel University Women's CLinic found 33 per cent with depressed enzyme saturation for thiamin, 37 per cent for riboflavin and 27 per cent for pyridoxine (Decker et at., 1975). A paper from the University of Kiel reported a significant correlation of low birthweight and low maternal thiamin intake in the first trimester, and referred to the "surprisingly high proportion" of 229 women from 6 German teaching hospitals whose thiamin intake was inadequate to ensure an acceptable level of enzyme saturation (Kubier & Moch, 1975). The German Ministry of Youth, Family and Health sponsored a study by the University of Heidelberg into the nutrition of young men and women 20 to 40 years of age, and 621 women participated. The report commented particularly on thiamin intakes of both men and women and said (Arab et at., 1982):

"Concurrently, in both sexes approximately 5 per cent show enzyme activation levels of transketolase that indicate extremely low coenzyme status, and another 25 per cent with borderline deficiency levels. The mean levels are comparable to those reported in the Swiss study of the Basel population but the prevalence of low levels is much greater in Heidelberg."

Dividing cells have much higher energy requirements than cells that are not dividing. Ovulatory maturation and embryonic development involve the higher rates of cell replication in the human life cycle and therefore a high local availability of intracellular energy. DNA and RNA and protein synthesis require energy. Thiamin is an essential coenzyme for 3 enzymes concerned with the release of intracellular energy by the oxidation of glucose. Research on wound repair and scar development has shown that enzyme saturation with thiamin can be the factor limiting DNA and protein synthesis (Alvarez & Gilbreath, 1982; Im et al., 1975). It would not therefore be surprising to find that low levels of enzyme saturation with thiamin might limit birthweight. In the volume on the American Recommended Dietary Allowances (US National Academy of Sciences, 1980) it is explained that the RDAs for thiamin of 1.1mg/day for women aged 19-22, and 1.0mg/day aged 23-50, were not adequate for enzyme saturation which indeed required a much higher intake. Because quite low levels of enzyme saturation are adequate for most body cells it does not follow, as assumed in the conception of the RDAs, that the rapid cell replication involved in the early stages of reproduction does not benefit from 100 per cent enzyme saturation. The average intake of thiamin of the reference women in Table 5.1 is seen to have been 1.2mg/day and of the mothers of the low birth-weight babies only 1.0mg/day, an intake close to the RDA of 1.0mg/day. Figure 5.5 shows that the risk of a low birthweight baby being born apparently decreased at least up to The American RDAs for B vitamins may be satisfactory for women not intending to reproduce, but the maternal intakes in Figures 5.5 suggest that they are too low during the period preceding pregnancy or during pregnancy. The RDAs for pregnancy are increased to 1.4mg/day for thiamin, 1.5mg/day for riboflavin, 15mg/day for niacin, and 2.6mg/day for pyridoxine, but these higher levels, if they were followed, would only take effect after any deficiency had already caused an irreversible slow-down in follicular or embryonic development and after the placenta had developed to protect the foetus.

Nutritional deficiencies have their effects partly by their direct influence on the gonads and embryo and partly by their effect on the endocrine system. In other mammals the hypothalamus is more sensitive to a deficiency of thiamin than are the gonads or embryo. It is likely therefore that the associations found in the Hackney study and illustrated in Figure 5.5 were a consequence of depression of the hypothalamic-pituitary-ovarian axis. The hypothalamus in other mammals reacts to a severe deficiency of any of these B vitamins by inhibiting gonadotrophin secretion and so causing infertility (Watteville et al., 1954). This reaction to an unsatisfactory diet, including a diet deficient in B vitamins, must have had survival value by inhibiting reproduction at times when fresh food was in short supply. However this depression of gonadotrophin secretion is not a simple on-off mechanism and there is a penumbra between fertility and infertility illustrated in Figure 5.6. Within this penumbra hormonal and nutritional levels are not low enough to cause infertility but are still too low so that follicular growth is retarded, cpus latea arew too small and embryonic growth is retarded.

The Heidelberg report concluded that the greatest nutritional threat to uire women"s health was too low a consumption of a number of nutrients notably the thiamin, pyridoxine, folic acid, calcium and iron. The report said:

"Despite seemingly sufficient energy intakes, there is apparently art insufficient intake of some the essential nutrients, particularly the B vitamins... The lower caloric needs of our passive society are met by lower intakes and when these are partially met by caloriclow sources which contribute few essential vitamins and minerals a deficiency is inevitable."

Particularly before and around the time of conception a diet with a high density of B vitamins and essential minerals is desirable to maintain high levels offor enzyme saturation.

Newborn head circumference in the Hackney study was significantly correlated with matemal intakes of thiamin and niacin (p>O.OO 1) as illustrated in Figures 5.7 and 5.8.

The high statistical significance is largely attributable to~ent the women who had babies with head circumferences below 34cm and particularly below 33cm. The average thiamin intake of the reference women in Table 5.1 was 1.2mg/day. This figure was only an average and made no allowance for variations in the requirements of individual women. As recommendation 1.2mg/day would have no safety margin to allow for individual can variations and it is seen in Figure 5.5 that the risk of low birthweight continues to decline up to intakes of 1.46mg/day. The mothers of the babies with the ice, largest birth head circumference in Figure 5.7 had an average thiamin intake mg of 1.28mg/day. Both Figures 5.5 and 5.7 suggest that thiamin intakes highericy than 1.2mg/day may be desirable for some women. Animal experiments showfor that the efficiency of energy utilization increases with thiamin intake with~rean rapidly diminishing, but nevertheless real, returns up to 2 or 3 times normal:he intake. The thiamin intake of the reference women of 1.2mg/day is the lowest intake that can be regarded as acceptable for women anticipating pregnancy, but intakes up to 2mg/day may be desirable for some women. Similarly the niacin intake of the reference women was 16.1mg/day which may be regarded as the minimum desirable intake of women anticipating pregnancy, but Figure 5.5 suggests that some women may benefit from intakes over 19mg/day. Intakes of niacin up to 25mg/day may be necessary to allow for variations between women. Individual requirements depend on many factors and in particular on total energy requirements.

MAGNESIUM, AND OTHER ESSENTIAL MINERALS AND BIRTH DIMENSIONS

The placenta can extract vitamins from the mother's blood and transfer them at higher concentration to the foetus. The vitamin concentration in cord blood may be 5 or 10 times the concentration in maternal blood. This capacity of the placenta is sometimes referred to as the vitamin pump. There is no similar placental pump for minerals. The concentrations of minerals in maternal and foetal blood are compared in Table 5.2 and the differences are seen to be very small, in most cases statistically non-significant, and in no way comparable to the large differences often reported for vitamins (Gontzea, 1965; Ham felt & Tuvemo, 1972; Van den Berg et al., 1978). The foetus is not well-protected by the placenta from low mineral concentrations in maternal blood.

TABLE 5.2

SIMILARITY OF LEVELS OF MINERALS IN MATERNAL AND CORD WHOLE BLOOD; 25 NORMAL BIRTHS, OXFORD, ENGLAND

.

	mcg/ml
	maternal		cord
	mean	s.d	mean	s.d.
iron	442	25	433	30
calcium	60	6	61	6
magnesium	36.2	3.9	36.7	3.4
zinc	7.7	1.1	7.6	0.7
copper	1.04	0.17	0.94	0.13
selenium	0.098	0.013	0.099	0.011
iodine	0.036	0.009	0.038	0.01
cobalt	0.021	0.007	0.024	0.005
manganese	0.012	0.005	0.015	0.007
chromium	0.007	0.003	0.009	0.005

Source: Bryce-Smith, 1985

In the Hackney study 54 out of 513 babies were born before 37 weeks' gestation (Doyle et al. 1989a). The mineralintakes of the mothers of these 54 preterm babies are compared in Table 5.3 with the intakes of the mothers of babies born after 37 weeks gestation. There were no significant associations of preterm birth with maternal energy or protein intake or with intake of any vitamin except thiamin. Length of gestation is, however, difficult to record accurately and it is likely that the lower significance of the correlations of nutrient intakes with length of gestation compared with birthweight is partly a consequence of errors in recording.

The close association of maternal mineral intake and birthweight are shown in Table 5.4 for the mothers with babies of less than median birthweight. The

TABLE 5.3

THE MEDIAN MINERAL INTAKES OF MOTHERS OF BABIES BORN PRETERM BEFORE 37 WEEKS OR AFTER 37 WEEKS GESTATION RANKED BY STATISTICAL SIGNIFICANCE OF DIFFERENCE; 513 LONDON MOTHERS

	n=54	n=459	p
	mg/day	mg/day
magnesium	227	259	0.005
sodium	2268	2566	0.006
chlorine	3624	4016	0.008
iron	10.4	11.7	0.010
calcium	805	889	0.020
copper	1.42	1.51	0.020
phosphorus	1120	1244	0.021

Source: Doyle et al., 1989a

TABLE 5.4

CORRELATIONS OF BIRTHWEIGHT WITH MATERNAL MINERAL INTAKES OF MOTHERS WITH BABIES BELOW THE MEDIAN BIRTHWEIGHT OF 3,270g; 255 LONDON MOTHERS

mineral	correlation
	r	less than p
magnesium	0.253	0.001
iron	247	0.001
phosphorus	0.243	0.001
chlorine	0.240	0.001
zinc	0.238	0.001
sodium	0.237	0.001
potassium	0.208	0.001
calcium	0.184	0.002

>Source: Doyle et al., 1990

highest correlation is again seen to be with magnesium. None of the correlations for the same minerals were statistically significant for the mothers of babies of birthweight above the median of 3,270g. The probability of birthweight being less than 3,OOOg by maternal magnesium intake is illustrated in Figure 5.9. The association of low birthweight and low maternal magnesium intake has been the subject of studies from a number of countries with some indication that it is the commonest mineral deficiency in many populations (Wynn & Wynn, 1988b). Trials of m2agnesium supplementation of women during pregnancy have reported significant reductions in the incidence of low birthweight and preterm birth (Spiitling & Spiitling, 1988). Magnesium is another nutrient essential for intracellular energy metabolism and participates in the same chemical reactions as thiamin. Deficiency of mag nesium depresses the level of thiamin in body tissues as illustrated in Figures 5.10 from animal experiments (Itokawa et al., 1973, 1974; Wynn & Wynn, 1988b). Thiamin cannot be used and is excreted if magnesium is deficient, with associated depression of intracellular energy metabolism. Magnesium is an essential cofactor for over 300 known enzymes, a much larger number than have been shown to require thiamin. It was noted in the last chapter that magnesium deficiency is mutagenic.

However maternal magnesium intake, and the maternal intake of all the other minerals in Table 5.4, were only associated with birthweight in the lower half of the birthweight range. This is illustrated in Figure 5.11 showing two cumulative average curves. These curves illustrate the important finding of the Hackney studies that the significant associations of maternal intakes of all nutrients and birth dimensions were limited to mothers of the smaller babies. In Figure 5.11 the cumulative average curve beginning from low birthweights falls to around 62 per cent of the reference diet value at birthweight around 2,000g. but rises to around the reference diet value somewhat below the median birthweight of 3,270g. There is a plateau at higher birthweights indicating no significant association between birthweight and magnesium intake. The statistical significance of the difference between the two cumulative average curves may be computed for any birthweight and is found to be significant only below the median birthweight. This is an important conclusion because it follows that any major influence of nutritional deficiency in causing reduced birth dimensions must have been confined to a minority of mothers with the smaller babies

WHEN DO NUTRITIONAL DEFICIENCIES HAVE THEIR EFFECTS?

There are several streams of evidence suggesting that the highest susceptibility to nutnent deficiency in the female is during ovulatory maturation and early embryonic development. Animal experiments show that deficiencies of some nutrients depress hormone levels and can cause infertility by preventing ovulation as illustrated diagrammatically in Figure 5.6. Lesser degrees of nutritional deficiency and depression of the hormonal status can cause delayed ovulation, reduced rates of early cleavage, low birthweight and congenital malformations in animals in the area described in Figure 5.6 as the penumbra. These results are produced before or around the time of mating. There is evidence that the supplementation of human pregnancy during the last two trimesters may benefit the mother and improve breast feeding but has very little effect on the foetus or pregnancy outcome, whereas the evidence of 50 years, from the Harvard studies of Burke to the London studies in Hackney of Doyle, show highly significant correlations of matemal nutritional status and birth outcome, associations that can only have had their origin very early in pregnancy. In the Hackney study vitamin and mineral supplements during the last two trimesters of pregnancy produced no significant effect on birth dimensions of the babies. In contrast vitamin supplements given before and around the time of conception in the trials of Smithells and colleagues in Leeds produced significant reductions in the risk of repetition of neural tube defects (Smithells et al., 1983).

There is a long historical record from famines and during wars and the aftermath of wars showing the effect of food shortage on pregnancy outcome (Wynn & Wynn 1979, 1981b, 1981c). Many epidemics of perinatal mortality and congenital malformations were recorded following these periods of food shortage. The malformations such as neural tube defects are known to be of early pregnancy origin. The epidemic in Holland of congenital malformations following the "hunger winter" of 1914975 is illustrated in Figure 5.12. It is seen that the highest incidence of deaths caused by malformations was among babies conceived during the food shortage or during the following 4 months. The most serious consequences followed when the food shortage was before and around the time of conception. If a woman was short of food before or around the time that she conceived the risks to her baby were much more serious than if she was well fed at conception but short of food during pregnancy. The Dutch experience was repeated in Germany and other European cities.

The Dutch hunger produced a fall in the birth-rate of about 50 per cent, caused in part by stillbirths and early miscarriages. Such food shortages are today only a memory in the westem world but how far does individual anorexia still cause infertility? The long history of anorexia nervosa and its effect on fertility goes back to the 19th century and also throws light on the effects of malnutrition on pregnancy outcome. The urinary and plasma gonadotrophins When Does a Woman's Diet Affect the Outcome of Her Pregnancy? in anorexic, underweight women are low and may be too low to support follicular development and ovulation (Vigersky & Loriaux, 1977; Brown et at., 1977). The hormones mainly affected are the pituitary hormones LH and FSH and the thyroid hormone T3, and it is the resting level of LII that is particularly correlated with body weight independently of calorie intake. It is primarily failure of the LH surge to reach a critical level necessary for ovulation that causes infertility (Warren, 1977). Self-imposed weight loss by dieting has been shown to be a common cause of infertility and amenorrhoea of more than 50 per cent of women attending some infertility clinics (Nillius, 1978; Knuth et at., 1977; Hirvonen, 1977; Bergh et at., 1978). Nearly 80 per cent of infertile women have been judged to be underweight. What then is underweight in this context?

Women's weight is a rough indicator of nutritional status and its decline was recorded during the European food shortages by hospital clinics. There have been many subsequent studies in Europe and America showing the associations of fertility and pregnancy outcome with prepregnancy weight. These studies point again to the importance of nutritional status before conception. In a Swedish study the mean prepregnancy weight of 161 mothers of low birthweight babies was 53.9kg (Bjerre & Bjerre, 1976). The average Swedish women's body weight at the onset of amenorrhoea has been shown to be 52kg (Fries, 1974). The mothers of the low birthweight babies had, therefore, prepregnancy weight only 1.9kg above the infertility threshold for Swedish women. The average prepregnancy weight of Swedish women is 62.4kg, 8.5kg higher than the mothers of the low birthweight babies. Of the 161 low birth-weight babies in this study 28 died, 14 had congenital malformations or cerebral palsy and 10 had minimal brain dysfunction at age 5. These were penumbral children.

Drugs may be used to induce ovulation. However a study in London teaching hospitals showed that amenorrhoeic women had a 25 per cent risk of having a small-for-dates, growth retarded baby following induced ovulation, and this risk increased to 54 per cent if the women were underweight (van der Spuy et al., 1988). This important study extended to women the findings of Kinzey and Srebnik (1963) who showed that the maintenance of pregnancy with estrone and progesterone in protein deficient animals produced low birth-weight offspring.

Prepregnancy maternal weight was highly correlated with birthweight in the Hackney study, with a correlation coefficient of 0.235 (Doyle et al., 1989b). This compares with correlation coefficients of 0.260 for pregnancies of American black mothers and 0.290 for white mothers in the US Collaborative Perinatal Study (Niswander & Jackson, 1974). The percentages of low birthweight babies among 7,054 American white women and 8,175 black women in the Collaborative Study are shown in Figure 5.13 for different prepregnancy weights. The association with height is eliminated in this figure by selecting only women in the height range 160 to 165cm (Niswander & Gordon, 1972). The 165 Hackney reference mothers had an average weight of 63kg (10 stone 9 ibs) a figure close to the average for Swedish women of 62.4kg.

Surveys over many years have shown an association between maternal height, birthweight and perinatal mortality. The Hackney study found a significant association between maternal height and nutrient intake, shown in Table 5.5. Maternal shortness was a risk factor associated with eating less of a poorer quality diet and a poorer social background. Prepregnancy weight is however in all surveys much more closely associated with birthweight and other birth dimensions than is prepregnancy height and the Hackney study found no correlation between social class and prepregnancy weight.

TABLE 5.5

CORRELATION OF MATERNAL HEIGHT AND MATERNAL NUTRIENT INTAKES WITH SIGNIFICANCE BETTER THAN p = 0.01; 513 LONDON WOMEN

fibre	0.162	0.001
magnesium	0.145	0.001
niacin	0.137	0.001
folic acid	0.136	0.001
thiamin	0.127	0.002
fat	0.121	0.003
phosphorus	0.120	0.003
pyridoxine	0.119	0.003
energy	0.112	0.006
social class	0.188	0.001

Source: Doyle W., personal communication.

Maternal height may be taken into account by using the body mass index (BMI) (weight in kilograms divided by height in metres squared kg/rn2). The 165 Hackney reference women who had babies within the optimum weight range~ had a BMI of 23.7kg/rn2. This is close to the average for American women of reproductive age of 24.0kg/rn2. The BMI is easy to assess and the figure of 24.0kg/rn2 may be regarded as a satisfactory value for women anticipating pregnancy. Prepregnancy weight is a risk factor with risk increasing as the BMI falls below 24kg/rn2. The American infertility threshold is reported to be 20.8kg/rn2 for 50 per cent fertility (Frisch, 1977), with only 10 per cent fertility at 18.2kg/rn2. There is nothing inevitable about an unfavourable outcome of the pregnancy of the low weight mother, nor does an ample prepregnancy weight eliminate all risk. There are other risk factors such as low B vitamin, magnesium or iron intakes. Effective action to reduce these risks is possible. This can be better understood from the systematic studies in animal husbandry.

In reacting to body weight the hypothalamus is reacting to past nutrition over a period of weeks and months. One problem in animal husbandry is deciding how long the period of nutritional preparation between pregnancies should be. Failed insemination particularly in cattle is expensive not only because artificial insemination has to be paid for but because it increases the period of unproductive maintenance. All mammals have an infertility threshold and both long and short term effects are described in the extensive literature of animal husbandry. For example the farming practice of "flushing" has been defined in the context of sheep farming (Coop, 1966):

"The practice of giving ewes which are in fairly poor condition an improved diet for a few weeks before mating so that they are in a rapidly rising condition when they meet the ram."

Ovulation rate and numbers of surviving young depend on the nutritional status of cows, ewes and other domestic animals and in particular on energy balance before mating (Haresign & Cole, 1981). However a longer tenn effect of poor nutrition can only be partly overcome by flushing. Low body weight in animals is almost synonymous with "poor condition", but flushing is not just a matter of fattening but of supplying all those nutrients, provided for example by fresh pasture, which are needed by the hypothalamus and pituitary to provide an excellent hormonal profile. There are, indeed, long and short term effects of diet on fertility and pregnancy outcome. The effects of a less than adequate diet, if it has not been too inadequate, can be offset by an excellent diet for a few weeks before and around conception.

APPETITE AND EXERCISE

A recommendation that anyone should consume a better diet may be defeated by lack of appetite. Anorexia nervosa cannot be cured simply by advising the sufferer to eat more. Loss of appetite is an important clinical sign but difficult to interpret. It is common in most diseases involving the digestive system. Poor appetite is, however, a problem of important minorities in developed countries who have no diagnosable digestive illness.

In the Hackney study the majority of low birthweight babies were born to a minority of women with abnormally low energy intakes and low intakes of other nutrients. The case histories of many of these mothers reported a "poor appetite". Of 1,582 pregnant women attending the Montreal Diet Dispensary 19.7 per cent had an intake below 1,719 kcal/day and 9.8 per cent below 1,453 kcal/day at first attendance and many of these women had poor appetites. A Canadian national survey showed that 25 per cent of pregnant women were consuming less than 1,655 kilocalories (Canada, 1975). American surveys show that, taking only women with incomes above the poverty level, there When Does a Woman's Diet Affect the Outcome of Her Pregnancy? are substantial minorities with very low calorie intakes (US National Center for Health Statistics, 1979).

Appetite is psychosomatic and the somatic component includes the effect on appetite of the food actually consumed. Appetite grows by what it feeds on. Poor quality diet can cause both conscious and unconscious loss of appetite. Thiamin deficiency causes loss of appetite which has been described as "more specific and severe" than the loss of appetite caused by deficiencies of other vitamins or of amino acids (Gubler, 1982). Deficiencies of riboflavin, niacin, pyridoxine, folic acid, biotin and vitamin C have been reported to depress appetite. Deficiencies of minerals including magnesium, phosphorus, sodium and zinc depress appetite as also do excesses of calcium, iron, manganese and sodium (Werbach, 1988; Tremoliere, 1977).

What is eaten depends on appetite to the extent that availability is under control. Nutritional requirements are increased by exercise which generally increases appetite. Experiments on athletes expending between 3,000 and 4,500 kilocalories a day found that they adapted their calorie food consumption with a time lag of about 2 days. They ate more or less according to their energy requirements hardly aware of their changes in appetite (Tremoliere, 1977). Exercise in a female athlete can, however, result in amenorrhoea as a consequence of depression of the hypothalamic-pituitary-ovarian axis with the same effect on the axis as a poor diet. Amenorrhoea has been reported to affect 50 per cent of competitive runners, 44 per cent of ballet dancers, 25 per cent of non-competitive runners, 12 per cent of swimmers and cyclists (Calabrese et al., 1983; Sanbom etal., 1982). Amenorrhoea was for long considered to be a benign side effect of athletic training, but there is now evidence that the low levels of oestradiol with which it is associated produce lower bone density and predispose to osteoporosis (Nelson et al., 1986). Comparison of two groups of eumenorrhoeic and amenorrhoeic women runners showed that the amenorrhoeic women had the lower average intakes of energy including carbohydrate, fat and protein. The study concluded that the amenorrhoea might be explained by inadequate diet, which would be expected to depress gonadotrophin secretion. Appetite alone does not ensure the consumption by some women athletes of a diet adequate to prevent amenorrhoea. Japanese studies have reported a serious aggravation of thiamin deficiency among poor male students from engaging in athletic pursuits (Hatanaka & Ueda, 1981). Riboflavin is another nutrient required in increased amount by women taking daily exercise. The present recommended dietary allowances in the U.K. and U.S.A. are borderline for the average sedentary woman, and riboflavin status rapidly deteriorates with moderate exercise such as jogging (Belko et al., 1983). Riboflavin deficiency, as we have seen, also causes hormonal imbalance.

Appetite alone can result in too high as well as too low a consumption and depends in complicated ways on food composition and even on the order in which different foods are consumed. Appetite cannot be relied upon to ensure an optimum food intake and hormonal balance.

FOOD EATEN BY MOTHERS OF LOW AND OF OPTIMUM BIRTHWEIGHT BABIES

It was seen in Table 5.1 that the mothers of low birthweight babies in Hackney ate less on average than the mothers of babies of optimum birthweight; the reference mothers ate more of virtually every essential nutrient. The foods recorded in the diaries of the reference women were different also in quality. The difference is summarized using food groups in Table 5.6. The first food group in this table is breakfast cereals including muesli, oats, nuts and seeds. This food group had the highest density of B vitamins and contributed more than 20 per cent of thiamin and niacin to the total diet and was also responsible for around 40 per cent of the difference in the intakes of B vitamins between these two groups of mothers. This same food group also made a major contribution to the intake of magnesium and other minerals, and a 22 per cent contribution to the difference in energy intake between the two groups. This first group alone emphasizes the importance of breakfast. The next food group eggs and egg dishes shows a difference of 121 KJ between the two groups of women, a difference much greater than for bacon and ham which was only l9kJ and was included in the meat group at the bottom of the table. The maternal consumption of wholemeal bread is also seen to be associated with higher birthweight and of white bread with lower birthweight. The first three groups in the table appear to be relevant to the breakfast menu.

The next two groups "sugar and jams", "biscuits and cakes" contributed together a quarter of the difference in calorie intake between the two groups of mothers but only about 4 per cent of the difference in magnesium intake and less still for the intake of vitamins. The reference mothers are then seen to have had appreciably higher intakes of fats and of fish than the mothers of the low birthweight babies. However the next important difference between the two groups of mothers was in the consumption of dairy produce which was responsible for 28 per cent of the difference in energy intake and a similar percentage difference for many minerals such as calcium and magnesium and vitamins such as riboflavin.

"Vegetables" made a major contribution to B vitamin intake and to the maternal intake of important minerals including magnesium. Vegetables were, indeed, the most important contributor of magnesium with dairy produce in second place and were more important than dairy produce as a source of B Vitamins.

TABLE 5.6

DIET BY MAJOR FOOD GROUPS OF MOTHERS OF LOW BIRTHWEIGHT BABIES 2,500g AND OF REFERENCE MOTHERS WiTH BABIES BORN BETWEEN 3,500 AND 4,500g; 193 LONDON MOTHERS

	intake of foods	in kilojoules/day
	a	b
food group	mothers of babies	mothers of babies	a as
	less than 2,500g	3,5OO-4,500g	percentage of b
	n=28	n=165
breakfast cereals,	339	632	53.6
muesli, oats, nuts
and seeds
eggs and egg dishes	179	300	59.7
wholemeal bread	128	203	63.1
sugar and jams	310	455	68.2
biscuits and cakes	515	722	71.3
fats (marg., butter,	328	456	71.9
oils)
fish	138	187	73.9
dairy produce	1,348	1,732	77.8
fruit	280	351	79.7
vegetables	934	1,087	85.9
potatoes	644	703	91.5
meat products	469	468	100.2
white bread	620	598	103.6
meat	873	842	103.7
soft drinks	286	232	123.3
total	6,820	8,170	83.5

Source: Doyle et al., 1990