Maternal Nutrition and Lactational Amenorrhoea: Perceiving the Metabolic Costs

Rose E. Frisch

Harvard Center for Population Studies, Cambridge, Massachusetts 02138

The disruptive effects of undernutrition and intensive physical work on female reproductive ability are well documented. Undernutrition and weight loss delay menarche and cause cessation of already established ovulatory cycles (1—3). High- energy outputs also affect menstrual periodicity and the onset of menarche. Studies of ballet dancers (4,5) and women athletes (6-8) show that training at young ages before menarche delays menarche and that dancers and athletes have a high incidence of irregular cycles and amenorrhoea. The delay of menarche can be as much as 0.4 year (5 months) for every year of training (6). Some athletes and dancers had menarche as late as ages 19, 20, and 21 years (4-6). These disruptive effects of undernutrition and intensive exercise on female reproductive ability are reversible with weight gain (1,3,9,10) and/or cessation or reduction of physical activity after varying periods of time (5,6). The endocrinological basis for these findings is now also well documented. Studies of pituitary response to exogenous luteinizing hormone (LHRH) show that there is hypothalamic dysfunction associated with weight loss in the range of 10% to 15% of normal weight for height, as well as with the more extreme weight loss (30%) associated with anorexia nervosa (2,9). The degree of hypothalamic dysfunction is directly related to the amount of weight loss (2,9). The endocrinological changes which are associated with the hypothalamic dysfunction are essentially a reversion to a prepubertal endocrinological state (11). What do these findings on dancers, runners, and too-thin women have to do with observations of varying length of lactational amenorrhoea? One connection is the unexplained differences in natural fertility (fertility of couples who do not volun- tarily control the number of births in any way) (12). The natural fertility of the Bush people of the Kalahari desert is four babies in a reproductive lifetime (13). In contrast, the natural fertility of the Hutterites, a well-to-do religious sect who do not believe in contraception, is 10 to 11 (14). Louis Henry, who first observed the differences in natural fertility in historical populations (12), explained the observed differences by differences in the birth interval. The birth interval difference is also observed in many developing countries today, such as Bangladesh, and 65 66 NUTRITION, ACTIVITY, MENSTRUATION among the Bush people of the Kalahari. In the latter population the birth interval is almost 4 years (13); in Bangladesh it is approximately 3 years (15). A major component of this long birth interval is a long lactational amenorrhoea (13,15). No explanation was offered historically for the observed differences in natural fertility of populations other than the statement of the longer birth interval or general statements on differences in health and nutrition, without specification of the mechanism.

NUTRITION AND THE REPRODUCTIVE SPAN The findings on the direct effect of undernutrition and physical activity on age of menarche and regular ovulatory cycles cited above suggested a direct pathway from food available per capita to fecundity to fertility (16-18), in addition to the classical pathway of Malthus through a rise in mortality (19). Gopalan and Naidu (20) suggested such a pathway in 1972. The average number of births to poor couples in many developing countries today is approximately six or seven (15,17), similar to that observed in the past for poor couples in countries which are now developed. As the result of the improvements in mortality following the proper introduction of public health mea- sures in developing countries, an average of six or seven children per couple results in a very rapid rate of population growth. However, paradoxically, this total fertility rate is far below the observed human maximum of an average of 10 to 11 children born to well-nourished, non-contracepting couples, such as the Hutterites. Historical data for women of mid-nineteenth-century England and Scotland showed that slow growth to maturity of women and men due to undernutrition, hard work, and disease is correlated with a reproductive span which is shorter and less efficient than that of a well-nourished population (17,18). The submaximally nourished females and males are identifiable by a later average age of completion of growth, 20 to 21 years and 23 to 25 years, respectively, compared with that of contemporary, well-nourished females and males who complete their growth by ages 16 to 18 years and 20 to 21 years, respectively. The historical data showed that the slower growing women subsequently differed reproductively from well-nourished females, not only in having longer birth intervals, but also in having shorter lengths of the entire reproductive span, in later ages of peak nubility, and in lower levels of age-specific fertility in the reproductive years (21). There is thus a biological syndrome as it were in which submaximal rates of growth to maturity in a population, or in some classes of a population, are subsequently associated with a pattern of late mean age of menarche, early mean age of menopause, longer birth intervals, and more relative and absolute sterility (18,21) (Figs. 1 and 2). Such a reproductive pattern is observed also among the poor populations of many developing countries today when data on age of menarche, age of menopause, length of birth intervals, and pregnancy wastage are available. The differences in the length of the reproductive span and the timing of each reproductive event have been published in detail (Figs. 1 and 2). New data now 20-24yr 100 . Peok . / Nubility

25-29 80 Peak _l Age 18 Nubilty Nubility / Stage I Age 22 Nubility t Stage I f- 60

Adolescent Subfecundity I 40 / Adolescent ' Subfecundity

20- Premenopausal Subfecundity 41 Age of last birth ^-Menopause I Menopause 15 20 25 30 35 40 45 50 AGE (YEARS)

FIG. 1. The mid-nineteenth-century curve of female "procreative power" or reproductive ability (variation of the rate of child- bearing with age; maximum fertility rate, 100) compared with that of the well-nourished, non-contracepting modern Hutterites. The Hutterite fertility curve results in an average of 10 to 11 children; the 1850 to 1870 fertility curve in about 6 to 8 children. (From ref. 17.) 68 NUTRITION, ACTIVITY, MENSTRUATION

Fatness (%fot) •—Weight .WelgJ>t —^-s "Height Peok Reproductive Ability =100

24 28 32 AGE-YEARS

FIG. 2. The synchronization of peak female reproductive ability with the attainment of mature height, weight, and relative fatness. Timing and levels for 1950 to 1975. Slower growth to maturity is associated with a shortened, less efficient reproductive span, a later age of peak reproductive ability, and a decreased peak value. establish that fatter women have a later age of menopause (22) and that the mean age of menopause is now 52.0 years for United States women. There is thus evidence for a secular trend in age of menopause as well as for age of menarche. The strong association between differing rates of growth to maturity and differ- ential timing and efficiency of reproductive events makes it reasonable to hypoth- esize that the length of the lactational amenorrhoea is also affected by the previous rate of growth of the mother and her physical state as determined by nutritional intake and energy output. In addition, the physical demands of the infant must be included in the equation. Since the prepregnancy weight of the mother and, inde- pendently, her weight gain during pregnancy are the determinants of the infant birth weight, all of these factors are intercorrelated.

ENERGY COSTS OF LACTATION Although the neuro-endocrinological control of lactation was not understood in the second half of the nineteenth century, there was widespread understanding of the energy costs of lactation and reproduction in general. Darwin noted that "it is difficult to get a cow to give much milk and to fatten readily." Darwin also observed that "hard living retards the period at which animals conceive" and that "domestic animals which have regular, plentiful food without working to get it are more fertile than the corresponding wild animals" (23). Darwin concludes from his list of NUTRITION, ACTIVITY, MENSTRUATION 69 examples: "All these facts may be merged under a more general principle, namely that natural selection is continually trying to economise in every part of the organization." The idea is "not to waste nutriment" (24). Estimates of the metabolic cost of lactation for present-day women make even clearer the importance of the concept that "the production of milk costs something" (25). Hytten and Thomson (25) further state: "Satisfactory lactation represents the greatest nutritional stress imposed by a physiological process on the human body." The required input, estimated at between 800 and 1,000 calories/day (25,26) above requirements for maintenance and activity, is more than the output because the conversion of nutriments from diet into milk is not 100% efficient. The production of a live infant also costs something: a human pregnancy requires approximately 50,000 calories over and above normal metabolic requirements (27). It is therefore physiologically improbable to assume that women who have grown up slowly because of marginal food supplies, physical labour, and/or disease and who have a late age of menarche (15-16 years) would have the same length of lactational amenorrhoea and the same birth interval as non-contracepting women who grew up rapidly on high levels of nutrition, both quantitatively and qualitatively (particularly approximately 40% calories from fat), even if the pattern of nursing were identical.

Length of Postpartum Amenorrhoea Without Lactation In accord with the above view of the biological consistency of reproductive patterns, data which are not confounded by a possible difference in suckling patterns show a significant difference in length of postpartum amenorrhoea. In the excellent study by Salber et al. (28) (to explore the basis of international variations in the rates of carcinoma of the breast), the authors note that among Boston women who did not lactate, 60% had a return of menses by the end of the second month and 91% had a return of menses by the end of the third month. At the end of 4 months, only 1.8% of these Boston women were still amenorrhoeic after delivery; after 7 months, none was still amenorrhoeic. In contrast, the authors cite the data of Potter et al. from 11 Indian villages: the only nonlactating women in the Indian series were those whose infants were stillborn or had died neonatally. At the end of 4 months, 20% of these Indian women were still amenorrhoeic; at the end of 7 months, 10% were still amenorrhoeic. The median length of amenorrhoea of the Indian women was 60 days, only 5 days longer than the Boston median, illustrating the importance of determining the distribution of values in the two samples. As the authors point out (28), the differences between nonlactators in Boston and India at 4 and 7 months cannot be explained by any difference in partial breastfeeding or patterns of feeding. It would be useful to collect detailed data on the length of postpartum amenorrhoea of nonlactating mothers for populations varying in nutritional status and the amount of physical labour. Height and weight for each woman (not just weight alone) should also be collected. Among the Indian women whose infants survived 1 month or more, 98% were still lactating at the end of the sixth month and 74% were still amenorrhoeic. In 70 NUTRITION, ACTIVITY, MENSTRUATION

Boston, only 30% of those who were still lactating were also amenorrhoeic at that time. Nearly all of the Boston lactating mothers at 6 months were partial breastfeeders (28). But this fact in itself raises many questions rather than necessarily being an explanation, as will be discussed below.

Seasonal Variation in Length of Lactational Amenorrhoea The data of Chen et al. (15) from a prospective study in rural Bangladesh are also in accord with the view of a biologically consistent pattern of reproduction, unless there are outside interventions. These data showed that the onset of menses and ovulation following lactational amenorrhoea was seasonally distributed; the frequency of termination rose at the peak of the major rice harvest. These data support the finding of seasonal hormonal changes (29) and seasonal changes in length of lactational amenorrhoea and the efficiency of milk production reported by Paul et al. (30), Lunn et al. (31), and Whitehead et al. (32). Interestingly, Chen et al. also observed a seasonal variation in number of births (15). These authors noted that "while fertility was high by Western standards, it was well below the biological maximum, particularly in the 20-24 age group" and was not explained by the use of contraceptives or induced abortion. The data of Wilmsen (33) for the Bush people of the Kalahari, who have an even longer birth interval (approximately 44 months) than the Bangladeshi, also show a seasonal change in the birth rate correlated with seasonal changes in the food supply.

Length of Lactational Amenorrhoea Versus Calorie Supplies Corsini (34) tabulated average duration of breastfeeding (x) and average length of postpartum amenorrhoea (y) in nursing women from various regions of the world. For 16 rural areas, the mean values were x= 19.28, SD 6.55 months; y= 12.23, SD 4.82 months. In 14 urban areas, x= 10.96, SD 5.98 months; y = 7.23, SD 3.96 months. The mean duration and the standard deviation of both length of lactation and length of lactational amenorrhoea vary by type of region, decreasing as one passes from rural to urban areas. Corsini represented the relationship by the linear regression equation y = 1.332 + 0.556x, where both variables are expressed in months. In this relationship, breastfeeding explained about 77.4% of the variation in length of amenorrhoea (r = 0.88). Bongaarts and Potter (35) fitted this relationship for a selected sample by an exponential equation, with an r of 0.96. Bongaarts notes that "the end of amenorrhoea occurs frequently while the woman is still breastfeeding" (35). There are certainly many anecdotal data of women who are breastfeeding on demand and are not supplementing lactation, who nevertheless have a return of menses within 1 to 3 months. There should be further research on the characteristics of these women, who apparently are reported only from well-nourished populations. The relationship of calorie supplies (C) to duration of lactational amenorrhoea (L) for various populations has not been estimated heretofore, as far as I know. The data are crude since the calorie supplies are national estimates [World Bank NUTRITION, ACTIVITY, MENSTRUATION 71

(36) or Food and Agriculture Organization (37)], and the data on lactational amenorrhoea compiled by Corsini (34) and by Bongaarts and Potter (35) are from studies which represent subsamples of a population. However, both sets of data show a significant inverse relationship (P<0.05) between the length of lactational amenorrhoea and estimated calorie supplies (Fig. 3). The regression equation for 13 countries (India and South Korea, rural and urban shown separately) shown in Fig. 3 is: L (months) = 26.5 - 0.00654 (±0.00243)C r = -0.598, t = -2.69, P < 0.05 The inverse relationship between length of lactational amenorrhoea and estimated calorie intake was also significant for the data compiled by Bongaarts (16 studies): L = 29.7 - 0.00805 (±0.00297)C r= -0.562, t= -2.72, P < 0.05 Regressions of length of lactational amenorrhoea (L) on calorie supplies (C) using data from FAO (37) for 1975 to 1977 (calories not available for three countries)

24.0

X INDONESIA (R)

£_20.0 xBANGLADESH (R)

X S. KOREA (R) x INDIA (R)

X GUATEMALA (R) x S. KOREA (U)

RWANDA(R) SENEGAL (R) x INDIA (U) 8.0 x PHILIPPINES (U) a: EGYPT(R)x USA(U) UJ O TURKEY(U)

4.0 X CHILE (U) xENGLAND(U)

1900 2100 2300 2500 2700 2900 3100 3300 3500 DAILY PER CAPITA CALORIE SUPPLY FIG. 3. Average duration of postpartum amenorrhoea (months) in nursing women versus estimated daily per capita calorie supplies for studies listed by Corsini (34); India (rural), weighted average of three studies; South Korea (urban), weighted average of two studies; Rwanda (rural), weighted average of two studies; United States (urban), weighted average of four studies. Omit- ted: South Korea (urban-rural mixed); Taiwan (three studies) no calorie supplies for 1974 available; Alaska; and Venezuela-Colombia (rural) mixed data. Daily per capita calorie supply is for 1974 from the World Bank (36). Studies cited by Corsini date from 1961 to 1975. Unfortunately, no calorie supplies were available for urban and rural areas of a country. 72 NUTRITION, ACTIVITY, MENSTRUATION gave essentially the same significant inverse relationship for both the Corsini (Eq. 1) and Bongaarts (Eq. 2) collected data (DF, degrees of freedom): L = 26.0 - 0.00633 (±0.00239)C r= -0.608, t= -2.65, P < 0.05 (DF = 12) L = 29.9 - 0.00788 (±0.O0313)C (2) r = -0.558, t = -2.52, P < 0.05 (DF = 14) The significant inverse relationship between calorie supplies and length of lactational amenorrhoea can of course be explained by time of supplementation of the infant and other behavioural differences among populations. On the other hand, the inverse relationship may also reflect a real effect of intake of calories and thus physical differences among mothers, particularly those who have large energy outputs, and their infants. (Needless to say, women would have a lower caloric intake than the number estimated for the whole population.) The physical differences, in turn, would result in endocrinological differences affecting the length of lactational amenorrhoea.

BREASTFEEDING FREQUENCY AND NIGHT FEEDING Proponents of the view that the suckling pattern is the overriding factor in determining the length of lactational amenorrhoea have placed great emphasis on night feeding (38,39). It therefore seemed useful to explore what is known about the sleep pattern of infants after birth. How often does an infant wake up at night at various ages after birth? What are the theories of why an infant wakes up? Can a mother train an infant of a particular age not to wake up at night? (This latter suggestion was greeted with incredulity by all young parents, breastfeeding mothers or non-breastfeeders.)

Non-Western Societies As far as I know the only data on breastfeeding at night in a non-Western society is that of the Bush people of the Kalahari. Although the great frequency of breastfeeding during 24 hr has been widely cited, the actual observations are from dawn to dusk. Understandably, Konnor and Worthman (38) state that "no systematic observations were made at night"; !Kung infants sleep on the same skin mat with their mothers outdoors on the ground around a fire. Howell (13) describes how adults wake from time to time during the night and keep the fire going (and personal communication). Data on breastfeeding during the night were obtained by interviews of 21 women, feeding infants as old as 3 years. Of the 21 women, 20 reported being awakened at least once each night by the infant for breastfeeding. This is not unexpected. "All 21 reported that the infant fed from the breast during the night without waking the mother from two to 'many' times or 'all night' " (38). There are, however, no actual observations, as there are for the daylight hours. NUTRITION, ACTIVITY, MENSTRUATION 73

Interestingly, Konnor and Worthman note that as the population has become sedentary, the mean birth spacing has shortened by 8 months (38, footnote 10).

Night Waking in Infancy The emphasis on the importance of suckling patterns, particularly the frequency of night feeds, in relation to the time of resumption of menses raises fundamental questions about why and when babies wake in the night. This discussion is necessarily limited and will only set forth some of the main ideas from summary papers recommended by knowledgeable persons in the field. Parmelee et al. (40) recorded the periods of sleep and wakefulness of 46 infants during their first 16 weeks of life. Twenty-six of these were breastfed, 10 not breastfed, and 10 were changed from breast to artificial feeding during the 16 weeks studied. All babies were fed on demand. The addition of solid food in the form of cereal or purees was noted for each child. Night sleep was defined as 7:00 P.M. to 7:00 A.M. By 8 weeks of age, the number of hours of night sleep became significantly different from the length of daytime sleep. There was a clear definition of a diurnal cycle for 66% of the infants in 5 weeks and for 98% of the children in 12 weeks. The ratio of day to night sleep changed from 0.93 in the first week to 0.46 in the sixteenth week. The average daily number of wake periods decreased from 6.86 (SD, 1.27) in the first week to 4.66 (SD, 0.76) in the sixteenth week. The average number of daily feedings corresponded closely to the number of wakeful periods and showed a similar decline; there were 6.73 (SD, 1.40) daily feedings in the first week and 4.55 (SD, 0.80) in the sixteenth week. The time of starting "solids" was not significantly correlated with measurements of sleep functions at 4 months of age. The dramatic change in the infants' sleep pattern in the first 16 weeks of life was not the result of a marked reduction in the total amount of sleep in this period. There is apparently a "steady and rapid maturation of the infant's ability to sustain a prolonged period of sleep in the first 16 weeks of life." There is a similar but not so dramatic increase in the ability of the baby to sustain a prolonged period of wakefulness. The development of the diurnal cycle, in the view of these authors, is explained in part by the maturation of the nervous system, as well as by the effects of metabolic processes which have circadian rhythms; although some learn- ing must also be involved in relation to family patterns of behaviour. The authors conclude: "of practical significance to the pediatrician and the parents is that by the end of the first 12 to 16 weeks of life normal babies may be expected to sleep prolonged periods of time and to do so primarily at night." Moore and Ucko (41) also note that infants fed on demand progressively lengthen the intervals between their night feedings and that "maturation plays a part in the process." The 160 London infants in their study represented the five major cate- gories of social class. Confirming the Parmelee (40) results, 70% of babies had 74 NUTRITION, ACTIVITY, MENSTRUATION

ceased waking in the night (defined as midnight to 5:00 A.M.) by the age of 3 months, and 83% had "settled" by the age of 6 months. These authors note that there is a distinct rise in night waking again from 5 to 9 months. A comparison of infants who had settled before the age of 13 weeks and those who had not showed that boys wake significantly more than girls (though there was no significant difference in age of first settling between girls and boys). Birth weight, weight at 3 months, or weight increment over this period had no relationship to age of settling. Asphyxia at birth was associated with a higher tendency to wake at night. Feeding difficulties reported at 3 months were significantly (P = 0.02) related to failure to sleep through the night, although the two problems did not always coincide. The average sucking time per feed, so far as the mothers could estimate it, showed no relationship to the tendency to continue waking at night. The mother's education or age was also not related to the infant's settling age nor to its tendency to wake. The authors conclude that there is a natural tendency for infants to lengthen their period of continuous night sleeping. Babies who received the least nursing had the greatest tendency to wake; those receiving a "rather excessive amount" (not defined) were the next most wakeful group; those who had between 10 and 20 min with the mother in addition to their feeding time settled best (by which I assume earliest?). As in the previous chapter, the settling process is regarded as connected with fundamental physiological changes.

LACTATION AND WEIGHT CHANGES AFTER PARTURITION Hytten and Thomson (25) found that a tendency to lose weight, in the absence of special feeding arrangements, is common in both lactating and non-lactating women during the months following parturition. Of 11 subjects from Aberdeen who were breastfeeding, all but 1 lost weight. These authors comment that "if protein, fat, and, possibly, carbohydrate in body stores are used up during lactation the mother must lose weight. Curiously there seems to be very little evidence in the literature on this elementary point." I quote Hytten and Thomson's remarks because I think that it is of the greatest importance to follow the changes in weight for height of an individual woman from parturition to the time of resumption of menstrual cycles, as set forth in Fig. 4. Such data are necessary for the elucidation of the mechanism controlling the time of resumption of menses in lactating women. These data are needed for different ethnic or racial groups. Only two published studies that I know of have attempted such an assessment. Harrison et al. in New Guinea (42) reported that body-fat reserves, measured by changes in weight, diminish during the course of lactation and that subcutaneous fat reserves diminished with parity. Carael (43) related a 3- to 4-kg difference in weight for equivalent height during the first year of lactation, with variations in the length of lactational amenorrhoea in areas differing in nutritional patterns. The length of postpartum amenorrhoea was shorter in areas with a relatively balanced diet in contrast to areas with seasonal protein and lipid deficiencies (43). 70 Goln kg Conceptus 3.3 ot term Maternal 5.2 Pregnancy 65 Fat 4.0, X

60 iBirth Weight „ no"—=2 X lactation kg JT 55 AveroAre l*^ 1 Resumption of cycles 50 Ave. Men'chis |Minimalimal MMatura e 45 cycles Minimal Men'che 40

I I 12 14 16 20 22 24 Birth 1 2 3 4 5 10 11 12 Age (years) Pregnancy Months-Lactotion To resumption of cycles

FIG. 4. Scheme of changes in weight (approximate) for a woman 160 cm tall from the age of onset of menstrual function to pregnancy and birth of the first child. The weight for height at which regular ovuiatory cycles resume for a lactating woman is not yet known for any population. Estimates of gain during pregnancy from Hytten and Thomson, ref. 49. 76 NUTRITION, ACTIVITY, MENSTRUATION

There has been confusion between the minimal weight for height necessary for menstrual cycles in nutritional or exercise amenorrhoea and the as yet unknown weights for height at which a woman resumes ovulatory menstrual cycles after or during lactation (44). These weights would not be expected to be the same (45). A woman who has become pregnant must have ovulated; by definition, therefore, her weight must have been above the minimum weight for height necessary for ovulation. British data show that during pregnancy women store additional fat. Thus, at the time of parturition, these women would have a relatively high fat/lean ratio; if a woman did not breastfeed or if her infant died soon after birth, she might be expected to conceive again in a short time (46). In Bangladesh, Mosley has indeed observed that the interval to conception is very short if an infant dies soon after birth (46,47). In the absence of hard data, the most reasonable assumption is that the majority of lactating women have weights for a particular height above the minimal weight necessary for ovulation, at least during the early part of the lactational period. Therefore lactating women would not necessarily have to gain weight before resuming regular cycles. Only those women who had been marginal in their nutrition before and during pregnancy and then also inadequately nourished while lactating might have to gain weight before cycles resume (45). The data published for postpartum Bangladesh women are average weight and average height data from which nothing can be ascertained about an individual woman (44).

CONCLUSION The general concept that lactation is an energy-demanding process dictates a research protocol which will correlate detailed physical changes in the mother, including changes in weight for height studied longitudinally, with endocrinological changes. The weight/height2 index gives some measure of relative leanness (or fatness). However, since weight can remain the same but body composition change from loss of fat and gain in leanness or vice versa (50), body-composition data will be needed in conjunction with the endocrine levels. Physical data on the infant and on suckling patterns must also be correlated with the maternal data. These combined data may help elucidate the mechanisms controlling the return of menses and the risk of another pregnancy. The suggestion that women should not supplement breastfeeding with solid food in order to prolong breastfeeding and the length of lactational amenorrhoea may be dangerous for the infant. Data from the Khanna study show that although supple- mented infants had a higher incidence of diarrhoea, they had a lower mortality than did unsupplemented infants of the same age (48).

ACKNOWLEDGEMENTS I am indebted to Grace Wyshak for computation of the regression equations. I thank Rosemary Moore for preparation of the manuscript. I thank Marianna Mar- guglio for assistance in compiling the calorie and amenorrhoea data. NUTRITION, ACTIVITY, MENSTRUATION 77

REFERENCES 1. Frisch RE, McArthur JW. Menstrual cycles: fatness as a determinant of minimum weight for height necessary for their maintenance or onset. Science 1974;185:949-51. 2. Vigersky RA, Andersen AE, Thompson RH, Loriaux DL. Hypothalamic dysfunction in secondary amenorrhea associated with simple weight loss. N Engl J Med 1977;297:1141-5. 3. Nillius SJ, Wide L. The pituitary responsiveness to acute and chronic administration of gonado- tropin-releasing hormone in acute and recovery stages of anorexia nervosa. In: Vigersky RA, ed. Anorexia nervosa. New York: Raven Press, 1977:225-41. 4. Frisch RE, Wyshak G, Vincent L. Delayed menarche and amenorrhea of ballet dancers. N Engl JMed 1980;303:17-9. 5. Warren M. The effects of exercise on pubertal progression and reproductive function in girls. J Clin Endocrinol Metab 1980;51:1050-7. 6. Frisch RE, von Gotz-Welbergen A, McArthur JW, et al. Delayed menarche and amenorrhea of college athletes in relation to age of onset of training. JAMA 1981 ;246:1559-63. 7. Dale D, Gerlach DH, Wilhite AL. Menstrual dysfunction in distance runners. Obstet Gynecol 1979;54:47-53. 8. Schwartz B, Cumming DC, Riordan E, Selye M, Yen SS, Rebar RW. Exercise-associated amenorrhea: a distinct entity? Am J Obstet Gynecol 1981;141:662-70. 9. Nillius SJ. Weight and the menstrual cycle. In: Understanding anorexia nervosa and bulimia. Columbus, Ohio: Ross Laboratories, 1983:77-81. (Report of the Fourth Ross Conference on Medical Research). 10. McArthur JW, O'Loughlin KM, Beitins IZ, Johnson L, Alonso C. Endocrine studies during the refeeding of young women with nutritional amenorrhea and infertility. Mayo Clin Proc 1976;51:607- 15. 11. Boyar RM, Katz J, Finkelstein JW, et al. Anorexia nervosa: immaturity of the 24-hour luteinizing hormone secretory pattern. N Engl J Med 1974;291:861-5. 12. Henry L. Some data on natural fertility. Eugenics Quarterly 1961;8:81-91. 13. Howell N. The demography of the DoBe !Kung. New York: Academic Press, 1979. 14. Eaton JW, Mayer JJ. The social biology of very high fertility among the Hutterites: the demography of a unique population. Hum Biol 1953;25:206-64. 15. Chen LC, Ahmed S, Gesche M, Mosley WH. A prospective study of birth interval dynamics in rural Bangladesh. Population Studies 1974;28:277-97. 16. Frisch RE. Demographic implications of the biological determinants of female fecundity. Soc Biol 1975;22:17-22. 17. Frisch RE. Population, food intake, and fertility. Science 1978; 199:22-30. 18. Frisch RE. Population, nutrition and fecundity. In: Dupaquier J, Fauve-Chamoux A, Grebenik E, eds. Malthus past and present. London: Academic Press, 1983:393-404. 19. Wrigley EA. Population and history. London: World University Library, 1969. 20. Gopalan C, Naidu AN. Nutrition and fertility. Lancet 1972;2:1077-9. 21. Frisch RE. Body fat, puberty and fertility. Biol Rev 1984;59:161-88. 22. Sherman B, Wallace R, Bean J, Schlabaugh L. Relationship of body weight to menarcheal and menopausal age: implications for breast cancer risk. J Clin Endocrinol Metab 1981;52:488-93. 23. Darwin C. The variation of animals and plants under domestication. 1868. New York: Appleton, 2d ed. 1894; vol. 2; 89-90. 24. Darwin C. On the origin of species. 1859. Facsimile 1st ed. Cambridge, Mass.: Harvard University Press, 1964:147. 25. Hytten FE, Thomson AM. Nutrition of the lactating woman. In: Kon SK, Cowie AT, eds. Milk: the mammary gland and its secretion; vol. 2. New York: Academic Press, 1961:3-46. 26. Food and Agriculture Organization of the United Nations. Calorie requirements. Rome: FAO, 1957; II. 27. Emerson K Jr, Saxena BN, Poindexter EL. Caloric cost of normal pregnancy. Obstet Gynecol 1972;40:786-94. 28. Salber EJ, Feinleib M, MacMahon B. The duration of postpartum amenorrhea. Am J Epidemiol 1966;82:347-58. 29. Vander Walt LA, Wilmsen EN, Jenkins T. Unusual sex hormone patterns among desert-dwelling hunter gatherers. J Clin Endocrinol Metab 1978;46:658-63. 30. Paul AA, Muller EM, Whitehead RG. The quantitative effects of maternal dietary energy intake 78 NUTRITION, ACTIVITY, MENSTRUATION

on pregnancy and lactation in rural Gambian women. Trans R Soc Trop Med Hyg 1979;73:686- 92. 31. Lunn PG, Austin S, Prentice AM, Whitehead RG. Influence of maternal diet on plasma-prolactin levels during lactation. Lancet 1980;l:623-5. 32. Whitehead RG, Rowland MGM, Hutton MA, Prentice AM, Muller EM, Paul AA. Factors influencing lactation performance in rural Gambian mothers. Lancet 1978;2:178-81. 33. Wilmsen EN. Seasonal effects on dietary intake on Kalahari San. Fed Proc 1978;37:65-72. 34. Corsini CA. Is the fertility reducing effect of lactation really substantial? In: Leridon H, Menken J, eds. Natural fertility. Liege: Ordina Editions, 1979:197-214. 35. Bongaarts J, Potter RG. Fertility, biology and behavior: an analysis of the proximate determinants. New York: Academic Press, 1983. 36. World Development Report 1979. Washington, DC: World Bank, 1979. 37. Food and Agriculture Organization of the United Nations. The state of food and agriculture 1981. Rome: FAO 1982. (FAO Agriculture Series; no. 14; data used for 1975-1977). 38. Konnor M, Worthman C. Nursing frequency, gonadal function and birth spacing among !Kung hunter-gatherers. Science 1980;207:788-91. 39. Short RV Breast feeding. Sci Am 1984;250:35-41. 40. Parmelee AH Jr, Wenner WH, Schulz HR. Infant sleep patterns: from birth to 16 weeks of age. J Pediatr 1964;65:576-82. 41. Moore T, Ucko LE. Night waking in early infancy: part I. Arch Dis Child 1957;32:333-42. 42. Harrison GA, Boyce AT, Platt CM. Body composition changes during lactation in a New Guinea population. Ann Hum Biol 1975;2:395-8. 43. Carael M. Relations between birth intervals and nutrition in three Central African populations (Zaire). In: Mosley WH, ed. Nutrition and human reproduction. New York: Plenum Press, 1978:365-84. 44. Huffman SL, Chowdhury AKMA, Mosley WH. Postpartum amenorrhea: how is it affected by maternal nutritional status? Science 1978;200:1155-7. 45. Frisch RE, McArthur JW. Difference between postpartum and nutritional amenorrhea. Science 1979;203:921-2. 46. Howell NJ. Toward a uniform theory of human paleodemography. In: Ward RH, Weiss KM, eds. The demographic evolution of human populations. London: Academic Press, 1976:25—40. 47. Mosley WH. The effects of nutrition on natural fertility. In: Leridon H, Menken J, eds. Natural fertility. Leige: Ordina Editions, 1979:83-105. 48. Wyon JB, Gordon JS. The Khanna study. Cambridge, Mass.: Harvard University Press, 1971. 49. Hytten FE, Thomson AM. Maternal physiological adjustments. In: Maternal nutrition and the course of pregnancy. Washington, D.C.; National Academy of Sciences, 1970:41-73. 50. ParfskovS J. Physical activity and body composition. In: Brozek J, ed. Human body composition: Approaches and applications. Vol. 7. Oxford: Pergamon Press, 1965:161-176.

COMMENTARY

C. ROBYN

I agree with the idea that intensive physical work and undernutrition have disruptive effects on the reproductive ability of women, but I do not think that the endocrinological basis of these findings is so well documented. I do not find reliable information in the two references (Vigersky et al. and Nillius) quoted. How can, for example, an LHRH test show a hypothalamic dysfunction? What kind of hypothalamic dysfunction is responsible for the absence of ovulation in secondary amenorrhoea associated with single weight loss? Would you consider that single weight loss, undernutrition, and anorexia nervosa result in the same hypothalamic dysfunction? Would this dysfunction also be the same in women athletes and ballet dancers? NUTRITION, ACTIVITY, MENSTRUATION 79

The birth interval is a reasonably good marker of the restoration of normal ovulatory cycles, although it may be influenced by other factors: contraceptive practices, sexual abstinence, taboos, etc. The return of menses, however, is a very poor one. Therefore, it is hazardous to rely on epidemiological data based on the return of menses, all the more so as the incidence of anovulatory bleeding among lactating mothers varies from population to population. The aetiological factors for these differences are unknown. Could you provide data on the relation between caloric supplies and the length of the lactational infertility, based on more accurate criteria than amenorrhoea? You seem to consider that the seasonal changes in fertility are related to the seasonal changes in food supply and body weight, basing your opinion on the data on seasonal fertility of the Bush people. But are there any seasonal changes in the light and dark cycles in the Kalahari? Seasonal changes in fertility are a well-known phenomenon among mammals: they are influenced by the relative length of the light and dark periods. I suggest you read the paper by Lincoln and Short (1). Would you agree that alternative explanations other than nutritional and body weight changes could account for the seasonal variation in fertility among the Bush people? In addition to the reduction in body weight and in body fat stores, there are increased concentrations of oestrone relative to oestradiol in the circulation of both amenorrhoeic runners and runners with menstrual cycles, in comparison with non-running control women. Short-term exercise also increases the serum levels of virtually all gonadotrophic and sex steroid hormones, of the catecholamines and opioid peptides, whereas it decreases the clearance rate of oestradiol. Thus, the cumulative effects of repeated exercise may produce profound hormonal changes that may lead to altered reproductive function. Could you comment on the possible relationships between body weight or body fat store changes and these complex hormonal changes occurring in women athletes? What is the proportion of women athletes with altered menstrual function? Have amenorrhoeic runners more than others a history of cycle irregularity? What is their endocrine profile—LH, FSH, prolactin, E2, El, testosterone, DHEA-S—in comparison with women having other types of hypothalamic amenorrhoea? Is this profile different from or similar to that of anorexia nervosa or single weight loss? Lactational infertility is the rule in all mammals. Would you extend your nutritional theory of lactational infertility in women to the lactational infertility of all mammalian species including marsupials (2)? Thus, a comparative evaluation of animal and human data may provide a useful complement to this discussion. This may be even more important, as animal data do not support the critical body weight or critical body fat hypotheses of puberty (3,4).

1. Lincoln GA, Short R. Seasonal breeding: nature's contraceptive. Recent Prog Horm Res 1980;36:l- 43. 2. Short RV Lactation, the central control of reproduction. In Breast-feeding and the mother, Ciba Foundation Symposium. Elsevier-Excerpta Medica/North Holland, Amsterdam: 1976:73-81. 3. Glass AR, Swerdliff RS. Nutritional influences on sexual maturation in the rat. Fed Proc 1980;39:2360. 4. Glass AR, Anderson J, Herbert D, Vigersky RA. Relationship between pubertal timing and body size in underfed male rats. Endocrinology 1984; 115:19-24. SO NUTRITION, ACTIVITY, MENSTRUATION

H. L. Vis It is true that in developing countries, in a poor environment, a late mean age of menarche and an early mean age of menopause, due to nutritional factors in a broad sense, have been observed. However, it is not proven that the long birth interval in developing countries is also directly due to unfavourable environmental factors. Even though the specialized literature dealing with the influence of nutritional status, energy expenditure, and stress on female reproductive ability is abundant, the endocrinological aspects involved in this subject are still poorly understood (11). Investigations into these problems are generally limited to a few research teams. Literature about this subject always reveals that one or the other parameter has not been taken into account. Ideally, the factors which should be studied are the following:

1. The nutritional status of the mother, the weight on height, and the presence of subcutaneous fat above a weight threshold (above 22% of the total weight) but also other nutritional parameters. In developing countries, maternal nutritional status depends not only on quantitative but also on qualitative factors. The effect of the qualitative aspect of the diet on fertility is not well studied. What are the consequences on fertility of a lack of protein (kwashiorkor), lack of trace elements (Zn, Cu, Se, etc.), lack of iodine (as in the area of endemic goiter and cretinism), lack of vitamins, etc.? 2. An estimate of the energy expenditure in the form of physical work and milk production. 3. Possibly stress factors. 4. The suckling pattern over 24 hr and not only during the day. Hormone determinations are necessary to control the data on the suckling pattern. Indeed, variations in the sensitivity of the hypothalamus to the nipple stimulation reflex according to the length of the lactation period may not be excluded. Hormone determinations are also necessary to ascertain ovulation. 5. "Social environment." Taboos, behaviours of the mother, and behaviours of the society in relation to pregnancy, delivery, breastfeeding, and infant rearing.

We have constructed two diagrams, showing the possible interactions in relation to the nutritional status of the mother on one side and on the other the interrelationship between nutritional status, energy expenditure, lactational amenorrhoea and suckling pattern (Figs. C-l and C-2). The important sentence of Rose Frisch, "A woman who has become pregnant must have ovulated; by definition, therefore, her weight must have been above the minimum weight for height necessary for ovulation," indicates that in their socio-economic and nutritional environment lactating women have had the weight criteria for ovulation. In Kivu, the nutritional status in rural women is poor, weight increase during pregnancy amounts to 5 kg, birth weight averages 3 kg, and the placenta weighs between 0.5 and 0.6 kg. If water retention and the weight of the amniotic fluid are taken into account, it is evident that under those circumstances pregnant women are not able to accumulate subcutaneous fat, which amounts to 3-4 kg in industrialized countries). The result is that, in Kivu, the mothers have the same weight after delivery as before pregnancy (10). Weight changes during the whole lactation period are negligible (3). As a consequence, if there are no weight losses during lactation, the extra nutrition necessary for milk production and/or for energy expenditure will come only from the daily food intake. In some marginal circumstances, the mother has to make NUTRITION, ACTIVITY, MENSTRUATION 81

Production Reproduction

Fertility pattern Energy output Pregnancy—Breastfeeding work at home ? and in the field 7 Nutritional status of the mother Quantitative and qualitative aspects t Food intake and food reserves Quantitative and qualitative

f Socio-economic environment Infections—Parasitosis Social behavior (Hookworm—Malaria) Taboos—Social pressure

FIG. C-1. Interacting factors and the mother's nutritional status. (From ref. 10.)

Production Reproduction

Energy output Long-lasting postpartum amenorrhoea

Suckling pattern Frequency, force, duration

Mother's behavior

Nutritional status of the mother Quantitative and qualitative aspects

Social pressure Taboos—Urbanization, etc.

FIG. C-2. System showing the suckling pattern as a pace-maker directed to energy expenditure (suckling pattern of low frequency) or to long lasting amenorrhoea (suckling pattern of high frequency). The social pressure and the nutritional status of the mother are the main factors which influence the suckling frequency. Social pressure may have a positive or a negative influence on the energy output. The weight of the infant only has an influence on the suckling duration. 82 NUTRITION, ACTIVITY, MENSTRUATION a choice between energy expenditure and lactation, a choice which can be made by changing the suckling frequency (Fig. C-2). It is possible, but still not proven by the data in the literature, that energy expenditure may directly influence, by hormonal changes in the hypothalamus, the duration of lactational amenorrhoea (11). In most poor Third World regions (at least in rural African areas) during the period of seasonal food scarcity, extra physical work is demanded from the women. Lee (5) pointed out that in such conditions the food intake is spent either on the production sector (physical labour) or on the reproduction sector (ovulation-lactation). This is the reason my main criticism of the work of Rose Frisch concerns the fact that correlations are made only with food energy intake and not with energy expenditure. Amenorrhoea as a consequence of famine is a good example of an amenorrhoea occurring without lactation (6). This amenorrhoea is always present when a very sizeable weight loss takes place (which is not the case in most of the studies done on lactating mothers in developing countries). Weight loss by itself, however, does not seem to be sufficient to explain the appearance of amenorrhoea. Let us take an example well known in paediatric literature: the siege of Leningrad 1941 to 1943 (1). "The people suffered from hunger" ... "Amenorrhoea was widely prevalent"... "Incessant bombing... air-raid alarms by day and night all have their effect on the nervous system of the women" (i.e., stress) "the women had to undergo physical exertions to which many of them had not been accustomed" (i.e., physical labour). As a matter of fact, during the siege of Leningrad the women who were able to be pregnant "did not suffer from the hunger that was the lot of the others." Also, in developing countries the women who are able to be pregnant are well above the nutritional threshold necessary to ovulate. Perhaps we are dealing with a selection of women who remain in equilibrium at, for us, a low nutritional threshold. In the work of Prema (9, Fig. 8), there seems to be a correlation between the mother's weight and the mean duration of unsupplemented lactation. Heavy mothers have a shorter unsupplemented lactation duration than light ones. It is possible that because of the early introduction of weaning food, breastfeeding frequency decreases; therefore, it would not be directly the weight but the change in suckling pattern that makes the differences. The main remark on the work of Prema is that the weight of the lactating mother is used as the only parameter which is taken into account in accordance with the duration of lactation and lactational amenorrhoea. But in the case of lactational infertility, several parameters are always involved, of which many (mainly suckling pattern or prolactin levels, for instance) are not mentioned. In this context we want to describe two examples of situations which are endlessly quoted: The first is that of the !Kung. Lee (5) estimated that carrying the child is in itself very great physical work which averages (in kilograms per kilometer per year) 14,400 between 0 and 1 year; 21,120 between 1 and 2 years; and 20,880 between 2 and 3 years. "Given a mean birth interval among nomadic San of four years for women of normal fertility During the ten-year period the average woman will raise three children and will have carried an average burden of 9.2 kg per day. Her burden will be least during the years when she is carrying a newborn and greatest when she is carrying a three-year old (plus the burden of being pregnant at the same time)." For the !Kung, energy expenditure seems to be a much more important factor than nutritional status. The lactating mothers do inhibit ovulation, by a suckling pattern of high frequency during a very long period of time. The second example is CaraeTs work (2), quoted several times by Rose Frisch and Prema Ramachandran. It deals with the difference in lactational amenorrhoea duration among three NUTRITION, ACTIVITY, MENSTRUATION 83 ethnic groups in Zaire (Havu, Ntomba, and Tembo). Let us take the two extreme groups: the Havu of Kivu and the Ntomba around the Tumba lake. In Kivu, the nutritional situation is quite bad, with seasonal variation in protein intake; and around Tumba lake the nutritional situation seems to be better, but important seasonal energy intake variations still persist (7). In both situations the amount of physical work done by the women is very important. Accurate data are lacking for the Kivu region, whereas the physical work performance of the women from the Tumba lake was very precisely measured by Pagezy (7). The Oto (the Ntomba of Carael) women carry baskets of 44.3 kg on average when coming from the field. They also carry the additional burden of a child weighing 5 to 12 kg, whereas their own mean weight is 53.9 kg. In Kivu, the situation seems to be comparable. But an important difference in behaviour is observed at the beginning of lactation between the two ethnic groups. Social, sexual, and nutritional taboos, considerably diminishing the energy expenditure, exist in the Tumba lake area (8), whereas these are absent in Kivu. We do not have data on the suckling pattern in the Tumba lake area, but we do know that weaning food of good quality (fish, palm oil) is introduced at the age of 6 months. It is probable that the frequency of breastfeeding decreases at the time. Finally, the parasites in the Tumba lake area are mainly hookworms and malaria (Plasmodium falciparum), both known to enhance malnutrition, in contrast with the ascaris found in Kivu. The weight of women does not differ much from one area to the other, so the main difference lies in the energy output and also in the suckling pattern of the infant. In fact, Pagezy (1983) found a difference of more than 5 to 10 kg between Bantus (Oto) and pygmoid mothers (Twa) of the Ntomba society, that were lactating and menstruating (cf. Table C-l).

TABLE C-1. Comparison of measurements among young menstruating nulliparous girls, nursing primiparous women aged 15 to 20 years', and nursing multiparous women', at end of dry season, 1979 and 1980b

Weight (kg) Skinfolds Lean arm Stature Stature Triceps + Subscapular circumference 150-160 cm 160-170 cm biceps (mm) (mm) (cm) Subjects Mean SD' Mean SD Mean SD Mean SD Mean SD

Oto women n = 21c 46.2 3.6 \ 49.3 3.5 L 13.0 2.6 8.6 2-1 L 21.8 1.3 n = 9d 52.8 7.51 57.6 8.2/ 17.3 7.41 11.2 2.7/ 22.5 2.11 n = 33* 47.7 4.6 54.7 4.4 12.3 4.4 I"* 8.1 2.6 }" 23.8 1.6/* Twa women n= 4C 43.0 4.2 11.4 6.4 7.6 0.8 20.4 1.9 n = 9" 39.6 6.3 46.1 3.1 14.3 4.2 9.2 2.2 20.5 1.31 n= 13* 42.6 3.4 49.0 6.0 11.6 3.8 8.9 2.6 22.5 1.4/t

'Nursing for at least 6 months. 6From ref. 8. cMenstruating nulliparous. ^Nursing primiparous "Nursing multiparous. 'Brackets indicate significant comparisons: *P < 0.05; "P < 0.01, Mann and Whitney U test; tP < 0.001, Mann and Whitney U test. 84 NUTRITION, ACTIVITY, MENSTRUATION

In conclusion, it is not the small discrepancies in the mothers' weights which determine the different durations of postpartum amenorrhoea between the Tumba and the Havu people. Because of the existence of strong social rules, the Tumba mothers diminish the energy expenditure and introduce weaning food from 6 months after delivery, with the consequence that they protect their nutritional status (8). The Havu mothers maintain a great energy output, and they also maintain a high suckling frequency. Supplementary food is introduced early, but a true weaning does not exist before the age of 2 years. The behaviours in the two ethnic groups are totally different. 1. Antonov AM. Children born during the siege of Leningrad in 1942. J Pediatr 1947;30:250-9. 2. Carael M. Relations between birth intervals and nutrition in three Central Africa populations. In: WH Mosley, ed. Nutrition and human reproduction. New York: Plenum Press, 1978:365-84. 3. Hennart Ph. AUaitement maternel en situation nutritionnelle critique: adaptations et limites [Thesis]. Brussels: Universite Libre de Bruxelles, 1983. 4. Hennart Ph, Ruchababisha M, Vis HL. Breast-feeding and post-partum amenorrhoea in Central Africa. J Trop Pediatr 1983;29:185-9. 5. Lee RB. The !Kung San. In Lee R, ed. Men, women and work in a foraging society. Cambridge: Cambridge University Press, 1979:281-308. 6. Leroy-Ladurie E. L'amenorrhee de famine (XVII-XX siecle). In Le territoire de l'historien. Paris: N.R.F.-Gallimard, 1973:331-48. 7. Pagezy, H. Seasonal hunger, as experienced by the Oto and the Twa of a Ntomba village in the equatorial forest, Lake Tumba, Zaire. Ecol Food Nutr 1982;12:139-53. 8. Pagezy, H. Attitude of Ntomba society towards the primiparous woman and its biological effects. J Biosoc Sci 1983; 15:421-31. 9. Prema K, Nadamuni Naidu A, Neela Kumari S, Ramalakshmi BA. Nutrition-fertility interaction in lactating women of low income groups. Br J Nutr 1981;45:461-7. 10. Vis HL, Hennart Ph, Ruchababisha M. Some issues in breastfeeding in deprived rural areas (maternal nutrition and breastfeeding in the Kivu, Zaire). Assignment children 1981 ;55/56:183- 200. 11. Vttoen MR Effects of undernutrition on reproductive function in the human. Endocr Rev 1983;4:363- 77.

P. G. LUNN AND R. G. WHITEHEAD Much of Dr. Frisch's discussion is concerned with the energy requirements of women during pregnancy and lactation. If the amount of energy consumed is not adequate for a woman's needs, it is assumed that maternal tissues will be mobilized to maintain pregnancy and lactation, resulting in a reduction in body weight and fat content below the critical value for ovulatory activity. Central to this argument is the need for an accurate knowledge of energy requirements of these reproductive processes, and we would suggest that such figures are not actually known. Certainly, the estimate quoted by Dr. Frisch, i.e., 800 to 1,000 kcal/day extra, for a successful lactation is now considered to be extremely high, and more recent estimates suggest an extra 500 kcal/day on top of a non-pregnant, non-lactation value of 2,100 kcal/day (1) or approximately 550 kcal/day, raising that intake to 2,750 kcal/day (2). In practice, however, such intakes are rarely seen even in women in industrialized communities, and in most developing countries energy consumption is frequently approxi- NUTRITION, ACTIVITY, MENSTRUATION 85 mately 1,000 kcal/day below these recommended allowances (3), yet very successful lacta- tions are the norm. Changes in weight and skinfold thickness of individual pregnant and lactating women have been followed longitudinally in the Gambian study and compared with maternal food consumption (4). Average energy intakes during the dry and wet seasons, respectively, were as follows: in pregnancy, 1,483 and 1,417 kcal/day; in first-trimester lactation, 1,773 and 1,474 kcal/day; and in subsequent trimesters, 1,662 and 1,413 kcal/day. Women who were pregnant during the dry season gained an average 1.4 kg/month but only 0.4 kg/month during the rains. Similarly, in the dry season, lactating women gained weight and deposited subcutaneous fat, but a loss of weight and a reduction in subcutaneous fat occurred during the wet season. The deterioration in the wet season, however, appears to be more associated with the exceptionally heavy agricultural work which the women were obliged to undertake at that time of the year rather than the diet per se, as a similar deterioration in nutritional status still occurred in later studies when the mothers were receiving a substantial dietary supplement. Over the whole year these alterations in maternal body weight and skinfolds balanced out, and there was also no detectable effect of parity on any index of maternal nutritional status. Theoretical calculations based on the summation of even the lowest estimates of the energy costs of resting metabolism, activity, and milk production clearly indicate that at the level of maternal energy intake observed in this study (which compares very closely with data from other poor countries) these are incompatible with successful child rearing. Yet until the introduction of weaning foods (at 3-4 months in The Gambia) infant growth rates actually exceeded standard values and made up a small deficit in birth weight. The almost inevitable conclusion from this data is that metabolic and/or behavioural changes in the mother and/or her offspring allow a significant adaptation to a poor food supply which enables these essential processes to continue normally despite an apparently high energy deficit. It is also evident from these results that it is not possible to deduce alterations in maternal weight and fat content on the basis of energy consumption by the mothers.

Energy Consumption and Lactational Amenorrhoea All the data quoted by Dr. Frisch are in keeping with the suggestion that improved dietary status results in a shortened duration of lactational amenorrhoea, and I know of no data which contradict this view. The conclusion is reinforced by observations of seasonal changes in recommencement of menstruation and of birth incidence, and we would agree that these probably occur as a result of seasonal variations in food supply and energy expenditure. It is, however, a considerable leap from this data to the assumption that improvements in food supply would necessarily lead to high maternal weight and fat content and as a consequence reduce the duration of postpartum amenorrhoea. In The Gambia, supplementation of the energy consumption of lactating women by 46% (1,568-2,291 kcal/day) produced a maternal weight gain of only 1.8 kg, but the duration of lactational amenorrhoea fell by 5 to 6 months. An alternative interpretation of the relationship between community energy intakes and the length of lactational infecundity is that the better food supply would alter the infants' suckling, either as a result of the child receiving more or better weaning food, thus making him less hungry, or if the food is eaten by the mother, by making milk more readily available to her offspring. In both cases the total suckling input would fall, leading to lower prolactin values and an earlier resumption of ovulatory activity. 86 NUTRITION, ACTIVITY, MENSTRUATION

The argument about night feeding rests on the assumption that because basal blood prolactin concentrations tend to be higher at night, suckling stimulation at this time may induce a larger release of the hormone (is there data on this?) and would therefore have a greater repressive effect on the ovaries. One interpretation of the studies quoted would be that well-fed contented babies soon find they do not need the night feeds because their milk supply during the day is adequate for their requirements. On the other hand, less well nourished infants may need to suckle at night to obtain their requirements, and if prolactin release is more sensitive at this time, such an action would boost milk supply and prolong amenorrhoea in precisely those individuals in whom it is most important. In the majority of lactating women, including those exposed to chronic mild to moderate malnutrition, there seems little direct evidence to support the conclusion that body weight and/or fat content are of major importance in determining the length of lactational amenorrhoea. However, in more severely malnourished women, where the high energy cost of lactation does result in significant changes in maternal weight, then presumably this mechanism comes into action. The fact that the median lengths of amenorrhoea of non-lactating mothers in Boston and India are very similar is in keeping with this suggestion, and in the absence of any data to the contrary it must be assumed that the individual Indian women with prolonged amenorrhoea were suffering from a more severe malnutrition than other members of their community. 1. National Research Council. Recommended dietary allowances, 9th rev. ed. Washington D.C.: National Academy of Sciences, 1980. 2. WHO/FAO. Energy and protein requirements. Geneva: WHO, 1973. (WHO Technical Report Series; no. 522). 3. Whitehead RG, Paul AA. Diet and the pregnant and lactating woman. In Turner MR, ed. Nutrition and health: a perspective. London: MTP Press, 1982:159-68. 4. Prentice AM, Whitehead RG, Roberts SB, Paul AA. Long-term energy balance in child-bearing Gambian women. Am J Clin Nutr 1981;34:2790-9.

A. S. McNEILLY Energy costs of lactation. There seems to be no discussion of the probability that there is a reduction in maternal nutritional requirement for personal metabolism while the mother is lactating. Such a possibility is discussed by Lunn, and if such a physiological adaptation were to take place, this would alter the extent of the apparent calorific drain on the mother. Length of postpartum amenorrhoea without lactation. The data presented here are very interesting and should be discussed in much more detail. Seasonal variation. Again, these are interesting observations, but they have not been related to any changes in the patterns of suckling, which may vary with nutrition, milk supply, etc. In women with prolonged breastfeeding it would be expected from data now available in well-nourished Western societies that amenorrhoea would cease while token breastfeeding was continued. Many of the problems of understanding the mechanisms whereby breast- NUTRITION, ACTIVITY, MENSTRUATION 87 feeding suppresses ovarian activity have arisen because of "much anecdotal data," which is often misleading and rarely properly documented. Thus, there is certainly a need for "research on the characteristics of these women, who apparently are only reported from well-nourished populations." Breastfeeding frequency and night feeding. This is an interesting section which tackles the problem of night nursing. However, night nursing does not necessarily mean in the middle of the night. It is clear from adequately collected data that night feeds occurring before 11 p.m. (23.00 hr) are sufficient to disrupt normal ovarian activity and resumption of fertility in Edinburgh and Australia (Melbourne) women. Feeds given during the night, i.e., between 11 p.m. and 6 a.m., were not any more important. A possible reason for this is argued by McNeilly. Lactation and weight changes after parturition. It would be interesting to know how the interval to conception in Bangladesh quoted by Mosley relates to the data on resumption of ovulation in the non-breastfeeding women in Boston and India.

J. HOBCRAFT The fact that there are differences between populations in the length of their birth intervals is not evidence for nutritional effects. Nor is variation in lactational amenorrhoea. Both are virtually preconditions for nutritional effects on length of lactational amenorrhoea but no more. Dancers and runners are usually very well nourished, making the connection even more specious. The evidence for a "biological syndrome" is hardly conclusive; variation in menopausal age with nutrition is not well established; and longer birth intervals can be attributable to many other things than nutritional differences—of course, we do not know. Seasonal variation—other possible explanations are again ignored. Calories versus amenorrhoea—this is not at all helpful; the statistical/epidemiological work is quite inadequate even given the caveats stated. I fail to understand why following "changes in weight for height of an individual woman" will give results different from following her weight alone. Presumably, this is confusing the individual woman with needed indices for statistical analyses which aggregate across women? Scatter diagrams by Huffman et al. (1) clearly show that no threshold of weight for height works in Bangladesh to discriminate menstruating from postpartum amenorrhoeic women. 1. Huffman SL, Chowdhury AKM, Chakraborty J, Mosley WH. Nutrition and postpartum amenorrhoea in rural Bangladesh. Population Studies, 1978;32:251-60. 88 NUTRITION, ACTIVITY, MENSTRUATION

G. S. MASNICK A 1980 literature review by John Bongaarts (1) concluded that "moderate chronic malnutrition has only a minor effect on fecundity (reproductive capacity), and the resulting effect on fertility (actual reproduction) is very small." One and one-half years later this same conclusion was voiced in an even more extensive summary of the literature (2). For Frisch, the most avid proponent of the "fatness and fertility" hypothesis, the Bongaarts review was less than conclusive (3,4). Yet the mounting weight of authority, if not of evidence, has led many to be sceptical about the hypothesis that lack of adequate nourishment will result in a positive check on population growth by depressing fertility as well as by increasing mortality. Now, new research findings from The Gambia have provoked a recon- sideration of the role of maternal nutrition in structuring birth intervals (P. G. Lunn, this volume; 5,6). For Frisch and her colleagues, this research of Lunn, Whitehead, and their colleagues comes none too soon; and while it should be welcome, it is not exactly vindicating. The present paper by Frisch summarizes her thesis why maternal nutrition is hypothesized to affect fertility. First is the argument of biological consistency. The relationship between fatness and both age at menarche and regularity of cycling is unequivocal; that between fatness and menopause is strongly suggested. For Frisch, it is "therefore improbable physiologically to assume that women who have grown up slowly because of marginal food supplies, physical labour, and/or disease, and who have late age of menarche (15-16 years), would have the same length of lactational amenorrhoea and the same birth intervals as non- contracepting women who grew up rapidly on high levels of nutrition (8). The second argument is one of empirical acceptability. Populations with longer average birth intervals generally also have lower nutritional intake and/or higher energy outputs in terms of work performed by women. The third level of argument is more theoretical and appeals to certain principles of evolutionary adaptability. Here it is emphasized that lactation is an energy- demanding process, as would be another pregnancy. Prolonged amenorrhoea among under- nourished women is viewed as nature's way of insuring that the mother's energy outputs are directed towards the surviving infant rather than towards sustaining a new pregnancy. Too rapid a succession of pregnancies in a malnourished woman would put both the surviving infant and the fetus, as well as the mother, at risk because she would be unable to convert energy in sufficient quantities to support three biological entities. If the fatness and fertility hypothesis is correct, the policy implications are significant. Food aid to populations who are pushing against their resource base will be in some ways counterproductive if fertility rates are increased as a consequence of raising caloric intake of women. Critics of the hypothesized relationship between maternal nutrition and fertility have dealt only with its empirical acceptability. Lack of large or consistent differences in postpartum sterility among groups of women selected to represent a continuum of weight for height is offered as a basis for rejecting the hypothesis (9). Most damaging of all, however, is the existence of an "alternative" hypothesis to explain observed differences between lactational amenorrhoea and birth intervals of well-nourished and poorly nourished women. Patterns of introducing supplementary foods into the diet of infants being breastfed, and other factors resulting in variations in the frequency and intensity of breastfeeding, have been clearly shown to exercise a major influence on lactational amenorrhoea. This "alternative" line of reasoning, however, does not explain differences in temporary postpartum sterility among groups of women who do not breastfeed by choice, because of inability or because then- baby has died. NUTRITION, ACTIVITY, MENSTRUATION 89

1. Bongaarts J. Does malnutrition affect fecundity? A summary of evidence. Science 1981;208:564- 9. 2. Breast-feeding, fertility and family planning. Popul Rep [J] 1981;24:525-75. 3. Frisch RE. Body fat, puberty and fertility. Biol Rev 1984;59:161-88. 4. Frisch RE. Malnutrition and fertility. Science 1982;215:1272-3. 5. Lunn PG, Austin S, Prentice AS, Whitehead RG. Influence of maternal diet on plasma—prolactin levels during lactation. Lancet 1980;(i)623-5. 6. Whitehead RG, Hutton M, Muller E, Rowland MGM, Prentice AM, Paul A. Factors influencing lactation performance in rural Gambian mothers. Lancet 1978;(ii)178-81. 7. Frisch RE. Maternal nutrition and lactational amenorrhea: perceiving the metabolic costs. Draft, 1984. 8. Huffman SL, Chowdhury AMK, Chapborty J, Mosley WH. Nutrition and postpartum amenorrhea in rural Bangladesh. Population Studies 1978;32:251-60.

J-P. HABICHT The thrust of this paper seems to be that inadequate energy stores or (?) deficient energy balance prolongs postpartum amenorrhoea. When both conditions exist (they often do not) and at the extreme, there is no doubt about the statement. (How women athletes and dancers fit into that picture is not clear to me because their stores and balance are often adequate.) The evidence to support the above claim in less extreme situations is peculiar. 1. Evidence of association between malnourished populations and long postpartum amenorrhoea. The consensus would be that this is due to the confounding of breastfeeding. Almost all of the evidence of association presented by Frisch can be explained that way without involving nutrition as a cause of prolonged postpartum amenorrhoea: poor people in developing countries who are malnourished also tend to breastfeed. Their longer length of postpartum amenorrhoea could be due solely to more breastfeeding [see Habicht (1) for discussion in which Frisch participated]. When breastfeeding, even in very well nourished populations, is such a strong contraceptive, one must take that confounding factor into account in one's discussion. Frisch does so only once, when comparing non-breastfeeding Boston women with Indian women who had lost a child at birth. The subsequent comparison of lactating women in Boston and India is confounded by "intensity" of suckling. I think the basic problem is that Frisch does not believe that lactational infertility is caused, above all, by breastfeeding. Perhaps this explains why normal expectations discussed below under item 5 are not attempted in this paper. 2. Seasonal variations in length of lactational amenorrhoea have been used by the Cambridge group (see P. G. Lunn, this volume) to great advantage in relating nutrition to lactational amenorrhoea. None of that thinking comes through in Frisch's presentation of this issue, and no other thinking relevant to the issue replaces it. 3. One-third of Frisch's paper is an argument against something to do with night feeding. It is difficult to understand what is being attacked. In any case, none of this section seems related to the issue of nutrition and postpartum amenorrhoea. [McNeilly et al. (this volume) feels night feeding is important for postpartum amenorrhea, and I could see that it might be an important confounding variable, but none of that is apparent in this section.] 90 NUTRITION, ACTIVITY, MENSTRUATION

4. The section on weight changes with lactation and the implications for subsequent fecundity baffle me. I would have expected a straightforward argument: "During lactation maternal energy stores decrease. If the stores were already marginal or inadequate at parturition, they will have inadequate stores thereafter. This inadequacy will prevent resumption of menses." Instead we have an argument culminating with "In Bangladesh, Mosley has indeed observed that the interval to consumption is very short if an infant dies soon after birth," which, of course, is due to a lack of breastfeeding and has nothing to do with energy stores. 5. Any discussion of malnutrition and postpartum amenorrhoea has to take into account lactational amenorrhoea caused by suckling and the mechanisms which mediate the two. How does malnutrition affect this mechanism? Or is there a different mechanism for malnutrition? How does this relate to meeting the nutritional costs of lactation? By what mechanism? 6. Nothing in the conclusion is developed in the paper. The second paragraph of the conclusion is particularly strange. It claims that early supplementation is better than full breastfeeding. If by "early" one means the second half of infancy, that is correct. But that is so obvious that the implication of the statement is that "early" is within the first few months. The association between such early supplementation and better survival is com- pletely mediated by socioeconomic status in developing countries [e.g., DaVanzo et al. (2)]. Such early supplementation of infants living in households with poor sanitation and water supply is associated with much higher mortality than that of babies in the same environment who are fully breastfed (3). 7. The non-breastfeeding Boston women menstruated sooner than the Indian women. Postpartum amenorrhoea is affected by age (Salber et al., ref. 28; R. E. Frisch, this volume). Were the Indian women who had lost their babies older than the Boston women? 1. Habicht J-P. Field studies introductory statement. In: Mosley WH, ed. Nutrition and human reproduction. New York: Plenum, 1978:345-52. 2. DaVanzo J, Butz WP, Habicht JP. How biological and behavioral influences on mortality in Malaysia vary during the first year of life. Population Studies 1983;37:381-402. 3. Butz WP, Habicht J-P, DaVanzo J. Environmental factors in the relationship between breastfeeding and infants' mortality: The role of sanitation and water in Malaysia. Am J Epidemiol 1984;119:515- 25.

K. PREMA RAMACHANDRAN Could the observed human maximum average of 10 to 11 children in Hutterites be to a large extent attributable to early introduction of supplements to breastfed infants and short duration of lactation and consequent rapid return of fertility? Is "submaximal nutrition" optimal? Is it desirable or normal to obtain 40% calories from fat? Do the secular trends seen in "well-nourished" women from developed countries represent adaptation to overnutrition, which is also a form of malnutrition? At the moment there is no answer to these questions. We need to know whether these are relevant questions NUTRITION, ACTIVITY, MENSTRUATION 91 that should be considered in our efforts to understand the effect of nutritional status on lactational fertility. The majority of the studies from developing countries suggest that the duration of postpartum amenorrhoea in non-lactating women is not significantly different from the duration of postpartum amenorrhoea in well-nourished women from developed countries. Unlike women from developed countries, most women from developing countries such as India do not lay down body fat during pregnancy (weight gain during pregnancy is between 5 and 7 kg). In the first 12 months of lactation they lose on average 1 to 2 kg. The average duration of lactational amenorrhoea is 10 months. So in fact, the return of menstruation occurs when women are losing weight. The time of introduction of supplements is the crucial factor. In poorer segments of the population the introduction of supplements before 6 months is associated with a higher prevalence of morbidity due to infections in the infant and shorter duration of lactational amenorrhoea. However, delay in the introduction of supplements beyond 6 to 9 months is associated with undernutrition and the consequent higher morbidity and mortality in the infants; the duration of lactational amenorrhoea, however, is not longer.

D. J. NAISMITH Frisch has produced convincing evidence that the initiation of fertility (menarche) is dependent on the attainment of a well-defined body composition around the age of puberty— a balance between lean body mass and body fat. Chronic undernutrition or strenuous physical activity, as practised by athletes and dancers, may delay menarche, apparently by distorting the ratio of lean tissue to fat. But can the resumption of fertility during lactation be thought of as a second (or third or fourth) menarche? Factors that may exert an influence on menarche may no longer do so in the altered hormonal environment of lactation. Exercise is an example. Whether a high level of energy expenditure is itself a factor that delays menarche, perhaps by altering hypothalamic function, or whether the abnormal proportions of muscle tissue and fat that result from intensive training is the factor responsible for the suppression of fertility has yet to be clarified. It is often assumed that athletes and dancers have very high intakes of energy. This may not be so. The obvious need to maintain a low body fat content may well be achieved by the deliberate restriction of food consumption. Breastfeeding mothers in The Gambia undertake hard physical work (P. G. Lunn, this volume) and have been described as "muscular," with little subcutaneous fat. In this respect they can be likened to the athletes and dancers. However, lactation is normally associated with the loss of body fat and (in animals) the conservation of lean tissue, a change in body composition that should maintain the state of amenorrhoea until breastfeeding is stopped or reduced by the introduction of weaning foods and a more normal body composition re- covered. This is clearly not the case.