Infant feeding and infant mortality in rural Chile.

Bull Org. mond. Sant 1 1973, 48, 203-210Bsd!. Wld HiMh Org. j Infant feeding and infant mortality in rural Chile S. J. PLANK1 c& M. L. MILANESI 2 During 1969-70, 1712 rural Chilean mothers were interviewed to see if their feeding practices contributed to infant mortality. There were three times as many deaths among babies given bottles before the age of3 months as among those who were wholly breast-fed. Nearly halfthe children had started bottlefeeding by that age, and it was thus a majorfactor in infant mortality. Additionalfoods, apparently, provided these babies with someprotection, whereas continued breast feeding did not. As living standards improved, weaning was accelerated and a higher proportion ofchildren werefed on the bottle alone. The anomalous consequence was that infant mortality rose with income. These observations support the opinion that early weaning is a concomitant ofeconomic development; they also suggest that the risk associated with bottle feeding may be reduced by adding other foods to the diet. Urgent attention to the epidemiology ofearly wean- ing would help prevent infant mortality and birth rates from rising in developing countries where the practice of bottle feeding can be expected to increase with economic progress. Although rural populations are generally not yet affected (Becroft & Bailey, 1965; Cantrelle & Leridon, 1971; Gordon et al., 1964; Harfouche, 1970; Martin et al., 1964; Taba, 1970; Thomson et al., 1966; Urrutia & Mata, 1969; Wyon & Gordon, 1971), both village and urban mothers wean their babies early in Chile (Ariztia, 1955; Monckeberg et al., 1967a 8). It has been suggested (Ariztia, 1955; Behar, 1964; Monckeberg, 1970) that this practice might be responsible for Chile's high infant mortality relative to its economic and educational levels and its comprehensive and extensive health services (Behm et al., 1964). As part of a study of health and fertility in rural Chile (Plank & Milanesi, 1973), infant feeding practices were investigated: (1) to measure the effect of early weaning on mortality, (2) to identify characteristics distinguishing mothers who practised early weaning, (3) to clarify the respective roles of matemal milk deprivation and factors associated 1Lecturer on Population Studies, Harvard School of Public Health, Boston, Mass., USA, 02115. ' Research Associate in Population Studies, Harvard School of Public Health, and Assistente Doutor, Faculdade de Saude Publica, Universidade de Sao Paulo, Brazil. s See also: Valiente, S. et ai. (1967) Estudio de la alimenta- cidn de prescolares de la Provincia de Curicd, Departamento de Nutrici6n, Santiago, Universidad de Chile, Publication 247/67 (mimeographed document). with bottle feeding, and (4) to measure the effect on survival of giving bottle-fed babies additional foods. METHODS In 1969-70, 96% of the women between 15 and 44 years of age in 15 rural Chilean communities were interviewed in their homes. Data were sought on their reproductive histories, on their housing, and on other environmental and socioeconomic factors that could influence their health and fertility. The 1 712 mothers who had had deliveries in the preceding 5 years were questioned about the obstetrical care received, their infant feeding practices, and the medical attention given the child. Each was asked to state the duration of breast feeding and the reason for discontinuing it; how soon other milk started to be given and its source; and the baby's age when foods other than milk were introduced. If a woman had had 2 or more deliveries in the period, only the last was the subject of inquiry. Thus, the sample was not representative of the total population under 5 years of age. Neonatal deaths and living children under 4 weeks of age were excluded from the analyses. The level of statistical significance was preset at 0.01. RESULTS AND DISCUSSION Weaning Almost 90% of the women thought that breast feeding was much less common in their localities now than it had been 15 or 20 years ago. This could not be 2997 -203- S. J. PLANK & M. L. MILANESI confirmed and may only reflect the general assump- tion that early weaning is a recent phenomenon-an assumption shown by Knodel & Van de Walle (1967) to be inapplicable in certain European areas. On the other hand, there may well have been a recent shift toward earlier weaning like that documented by Sanjur et al. (1971) in a Mexican village. The pre- sumed decrease in breast feeding in the study area was ascribed to the convenience of bottles by the majority of respondents, but not one of them cited this as the determining factor in her own case. A number of benefits were attributed to breast feeding; only 2% of the women were not in favour of it, alleging that human milk was " too weak " or transmitted infections to the baby. More than two- thirds felt that breast feeding should be continued for at least a year, and fewer than 4% advocated its discontinuation in the first 6 months. In practice, however, only 40% of the children who had reached the age of 1 year were receiving any breast milk at all by then and nearly half of them had been completely weaned by 6 months. Of the mothers who had weaned their infants at under 6 months, more than half claimed that their milk production was inadequate. But, in view of the close association between early weaning and socio- economic conditions (see below), "inadequate" would seem to be a very relative criterion. Of the babies weaned at between 6 months and 1 year, 11 % were taken off breast feeding because of a new pregnancy. Only 1 of the 18 women concerned, however, had exclusively breast-fed her child during the 2 preceding months, as compared with 49 of 158 women weaning their children for other reasons during the second 6 months. It seems that an increased probability of conception ought to be counted among the risks of starting supplements to breast feeding (Jain et al., 1970; McKeown & Gibson, 1954; Perez et al., 1971; Sharman, 1951; Udesky, 1950). Employment of the mother, often cited as an important reason for weaning (Ariztia, 1955; Aykroyd, 1971; Bower, 1958; Vahlquist, 1955), was an insignificant factor in the communities studied. Of the 17% who were gainfully employed, fewer than half worked outside the home, and only 16 women claimed that their jobs kept them from breast feeding. Breast-feeding practices showed little varia- tion according to marital status, and, contrary to what has been observed in some other regions (Knutsson & Melbin, 1969; Cowgill & Hutchinson, 1963), boys were suckled no longer than girls. Table 1. Proportion of infants surviving at 6 months who were weaned before that age, by maternal age and parity, rural Chile, 1969-70* Maternalage ~~Number Proponion(% (years) Parity of children weaned before(years) ~~~included 6 monthsin study < 25 1-2 240 57.5 3-4 129 46.5 5-7 28 46.4 >8 2 50.0 Total: 399 Av.: 53.1 25-34 1-2 108 56.5 3-4 138 45.6 5-7 189 42.8 > 8 112 48.2 Total: 547 Av.: 47.3 > 35 years 1-2 33 39.4 3-4 34 55.9 5-7 82 36.6 >8 188 38.5 Total: 337 Av.: 39.9 all age groups 1-2 381 55.6 3-4 301 47.2 5-7 299 41.5 > 8 302 42.2 Four mothers did not supply complete information. With advancing age and parity, the duration of breast feeding was significantly longer (Table 1). Of the two variables involved, parity was shown to be the more significant (Mantel, 1963). From the physio- logical standpoint, early weaning would be expected to result in a higher parity at younger ages by reduc- ing birth intervals, as Kamal et al. (1969) observed among Egyptian women. The pattern found in Chile thus appears to be determined by social rather than biological factors, the high parity, late weaning group being composed of poorer, less educated women who married earlier, used effective contraception less, and depended to a greater extent on breast feeding for economic and family planning reasons. It is possible, however, that the generally younger age at first preg- nancy of high parity Chilean women favoured the establishment and maintenance of a more adequate production of milk (Dean, 1951; Hytten, 1954; Miller, 1952). Breast feeding decreased significantly as maternal education and paternal income rose (Table 2). Although these two variables were interdependent, the correlation with education continued to be highly 204 INFANT FEEDING AND INFANT MORTALITY IN RURAL CEILE Table 2. Proportion of infants surviving at 6 months who were weaned before that age, by maternal edu- cation and paternal income, rural Chile, 1969-70 ' Maternal Paternal Number Proportion (%) education income of children weaned before(years) (escudos a) includy 6 months <3 <300 197 35.5 300-549 108 40.7 >550 9 44.4 Total: 314 Av.: 37.6 3-5 < 300 166 45.8 300-549 138 46.4 550 24 33.3 Total: 328 Av.: 45.1 >6 < 300 91 54.9 300-549 83 59.0 > 550 115 70.4 Total: 289 Av.: 62.3 all groups < 300 454 43.2 300-549 329 47.7 > 550 148 62.8 Children of married women who provided income data. a Ten escudos were worth about USS1.00. significant when the data were standardized for income, whereas earnings were no longer significant when adjustment was made for years of schooling (Mantel, 1963). Education was also more closely correlated with the duration of breast feeding than was parity, again suggesting that infant feeding practices in the study population were largely deter- mined by psychosocial factors. Dried milk Of children being weaned or already weaned, 66% received partially defatted dried milk regularly under the National Health Service's distribution pro- gramme. Another 6% were eligible, but according to their mothers they refused the milk, or it " didn't agree " with them. The total figure, nevertheless, fell short of the Service's target of supplying dried milk for 85% of weanlings and preschool children. There was some criticism of the programme for not doing more, and the target figure was later raised to 100%. According to Jelliffe & Jelliffe (1971), however, the distribution of dried milk probably encourages early weaning. Their hypothesis was not tested directly, but only 22% of infants under super- vision at local health posts and thus eligible for dried milk began bottle feeding before 4 weeks as compared with 33 % of other infants. Their mothers tended to be older, poorer, and less educated than others, how- ever, and possibly would have continued breast feed- ing even longer if alternative milk had not been so readily available. Weaning and mortality Postneonatal deaths were significantly more fre- quent (relative risk, 3 to 1) among infants who started bottle feeding in the first three months than among those exclusively breast-fed during that time (Table 3). The investigation did not include an evaluation of the part played by qualitative and quantitative deficiencies associated with bottle feeding. These are known to interact (Newman, 1906; Scrimshaw et al., 1968), and Monckeberg (1970), in an investigation carried out in rural Chile in summer, found bacteriological contamination in 80% of bottles used for feeding babies and also observed that undernutrition was common (Monc- keberg et al., 1967b). An indeterminable part of the higher mortality associated with bottle feeding was an artifact due to the inclusion of low weight, high risk infants for whom supplementary milk was medically prescribed. However, in view of the socioeconomic correlates of bottle feeding and the related mortality (see below), this factor was probably relatively unimportant. Judging from the death rates, infants given bottles as a supplement to breast feeding fared no better than those who received no breast milk at all beyond Table 3. Mortality in the first year among infants surviving at 4 weeks, 3 months, and 6 months of age, by type of milk given, rural Chile, 1969-70 Deaths between age indicated Number and 52 weeksAge Type of milk of survivors Rate No. 1000 survivors 4 weeks breast only 1231 36 29.2 breast plus bottle 125 7 56.0 bottle only 215 13 60.5 3 months breast only 798 1 1 13.8 breast plus bottle 240 9 37.5 bottle only 413 16 38.7 6 months breast only 398 4 10.0 breast plus bottle 285 4 14.0 bottle only 604 12 19.9 -20-5 S. J. PLANK & M. L. MILANESI the fourth week. Potentially protective factors in human milk (Mata et al., 1967, 1971) seemed to have no bearing on postneonatal mortality. Earlier au- thors (Grulee et al., 1934; Howarth, 1905; Newman, 1906; Robinson, 1951) reported better survival rates with " mixed " breast and bottle feeding, but defined their cohorts retrospectively, thus including many infants who had been suckled exclusively for months before starting bottle feeding. In the Chilean study, a prospective approach was adopted and it was found that " mixed " feeders had no advantage as regards survival. As it was possible that deaths among " mixed diet" children occurred only after breast feeding stopped, all records were analysed week by week up to the end of the ninth month for each child to determine the total time spent on each regimen and the corresponding death rates. There were 86.5 deaths per 1 000 person-years of breast plus bottle feeding and 85.4 deaths per 1 000 person-years of bottle feeding alone. Between 4 and 12 weeks of age, which might be a critical period as regards the need for mother's milk and when other foods would not yet have intervened, the respective rates were 167.1 and 169.3. Again, there was no evidence that the continuation of breast feeding afforded any protec- tion once bottle feeding began. Non-milk foods The modal age for introducing any food other than milk (and the starchy additives occasionally used to stretch out powdered milk) was 6 months. Fewer than 40% of the infants were given additional foods any earlier. Mothers who started bottle-feeding early tended to introduce other foods sooner, too. Only 24% of the infants who were breast-fed, but not bottle-fed, were given anything but milk during the first 6 months, as compared with 45% of the bottle-fed infants. Newman (1906) observed undernutrition and ex- cess mortality among English infants who were exclusively breast-fed for 9 months or more, and Wyon & Gordon (1971) found that among Punjabi children, despite a lower incidence of diarrhoea, the risk of death was significantly greater if a milk-only diet was continued beyond the sixth month. Mater- nal milk alone evidently cannot meet the infant's needs after the first 6 months (Dean, 1959). The anomalous finding in Senegal (Cantrelle & Leridon, 1971) of a higher mean age of weaning for children dying in infancy than for those who survived suggests that there, as in Chile, prolonged breast feeding may have been associated with a significant delay in starting other foods. Because of the practice of early weaning in Chile, it was possible to extend the association between non-milk foods and survival back into the second trimester. Exclusively breast-fed children were not included in the analysis because they differed significantly from the high-risk bottle-fed children in respect of mortality as well as feeding. Although the death rates for bottle-fed babies were similar, whether they continued to receive breast milk or not, mortality was about 70% higher during both the second and third trimesters for those of them who were not being given any other food (Table 4). The difference, though not statistically significant, concorded with the earlier finding that, in Table 4. Total deaths and death rates per 1 000 person-years on different types of diet in the second and third trimesters of life, rural Chile, 1969-70 Basic diet 1: Basic diet 2: breast plus bottle bottle only Basic diets 1 and 2 Age (weeks) with without with without with without other food other food other food other food other food other food 13-25 person-years 19.2 50.9 37.6 100.3 56.8 151.3 deaths 1 3 2 10 3 13 death rate per 1 000 person-years 52.2 58.9 53.1 99.7 52.8 85.9 26-38 person-years 51.9 15.6 125.3 29.0 177.2 44.7 deaths 3 2 6 2 9 4 death rate per 1 000 person-years 57.8 128.0 47.9 68.9 50.8 89.5 both groups person-years 71.0 66.5 163.0 129.4 234.0 195.9 deaths 4 5 8 12 12 17 death rate per 1 000 person-years 56.3 75.1 49.1 92.7 51.3 86.8 206 INFANT FEEDING AND INFANT MORTALITY IN RURAL CHILE Table 5. Postneonatal deaths and death rates, by maternal age and 1969-70 * parity, rural Chile, Maternal age PaiyNumber of Deaths between Death rates per(years) Parity children included 4 and 52 weeks 1000 childrenin study < 25 1-2 240 8 3-4 129 4 5-7 28 1 >8 2 0 Total: 399 Total: 13 32.6 25-34 1-2 108 6 3-4 138 6 5-7 189 7 >8 112 5 Total: 547 Total: 24 43.9 >35 1-2 33 2 3-4 34 2 5-7 82 2 > 8 188 13 Total: 337 Total: 19 56.4 all age groups 1-2 381 16 42.0 3-4 301 12 39.9 5-7 299 10 33.4 >8 302 18 59.6 Four mothers did not supply complete information. terms of mortality, the risk from introducing addi- tional foods is outweighed by the benefit. Other factors associated with infant mortality These data tend to corroborate the impression that " suitable infant feeding ... is a greater factor than any other single thing" in preventing first-year deaths (Newman, 1906). Even the higher mortality among the children of older mothers (Table 5) appeared to depend on diet. While it is true that these mothers breast-fed their children longer, the crucial factor was evidently the introduction of bottle feeding rather than the duration of breast feeding, and mothers of all ages began bottle feeding at about the same time. Those over 35, however, were significantly less likely (31-.5% as compared with 41.6% ) to have introduced other foods by the end of the first 6 months, and nearly all the excess infant mortality related to maternal age occurred later than that. As shown in a study of infant mortality and standards of living by Behm (1962), economic forces greatly affect a child's chances of survival. However, at the economic level of these rural communities (only 10% of the fathers earned more than the equivalent of US$75 a month), whatever protection additional income might have afforded was evidently nullified by accelerated weaning. Although the less impoverished tended to introduce non-milk foods sooner, too, the proportion of infants receiving bottles but no additional foods at 6 months-those in greatest danger-rose from 34% to 44% as income increased; this was paralleled by a rise in the infant death rate from 42 to 54 per 1 000 infants (Table 6). Among children whose mothers had had more than 6 years' schooling, the effects of the progres- sively earlier weaning associated with better maternal education were apparently attenuated by the timely introduction of other foods. As the mothers' educa- tional level rose, the proportion of children who were bottle-fed but received no other food at 6 months increased from 33% to 45% and then fell to 37%, and the mortality followed a similar curve (Table 6). Sanitary conditions-presumably because of their association with income and education-were better in the homes where infant deaths occurred. There 207 S. J. PLANK & M. L. MILANESI Table 6. Postneonatal deaths and death rates, by income, rural Chile, 1969-70 v maternal education and paternal Maternal education Paternal income Number of Deaths between Death rates per (years) (escudos a) children included 4 and 52 weeks 1000 childrenin study <3 <300 197 7 300-549 108 4 >550 9 1 Total: 314 Total: 12 38.2 3-5 < 300 166 10 300-549 138 7 > 550 24 3 Total: 328 Total: 20 61.0 >6 < 300 91 2 300-549 83 5 > 550 115 4 Total: 289 Total: 11 38.1 all groups < 300 454 19 41.9 300-549 329 16 48.6 > 550 148 8 54.1 Children of married women who provided income data. a Ten escudos were worth about US $1. were 48.1 post-neonatal deaths per 1 000 infants in houses with running water (17% of the total) as compared with 32.3 in those without it; the rate was 36.0 in houses with some sort of sewage system (83% of the total) as compared with 30.0 in those that had none. The families affected lived in less overcrowded conditions than others, with fewer members per room and per bed, and their homes were more likely to be supplied with electricity. Prenatal medical care was taken as an index of the accessibility and use of health services. The risk of postneonatal death (relatively unaffected by prenatal care) was higher if mothers had visited a private physician or prenatal clinic during pregnancy than if they had had professional care only at deli- very or not at all (37.5 as compared with 27.3 per 1 000 infants). The inverse relationships of the infant mortality rates to family income, environmental factors, and medical care reinforce the conclusion that the dif- ferential mortality observed was attributable to bottle feeding and neglect of supplementary foods. ACKNOWLEDGEMENTS The authors gratefully acknowledge the support re- ceived from the Rockefeller Foundation and the US Agency for International Development, the technical assistance of the Department of Public Health and Social Medicine of the University of Chile, the cooperation of the Directors and personnel of the Chilean National Health Service, and the statistical help given by Dr James Warram. 208 INFANT FEEDING AND INFANT MORTALITY IN RURAL CHILE 209 RJfSUMt ALIMENTATION ET MORTALITt DES NOURRISSONS EN MILIEU RURAL AU CHILI Les pratiques en matiere d'alimentation des nourris- sons, leur contexte et leurs consequences ont et6 6tudi6s en 1969-70 dans 15 collectivites rurales du Chili grace a une enqu8te a domicile aupres de 1712 femmes ayant accouch6 au cours des cinq annees pr6c6dentes. Sur 1451 enfants ayant atteint au moins l'age de 3 mois, pres de la moiti6 avaient deja re;u un allaitement artificiel. Le risque de mortalit6 post-n6onatale 6tait chez eux trois fois plus elev6 que chez les enfants nourris uniquement au sein durant cette periode. Manifestement, le danger pour la survie des enfants resultait de la mise en route du processus de sevrage et non de son achevement; I'allaitement naturel poursuivi conjointement a l'allaite- ment artificiel ne conferait aucune protection apparente. En revanche, l'administration aux nourrissons d'ali- ments autres que le lait semblait accroitre leurs chances de survie; durant les 2e et 3e trimestres de la vie, les taux de mortalite etaient d'environ 70% plus 6leves chez les nourrissons recevant du lait pour seule nourriture que chez ceux qui beneficiaient d'un regime varie. L'emploi de l'allaitement artificiel etait d'autant plus precoce que l'education de la mere etait plus pouss&e et que les revenus du chef de famille etait plus eleves. Cette tendance se r6fletait dans les taux de la mortalit6 post- n6onatale. Bien qu'allant de pair avec un souci accru de donner precocement au nourrisson une nourriture variee, avec de meilleures conditions d'hygiene, un recours plus regulier aux soins medicaux et une frequence moindre de l'alcoolisme, l'el6vation du niveau de vie avait comme cons6quence paradoxale une mortalit6 plus forte chez les noufrissons. Les meres relativement agees ou ayant eu de nom- breuses grossesses allaitaient leurs enfants plus longtemps que les autres mais introduisaient les biberons dans l'alimentation a peu pres au meme moment. De ce fait, les taux de mortalit6 etaient du meme ordre chez tous les nourrissons pendant les 6 premiers mois, independam- ment de l'age ou du degre de parit6 de la mere. Apres cette periode, cependant, les femmes agees de plus de 35 ans etaient moins enclines a varier l'alimentation de leurs enfants et le taux de mortalite s'elevait chez ces derniers. Dans les collectivit6s etudiees, les conditions econo- miques, les services de sante et l'hygi6ne etaient d'un niveau superieur a celui de la plupart des zones rurales des pays en developpement, mais insuffisant pour que le sevrage precoce des nourrissons ne mette pas leur exis- tence en grave danger. On peut s'attendre a ce que dans d'autres regions les progres economiques entrainent aussi un abandon premature de l'allaitement maternel. Les pr6sentes observations donnent a penser que les incon- v6nients de cette pratique pourraient etre attenu6s par l'introduction d'aliments autres que le lait dans le regime des nourrissons. REFERENCES Ariztia, A. (1955) Quart. Rev. Pediat., 10, 150 Aykroyd, W. R. (1971) Amer. J. clin. Nutr., 24, 480 Becroft, T. & Bailey, K. U. (1965) J. trop. Pediat., 11, 28 Behar, M. (1964) Amer. J. publ. Hlth., 54, 1100 Behm, H. (1962) Mortalidad infantil y nivel de vida, Santiago, Universidad de Chile Behm, H. et al. (1964) Recent mortality trends in Chile, Washington, D.C., United States Department of Health, Education, and Welfare (United States National Centre for Health Statistics, Series 3, No. 2) Bower, B. D. (1958) Lancet, 1, 1379 Cantrelle, P. & Leridon, H. (1971) Population Studies, 25, 505 Cowgill, U. M. & Hutchinson, G. E. (1963) Hum. Biol., 35, 90 Dean, R. F. A. (1951) Spec. Rep. Ser. med. res. Coun. (London), No. 275, p. 346 Dean, R. F. A. (1959) Quart. Rev. Pediat., 14, 182 Gordon, J. E. et al. (1964) Bull. Wld Hlth Org., 31, 9 Grulee, C. G. et al. (1934) J. Amer. med. Ass., 103, 735 Harfouche, J. K. (1970) J. trop. Pediat., 16, 133 Howarth, W. G. (1905) Lancet, 2, 210 Hytten, F. E. (1954) Brit. med. J., 1, 912 Jain, A. K. et al. (1970) Demography, 7, 255 Jelliffe, D. B. & Jelliffe, E. F. (1971) Amer. J. clin. Nutr., 24, 1013 Kamal, I. et al. (1969) Amer. J. Obstet. Gynec., 105, 314 Knodel, J. & Van de Walle, E. (1967) Population Studies, 21, 109 Knutsson, K. E. & Melbin, T. (1969) J. trop. Pediat., 15, 40 Mantel, N. (1963) J. Amer. statist. Ass., 58, 690 Martin, W. J. et al. (1964) J. trop. Pediat., 10, 82 Mata, L. et al. (1967) In: Wolstenholme, G. E. & O'Connor, C. M., ed., Nutrition and infection, Boston, Little Brown (Ciba Foundation Study Group, No. 31) 210 S. J. PLANK & M. L. MILANESI Mata, L. et al. (1971) Amer. J. clin. Nutr., 24, 249 McKeown, T. & Gibson, J. R. (1954) J. Obstet. Gynaec. Brit. Emp., 61, 824 Miller, R. A. (1952) Arch. Dis. Childh., 27, 187 Monckeberg, F. (1970) In: Gyorgy, P. & Kline, 0. L., ed., Malnutrition is a problem of ecology, New York, Karger, p. 23 Monckeberg, F. et al. (1967a) Rev. Chil. Pediat., 7, 499 Monckeberg, F. et al. (1967b) Rev. Chil. Pediat., 7, 522 Newman, G. (1906) Infant mortality, London, Methuen Perez, A. et al. (1971) Population Studies, 25, 491 Plank, S. J. & Milanesi, M. L. (1973) Soc. Biol. (in press) Robinson, M. (1951) Lancet, 1, 788 Sanjur, D. et al. (1971) Bol. Ofic. Sanit. Panamer., 71, 281 Scrimshaw, N. S. et al. (1968) Interactions ofnutrition and infection, Geneva, World Health Organization (Mono- graph Series, No. 57) Sharman, A. (1951) J. Obstet. Gynaec. Brit. Emp., 58,440 Taba, A. H. (1970) J. trop. Pediat., 16, 212 Thomson, A. M. et al. (1966) J. Obstet. Gynaec. Brit. Cwlth., 73, 724 Udesky, I. C. (1950) Amer. J. Obstet. Gynec., 59, 843 Urrutia, J. J. & Mata, L. (1969) Arch. tat.-amer. Nutr., 19, 173 Vahlquist, B. (1955) Quart. Rev. Pediat., 10, 101 Wyon, J. B. & Gordon, J. E. (1971) The Khanna study, Cambridge, Harvard University