C International Epidemiological Association 1998

Printed In Great Britain

International Journal of Epidemiology 1998:27.255-260

A longitudinal study of prolonged breastfeeding in relation to child undernutrition

Wafaie W Fawzi,a M Guillermo Herrera, a ' b Penelope Nestel,b Alawia El Aminc and Kamal A Mohamed c
Background In a number of studies prolonged breastfeeding was associated with a higher risk of undernutrition, although most of these studies are limited by their crosssectional design which does not allow examination of temporal relationships between full weaning and undernutrition. Methods The relationship between prolonged breastfeeding and child growth was examined prospectively among children under 36 months old who participated in a large cohort study. At baseline and at each of three 6-monthly follow-up visits breastfeeding status was assessed and all subjects were weighed and measured. Results Undernourished children were more likely to be breastfed for a longer period of time compared with normal children. We found a small difference between breastfed and fully weaned children in the gain in height over the following 6-month period; however, breastfed children were likely to gain significantly less weight, particularly among children who were aged 6-12 months. Similar findings were noted when these associations were examined among children who were normally nourished at the time of breastfeeding assessment. The inverse association between breastfeeding status and weight gain was significantly larger among children of poor or illiterate mothers compared with children of relatively more affluent or literate mothers, respectively. Conclusions Our findings suggest that the inverse association is not causal, and may be explained by poorer complementary feeding among breastfed compared with weaned children. Children from poorer households and whose parents are illiterate are more likely to have less than adequate complementary feeding. The importance of adequate complementary feeding in the second half of infancy needs to be stressed in nutrition education programmes. Keywords Breastfeeding, nutritional status, wasting, stunting, developing countries, Sudan Accepted 21 July 1997

Downloaded from http://ije.oxfordjournals.org/ by guest on July 19, 2012

The benefits of breastfeeding in reducing the risk of infection and mortality and improving growth in the first 6 months of life have been known for some time.1'2 The relationship of breastfeeding and nutritional status in older children has been controversial.3 Prolonged breastfeeding, usually defined as breastfeeding beyond 6 months or one year of age, has been reported to be associated with a higher risk of undernutrition.4'5 In other studies, however, prolonged breastfeeding was protective6 or
a

Department of Nutrition, Harvard School of Public Health, Boston, MA, b Harvard Institute for International Development, Cambridge, MA, USA. c Ministry of Health, Khartoum, Sudan. Reprint requests to Dr Wafaie Fawzi, Department of Nutrition, Harvard School of Public Health, 665 Huntlngton Avenue, Boston, MA 02115, USA.

had no impact on nutritional status.7 Using data from the Demographic Health Surveys (DHS) carried out in 19 developing countries, it was recently reported that children who were breastfed beyond the first year of life were shorter and lighter compared with non-breastfed children.8 Most of the studies that have examined this question, including the large and representative DHS, are limited by their cross-sectional design which does not allow examination of temporal relationships between full weaning and undernutrition. In addition, only a few of these studies included adjustment for possible confounding variables such as socioeconomic status, and fewer still included an examination of the findings within strata of important modifiers. In this report, the relationships between breastfeeding and nutritional status were examined among children who participated in a large cohort study carried out in the Sudan. We

255

256

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY children were still breastfed at ^ 3 6 months of age, all analyses in this paper are limited to children who were <36 months. At each visit, interviewers asked whether each child was exclusively breastfed, partially breastfed (breastfed and bottle fed or breastfed and receiving solid diet), exclusively bottle fed, or fully weaned on solid foods. We defined breastfeeding in two ways and examined the relationship of each definition with nutritional status. In the first definition, we merged partial and exclusive breastfeeding and compared it with receiving no breast milk at all. Given that only a few children at any of three periods were either exclusively breastfed (n = 212 child-periods), exclusively bottle-fed (n = 38 child-periods), or both breast- and bottle-fed (n = 148 child-periods), we excluded these three categories from the second definition and compared children who were on breast milk and solid foods (n = 6808 child-periods) to those on solid foods alone (n = 21 573 child-periods). Results with both definitions of breastfeeding were similar so we present the findings using the second definition. We examined whether the nutritional status and socioeconomic status of children at the beginning of a 6-month period are predictive of the risk of 'being weaned' by the end of the period. For this analysis, logistic regression was used and the data were limited to children who were breastfed at the beginning of a round, and hence at risk of weaning. We examined this question by round of data collection to account for the independence between observations across rounds. The predictors of weaning identified in the results by round were essentially the same as those when the total number of child-periods were used, hence we present the latter analysis. Linear regression was used to examine the relationships between breastfeeding status at a particular round and height or weight at the same visit (cross-sectional analyses), as well as changes in these measurements over the following 6 months (longitudinal analyses). These analyses were carried out in 3-month or 6-month strata by age at the beginning of a round. Since undernutrition was associated with continuation of breastfeeding, we repeated the same analyses among children who were normally nourished at the beginning of each round. We next explored whether the relationship of breastfeeding to height or weight gain over a 6-month period was modified by levels of wealth or maternal literacy. These results were adjusted for a number of potentially confounding variables at baseline, namely age by month (within a 3- or 6-month stratum), sex, dietary vitamin A intake, capsule (vitamin A or placebo), morbidity, household wealth, availability of water in the house, and maternal literacy. Variances generated by logistic and linear regression models were used to estimate 95% confidence intervals. We used a probability level of <0.05 to define statistical significance. Data were analysed using Statistical Analysis System software, SAS Institute, Cary, NC. The study was approved by the Committee on the Use of Human Subjects in Research at the Harvard School of Public Health, the Director General of Primary Health Care at the Ministry of Health in the Sudan, and by both Directors of Health for Khartoum and Central Regions.
Downloaded from http://ije.oxfordjournals.org/ by guest on July 19, 2012

prospectively examined the relationship between breastfeeding status and gain in height or weight over the following 6 months in the total study population, among children who were normally nourished at the time of breastfeeding assessment, and in groups of children stratified according to household wealth or maternal literacy.

Methods
The study was initiated in June 1988 to examine the relationships between vitamin A supplements and the health and survival of preschool children. Additional details on the study design and the study population have been published elsewhere. ' The study population comprised 28 753 boys and girls aged 6-72 months in five rural councils in northern Sudan. Equal numbers of boys and girls were enrolled. About 28% of the mothers were literate and 50% of the households had running water in the house. Interviewers enrolled all eligible children at baseline (round 1) and subsequently visited each household three times at 6-monthly intervals. Field personnel were divided into six teams, each consisting of two interviewers, two anthropometridans and a supervisor. At each round, each household was visited; anthropometric measurements were done by each team at a central location after household visits were completed. Follow-up rates at rounds 2, 3, and 4 were 92.1%, 87.5%, and 84.2% respectively. Children who were not available at round 4 included those who had died, those who were diagnosed to be xerophthalmic at rounds 2 or 3 and were excluded from further follow-up, and those lost to follow-up, mostly due to absence of the mother from the home at the time of the follow-up surveys. At each round, the interviewers measured each child's weight and height. Weight was measured using a Salter scale to the nearest 100 g. Height (or recumbent length for children <85 cm) was measured to the nearest 1 mm with a locally made anthropometer. The Centers for Disease Control Anthropometric Software Package (CASP) was used to calculate anthropometric indicators. All children with Z scores below -2 were considered undernourished: such children were considered wasted if only deficient in weight-for-height; stunted if only deficient in height-for-age; or both stunted and wasted if they were deficient in both parameters. Children at or above -2 Z scores in both weight-for-height and height-for-age were classified as not undernourished or normally nourished. Interviewers also assessed child dietary vitamin A intake at each round by administering to mothers a simple 24-h recall questionnaire. Approximate dietary intake of total vitamin A was computed by multiplying the nutrient content of each food item by an assumed average portion. 10 At baseline, information was collected on potential determinants of nutritional status, including household wealth subjectively assessed on a four-point scale, availability of water in the house, and maternal literacy. The latter three variables were used as indicators of socioeconomic status. Morbidity at each visit was assessed by asking the mother whether in the preceding 7 days the child had diarrhoea, cough, fever, or measles. These symptoms were grouped into the following six mutually-exclusive categories: diarrhoea alone, diarrhoea with fever, cough alone, cough with fever and/ or diarrhoea, fever alone, and measles irrespective of other symptoms associated with it. Given that a small proportion of

Results
In all, 81% of children were breastfed by 12 months of age, but the prevalence declined to 62% and 27% by 18 and 24 months

PROLONGED BREASTFEEDING AND GROWTH

257

Table 1 Risk factors of stopping breastfeeding during a 6-month period among children who were breastfeeding at the beginning of the period 3 Covarlate Nutritional status Normal Wasted Stunted Wasted and stunted Unknown Wealth Mother literate Yes No Water In house Yes No No. (%) 1672 (71.8) 285 (75.2) 1989 (73.3) 345 (78.2) 358 (75.4) 6335 (74.2) 1260 (76.1) 3383(72.4) 1888(74.2) 2761 (72.9) Multivariate relative risk (95% CI) 1.0 (referent) 1.17 (0.90-1.52) 0.52 (0.45-0.60) 0.73 (0.56-0.94) 0.76 (0.59-0.97) 0.92 (0.84-0.99) 1.0 (reference) 0.87(0.75-101) 1.0 (reference) 1.14 (1.0-1.31)

of age respectively. There was no difference between boys and girls in the likelihood of stopping breastfeeding at any age. Compared with normally nourished children, those who were stunted were half as likely to be fully weaned (95% CI: 0.45, 0.60), while those who were both wasted and stunted were about 0.73 (95% CI : 0.56, 0.94) times as likely to be weaned in the following 6 months (Table 1). Children who came from relatively affluent families and children of literate mothers were weaned earlier than children from poorer backgrounds and those whose mothers were illiterate. Having running water in the house was also associated with earlier weaning. We examined the relationship between breastfeeding status and height or weight measured at the same time, as well as change in these measurements over the following 6 months. In cToss-sectional analyses, breastfed children were significantly shorter and lighter than weaned children (Table 2). There were small differences between breastfed and weaned children in the gain in height over the following 6-month period (Table 2). Breastfed children were also likely to gain less weight, compared with weaned children, particularly among those who were aged 6-12 months. Among children 9-12 months old, those who were breastfed gained 308 g less (95% CI :-414, -201) compared with weaned children. Given that undernourished children were more likely to be breastfed for a relatively longer period, we examined the relationship between breastfeeding status and gain in height or weight among children who were normally nourished at the time of breastfeeding assessment (Table 3). The change in height was negligible between breastfed and weaned children. Breastfed children tended to gain less weight and the difference

' From a logistic regression model that included age as a continuous variable, sex, morbidity (five dummy variables), dietary vitamin A intake as a continuous variable, capsule (vitamin A/placebo), maternal literacy, household wealth as subjectively assessed by the field worker (four-level ordinal variable with the relatively affluent as the reference group), and availability of water m the house. Only children who were breastfed at the beginning of a round were included in the model.

Downloaded from http://ije.oxfordjournals.org/ by guest on July 19, 2012

Table 2 Relationship of breastfeeding to baseline height/weight and to change in these measurements over a 6-month period Cross-sectional analyses Breastfed mean (SD) Weaned mean (SD) Change over 6 months Multivariate analysts (95% CI) b 0 (-12, 11) -6 (-10, -2) -2 (-5, 2) -2 (-4, 0) -3 (-5,-1) -2 (-4,0) -1 (-3, 0) -278 (-522, -33) -308 (414, -201) -73 (-169, 16) -99 (-155,^12) -54 (-120, 13) -87 (-154,-20) -120 (-170,-70)

Characteristic/Age Height (mm) 6.0-8.9 mo 9.0-11.9 mo 12.0-14.9 mo 15.0-17.9 mo 18.0-20 9 mo 21.0-23.9 mo 24.0-35.9 mo Weight (grams) 6.0-8.9 mo 9.0-11.9 mo 12.0-14.9 mo 15.0-17.9 mo 18.0-20.9 mo 21.0-23.9 mo 24.0-35.9 mo

Number of children
214

Multivariate analysis (95% CI)a -^0 (-55, -25) -38 H 5 , -32) -21 (-26,-17) -19 -30 -46 -67 (-22,-15) (-34,-26) (-49, -42) (-70, -64)

1246 1496 2412 2538 3844 14212

213

684(35) 703(39) 726 (35) 743 (37) 744 (54) 746 (56) 755 (53) 7573 (1044) 7837 (1109) 8204 (1113) 8539(1149) 8662 (1382)
8596 (1418)

738 745 751 764 778 796 836

(73) (64) (54) (47) (48) (47) (55)

1247 1496 2412 2540 3848 14 212

8482 (1635) 8572 (1609) 8669 (1449) 8959(1330) 9307 (1371) 9739 (1407) 10 645 (1063)

-704 (-1102, -307) -673 (-841, -506) -344 (^175,-214) -368 (-463, -274) -529 (-636,-421) -1016 (-1125,-907) -1406 (-1500, -1313)

8877 (1444)

" From individual linear regression models for each 3-month stratum. For each model, the relationship between breastfeeding and measurement (height or weight) at the following round was examined and adjusted for measurement at the previous round, age as a continuous variable, sex, morbidity (five dummy variables), dietary vitamin A as a continuous variable, capsule (vitamin A/placebo), maternal literacy, wealth, and availability of water in the house. b From individual linear regression models for each 3-month stratum. For each model, the relationship between breastfeeding and measurement (height or weight) at the same round was examined and adjusted for age as a continuous variable, sex, morbidity (five dummy variables), dietary vitamin A as a continuous variable, capsule (vitamin A/placebo), maternal literacy, wealth, and availability of water In the house.

258

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 3 Differences in height/weight change over a 6-month period among breastfed and weaned children who were normally nourished Age
6 0-11.9 mo 12.0-17.9 mo 18.0-23.9 mo 24.0-35.9 mo
3

No. children 1010 1986 2640 7013

Change in height (95% CI)a 0 (-5. 4) -1 (-3, 1) 0 (-2, 2) -6 (-9, -2)

Change in weight (95% CI)a -263 (-381,-144) -36 (-151, -22) -11 (-84, 62) 6 (-128, 140)

Table 4 Differences in weight and height changes over a 6-month penod between breastfed and weaned children categorized by household wealth and maternal literacy Age No. children Poor b 6-11.9mo 611 12-17 9 mo 1725 18-23.9 mo 2860 24-35.9 mo 6296 Change in height (95% CI)a -10 (-17,-30) -1 (-*. 2) -2 (-4, 1) -3 (-5, 0) -3 (-8, 2) -2 (-4, 0) -3 (-5,-1) 0 (-2, 3) -6 (-11,-2) -2 (-4, 1) -2 (-4, 0) -1 (-3, 1) -3 (-10, 5) -2 (-6. 1) -1 (-7,-1) Change In weight (95% CI)a -449 (-615, -284) -133 (-211, -54) -94 (-167, -21) -205 (-279, -131) -235 (-357, -114) -56 (-116, 5) -44 (-105, 18) -38 (-106, 30) -365 (-477, -252) -118 (-174, -62) -80 (-137, -24) -133 (-193, -74) -188 (-387, 12) -14 (-105, 76) -31 (-115, 53) -88 (-179, 4)

From individual linear regression models for each 6-month siratum For each model, the relationship between breastfeeding and measurement (height or weight) at ihe following round was examined and adjusted for measurement at the previous round, age as a continuous variable, sex, morbidity (five dummy variables), dietary vitamin A as a continuous variable, capsule (vitamin A/placebo), maternal literacy, wealth, and availability of water in the house.

Relatively affluent1"
933 6-11.9 mo 2477 12-17.9 mo 4085 18-23.9 mo 9101 24-35.9 mo Mother illiterate 1151 6-11.9 mo 3008 12-17.9 mo 4929 18-23.9 mo 10 768 24-35.9 mo Mother literate 391 6-11.9 mo 1173 12-17 9 mo 1977 18-23.9 mo 4539 24-35.9 mo
a

was greatest for children between the aged 6-12 months: -263 g (95% CI : -381, -144). We next examined the association of breastfeeding and gain in height or weight using the total population stratified by a subjective assessment of household wealth and by maternal literacy (Table 4). We noted an inverse association between breastfeeding status and gains in height or weight that was apparently of larger magnitude for the youngest children (612 months at baseline) and those who had poorer or illiterate mothers compared with older children or those whose mothers were more affluent or literate. In general, the biological significance of these differences were stronger for weight than for height. Children from poorer households who were breastfed at the age of 6-12 months gained 10 mm less than weaned children (95% CI :-17, -30), compared with 3 mm difference among children from more affluent households (95% CI : -8, 2); a t-test was statistically significant (P < 0.0001). Breastfed children in the same age range and from poorer households gained 449 g less than weaned children (95% CI : -615, -284), compared with a 235 g difference among children from more affluent households (95% CI : -357, -114); a t-test was statistically significant (P < 0.0001). Similarly, children of illiterate mothers who were breastfed gained less weight compared with children in the same age group but whose mothers were literate. Children who were breastfed at between 6 and 12 months and whose mothers were illiterate gained 365 g less compared with weaned children (95% CI : -477, -252), while breastfed children in the same age bracket and whose mothers were literate gained 188 g less compared with weaned children (95% CI :-387, 12). A t-test was also statistically significant (P < 0.0001).

Downloaded from http://ije.oxfordjournals.org/ by guest on July 19, 2012

-1 M. 2) From individual linear regression models for each stratum. For each model, the relationship between breast feeding and measurement (height or weight) at the following round was examined and adjusted for measurement at the previous round, age as a continuous variable, sex, morbidity (five dummy variables), dietary vitamin A as a continuous variable, capsule (vitamin A/placebo), maternal literacy, wealth, and availability of water in the house. Household wealth was subjectively assessed by the field worker on a fourpoint scale which is grouped in these analyses into two levels

Discussion
The association between current breastfeeding and height or weight has been examined in a number of studies with conflicting results:3 an inverse association was reported in the majority of the studies, 4 ' 5 although in a few studies a protective6 or null 7 protective association was observed. Most of the published studies are limited by their cross-sectional design which does not account for the possibility that undernutrition may be the reason for prolonged breastfeeding, rather than vice versa. We found a similar inverse association in this study when we examined the data cross-sectionally. We also found that undernourished children were more likely to be breastfed for longer durations

compared with normally nourished children. The latter finding was also reported in a study from Guinea Bissau" providing one explanation for this inverse association. Longitudinal data are necessary to examine the relationship between prolonged breastfeeding and nutritional status. To our knowledge, this is the first large longitudinal study to examine this question. The consistent and significant inverse associations that we and others observed with cross-sectional analyses were not observed when the data were examined prospectively. There were inverse associations with height and weight gain in favour of weaned children which were more notable among the youngest children in the sample: 6-12 months old. To exclude the effect of prolonging breastfeeding because a child was undernourished, we examined these associations among children who were normally nourished at the time of the breastfeeding assessment. We noted a consistent finding of a more marked inverse relationship with respect to weight gain among children who were aged 6-12 months. Given that a small proportion of these children were below 9 months of age, these results are largely reflective of the associations among children who were 9-12 months old. The apparently greater risk associated with breastfeeding occurs shortly after the age range (46 months) during which complementary feeding is introduced in many developing countries. These Findings may be confounded by

PROLONGED BREASTFEEDING AND GROWTH

259

differential levels of food intake between breastfed and weaned children around the time of introduction of complementary feeding. Unfortunately, dietary intake was not carefully assessed in any of the published studies including this one. We have data on dietary vitamin A intake in this population; while this is not a good marker of total food intake, it may shed some light on the dietary patterns between breastfed and weaned children. Vitamin A from breast milk was not included in this calculation of intake. Breastfed children consumed significantly less vitamin A from foods (excluding breast milk) compared with weaned children, suggesting that breastfeeding mothers may be under the impression that breast milk provided adequate nourishment and have thus not provided, out of lack of information or because of poverty, adequate complementary feeding. To examine this explanation for our findings, we assessed the association of breastfeeding status and nutritional status among children categorized by household wealth or maternal literacy. Marked differences in weight gain and height gain favouring weaned children were also noted among children <12 months of age in all strata. However, these differences were significantly greater for children who were from poorer households compared with those from relatively affluent ones, and similarly for children of illiterate mothers relative to those bom to literate mothers. This finding supports the explanation that breastfed children were more likely to be provided with more inadequate complementary feeding. Women with minimal formal education are less likely to have access to information about nutrition; also parents from poorer backgrounds are less able to provide for adequate complementary feeding, and infants and young children in such households have to rely on breastfeeding for a longer duration (past 6 months of age) with minimal additional foods. These findings are supported by the results of a study in which the growth and lactation performance of breastfed infants in the US were compared with a group of poor Peruvian infants. 13 Compared with US infants, those from Peru started faltering after 6 months of age, although both groups had comparable milk volume and density. The Peruvian group, however, received less energy and much less protein from non-breast milk sources at 9 and 12 months of age than did the US infants. Our findings are also in accord with the only other prospective longitudinal study from Guinea Bissau." In cross-sectional analyses in the latter study, breastfed children at all ages were more likely to be malnourished compared with weaned children. Children older than one year who were weaned during the study were weighed before (0-100 days) and after (30-130 days) weaning. There was no change in Z score at the time of weaning. Given the observational design of our study and the one from Guinea Bissau, it is difficult to exclude the possibility of residual confounding by mismeasured variables of socioeconomic status or morbidity. In addition, the contribution of complementary feeding to the relationships of prolonged breastfeeding and nutritional status should be studied in future studies. Prolonged breastfeeding has other benefits of public health significance. The possible protective effects of prolonged breastfeeding on child mortality and morbidity have been controversial. In a few studies, children who were breastfed past 6 months of age experienced a significant reduction in mortality, 14 or reductions in diarrhoea 15 and mortality," while in other studies no such findings were observed. 1 ' 16 Prolonged breastfeeding was not associated with mortality in the Sudan study population

(WW Fawzi et al., unpublished results). However, harmful prospective relationships between malnutrition and child mortality were noted in the Sudan study 17 and in another from Bangladesh, 18 and these were significantly reduced among children who were breastfed compared with fully weaned children. These findings suggest that breastfeeding is associated with a higher likelihood of survival among malnourished children. Breastfeeding is also associated with family planning benefits in many traditional societies with secondary benefits in terms of wider child spacing and better child and maternal health. In conclusion, there is no evidence that prolonged breastfeeding perse is associated with poorer nutritional status. Rather we found a higher risk of malnutrition among breastfed infants from poorer households and whose parents are illiterate and are thus less likely to have adequate complementary feeding. Breastfeeding for up to 18 months of age and beyond is common in many developing countries. Given that prolonged breastfeeding may have additional benefits including reductions in child morbidity and mortality, rather than ignoring the importance of prolonged breastfeeding, public health workers need to reinforce the importance of adequate complementary feeding beyond 46 months of age.

Downloaded from http://ije.oxfordjournals.org/ by guest on July 19, 2012

Acknowledgements
This study was carried out under cooperative agreement No. DAN-0045-G-SS-6067 of the Office of Nutrition, US Agency for International Development, Washington, DC, and the Harvard Institute for International Development, Cambridge, MA. We would like to thank the field teams and the mothers and children who made the study possible.

References
' Feachem RG, Koblinsky MA. Interventions lor the control of diarrhoeal diseases among young children: promotion of breast-feeding. Bull World Health Organ 1984;62:271-91.
2

Asha Bai PV, Leela M, Subramaniam VR. Adequacy of breast milk for optimal growth of infants. Trop Geogr Med 1980:32:158-61 Grummer-Strawn LM. Does prolonged breastfeeding impair child growth? A critical review Pediatrics 1993;91:766-71 Victora CG, Vaughan JP, Martines JC, Barcelos LB. Is prolonged breastfeeding associated with malnutrition ? Am J CUn Nulr 1984;39:307-14. Brakohiapa LA, Yartey J, Bille A et al. Does prolonged breastfeeding adversely affect a child's nutritional status? Lancet 1988:1:416-18.

Taren D, Chen J. A positive association between extended breastfeeding and nutritional status in rural Hubei Province, People's Republic of China. Am J Clin Nutr 1993:58:862-67. Rao S, Kande AN. Prolonged Breastfeeding and malnutrition among rural Indian children below 3 years of age Europ J Clm Nutr 1992; 46:187-95.

Caulfield LE, Bentley ME, Ahmed S. Is prolonged breastfeeding associated with malnutrition? Evidence from nineteen Demographic and Health Surveys. Int J Epidemiol 1996:25:693-703. Herrera MG, Nestel P, El Amin A, Fawzi WW, Muhammad KA, Weld L. Vitamin A supplementation and child survival. Lancet 1992; 340:267-71. Fawzi WW, Hen-era MG, Willetl WC et al. Dietary vitamin A intake and the nsk of mortality among children. Am J Clin Nutr 1994:59: 401-08.

10

260

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

11

Moblak K, Gottschau A, Aaby P, Hojlyng N, Ingholt L, Jose da Silva AP. Prolonged breastfeeding, diarrheal disease, and survival of children m Guinea-Bissau. BrMed J 1994;308:1403-06.

" Clemens JD, Stanton B, Stoll B, Shahid NS, Banu H, Chowdhury AKMA. Breastfeeding as a determinant of severity in shigellosis Am J Epidemiol 1986; 123:710-20.
16

Fawzi WW, Herrera MG, Nestel P, El Amin A. Mohamed KA. Risk faaors of low dietary Vitamin A intake among children in the Sudan. East Afr Med J 1997:74:227-32 13 Dewey KG. Peerson JM, Heinig MJ et al Growth patterns of breastfed infants In affluent (United States) and poor (Peru) communities: implications for timing of complementary feeding. Am J Clm Nutr 1992;56:1012-18.
14

12

Yoon PW, Black RE, Moulton LH, Becker S. Effect of not breastfeeding on the risk of diarrheal and respiratory mortality in children under 2 years of age in Metro Cebu, the Philippines. Am J Epidemiol 1996; 143:1142-48. l7 F a w a WW, Herrera MG, Spiegelman D, El Amin A, Nestel P, Mohamed KA. A prospective study of malnutrition in relation to child mortality In the Sudan. Am J Clin Nutr 1997:65:1062-69. Briend A, Wojtyniak B, Rowland MGM. Arm circumference and other faaors in children at risk of death in rural Bangladesh. Lancet 1987:11: 725-28

Briend A, Wojtyniak B, Rowland MGM. Breastfeeding, nutritional state, and child survival in rural Bangladesh. Br Med J 1988:296: 879-82.

18

Downloaded from http://ije.oxfordjournals.org/ by guest on July 19, 2012