O International Epldemiological Association 1998 Printed In Great Britain
International Journal of Epidemiology 1998:27.4*4-^89
Food diversity versus breastfeeding choice
in determining anthropometric status in
rural Kenyan toddlers
Adelheid Onyango, a Kristine G Koskia and Katherine L Tuckerb
Background Prolonged breastfeeding in developing countries is routinely recommended as a
valuable and cost-effective public health measure to promote early childhood
growth. However, the effects of breastfeeding beyond 12 months are unclear,
with some studies showing positive, and some showing negative effects. The role
of complementary foods for children 1-3 years has been less studied.
Methods
We examined feeding behaviour and illness data in relation to anthropometric
status among 154 rural western Kenyan children, aged 12-36 months.
Results
There was little difference in anthropometric status between partially breastfed
and fully weaned children. Rather, dietary diversity (number of different foods
consumed) was strongly and consistently related to anthropometric status in this
age group. In addition, early complementation with starchy gruels was associated
with stunting.
Conclusions Public health efforts which focus only on prolonged breastfeeding (>12 months)
in developing countries will not ensure adequate early childhood growth. Important complementary feeding recommendations that promote diet diversity, through
the inclusion of a variety of foods in the diets of children in the 1-3 year age
group, are needed.
Keywords
Breastfeeding, weaning, dietary diversity, complementary feeding, anthropometry,
Africa
Accepted
28 August 1997
Exclusive breastfeeding is currently recommended in developing countries for the first 4-6 months. With appropriate complementary food, the continuation of breastfeeding is recommended
for up to 2 years or more. 1,2 Prolonged breastfeeding has been
associated with improved child survival in some studies.3"5 However, the value of breastfeeding for nutritional status beyond 12
months has recently been questioned. Some studies have shown
no benefit to nutritional status, 3 4 and others have shown
poorer nutritional status among breastfed versus fully weaned
children aged 12 months or older. 6 " 10 Two reports on subSaharan African children suggested that the negative association
is strongest between 12 and 24 m o n t h s . " 1 2 Caulfield et al.xl
have hypothesized that mothers in sub-Saharan African
and some other countries may be more likely to continue
" School of Dietetics and Human Nutrition. Macdonald Campus ol McGUI
University.
h
Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts
University.
Reprint requests: Dr Katherine L Tucker. Jean Mayer USDA Human Nutrition
Research Center on Aging at Tufts University, 711 Washington St.. Boston,
MA 02111. USA.
breastfeeding if their children are small rather than breastfeeding choice perse leading to poorer growth. A recent study by
Marquis et al. provides some support for this hypothesis in
Peruvian toddlers.
Many studies on prolonged breastfeeding and nutritional
status have controlled for maternal and household confounders,
but few have carefully examined the role of non-breast milk
food intake among breastfeeding children aged 12 months or
older, although several have noted that complementary foods
play an important role. Low food intake in Mali 14 and low
energy intake in Ghana 7 have been associated with prolonged
breastfeeding. A positive association between intake of complementary foods (yes or no) and weight for height was reported
in Yemeni toddlers, 12-23 months old. 15 Chinese children
(12-47 months old) whose complementary diets consisted of
3>3 food groups at 12 months had better height for age scores
compared to those provided with < 3 food groups. 16 Cousens
et al}1 found that when prolonged breastfeeding was accompanied with complementary solid foods, there was a reduction
in clinical malnutrition in Burkina Faso. In this study, we examine the relationship between prolonged breastfeeding,
diversity of weaning food use, and the timing of introduction of
484
WEANING FOODS AND ANTHROPOMETRIC STATUS IN KENYAN TODDLERS
complementary foods with anthropometric measures of
nutritional status among 154 children aged 12-36 months in a
rural district in western Kenya.
Methods
Data were collected in September through November, 1988
from families which included both parents or defacto female
heads of household in six villages in Busia district of western
Kenya. A door-to-door survey of all village households identified 194 children in the target age range of 12-36 months. All
were invited to join the study. Only one household refused.
Another was excluded due to a congenital malformation. Another 38 had incomplete data, (most because of missing visits
due to extended visits to relatives outside the study location),
leaving 154 with complete data for this analysis. Comparison of
basic demographic data did not show differences between those
with or without complete data. Details of the field methodology
have been published. 18 Breastfeeding status was ascertained at
the second visit, along with anthropometric measurement.
Three non-consecutive 24-hour dietary recall interviews were
used to describe food intake for all study children. Mothers
verbally reported all foods and beverages consumed by the child
along with recipes for each prepared food item and proportions
consumed by the child. Older siblings helped to report foods the
child had consumed without the mother's observation. Frequency and quantity of breast milk intake were not estimated.
Mothers were asked to recall when they had first introduced
porridge, fruit and other foods into their children's diets, and if
no longer breastfeeding, why. They were also asked to assess
their child's appetite by indicating where it fell on a five-point
Likert scale, from poor to excellent.
Dietary data were processed using the CANDAT nutrient analysis system. 19 Nutrient composition data for foods consumed in
East Africa20'21 were incorporated into the program and used in
the analysis of nutrient intakes by the children. Daily nutrient
intake estimates were compared to the WHO/FAO recommendations for specified age categories.22 A dietary diversity
score was computed by counting the number of unique food
items consumed by the child during each of the 3 days of recorded intake. The score used here is an average of the 3 days.
Information on parental and household characteristics was
obtained by questionnaire interview. On each of the three visits,
mothers reported whether or not the study child had been ill
with diarrhoea or malarial fever during the past month, or since
the previous visit, and how long the illness lasted. The intervals
between the second and third visits varied from 2 weeks to a
month for different households. Differences in study duration
were, therefore, corrected by expressing days of illness for each
child as a proportion of the total number of days of morbidity
recall for each household.
Anthropometric measurements were taken according to
United Nations procedures.
Mid-upper arm circumference
(MUAC) measurements were taken with a standard arm band
provided by Ross Laboratories (Columbus, Ohio) and triceps
skinfold (TSF) measurements, with Lange skinfold calipers.
Children were weighed on a hanging Salter (UK) scale. A length
board was locally constructed following UN guidelines.23 Supine
lengths of all children who had not reached their second birthday were taken, while standing height was recorded for those
485
who were over 2 years old. Precise birth dates were read from
clinic cards kept by all except three mothers. Height for age
(HA), weight for age (WA) and weight for height (WH) were
interpreted according to the National Center for Health and
Statistics (NCHS) reference standards, using the Anthropometric Software Package, prepared by the Centers for Disease
Control. 24
Statistical analysis was done with SPSS for Windows (version
7.0). 25 Breastfeeding was defined as a dichotomous variable:
partially breastfed (PBF) or fully weaned (FW)—no child was
exclusively breastfed. Characteristics of the household, mother
and child were compared across these two groups using general
linear models, controlling child's age in months to obtain adjusted means. For dichotomous variables, adjusted comparisons
were made using logistic regression. Logistic regression analysis
was also used to determine the combination of factors associated
with prolonged breastfeeding, and reasons for termination of
breastfeeding before age 24 months, as reported by mothers,
were summarized. The proportion of children in each group
consuming individual food items was tallied in order to describe
and rank the foods contributing to these weaning and postweaning diets.
Anthropometric measures (NCHS Z-scores for WA, HA and
WH; actual measures for triceps skinfold and mid-upper arm
circumference) were regressed on feeding and illness variables,
adjusting for age of child, household income and mother's
education. Breastfeeding at the time of measurement (no = 0 or
FW; yes = 1 or PBF) was included in all equations. Other feeding
variables included dietary diversity, early introduction of cereal
(<4 months of age), late introduction of cereal (>6 months);
and similarly, early and late introduction of fruits. Illness
variables included the proportion of time ill with diarrhoea and
of time ill with malaria during the study period. All these main
effect variables were initially kept in the equation and interactions between each of them and breastfeeding status were
allowed to enter, using a forward selection approach. In subsequent model reduction steps, interactions significant at P < 0.05
were kept, along with their component variables, child's age,
household income and mother's education; the remaining main
effect variables were tested and, if non-significant at P < 0.05,
removed by backward elimination. Mean energy and nutrient
intakes and anthropometric measures were then compared in
analysis of covariance models by feeding status group and diet
diversity level (* or >5), controlling child's age, household
income and mother's education.
Results
Characteristics by feeding pattern
Ninety-eight (64%) of the 12-36-month-old children in this
rural Kenyan sample were fully weaned (FW) and 56 (36%)
were receiving breast milk (PBF) in addition to the consumption of other foods from the household diet. Ninety-four per cent
of children aged between 12 and 17 months received breast
milk. Termination of breastfeeding occurred most frequently
between 18 and 23 months and within this range, 50% were
FW and 50% PBF. Ten per cent of children over 24 months of
age continued to breastfeed.
Table 1 summarizes general characteristics by breastfeeding
status. The means of continuous variables, adjusted for child's
486
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 1 Characteristics by feeding pattern*
Partially breastfed
(n = 56)
Household variables
Amount spent on food/week (Kshil)
Father's education level
No. children <5 years
No. children who have died
Female head (% yes)
Maternal variables
Age (years)
Years in school
Pregnant (% yes)
Child variables
Dietary diversityd
Appetite rating e
% time ill with diarrhoea
% time ill with malaria
Male (% yes)
Fully weaned
(n = 98)
184 (12)
2.0 (0.2)
2.2(0.1)
0.7 (0.2)
151
<17>.
1.5 (0.3)
1.7(0.1)
1-2(0.2)
F
P-value
1.9
0.17
0.16
0.01
0.09
0.22
2.0
6.8
3.0
37
49
OR = 1.9C
(0.7-5.2)
31 (1.1)
27(0.7)
4.0 (0.4)
28.6
6.8
6.0
4.5
6.1
8.6
6.8
2.6 (0.6)
0
5 0 (0.3)
4.0 (0.2)
4.6(1.2)
7.6(1.7)
41
3.1
(0.2)
(0.1)
(0.8)
(1 1)
58
3.2
0.9
0.2
OR= 1.5C
(0.5-3.9)
0.01
0.08
0.01
0.08
0 34
0.66
0.45
* Means (SE) and percentages adjusted for age of child in months
b
0 = no school 1 = primary, 2 = secondary, 3 = advanced.
c
OR = odds ratio, from logistic regression, adjusting child's age; (95% confidence interval)
Count of food items consumed/day, average of 3 days
c
Mother's rating from 1 = poor to 5 = excellent
age, were compared using analysis of covariance. There were no
differences between feeding groups for time iU with diarrhoea
or malaria or by household headship; and although FW children
tended to be male, to have fathers with more education and to
live in households which spent more on food, these differences
were not significant. Households with FW children had significantly more children under 5 years of age, a history of fewer
child deaths, and younger, more educated mothers. Twentyeight mothers in our sample were currently pregnant and all of
these had weaned their children. Diets of FW children had
significantly greater diversity from non-breast milk foods; however, with the inclusion of breast milk, there was no difference
in diversity score between the two groups. The appetites of FW
children were rated by mothers as better than those of the PBF
children.
These observations were further substantiated by multiple
logistic regression of feeding status on the above socioeconomic
and demographic variables. A final model included older age of
the child, greater number of children in the family under 5 years,
maternal pregnancy (all at P < 0.05) and younger maternal age
(P < 0.1) as having independent significant associations with
likelihood of being FW (data not shown).
Among those who were FW, 80% had stopped breastfeeding
before they were 2 years old. The reported reasons for termination were subsequent pregnancy (54%), child refused or could
not continue to breastfeed (13%), unspecified parental preference (11%), to get children to eat other foods (9%), insufficient
milk (8%) and maternal illness (5%).
Non-breast milk food intake
Maternal recall of infant feeding practices suggested that
exclusive breastfeeding was rarely practised beyond 3 months.
Porridge was introduced during the first month of life for 23%
of children, and by 4 months for 86%. Fruit was introduced
between 3 and 6 months for the majority of children. About
15% of mothers reported giving fruit to their children before
3 months of age and, at the other extreme, about 15% delayed
offering fruit until 12 months of age or later.
Gruel, prepared from one or a combination of maize,
sorghum, millet and cassava flour, was a constituent of all
children's diets (Table 2). Green leafy vegetables were used by
69% of the sample. Fewer children consumed fruits. Among the
starchy staples, kidney beans contributed to the diets of 47% of
the sample, and sweet potatoes to 40%. Fish, consumed by 39%
of the children, was more commonly used than meat. Sugar
was a regular component of most children's diets. Milk consumption, both fresh and sour, was reported for approximately
half the children.
There were few obvious differences in food intake pattern
between the two groups, and there were no foods confined
exclusively to one group, despite the age differences associated
with breastfeeding status. PBF children were slightly less likely
to consume most food items; differences were greatest for
bread, beef, dark greens, milk, cooking fat, fish, eggs and tea.
Among the FW group, 82% of energy was derived from
carbohydrate sources, with approximately 10% each from fat
and protein.
WEANING FOODS AND ANTHROPOMETRIC STATUS IN KENYAN TODDLERS
Associations with anthropometric status
Table 2 Foods consumed by Kenyan children, aged 12-36 months 3
Partially breastfed
(n = 56)
No. (%)
Weaned
(n = 98)
48 (86)
45 (80)
43 (77)
14(25)
87 (89)
81 (83)
81 (83)
23(23)
33 (59)
14(25)
6(11)
5(9)
73 (74)
25 (45)
22 (39)
47 (48)
40 (41)
22 (22)
Porridge
Cassava Dour
Maize flour/meal
Sorghum flour
Millet flour
Results from the regression analyses of anthropometric measures on food intake and illness variables are presented in Table 3.
The strongest result was that dietary diversity was consistently
and positively associated with each of the five anthropometric
outcome measures. Breastfeeding status was significant only as
a component of interactions in the equations for TSF and MUAC.
TSF was greatest among FW children with low diarrhoeal rates,
in comparison with all others. MUAC was significantly greater
with age among FW, but not PBF children; and was lowest
among PBF children to whom cereal had been introduced before 4 months of age. Late introduction of cereal (>6 months
versus earlier) was significantly and positively related to HA.
Percentage of time ill with diarrhoea was negatively associated
with WA, WH and TSF; malarial morbidity was negatively associated with MUAC.
NO. (%)
Fruits and vegetables
Dark greens
Papaya
Bananas
Guava fruit
31 (32)
15(15)
12(12)
Starchy staples
Beans
Sweet potatoes
Bread
3 (5)
Nutrient intake by feeding pattern
Mean percentages of WHO/FAO recommendations for nonbreast milk food intake of energy, protein, vitamin A, vitamin C,
thiamin, riboflavin, niacin, iron and calcium are presented in
Table 4 for the PBF and FW groups by level of dietary diversity
( « versus >5). Of these nutrients, mean intakes were particularly low for vitamin A, riboflavin and niacin. Vitamin C
intakes may be artificially high, due to the inclusion in the database of raw, rather than cooked values for many foods. Much of
the vitamin C would be lost in the long cooking methods used
by these families. Overall, PBF children consumed lower proportions of their requirements from these foods; differences
were significant for all nutrients reported except vitamins A and
C. It is not possible to comment on the overall adequacy of
nutrient intakes by PBF children without estimates of breast
milk contribution to the total. However, it is clear that dietary
diversity plays a major role in the adequacy of intake among
Meats
Fresh fish
Dried fish
Beef
Chicken
18(32)
15(27)
43 (44)
35 (36)
22 (22)
15(15)
15 (15)
3(5)
6(11)
3(5)
Egg
Miscellaneous
Sugar
Milk (fresh and sour)
Cooking fat
35 (63)
24 (43)
8(14)
Tea
64 (65)
55(56)
25 (26)
11 (11)
1 (2)
487
' Number of children fed food item at least once during the three 24-hour
recalls.
Foods were reported if at least 10% of one group used them.
Table 3 Relationship of feeding pattern and illness to anthropometric measures0
WA Z-score
Age (months)
Breastfedb
Diet diversity0
Diarrhoea
Malariad
Early cereale
Late cerealc
Interactions; BF*
b
c
d
e
t
0.41
-2.4
0.001
0.02
_
-
P
0.70
0.10
0.008
_
-
-
-
-
-
-
-
-
-
1.0
2.3
0.02
-
-
34
P
0.68
0.39
b
t
0.007
0.45
0.17
-0.38
1.6
2.7
Early cereal
-
-
-
-
-
Age
-
-
-
-
-
-
-
-
-
-
Diarrhoea
Adjusted R2
1
b
0.007
0.21
0 19
-2.8
WH Z-score
b
t
0.01
0 91
-0.12 -0.60
0.12
2.6
-3.2 -3.2
-
0.86
HA Z-score
0.15
0.09
0.18
Mid-upper arm
circumference
Triceps skinfold
P
004
b
0.09
0.30
0.05
0.001
2.8
3.1
0.17
2.8
_
-
-
-
-
-2.2
0.15
-2.4
0.63
0.02
0.53
-
-
_
-
-
-
-
-
-
- -0.78
- -0.08
-1.9
-2.2
0.05
0.03
-
-
t
b
0.07
2.1
-0.64 -1.05
0.24
2.0
-12.0 -3 3
P
0.36
0.55
0.01
0.001
-
-
-
-
-
10.4
0.19
2.1
-
-
0.04
-
t
4.4
P
0.000
0.002
0.006
0.25
Adjusted for child's age, household Income and mother's educalion. Other variables tested that were not significant: non-breast milk energy Intake, early and
late introduction of fruit, and appetite rating by mothers.
Breastfeeding at time of measurement no = 0, yes ° 1.
Count of number of different foods consumed.
Proportion of days ill with malaria during observauon period.
Early cereal = 1 If <4 months. Late cereal = 1 If >6 months.
488
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 4 Intake of seleaed nutrients from foods other than breast milk as % of 1988 FAO/WHO Standards (mean ± SD)
Partially breastfed
<5
Energy
Protein
Vitamin A
Vitamin C
Thiamin
Riboflavin
Niarin
Iron
Calcium
n
49 ± 18
59 ± 22
27 ± 31
160 ± 164
68 ± 27
36 ± 21
35 ± 15
87 ±87
39 ± 31
P-value*
Fully weaned
Diet diversity
>5
71 ± 27
94 ± 4 4
57 ± 59
<5
82 ± 4 3
125 ± 6
62 ± 66
316 ±243
111 ± 5 4
53 ± 32
68 ± 33
165 ± 102
84 ± 53
230 ± 194
95 ± 34
56 ±27
52 ±21
130 ± 71
83 ±68
32
Diet diversity
>5
95 ± 30
135 ±63
83 ±65
489 ± 364
140±44
80 ± 38
81 ± 33
187 ±92
121 ± 52
24
23
Diet diversity
(<5 vs. >5)
0.002
0.04
0.03
0.05
0.000
0.000
0.006
0.06
0.000
Breastfed vs.
fully weaned
0.000
0.001
0.4
1.0
0.006
0.006
0.001
0.04
0 002
75
' Adjusted for age No interactions were significant, /"-values reflect independent main effects.
Table 5 Anthropometric measures by feeding pattern (mean ± SD)
Partially breastfed
Diet diversity
•S5
WAZ Score
HAZ Score
WHZ Score
Triceps skinfold (mm)
Mid upper arm circumference (cm)
n
-1.6± 1.0
-1.8± 1.0
-0.7 ± 0.9
9.7 ± 1.6
13.9 ±0.8
32
>5
-1.1
-1.6
-0.2
10.7
147
± 1.1
± 1.1
± 1.0
±2.4
± 1.3
24
Fully weaned
Diet diversity
<5
-1.8 ±1.3
-2.6 ± 1.4
-0.3 ± 0.9
10.9 ±2.2
142 ± 1.5
23
>5
-1.0± 1.0
-1.7 ± 1.1
-0.05 ±0.8
1 1.4 ±2.5
14.9 ± 1.2
P-valuea
Diet diversity
(>S5 vs. >5)
0.005
0.02
0.03
0.20
0.01
Breastfed vs.
fully weaned
0.35
0.12
071
0.84
0.07
75
* Adjusted for child's age, household income and mother's education.
No interactions were significant /"-values reflect independent main effects.
both feeding groups; differences by diversity were significant for
all nutrients. Among the PBF, those with low dietary diversity
had intakes from approximately 20% to more than 50% below
those of PBF children with high diversity. Interactions between
diversity level and feeding status were not significant for any
nutrient, suggesting independent effects.
Discussion
Recognizing the importance of complementary feeding for
growth in the older child, WHO and UN1CEF2 recommend that,
in addition to breastfeeding, nutritious foods should be included
in children's diets from 6 months of age onwards. The results
reported here support the importance of this guideline, but further strongly suggest that more emphasis be placed on increasAnthropometry by feeding pattern
ing dietary diversity among all weaning age children.
Mean anthropometric measures are presented by feeding status
Among these Kenyan children, dietary diversity was the strongand diversity level in Table 5. Mean levels for WA and HA est and most consistent predictor of anthropometric status. Few
suggest that malnutrition, particularly stunting, is a problem in
studies have examined this variable among weaning age chilthis community, with most children falling between -1 and -2 dren, but a similar relationship between dietary diversity and
Z-score. Mean WH Z-scores range from -0.7 to -0.05. There anthropometric status was reported in China. 16 Dietary quality,
were no significant differences by breastfeeding status nor any which is greatly enhanced by the inclusion of a variety of foods,
significant interactions between feeding status and diversity is of critical importance to weaning age children; and after 12
level, although the interaction for HA approaches significance months of age, it may be of greater importance than whether or
(P = 0.1); the FW with low diversity had the most stunting. not breastfeeding is continued.
Again, diet diversity level was strongly and consistently associated
Prolonged breastfeeding in developing countries has been
with all measures of nutritional status, regardless of breast- promoted partly due to the protective role it plays against the
feeding status.
consequences of diarrhoea. 3 ' 5 ' 26 In our study, diarrhoeal disease
WEANING FOODS AND ANTHROPOMETRIC STATUS IN KENYAN TODDLERS
was, along with low dietary diversity, an important contributor
to lower measures of the anthropometric indicators of current
nutritional status (WH and TSF). However, this interacted with
feeding method only for the latter; TSF was greater among children who were FW with low diarrhoeal morbidity than among
all other combinations.
Introduction of complementary foods is currently recommended for children between 4 and 6 months of age. 1 ' 2 In these
Kenyan children, there is a strong tendency toward early introduction (during the first 3 months) of cereal gruels. UNICEF 2 7
has reported that fewer than 26% of infants aged 0-3 months
in sub-Saharan Africa are exclusively breastfed. Early introduction of gruels is common throughout Africa and other developing countries. In Nigeria, most children are fed pap, from maize,
sorghum or millet flour; 2 8 2 9 in urban Ethiopia, atmit, a barley
flour cereal; 30 and in Cambodia, a rice-based porridge, bobor?^
Grain-based foods are also important early weaning foods in
Brazil; 3 2 and in Costa Rica.33 The positive relationship between
late cereal introduction and HA, and the negative interaction
between early cereal introduction and prolonged breastfeeding
seen in our results, suggest that the common practice of early
introduction of cereal may have long-term negative effects.
In this population, breastfeeding under 12 months of age is
almost universal, and this practice should dearly be encouraged. However, early introduction of complementary foods is
common, with some evidence of detrimental effects. Efforts to
discourage this early introduction of gruels may prove to be
beneficial. For children over 12 months of age, breastfeeding
status appears to be less important than the composition of the
diet, suggesting that efforts to improve dietary diversity are
likely to be very important to the improvement of nutritional
status among these children.
References
489
Nube M, Asians-Okyere WK. Large differences in nutritional status
between fully weaned and partially breastfed children beyond the age
of 12 months. Eur J Clm Nutr 1996:50:171-77.
10
Thoren A, Stlntzing G. Value of prolonged breastfeeding. Lancet
1988:11:788.
11
Ngandu NH, Watts TEE. Child growth and duration of breast feeding
in urban Zambia. J Epidemiol Community Health 1990:44:281-85.
12
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.
l3
Dettwyler KA. Breastfeeding and weaning in Mali: cultural context
and hard data. Soc Sa Med 1987:24:633-44.
14
Marquis GS, Habicht J-P, Lanata CF, Black RE, Rasmussen KM.
Association of breastfeeding and stunting in Peruvian toddlers: An
example of reverse causality. Int J Epidemiol 1997:26:349—56.
15
Jumaan AO, Serdula MK, Williamson DF, Dibley MJ, Binkin NJ,
Boring JJ. Feeding practices and growth in Yemeni children. J Trop
Pediatr 1989:35:82-86
16
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.
17
Cousens S, Nacro B, Curtis V et al. Prolonged breastfeeding: no association with increased risk of clinical malnutrition in young children
in Burkina Faso. Bull World Health Organ 1993:71:713-22.
l8
Onyango A, Tucker KL, Eisemon TO. Household headship and child
nutrition: a case study in western Kenya. Soc Sa Med 1994:39:1633-39.
19
CAN'DAT. A Research-Oriented Nutrient Calculation System
Ontario: Godin a n d London Inc., 1989.
20
West CE, Pepping F, Temaliwa CR (eds). The Composition of Foods
Commonly Eaten in East Africa. Netherlands: Wagemngen Agricultural
University, 1988.
21
London.
J a n s e n , AAJ, Horelli HT, Quinn VJ. Food and Nutrition in Kenya
A Historical Review. Faculty of Medicine, University of Nairobi, 1987.
22
Cameron M, Hofvander Y. Manual on Feeding Infants and Young Children. Oxford: Oxford University Press, 1983.
23
United Nations. How to Weigh and Measure Children. Preliminary Version.
New York: United Nations, 1986.
24
J o r d a n MD. Anthropometric
Software Package Tutorial Guide and
Handbook. Atlanta, Georgia: Centers for Disease Control, 1987.
1
FAO/WHO. World Declaration and Plan of Action for Nutrition. Rome,
1992.
25
2
World Health Organization (WHO)/UN1CEF. The Innoccnti Declaration.
Produced and adopted by WHO/UNICEF policy makers' meeting on
breastfeeding in the 1990s- A Global Initiative. Florence, Italy, July/
August 1990.
SPSS for Windows, version 7.0. Englewood Cliffs, New Jersey: PrenticeHall, 1995.
26
Briend A, Wojtyniak B. Rowland MG. Breastfeeding, nutritional state
and child survival in rural Bangladesh. Br Med J 1988:296:879-82.
Habicht J-P, DaVanzo J, Butz WP. Does breastfeeding really save lives
or are apparent benefits due to biases? Am J Epidemiol 1986:123:
279-90.
27
UNICEF. The State of the World's Children. 1994.
28
Uwaegbute AC. Weaning practices, weaning foods of t h e Hausas,
Yorubasand Ibos of Nigeria. Ecol Food Nutr 1991:26:139-53.
29
Oni GA, Brown KH, Bentley ME, Dickin KL, Kayode B, Alade I. Feeding practices and prevalence of hand-feeding of infants a n d young
children in Kwara State, Nigeria. Ecol Food Nutr 1991:25:209-19.
3
4
Briend A, Bar! A Breastfeeding improves survival but not nutritional
status of 12-35-month-old children in rural Bangladesh. Am J Clin
Nutr 1989:43:603-08.
5
Vlctora CG, Fuchs SC, Kirkwood BR, Lombardi C, Barros FC. Breastfeeding, nutritional status, and other prognostic factors for dehydration among young children with diarrhea in Brazil. Bull World Health
Organ 1992:70:467-75.
30
A l m e d o n AM. Infant feeding in urban low-income households in
Ethiopia: I. The weaning process. Ecol Food Nutr 1991:26:97-109.
6
Victora CG, Vaughan JP, Manines JC, Barcelos LB. Is prolonged
breastfeeding associated with malnutrition? Am J Clin Nutr 1984;
39:307-14.
31
Stone T, Senemaud B, Thai EV. Return to normalcy: Nutrition a n d
feeding practices in P h n o m Penh, Cambodia. Ecol Food Nutr 1989,
23:249-59.
7
Brakohiapa LA, Yartey J, Bille A et al. Does prolonged breastfeeding
adversely affect a child's nutritional status? Lancet 1988:11:416-18.
32
Miotto-Wright M, Dutra d e Oliveira J. Infant feeding in a low-income
Brazilian community. Ecol Food Nutr 1989:23:1-12.
8
Michaelsen KF. Value of prolonged breastfeeding. Lancet 1988,11:
788-S9.
33
M u n o z LM, Ulate E. Breastfeeding patterns of urban low to middle
income w o m e n in Costa Rica. Ecol Food Nutr 1991:26:59-67.