WOMEN'S WORK AND CHILD NUTRITION IN HAITI
by
PATRICIA ANN HAGGERTY
B. A., Boston College
(1977)
SUBMITTED TO THE DEPARTMENT OF
NUTRITION AND FOOD SCIENCE
IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS OF THE DEGREE OF
MASTER OF SCIENCE
IN NUTRITIONAL BIOCHEMISTRY AND NETABOLISM
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September 1981
Signature of Author _____ _ ____ ____
Certified
Department of icioJ and Food Science
August 21, 1981
Barbara A. Underwood, Ph.D.
Thesis Supervisor
by
Accepted by
Archi ve Anthony J. SinskeyMASSACHUSETTS NSTITU Clairman, Departmental Committee on StudentsOF rECHNOLOGY
SEP 24 1981
LIBRARIES
2WOMEN'S WORK AND CHILD NUTRITION IN HAITI
by
PATRICIA ANN HAGGERTY
Submitted to the Department of Nutrition and Food Science
in Partial Fulfillment of the Requirements of the Degree
of Master of Science in Nutritional Biochemistry: and Metabolism
ABSTRACT
A study was undertaken to investigate the associations between women's
work activities and the nutrition of children in the first two years of
life. Early growth and mortality of a cohort of 189 children were studied
according to a child welfare function defined by maternal occupation,
literacy, time spent in child care and household resources.
A sample of 530 low-income women living in Cite Simone, Haiti, were
randomly selected to participate in the partly retrospective, partly
cross-sectional mortality-nutrition survey. Eighty percent of the women
had given birth within 18 months prior to the survey period. Maternal
pregnancy histories were included in a questionnaire'which obtained
information on demographic, socioeconomic and nutritional characteristics
of the women involved. Data on child care and infant feeding practices
was collected by observing as well as questioning the mothers. Anthro-
pometric measurements were made on a subsample of the women and their
recently born children to assess nutrition status in the community.
Life table methodology was used to analyze mortality of the study
children and the introduction of bottle feeding by age of children in
months. Growth in weight and height was evaluated according to the
Waterlow classification of nutritional status. Ten percent of the child-
ren (0-23 months) were identified as wasted and 17 percent stunted. First
year mortality risk was found to be 163 per 1000 live births while that
to 23 months was 205 per 1000.
Children of women who earned the lowest wages or did not work for
money showed the poorest growth performance overall. Analysis of vari-
ance revealed that work status and location of mother's work significantly
affected length for age in the second year of life. Income appeared to
be the major factor responsible for better nutritional status after the
sixth month. In the first six months, though, time women spent with
their infants was of relatively greater importance in establishing
satisfactory growth.
Risk of death was markedly higher among children of the poorest
working women than among children of women who did not work for money.
In households with cement floors, mortality of the study children was far
below that of children living in dirt-floored houses. Children of
literate women were subject to a higher mortality risk than children of
illiterate women. Wide-spread introduction of bottle feeding and supple-
3mentary foods in the first three months of life is considered to be a
major cause of the high risk of death in this community.
Thesis Supervisor: Barbara A. Underwood, Ph.D.
Title: Associate Professor
Department of Nutrition and Food Sciences
4Acknowledgements
The author wishes to express the deepest gratitude to Dr. Gretchen
Berggren for her gracious support throughout the progress of this thesis.
She provided a personal and professional account of nutritional problems
in Cite Simone.
Equal thanks go to Dr. Douglas Ewbank for his instructive assistance
in the data handling and statistical methodology.
Many thanks also are conveyed to Dr. Barbara Underwood, my advisor.
Her wisdom and experience as a scientist have been a guide throughout
my research at M.I.T.
5Table of Contents
Page
2Abstract . . . . . . . . . . . . . .. . . . . .
Acknowledgements . . . . . . . . . . . . . ..
Table of Contents . . . . . . . . . . . . ..
List of Tables . . . . . . . . . . . . . . ..
List of Figures . . . . . . . . . . . . . . .
I. Introduction . . . . . . . . . . . . .
II. Conceptual Framework . . . . . . . . .
III. Literature Survey . . . . . . . . . ..
IV. Background . . . . . . . . . . . . . .
V. Methodology . . . . . . . . . . . . ..
Survey Questionnaire . . . . . . ..
Anthropometric Data . . . . . . . .
Data Analysis . . . . . . . . . . .
VI. Results . . . . . . . . . . . . . . ..
Descriptive Sample Characteristics.
Infant Feeding Practices . . . . ..
Anthropometric Analysis . . . . . .
Mortality . . . . . . . . . . . . .
VII. Discussion...... . . . . . . . . . o
VIII. Summary and Conclusions . . . . . . ..
IX. Recommendations for Future Research. .
Footnotes . . . . . . .. . . . . . . . . . .
References . . . . . . . . . . . . . . . . ..
Appendix A . . . . . . . . . . . . . . . . ..
Appendix B.
4
5
6
8
9
12
14
23
27
27
28
28
33
33
34
41
58
69
77
81
84
85
90
91
6List of Tables
Page
. . . . 35
Table No.
1 Bottle introduction by age (in months) and
time of introduction..... ..........
2 Life table of bottle introduction .. ....... ..... 37
3 Foods eaten by children by age and type of food.... ... 38
4 Mean length for age, weight for length and weight
for age of children by age and sex.. ........
5 Analysis of length for age, weight for length and
weight for age of children by work status of
mother and age of child........... ..........
6 Analysis of length for age, weight for length and
weight for age of children by location of mother's
work and age of child... ..........
7 Analysis of length for age, weight for length and
weight for age of children by time mother spends
with child... ....... ........
8 Analysis of variance in length for age of children
12-23 months by mother's work status and location
of work......... . ...........
9 Analysis of variance in weight for length of
children 12-23 months by mother's work status
and location of work. ........ . . .0. .0....
10 Analysis of variance in weight for age of children
12-23 months by mother's work status and location
of work....... ..........
11 Analysis of Waterlow classification of nutritional
status of children 0-23 months by work status of
mother............. ..........
12 Analysis of Waterlow classification of nutritional
status of children 0-23 months by number of days/
week mother works outside of home. .........
13 . Analysis of Waterlow classification of nutritional
status of children by age of child and location of
mother's work...............&.. ... a..........
42
44
45
47
50
51
52
54
55
56
7Table No. Page
14 Analysis of Waterlow classification of nutritional
status of children 6-23 months by number of days/
week mother works outside of home. . ..... . . ... .57
15 Life table and survival analysis (weighted),
cohort of children 0-23 months ... .......... 59
16 Life table and survival analysis, children of
marchands... .......... . ........... 60
17 Life table and survival analysis, children of
dependents and housewives. ................... ...... 61
18 Life table and survival analysis, children of
households with cement floor.......... .............. 63
19 Life table and survival analysis, children of
households with dirt floor ..... . ......... 64
20 Life table and survival analysis, children of
literate mothers.. ........ .. ......... 65
21 Life table and survival analysis, children of
illiterate mothers.. ........ .......... 66
22 Summary of mortality analysis.... ......... 6868
8List of Figures
Figure No. Page
1 Map of Haiti. ....... . ........... 24
2 Patterns of Milk and Cereal Feeding During
First Twenth Months of Life. ... . ....... 40
3 Distribution of % Length for Age Measurements
of Study Children. .. ............ ... 48
4 Distribution of % Weight for Length Measurements
of Study Children.so......... ............... .... 49
9I. INTRODUCTION
Throughout the developing world low income communities are nlagued
by the problems of excessive infant mortality and marked early growth
failure. Efforts to mitigate these problems have been largely based
on promotion of breast-feeding through delivery of health services and
nutrition education. Nutritionists emphasize that exclusive breast-
feeding is essential for at least the first two months of life to
support growth of the newborn equivalent to that of reference healthy
infants. To maintain satisfactory growth they further recommend that
additional foods be introduced into the infant's diet soon after the
first three months. The exact age at which complementary foods are re-
quired is intensely debated at present, but it is widely accepted that
three to five months is the appropriate range (Underwood and Hofvander,
1980; Waterlow et al., 1980). Breast-feeding alone for the first months
of life has also been found to reduce the risk of death among popula-
tions of infants. Conventionally, lower mortality rates among breast-
fed infants have been attributed to the increased protection against
infection afforded by breast milk (Wray, 1978; Rowland et al.
1977).
Under adverse ecological conditions, however, breast-feeding alone
unfortunately is not a guarantor of survival and satisfactory growth.
Frequently even the "best" combination of breast milk and complementary
food does not save the infant from malnutrition and eventual death.
Socioeconomic and demographic characteristics of low income households
are equally critical determinants of child welfare. In Haiti, child
welfare is affected in particular by women's work.
10
Although Haiti is essentially a breast-feeding population, infant
mortality is exceedingly high (nearly 200 per 1000 live births), and
children are grossly undernourished. Nearly all women supplement
their breast milk in the first month of life, despite promotion of
breast-beeding within the health services. The women of Haiti, among
the poorest women in the world, are subject to serious time, income and
resource constraints. 'These constraints preclude any chances of estab-
lishing exclusive breast-feeding. From this perspective, an alterna-
tive approach is to focus on women's constraints and on ways to al-
leviate them. Indeed, this approach may be the practical one for de-
veloping societies in which women are overburdened with the responsi-
bilities of household management and child care.
In the ensuing chapters, child welfare will be studied with re-
spect to maternal time use, occupation, education and household re-
sources. The parameters of child welfare examined are growth in the
first two years of life and infant mortality. On the basis of a con-
ceptual framework outlined in Chapter Two, the following hypotheses
will be tested:
(1) In Haiti, survival in the first year is dependent upon
maternal occupation, literacy and household resources.
(2) Growth in height and weight of Haitian infants, relative
to international standards, is affected by maternal
occupation, time spent in child care and location of
maternal work.
(3) The net effect of women's income earning activities
on child nutrition is unpredictable. Women's work may
11
improve a child's welfare through associated income
benefits. Conversely, working by the mother may be
detrimental to infant health due to the time constraint
imposed by working and/or to a reduction in breast-
feeding. Thus an important component -in child nutrition
is the pattern of infant feeding.
In Chapter Three a review of the literature on women's work.and
child nutrition is presented. Surprisingly little research has been
done on the relationships between women's work and time use charac-
teristics and infant mortality. A greater body of literature docu-
ments the multiple (and frequently obscure) associations of child
dietary intake and growth with women's labor. In Chapters Four through
Six the background, methodology and results of the child nutrition-
mortality study in Haiti are described. A discussion and interpreta-
tion of the study results is presented in Chapter Seven. The paper
will close with general suggestions for future research and some
particular suggestions for health promotion efforts in Haiti.
12
II. CONCEPTUAL FRAMEWORK
In the low income household, the production of infant welfare is a
function of time spent in infant care and resources available for the
child's health and nutritional needs. In lesser developed countries,
early growth and mortality are the relevant welfare parameters affected
by time and resource constraints. Income is the major resource required
to buy the medical care, food, utensils and housing that a child's health
demands. Time spent in child care includes that by the mother and
other household members; it encompasses hours spent in breast-feeding,
supervision of the child, preparation of bottles and weaning foods, tak-
ing the child to health clinics and so on. During the first year of
life time spent by the mother (as opposed to other family members) is
considered superior both for the physiological and mental development of
the child.
In addition to the child's welfare requirements the household also
demands a certain proportion of a woman's time and energy. Ultimately
tradeoffs need to be made between the child's needs and those of the
household. The conflict occurs chiefly when the household requires
absolute or additional income, and the woman is forced to work outside
the home. Whether the child's health is improved or not when this
occurs depends on the quality of the substitute child care and the
efficiency with which the extra income is used. Children of working
mothers may live longer and grow faster than children of non-working
mothers if the income is used to buy more food of better quality for the
infant. The income effect could also be reflected in increased medical
care, better floors, roofs, toilet facilities, piped-in water and the
13
like. On the other hand, a working woman presumably has less time to
spend with the child, breast feeds less and leaves the child in the
care of older siblings for longer periods during the day. These cir-
cumstances could jeopardize the child's health. Finally, the mother's
own health can be affected by working, either positively or negatively,
and this has nutritional implications for the child.
For the household in which the mother is the sole income earner,
the tradeoff between time and income in the production of infant health
is more conspicuous. Children in this group seem particularly vulner-
able since their mothers frequently work at the lowest wages and for
long hours, leaving little time or energy to prepare weaning foods
properly, breast feed long enough or to use food and resources effi-
ciently. Even for the case of the working mother without an income
earning partner, the net importance of time and income depends upon a
number of other individual and household factors. In the following
pages some of the most important of these, based in empirical studies
in the literature, ,are discussed.
14
III. LITERATURE SURVEY
Within the general framework of child care and infant welfare just
presented, a variety of factors interactively contribute to the impact
of women's time use on infant nutrition. As maternal variables, these
include: (1) participation of women in the wage labor force; (2) com-
patibility of a working woman's job with child care; (3) the presence
of older children in the household; (4) the woman's value of time in and
outside of the home; and (5) the mother's own "human" resources, in-
cluding her knowledge, education, and nutritional status. Each of
these has implications for child growth or survival.
3.1 Participation of women in the labor force
Participation of women in the market labor force is accompanied
by changes in household income, food purchases, nutrient consumption
and infant feeding practices. In both urban and rural communities,
working away from home has been associated with earlier introduction
of supplementary foods. Nerlove (1974), in a cross-cultural study of
societies in which women contribute substantially to subsistence
activities, found that women who begin supplementary feeding of their
infants before one month of age participate to a greater extent in sub-
sistence activities than do women who introduce additional foods after
one month. The weaning period is known to be associated with severe
diarrheal disease and growth failure (Rowland et al., 1977; Rowland,
1980; Gordon et al., 1967). Furthermore, the period from one to five
months is the main danger period for death from diarrhea and pneu-
monia (Cooper, 1980). For the infant under six months then, whose
risk of death is more than twice that of counterparts six months to
15
one year old (Cooper, 1980), edrly weaning increases the risk of death
due to infectious disease.
Conversely, income increases may translate into improved surviv-
ability of the infant. This effect occurs regardless of breast-feeding
patterns. In a Colombian study, infants of mothers from a lower socio-
economic group had higher rates of post-natal death (one month after
birth or subsequently) than infants of mothers with higher incomes
(Oberndorfer and Mejia, 1968). In the same study, there was no evident
difference in the time of breast feeding between the two groups.
Increments in income appear to have the greatest positive impact
on child nutrition when the mother earns the income (Kumar, 1978).
Levinson (1974) found that calorie intakes of rural Punjab children in
low income families significantly improved when the mother worked, due
to income increases. (The effect was absent among children of the
higher income group.) Boulier (1976) reports increased weekly childcare
time by Filipino women when a mother's own wage increases. Neither
of these studies, though, demonstrated consequences for child growth.
In the Philippines, women's participation in the labor force
was found to have a positive effect on weekly food expenditure of study
households (Popkin and Solon, 1976). The impact on calorie, protein,
iron and vitamin A intakes of children of working mothers was negative.
Moreover, this result was pronounced among children of low income work-
ing mothers when the effect of food expenditures was separated from
that of mothers being away from home, while nutrient intakes of children
of higher income working mothers improved.
A study conducted in Kerala, India, explored the consequences of
some of these effects due to maternal work participation on child
16
growth. Kumar (1978) found that among working women aggregate child
nutrition (weight for age) was lower than that among non wage-earning
women. Increases in income, however, were translated into improved
weight for age among children of wage earning women. These seemingly
contradictory results were explained when the study households were
disaggregated. Lower levels of child nutrition were more apparent
within households where women were the sole wage earners. Unable to
earn enough money to relieve the income constraint for which they
entered the labor force, these women imposed a time constraint on
the production of child nutrition as well.
Thus it appears that participation of women in the labor force
per se may be detrimental to nutritional intakes of children, possibly
through reduction in time spent with the child. Diarrheal disease
associated with early weaning increases the probability of death in the
first year of life. Nevertheless, higher levels of income do improve
the chances of an infant's survival. Maternal income can also im-
prove nutrient intakes through increases in the purchase of food.
The effects are pronounced among low income groups, where the marginal
values of time and money are greater. The time constraint imposed by
working is potentially more detrimental to child growth when the mother
is the sole supporter of the family.
3.2 Job compatibility
For the working woman with no other source of income, nutrition-
al status of the child could be improved if the job is compatible with
child care. Compatibility of work with child care and breast feeding
is based on the hours during which the job is available, the location
17
(and accessibility) in relation to home and the ability of the woman
to take the infant with her while she works. In a theoretical model
of infant nutrition and breast-feeding behavior proposed by Butz,
the effects on breast feeding of maternal employment depend on com-
patibility of the job. Jobs that are located close to the home have
been found to increase breast-feeding in Malaysia and the Philippines
(Butz and DaVanzo, 1978; Popkin and Solon, 1976). On the other hand,
compatible jobs may not necessarily increase breast-feeding but may
improve the efficiency of the mother in allocating child care responsi-
bilities to others (Popkin, 1978, 1976). (In the Philippines, the
net impact of a compatible job was to raise overall child care time.)
In either case a compatible job for the woman of a single parent
household would improve her child's situation.
3.3 The presence of older children
The presence of older siblings in the low income household is
also an important component in the production of infant nutrition.
Children have a considerable influence on the woman's allocation of
time. By substituting for the mother's home production and child care
time, older children, especially teen age girls, increase the partici-
pation of women in the labor force and the number of hours mothers spend
in income earning activities (Boulier, 1976; Popkin, 1978; Popkin et
al., 1979; Popkin, 1976). The consequence for breast feeding partici-
pation is likely to be negative, while that for growth is less pre-
dictable. Depending upon the strength of the income effect and the
quality of the substitute care, growth could be better or worse.
18
3.4 Woman's value of time
Presuming that the mother is an important decision maker in these
child care-income tradeoffs, the value of her time is a key factor in
the choices that are made. The value of time spent in child care
activities, particularly in the time intensive activity of breast-
feeding, is relative to the costs of substitute child care. These costs
include the actual market prices and "opportunity costs" of human and
non-human substitutes for child care (Popkin et al., 1979). Human
costs are affected by wage rates and employment opportunities, and de-
pend on age, sex, education and health of the individual. The degree
to which the mother's value of time increases is expected to have a
stronger impact on breast-feeding--higher paid women have more to
lose in terms of job tenure and career advancement (Popkin, 1978).
However, once a certain wage is reached that satisfies a level of needed
income, a woman may be able to work fewer hours and breast-feed more.
Changes in the woman's value of time may not affect breast-feeding
if time spent in other home production activities is adjusted. Higher
income could increase the number of time-saving household assets, such
as stoves, piped water, electricity and indoor toilets. This kind of
capital, in addition to saving time, could significantly reduce the
child's risk of infection by providing a more sanitary environment.
Child growth is implicitly affected also (Popkin, 1980), but just where
the linkages occur is surprisingly under-researched.
3.5 The mother's "human" resources
The influence on child care time and infant nutrition of the
mother's human capital has been studied to a greater extent. These
19
characteristics include the education level and nutrition status of
the mother. Education is expected to increase a woman's efficiency in
her child care, home production and income earning activities. Her
earnings potential is also expected to rise. Nevertheless, in many
urban communities of the developing world education has had a sizeable
negative impact on breast-feeding length and participation. Earlier
supplementation of infants' diets and reduced length of breast-feeding
are consistently documented in the literature (Popkin, 1978; Kumar,
1978; Mudambi, 1981; Underwood et al., 1981). This result is fre-
quently attributed to .increased exposure to modern conveniences and
Western values that accompanies education.
The consequence for infant nutrition, like the work force par-
ticipation variable, is not the same under all conditions. Education
can have a positive affect on nutrient intake of young children, as
was observed in the Philippines (Popkin, 1980). It has also been
associated with significantly lower levels of pre-schooler malnutri-
tion in Sudan (Nutrition Reviews, 1980). The benefits of education
in terms of income and increased efficiency of time and resource
allocation could theoretically outweigh the disadvantages of earlier
supplementation of the infant's diet.
A contemporary theory of maternal education as a factor in child
mortality suggests that education of women greatly changes the tra-
ditional balance of familial relationships, with profound effects on
child care (Caldwell, 1981). Studies conducted in West Africa,
Bombay and many Latin American countries, confirm that mortality of
children among mothers with no education was almost twice as high
20
for those with elementary education and more than four times as high
as for those with secondary education. Education of a woman is likely
to challenge traditional modes of decision making and child care that
are disadvantageous to the child. Subsequent changes in nutrition
occur through increases in food consumption, improvement in the quality
of food and greater equality of food distribution within the household.
Of important consequence to breast-feeding duration, quantity and
quality of breast milk production and birth weight of infants is
maternal nutrition. The effects of the mothers dietary intake are
distinct from those of her nutrition status. The influence of food
intake on birthweight is small, for example, whereas the influence of
maternal size on birthweight is considerable (Thomson et al., 1979).
The incidence of low birthweight babies appears to be greater when
maternal height and pre-pregnant weight are low. Implicit in this re-
lationship is that long term nutritional factors associated with the
mother's height and weight may have an effect on the size of the
babies she is likely to produce. Birth weight has been directly re-
lated to the child's neurological and mental development, to size
attained in later life, and (inversely) to peri-natal mortality (Thompson
et al., 1979). A positive correlation between birthweight and age at
birthweight doubling time has also been observed (Neumann and Alpaugh,
1976).
Maternal dietary intake affects Iactationnperformance. In a re-
view of studies addressing this issue Wray (1978) concludes that the
quality of breast milk is affected only when the diet of the mother
is grossly inadequate, while the quantity of milk and the duration of
21
lactation disease with less severe dietary inadequacy. The response
to maternal dietary intake has been observed in many parts of Africa,
where decreases in energy intakes during seasons when energy expencii-
tures are highest have been correlated to low birthweight and diminished
breast milk output (Whitehead et al., 1978; Paul et al., 1979; White-
head, 1979(a)).
Inadequate breast milk output by women in developing countries
has been recognized as a major cause of growth faltering very early
in life. Infants fed an exclusive breast milk diet have been found to
falter in growth as early as the second month (Froozani et al., 1980).
Studies based on comparisons of breast milk intakes with independent
estimates of nutrient requirements for growth demonstrate that breast
milk alone becomes insufficient to support satisfactory growth after
three months of age (Waterlow and Thompson, 1979; Whitehead et al., 1980).
Although breast milk of a healthy woman is nutritionally suffi-
cient for growth of the newborn, the age at which additional foods
should be introduced into the infant's diet is not known with certainty.
The appropriate age complementary foods should be started is difficult
to establish because of the great variations in individual requireuents --
infants ingesting similar milk volumes exhibit unpredictable differences
in growth. 1 Clearly, factors other than the adequacy of breast milk are
operative under varying ecological conditions. A more important issue
here, however, is that weight for age of the infant is correlated with
breast milk intake from birth to twelve months (Whitehead et al., 1978).
Thus, smaller infants, by making less nutritional demands on the mother or
by reducing lactation capacity, never receive sufficient breast milk
22
supplies to allow for catch up growth.
The influence of maternal nutrition on infant survival is implicit
in its effects on lactation performance. A marked negative correlation
between infant mortality rates and breast feeding duration has long been
apparent (Cunningham, 1981; Wray, 1978; Woodbury, 1922). Breast milk
contains a number of important immunoglobulins which protect the infant
against infection and death due to diarrheal and respiratory disease (Wray,
1978; Cooper, 1980; Hill, 1979). Furthermore, survivability of infants
breast fed has been better than that of bottle fed infants for decades.
Wray suggests that the nutrient density of breast milk compared to bottle
feeds is the factor responsible for this phenomenon. Only recently have
bottle fed babies begun to catch up in survival possibly due to improved
quality of bottle mixes. Whether nutrition of the mother influences sur-
vival of infants in ways other than its effects on breast feeding is a
neglected issue in the literature.
In sum, these findings make it clear that a network of factors is
responsible for the interactions between women's work, human and household
resources, and child nutrition. None of the studies discussed here were
able to investigate all of the factors operating within a particular
cultural milieu. Nevertheless, the evidence allows researchers in this
area to focus on topics poorly studied and on methods which assess nutri-
tional and socioeconomic information.
In the following chapters an analysis of women's work and child
nutrition in Haiti is presented. Only some of the outstanding variables
known to affect child health will be tested directly. These include women's
participation in the labor market, job compatibility, household resources,
time spent in child care, bottle feeding practices and maternal literacy.
23
IV. BACKGROUND
Haiti is a small, mostly mountainous country in the Caribbean Sea,
situated among Puerto Rico, Cuba and Jamaica (see Figure 1). A 1978 popu-
lation of 5.3 million persons inhabits 11,000 square miles, making Haiti
one of the most densely populated countries in the world. About one-fourth
of the population lives in urban areas, the largest of which is the capital,
Port-au-Prince. French is the official language but is spoken by only
about ten percent of the people. For the remainder of the predominantly
neo-African people, Creole is the language as well as the way of life
(Berggren and Berggren, 1973; Weil et al., 1973; Rubin and Schaedel, 1975;
Haiti Nutrition Status Survey, 1979).
Haiti is also the poorest nation in the Western hemisphere. Annual
per capita income is less than $100 in the rural areas and just slightly
higher in the cities (Berggren et al., 1981; Boulos and Boulos, 1980).
Except for a small, elite group of wealthy people in Port-au-Prince,
poverty exists throughout the Republic (Weil et al., 1973). Haiti has
often been described as a fifth world country, so poor that it doesn't
even belong to the third world (Bordes and Couture, 1978). The infant
mortality rate is approximately 150 per thousand live births, life
expectancy about 48 years, and the birth rate 37 per thousand population
(Berggren et al., 1981). Chronic malnutrition is a severe problem, aggra-
vated by poor sanitation and extreme overcrowding.
Only a small percentage of the people have stable, salaried employ-
ment. Most support themselves from day to day farming, never counting on
a fixed sum of money. Haitians live by the system "Bon Die bon" (the
Lord is good). Annual revenues are far from sufficient to satisfy even
Q"A TOAWTU TUQA J)SI.A;1ND iQA ~
ort-do-Pa aint Louls-du-Nor4
A -. tA '~"~'~''Cap-Htln
M1le Snt-Ncoa Ca-ne a Rouge
A FOUX Gros-Morne bFort-LibertdCAP Fort Libertd
STrou
4 e ~ $A ~Grande-Rivibre-du-Nord du Nord
Gonaives ;N
Gofo d Is \*Saint-Michel-de-I Atalaye
GonAve DessaInI
i. san c Petite-Rivibre-de-l'Artibonite D 0 M N I C A N
saint-Marhe0
1A DUsarmes -
OANDE sarme G R E P U B L I C
CAYEMITE lte srebalais Lascahobs
- J0r6 i SAMannevile,
DAME-MARIE -tPc
Dame-Marie S Pr-uPic
SAtfrd "H-lanaWi N AVA M tAAS SIF DU SUD Anse-A V au LeogAne- Petlonville A
MiragoAne Pett- Grand-
T A qunGoAve MAPort-A-Piment * Coteaux -W -Sa tell
L es Cayes acme
18- _LE A VACH E- -
SC AR I BBE AN SE A10 20 40
47,IScale of Mles
5 4'72
)
25
the primary needs of the people. Scarce resources are used poorly be-
cause, it is said, Haitians lack discipline and a knowledge of funda-
mental home economics (Boulos and Boulos, 1980).
The study was carried out in Cite Simone, an impoverished urban
community on the outskirts of the capital. Citd Simone was originally
constructed in 1958 by the Haitian government to accommodate inhabitants
of a shantytown, La Saline. The people were moved in without an over-
all scheme, without sanitation facilities and without electricity. Houses
were packed one on top of the other along the top of the marshy terrain.
Gradually -Cite Simone became a retreat for all the spillovers and unfor-
tunates from Port-au-Prince. Between 1958 and 1976 the population had
multiplied by fifty percent (Boulos and Boulos, 1980).
Living conditions in the city typify the deprived environment in
which malnutrition and infection are synergistically nurtured. There are
no indoor toilets and piped in water is rarely found in the home. Garbage
heaps near the home or neighbors' backyards are the usual latrine reposi-
tories. (Haitians say "Ils vont dans la nature.") Children sleep on low
beds or banana mats on dirt floors, with dogs and chickens in the same
quarters. The moist floors become mud when the storm sewers overflow in
the rain. Woman are strongly influenced by their peers with whom they
constantly interact in an overcrowded environment.
Infant feeding practices conform to the norms set by peer groups of
women. These practices are partly forced by a scarcity of resources and a
belief system based on experience. Fuel is the major constraint for most
women. Charcoal is used as the primary fuel but as the Haitian hills are
denuded its cost has increased. The women generally build one fire per
day on top of which the main meal is prepared, usually rice and beans.
26
An effort is made to feed the child as much as possible at this meal as
children are likely to receive no other food during the day. Women are
afraid to leave food out overnight, knowing that spoiled food gives the
child diarrhea. Meat and eggs are believed to be bad for the baby, and
are withheld for fear that they cause diarrhea. Bottle feeding in public
is preferred to breast feeding, despite the expense, and despite the
scarcity of fuel for its preparation. Most city women have abandoned total
breast feeding in the first month. Only the poorest of poor breast feed
exclusively. Children raised among these conditions are subject to re-
peated episodes of diarrheal attack and gastrointestinal infections. Sick-
ness and death due to respiratory or intestinal disorders are not uncommon.
In 1968 the Department of Community Health began delivery of health
and nutrition services to Cite Simone. Health centers were built in Boston
and in Brooklyn, one in each quarter, to provide medical, social and
nutrition education services to the inhabitants of the surrounding zones.
Children are brought to health rallies four times per year to be weighed,
measured and counselled. The study was undertaken to determine what
changes, if any, have taken place in nutrition and child mortality since
the inception of these services. The results in this paper refer only to
the children born since January 1979.
27
V. METHODOLOGY
Sample Selection
The study was conducted on 530 women representing a population of
approximately 20,000 from Boston and Brooklyn, two neighboring quarters
of Cite Simone. Eighty percent of the sample comprised randomly se-
lected women who had given birth since January 1st, 1979, 18-22 months
preceding the study period. Twenty percent had not given a recent
birth. The women were chosen from family register sheets that were up-
dated by collaborators within the community. All women were categorized
by whether or not they had delivered a baby, and for those who had, by
the child's birth month and year. Using a random numbers table, women
were selected for each category and results were weighted accordingly.
Second and third replacements (representing a less mobile sector of
the population) were selected in the same way.
Survey Questionnaire
Information on mortality and nutrition of women and children was
obtained between August and September, 1980, using a pre-coded question-
naire. Interviews were administered by part-time resident home visitors,
2
called "volunteer community collaborators."* Doctors supervised .
coding of responses (done by highschool level coders) and were especial-
ly careful that responses from the pregnancy histories were coded accur-
ately. Data was collected on maternal education, occupation, marital
status, household characteristics (including family composition, toilet
facilities, water source and usage, radio ownership and type of floor),
use of health services and pregnancy history. For children who died,
28
a special section obtained information on type of sickness, medical
care, age of introduction of bottles and supplementary foods, and usual
supervision of the child before death. For surviving children born
since January 1979, mothers were asked about the type of feeding
(breast or bottle) a child received the day before the interview; the
age in days a child first received the breast and/or bottle; the age in
months a child was weaned (if applicable); the type and quantity of other
foods in the child's diet the previous day; and the number of hours a
mother spent with the child during the day. Observers noted the con-
tents of bottle feeds. Maternity histories, based on the mother's
recall, recorded all births and deaths of children up to the sixth most
recent child; causes of death; whether or not a child was breast-fed;
location of births and deaths; and the type of medical care, if any,
that a dying child had received.
Anthropometric Data
Within a month of the interview, height and weight measurements
were taken from a subsample of the women and last born children. In
the company of a physician standardized scales with weights were used
by an auxiliary staff previously trained in weighing and measuring.
Babies' weights were taken using a pan-type scale; their reclining
lengths were measured along a board.
Data Analysis
(1) Feeding Practices
Infant feeding practices during the first year-and-a-half of life
(other than breast feeding) were analyzed by tabulating the average
29
number of servings per day of nine categories of infant foods by age
of children. In this manner the importance of various categories of
bottle feeds and weaning mixes at specific stages of the child's matura-
tion were observed. The pattern of introduction of bottle feeding was
obtained by constructing a life table similar to that used for mortality
analysis (see below). The advantage of using the life table method is
that it presents the monthly probability of being introduced to a bottle
as well as the cumulative probability of being bottle fed at any given
age, for the cohort of infants in the study.
(2) Nutritional Status
Nutritional status of children 0-23 months old was assessed by
physical growth and category of malnutrition. Growth was evaluated
based on reference population standards of the National Center for
Health Statistics Center for Disease Control (NCHS/CDC). These standards
are used to compare measurements of height for age, weight for age and
weight for height, and are recommended currently by the World Health
Organization (WHO, 1979). Although a single measurement of weight for
age masks the. difference between a tall, thin child and a short, well-
proportioned child, this shortcoming is mitigated when single weight
measurements are taken on a large number of children (Jelliffe, 1966).
In this study, weight for age is useful as an indicator of the approxi-
mate nutritional profile of the community. Weight for height is used
as an index of acute undernutrition while height for age is a standard
indicator of chronic malnutrition. Growth was analyzed by work status
of the mother, location of mother's work, and number of hours per day
women spent with their children.
I I
I I
30
The Waterlow cross classification of nutritional status was used
to identify types of malnutrition prevalent among the children. Under
this classification, a weight for height measurement less than 80% of
the NCHS/CDC reference median is defined as wasting; height for age
less than 90% of the reference median is defined as stunting; and weight
for height less than 80% together with height for age less than 90% of
the reference population is classified as concurrent wasting and stunt-
ing. A normal child is neither wasted nor stunted. The percentage of
children wasted, stunted, or wasted and stunted was analyzed by
mother's work, location of work and number of days worked outside the
home.
(3) Mortality
Cohort life tables were constructed to investigate mortality and
survival in the first 23 months of life among the study children. The
life table method was chosen in order to utilize all births that were
observed between January 1st, 1979, and September 1980. Cohort mortality
rates, unlike mortality rates based on population censuses, only measure
the proportion of study children dying at a given age. They are useful
as an indicator of death rates which may prevail in the population but
are not, in fact, the population's true mortality.
The World Health Organization defines age-specific mortality as
the number of deaths occurring in one year, all of whom are aged x at
death, divided by the mid-year population all of age x precisely (WHO,
1977). Age-specific mortality is accurately estimated for a population
only when a complete record of births and deaths for the years of
31
interest are available. In this study only births and deaths of the
cohort of children were utilized. Thus cohort infant mortality is de-
fined as the cumulative probability of dying in the first year of life
(0-11 months, or 12 full months).
Infant mortality is particularly significant for Haiti because of
the widespread early introduction of bottle feeding. High rates of
infant mortality have been repeatedly shown to reflect the cumulative
effect of artificial feeding (Woodbury, 1922). In order to determine
what maternal and socioeconomic variables may influence early artificial
feeding and assdciated high death rates, mortality was studied in terms
of maternal occupation, literacy and household floor type.
The life table presents the probability of dying during a given
interval (in this case three-month intervals), sometimes called the
death density function, and the probability of survival for at least a
given time t, also known as the survivorship function (Gross and Clark,
1975). (Survivorship is also equal to one minus the cumulative death
density.) In Appendix A details of the life table construction are
illustrated. Weighted values were used for life tables when the effect
of a particular variable on death of the study children was being ob-
served.
(4) Statistics
Chi square analysis and T tests for the two sample location
problems were used to test for differences in nutritional status by
child's age, maternal work, time or education variable, or floor type.
Analysis of variance was used to sort out the variation observed in
mean percentage weight for age, height for age and weight for height,
32
and to test for significant differences among samples.
Mortality differences were tested by calculating the standard
error of the estimated probability of survival to time t, SE(St). A
test statistic (similar to the T test) with a normal distribution could
then be used to find significant differences in mortality by maternal
occupation, literacy, and household floor type. An example of this
calculation is given in Appendix B.
33
VI. RESULTS
Descriptive Sample Characteristics
Of the 530 women studied, 70 percent had no primary school educa-
tion. Of the 30 percent who had some primary schooling, only 2 percent
had completed it. More younger women (15-19 years) have some primary
education; but this becomes difficult to complete with age, as 61 per-
cent of these women started but did not finish school. Having a primary
school diploma does not improve living conditions in the sample. Over-
all 29 percent of women claim to read a book in French and Creole.
Approximately 60 percent of the sample women participate in in-
come earning activities. The majority of these women consider
themselves marchands, or merchants. Proportionately there are
more marchands and fewer dependents or housewives in Brooklyn than there
are in Boston, reflecting the fact that Brooklyn is the poorer community.
By and large marchands are petty merchants and may or may not work at
home. Forty-four percent of the women said they worked in the market. Women
who work at home sell shelled and roasted peanuts, clairin (an herbed
alcoholic beverage sold as a traditional or medicinal drink by the
shotglass) and Argo, a laundry starch that is bought and then resold. 3
Marchands make only a small amount of money, sometimeq none at all. Some
women (4 percent) engage in cottage industry, working mostly in em-
broidery and piecework. Only 6 percent of the ever-pregnant women were
factory workers and less than one percent were domestiques (maids).
This is not surprising, since factory workers and domestiques make more
money and are less likely to be living in the slum.
i i
34
Nearly 30 percent of the women who were ever pregnant spend whole
days working away from the home. The remainder work at home or are de-
pendents and housewives and do not work for money. The peak age for
working outside of the home is 25-29 years.
Working characteristics of the women are largely determined by a
complex system of cohabitation. Marriage in the formal sense is rare
due to the excessive costs. Just 13 percent of the sample women
were married. Three-quarters of the women studied reported living in a
placage or consensual union. In a placage a woman may or may not be
living with her partner. With increasing age, a woman is more likely
to be abandoned by her (common law) spouse. Over 50 percent of the
sample women were, by the age of 45, living without a husband (place
separe). Thus by this age an ever higher proportion of women enter the
labor market. In Boston, for example, 66 percent of women aged 40-49
are market workers and 27 percent are dependents or housewives, while
among women 20-29 the respective proportions are 27 percent and 55 per-
cent. Hence entry into the labor force is partly forced by age and the
tendency of husbands to leave home.
Poverty is reflected overall in the lifestyle of the marchand.
In Brooklyn conditions are worse, mortality is higher and a greater
number of marchands struggle to earn a meager living. As the women
age, their hardships only worsen.
Infant Feeding Practices
Bottle feeding in the first three months of life was nearly uni-
versal among the sample of ever pregnant women. As shown in Table 1,
over 70 percent of all study children were introduced to the bottle in
the first three months of life. Among the newborn infants, 88 percent
))
Age During Which Bottle Introduced
never rec'dAge in Total
N 0-2 3-5 6-8 9-11 12-14 15-17 18-20 21- 23 bottle
88.24
76.32
74.51
65.08
72.41
65.00
68.57
100.00
71.38
10.53
7.84
11.11
1.72
7.50
5.71
0.00
6.91
0.00
1.59
1.72
2.50
5.71
0.00
1.59
1.72
2.50
0.00
0.00
1.00 1.00 0.33
*Values represent percentage of N
) I 1)
Months
0-2
3-5
6-8
9-11
12-14
15-17
18-20
21-23
Total
17
38
51
63
58
40
35
2
304
0.00
2.50
0.00
0.00
11.76
13.16
17.65
20.63
22.41
17.50
25.71
0.00
19.08
2.50
0.00
0.00
0.33
0.00
0.00
0.00
0.00
0.00
Table 1
BOTTLE INTRODUCTION
BY AGE (in months) AND TIME OF INTRODUCTION*
36
had already been given a bottle. Only 19 percent of infants had never
received a bottle. It is likely that these were children of the poor-
est mothers who could not afford the bottle ingredients.
A life table analysis of bottle introduction shows the probabili-
ty of bottle introduction at each age interval for the study children
(Table 2). During the first three months of life a study infant is 72
percent likely to receive his or her first bottle. Between the third
and fifth month the proportion of children just receiving their first
bottle is 25%. By the end of the first year the cumulative probability
of having been introduced to the- bottle is 81%.
Analysis of the types of additional foods in the diets of children
0-20 months revealed an interesting trend toward greater cereal con-
sumption and lower milk consumption with increasing age. Also evident
was the .trend through the first 20 months toward one meal per day.
Table 3 presents the average number of servings per day and the pro-
portion of children, by age, receiving each food for nine categories of
foods.
Seventy-one percent of infants 0-2 months of age receive an average
of nearly two servings per day of milk and dairy products (excluding
breast milk). Milk products represent bottle feeds, which contain
sugar, Argo and a small amount of Carnation evaporated milk. More than
half of the newborns are fed a serving of bouilli or soup.. Bouilli is
a watery, starchy gruel made of corn primarily, cassava, arrowroot or
plaintain. Its caloric density is approximately 30 kcals per 100 gms.
Soup is generally a watery bouillon with bread broken into it. Thirty-
eight percent of infants at this age receive an average of less than one
VZZB' 9L'LTa
1008S' 666TO
788LO 911Z8
Z9Ii? BESZO
00000000OO'T
x l xd xb
(xb-I) (1Xz/xD)
S rt 0 rtO0
Z ~ rv'PO rt "0
rtrr
mD 0 0
0OrD
0 rt n
r
Crt
XZ xtK I
ft
0
0 S" D
00
OD
nt
0<
0
nt
ft
H'A
(D
0
C r
<~CD
H-
0'
H'rt
(D
nt
DQ
N0ILIIlQ0UILU T'lMOEa o 2TIEVI Sax?
0000'aI
0000 "'T
6Z66'
69Z6
00001,
0000"
TVLOa
90'0'
Tt'L0
Z9TL'*
0
CT-I
i7Z
goo
9S"7
9,S '
LOC.
0
6
LT
IC
L 17
69
98
'70C
0
6
8
17I
91
UT
9z
LITZ
0
6
L
CT
1
6
9
z
0
0
T
LTZ
0z-T
11-6
9 -o
Sd-I
O n
t
H0
rtH
HO
o r?'
0l %4
zD -S
ft
CL
0
ft
ft
(D
a XP6
38
Table 3
Foods Eaten by Children by Age and Type of Food-
Age inib
Months '-Types of Food
1 2 3 4 5 6 7 8 9
0-2 #reported 11 3 8 0 15 0 1 0 7
(n=21) avg. amt. .95 .30 .70 0 1.90 0 .10 0 .50
proportion .5238 .1429 .3820 0 .7143 0 .0476 0 .3333
3-5
(n=40) # reported 17 7 27 1 18 2 4 0 12
avg. amt. .75 .30 1.15 0.05 1.30 .10 .15 0 .50
proportion .425 .175 .675 .025 .45 .05 .1 0 .3
6-8
(n=54) # reported 26 15 27 4 23 4 2 2 22
avg. amt. .90 .45 .75 .15 1.30 .10 .05 .05 .70
proportion .4815 .2778 .5 .0740 .4259 .0740 .0370 .0370 .4074
9-11 # reported 34 34 50 8 23 11 2 0 13
(n=67) avg. amt. .80 .75 1.05 .20 1.20 .35 .05 0 .30
proportion .5075 .5075 .7463 .1194 .3433 .1642 .0300 0 .1940
12-14 # reported 28 39 53 12 28 5 2 0 16
(n=65) avg. amt. .65 1.10 1.35 .30 1.40 .15 .05 0 0.25
proportion .4308 .6 .8154 .8146 .4308 .0739 .0308 0 .2462
15-17 # reported 11 21 41 4 14 0 3 1 9
(n=44) avg. amt. .40 .85 1.50 .15 .85 0 .10 0 .25
proportion .25 .4773 .9318 .0909 .3182 0 .0682 .0227 .2045
18-20 # reported 11 21 40 3 12 1 2 0 12
(n=39) avg. amt. .55 .90 1.65 .15 .70 .05 .10 0 .50
proportion .2821 .538 1.0254 .0769 .3077 .0256 .0513 0 .3077
- Ce] Ll contents:
count
avg. # servings/day
proportion of n
- Code:
1 = soup or bouilli
2 = bean sauce and beans
3 = cereals
4
5
6
7
8
eggs, meat, fish
milk, cheese
fruits and fruit juices
vegetables
table sugar, in food or with
milk
9 = other food
39
serving per day of cereal products.
Between 3 and 5 months, 67 percent of children receive one or
more servings of cereals. A lower proportion of children (45 percent)
are receiving milk products, and in smaller amounts (nearly one serv-
ing less per day than that of infants 0-2 months). The proportion
of children consuming soup and. bouilli is slightly higher (42 perent).
By the age of 11 months, the proportion of children receiving
milk products has again decreased. Only 34 percent of infants are fed
a serving of milk. Cereals are proportionately greater in the diet,
with 75 percent of children receiving one or more servings. Builli
and soup consumption has increased to 51 percent. By this age sauce
pois, or bean sauce, has become significant in the diet (51 percent con-
sumption). Bean sauce represents an additional starchy component of
the diet as it is always served with rice or corn.
From 12 to 14 months, 82 percent of children are receiving an
average of 1 servings per day of cereals. Sixty percent of one-year-
old children are fed one or more servings of sauce pois. Milk con-
sumption has declined proportionately as well as in average number of
daily servings. By the age of 20 months all children are fed nearly
two servings per day of cereals, over half receive a serving of sauce
pois, and less than a third receive a small amount of milk. These
striking trends in cereal and milk feeding are illustrated graphically
in Figure 2.
Overall, mothers place relatively more importance on feeding milk
products to the younger infants. Cereal consumption increases rapidly
i I
FIGURE 2
100 Patterns of Milk and Cereal Feeding During First Twenty Months of Life
(Width of bar indicates average number of servings per day)
90 Milk products (excluding breast milk)
~ 80 ~S9Cereals
,a 8
0
Pz4
~3 70
4) 60
S 50-
4-
0 20
10
0-2 3 56-8 9-1 12-14 15-17 18-20
Age of Children
(months).
I
)
41
with age, as bottle feeding declines. Sauce pois becomes a significant
weaning food between 9 and 11 months. Bouilli is an important dietary
component throughout the first year, but the amount fed remains approxi-
mately constant with age. By the end of the second year of life,
children, like adults, are likely to be eating one meal per day.
Anthropometric Analysis
An analysis of growth by age and sex revealed that growth failure
after the first six months is more severe among female children. Mean
percent length for age, weight for length and weight for age by age in
months and sex for all study children are shown in Table 4. Faltering
in length for age below 95 percent of the reference median begins in
both boys and girls at about the ninth month and continues to decline
through the seventeenth month. There appears to be some catch-up
growth in males after 17 months but feamles continue to falter through
the twentieth month. Mean weight for age for males is below 80 percent
of the reference median from 10 to 17 months, while females fall be-
low this point from 9 to 21 months. Weight for length of boys is below
94 percent of the reference median betwen 8 and 12 months, rises slight-
ly at 13 to 15 months and declines again through the twentieth month.
Girls fall below 94 percent of the reference weight for length beginning
in the seventh month and show no catch up growth in the second year.
Thus failure to gain satisfactory weight and height begins earlier and
continues longer among female children. The growth of male children
tends to improve in the latter part of the second year.
iIi
42
Table 4
Mean Length for
Children by
Age, Weight for Length and Weight fQr Age of
Age and Sex (% of NCHS Reference Median)
Age in
Months Length/Age Weight/Length Weiht/Ae
Males
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
95.,6
94.5
95.3
96.4
96.4
96.4
95.6
95.0
93.6
92.1
91.1
90.3
91.1
91.5
91.7
92.4
92.7
94.3
106.0
107.9
103.6
98.2
95.7
94.4
92.1
90.5
90.4
91.2
91.9
93.2
95.7
94.8
91.8
91.2
94.1
93.7
109.1
91.1
90.6
84.4
85.1
84.7
82.6
80.3
79.5
78.3
76.9
77.0
76.6
77.7
76.4
77.0
82.6
88.3
101.4
147.3
Females
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
100 .0
99.6
99.1
97.1
96.5
95.5
94.8
93.7
93.5
93.1
93.5
92.8
93.2
93.3
92.7
91.4
91.7
91.0
92.8
104.8
102.5
99.6
95.5
91.9
92.8
91.1
90.0
91.1
89.0
88.4
89.7
87.7
88.3
91.8
89.1
88.4
89 .2
88.5
96.5
93.8
90.9
87.2
82.7
81. 7
79.3
75.4
76.6
74.7
75.8
77.5
78.4
77.2
80.7
76.9
76.9
76.4
79 .3
79.3
i 1
43
When the sexes were combined, growth parameters were analyzed by
working status of the mother and age of the child in one-year inter-
vals (Table 5). By the second year, mean percents length for age,
weight for length and weight for age have fallen among all children.
The- most dramatic decline between the first and second year is in
percent weight for age, from 85 percent to 78 percent. From 0-11
months, infants of marchands are significantly lower in percent weight
for age than children 0-11 months of all other mothers. In the second
year, children of other workers (predominantly factory workers) have
significantly higher mean percents length for age, weight for length
and weight for age than either children of marchands or dependents
and housewives. Surprisingly, children of the latter two samples are
growing at about the same rate. This was unexpected since marchands
work for money while dependents and housewives do not. Over all ages
children of factory workers show substantially better growth in weight
for age.
Growth was then examined by location of mother's work (Table 6).
During the first year children of mothers who work at home have a sig-
nificantly higher mean percent weight for age than all other children
of working mothers. By the second year this difference is actually
reversed. Children of women who work in factories or in the market
have approximately maintained 80 percent weight for age, while children
of women who work at home have dropped in weight for age from 86 to
76 percent. In the second year children of women working at home also
grow considerably worse in length for age. Working outside the home is
apparently advantageous for growth in the second year.
44
Table 5
Analysis of Length for Age, Weight for Length and Weight for Age
of Children- by Work Status of Mother and Age of Child
Age of Mother's % % %
Children Work Status Length/Age Weight/Length Weight/Age
0-11 Mos. Marchand 95.75 + 7.05 95.37 + 12.86 81.88 + 13.06-
(n45)
Other Worker 95.84 + 4.02 94.75 + 20.13 84.30 + 8.937
Dependent or 96.51 + 4.78 96.04 + 12.37 87.09 + 14.58J
Housewife
(n-46)
Total 97.11 + 7.29 95.63 + 13.31 84.50 + 13.55
(n-99)
12-23 Mos. Marchand 92.29 + 4.47] 91.05 + 10.74] 76.17 + 13.25]
(n-35)
Other Worker 97.41 + 1.61- 98.21 + 16.02_/ 89.49 + 22.29-
(n-8)
Dependent or 91.57 + 5.63] -89.77 + 11.37]- 77.02 + 14.36]
Housewife
(n46)
Total 92.38 + 5.71 91.03 + 11.58 77.81 + 14.76
(n-89)
0-23 Mos. Marchand 94.36 + 6.51 93.52 + 13.87 79.16 + 13.70
(n=80)
f/
Other Worker 96.69 + 3.34 96.49 + 10.47 86.89 + 14.73]-
(n-16)
Dependent or 94.10 + 6.56 92.98 + 12.60 82.15 + 16.34
Housewife
(n-92)
Total 94.43 + 6.34 93.51 + 13.00 81.28 + 15.14
(n-188)
aI
- Based on NCHS/CDC reference median; values represent mean + 1 std. dev.
b/ *
- Weight/Age significantly dif ferent betweens marchands and all others @ p= .04.
-ELength/Age significantly different between other workers and marchands @ p<.005;
between other workers and dependents/housewives @ p < .005.*
4'Weight/Length significantly different between other workers and marchands
@ p - .07; between other workers and dependents/housewives @ p = .04.*
Weight/Age significantly different between other workers and marchands @
p < .025; becween other workers and dependents /housewives 9 p <025.*
- Weight/Age significantly different between other workers and all others @
p = .05.*
Significance of one-sided T-tests within age groups.
III:,
45
Table 6
Analysis of Length for Age, Weight for Length, and Weight for Age
of Children-A by Location of Mother's Work and Age of Child
Age of Location % % %
Children, of Work Length/Age Weight/Length Weight/Age
0-11 Mos. Home 96.86 + 7.72 95.92 + 12.12 85.85 + 13.09
(n-50)
Factory 94.80 + 4.00 91.30 + 5.62 79.25 + 7.31
(n=5)
Market & 94.77 + 3.81 93.89 + 13.16 80.59 + 15.09
Elsewhere -
(n=31)
Total 95.99 + 6.42 94.92 + 12.28 83.57 + 13.64
(n=86)
12-23 Mos. Home 91.36 + 5.44- 92.26 + 10.93 75.85 + 14.14
(n56)
Factory 96.40 + 2.09 91.98 + 7.31 80.34 +17.36
(n6)
Market & 94.00 + 4.56 92.12 + 13.88 82.29 + 14.37
Elsewhere -
(n=20)
Total 92.37 + 5.09 92.21 + 11.53 77.75 + 14.42
(n=82)
0-23 Mos. Home 94.03 + 7.19 93.48 + 11.95 80.61 + 15.14
(n-100)
Factory 95.61 + 3.17 91.67 + 6.96 79.85 + 10.98
(n=ll)
Market & 94.45 + 4.43 93.12 + 14.43 81.15 + 14.53
Elsewhere
(n-51)
Total 94.26 + 6.24 93.25 + 12.49 80.73 + 14.73
(n-168)
- Based on NCHS/CDC reference median; values represent mean + 1 std. dev.
b/Length/Age significantly different between home and all others @ p<.Ol.*
c/Weight/Age significantly different between, home and all others @ p<.05 (0-1l mos.)*
d/Weight/Age significantly different between home and all others @ p<.05 (12-23 mos.)*
Significance of one-sided T-tests within age groups.
I I
46
To further investigate what might account for the difference in
first-year growth of children whose mothers stay at home, length for
age and weight for length were studied by the number of hours per day
women spent with their children (Table 7). Generally women spent the
whole day (24 hours) or less than a third of the day (0-7 hours) with
the child. Interestingly, length for age and weight for length were
significantly better in the first year of life (p = .10 and .11, re-
spectively) among children whose mothers spent the entire day at home.
There were no differences in growth in the second year or over all
ages.
Tests of occupational characteristics were used to partial out
the differences observed in second-year growth. Analyses of variance
in length for age, weight for length and weight for age were done for
all children 12 to 23 months old (Tables 8-10). Work status and loca-
tion of mother's work proved to be significant sources of variation
among samples in length for age. Neither of the null hypotheses could
be rejected for weight for length or weight for age.
Waterlow classifications using the parameters of length for age
and weight for length were selected to look at the incidence of mal-
nutrition. The data for all children 0-23 months are plotted in
Figures 3 and 4
In this study, 17 percent of the sample children (n=189) were
identified as stunted , falling below 90% of the NCRS reference median
length for age. Ten percent of children were below 80% of the refer-
ence median weight for length, identified thus as wasted. Broken
down by work status of the mother, 70 percent of the children of
.11
47
Table 7
a/
Analysis of Length for Age and Weight for Length of Children-
by Time Mother Spends with Child and Age of Child
Age of Hours/Day % %
Children With Child Length/Age Weight/Length.
0-11 Mos. 0-7 94.62 + 4.56- 92.04 + 8.19
(n=20)
24 96.55 + 6.99] 96.12 + 13.01]
(n=81)
Total 96.17 + 6.59 95.31 + 12.23
(n=101)
12-23 Mos. 0-7 92.91 + 6.30 91.75 + 9.26
(n=22)
24 92.13 + 5.10 90.56 + 11.55
(n=72)
Total 92.31 + 5.40 90.84 + 11.07
(n=94)
0-23 Mos. 0-7 93.85 + 5.30 92.17 + 9.60
(n=42 )
24 94.55 + 6.55 93.56 + 13.52
(n=153)
Total 94.40 + 6.30 93.28 + 12.86
(n=195)
a/Based on NCHS/CDC reference median; values represent mean + 1 std. dev.
b/ *
- Length/Age significantly different between 0-7 and 24 hours at p = .10
- Weight/Length significantly different between 0-7 and 24 hours at
p = .11*.
*Significance of one-sided T-Tests within age groups.
i I
48
Figure 3
Distribution of Percent Length for Age
'Measurements of Study Children
a a
*peGe
009.
*e*S
S
*0
* 0
S. *
-. p
* - S
* a* a
*
. 0
a. * *
* 0
* me
0 5 -- -a ---
* -. 5 0
. a a
-vK - "-
a4S
-. S
, # I
MEDIAN
STUNTING
*
12 18
130
120
0
a
C
0
CD
0)
"M
C
z
110
100
s0
80
S
;a4p
a 0
-m
.o
a
-e
0 6
70
0 6
AGE
IL
0
a 6I
a
d
49
Figure 4
Distribution of Percent Weight for Length
Measurements of Study Children
.5 '.
* a
* *:'. 0 S
- .S.
- 6 -
* ,9
* ,*6
p
ar-a 0
5 .
. , *
12
0
I
,
18
S. MEDIAN9
** 0
o0
WASTING
24
AGE
. 0 
-.
mc
44
0)
C
Ma
N
2
140
120-
100-
80-
p
0
6 9*
* S.
0
a
6 0-
-T
0
I
-V
6
50
Table 8
Analysis of Variance in Length for Age of Children 12-23 Months by
by Mother's Work Status and Location of Work
Source of
Variation
Sum of
Sauares DF Mean Square
Significance
F of F
Among Samples
Work Status
Location
Within Scrples
Work Status
Location
Total
Work Status
Location.
233.2739
207.7090
2124.2774
2046.7399
2357.5513
2254.4489
4.72
4.01
<.025
.05
116.6370
103.8545
24.7009
25.9081
2
2
86
79
88
81
i i
51
Table 9
Analysis of Variance in Weight for Length of Children 12-23 Months by
Mother's Work Status and Location of Work
Source of Sum of Significance
Variation Squares DF Mean Square F of F
Among Samples
Work Status
Location
Within Samples
Work Status
Location
Total
Work Status
Location
485.4628
.6194
11532.2900
10502.3310
11532.2900
10502.9504
1.81
.002
NS
NS
242.7314
.3097
134. 0960
132.9409
2
2
86
79
86
81
Table 10
Analysis of Variance in Weight for Age of Children 12-23 Months by
Mother's Work Status and Location of Work
Source of
-riatioinfn
Sum of
S2uares DF
Significance
Mean Square F of F
Among Samp Les
Work Status
Location
Within SampLes
Work Status
Location
TotaZ
Work Status
Location
1214.2238
654.6406.
18735.7540
16426.9760
19949.9770
17081.6166
2.79
1.54
NS
NS
607.1119
327.3203
217.8576
207.9364
2
2
86
79
88
81
V Ct L JL CL L- JLULL LJ %4 LA CL JL - %-J A.F J6,
.-I j I .
II
53
marchands, 93 percent of the children of other workers, and 61 percent
of the children of dependents and housewives were normal (Table 11).
Chi-square tests showed that more children of dependents and housewives
were wasted, stunted and concurrently wasted and stunted, but only at
significance levels of about .20. When looked at by number of days
per week mothers work outside of the home, 68 percent of children whose
mothers worked 0 days per week were normal compared to 79 percent of
children whose mothers worked 1-7 days per week (Table 12). Chi-square
testing with 4 df revealed significant differences in wasting and con-
current wasting and stunting, favoring workdays outside of the home.
Fewer children whose mothers worked up to seven days per week were mal-
nourished under the Waterlow classification.
Further disaggregation by age of child and location of mother's
work revealed differences in nutritional status similar to those re-
ported for growth (Table 13). A chi-square test verified that sig-
nificantly more children of women who worked at home were stunted.
This difference did not disappear where the age groups were combined.
By the second year of life, children whose mothers earn a doorstep liv-
ing become increasingly malnourished and the percentage of them classi-
fied as normal falls from 84 to 47 percent.
Working outside of the home actually affects child nutrition be-
fore the second year of life. When the Waterlow classification for
children 6-23 months was broken down by number of days per week women
worked, more children were wasted and stunted among women who worked
0 days (Table 14). Only 63 percent of children whose mothers stayed
at home were normal, compared to 79 percent of those whose mothers
54
Table 11
a/
Analysis of Waterlow Classification of Nutritional Status
of Children 0-23 Months by Work Status of Mother
Mother's
Work
Status Total N Wasted Stunted
Wasted
and
Stunted Normal
Marchand
Other Worker
Dependent/
Housewife
77
15
90
8 14
(10%) (18%)
0
(0%)
1
(7%)
10 20
(11%) (22%)
a Based on NCHS/CDC reference median: <90%
<80%
<90%
<80%
ref. height/age =
ref. weight/ht. =
height/age and
weight/height =
(x2 tests showed lower levels of nutritional status among dependents /
housewives' children only at p > .20)
I
(1%)
0
(0%)
5
(6%)
54
(70%)
14
(93%)
55
(61%)
stunted
wasted
wasted
and
stunted
.11
55
Table 12
ANALYSIS OF WATERLOW CLASSIFICATION OF NUTRITIONAL STATUS
OF CHILDREN 0-23 MONTHS
BY NUMBER OF DAYS/WEEK MOTHER WORKS
Workdays Total N Wasted Stunted Wasted & Stunted-
0
1-4
128
5
5
6
7
15
24
14
581-7
12
(9%)
2
(40%)
0
(0%)
2
(8%)
0
(0%)
4
(2%)
25
(20%)
1
(20%)
2
(13%)
1.
(4%)
3
(21%)
7
(2%)
4
(3%)
1
(20%)
0
(0%)
0
(0%)
0
(0%)
1
(2%)
Normal
87
(68%)
1
(20%)
13
(87%)
21
(88%)
11
(79%)
46
(79%)
NCHS/CDC reference median: <90%
<80%
<90%
ref . height/age
ref. weight/ht
height/age
and
<80% wt/ht
/ 2 (4df) significant at p = .06
X2 (4df) significant at p = .13
= stunted
= wasted
= wasted
and
stunted
Based on
56
Table 13
ANALYSIS OF WATERLOW CLASSIFICATION OF NUTRITIONAL STATUSa!
OF CHILDREN BY AGE OF CHILD AND LOCATION OF MOTHER'S WORK
Wasted
Age of Location
Chi1dA Jnork TotalV M wasqt- d Stumntrd
and
Srin ri
Based of N /1
ie median, <90%
<80%
<90%
<80%
ref. height/age
ref. weight/ht
height/age and
weight/ht
b/ X 2 (ldf) significant a
c/ X2 (Idf) significant at
= stunted
= wasted
= wasted
and
stunted
.11
Nnrman1
0-11 Home 50 1 7 10 42
mos. (2%) (14%) (0%) (84%)
Factory 5 0 1 0 4
(0%) (20%) (0%) (80%)
Market & 31 3 4 1 23
Elsewhere (10%) (13%) (3%) (74%)
12-23 Home 56 9 19 3 25
mos. (16%) (34%) (5%) (47%)
Factory 6 0 0 0 6
(0%) (0%) (0%) (100%)
Market& 2 3 0 15
Elsewhere 20 (10%) (15%)_j (0%) (75%)
0-23 Home 106 10 26 3 67
mos. (9%) (25%) (3%) (63%)
Fac tory 11 0 1 0 10
(0%) (9%) (0%) (91%)
Market.& 51 5 7 1 38
Elsewhere (9%) (14%)_j (2%) (75%)
57
Table 14
.a/
Analysis of Waterlow Classification of Nutritional Status
of Children 6-23 Months by Number of Days/Week Mother Works
Workdays Total N Wasted Stunted Wasted & Stunted Normal
0 100 11 22 4 63
(11%) (22%) (4%) (63%)
1-4 5 2 1
(40%) (20%) (20%) (20%)
5 15 0 2 0 13
(0%) (13%) (0%) (87%)
6 17 1 0 0 16
(6%) (0%) (0%) (94%)
7 11 0 30 8
(0%) - (27%) (0%) (73%)
1-7 48 3 6 C 1 38
(6%) (13%) - (2%) (79%)
a!
a Based on NCHS/CDC reference median: <90% ref. height/age = stunted
<80% ref. weight/ht. = wasted
<90% height/age and
<80% weight/height = wasted
and
stunted
b/ s
- X2(4df) significant at p = .06
c/ x2 (4df) significant at p = .13
58
worked one or more days outside the home. The difference in percent
of children wasted was significant at the 0.02 level when the chi-
square test had 4 df. The difference in percent of children stunted
was less strong (p value 0.17) in a chi-square test with 1 df (0 days
versus 1-7 days). The pattern clearly suggests that after six months
children of women making more money by working outside of the home,
have better nutritional status.
Mortality
In this study mortality among all infants 0-23 months old was
205 per 1000 live births (Table 15). Mortality at the end of the
first year was approximately 163 per 1000 live births. The' proba-
bility of a study infant dying in the first three months of life was
8.5 percent. As expected, the proportion of infants dying decreased
with each succeeding three-month age interval. Although the proba-
bility of survival at 9-11 months was 98 percent, only 95 percent of
children 12-17 months survived.
When survival was analyzed by work status of the mother, mortality
among infants of marchands was 172 per 1000 live births compared to 129
among the infants of dependents and housewives (Tables 16 and 17).
This difference was statistically significant at a level of 0.14.
Through 23 months of life, however, mortality between children of
marchands (217) and children of dependents and housewives (177) showed
a statistical difference only at a level of 0.23. The probability of
survival at given age among children of dependents and housewives be-
gins to decline at 12-17 whereas among marchands' children, survivability
59
Table 15
LIFE TABLE AND SURVIVAL ANALYSIS (WEhIGHTED)
COHORT OF CHILDREN 0-23 MONTHS
U4 ) Hr
C) 4 44 dL4 'H '
44-a aC
Cd0 W-i 4- o]0 0
CdQ ::j cd cu Z-W 0 0 w )>d> >
> > 'H CdC) w r-i 0 0mO 4''
4-i H"0 4CCdd C)"0 4) w41 'r-4 > -W44-1
> w" 0 >Cd wCw 4-) W p p'> Cd$.p ca m
HmC) .4) c'H4. 4WJ mlo 0 0'o ' w
'H du24O-j d0 w"0 Q0 m 0 C fl f5
w) w r_: O44Q) pC ;'n4-40Cd L64
z**=C)P-4X= 0.4 cnW u 0 u 0
(Cx/Zx') (1-qx)
x Cx Tx Wix Lx Lx' qx px Px I-Px
0-2 60.30 94.96 155.26 760.44 712.96 .0846 .9154 1.0000 '0000
3-5 34.8 6 107.74 142-60 605.18 551.31 .0632 .9638 .9154 .0846
6-8 15.58 .81.08 96.66 462.58 422.04 .0369 .9631 .8823 .1177
9-11 4.78 102.44 107.22 365.92 314.70 .0152 .9848 .8497 .1503
12-17 8.34 179.58 187.92 258.70 168.91 .0494 .9506 .8368 ..1632
18-2.2 0.O0 70.78 70.78 70.78 35.39 .0000 1.0000 .7955 .2045
60
Table 16
LIFE TABLE AND SURVIVAL ANALYS IS
CHILDREN OF MARCHANDS
(u)4-1 41 4-
( C x + T H ( C x
- 4 4 71 14 .84 94 H. 0 .0 C 4WC>000
HLH1 H-H
3- 2 0 2 1 4 13- .1 .HIW :1 -HHr-H
- 4 224 b"06 . 7)3cd 06
H c 93H 8. .03 .00 .>>0)>fr, > 4--H H "-1i H H 0 H-H
4ro p0 cd t 4'"0 Q) 10 44 H4"a> 4J 4J
11c) Z coW0> Q)0)Q) (U14> cap tUc
H2-1421 c3n 4 39.M .0 OH .87 .- 72
H-2o20 ,o z2) 60 ."00.0C.86 .c4
QO4J P - 6 0cd 0S-4p
60 4 J >4 r o x $ p :jL6
(Cx+Tx) (G x/Zx1) (jl.-cjx)
x Cx Tx Wx PZx 9,x _- qx px Px 1-Px
0-2 14 14 28 171 164 .0854 .9146 1.000G .0000
3-5 3 25 28 143 130.5 .0230 .9771 .9146 .0854
6-8 4 18 22 115 106 .0377. .9623 .8937 .1063
9-11 3 25 28 93 80.5 .0373 .9627 .8600 .1400
12-17 2 51 53 .65 39..5 .0506 .9494 .8379 .1721
18-22 0 12 12 12 6 .0000 1.0000 .7860 .2140
I..
II
61
Table 17
LIFE TABLE AND SURVIVAL ANALYSIS
CHILDREN OF DEPENDENTS AND HOUSEWIVES
co 0
tz 0 4-414-4
01 Z 0 0
HL4 H I p 0 ct
H$4 w a4 , 4.0.
0 41W3W0 0
to 0 = COO 00CpC p
H 0 H oC0124100CCc
> > 4H H "0H H 0 HH
41~- Hm0 $mc 0 "0 0" 4-H :j 14
" C)fr_:J0)::1C0) 04 It . > cn~j cc
HJ W)~ 1 W ~ 4c a" P, 0 OH H0 U4
(Cx+Tx) (Cx/Zx') (1-qx)
x Cx Tx Wx Zx gv _,x .px Px 1-Px
0-2 9 8 17 200 196 .0459 .9541 1.000 .0000
3-5 9. 17 26 183 174.5 .0516 .9484 .9541 .0459
6-8 4 31- 35 157 141,55..0283 .9717 .9049 .0951
9-11 1 39 40 122 102.5 .0098. .9902 ..8793 .1207
12-17 3 54 57 82 55 .0545 .9454 .8706 .1294
18-22 0 25 25 25 12.5 .0000 1.0000 .8231 .1769
62
begins a downward trend as early as 6-8 months. The proportion of
children 12-17 months surviving is 95% for both groups of children.
Apparently mortality is worse among children of marchands in the first
year, but in the second year survival between the two groups approach-
es equivalence.
Broken down by household floor type, mortality was markedly
different between children of households with cement floors and those
with dirt floors (Tables 18 and 19). This variable can be viewed
either as a socioeconomic index or a characteristic of the hygienic
environment. Death among the infants living in cement-floored houses
was 107 per 1000 live births. This compared to 184 among the infants
in houses with mud floors. The difference was significant at a level
of 0.02. The probability of death by 23 months was 166 per 1000 in
houses with cement floors, against 218 per 1000 in houses with dirt
floors. At a level of 0.12, this difference was weaker in significance
than the 0-11 month comparison. Only 3% of the infants with cement
floors died in the first three months whereas 8.5 percent of those
in houses with mud floors died before 3 months of age.
Surprisingly, when survival was analyzed by literacy of the sample
mothers, cumulative mortality among children of literate women was
higher at all ages (Tables 2Q and 21). Literacy was used as the closest
possible indicator of education since only a small percentage of the
women had ever entered primary school, and even fewer had completed it.
Infant mortality was 150 per 1000 among literate women and 132 per
1000 among illiterate women. This difference was statistically sig-
nificant at just the 0.15 level. Twenty-three month mortality was 228
. H 
.
Ii
63
Table 18
LIFE TABLE AND SURVIVAL ANALYSIS
CHILDREN OF HOUSEHOLDS WITH CEMENT FLOOR
>4 4
cu HO 4 J 44
Cd U>)0 0) 34
430 0040 .O-4
H 4 -H w 4 4 ,.
"O 0 w 40) =0)w co0Q0
HI ccf-H r.H" H) 0 0 w> > N
H 430 -- &H r - - H H -
-H H p r rj HotO1-w- > -H-
cu 4-W 4W-H P t 0 0-H 0 d q )
-H H0 0H043 043 . 04 >En>>
43 Hc O W -d N' CLJ"0 4 4H43r4
Ho 71 p3 CO N 43;300::0OH H4H
(Cx+Tx) (Cx/9kx') (1-qc)
x Cx Tx Wx kx 9,x' gx p-x Px 1-Px
0-2 7 11 18 214 208.5 .0336 .9664 1.0000 .0000.
3-5 9 23 32 196 '184.5'.0488 ..9512 .9664 .0336
6-8" 3 29 32 164 149.5 .0201 .9799 .9192 .0808
9-11 1 38 39 132 113 .0088 .9912 .9008-.0992.
12-17 4 65 69, 93 60.5 .0661 .933.9 .8928 ..1072
18-22 0 24 24 24 12 .0000 1.0000 .8338 .r1662
ii
64
Table 19
LIFE TABLE AND SURVIVAL ANALYSIS
CHILDREN OF HOUSEHOLDS WITH DIRT FLOOR
Ha >w 4- 4--
Hw CdCd 4- CL
(Cx+T) 0( 0 ' (-x
9-3 3 3344t c8 9 .33 974 4 .
H-2 3 7 44. . 8
- 42 0 2 ..
cQ d:j -H)c 0
> p4 Cu -wZ44 044:0 Otw
-) 0 0 w4H 4 > >'Z >
44 ~ - pI" w 71O )" 41 4r44
"0 >Z > 000)w - w$I >cd Ccdt
p p -w mr'r4N w40 COW 0 0 -H HZ0 H0)(
zpw 0,l0ZoZO 0 d P4> zc
0) C C-W w 44 r M1Q 4 0 t 0
~< E-4H z ~~P-1 u 0 Q Q
(Cx+Tx) (Cx/kx') (1-qx)
x Cx Tx Wx ix ixf gx ]px Ps 1-Px
0-2. 16 13 29 194' 187.5 .0853 .9147 1.0000 .0000
3-5 5 20 25 165. 155 .0323 .9677 .9147 ..0853
6-8 6 26 32 .140 127 ,.0472 .9.528 .8852 .114.8
9-11 3 30 33 . 108 93 ..0323 .9677. .8434 .1566
12-17 2 53 55 75 48.5 .0412 .9588 .8161. .1839
18-22 0 20 20 20 10 .0000 1.0000 .7825 .2175
65
Table 20
LIFE TABLE AND SURVIVAL ANALYSIS
CHILDREN OF LITERATE MOTHERS
r0r r_4
to d LH 44
(C )C/x) (-x
3-5 6 4 0 11- 10 .P41 942 .949 P33
cd _ 'H-
6-3 7 92 03. .3a .
wH cw 0 0N00 > >
9-02 2 24 72 60 .32.670 .20
'H 31 0 4 32. .0.07
N42 'Hz4 1 1 41 . 0 .000 .7774.2283
0) C 41'H 'H '0 'H 'H1 'H'
41 'H'14 N20C112 105J 0 571 1H.9 ~429 .69 .314
6-8 17 0 92 83,54 .035N9641 91N7C088
9-11412c2'H24 J72 61 .028'H672H879'H121
tO 4 1 34 48 2.X.093 .07N.8)1 .49
08-22 4 10 14 4 7 1260121.0 96 1.0000777 .0000
iI
66
Table 21
LIFE TABLE AND SURVIVAL ANALYSIS
CHILDREN OF ILLITERATE MOTHERS
4-J
46J4
P0) : 'H
x
0-2
3-5
6-8
9-11
12-17
Cx
19
8
6
2
3
4-4
'HI
'H0
44J
SCU
Tx
14
29
38
47
87
0
4J
to
T4 0)
C 0
44 p.
C (0)
0>
w- 4-N
rnUl4J
> 014
444 215
CU9 171
90 122
18-22 0 32 32 32 16 .0000 1.0000
0
(13
"0
Q)
)4
CU
"30
1440
2780
0
0
P4
(1-qx)
px
.9316
.9663
.9694
.9864
.9618
0
0
P44
.063
5 0332.036 
5 .0136
5 .0382
4-)
I'H
~0
CU
u0
P
.002
.8727
.86 08
4J
-H
r-Px
0
clCU
0
PSP
000
,069.0
.0998
.1273
.1320
.8279 .1721
67
and 172 among children of literate and illiterate mothers, respective-
ly, a difference without statistical significance. Between 0 and 2
months, 3 percent of children of literate women died compared to 7 per-
cent of children of illiterate women, but after three months the pro-
portion dying among the literate mothers was consistently higher. By
the second year, 9 percent of children in the literate group die in
the next six months, while only 4 percent of children of illiterate
women die between 12 and 17 months.
A summary of the mortality analyses f or the cohort of study child-
ren is presented in Table 22. The strongest differences appear when
mortality is broken down by floor type. Differences with weak sta-
tistical significance occur in the 0-11 month analysis by occupation
and in the 0-23 month analysis by literacy.
68
TABLE 22
a!Summary of Mortality Analysisa
Cohort b/ Significance
Mortality T of T
0-11 Months
Occupation
Marchands 179
1.06 0.14
Dependents /129
Housewives
Floor Type
Cement 107
Dirt 183 2.02 0.02
Literacy
Literate 150
Illiterate 132 0.44 0.33
Overall 163
0-23 Months
Occupation
Marchands 214
0.73 0.23Dependents/ 177
Housewives
Floor Type
Cement 166
Dirt 218 1.16 0.12
Literacy
Literate 228
Illiterate 172 1.05 0.15
Overall 205
a!
- Cohort mortality expressed as deaths per 1000 live births; values taken
from final column of respective life tables.
-- Sample test statistic calculation given in Appendix B.
i I
69
VII. DISCUSSION
The study established that women's work is clearly related to
growth and mortality at specific times in the first two years of life of
Cite Simone children. The results can be discussed within the general
framework of women's work and child nutrition outlined in Chapter Two.
Early introduction of bottle feeding and supplementary foods can-
not be said to be a consequence of the need for women to work. Feeding
practices are dictated by norms among the women and resource constraints,
rather than time limitations imposed by a job. Of all infants 0-2 months
old, 88% were being bottle fed. The bottle is the feeding method of
choice among Cite Simone women. It symbolizes prestige and modernity.
Furthermore, bottle feeding is initiated in the maternity units, and
all women who deliver their babies there come home with a bottle. Not
unexpectedly, other women in the community are motivated to do the same.
Bottle mixes are nevertheless notoriously deficient in nutrients
needed to support satisfactory growth. Women stretch their supplies
of canned evaporated milk as much as possible due to the cost of milk.
With increasing age children receive less milk and proportionately
more cereals, as was shown in Figure 2. Cereals are cheaper and more
easily prepared at the main midday meal. Curiously, women realize that
the younger babies have a higher priority for what little milk there
is, and that they must eat several times per day.
Although the women continue to breast feed through 16 to 18 months,
the early diet of supplementary foods is certainly a major cause of an
enormous incidence of diarrheal morbidity in this community. Early
feeding of additional foods is likely to reduce the amount of immunity
[I
70
to disease afforded by breast milk (Jelliffe and Jelliffe, 1971). This
can occur by a decrease in the amount of breast milk ingested, or an
increase in the amount of harmful bacteria in supplementary foods in-
gested. Such feeding practices are also likely to contribute to the
high mortality in the early months of life prevalent among the study
children.
Participation by mothers in the wage labor force affected welfare
of the study children depending on the occupation of the woman, the lo-
cation of her work and whether or not she worked outside of the home.
Working had stronger effects during the second year of life. Further-
more, occupation and location of work were found to affect length for
age and not weight for length or weight for age in analysis of variance.
These findings are related, since length takes a longer time to be sig-
nificant as a determinant of nutritional status. Work characteristics
appear to affect growth predominantly through an income effect.
Children of marchands were lower than all other children in weight
for age in the first year and in length for age, weight for length and
weight for age in the second. Marchands earn an extremely low wage,
often not enough to cover the costs of bottled milk. The poorer and
more self-supporting the women are, the greater is the necessity that
they work in the market. This is evidenced by the fact that more women
are marchands in Brooklyn, where socioeconomic conditions are poorer.
Of all women who reported living in single or informal unions (spouses
away), 83 percent were marchands. Factory workers, artisans,, coutures
or other employed women, who earn a higher wage, were either married or
in placage unions (except for some who were widowed, separated or
71
divorced). None reported living in single or informal unions. Mar-
chands are therefore forced to work in the market and at the lowest
wages. Kumar (1978) came to the same conclusion in her study of poor
working women in Kerala, India. It is likely that the time they spend
in child care is limited by the number of hours they work inside and
outside the home. The income they earn is not enough to provide suf-
ficient food for their children or a clean environment. Dirt floors
are more frequently seen in the homes of marchands. Their children
play in a moist, contaminated environment. By the second year of life
more of these children are stunted and wasted than children of other
workers.
Children of women who make more money (largely factory workers)
show better growth performance and have a lower incidence of malnutri-
tion than either children of marchands or dependents and housewives.
The effect predominates in the second year of life in the parameters
of length for age, weight for length and weight for age. The effect
does not occur in the first 12 months. Income appears to have a strong
impact on nutritional status in the second year.
The location of women's work is a determinant of child growth with
an income and a time effect. Among mothers who work at home, fewer
children than among women who work in factories and in the market are
wasted and stunted in the first year of life. Children of mothers who
work in the home also have higher weight for age in the first year. By
the second year, though, these children are significantly lower in
weight for age and length for age, and proportionately more are stunted.
:11
72
There are two possible explanations for these results. One is
that feeding patterns in the second year of life cause children of
mothers who work at home to falter in growth more severely than other
children. The second explanation is that time spent in child care is
critical for growth in the first year of life, and mothers who work at
home have more of this time available. The first of these reasons is
rejected based on the relatively homogeneous pattern of feeding known
to occur in Cite Simone; these feeding practices were described above.
The second explanation is proposed as the plausible one based on the
analysis of growth by time mothers spent with their children.
Analysis of hours per day women spend with children revealed that
in the first year children whose mothers spent the whole day with them
had significantly higher length for age and weight for length measure-
ments than children whose mothers spent only 0-7 hours with them. This
difference disappeared in the second year. Thus childcare time by
mothers in this study is a more important determinant of growth during
the first year while income assumes relatively greater importance there-
after.
Looking at the Waterlow classification of nutritional status by
number of days women worked outside the home suggests that income, in
fact, might become the predominant effect after the sixth month of life.
Of all children 0-23 months, a lesser percentage were wasted or
wasted and stunted among those whose mothers worked one or more days
outside the home. This corroborates the positive effect of income on
nutrition, since women who work outside the home earn more money than
those who do not. When the same analysis was performed on the 6-23
73
month old children, working outside of the home again had a positive
impact on nutrition. Among the study infants, then, the first six months
of life is likely to be the period of greatest risk of growth faltering
due to a maternal childcare time constraint imposed by working.
Deaths among the study children and survivalship were analyzed with
respect to women's occupation, literacy and household floor type. Mortal-
ity in the first year of life for all children was found to be 163 per
1000 live births. This is slightly higher than the infant mortality of
147/1000 reported in the 1978 Haiti Nutrition Status Survey by the Bureau
of Nutrition (1979) for metropolitan Port-au-Prince. Mortality through
the 23rd month for the cohort was approximately 205. The widespread
practice of artificial feeding beginning in the first three months of
life can be considered a major cause of the higher death risks observed.
Bottle feeding is known to have a cumulative effect on the risk of death
in subsequent months of the first year of life (Woodbury, 1922). The
earlier infants are introduced to bottles, the higher is the relative
mortality in any given month.
Higher rates of infant mortality among artificially fed children
compared to breast-fed children have been consistently documented in the
literature (Jelliffe and Jelliffe, 1981; Hill, 1979; Wray, 1978; Woodbury,
1922; Gordon et al, 1967). Unfortunately in this study deaths of
exclusively breast-fed versus artificially fed infants could not be com-
pared since nearly all women bottle fed their infants. An extremely high
incidence of respiratory and gastrointestinal infection, and deaths due
to these diseases, were observed. Unquestionably, infectious disease
played a predominant role in lowering the nutritional status of the
Ii
74
children. The synergistic effects of malnutrition and infection have long
been recognized (Scrimshaw, 1970; Mata et al, 1971).
Relative mortality risks of artificial feeding can be disaggregated
by socioeconomic class. Woodbury (1922) found that for the first nine
months of life mortality risks among artificially fed infants were highest
in the lowest income group. In this study, analysis of mortality by
occupation of women and floor type indicated that differences did occur
between socioeconomic groups.
For the first 12 months children of marchands experienced a higher
risk of death than children of dependents and housewives. The difference
was weakly significant at a level of 0.14. A greater proportion of deaths
in the first three months among the marchands' children can be considered
largely responsible for this difference. The time and income constraints
speculated to be causally associated with lower nutritional status in this
group may also have an effect on mortality. Dependent women and housewives
could be more apt to use the health services as well. Mortality for 23
months differed less between the groups, probably because the high risk
at 0-2 months among children of marchands was felt less strongly.
Children of women who reported having cement floors had a much lower
risk of death at the end of the first year than children who lived in
houses with dirt floors. Their risks differed significantly by the end
of 23 months as well. Children in houses with dirt floors are subject to
greater bacterial contamination because storm sewers overflow when it
rains, the sewerage trickles through the homes, and dirt floors become mud
floors. This problem is severe and continuous, as Cite Simone is built
on swampy ground. The study did not determine if women who had higher
paying jobs more frequently reported having cement floors. If women with
.1i:
75
better jobs can afford cement floors then income again might be a dif-
ferentiating factor.
Literacy of mothers was unexpectedly associated with higher mortality
at 23 months (p value 0.15). As an index of education, these results are
contrary to Caldwell's findings in Africa (1981), where maternal education
was inversely correlated to risk of young child death. Admittedly, though,
literacy is a poor index of education. Mortality could not be assessed
according to education because of the small number of women who completed
primary school. It is likely that the study women exaggerated their abil-
ity to read. Nevertheless, literate mothers may be more inclined to buy
Argo and canned milk in an effort to show other women how modern they are.
Their children could be receiving bottle feeds earlier and more often.
Analysis of many other variables related to women's work, and human
and capital resources would have helped explain the growth and mortality
results found in this study. Parity of the mother, her age, and the num-
ber of surviving children in a family are important potential determinants
of mortality in the first year. Gordon et al (1967) found in India that
death rates were higher for infants of first and late pregpancies and
among very young and old women. Frequency of visits to health centers
would have provided comparative information on child nutritional status
and mortality based on use of health facilities. Berggren et al (1981)
report that disease-specific mortality rates among rural Haitian children
have fallen progressively during five years since the delivery of a primary
health care program. Child-care time and employment characteristics of
other family members are factors repeatedly shown to influence child
growth (although the net effect is variable). Due to time and data
constraints, these and other important variables could not be examined.
76
The complexity of factors contributing to child welfare can only partially
be disentangled, even within a well-defined study framework.
The results described herein confirm that in low income communities
women's work profoundly affects growth and nutrition status at various
stages in the early life of a child. Infant feeding practices assume
a relatively minor role in determining growth differentials given the
homogeneity of bottle feeding which prevailed. The uniqueness of the pre-
sent study is the inclusion of mortality in the time/inc.ome analysis of
child welfare.
At.
77
VIII. SUMMARY AND CONCLUSIONS
The purpose of this study was to assess the continuing problems of
growth failure and excessive infant mortality from the point of view of
poor women's work. Broadly, this approach is a departure from the con-
ventional reliance on knowledge of breast and bottle feeding patterns to
explain infant malnutrition in low income societies. The framework of
women's work and child welfare provides a method for researching poor
growth and survival more viable and inclusive than that based on measure-
ments of breast milk adequacy. In communities where infant feeding is
traditionally mixed between breast milk and supplementary foods.
this approach is particularly relevant.
The specific objective of the study was to analyze growth and mortal-
ity of a sample of Haitian children 0-23 months old with respect to maternal
occupation, socioeconomic level, literacy and time use. The results are
summarized along two lines as follows:
Growth and Nutrition Status
1) Growth faltering and malnutrition were severe problems among the sample
of children studied (n = 189). Seventeen percent of all children were
identified as stunted, the majority of whom were 10 months of age or older.
Ten percent of the children were classified as wasted, the largest propor-
tion being between 10 and 18 months. Through all analyses, mean percents
weight for length, weight for age and length for age decline from the
first to the second year.
2) Women's working status has a strong impact on growth of children in
the first 23 months. Children of women who work at the higher paying jobs
i I
78
show better growth performance and a lower incidence of wasting and stunt-
ing. Children of women who work at the lowest wages (marchands) or who do
not work at all (dependents and housewives) have relatively poorer nutri-
tional status and proportionately more are malnourished.
3) The location of women's work affects growth through an income effect.
Location of work was not an indicator of job compatibility with child care.
Women who work outside of the home have fewer malnourished children and
growth faltering among them is less apparent.
4) The number of days worked by women outside of the home is positively
associated with a higher proportion of children identified as normal under
the Waterlow classification. This is a further effect of income.
5) Growth differences associated with working characteristics of mothers
depend upon the age of the child. Children of women earning higher wages
grow better in the parameters of length for age, weight for length and
weight for age in the second year (12-23 months). Children of the lowest
wage earners falter in weight for age in the first year as well as the
second. Working outside of the home favors better growth in height and
weight during the second year but is associated with lower weight for
age in the first.
6) During the first year of life time spent by the mother in child care
is more important for satisfactory growth than income. Children of women
who spent full days with them had significantly higher percents length for
age and weight for length than those whose mothers spent only up to 7 hours.
The difference disappears by year two. The first six months is indicated
(by comparison of Tables 7 and 14) as the period during the first year
79
when time is most critical for growth. Income assumes the predominant
role thereafter.
Mortality
1) Among all study children the cumulative risk of death in the first
year was 163 per 1000 live births. From 0-23 months the mortality risk
was 205 per 1000. The risk of death during the first three months.was
highest, as expected.
2) Mortality for the first year was significantly higher among children
of marchands (172) than children of dependents and housewives (129). It
can only be speculated that occupation had an effect on death, as weighted
values were not used in these estimates.
3) Children from homes with cement floors were subject to a significantly
lower risk of death from 0-11 months than children living in dirt floored
homes. The probabilities were 107 and 183, respectively. Risk to 23
months was also significantly different between the two groups. Again,
floor type cannot be considered the determinant of risk of death, but
suggests that there may be a causal association in the population.
4) Literacy of mothers was related to a higher 0-23 month mortality.
Mortality among children of literate mothers (228) was higher than that
for all children combined (205). Literate women may be more apt to feed
their children bottles.
Early introduction of bottles and supplementary foods is regarded as
a major cause of poor child health in this community. Eighty-eight percent
of infants 0-2 months old are introduced to bottle feeding. Milk is given
I I
80
to fewer children and in proportionately smaller amounts with increasing
age of the sample. By 18 months only 30 percent of children are receiv-
ing any milk at all. Cereals are fed in the first month and increasingly
replace milk as the main component of the child's diet. To the women of
Cite Simone, the message that "breast is best" has little practical
reality.
81
IX. RECOMMENDATIONS FOR FUTURE RESEARCH
It is important to apprehend more fully how women's responsibilities
in child care and household management effect nutrition and health of
children in the first two years of life. The results from the present
study suggest that a better understanding of women's constraints in the
low income household will influence the effectiveness of programs designed
to improve child nutritional status. A greater comprehension of the social
interactions among poor women in small communities will contribute to more
viable programs which impact on maternal behavior regarding young child
feeding. Further, programs may be constructed which improve child health
but which do not endanger the additional needs of other household members.
Finally, a fuller appreciation of the relationships between women's acti-
vities and child nutrition will guide policy makers in targeting their
programs to groups with the most immediate needs. Such research would
highlight the ecological conditions which provoke these needs.
The following areas of study are recommended to address the dynamics
of child health and women's work:
1) Investigation of the socioeconomic and demographic features of female-
headed households should be undertaken at the community, household and
individual levels. Wages and working conditions of women singly support-
ing a household should be examined in relation to time spent at work and in
child care. Family structure and age composition of such households should
be assessed. The contribution made by other household members to income
and child care must be considered in evaluating the economic needs of these
families.
82
2) Evaluation of the epidemiology of nutritional diseases as influenced
by women's work should be undertaken in low income communities. Housing
conditions, the availability of water, electricity, fuel, and sanita-
tion facilities should be surveyed in an effort to identify the severest
cases of malnutrition and infection. How the location of women's work
is related to health conditions inside the home will be an important
determinant of infant feeding'and health.
3) Research efforts should concentrate on the social interaction and
belief structure regarding infant care and feeding among small groups of
women in poor communities. The impact of peer pressure on patterns of
breast and bottle feeding should be carefully studied in order to deliver
realistic, effective programs aimed at feeding. If women are determined
to bottle feed their babies, instruction in the cleanest and most nutri-
tional method of bottle feeding is a more practical program of nutrition
education than one which chiefly encourages breast-feeding.
4) The influence of women's work on the use of local health facilities
needs tobe studied. Households should be disaggregated by union status
of women and socioeconomic level to determine which families derive greater
benefits from health clinics and nutrition rehabilitation centers. For
women working the longest hours and at the lowest wages, constraints of
time and money may preclude them from using well-equipped health services.
5) Finally, research needs to focus on working women's constraint s and on
ways to relieve these constraints. This requires an assessment of house-
hold resources and activities in relation to child care. Technical
innovations and simple household goods which would relieve the child care
J1
83
time constraint would improve child welfare in families where the woman
must work.
In addition to these areas of suggested research, some specific
recommendations are made to the government of Haiti to alleviate child
malnutrition in Cite Simone. These are as follows:
1) The widespread use of Argo in bottle feeds should be strongly dis-
couraged. Mothers should be taught in the proper methods of preparing
bottles, with an emphasis on sanitation. The association of diarrhea
with the feeding of dirty, diluted bottle mixes should be stressed in
this education.
2) Bottle feeding in the maternity clinics should be eliminated and
breast feeding of the newborn promoted. It is important to recognize that
women who deliver their babies at home are influenced by this practice
and imitate it in an effort to be contemporary.
3) Children of marchands are the most severely malnourished children.
They are a target group in need of immediate nutritional rehabilitation.
In cases where mothers cannot come to the nutrition center for follow-up
care and counselling, the service should be delivered to their homes.
Volunteer community collaborators could be trained for this purpose.
4) In households where women must work outside of the home, alternative
child care services in the first year of life would improve subsequent
growth performance.
.I i :
84
FOOTNOTES
For a more extensive review of this controversy, one is referred to
Underwood and Hofvander, 1980; Waterlow et al., 1980; Cooper, 1980;
Waterlow, 1981; Whitehead et al., 1980; Waterlow and Thompson, 1979.
2These community volunteers served as liasons between the health service
and the home. They assisted in weighing and measuring at the health
clinics, updated family registers, and were able to identify high risk
mothers. One liason was assigned to 500 persons.
3Argo is considered by the women of Cite Simone to be a very prestigious
item. It is used in bottle mixes with sugar and a small amount of
canned evaporated milk (less than 1 ounce). Home visitors from Foster
Parents Plan International have reported that women display empty Argo
boxes on their shelves to impress their peers (Berggren, personal
communication).
85
REFERENCES
Berggren, W., D. Ewbank and G. Berggren. 1981. Reduction of mortality
in rural Haiti through a primary-health-care program. New" England
J. Med. 304:1324-1330.
Berggren, W. and G. Berggren. 1973. Preliminary statistics from the
Harvard-Haiti Field Study. Cambridge: Department of Population
Sciences, Harvard University.
Bordes, A. and A. Couture. 1978. For the People, for a Change: Bringing
Health to the Families of Haiti. Boston: Beacon Press.
Boulos, C. and L. Boulos. 1980. Enquete sur le Statut Socio-Economique
des Menages de Boston. Complexe Mldico-Social, Cite Simone, unpublished.
Boulier, B. 1976. The influence of children on household economic acti-
vity in rural Philippines. Paper presented at the Agricultural Develop-
ment Council Service on Household Studies, Singapore, August, 1976.
Butz, W. and J. DaVanzo. 1978. Contracepting, breast-feeding and birth-
spacing in Malaysia: A model of decision making subject to economic
and biological constraints. Paper presented at conference on "Women
in Poverty: What Do We Know?", International Center for Research on
Women, Washington, D.C., May, 1978.
Caldwell, J. C. 1981. Maternal education as a factor in child mortality.
World Health Forum. 2(l):75-78.
Chen, L., S. Ahmed, M. Gesche and W. Mosley. 1974. A prospective study
of birth interval dynamics in rural Bangladesh. Pop. Studies. 28(2):
277-297.
Cooper, E. 1980. Faltering growth and human milk. The Lancet. December
20/27:1366.
Cunningham, A. 1981. Breast feeding and morbidity in industrialized
countries: An update. IN: D. B. Jelliffe and E. P. Jelliffe (editors),
Advances in International Maternal and Child Health, Vol. 1, pp. 128-167.
London: Oxford University Press.
Froozani, M. D., H. Malekafzali, and B. Bahrini. 1980. Growth of a group
of low income infants in the first year of life. J. Trop. Ped. 26(June):
96-98.
Gordon, J. E., J. B. Wyon and W. Ascoli. 1967. The second year death rate
in less developed countries. Am. J. Med. Sci. 254(3):357-380.
Gross, .A. and V. Clark. 1975. Survival Distributions: Reliability Appli-
cations in the Biomedical Sciences. New York: Wiley and Sons.
.1:1
Ii i
86
Haiti Nutrition Status Survey: 1978. 1979. Bureau of Nutrition, Depart-
ment of Public Health and Population, Republic of Haiti, in cooperation
with U. S. Department of Health, Sdudcation and Welfare, PHS/CDC, and
The Agency for International Development, Washington, D. C.
Hill, A. 1979. Mortality and Morbidity infBreast Fed vs. Artificially
Fed Infants, unpublished.
Jelliffe, D. 1976. World trends in infant feeding. Am. J. Clin. Nut.
29:1227-1237.
Jelliffe, D. B. 1966. The Assessment of the Nutritional Status of the
Community. Geneva: World Health Organization.
Jelliffe, D. B. and E. P. Jelliffe. 1971. The uniqueness of human milk.
Am. J. Clin. Nut. 24:968-1024.
Kumar, S. K. 1978. Role of the household economy in child nutrition at low
incomes, a case study in Kerala. Occasional Paper No. 95, Department of
Agriculture and Economics: Cornell Univ., Agriculture Experiment Station,
Ithaca, N.Y., December, 1978.
Levinson, F. J. 1974. Morinda: An economic analysis of malnutrition among
young children in rural India. Cambridge: Cornell-MIT International
Policy Series, 1974.
Malnutrition in Sudanese children as a function of socioeconomic change.
(editorial). 1980. Nutrition Reviews. 38:2.
Mata, L., J. Urrutia and A. Lechtig. 1971. Infection and nutrition of
children of a low socioeconomic rural community. Am. J. Clin. Nut.
24:249-259.
Moves to the cities: Nutritional causes-and consequences. (editorial).
1980. Nutrition Reviews. 38(11):367-369.
Mudambi, S. R. 1981. Breast feeding practices of mothers from mid-western
Nigeria. J. Trop. Ped. 27(2).
Nerlove, S. B. 1974. Women's workload and infant feeding practices: A
relationship with demographic implications. Ethnology. 13(a):207-214
Neumann, C. and M. Alpaugh. 1976. Birthweight doubling time, a fresh
look. Pediatrics. 57(4):469-473.
Oberndorfer, L. and W. Mejia. 1968. Statistical analysis of the duration
of breast-feeding. J. Trop. Ped. March:27-41.
Paul, A. A. and E. M. Miller. 1980. Seasonal variations in dietary intake
in pregnant and lactating women in a rural Gambian village. IN: Mater-
nal Nutrition During Pregnancy and Lactation: A Nestle Foundation Work-
shop Lutry/Lausanne April 26 and 27, 1979. H. Albi and R. Whitehead
(editors), pp. 105-116. Bern Stuttgart Vienna: Haus Huber Publishers.
iI
87
Paul, A., E. Muller and R. Whitehead. 1979. The quantitative effects of
maternal dietary energy intake on pregnancy and lactation in rural
Gambian women. Transactions of the Royal Society of Trop. Med. and
Hygiene. 73(7):686-691.
Popkin, B. 1976. The role of the rural Filipino mother in the determina-
tion of child care and breast feeding behavior. Institute of Economic
Development and Research Discussion Paper No. 76-12, U. of Philippines.
Popkin, B. 1978. Economic aspects of breast-feeding behavior: The case
of rural housewives in Laguna, Philippines. IN: W. H. Mosley (editor),
Nutrition and Human Reproduction, pp. 461-497. New York: Plenum Press.
Popkin, B. 1980. Time allocation of the mother and child nutrition. Ecol.
Food Nutr. 9:1-14.
Popkin, B. M., R. E. Bilsborrow, M. E. Yamamoto and J. Atkin. 1979. Breast
feeding practices in low income countries: Patterns and determinants.
Carolina Pop. Center Papers, No. 11, Chapel Hill, North Carolina,
October.
Popkin, B. M. and F. Solon. 1976. Income, time, the working mother and
child nutriture. Env. Child Health, August.
Rowland, M. G. 1980. Weaning and diarrhoea in the Gambia. Br. Med. J.
280(6222) :119.
Rowland, M. G., T. J. Cole and R. G. Whitehead. 1977. A quantitative
study into the role of infection in determining nutritional status in
Gambian village children. Br. J. Nutr. 37:441-450.
Rubin, V. and R. Schaedel. 1975. The Haitian Potential. New York:
Teachers College Press.
Scrimshaw, N. 1970. Synergism of malnutrition and infection. J. A. M. A.
212 (10):1685-1692.
Scrimshaw, N. S., J. E. Gordon and C. E. Taylor. 1968. The interaction of
nutrition and infection. WHO Monograph Series No. 57. Geneva: World
Health Organization.
Thompson, B. A. 1967. Infant feeding and child care in a West African
Village. J. Trop. Ped. Sept.0:124-138.
Thomson, A. M. and A. E. Black. 1975. Nutritional aspects of human lacta-
tion. Bull. World Health Org. 52:163-176.
Thomson, A. M., G. H. Beaton, J. P. Habicht and B. A. Underwood. 1979.
Nutrition during pregnancy: A review. University of Newcastle upon
Tyne, England, September, unpublished.
.1  .
88
Underwood, B. and Y. Hofvander. 1980. Appropriate timing of complementary
food for the breast-fed" infant: A background paper. Committee Report
to Consultative Group on Infant and Young Child Feeding of the Sub-
Committee on Nutrition, unpublished.
Underwood, B., H. VanArsdell, E. Blumensteil, and N. Scrimshaw. 1981.
Implications of available information on breast-feeding worldwide. IN:
J. T. Bond, L. J. Filer; G. A. Leville, A. Thomson and W. B. Weil (edi-
tors), Infant and Child Feeding, pp. 77-96. New York: Academic Press.
Waterlow, J. C. 1981. Faltering growth and human milk. The Lancet. Jan.
10:100-101.
Waterlow, J. C., A. Ashworth-Hill and M. Griffiths. 1980. Faltering in
infant growth in less developed countries. Dept. of Nutrition, London
School of Hygiene and Tropical Medicine." WClE7HT, September, unpublished.
Waterlow, J. C. and F. L. Rutishauser. 1974. Malnutrition in man. IN:
J. Cravioto et al. (editors), Early Nutrition and Mental Development.
Symposium of the Swedish Nutrition Foundation, No. 12:13-26. Sweden:
Aimquist and Wiksell.
Waterlow, J. C. and A. M. Thomson. 1979. Observations on the adequacy of
breast feeding. Lancet. 2:238-242.
Weil, T. et al. 1973. Area Handbook for Haiti. 180 pp. Washington, D.C.:
The American University.
Whitehead, R. G. 1979(a). Infant feeding practices and the development
of malnutrition in rural Gambia. Food and Nutrition Bulletin. 1(4):
36-41.
Whitehead, R. G. 1979(b). Nutrition and lactation. Postgraduate Med. J.
55:303-310.
Whitehead, R. G., A. A. Paul and M. G. M. Rowland. 1980. Lactation in
Cambridge and in the Gambia. IN: B. Wharton (editor), Topics in Pedia-
trics 2. Nutrition in Childhood, pp. 22-23, Turnbridge Wells: Putnam
Medical.
Whitehead, R. G., M. Hutton, E. Miller, M. Rowland, A. Prentice and A.
Paul. 1978. 'Factors influencing lactation performance in rural Gambian
mothers. The Lancet. 2:178-181.
Woodbury, R. 1922. The relation between breast and artificial feeding
and infant mortality. Am. J. Hygiene. 2:668.
World Health Organization. 1977. Manual of Mortality Analysis. Geneva:
World Health Organization.
89
World Health Organization. 1979. WHO Collaborative Study of Breast-feeding.
Methods and Main Results of the First Phase of the Study. Preliminary
Rep. Maternal Child Health (MCH)/79.3.
Wray, J. D. 1978. Maternal nutrition, breast-feeding and infant survival.
IN: W. H. Mosley (editor), Nutrition and Human Reproduction. New York:
Plenum Press.
90
Appendix A
Table 20 (page 65) shows the life table constructed for the cohort
of children of literate mothers. The age interval (x) is taken as the
number of completed calendar months. C is the number of children who
die during interval x. The number of children lost or still surviving
to the study cut-off date is labelled TX. W is the sum of C and T,
or the total number of observations terminating during the interval.
The number of children exposed to the risk of death at age x (Z ) is
calculated by summing W + (W + 1) + (W + 2) + ... (W + [18 - 22]).
The number of children at risk of dying during interval x(Z ') is taken
as Z -IT. The probability of dying during interval x -- (q ) -- is
simply C, +Z '. The probability of surviving during interval x (p --
is the complement of q , or 1 - q . The probability of remaining alive
to interval x -- (Ps) -- is calculated by multiplying P by p 1 , (Chen
et al., 1974). The final column (1 -P) is the complement of P-, or theX x
probability of dying up to the beginning of interval x.
91
Appendix B
Sample Calculation of Standard Error of Survival to Time t
and Test for Significance Between Two Samples
General Formula (uses unweighted values)
SE(S )
t
- 2C -T
= SEq/[Z -T /2][Z - 20-Kx Xt Zaqx zx . Tx /] 2
=AS
t
2C -T
aq /( - x
2
C
St T x x .
x
2
where: C = observed deaths
Tx= censored observations (survived or withdrawn at study cut-of f)
= group size (# entering interval)
so, Z =Zx -C - T
(1) Standard error of survival to 23 months: cohort of children of liter-
ate mothers.
= .7717St
^2(4)-10 2(6)-14SE (S ) = (.7717) [.0331/(126- 2 1+ [.0571/(112t 2. (contd.)
[.0359/(92 - 2(3)-7)I + [.0328/(72.- 2(2)-22 + [.0923/(48 2-(3)-14
= (.7717) Y.00026063+ .00050531+ .00036821+ .00040494+ .001775
= .04771
(2) Standard error of survival to 23 months; cohort of children of
illiterate mothers.
S=t ()Q = .8279
Ii
92
SE (S ) = (.8279)/ [.0683/(285- 2 )]+[0337/(252 - 2)] + (contd.)t 2
[.0306/(215-2(638)]+[.0136/(171 - 2(2)47)]+].O382/(1222(3)-872 2 2
= (.8279) /.00025018+ .0013037 + .00013421+ .00007065+ .0023508
- .02451
(3) One sided T
survival to
mothers.
test for significance of difference in probability of
23 months, between children of literate and illiterate
.2283 - .1721
T 1.05
(.0477)2+ (.0245) 2
p = .15