Content uploaded by Teresita Gonzalez de Cosio
Author content
All content in this area was uploaded by Teresita Gonzalez de Cosio on Jun 16, 2017
Content may be subject to copyright.
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
30:195-220
(1987)
Maternal Nutrition and Birth Weight
REYNALDO
MARTORELL
AND
TERESA
GONZALEZ-COSSIO
Food
Research Institute, Stanford University, Stanford, California
94305
(R.
M.);
Institute of Nutrition of Central America and Panama (INCAfl,
Guatemala City, Guatemala (TG.-C.)
KEY
WORDS
Nutrition Growth, Child development, Child mortality, Birth weight,
ABSTRACT
This
is
a review of short- and long-term consequences of low
birth weight (LBW) and of some of its determinants. The results of observa-
tional studies and of nutritional experiments aimed at clarifying the relation-
ship between maternal nutrition and birth weight are reviewed. Areas for
future research are noted.
LBW
is
a major public health problem in developing countries. LBW
is
strongly associated with infant, and especially neonatal, mortality, and there
is
some indication that
it
is also related to preschool mortality rates
(1-4
years). Long-term measures of physical growth, immune response, and men-
tal development are all adversely affected by LBW. The duration and magni-
tude of these effects appear to be a function of the type and severity of the
weight deficit
at
birth.
Determinants
of
LBW in developing countries include maternal nutritional
status before and during pregnancy, maternal diet, and perhaps exposure
to
smoke and infection rates during pregnancy. Studies from around the world
have demonstrated that dietary intervention during pregnancy
is
almost
always associated with increased birth weights.
It
is now known that nutri-
tional supplementation interacts with maternal nutritional status; newborns
of well-nourished women benefit little, if any, from dietary supplementation
of their mothers during pregnancy, whereas newborns of poorly nourished
women (thinnest and shortest) show substantial improvements in birth
weight. However, when best to begin dietary supplementation during preg-
nancy remains an unresolved issue.
Many areas are deserving of further research. Long-term studies of the
significance of birth weight, particularly in poor countries, are needed. Ex-
periences with dietary interventions during pregnancy in areas where mean
birth weights are very low (i.e., at
or
below
2,800
g) need to be documented.
Advances need to be made in the design of locally appropriate dietary inter-
ventions
€or
pregnant women and of risk scales capable of identifying those
women who are more likely
to
benefit from these programs. The influence
of
a number of factors on birth weight needs to be clarified intrauterine infec-
tion, maternal physical activity, lactation during pregnancy, exposure to
cooking smoke, and adolescent pregnancy. Finally, the impact of nonnutri-
tional interventions on birth weight needs to be assessed. These include, for
example, family planning, medical care, health and nutrition education, and
socioeconomic development programs.
The World Health Organization
(WHO)
has recently made an attempt to estimate
the incidence of low birth weight
(1980).
After reviewing the available evidence, the
Q
1987
Alan R.
Liss,
Inc.
196
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
vol.
30,
1987
authors of the review emphatically noted
that
there was a Severe “shortage of
adequate data, especially in the developing world” (p. 198). There were many
countries for which no data were available. Where data existed, problems such
as
small sample sizes, lack of representativeness of the sample, and poor quality of the
data severely limited their usefulness. The authors pointed out that by emphasizing
the lack
of
information they hoped
to
stimulate more data collection and draw
attention
to
the effort by
WHO
in sponsoring research on birth weight.
This review focuses on the relationship between maternal nutrition and birth
weight. The first part
is
a discussion of the significance of low birth weight and
specifically deals with estimates of the worldwide incidence
of
low birth weight, the
relationship between birth weight and mortality, and the growth and development
of low-birth-weight babies. The literature on the determinants
of
birth weight
is
extensive and no attempt
is
made here to review this subject thoroughly. Rather,
the second part
of
this review offers
a
brief summary of the principal fadors known
to affect birth weight. The extent
to
which nutritional problems account for lower
birth weights is the key subject reviewed in the third section. Various sources
of
evidence are considered wars and famines, observational studies, and intervention
trials. Finally, the review concludes with a summary of what is known and a
discussion of research needs.
SIGNIFICANCE
OF
BIRTH
WEIGHT
Numerous studies have shown that a greater percentage of newborns in develop-
ing countries weigh
less
than
2,500
g
at
birth than in developed countries
(WHO,
1980). These babies are more likely
to
die than heavier babies and, if they survive,
are also more likely
to
suffer short- and long-term sequelae (Taffel, 1980; McCormick,
1985).
Incidence of
low
birth weight
There
is
still some confusion regarding terminology. Current usage is shown in
Figure
1.
Whereas prematurity was
at
one time defined as
a
birth weight of
2,500
g
or less, today
the
term “prematurity” means
a
gestational age less than 37 weeks
and “low birth weight” means
a
weight less than
2,500
g regardless of gestational
age
(WHO,
1980; Bergner and Susser, 1970). Term babies (i.e., born at 37 weeks
of gestation) with
low
birth weight
(<
2,500
g)
are
said to have experienced “intra-
4.0
3.5
;;
3.0
g
2.5
2.0
f
1.5
1.0
0.5
0
MEAN
BIRTH
WEIGHT
Y
v
a
-
L
+
”LOW
B!9’H
WEIGHT”
28
30
32
34
36
38
40
42
44
WEtKS
GESTATION
AGE
Fig.
1.
The
distribution
of
birth weight by gestational age in Sweden, 1958-1959
(from
Pettersson
et
al.,
1978).
Martorell and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
197
uterine growth retardation
(IUGR).”
Related terms, “small for gestational age
(SGAY’
and “small for dates” are used
to
denote preterm
or
full-term newborns with
weights significantly less than the expected mean for gestational age, the exact
criterion varying in different studies. Synonyms for
SGA
and
IUGR
include fetal
undernutrition and small for dates (Westwood et al., 1983).
The use of a birth weight cutoff point, such as less than 2,500 g, to assess fetal
undernutrition has limitations.
For
instance,
it
fails
to
distinguish prematurity from
intrauterine growth retardation since both types of newborns are found below the
2,500 g limit (Fig.
1).
The use of this cutoff point also misclassifies as “normal”
a
small percentage of newborns whose birth weights are above 2,500 g but who are
nonetheless growth retarded because their weights are below 2 standard deviations
from the mean weight for gestational age.
In
Figure 1, this occurs from gestational
ages of 40 weeks
or
greater. Thus,
it
is best
to
use both birth weight and gestational
age in classifications of newborn status.
As
noted above,
WHO
recently carried out a comprehensive review of the litera-
ture on the incidence of low birth weight in the world
(WHO,
1980). Table
l
shows
the incidence estimates for various regions of the world for 1979. Some
20.6
million
low-birth-weight babies are estimated to have been born in the world in 1979 and
19.3 million
of
these,
or
94%, were from developing regions. The incidence of low
birth weight was estimated to average
18%
in
the poorer countries and 7% in
industrialized nations. These figures are in close agreement with estimates previ-
ously made for 1975. Lechtig et al. (1977) estimated a world incidence of 22 million,
with 19.1 and 7.4% of low birth weight in developing and developed countries,
respectively. Petros-Barvazian and BBhar (1978) estimated a world incidence of
17%
(23.4 million).
Belizan et al. (1978) and Villar and Belizan (1982) point out that a higher propor-
tion of low-birth-weight babies in developed countries are premature.
For
example,
55%
of all low-birth-weight babies born in the
U.S.
in 1979 (Taffel, 1980) were
premature compared to only
18%
of low-birth-weight cases in
a
Guatemalan Indian
TABLE
1.
Estimates
of
the incidence
of
low birth weight in the world in 1979l
Live births Low-birth-weight infants
Region (thousands)
(%)
(thousands)
Africa 21,081 15 3,165
North Africa 4,540 13 592
Western Africa 6,713 17 1,152
Eastern Africa 6,240 14 874
Middle Africa 2,340 15 362
Southern Africa 1,248 15 185
Northern America 3,684 7 269
Latin America 12,410 11 1,392
Middle America 3,649 15 566
Caribbean 870 13 117
Tropical South America 6,948 9 631
Temperate South America 943 8 78
Asia 72,940 20 14,785
Western South Asia 3,840 16 611
Middle South Asia 33,744
31
10,470
Eastern South Asia 12,456 18 2,259
East Asia 22,900 6 1,445
Europe 7,006 8 536
Northern Europe 984
6
61
Western Europe 1,836 6 116
Eastern Europe 1,962 8 159
Oceania 506 12 62
Union of Soviet Socialist
Republics 4,752 8 380
World 122,300 17 20,600
Developed countries 17,400 7 1,290
‘Data
from
WHO
(1980).
Southern Europe 2,224 9 200
198
YEARBOOK
OF
PHYSICAL
ANTHROPOLOGY
[Vol.
30,
1987
community (Mata, 1978). Villar and Belizan (1982) also make the observation that
the proportion of all newborns who
are
both premature and of low birth weight
is
about 6.7% in developing countries and about 3.3%
in
the
developed countries. Term
babies with low birth weight, on the other hand, make up 17% of all cases in data
from developing countries compared
to
only 2.6% of cases in developed countries.
The
proportion of newborns who
are
premature varies little from country
to
country
in the Third World and
bears
no relationship to
the
total
incidence of low birth
weight. Thus, one observes substantial intercountry variation in the prevalence of
low birth weight and mean birth weight but not in mean gestational age Crable 2).
Accordingly, Villar and Belizan (1982) argue that variation in the incidence of low
birth weight in developing nations
is
largely
a
function of the frequency of term low-
birth-weight babies (i.e.,
mGR
infants).
Birth weight distributions
are
shown in Table
3
for selected populations. Ethnic
differences in the incidence
of
low birth weight in the
US.
are large (Shiono
et
al.,
1986; Williams
et
al., 1986; TafYel, 1980). For instance, Table
3
shows that 6% of
newborns have low birth weights in the white population compared to 12.1% in the
nonwhite population (Tdel, 1980).
The
data for Nairobi, Kenya, are a fair represen-
tation of the average distribution of birth weights in developing countries with
17.5% below 2,500 g; the data for India illustrate the large distributional differences
to
be found between socioeconomic classes in developing countries. The high preva-
lence of low birth weight in the Guatemalan sample (41.6%) and the low-income
Indian sample (52.3%) are illustrative of the very high prevalence to be found in the
very poor in middle-income countries and in many of the very low-income countries,
such
as
India and Bangladesh.
TABLE
2.
Low
birth weight and gestational age in different countries'
Ireland
5.9 3.52 40.5'
Poland
6.8 3.4 40.1
Venezuela
7.9 3.2 39.7
Greece
9.5 3.3 39.8
Taiwan
11.3 3.1 40.0
Japan
11.3 3.0 40.2
Philippines
14.2 2.9 39.6
Malaysia
17.3 3.0 40.2
India
29.0 2.8 40.6
'Data
from
Rooth
(1980).
'Median.
Iran
13.2 3.0 39.4
TABLE
3.
Birth weight distribution in selected societies
-
U.S. 1976'
Hyderabad, India3
white races Kenya"
Guatemalan
Birth All Other Nairobi Low-middle Low Indians
weight
(g)
income (n
=
43014
-
_____
income
-
--___
1.16
6.51
16.7 39.6 33.95
<
1,550 0.9 2.2
1,501
-2,000 1.2 2.4
2.001-2,500 4.0 7.5 8.7
2,501-3,000 15.5 24.3 29.5 39.4 31.7 48.14
3,001-3,500 37.3 38.7 38.4 30.1 11.6 10.00
>
3,500 41.0 24.8 14.6 7.8 3.4 0.23
Prevalence of
6.1 12.1 17.5 22.7 52.3 41.62
6.0 12.7
low birth
weight
(i
2,500)
'Taffel(1980).
:Meme (1978).
.'Data given in Petros-Rarvazian and Behar (1978).
'Mata (1978).
Martorell and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
20
-
e
I
10
-
199
LAW
birth weight and mortality
Only about 7% of U.S. babies have low birth weights, yet this group accounts for
more than half of
all
infant deaths and nearly three-quarters of all neonatal deaths
(Taffel, 1980). Lee
et
al.
(1980) found that the neonatal mortality of populations in
developed countries can be quite accurately predicted by the very low-birth-weight
rate (i.e., live births with birth weights of
1,500
g
or
less per
1,000
live births-
VLBW). They demonstrated this fact for three types of populations: a New York
hospital over a 10-year period, the states of the
U.S.
in 1975, and for 13 industrialized
nations. Figure 2 shows the data
for
the
13
countries for which birth weight
distributions were published in the
United Nations Demographic Yearbook
(UN,
1976). The correlation between the neonatal mortality rate and the VLBW rate
is
0.88 if Hungary is included and
0.75
if
it
is
not. The slope of the relationship
is
around 1; that is, modality increases by one death per 1,000 newborns for every
increase of
1%
in the prevalence of VLBW.
The neonatal mortality rate
mMR)
in the U.S. has been declining since the 1960s,
and
a
number of studies have been carried out to determine the extent
of
the decline
to
be attributed
to
changes in the birth weight distribution
or
to
improved survival
among low-birth-weight babies (Hack et al., 1979). Williams and Chen (1982) attrib-
uted 15% of the decline in
NMR
in California from 1960 to 1977 to changes in the
birth weight distribution. Kleinmen et al. (1978) reported that such shifts accounted
for
21% of the decline in NMR in six U.S. states and David and Siege1 (1983) gave
a
figure of
19%
for the state of North Carolina. Goldenberg et al. (1983) found that
NMR
declined by
50%
from 1970 to
1980
in Alabama but that shifts in the birth
weight distribution could only account for 12% of the decline in
NMR
in whites and
for none in the case of nonwhites. These studies indicate that the decline in
NMR
in
the U.S. over the
last
decade cannot be explained satisfactorily by shifts in birth
weight distributions, but rather by improvements in obstetric and neonatal care.'
If
medical care improvements have been the principal explanatory factors, then only
30
HUNGARY
GFR.
CZECH. POLAND
GDR
.
.CANADA
NEW ZEALAND
DENMARK
ONORWAY
o/
5
10
15
20
VLBW RATE
Fig.
2.
Relationship between neonatal mortality rates
(NMR)
and verylow-birth-weight
(VLBW)
rates
for 13 countries from 1972 to 1974;
NMR
is defined as deaths per 1,000 live births and
VLBW
as number
of
live births with weight
<
1,500
g per
1,000
live births (from Lee et
al.,
1980).
'The salient technological advances in obstetric and neonatal care have been reviewed by Goldenberg et al.
(1983).
The new advances have been particularly effective in improving the life expectancy
of
premature newborns.
200
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
[Vol.
30,
1987
a
modest role could
be
attributed to the effects on birth weight on improvements in
the standard of living and in nutritional status through programs such as WIC
(Women, Infants and Children program) (Goldenberg
et
al., 1983).
On the other hand, differences in birth weight distributions appear to provide an
adequate explanation for differences in
NMR
between Norway and the
US.
New-
borns in Norway
are
heavier than in the
US.
and
NMR
is higher in the
US.
However,
NMR
is similar within birth weight categories (Wallace
et
al., 1982;
Erickson and Bjerkedal, 1982).
In developing countries, the availability of sophisticated neonatal care of the low-
birth-weight baby is very limited. This technology is expensive and not likely to be
widely available for many years. Research to date suggests that efforts to lower the
NMR
should rely instead on improving birth weights through adequate prenatal
care, emphasizing improvements in the nutritional status of the mother, and preven-
tive medicine (e.g., tetanus vaccine).
The relationship between birth weight and neonatal mortality for various popula-
tions is examined in closer detail in Table
4.
For all samples, neonatal mortality
rates
are several times higher for low-birth-weight newborns, particularly for those
below 1,500 g.
The
four samples from the
US.
demonstrate the improvements in
NMR
that have taken place in recent years. Birth-weight-specific NMR
is
higher in
the samples from Ribeirao Preto (Brazil) and Nairobi (Kenya) than in the data from
the
US.
Neonatal death rates will thus be higher in developing countries
as a
result
both
of
higher birth-weight-specific
NMR
and of greater prevalence of low-birth-
weight babies.
The data in Table
4
from North Carolina and Alabama in the
US.
show that birth-
weight-specific mortality
rates
for non-whites
are
lower than for whites at low birth
weights (i.e., <2,500 8). This
is
puzzling because the quality of prenatal care is
likely
to
be better for whites than for nonwhites. One possible explanation is that
a
greater proportion of
the
low-birth-weight babies in the nonwhite group are IUGR
and fewer
are
premature than is the case in the white group. Since the mortality
prognosis
is
better for IUGR newborns than for prematures (Starfield et al., 1982),
the overall effect is for
less
overall mortality in nonwhites in the lower birth weight
categories, overriding any effects of differential prenatal care.
Support for the above conclusion
is
provided in Figure
3,
where the natural
logarithms of
the
relative risk
of
neonatal mortality of India and the
US.
are
compared at various birth weight categories. The reference mortality experience for
the calculation of relative risk in this case is for the group of newborns weighing
2,501-3,000 g. Below 2,500 g, relative risks are higher in the
US.
than in India for
the reasons given above. This trend reverses
at
birth weights above 3,000
g
as
factors such as perinatal health care and vaccination coverage become the major
determinants
of
mortality.
In spite of the small sample of newborns (n
=
4301, the cohort study carried out by
Mata (1978) in Santa Maria Cauque, Guatemala, offers overwhelming evidence of
the association between birth weight and mortality rates not only during the neo-
natal but in the postneonatal period as well. These data are shown in Table 5. Low-
birth-weight babies accounted for
88%
of
neonatal deaths (14/16), 60% of postneona-
tal
deaths (15/25), and overall for 71% of infant deaths (29/41). Infant mortality rates
were 516 per 1,000 live births if both low birth weight and prematurity occurred (161
311,
84
if only low birth weight was present (12/143), and 50 if the babies were of
normal birth weight and gestational age (12/242). Prematurity was more predictive
of death than low birth weight when <37 weeks and <2,500 g were used as
respective cutoff points.
Studies in Nigeria (Morley, 1973) replicate the Guatemalan findings with regard
to the relationship between birth weight and neonatal and postneonatal mortality
rates
as
shown in Table 6. Studies of survival after the
first
year of life
as
a function
of birth weight
are
few. Rantakallio (1985) reports data from
a
14-year followup of
12,000 Finnish children of different birth weight categories.
The
results show that
those born with birth weights 2 standard deviations below the mean had signifi-
TABLE
4.
Birth-weight-specific neonatal mortality rate
(NMR)
in selected populations
Ribeirao
North Carolina3 Alabama4
Preto
Nairobi
1977 1980
project' Kenya5'
Birth California
weight
U.S.
project'
(9)
19601 1969-1970
White Non-white White Non-white
1968-1970 1975
<
1,000
912.8 940.0 803.1 696.2 577.8 544.0 890.6 852.9
1,001-1,500 521.5 434.8 281.4 187.5 223.6 111.6 696.3 638.3
1,501-2,000 180.6 106.9 67.7 53.9 55.8 27.1 391.0 166.7
2,001-2,500 41.4 28.2 18.8 11.2 12.3
11.1
68.1 29.0
2,501-3,000 9.9 5.4 4.9 5.3 16.8 8.5
3,501-4,OOO 3.6 2.2 2.1 1.5 6.0 8.6
>
4,000
5.0
3.1 2.0,2.6 4.3,4.8 3.2 4.3 8.7
<
2,500 171.6 129.9
- -
71.4 63.8 215.7 261.6
3,001-3,500 4.7 2.5 1.8 2.2
3.0
3.4
8.5 9.8
>
2,500
5.5
3.1
-
-
3.0
3.5 10.0 9.1
Total
18.4 12.7 8
approx
15
approx
6.8 10.6 28.2 53.3
"Chase (1972).
'Puffer and Serrano (1975).
3David and Siege1
(1983).
NMR given for birth weights over
4,000
g are for
4,001-4,500
and
4,501-5,000
g categories respectively. Total
NMR
values
for
Northern Carolina were
read from graphs given
by
the authors.
4Goldenberg et aL(1983).
5Meme (1978).
TABLE
5.
Infant
deaths
by birth weight in
a
Guatemalan Indian population,
1964-1973l
Neonatal period Postneonatal period
0-28
days
29
days-11 months Infant mortality
Number
born
Birth weight (g) alive Deaths Rate Deaths Rate2 Deaths Rate
<
1,501
1,501-2,000
2,001-2,500
2,501-3,000
>
3,000
5
28
146
207
44
600
214
34
10
0
500
409
35
44
23
4
15
10
11
1
800
536
68
53
23
Total
430 16 37 25 60 41 95
'Data
from
Mata
(1978).
'Denominator is number born alive minus neonatal deaths
202
1-
0
-1
-
-2
-
-3
-
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
--
I
Ill
II
I1
[Vol.
30,
1987
Fig.
3.
Log
of
relative
risk
(RR)
of
neonatal mortality by birth weight (data from Ashworth and Feachen,
1985).
TABLE
6.
Survival
by birth weight
in
Imesi, Nigeria’
Neonatal Postneonatal
mortality2
______~__
Birth weight
(g)
mortality’
__
<
1,000
1,000-
1,500-
2,000-
2,250-
>
2.500
750
200
21
1
49
13.7
15.2
1,000
444
286
103
111
38
’From
data
in
Morley
(1973).
‘Deaths
per
1,000
total
births
cantly higher mortality rates
(P
<
.05) than children
of
average birth weight (25th-
75th percentiles). Conversely, children born with birth weights exceeding 2 standard
deviations above the mean had significantly lower mortality rates
(P
<
.05) than
children
of
average birth weight.
Growth.
and
development
of
lowbirth-weight babies
Recent advances in obstetrics and neonatology have made
it
possible for many
VLBW
babies to survive (Hoskins et al., 1983; Rothberg et al., 1983; Bennett
et
al.,
1983; Gross et al., 1983). Concerns have been raised that this feat will lead
to
an
increase in the “number of permanently handicapped and damaged children” (Hack
et al., 1979). Another issue
is
the enormous financial cost and manpower needed to
sustain the life of these small and immature newborns (Hoskins et al., 1983; Hack
et al., 1979). There is also an emotional cost brought on by the prolonged mother-
infant separation (Hoskins et al., 1983). In spite of a decreased incidence
of
long-
term sequelae, very low weight
at
birth (e.g., <1,000 g)
is
still
associated with
significant neurologic and sensory handicaps (Nickel et al., 1982). This has led to as-
yet-unresolved questions
as
to
the “reasonable and ethical” birth weight cutoff above
which intensive care efforts would be justified (Bennett et al., 1982).
A
figure of 800
g had been proposed (Britton et al., 1981) but some argue that newborns weighing
less than this amount do have
a
“hopeful outcome” (Bennett et al., 1983; Driscoll
et
al., 1982).
The sequelae
of
low-birth-weight survivors are many and include mental retarda-
tion and other physical, immunological, and neurological impairments (Taffel, 1980;
Martorell
and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
203
Ferguson,
1978;
McCormick,
1985).
According
to
Westwood et al.
(19831,
previous
follow-up studies of survivors contain %erious methodological defects” in that they
did not differentiate between premature and small for gestational age (SGA) new-
borns and did not control
for
effects
of
intrauterine infections, neonatal asphyxia,
and other neonatal complications. Term-SGA newborns are generally thought to
have higher survival probabilities than prematures but a higher incidence
of
serious
growth and development sequelae. Long-term follow-up studies
of
SGA babies have
shown that these children rarely catch up in physical growth (Westwood et al.,
1983;
Garn et al.,
1977).
In a follow-up through adolescence, Westwood et al.
(1983)
showed
that
SGA
newborns had normal maturational development (e.g., timing
of
puberty
events, bone age) but much smaller stature. Mata
(1978)
has found that growth
during the
first
36
months of life is strikingly related to weight at birth. He
examined growth in height, weight, head circumference, and other measures for
children with birth weights within the following categories:
2,000, 2,001-2,500,
2,501-3,000
and over
3,000
g. Children whose weights
at
birth were larger had
consistently better growth at all ages and
for
all measures.
In assessing the postnatal growth of SGA infants,
it
is
useful
to
distinguish
between chronic and acute intrauterine undernutrition.
If
the nutritional insult
begins early in gestation, newborns will be born shorter
for
gestational age but with
an adequate weight for length (i.e., adequate ponderal index, API, as measured by
weight/height3) (Villar and Belizan,
1982).
Newborn length is arrested because the
peak velocity
of
intrauterine growth in length occurs during the first half of preg-
nancy (Tanner,
1978).
If,
on the other hand, the nutrient deprivation occurs late in
gestation, when the peak velocity
of
intrauterine growth in weight occurs, the
neonate may have adequate length but will be low weight for height (i.e., low
ponderal index, LPD. In studies carried out in Mexico and Bolivia, Haas et al.
(1987)
found that intrauterine growth retarded-adequate ponderal index [IUGR-API] new-
borns had nearly twice the mortality (i.e., deaths during
first
2
days)
of
the full-
term, appropriate-weight infants whereas the IUGR-LPI infants had
2.9-5.7
times
the mortality of the full-term, appropriate-weight infants. IUGR-LPI and IUGR-API
will have different growth patterns, as has been shown in the Guatemalan group
of
neonates followed up to the age of
3
years (Villar et al.,
1984).
IUGR-LPI neonates
showed catch-up growth in weight, length, triceps skinfold, and head circumference
early in life. Length was similar to that
of
normal-birth-weight infants throughout
the study period. Head circumference of IUGR-LPI neonates was normal up to the
age of
18
months, when growth rates decreased to levels similar to those observed
in IUGR-API neonates, and remained low until the end of the study period. The
IUGR-API newborns, on the other hand, did not show evidence
of
catch-up growth
and remained shorter and lighter than the normal-term newborns (Villar et al.,
In terms of cognitive performances, studies have shown that IUGR infants have
poorer achievements and a very marked incidence
of
academic difficulties at school
(Westwood et al.,
1983).
In this area, it is apparent that the sequelae of IUGR depend
also on the timing
of
the nutritional insult. Harvey et al.
(1982)
followed, to about
5
years
of
age, a group
of
SGA newborns whose head growth in utero had been
monitored by ultrasound. Those whose growth in head circumference was poor prior
to
26
weeks of fetal age (i.e.,
MI)
had considerably poorer cognitive performances
than those whose head growth faltered later in gestation (or LPD.
However, the long-term association between IUGR and cognitive development
should be analyzed, controlling for factors known to be associated with both aspects,
in order
to
test whether delays in cognitive development are explained by factors
other than IUGR. Socioeconomic level
is
the most important confounding variable
in this relationship, and where
it
has been taken into account, IUGR-LPI and normal
infants are observed to be similar in mental development, whereas IUGR-API are
retarded as shown by tests given as early as
24
months of age (Villar et al.,
1984).
Thus, IUGR-API newborns have poorer postnatal growth and development than
1984).
IUGR-LPI.
204
YEARBOOK
OF
PHYSICAL
ANTHROPOLOGY
[Vol.
30,1987
Regarding immune response, research has shown that in children with
IUGR,
decreased cellular immunity may persist for months or even years (Ferguson,
1978;
Chandra,
1975).
SGA babies may therefore have increased susceptibility to infec-
tions. Many children who develop marasmus begin
as
low-birth-weight babies CMor-
ley,
1973)
and Severe undernutrition is associated with decreased immunocompetence
and greater incidence of infection.
LAW
birth weight across generations
Some recent studies have found that poor intrauterine growth and development
influences
later
reproductive performance in female infants. Hackman et al.
(1983)
studied
a
cohort of
748
white women in Seattle for whom maternal birth weight (as
recorded on birth certificates) was known. Maternal birth weight was significantly
related
to
stature, prepregnancy weight, pregnancy weight gain, baby’s birth weight,
gestational duration, the baby’s need for neonatal intensive care, transient tachy-
pnea of the newborn, and idiopathic respiratory distress syndrome. The relationship
between maternal birth weight and infant birth weight was confirmed by Klebanoff
et al.
(1984)
using the Buffalo cohort of the Collaborative Perinatal Project (n=
1348).
Compared with infants of mothers who had weighed
8
pounds or more at birth,
infants of mothers who had weighed
6-7.9
pounds,
4-5.9
pounds, and less than
4
pounds at birth were
99
g,
244
g, and
170
g lighter, respectively. These two studies
are in agreement with other studies which indicate
a
relationship between condi-
tions in the mother’s childhood
to
her later bearing low-birth-weight infants or
infants with neural tube defects (Hackman et al.,
1983).
Environmental
as
well
as
genetic factors probably explain these relationships but little can be said about their
relative importance based on the studies to date. Hackman
et
al.
(1983)
have
proposed that “reduced birth weight may interfere with the growth and development
of one or more organ systems, including the reproductive andor endocrine systems”
(p.
2019).
The transgenerational effeds found by Hackman et al.
(1983)
and Kleban-
off et al.
(1984)
may provide an explanation for the fact that improved standards of
living and better health and nutrition in modern societies did not lead to current
levels of adult stature in one single generation; rather, secular trends were extended
over several generations.
It
may be that several generations are required
to
“wash
out” deleterious effects linked to poor intrauterine development.
DETERMINANTS
OF
LOW BIRTH WEIGHT
Numerous factors have been found to be associated with the prevalence of low
birth weight (Rasmussen et al.,
1985).
These include environmental characteristics;
medical complications of pregnancy; adverse maternal practices; genetic factors;
demographic, social, and cultural characteristics; and nutritional variables.
Environmental factors include high altitude (Haas,
1980)
and exposure to specific
toxic agents (Institute of Medicine,
1985).
Numerous medical complications have
been found to be associated with low birth weight and
a
partial list is given in Table
7.
A
high proportion of newborns in Guatemalan and in African communities have
been shown
to
have elevated cord serum concentrations of immunoglobulin
M,
presumably reflecting fetal responses to infectious agents (Mata et al.,
1977).
These
observations are preliminary, and the extent to which intrauterine infections are
a
cause of low birth weight remains unclear (Kramer,
1985).
It has been postulated that some bacteria
are
sources of phospholipase, an enzyme
observed to be related to preterm labor perhaps through its influence on the synthe-
sis of prostaglandins by
the
placental membrane. Prostaglandins play an important
role in
the
initiation of labor (Bejar
et
al.,
1981).
The role of chronic parasitic
infections, such as malaria, in causing low birth weight is well known and represents
a
significant problem in parts of the Third World (Kramer,
1985).
The use of addictive drugs (Kramer,
1985)
and consumption of large amounts of
alcohol (Chernick et al.,
1983)
during pregnancy have been found
to
be
associated
with lower birth weights. Cigarette smoking during pregnancy has also been found
to
be
associated with low birth weight across races, population groups, and socioeco-
Martorell and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
205
TABLE
7.
Medical complications
of
pregnancy associated with low birth weight (after Miller, 1983;
Grunberger et al, 1979; Institute
of
Medicine, 1985)
Intrauterine infections
Maternal morbidity (e.g., malaria, rubella)
Preeclamsia
Hypertension at delivery
(>
90
mm Hg for diastolic values)
Hypotension
(<
110/60
mm Hg)
Chronic hypertension
Severe vaginal bleeding in third trimester
Abnormally high glucose tolerance curves
Uncontrolled diabetes
Malformation
of
placenta, cord,
or
uterus
Placenta previa or abruptio placentae
Anemia
Leukemia
Malignant solid tumor
Large uterine fibroids
Continuous medication with corticosteroids, immunosuppressive, teratogenic,
or
fetal growth-retarding
Maternal-fetal blood incompatibility producing disease in newborn
DES
exposure
Inadequate blood volume expansion
Oligoh
y
dr amnios/pol yhy dr amnios
Hyperemesis
drugs
nomic levels (Scholl et al., 1986; Garn and Rosenberg, 1986). A recently completed
randomized clinical trial has shown that reductions in maternal smoking during
pregnancy result in heavier newborns (Sexton and Hebel, 1984; Hebel
et
al.,
1985;
Nowicki et al., 1984). Though few women smoke in developing countries, many are
continuously exposed to cooking smoke. In parts of Central America, the very poor
live in single-room huts and the cooking is sometimes done inside the hut on open
fires. Carbon monoxide levels may be high in these situations, particularly in poorly
ventilated homes, and may lead to high carboxyhemoglobin levels in maternal blood.
The significance of cooking smoke for birth weight has not been evaluated but is
presumed
to
be great since low birth weight has been found to be associated with
exposure
in
pregnancy to sidestream cigarette smoke (Martin and Bracken,
1986).
Other examples of adverse practices include restrictive dietary practices. Western
physicians used to routinely recommend “dieting” to pregnant mothers. In many
societies, there
are
taboos against the consumption of many food items by pregnant
women
(WHO,
1982; Ebrahim, 19801, but the degree to which these customs affect
birth weight remains
to
be shown. A similar point is made by a National Academy
of Sciences (1983) report with regard to food restriction during lactation.
It
is pointed
out that although the literature contains numerous discussions on the subject, often
with ethnographic examples, “little solid research addresses the impact of cultural
food restrictions
or
dietary changes in general on the health and nutritional status
of lactating women” (p.
18).
Genetic factors also play an important role in explaining variations in birth
weight. Race
is
often an important factor though one must be careful to adjust
for
socioeconomic status. Thus, in the
U.S.,
it has been observed that blacks have
smaller newborns after controlling for the confounding effects of socioeconomic
variables and maternal anthropometry (Institute of Medicine, 1985). Gender, congen-
ital malformations, inborn errors of metabolism, and multiple births are examples
of
characteristics
or
conditions that strongly influence birth weight. Maternal
stat-
ure, even in developed countries, is a significant predictor of birth weight (Anderson
et al., 1984; Lechtig et al., 1977). Although maternal stature probably reflects the
genetic potential for size in well-nourished populations, in poor societies
it
will also
be reflective of the history of maternal undernutrition. Studies have also shown
strong associations between maternal stature and infant mortality in developing
countries. In a sample of women living in coffee plantations in Guatemala, infant
mortality was 205 deaths per
1,000
live births for the shortest one-third of the
women, 150 for those in the middle third, and 101 for those in the tallest third
206
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
(Vol.
30,
1987
(Martorell et al., 1981). Differences in socioeconomic conditions between the three
groups
of
mothers were minor
as all
were wives of salaried agricultural laborers
living in similar dwellings provided by the landlords. Future studies need to identify
the mechanisms for this association.
Demographic and socioeconomic factors are strong predictors of birth weight.
These relationships tend to be complex and to be mediated by numerous mecha-
nisms. Birth order and maternal age are among the demographic factors associated
with low birth weight.
A
study by the National Center for Health Statistics (1980)
showed that the relationship between birth order and low birth weight varied with
maternal age. For infants born to teenage mothers and mothers in their early
20s,
the incidence of low birth weight rose sharply for third and higher births. In
contrast, the proportion of low birth weight infants generally decreased for higher-
order births to women 35 years of age and over.
For
infants of women 25-34 years
of age, the relationship was U-shaped, with second- and third-born children showing
the least prevalence of low birth weight. Newborns of very young mothers bear a
higher risk of being premature and low in birth weight Vaffel, 1980; Scholl et al.,
1984; Scholl et al., 1987; Horon et al., 1983). Infants
of
women older than 35 years
have a higher incidence
of
low birth weight than infants of women 20-34 years of
age (NCHS, 1980). The mechanisms responsible for these effects remain unclear.
Frisancho and co-workers (1983,1984, and 1985) have studied the problem of teenage
pregnancy in Lima, Peru, and believe that requirements for growth during adoles-
cence are an important constraint for fetal development. Gynecological immaturity
and numerous social factors such as inadequate prenatal care, low income and
education levels, unmarried
status,
and inadequate social support systems may also
play an important role in determining low birth weight in very young mothers.
MATERNAL NUTRITION AND
BIRTH
WEIGHT
Evidence of a relationship between maternal undernutrition and low birth weight
can be inferred from
three
kinds of studies: experiences during wars or famines,
observational studies, and nutrition experiments.
Wars
and
famines
During and immediately following World War
11,
previously well-nourished Euro-
pean populations experienced severe, sudden food shortages. The impact of these
deprivations on birth weight
was
assessed in studies in Leningrad (Antonov, 1947),
Holland (Smith, 1947; Stein et al., 19751, and Wuppertal (Dean, 1951). The most
severe situation probably occurred in Leningrad, where the famine lasted from
August 1941 to January 1943; conditions were particularly severe during the winter
of
1941-1942, as the Germans laid siege to the city. Antonov (1947) compared the
birth weights of infants born during the famine to those observed prior
to
the famine.
During the worst of the famine, the mean birth weight was reduced by more than
500
g and the prevalence of low birth weight reached 49%. The Dutch famine caused
birth weights to be reduced by about
300
g but only in those mothers who were
exposed during the third trimester
of
pregnancy (Smith, 1947; Stein et al., 1975).
Exposure to the famine
at
the earliest stages of pregnancy resulted in an excess
of
premature births, early neonatal deaths, and central nervous system malformations
(Ribeiro et al., 1982). In Wuppertal, mean birth weight during the worst period
of
the famine, when intakes may have been less than 1,000 kcallday, was reduced by
170
g
in private patients and by 277 g in other patients. When official rations were
increased to between 1,050 and 1,550 kcallday in 1947, the reduction in birth weight
was
81
g in private patients and 117 g in other patients (Dean, 1951).
These wartime experiences show clearly that acute starvation leads to substantial
reductions in birth weight. Factors other than food restrictions, such as extreme
physical labor and bitter cold in Leningrad, may have also contributed to the effects
on birth weight. In comparing these studies to the situation in developing countries
other aspects should be kept in mind. The European populations were well nourished
prior to the famine period and maternal size had not been limited by a prior history
Martorell
and
Gonz6lez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
201
of undernutrition. Mean birth weights in these areas prior to the famine were
between
3,300
and 3,500 g; as high, therefore, as they are today. On the other hand,
the dietary restrictions were more severe than those which populations in develop-
ing countries usually encounter.
Observational
studies
There are marked seasonal changes in food availability in some countries. In
Gambia,
it
has been shown that dietary intakes
are
much less during the rainy
season than during the
dry
season (Roberts et al., 1982). Physical activity is also
sharply increased during the rainy season because of agricultural tasks, and the
incidence of infections such as malaria and diarrhea is also increased. As a result,
maternal body weight, skinfolds, breast milk production, and birth weight are
reduced. During the
dry
season, from January to May, mean birth weights were
found to rise from about
3,000
to over 3,200 g. In contrast, mean birth weights
during the rainy season were about 2,900 g. Data from Ethiopia suggest that heavy
physical work during pregnancy will depress mean birth weights by about
200
g;
these studies compared groups of women consuming similar diets but following
different work patterns (Tafari et al., 1980).
Numerous authors have related dietary intakes during pregnancy to birth weight,
and some, but not all, have found significant associations with nutrient intakes
or
with indices of dietary quality (Bergner and Susser, 1970; Phillips and Johnson,
1977). Among the factors contributing to the conflicting nature of the reports are the
problems of measuring dietary intakes. Reliable measures of dietary intakes, partic-
ularly in nonliterate populations, are difficult to obtain. Ideally, estimates of dietary
intake
at
various times during pregnancy would be required to estimate the “usual”
intake of individuals. Unfortunately, this has rarely been done. There are numerous
other problems that cloud the interpretation of studies of maternal diets and birth
weight; pregnancy-induced changes in maternal nutrient stores, nutrient utilization,
and in physical activity are rarely taken into account in studies of diet and birth
weight. Adaptations during pregnancy in basal metabolism also have not been
considered in most studies (Lawrence et al., 1984). Data from Scottish women
(Durnin et al., 1985) suggest that depressions of metabolic rates in the
first
half of
pregnancy result in considerable energy savings such that total energy needs during
pregnancy in healthy, well-nourished women may be closer
to
20,000 kcal than to
the 80,000 kcal specified by
FAO/WHO
(1985).
The
relationship between maternal anthropometry and birth weight has also been
studied. Maternal height, prepregnancy weight, and weight gain have been shown
to be significant predictors of birth weight (Dougherty and Jones, 1981; Luke and
Petrie, 1980; Naeye, 1979; Winikoff and Debrovner, 1981; Lechtig et
al.,
1978; Mbise
and Boersma, 1979; Kramer, 1985). The relationship is complex and clearly differs
depending upon maternal body size and shape.
For
instance, the negative impact of
low pregnancy weight gain on birth weight is stronger in women with low prepreg-
nancy weights (Brown et al., 1981). One useful spin-off of these types
of
investiga-
tions
is
that anthropometric indicators, together with health, social, and demographic
indicators, can be used to compose risk scales to identify women who are more likely
to deliver low-birth-weight babies. Europeans were pioneers in developing risk
scales for low birth weight, with Pappiernik-Berkhauer and Giffei (1969) reporting
methods to predict prematurity and IUGR.
As
relationships are likely to vary by
population, area-specific studies are required.
One simple interpretation of the anthropometric studies is that they indicate that
mothers with greater body size have larger babies because they have greater nu-
trient and fat stores. However, anthropometric characteristics also reflect genetic
components and
are
also related to socioeconomic conditions. Thus, maternal anthro-
pometry may be related to birth weight through several mechanisms.
The interval between births has been found
to
be consistently associated with
birth weight in studies in developing countries (Morley, 1973; Eisner et al., 1979).
The usual interpretation
is
that short birth intervals do not allow for a full recuper-
208
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
[Vol.
30,
1987
ation of maternal body stores and thus affect birth weight.
A
failing of many studies
according to Kramer (1985) is that birth interval, which is influenced by the duration
of gestation, has been used rather than pregnancy interval (i.e., end of first preg-
nancy
to
beginning
of
second pregnancy). Thus, further research is required
to
clarify this relationship.
Intervention studies
There have been only a handful of studies where investigators have offered a
nutritional treatment
to
pregnant women and then measured the effects on birth
weight.
Developed countries
Results of three recent intervention trials are available from cities in developed
countries: New York (Rush et al., 1980), Montreal (Rush, 1981) and Birmingham,
England (Viegas et al., 1982b). In addition, evaluations of the impact of the WIC
program on birth weight have been published.
The New York study was a controlled intervention trial (Rush et al., 1980). Women
were randomly assigned to two experimental treatments and a control, and proce-
dures were carried double-blind. The study women were poor black women selected
on the basis of criteria thought to predispose them to low birth weight and included
at least one of the following: a prepregnant weight less than 110 lb, a history
of
previous delivery of an infant of low birth weight, low weight gain at the first
prenatal visit,
or
low protein in the diet. The last criterion was based on a single 24-
hour recall dietary survey, and its value
to
select women at risk is questionable.
One group of women received a “supplement” consisting of two
8-oz
cans
of
liquid
providing a large amount of animal protein (40 g) and 470 kcal. The second group
received a “complement,” two
8-02
cans providing 6 g of animal protein and 322
kcal. Both experimental groups as well as the control group received regular clinic
care and vitamins and minerals. Daily dietary intakes were 2,326 kcal and 102 g of
protein for the supplement group, 2,272 kcal and 83 g of protein for the complement
group, and 2,065 kcal and 79 g of protein for the control group. The findings were
unexpected in that mean birth weights were
lower
by 32 g in the supplement group
than the control group
(P
>
.05). There was also a tendency for an increase in the
number of very early premature deliveries and in neonatal deaths in the supplement
group. On the other hand, the group receiving complement had birth weights that
exceeded those of the control group by 41 g
(P
>
.05). The mean birth weight of the
control group was 2,970 g. One interpretation of the study
is
that excessive protein
intake is harmful, though
it
should be noted that the effects found were small and
non-significant. The complement, more comparable to actual diets in terms of its
protein-energy ratio, resulted in a possible improvement in birth weight, though a
small one.
A
program of nutritional counselling and dietary supplementation in Montreal
was found to be associated with a small increase in birth weight (Rush,
1981).
The
program was designed
as
a practical intervention and not for scientific purposes.
Thus, only about 75%
of
the participants received dietary supplements and their
actual contribution to dietary intakes is unknown. In the analyses, the researchers
could not take into account who received food and who did not. The women were
said to be of lower socioeconomic status and were not selected because they were
undernourished. Participants were compared
to
nonparticipants in a matched-pairs
analyses. The mean birth weight was 3,291 g for participants and 3,251 g for non-
participants
(P
<
.05,
n
=
1,213 pairs). The effect was found to be conditional in the
following way. The differences in birth were greatest for newborns of women in their
first pregnancy
(+
61 g) and for women who weighed less than 140 lb at conception
(+53 g).
Viegas et al. (1982a,b) have carried out studies in East Asian women living in
Birmingham, England. In
a
first
study, 153 women were assigned to three groups:
(1)
vitamins only, (2) vitamins
+
273 kcal, and
(3)
vitamins, 26 g protein, and 273
Martorell
and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
209
kcal (Viegas et al., 1982a). Many of the women in this study were said to live in
“overcrowded” conditions. Beginning in the 18th-20th week of gestation the supple-
ments were delivered to the patients’ homes at 5-week intervals. The energy and
the
protein-energy supplements, which were
in
liquid form, were designed
to
satisfy
the extra dietary allowances
of
energy and protein for pregnant women recom-
mended in Britain.
No
differences were found in the birth weight of infants born to
women in each of the three groups. Mean birth weights were about 3,000 g.
A
second
study focused on 45 mothers who
at
28 weeks of pregnancy were judged to be at risk
of
delivering a small-for-date infant on the basis
of
weekly changes in the triceps
skinfold? The 45 women were assigned to each
of
the three experimental groups as
in the first study but beginning at 28 weeks
of
gestation.
Also,
85 women judged not
to be at risk on the basis of changes in the triceps skinfold were assigned to each of
the three groups. Protein-energy supplementation was associated with better weight
gain but only in the high-risk group.
For
mothers at risk, birth weights were 310 g
larger
for
the 14 cases in the protein-energy group when compared to the 31 cases in
the other two groups
(P
<
.02). Differences in birth weight were not statistically
significant among groups for mothers not at risk. The authors argued that their
study suggests that prenatal supplementation has a differential effect depending
upon the mothers’ nutritional status. The two studies by Viegas et al. (1982a,b) are
important contributions to the literature. However, the sample sizes are small and
the results
are
best viewed as preliminary.
For
example, the possibility cannot be
excluded that the lack of demonstration of an effect
of
dietary supplementation in
the
first
study and in the population not at risk in the second study is simply a
reflection
of
poor statistical power.
Dwyer (1983) has reviewed the literature on the evaluation of the impact
of
the
WIC program on birth weight. The aim of the program is to provide food supple-
ments
to
low-income pregnant
or
lactating women,
to
infants, and to children who
have not reached their fifth birthday and who are considered to be at risk of
undernutrition. Surprisingly few health impact evaluations have been carried out.
Dwyer (1983) has identified major shortcomings in these studies; some of these stem
from the fact that some evaluations were carried out before the program was fully
implemented and others relate to deficiencies in design including inadequate con-
trols and the unrepresentativeness of the subjects chosen
for
the studies. These
studies have overwhelmingly suggested a beneficial effect on birth weight.
For
example, in a study carried out in Massachusetts by Kennedy et al. (19821, mean
birth weights among WIC women were 120 g higher than those
of
newborns of
women who did not participate in WIC (3.27 kg vs. 3.15 kg,
P
<
.001). However, as
Dwyer (1983) points out, “Intervening variables could not be controlled completely,
possibly leading to overattribution of effects on birth weights
to
the program” (p.
141).
Retrospective data
from
a
recent national investigation
of
the
WIC
program
indicates that WIC benefits were associated with increased mean birth weights in
the range of 23-47 g (Rush, 1986).
As
Rush (1986) notes, “the most serious possible
limitation
of
this analysis is that effects attributed to the intensity of WIC service
to
eligible women.
.
.
are really due
to
something else, e.g., improved perinatal care”
(pp. 111-116). Prospective data from this same evaluation failed to find differences
in birth weight between a national probability sample
of
pregnant women enrolled
in the WIC program and a sample
of
low-income pregnant women not enrolled in
WIC (Rush, 1986).
Studies in developing countries
It
is also the case that only a few field studies of nutritional supplementation and
outcome of pregnancy have been carried out in developing countries. These include
studies carried out in Taiwan (Herriot et al., 1978; Adair and Pollitt, 1985; Pollitt
‘The authors state a cutoff point
of
less than
2
mm per week. Details as to the monitoring
of
skinfold changes and
the quality
of
the measurements
were
not provided.
210
YEARBOOK
OF
PHYSlCAL ANTHROPOLOGY
[Vol.
30, 1987
and Mueller, 1982; McDonald et al., 1981), Bogota, Colombia (Mora
et
al., 19791,
Guatemala (Habicht
et
al., 1974; Lechtig et al., 1975a), Gambia (Prentice
et
al.,
1983), and Mexico (Chavez and Martinez, 1979).
The Taiwan study was
a
randomized, controlled, double-blind trial. The population
was poor and rural but not necessarily undernourished (Adair and Pollitt, 1985;
McDonald
et
al., 1981).
Two
hundred ninety-four women were randomly assigned
to
one of the two treatment groups. The daily supplement for group A provided
800
kcal and
40
g of proteidday and for group B it was
80
kcallday with no protein.
Supplementation began
3
weeks after delivery and continued through lactation and
through the pregnancy and lactation of a second study infant. Inadequacies in the
dietary methodology do not allow for estimates of the true contribution of the
supplement
to
diets. Herriot et
al.
(1978) limited the analysis
to
only those women
who used
50%
or more of the supplements and found that birth weights of the second
study infants in group A were higher by 94 g than those of infants born to mothers
of group
B.
After adjusting for the fact that group
B
first newborns weighed 43 g
more than group A newborns (i.e., before supplementation),
a
51
g difference was
left that could
be
attributed to the intervention.
McDonald et al.
(1981)
included all subjects of the Taiwan study and focused on
the analysis of change scores in birth weight (i.e., the birth weight of the second
minus that
of
the
first
infant). Presumably change scores are a more effective way
of controlling for factors constant
to
the mother (e.g., maternal height or socioeco-
nomic status,
SES).
There were no significant differences in change scores between
treatment groups (i.e.,
A
vs. B). However, second male infants were 162 g heavier
(P
<
.05)
than first male infants in group
A.
A
randomized intervention trial was carried out in Bogota, Colombia, by Mora
and co-workers (1979). Women from families from poor neighborhoods who were in
the first or second trimester
of
pregnancy and who had
at
least
50%
of family
children with
a
weight for age less than
85%
were the subjects of the study. The
average daily diet provided 1,600 kcal and 35 g of protein. Families were randomly
assigned to supplementation and control groups at the beginning of the third trimes-
ter
of pregnancy. Foodstuffs were supplied to the families through neighborhood
stores. The increments in net dietary intake resulting from the supplementation
were estimated to be 155 kcal and 20 g of protein per day. The results of the study
are presented in Table
8
for all 407 cases and for the 339 full-term newborns. There
were differences of
51
(total sample, sexes combined) and
50
g (full-term, sexes
combined) between the unsupplemented and the supplemented groups, but these
differences were not statistically significant. Among full-term males, however, sup-
plementation did result in a statistically significant increment of 95 g
(P
<
.05).
The
mean weight gain of full-term mothers during the last trimester of pregnancy was
greater in supplemented than in unsupplemented cases, but only for those delivering
full-term male infants
(P
<
.05).
In a later publication, Mora et al. (1981) argued that
fetal growth of males toward the end of pregnancy is more rapid and hence more
TABLE
8.
Mean birth weight
(R,
by
sex and supplementation
group
for the total sample and for full-term
newborns
(after
Mora
et
aL,
1979)'
---
Males Females
Both
sexes
n
X
SD
n
X
SD
n
X
SD
Unsupplemented
93 2,951 411 107 2,905 374 200 2,927 392
Supplemented
107 3,040 414 100 2,911 322 207 2,978 377
Total
200 2,999 414 207 2,908 349 407 2,953 385
Unsupplemented 74
2,966 322
88
2,942 303 162 2,953 311
Supplemented
95 3,061 374 82 2,935 320
177
3,003 355
169 3,019 354 170 2,939 310 339 2,979 335
Total
'Differences between supplemented and unsupplemented groups. The difference between supplemented and
unsupplemented full-berm males
was
95
g
(singletail t-test
P
<
.05).
Differences between sexes are significant (t-test
P
<
,051
among the supplemented groups.
-_______
-____
Group
A.
Total sample
B.
Full-term
--
Martorell
and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
211
susceptible
to
adverse environmental influences than that of females. They postu-
lated that male fetuses exhibit greater sensitivity to nutritional supplementation of
their mothers during pregnancy.
Other analyses from the Bogota study indicate that the stillbirth, neonatal, and
perinatal mortality rates were all lower in the supplemented group. Because of
small sample sizes, none of these differences was statistically significant (Mora et
al., 1978). Herrera et al. (1980) have also shown that the effect of the food supplemen-
tation program on birth weight was evident only in women with low weight for
height
(<
360 g/cm). Differences in mean birth weights between supplement and
control groups were
181
g
(I‘
<
.01)
in women with low weight for height but only
22
g
(P
>
.05)
in heavier women.
The Guatemala study was carried out in four villages from an area of endemic
chronic undernutrition (Lechtig et al., 1975a).
Two
types of liquid supplements were
used. In two of the villages, a high protein-energy supplement called “atole” contain-
ing 11.5 g
of
protein and 163 kcal per 180 ml was made available. The other two
villages were provided with a fruit-flavored
drink
called “fresco” that contained no
protein and supplied only 59 kcal per 180 ml. The supplements also contained
vitamins and minerals. Attendance
at
the food supplementation centers and con-
sumption of the supplement were free and voluntary and, as a result, a wide range
of supplement intake during pregnancy was observed. The supplements were given
twice daily, 7 days a week. Subjects were provided with a cup containing 180 ml,
and more was given if requested. Leftovers were measured and actual intake on an
individual basis was recorded to the nearest 10 ml. The measurement of individual
supplement intake
is
a unique feature of this study.
The Guatemala study was originally designed
to
be a contrast between the treat-
ment villages (i.e., those receiving atole) and the control villages (i.e., those receiving
fresco). The design was conceived at
a
time when protein was thought to be more
limiting than energy in most diets in developing countries. At one time, the possibil-
ity of using cyclamates as
a
sweetener in the fresco was contemplated but the idea
was discarded because of concerns for
its
carcinogenic potential. As the study un-
folded, the investigators abandoned the planned village-level comparisons as they
discovered that both the atole and the fresco were producing an effect, an observation
explained as an effect of
calorie^.^
Instead they adopted the strategy of relating
actual supplement intakes during pregnancy to birth weight. This approach left
open the possibility that the observed association between supplement intake and
birth weight was due to self-selection. The investigators addressed this issue by
controlling for potential confounding factors including obstetric factors, home diet,
morbidity, and socioeconomic status.
In one analysis, the birth weights of infants born to mothers consuming more than
20,000
kcal from the supplements during the last two trimesters
of
pregnancy were
compared to those born to mothers ingesting smaller amounts of energy. The new-
borns of the better-supplemented group had
a
mean birth weight of 3,105 g while
the poorly supplemented group had a mean
of
2,994 g
(111
g difference,
P
<
.05).
As
shown in Table 9, the findings were similar in atole and in fresco villages. The
difference,
111
g of birth weight, is one of the largest effects found in the studies
carried out to date. An intake of 10,000 kcal during pregnancy was associated with
29 g of added birth weight, a figure that became 30 g after controlling for potentially
confounding factors. The prevalence of low birth weight was 17% in the low-intake
group and
8%
in the high-intake group. The.results suggest that providing energy
(fresco)
is
just as effective as providing energy and protein (atole). However, birth
weights were higher in atole than in fresco villages in
both
the low- and the high-
supplementation groups. This may mean that there were better preexisting mater-
nal conditions in atole villages rather than a simple protein effect.
If
protein were
the sole explanation for the effect, one would have expected the low-supplementation
3The atole had nearly three times the amount of calories as a similar volume of fresco. However, actual intakes of
calories from the supplements were
not
nearly as different because mothers drank more fresco than atole.
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
[Vol.
30, 1987
TABLE
9.
Supplementation duringpregnancy and birth weight for
two
levels of caloric supplementation
(from
Lechtig et al.
1975~)
2 12
Mean
birth
weight
b)
Supplemented
calories during Atole
Fresco
All
pregnancy
High
3,1731 3,035 3,105
2
20,000 (102)
(68)
(170)
Low
3,042 2,948 2,994
<
20,000 (117) (118) (235)
Total
3,107 2,992 3,049
rt
469
(219)
(186)
(405)
'Sample siae
in
parenthesis.
villages villages villages
__~
_______
groups in atole and fresco villages to be similar and the high-supplementation
groups
to
be different. Nonetheless, the possible role of protein in this study cannot
be excluded and further research on this important topic
is
highly desirable.
The Guatemalan study also found that the impact of the supplement was greatest
in those women who apparently needed it the most, that is, shorter women and
women of poorer socioeconomic status (Lechtig
et
al., 1975b). Finally, the Guatema-
lan study suggests that supplemental energy taken early in pregnancy is
as
effective
as energy taken in the third trimester of pregnancy in
terms
of effects on birth
weight (Habicht et al., 1974).
Prentice et al.
(1983)
have published the results of an intervention study in the
village of Keneba in Gambia. The investigators had baseline data for a period of 4
years prior to the supplementation program and then carried out
a
food supplemen-
tation program for 2 years. Daily diet before supplementation was 1,48011,300 kcal
during the drylwet season. Groundnut-based biscuits and a vitamin-fortified tea
drink were provided. The supplement was offered on 6 days each week and consump-
tion was carefully supervised and measured in
a
supplementation center. On aver-
age women were attending by the 16th week
of
pregnancy and attendance by
pregnant women was very high. Amounts of 950 kcallday were offered daily during
the dry season and 1,110 kcallday were provided during the wet season when food
was
less
plentiful. Average supplement consumption was 673 kcallday, but because
of a reduction in home food consumption, the actual contribution
to
the total dietary
intake was 431 kcallday. Still, this is the most substantial dietary supplement
consumed by women in any of the studies.
A
total of
181
singleton births were studied during the pre-supplementation phase
and 97 during the supplementation phase. The mean birth weight during the dry
season of the intervention phase was similar
to
that measured during the
dry
season
of
the baseline phase. However, the mean birth weight was significantly greater for
the rainy season
of
the intervention phase compared to the rainy season of the
baseline phase. Specifically, mean birth weights were greater by 224 g (2,743 vs.
2,967,
P
<
.001) for birth weights of known gestational age and after adjustment for
confounding factors. The authors argue that the women were in positive energy
balance during the
dry
season and that as a result the supplements had no effect
during that period. On the other hand, conditions in the rainy season were harsh,
food was scarce, and demands for physical activity were greatly increased. For these
reasons, the authors argue, the food supplements produced an effect during the
rainy season. The findings are provocative, but the lack of
a
control group poses
problems in the interpretation of the results. There
is
no way, for example, to
effectively control for time trends in other factors which also influence birth weight.
There
is
very little additional information available. Iyengar (1967) carried out
a
hospital study in India on a limited number of subjects. Women were hospitalized
before the 34th week of pregnancy and fed two types
of
diets. One group was given
a diet providing 2,100 kcallday and 60 g of protein per day (n
=
13)
while the other
(n
=
12) was given milk in addition
(35
g of protein and 350 kcal). Birth weights
were similar in the two groups but higher by 324 g when compared to those of
infants born to 26 women who delivered
at
the nursery but who had not been
Martorell
and
Gonzalez-Cossio]
MATERNAL NUTRITION AND BIRTH WEIGHT
213
hospitalized. The potential factors accounting for the differences between the hospi-
talized and nonhospitalized women are
so
many that little can be concluded from
this study.
Chavez and Martinez (1979) carried out a small study in Mexico, and
it
has been
reported that a change of 180 g in birth weight and a
29.6%
reduction in the
incidence of low birth weight was produced by supplementation during pregnancy.
The study compared
the
weight of newborns born to samples of women in the village
before and after a supplementation program.
SUMMARY
AND
DISCUSSION
The notion that the human fetus
is
an effective parasite who will get what
it
needs
at the expense of the mother was
a
popular one among scientists until recently. The
studies carried out during World War
I1
served to dispel this notion because acute
starvation was shown to dramatically reduce size at birth. Obstetricians in the
40s
and
50s
widely believed that birth weight could be influenced by maternal diet and
many sought to limit weight gain during pregnancy by asking mothers to eat less.
Physicians desired smaller birth weights to lower the incidence of maternal compli-
cations during pregnancy. A great deal has been learned in the last two decades,
The implications
of
low weight at birth are now clear and dietary restrictions during
pregnancy are now only an embarrassing memory. On the other hand, there are
many issues that remain unresolved. A few of these, those which we consider to
have great practical significance, are discussed in this section.
Birth weight data
The review by
WHO
(1980) clearly documented that low weight at birth is
a
major
public health problem in the developing world. At the same time, the review made
painfully clear that data on birth weight are scarce, even in Third World countries
with relatively advanced economies. Efforts to develop simple but appropriate data
collection systems are therefore required. As little information generally filters from
rural areas, where deliveries usually take place at home, it is crucial that data
collection systems rely on the use of paramedical personnel and traditional birth
attendants as data collectors. The effort to collect data should not be an end in itself,
but rather the immediate rationale should be
to
identify those women and newborns
in need of special attention. At ministerial levels, the routine compilation of these
data could serve to design, monitor, and evaluate programs oriented to improve
infant health.
Significance
of
birth weight
There is no question that low birth weight is adversely correlated with growth,
development, and survival during infancy. With notable exceptions, most of the
information available comes from studies in developed countries. Conditions are
so
different between rich and poor nations that one should be extremely cautious when
extrapolating findings from one to the other.
It
is
likely that in an environment of
endemic undernutrition and high rates of infection, weight at birth plays a greater
role in determining survival and postnatal development than
it
otherwise does in
more favorable environments.
In spite of some prior research, the following questions seem
to
us to deserve
greater attention among researchers:
To
what extent is low birth weight
a
factor
in
the epidemiology of severe undernutrition? What is the relative importance of birth
weight and postnatal nutrition in terms of predicting psychological test performance
and school achievement? Is the 1-4-year mortality rate predicted by birth weight in
various environments? What are the implications
of
birth weight for growth beyond
infancy, for the development of the immune system, and for child health? Few
investigators have looked at anthropometric characteristics other than birth weight
in assessing risk. The evidence to date suggests future studies need to consider,
at
the very least, gestational age, birth weight, and birth length. Finally, more studies
are required to investigate the effects of low birth weight across generations. In
2
14
YEARBOOK
OF
PHYSICAL ANTHROPOLOGY
[Vol.
30,1987
addition to examining reproductive performance as a function of maternal birth
weight, other outcomes such as infant mortality and child growth and development
need to be included.
Maternal nutrition and birth weight
Studies from around the world
are
now available to demonstrate conclusively that
maternal diet during pregnancy does make
a
difference in terms of birth weight.
However, the impact on birth weight clearly varies in different settings.
In famine,
as
shown by the World War
11
studies, birth weights are markedly
depressed if the mother
is
starved during the last trimester of pregnancy. Starvation
early in pregnancy followed by adequate food in the last few weeks of pregnancy
does not apparently affect birth weight although
it
has a negative impact on pre-
maturity rate and survival. Of course, all these observations are only applicable to
mothers with adequate lifetime nutrition prior to the period of starvation.
Recent studies from developed countries carried out in mothers who were poor but
not necessarily undernourished indicate
a
rather small effect on birth weight. On
the other hand, effects on birth weight were somewhat larger in some of the studies
from developing countries (e.g., Guatemala and Gambia). Many investigators have
reported that mothers with poor nutritional status benefit the most from food
supplementation. Women who are short or thin and who receive a food supplement
during pregnancy, for example, have been shown
to
have babies who are substan-
tially heavier than those born to similar women who did not receive
a
supplement.
Most of the supplementation trials have been canied out in populations where the
average birth weight exceeds
3,000
g and where the prevalence of low birth weight
is much lower than the estimated average value of
17%
for the developing countries.
Food supplements have not been tried in
areas
where birth weights
are
very low
(e.g., at or below
2,800
g) and where the expected effect would
be
greater.4
It
is
of
great interest that in the Gambian supplementation trial, effects of food supplemen-
tation during pregnancy were extraordinarily large during the rainy
season
when
mean birth weights are generally between
2,800
and
2,900
g and totally absent
during the dry season when birth weights usually exceed
3,000
g.
We have learned that there is
a
differential response to food supplementation.
Therefore, the cost effectiveness of food aid programs to pregnant women can be
enhanced by selecting preferentially those women in whom
a
greater impact of food
on birth weights would be expected.
Maternal anthropometry
Undernutrition in childhood can lead to short stature as an adult. A poor diet as
an adult will be reflected in
a
low prepregnant weight for height and in
a
low weight
gain during pregnancy. Undernutrition throughout the entire life of the mother will
therefore affect the development of her children.
A question of great practical and scientific importance for the development of risk
indicators of low birth weight
is
the relative importance of height, prepregnant
nutritional
status
(e.g., weight for height), and of changes during pregnancy (e.g.,
weight gain). For example, does a history of prior undernutrition,
as
reflected by
short maternal stature, place constraints on the degree of improvement that could
be achieved by dietary improvements during pregnancy? Are dietary intakes during
pregnancy in very short mothers limited by pressure from
a
relatively large fetus
and the fact that diets in developing countries
are
usually bulky?
Timing of
food
supplementation
It
has
already been mentioned that the Holland study showed that starvation of
previously well-nourished women affected birth weight only if it occurred during the
‘There
are
supplementation trials currently being carried out in populations in Bombay, India, and in East Java.
Indonesia.
Results
from these studies are not yet available.
Martorell
and
Gonzalez-Cossio]
MATERNAL NUTRITION
AND
BIRTH WEIGHT 215
last
trimester of pregnancy. The Bogota and the Birmingham studies also suggested
that supplementation in the latter part of pregnancy will improve birth weights.
Finally, preliminary analyses have been reported from the Guatemalan study that
suggest that
it
makes no difference in terms of birth weight when energy is provided
to mothers during pregnancy. On the other hand, Guatemalan data suggest that
supplementation early in pregnancy has a positive impact on duration of gestation
and that the earlier in gestation that calories are offered the greater the impact on
length
of
gestation (Delgado et al.,
1982).
More data analyses are therefore required
on the issue
of
timing as the results will inform planners as
to
when in the course of
pregnancy
it
is
best to intervene.
A
similar situation exists with regard to the differential impact of maternal
undernutrition during pregnancy on boys and girls. Is
it
the case that maternal
undernutrition
is
more likely
to
affect the weight of male newborns and therefore
that dietary improvements in pregnancy are more likely to favor boys? Though some
of the studies have reported such findings, other authors have not systematically
addressed this issue.
Specific nutrient effects
The New York study raises the possibility that in women who are not deficient in
protein, very high amounts of protein will be harmful to the fetus, specifically those
of women with a previous history of low birth weight. Though the Guatemala study
suggests that additional energy rather than more protein was responsible for the
effects on birth weight, extrapolation
of
these findings to other settings is proble-
matic. The literature offers limited guidance regarding the ideal mix of protein and
energy in populations consuming different diets. There is even less information
regarding the improvements in birth weight
to
be specifically attributed to iron,
folate, and other nutrients.
Other factors affecting birth weight
There is also a need for clarifying the relative importance of a number
of
determi-
nants of maternal dietary intake, birth weight, and gestational age (after controlling
for the important confounding factors) especially in women of developing countries.
Infection
The role of intrauterine infections as causes
of
intrauterine growth retardation
and prematurity
is
not well understood, especially in developing countries. Little is
also known about the effects of other disease states such as diarrhea and
colds.
Do
these problems affect maternal diets and nutrient utilization and in this manner,
birth weight?
Seasonal aspects
Data from Gambia suggest
that
there are marked seasonal patterns in nutritional
intake, energy expenditure, and nutritional status. Birth weights differ markedly
by season.
Most
researchers in developing countries have generally disregarded
seasonal effects.
It
is desirable that investigators explore the extent to which such
patterns exist in