See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/12726792

Early Childhood Nutrition, Education and Fertility Milestones in Guatemala

Article  in  Journal of Nutrition · December 1999

DOI: 10.1093/jn/129.12.2196 · Source: PubMed

CITATIONS READS
24 74

5 authors, including:

Usha Ramakrishnan Huiman Barnhart

Emory University Duke University Medical Center
241 PUBLICATIONS   6,022 CITATIONS    193 PUBLICATIONS   10,980 CITATIONS   

SEE PROFILE SEE PROFILE

Dirk G Schroeder Aryeh Stein

Emory University Emory University
76 PUBLICATIONS   4,313 CITATIONS    272 PUBLICATIONS   11,210 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

FAMINE – Prenatal under-nutrition exposure and adult mortality View project

Predictors of childhood obesity in Colombia View project

All content following this page was uploaded by Huiman Barnhart on 06 January 2014.

The user has requested enhancement of the downloaded file.

 Community and International Nutrition

Early Childhood Nutrition, Education and Fertility Milestones in Guatemala1

U. Ramakrishnan,2 H. Barnhart,* D. G. Schroeder, A. D. Stein and R. Martorell
Department of International Health and *Department of Biostatistics, The Rollins School of Public Health at
Emory University, Atlanta, GA 30322

ABSTRACT Data on fertility milestones were collected in 1994 and linked to information collected in a trial
conducted in eastern Guatemala between 1969 and 1977, to examine whether early childhood nutrition was
associated with the timing of fertility milestones. In the original trial, two pairs of villages were randomly allocated
to receive either a high energy, high protein supplement (Atole) or a low energy, no-protein supplement (Fresco).
Mean age at follow-up was 23.47 y (n 5 240). About 62% of women had experienced first birth (median age at first
birth 5 19.83 y). The median intervals from menarche to first intercourse and from first intercourse to first birth were
5.67 and 0.95 y; they were 1.68 and 0.06 y shorter, respectively, for the Atole group than for the Fresco group.
Women who had received Atole in utero and/or during early childhood experienced earlier milestones even after
adjusting for socioeconomic status (SES), education and age at the prior event. Median age at first birth was 1.17 y
earlier for the Atole group. Better growth during early childhood (not severely stunted) led to earlier milestones

Downloaded from jn.nutrition.org by guest on June 13, 2013

(median age at first birth was 1.04 y earlier), primarily among women with illiterate fathers. Completion of primary
school significantly delayed fertility milestones; the median age at first birth was 4.27 y later for those who
completed primary school compared with those who did not (P , 0.05). In sum, improved nutrition during early
childhood results in earlier fertility milestones, but the effects of schooling in delaying fertility milestones are greater
in magnitude. Intervention programs that improve early childhood nutrition should be accompanied by investments
in education that ensure that girls complete primary school. J. Nutr. 129: 2196 –2202, 1999.

KEY WORDS: ● childhood stunting ● improved growth ● education ● fertility ● food supplements
● Guatemala

The improvement of childhood nutrition is viewed by
many developing countries and international agencies, such as
the World Bank and UNICEF, as an important strategy for
enhancing human capital and promoting economic growth
(Martorell 1996). One argument made against investments in
early childhood nutrition is that they may lead to increased
fertility and population growth (Bongaarts 1980), a major
problem in many developing countries with limited resources.
Specifically, improved childhood nutrition could lead to an
earlier occurrence of key fertility milestones, such as age at
menarche, first intercourse, first pregnancy and first birth. The
earlier occurrence of first pregnancy and first birth would then
be expected to result in higher total fertility rates (Balakrish-
nan et al. 1988, Freedman et al. 1981). A counterargument is
that improved maternal nutritional status results in improved
birth outcomes, such as higher birth weight and reduced infant
mortality (Martorell et al. 1981, Rivera et al., 1996), and
women whose children survive are more likely to participate in
family planning (Rahman 1998).
Studies about nutrition and fertility milestones are limited
mainly to menarche. In North America and Europe, age at
menarche has declined by ;3 y since the beginning of this
century (Tanner 1968 and 1975, Zacharias and Wurtman
1969). This has been attributed to improvements in the stan-
dard of living, including nutrition (Ostersehlt and Danker
Hopfe 1991, Prado 1986, Veronesi and Gueresi, 1994, Wolan-
ski, 1985). Conversely, Chowdhury et al. (1978) reported
delays in the age at menarche in famine/war situations involv-
ing acute food shortages. In a study of a rural Indian popula-
tion in Guatemala, Delgado et al. (1985) found that taller and
heavier girls were more likely to experience earlier menarche
and, more importantly, that an early age at menarche was
associated with earlier marriage. A major limitation of most
studies to date, however, is their cross-sectional study design,
which does not permit inferences of causality (Chowdhury et
al. 1978, Delgado et al. 1985 Frisch 1972, Frisch and
McArthur 1974, Hillman et al. 1971, Tanner 1975, Zacharias
et al. 1976). The only prospective study about menarche is our
own from Guatemala. Girls who were severely stunted at 3 y of
age, defined as height-for-age Z-score (HAZ)3 , 23 based on
the international National Center for Health Statistics
(NCHS)/WHO reference (WHO 1986), attained menarche
;7 mo later (P , 0.05) than girls who were not stunted (HAZ
. 22) (Khan et al. 1996). On the other hand, nutritional
supplementation was not associated with age at menarche
(Khan et al. 1995).

1 3
Supported by grants from UNICEF, Thrasher Research Fund and the Na- Abbreviations used: AHR, adjusted hazard ratio; CI, confidence interval; HAZ,
tional Institutes of Health (HD-22240, HD-22297, HD-33468, HD-34531). height-for-age Z-score; INCAP, Instituto de Nutrición de Centro América y Panamá;
2
To whom correspondence should be addressed. NCHS, National Center for Health Statistics; SES, socioeconomic status.

0022-3166/99 $3.00 © 1999 American Society for Nutritional Sciences.

Manuscript received 5 April 1999. Initial review completed 20 May 1999. Revision accepted 24 August 1999.

2196
 FERTILITY MILESTONES AND EARLY CHILDHOOD NUTRITION
It is not known whether these variations in age at men-
arche by early childhood nutrition extend to the timing of first
pregnancy and birth, which are outcomes influenced by bio-
logical and social factors. Several biological factors that may
be determined in utero and early in life may affect future
ability to conceive and deliver a healthy newborn and thereby
influence age at first birth. Recently, Lumey and Stein (1997),
using a retrospective cohort study design, found that poor
nutrition during pregnancy, measured by exposure in utero to
the Dutch Famine of 1944 –1945, did not alter age at men-
arche or age at first birth, which led the authors to conclude
that in utero exposure to nutritional deprivation in previously
well-nourished women does not reduce fertility. However,
those results must be verified in other settings, especially those
in which malnutrition is common.
Social factors, such as socioeconomic status (SES) and
education are well-known predictors of fertility behavior (Ann
et al. 1983, Castro 1995, dos Santos Silva and Beral 1997).
Analyses of cross-sectional data from Demographic and Health
Surveys in 26 developing countries found that greater school-
ing in women was consistently associated with lower fertility
(Castro 1995). Some studies show that education delays the
age at first birth across varied settings (DeWit and Rajulton
1992, dos Santos Silva and Beral 1997, Engle and Smidt 1996,
Nath et al. 1999, Nguyen et al. 1993). These effects may be
due to the effect of female education on age at marriage, family
size preferences and/or contraceptive use. Although the ben-
efit of women’s education in reducing overall fertility is well
known, the role it plays in the relation between nutrition and
fertility milestones remains unclear.
In summary, although the link between nutrition and fer-
tility has been long debated, no prospective studies have
examined how improved nutrition during early childhood
influences future reproductive performance. The objectives of
this study were twofold: 1) to examine the relationship be-
tween early childhood nutrition and the timing of fertility
milestones that occur after menarche; and 2) to examine the
role of education and the extent to which it modifies the
relation between nutrition and fertility milestones in a devel-
oping country setting with high rates of malnutrition.
SUBJECTS AND METHODS
Study sample. This study is a follow-up of participants in a
longitudinal, community-based, food supplementation trial that was
conducted in eastern Guatemala by the Instituto de Nutrición de
Centro América y Panamá (INCAP) between 1969 and 1977. Four
study villages were stratified by size (two large and two small) and
were randomly allocated to receive either a high energy, high protein
supplement (Atole) or a low energy, no-protein supplement (Fresco).
Atole provided 682 kJ of energy and 11.5 g of protein per cup (180
mL), whereas Fresco provided 247 kJ per cup and no protein. Both
supplements contained vitamins and minerals and were administered
twice daily. Data on maternal nutrition and on growth and matura-
tion, diet, illness and psychosocial development of children from
birth to 7 y of age were collected. Additional details about the
original intervention study are described elsewhere (Martorell et al.
1995a). Since 1988, a series of follow-up studies of the participants of
the original study have been conducted in the same villages in
collaboration with INCAP. This report utilizes data from the original
intervention trial and the 1988 and 1994 follow-up examinations
described below.
A total of 954 women who were born between January 1966 and
August 1977 in the four study villages were exposed to supplemen-
tation either partially or totally, in utero and during the first 3 y of life
in the original trial. Data on HAZ at or around 3 y, paternal illiteracy
and SES during early childhood, and SES in 1988, were available for
395 women. Fertility data were available for 61% of this sample (n
2197
FIGURE 1 Conceptual framework. Early childhood nutrition was
measured as the type of supplement or severe stunting at or around 3 y
of age.
5 240). Comparison of the final analytical sample with women for
whom follow-up data were not available (n 5 714) indicated no
differences in key variables such as type of supplement, level of
stunting, SES during early childhood and paternal illiteracy.
Data collection. Reproductive histories were collected for all
women of reproductive age who were residents of the four study
villages between October and December 1994. Trained field workers
interviewed women at home using pretested questionnaires. The
study protocol received ethical clearance from Emory University and
INCAP, and informed consent was obtained from all subjects. Events
recalled included age at menarche (to the nearest month), date of first
intercourse, all unions and marriages, all pregnancies and their out-
Downloaded from jn.nutrition.org by guest on June 13, 2013
comes (abortions, miscarriages, still births and live births) and details
of past and current contraceptive use. In addition, data on the highest
grade of schooling completed and current marital status were ob-
tained.
Conceptual framework. We considered a series of successive
time intervals among the following events: menarche, first inter-
course, first pregnancy and first birth (Fig. 1). In addition, cumulative
time intervals (menarche to first birth and age at first birth) were also
considered. We hypothesized that either the type of supplement
and/or severe stunting, both of which are measures of early childhood
nutrition, were the key biological variables that could influence the
timing of all the fertility outcomes, especially from menarche to first
pregnancy (Stein and Kline 1991). The social variables that we
considered were education (measured either as paternal literacy dur-
ing early childhood or the completion of primary school by the study
subject) and SES. Education could influence the timing of first
intercourse and/or first pregnancy, but not menarche. We therefore
examined the association between early childhood nutrition and
fertility outcomes in models that included either the antecedent
(paternal literacy and SES during early childhood) or concurrent
(completion of primary schooling and current SES) social variables.
Because there is no variability in the time interval from first preg-
nancy to first birth in our data, we used time to first birth in lieu of
first pregnancy.
Variable definitions. Fertility outcomes. Age at menarche, first
intercourse, first pregnancy and first birth were calculated to the
nearest month by subtracting the reported date of these events
(month, year) from the date of birth in the original longitudinal
study. More than 90% of the first pregnancies were first births and
time from first pregnancy to first birth was 9 mo for 95% of first births.
Therefore, time to first pregnancy was not considered, due to lack of
variability. The main outcome variables were the time intervals from
menarche to first intercourse and from first intercourse to first birth.
We also considered cumulative outcomes about menarche and first
birth, i.e., time from menarche to first birth and age at first birth.
Atole. This was defined as a dichotomous variable to indicate the
type of nutrition intervention received during early childhood (1 if
Atole, 0 if Fresco).
Severe stunting. If present at 3 y of age, this was used as a
cumulative measure of childhood nutritional status. Poor linear
growth or stunting is the best summary measure of chronic exposure
to poor diet, high rates of infection and inadequate care during early
childhood (UNICEF 1990). Previous analyses have shown that nearly
all growth retardation in our study population occurred by age 3 y
(Martorell et al. 1995b). Length measurements, obtained prospec-
tively during early childhood in the original longitudinal study, were
expressed as HAZ using the international NCHS/WHO reference
(WHO 1986). For women who did not have a length measurement at
2198 RAMAKRISHNAN ET AL.
TABLE 1
Selected characteristics of study women, by village and nutritional supplement1
Large Fresco
Characteristics (n 5 56)
Measured at follow-up
Age, y 23.1 6 2.9
Socioeconomic status2 0.57 6 0.88
Not completed primary school,2 % 51.8
Measured in early childhood
Socioeconomic status2 0.43 6 0.92
Paternal illiteracy,2 % 42.9
Severely stunted,2,3 % 44.6
1 Values for age and socioeconomic status are means 6 SD.
2 P , 0.05 for differences between villages.
3 Height-for-age Z-score , 23.
exactly 3 y, the next best measurement between 24 and 48 mo of age
was selected, following an algorithm that has been used in previous
analyses (Schroeder et al. 1995). Severe stunting was defined as a
HAZ , 23.
Socioeconomic status (SES). SES during early childhood and
adulthood were expressed as separate scores derived from principal
components analysis in previous studies from data on household
characteristics and possessions that were collected during 1975 and
1988, respectively (Rivera et al. 1995).
Paternal illiteracy. This factor was based on whether the father
could read or write at the time of the 1969 –1977 longitudinal study.
The level of education for the subject’s mother was not used because
of the extremely low prevalence and resulting lack of variability.
Schooling. This was categorized as “completed primary school” if
the highest grade of schooling completed was $6 y. Examination of
the effect of schooling on fertility milestones as a continuous variable,
i.e., years of schooling, indicated a strong threshold effect at ;6 y,
i.e., completion of primary school.
Data analysis. Survival analysis techniques were used to account
for right censoring (Fleming and Harrington 1991). Median values
obtained from Kaplan-Meier curves were used to compare the timing
of fertility outcomes within strata of the independent variables of
interest. Cox proportional hazard models were fitted to examine the
effect of the nutrition intervention (Atole vs. Fresco) and severe
stunting on the successive and cumulative time intervals, adjusting
for the following effects: 1) paternal literacy and SES measured during
early childhood and, 2) completion of primary school by the subject
and current SES. All analyses were restricted to those women who
had attained the prior milestone (all women had experienced men-
arche). In addition, we adjusted for the age at the prior milestone in
all analyses except that for age at first birth (Brewster et al. 1998). For
example, we controlled for age at first intercourse in the model for
time from first intercourse to first birth. In the analysis of age at first
birth, however, we also controlled for age at menarche. We examined
interactions between type of supplement and education, and between
severe stunting and education. Because the unit of randomization in
the original intervention trial was the village, the responses of indi-
viduals within a village are likely to be correlated. Therefore, robust
variance estimators (Lipsitz et al. 1994) were used to obtain P-values
for the parameter estimates in the proportional hazard models in the
overall analysis, which pools all four villages. All analyses were
performed using SAS (1997) and S-plus software (MathSoft 1996).
All tests were two-tailed, and a P-value # 0.05 was considered to
indicate significance. Missing data were assumed to be missing at
random (Rubin 1987).
RESULTS
Mean age at follow-up was 23.47 y (range, 19.22–28.76 y)
and was similar across all villages (Table 1). However, there
Village
Large Atole Small Fresco Small Atole
(n 5 70) (n 5 57) (n 5 57)
23.7 6 3.1 23.4 6 2.6 23.6 6 2.7
0.04 6 0.83 20.34 6 0.83 20.10 6 0.81
78.6 45.6 63.2
20.23 6 0.98 20.55 6 0.87 20.08 6 0.75
50.0 19.3 35.1
18.6 59.6 21.1
were significant differences among villages in SES, education
and nutritional status, both at the time of the original inter-
Downloaded from jn.nutrition.org by guest on June 13, 2013
vention trial and at follow-up. The large villages had higher
SES scores than the small villages; the difference between the
large and small Fresco villages was much greater than that
between the pair of Atole villages. Although overall SES
improved in all villages between 1975 and 1988, the between-
village differences persisted. Village-level differences in the
proportion of illiterate fathers and schooling were also present.
The small Fresco village, which had the poorest SES, also had
the lowest proportion of illiterate fathers and of study subjects
who did not complete primary school. About one third (35%)
of the study sample were severely stunted (HAZ , 23) at or
around 3 y of age. As reported elsewhere, the Atole villages
had significantly better nutritional status at 3 y of age than the
Fresco villages (Schroeder et al. 1995).
Overall, 65.0 and 62.1% of women had experienced first
intercourse and first birth at follow-up, respectively. The me-
dian time intervals from menarche to first intercourse and from
first intercourse to first birth were 5.67 and 0.95 y, respectively.
The median time from menarche to first birth was 6.32 y and
the median age at first birth was 19.83 y.
Comparison of key outcomes by type of supplement (Table
2) indicates that the median time intervals from menarche to
first intercourse and from first intercourse to first birth were
1.68 and 0.06 y shorter among those who received Atole than
those who received Fresco; median age at first birth was 1.17 y
earlier. However, there were strong between-village differ-
ences, especially for the time from menarche to first inter-
course, and from menarche to first birth. Specifically, all fer-
tility outcomes were significantly delayed in the small Fresco
village. In this village, only 42 and 38% of women had
experienced first intercourse and first birth, respectively. The
median interval from menarche to first intercourse was 0.5 y
earlier among those who were not severely stunted compared
with those who were. A similar difference was seen for the
time from menarche to first birth. Median age at first birth was
1.04 y earlier among those who were not severely stunted
compared with those who were. These differences, however,
were not significant (P 5 0.25).
Time from menarche to first intercourse was ;2.11 y
greater for women whose fathers were literate compared with
those with illiterate fathers; the median age at first intercourse
was 19.75 and 18.17 y, respectively (P 5 0.005). The time
from first intercourse to first birth was slightly smaller (0.1 y; P
 FERTILITY MILESTONES AND EARLY CHILDHOOD NUTRITION 2199

TABLE 2
Median values for various fertility milestones by education, severe stunting, village and type of supplement1

First Menarche
Age at Menarche to first intercourse to to first Age at first
Independent variable menarche, y intercourse, y first birth, y birth, y birth, y

Type of Supplement
Atole 13.58 (127) 4.67 (127)* 0.91 (59)* 5.73 (127)* 19.46(127)*
Fresco 13.50 (113) 6.35 (113) 0.97 (97) 7.13 (113) 20.63(113)
Village
Large Fresco 13.42 (56) 6.92 (56)* 0.88 (35) 7.79 (56)* 21.57 (56)*
Large Atole 13.92 (70) 4.67 (70) 0.98 (57) 5.54 (70) 19.11 (70)
Small Fresco 13.58 (57) NA (57)2 0.93 (24) NA (57)2 NA (57)2
Small Atole 13.58 (57) 4.67 (57) 0.97 (40) 6.11 (57) 19.72 (57)
Early childhood nutritional status
Not severely stunted 13.50 (156) 5.50 (156) 0.97 (105) 6.98 (156) 20.30 (156)
Severely stunted 13.75 (84) 6.00 (84) 0.94 (51) 7.45 (84) 21.34 (84)
Paternal literacy
Illiterate father 13.33 (90) 4.56 (90)* 1.02 (72)* 5.87 (90) 19.40 (90)*
Literate father 13.67 (150) 6.67 (150) 0.89 (84) 7.91 (150) 21.67 (150)
Schooling
Not completed primary school 13.75 (146) 4.25 (146)* 0.99 (113) 5.47 (146)* 19.35 (146)*
Completed primary school 13.42 (94) 10.42 (94) 0.87 (43) 11.18 (94) 23.62 (94)

Downloaded from jn.nutrition.org by guest on June 13, 2013

1 Values are medians; sample sizes are shown in parentheses; *denotes P , 0.05 for comparison of Kaplan-Meier curves for each independent
variable.
2 Median values are not available because fewer than 50% of subjects reached the milestone.

, 0.05) among women with literate fathers compared with
those with illiterate fathers, but time from menarche to first
birth and age at first birth were delayed by ;2 y among those
with literate fathers. The effects of schooling were greater. The
median time from menarche to first intercourse was 6.17 y
greater for women who had completed primary school com-
pared with those who had not (Table 2); the median age at
first intercourse was 20.75 and 18.13 y, respectively (P
5 0.007). Schooling had no effect on time from first inter-
course to first birth. Overall, there was evidence of a strong
effect of education, especially for completing primary school,
in delaying the occurrence of first birth.
After adjusting for concurrent covariates (current SES and
completion of primary schooling), the hazard ratio for Atole
for time from menarche to first intercourse was 1.98 [95%
confidence interval (CI):1.38, 2.84]. Adjustment for anteced-
ent covariates (parental literacy and SES during early child-
hood) yielded similar estimates (Table 3). The hazard ratio for
the time from first intercourse to first birth was 1.12 (95% CI:
1.10, 1.23), adjusting for the concurrent covariates. The ad-
justed hazard ratios (AHR) for the cumulative time intervals
(time from menarche to first birth and age at first birth) were
similar to those reported for the time from menarche to first
intercourse. There were no significant interactions between
type of supplement and education.
In the adjusted models, paternal illiteracy was significantly
associated with shorter time from menarche to first inter-
course, but longer time intervals from first intercourse to first
birth. The AHR for schooling indicated that not completing
primary school was significantly associated with earlier out-
comes in all models except for the time interval from first
intercourse to first birth. SES was not significantly associated
with the attainment of milestones.
There was a significant interaction between education and
severe stunting during early childhood for time from menarche
to first intercourse but not for time from first intercourse to first
birth. Better growth, i.e., not being severely stunted at or
around 3 y of age, led to earlier milestones only among women

TABLE 3
Effects of supplementation and severe stunting during early childhood on fertility milestones1

Menarche to first First intercourse to Menarche to first

# Model1,2 intercourse2 first birth3 birth2 Age at first birth2

1. Type of supplement
Paternal illiteracy
2. Type of supplement
No primary school
3. Severe stunting
Paternal illiteracy
4. Severe stunting
No primary school
1.98 (1.38, 2.84)* 1.16 (1.07, 1.24)* 2.03 (1.46, 2.82)* 1.95 (1.35, 2.82)*
1.65 (1.06, 2.55)* 0.68 (0.54, 0.85)* 1.46 (0.88, 2.43) 1.48 (0.89, 2.46)
1.80 (1.37, 2.38)* 1.12 (1.01, 1.23)* 1.85 (1.39, 2.46)* 1.76 (1.30, 2.39)*
2.29 (1.30, 4.04)* 1.00 (0.90, 1.10) 2.43 (1.54, 3.84)* 2.50 (1.49, 4.19)*
0.85 (0.67, 1.06) 0.77 (0.57, 1.05) 0.83 (0.70, 0.99)* 0.86 (0.69, 1.07)
1.74 (1.11, 2.73)* 0.65 (0.53, 0.79)* 1.56 (0.90, 2.70) 1.57 (0.90, 2.73)
0.79 (0.58, 1.08) 0.83 (0.55, 1.27) 0.77 (0.63, 0.95)* 0.80 (0.62, 1.05)
2.58 (1.33, 5.02)* 1.06 (0.92, 1.22) 2.75 (1.55, 4.89)* 2.82 (1.51, 5.28)*

1 Values are adjusted hazard ratios with 95% confidence intervals in parentheses for type of supplement (Atole 5 1; Fresco 5 0) or severe stunting
(Yes 5 1; No 5 0); *denotes P , 0.05.
2 Models incorporate adjustments for socioeconomic status, village and age at menarche (n 5 240).
3 Models incorporate adjustments for socioeconomic status, village and age at first intercourse (n 5 156).
2200 RAMAKRISHNAN ET AL.

TABLE 4
Joint effects of severe stunting and education on time from menarche to first intercourse, time from menarche
to first birth and age at first birth1

Time from menarche to first Time from menarche to first

intercourse birth Age at first birth

Model n Median, y AHR (95% CI) Median, y AHR (95% CI) Median, y AHR (95% CI)

Severe stunting and paternal illiteracy

Severely stunted and illiterate father 25 6.00 1.31 (0.71, 2.41)* 7.11 1.16 (0.57, 2.38)* 19.61 1.20 (0.56, 2.58)*
Severely stunted and literate father 59 6.42 1.00 (0.76, 1.31) 7.45 0.97 (0.76, 1.22) 21.43 1.01 (0.75, 1.36)
Not severely stunted and illiterate father 65 4.50 1.97 (1.13, 3.44) 5.73 1.74 (0.94, 3.23) 19.35 1.77 (0.94, 3.32)
Not severely stunted and literate father 91 7.25 — 7.95 — 22.00 —
Severe stunting and schooling
Severely stunted and no primary school 57 4.42 1.96 (0.91, 4.23)* 5.59 1.99 (1.08, 3.68)* 19.51 2.12 (1.03, 4.39)*
Severely stunted and primary school 27 NA2 0.26 (0.16, 0.41) NA2 0.23 (0.12, 0.46) NA2 0.24 (0.11, 0.51)
Not severely stunted and no primary school 89 4.08 1.86 (1.16, 2.99) 5.35 2.00 (1.32, 3.05) 19.21 2.05 (1.29, 3.25)
Not severely stunted and primary school 67 7.58 — 8.51 — 22.13 —

1 All models incorporate adjustment for socioeconomic status, village and age at menarche (n 5 240); *denotes P , 0.05 for the overall interaction
term between stunting and paternal illiteracy in Model 1 and between stunting and schooling in Model 2 for all 3 outcomes.
2 Median values are not available because fewer than 50% of subjects reached the milestone.
AHR, adjusted hazard ratio; CI, confidence interval.

Downloaded from jn.nutrition.org by guest on June 13, 2013

with illiterate fathers (Table 4). The effect of paternal literacy
in delaying fertility milestones was greater among those who
were not severely stunted during early childhood. The median
time from menarche to first intercourse was 2.75 y greater for
subjects with literate fathers compared with those with illit-
erate fathers among those not severely stunted during early
childhood. The difference associated with paternal illiteracy
was only 0.42 y among those who were severely stunted during
early childhood. Similar results were seen for time from men-
arche to first birth and age at first birth. Completion of primary
schooling by women in the study increased the time from
menarche to first intercourse, with a larger effect among those
who were severely stunted during early childhood. Similar
effects were seen for the cumulative time intervals. Overall,
completing primary schooling delayed the median age at first
birth by at least 3 y.
DISCUSSION
This study is the first to examine the relationship between
early childhood nutrition and fertility outcomes using prospec-
tively collected data in a developing country. The mean age at
first birth and time from menarche to first birth are similar to
those reported elsewhere in Guatemala (Delgado et al. 1985,
Engle et al. 1996, Macro International 1990) and the Central
American region (Morabia and Constanzia 1998). We found
that exposure to food supplementation resulted in shorter time
intervals between menarche, first intercourse and first birth
and overall earlier occurrence of first birth. Better growth, i.e.,
not being severely stunted during early childhood led to earlier
milestones, but these differences were significant only among
women with illiterate fathers. Women who were not severely
stunted during early childhood had later outcomes compared
with those who were severely stunted, if their fathers were
literate. Completion of primary school was associated with
significantly later milestones.
Lumey and Stein (1997), found that poor nutrition in utero
(measured by ecological exposure to famine) did not affect age
at first birth in Dutch women. However, several differences in
study design (natural experiment vs. intervention trial; in
utero vs. in utero and early childhood exposure), study setting
(differences in the prevalence of malnutrition and other social
factors that determine fertility milestones such as age at first
birth) and age at follow-up (older vs. younger) between the
two studies may account for these differences between that
study and ours.
Earlier work by Khan et al. (1995) found no significant
effects of the type of supplement consumed during early child-
hood on age at menarche. The current findings, however,
show that women who had received Atole in utero and/or
during early childhood experienced first intercourse or first
birth at an earlier age than those who received Fresco, even
after adjusting for socioeconomic status and paternal illiteracy
during early childhood. The effect of type of supplement was
seen for both times, i.e., time from menarche to first inter-
course and time from first intercourse to first birth. The effect
of supplementation may be due in part to village-level differ-
ences in other known and unknown nonnutritional or behav-
ioral/cultural factors. Our results show that although age at
menarche was similar across villages, the more culturally de-
termined fertility milestones, such as the time interval from
menarche to first intercourse, were significantly later in the
small Fresco village. Earlier work has also shown that this
village is quite distinct both socially and ecologically from the
other villages (Bergeron 1993). This village had a strong
tradition of higher schooling that dates back to even before
the intervention study. Both paternal literacy and completion
of primary schooling were the highest in this village. Unfor-
tunately, we lack the data to examine conclusively whether
these village-level differences in fertility milestones existed
before the original intervention study. Although the village-
level allocation of treatment was appropriately accounted for
in our analysis, the small sample size (two pairs of villages)
limits our ability to control for these inherent village-level
differences.
In the case of severe stunting, previous work by Khan et al.
(1996) using data from the same INCAP cohort, found a
difference of ;7 mo in the mean age at menarche between
subjects who were severely stunted (HAZ , 23) at 3 y of age
compared with those who were not stunted (HAZ . 22).
However, we found a smaller nonsignificant difference (3 mo;
P 5 0.1), which may be due to the small sample size and the
 FERTILITY MILESTONES AND EARLY CHILDHOOD NUTRITION
fact that we compared two groups, i.e., severely stunted or not,
instead of three groups as in Khan et al. (1996).
Our results demonstrate the importance of social determi-
nants, especially education, in delaying fertility milestones.
Paternal literacy was associated with a longer time from men-
arche to first intercourse, but was not associated with earlier
first birth after adjusting for other covariates. Completion of
primary school had a larger effect and significantly delayed the
timing of first intercourse and first birth, even after adjusting
for socioeconomic status and age at menarche. It should be
noted, however, that schooling delayed only the time to first
intercourse from menarche, but not the time to first birth from
first intercourse. Contraceptive use was extremely low in this
population, which may explain the lack of effect of schooling
on the second interval. The other reason may be related to the
accuracy of recall of age at first intercourse. The interval from
first intercourse to first birth is short, suggesting that women
may have reported age at first intercourse based on the timing
of first pregnancy and/or birth. Age at menarche was not
associated with completion of primary school, which suggests
that it is unlikely that early maturers dropped out of school
earlier; rather, we hypothesize that either better school per-
formance or staying in school longer were associated with a
reduced risk of having sex. This finding is consistent with
previous studies that have examined the role of social factors
in predicting fertility milestones (Ann et al. 1983, Castro
1995, dos Santos Silva and Beral 1997, Engle and Smidt
1996).
A major contribution of our study is evidence of a strong
interaction between early childhood stunting and education.
Our findings suggest that the observed effect of stunting on
time from menarche to first intercourse may also be due to
behavioral rather than biological mechanisms. Better growth
led to earlier outcomes only among those whose fathers were
illiterate, suggesting familial influences. On the other hand,
shorter girls who remain in school may be perceived as being
less attractive and less mature for childbearing, and therefore
have later outcomes.
In summary, this paper suggests that improved nutrition,
especially food supplementation during early childhood, re-
sults in earlier attainment of fertility milestones. However,
these effects may be countered by social factors, especially
education. The effect of completing primary school in delaying
fertility milestones was larger in magnitude than that of im-
proving nutrition. Because they come from only one setting
and are based on relative small samples, these findings should
be considered with caution, and more work is required to
improve our understanding of the underlying biological and
social mechanisms that explain the sequence of events that
lead from early childhood to age at first birth. In the mean-
time, programs that aim to improve early childhood nutrition
should be implemented in conjunction with efforts to make
sure that every child attends and stays in school. It should also
be noted that earlier work has shown that “educability,” i.e.,
performance on tests of knowledge, numeracy, reading and
vocabulary that were administered during adolescence, im-
proved among those who received Atole compared with
Fresco, while controlling for the level of schooling (Pollitt et
al. 1995). To benefit from improved educability, investments
in education such as the provision of good schools as well as
improving the availability of family planning services would be
required to see long-term benefits. In conclusion, the combi-
nation of improved nutrition and better education will result
not only in later fertility milestones, but also healthier and
better-nurtured children.
2201
ACKNOWLEDGMENTS
This study would not have been possible without the strong
institutional support and role of INCAP in coordinating and orga-
nizing data collection and entry over the years. The participation of
the women and families in the study villages is gratefully acknowl-
edged. The contributions of Joan Herold, Associate Professor, De-
partment of Behavioral Sciences and Health Education at the Rollins
School of Public Health in designing and providing oversight for data
collection and management for the 1994 fertility survey and the
assistance of Clark Denny and Morgen Hughes in data management
and analysis are gratefully acknowledged. Finally, the valuable con-
tributions and recommendations made by the anonymous reviewers
are recognized.
LITERATURE CITED
Ann, T. B., Othman, R., Butz, W. P. & DaVanzo, J. (1983) Age at menarche in
peninsular Malaysia: time, ethnic differentials and association with ages at
marriage and at first birth. Malays, J. Reprod. Health 1: 91–108.
Balakrishnan, T. R, Rao, K. V., Krotki, K. J. & Lapierre-Adamcyk, E. (1988) Age
at first birth and lifetime fertility. J. Biosoc. Sci. 20: 167–174.
Bergeron, G. (1993) Social and economic development in four Ladino com-
munities of eastern Guatemala. A comparative description. Food Nutr. Bull.
14: 221–236.
Downloaded from jn.nutrition.org by guest on June 13, 2013
Bongaarts, J. (1980) Does malnutrition affect fecundity ? A summary of evi-
dence. Science (Washington, DC) 208: 564 –569.
Brewster, K. L., Cooksey, E. C., Guilkey, D. K. & Rindfuss, R. R. (1998) The
changing impact of religion on sexual and contraceptive behavior of adoles-
cent women in the United States. J. Marriage Fam. 60: 493–504.
Castro, M. T. (1995) Women’s education and fertility: results from 26 Demo-
graphic and Health Surveys. Stud. Fam. Plann. 26: 187–202.
Chowdhury, A.K.M.A., Huffman, S. L. & Curlin, G. T. (1978) Malnutrition,
menarche and marriage in rural Bangladesh. Soc. Biol. 24: 316 –325.
Delgado, H. L, Hurtado, E. & Valverde, V. (1985) Physical growth, age at
menarche and age at first union in rural Guatemala. Ecol. Food Nutr. 16:
127–133.
DeWit, M. L. & Rajulton, F. (1992) Education and timing of parenthood among
Canadian women: a cohort analysis. Soc. Biol. 39: 109 –122.
dos Santos Silva, I. & Beral, V. (1997) Socioeconomic differences in repro-
ductive behavior. IARC Sci. Publ. 138: 285–308.
Engle, P. L. & Smidt, R. K. (1996) Influences of Adolescent Childbearing and
Marital Status at First Birth on Rural Guatemalan Women and Children. The
Population Council, New York, N.Y. International Center for Research on
Women, Washington DC.
Fleming, T. R. & Harrington, D. P. (1991) Counting Processes and Survival
Analysis. John Wiley & Sons, New York, NY.
Freedman, D. S., Thornton, A. & Wallisch, L. (1981) Age at first birth and family
size: evidence from a longitudinal study. Soc. Biol. 28: 217–227.
Frisch, R. (1972) Weight at menarche: similarities for well nourished and
undernourished girls at different ages and evidence for historical constancy.
Pediatrics 50: 445– 450.
Frisch, R. E. & McArthur, J. W. (1974) Menstrual cycle: fatness as a determi-
nant of minimum weight for height necessary for their maintenance or onset.
Science (Washington, DC) 185: 949 –951.
Hillman, R. W., Slater, P. & Nelson, M. J. (1971) Season of birth, parental age
menarcheal age and body form: some interrelationships in young women.
Hum. Biol. 42: 570 –580.
Khan, A. D., Schroeder, D. G., Martorell, R., Haas, J. D. & Rivera, J. (1996)
Early childhood determinants of age at menarche in rural Guatemala. Am. J.
Hum. Biol. 8: 717–723.
Khan, A. D., Schroeder, D. G., Martorell, R. & Rivera, J. A. (1995) Age at
menarche and nutritional supplementation. J Nutr. 125: 1090S–1096S.
Lipsitz, S. T., Dear, K.B.G. & Zhao, L. (1994) Jacknife estimators of variance for
parameter estimates from estimating equations with applications to clustered
survival data. Biometrics 50: 842– 846.
Lumey, L. H. & Stein, A. D. (1997) In utero exposure to famine and subsequent
fertility: the Dutch Famine Birth Cohort Study. Am. J. Pub. Health. 87: 1962–
1966.
Macro International Inc. (1990) Guatemala 1987: results from the demographic
and health survey. Stud. Fam. Plann. 21: 55–59.
Martorell, R. (1996) The role of nutrition in economic development. Nutr. Rev.
54: S66 –S71.
Martorell, R., Delgado, H. L., Valverde, V. & Klein, R. E. (1981) Maternal stature,
fertility and infant mortality. Hum. Biol. 53: 303–312.
Martorell, R., Habicht, J.-P. & Rivera, J. A. (1995a) History and design of the
INCAP longitudinal study (1969 –77) and its follow up (1988 – 89). J. Nutr. 125:
1027S–1041S.
Martorell, R., Schroeder, D. G., Rivera, J. A. & Kaplowitz, H. A. (1995b) Pat-
terns of linear growth in rural Guatemalan adolescents and children. J. Nutr.
125: 1068S–1077S.
MathSoft Inc. (1996) S-Plus Software, Version 3.4. MathSci, Seattle, WA.
Morabia, A. & Constanza, M. C. (1998) World Health Organization collaborative
2202 RAMAKRISHNAN ET AL.
study of neoplasia and steroid contraceptives. International variability in ages at
menarche, first livebirth and menopause. Am. J. Epidemiol. 148: 1195–1205.
Nath, D. C., Land, K. C. & Goswami, G. (1999) Effects of the status of women
on the first-birth interval in Indian urban society. J. Biosoc. Sci. 31: 55– 69.
Nguyen, L., Nguyen, M. T., Swenson, I. & Pham, B. S. (1993) Selected
determinants of fertility in Vietnam: age at marriage, marriage to first birth
interval and age at first birth. J. Biosoc. Sci. 25: 303–310.
Ostersehlt, D. & Danker Hopfe, H. (1991) Changes in age at menarche in
Germany: evidence for a continuing decline. Am. J. Hum. Biol. 3: 647– 654.
Pollitt, E., Gorman, K. S., Engle, P. L., Rivera, J. A. & Martorell, R. (1995)
Nutrition in early life and the fulfillment of intellectual potential. J. Nutr. 125:
1111S–1118S.
Prado, C. (1986) Secular change and genetic influence in the menarcheal age.
Acta Med. Auxol. (Milan) 18: 117–121.
Rahman, M. (1998) The effect of child mortality on fertility regulation in rural
Bangladesh. Stud. Fam. Plann. 29: 268 –281.
Rivera, J., Flores, M., Martorell, R., Ramakrishnan, U. & Melgar, P. (1996)
Generational effects of supplementary feeding during early childhood. In: 5th
International Workshop on Maternal and Extrauterine Nutritional Factors.
Their Influence on Fetal and Infant Growth, Universidad de Salamanca, Spain
(Battaglia, F., Falkner, F., Garza, C., Salle, B. & Rey, J. eds.), pp. 197–204.
Ediciones Ergon, Madrid, Spain.
Rivera, J., Martorell, R., Ruel, M., Habicht, J.-P. & Haas, J. D. (1995) Nutritional
supplementation during the preschool years influences body size and com-
position of Guatemalan adolescents. J. Nutr. 125: 1078S–1089S.
Rubin, D. B. (1987) Multiple Imputation for Nonresponse in Surveys. John
Wiley & Sons, New York, NY.
View publication stats
SAS Institute Inc. (1997) SAS/STAT Software: Change and Enhancements
through Release 6.12. SAS Institute, Cary, NC.
Schroeder, D. G., Martorell, R., Rivera, J., Ruel, M. & Habicht, J.-P. (1995) Age
differences in the impact of nutritional supplementation on growth. J. Nutr.
125: 1051S–1067S.
Stein, A. D. & Kline, J. (1991) Pre-pregnant body size and spontaneous
abortion of known karyotype. Early Hum. Dev. 25: 173–180.
Tanner, J. M. (1968) Earlier maturation in man. Sci. Am. 218: 21–27.
Tanner, J. M. (1975) Age at menarche. Evidence of the rate of human matu-
ration, unpublished paper presented at a meeting of the British Society for
Population Studies, London School of Economics, England.
United Nations International Children’s Emergency Fund (1990) Strategy for
improved nutrition of children and women in developing countries. UNICEF,
New York, NY.
Veronesi, F. M. & Gueresi, P. (1994) Trend in menarcheal age and socioeco-
nomic influence in Bologna (Northern Italy). Ann. Hum. Biol. 21: 187–196.
Wolanski, N. (1985) Secular trend, secular changes or long-term adaptational
fluctuations? Acta Med. Auxol. (Milan) 17: 7–19.
World Health Organization (1986) Use and interpretation of anthropometric
indicators of nutritional status, WHO Working Group. Bull. WHO 64: 929 –
941.
Zacharias, L. W., Rand, W. M. & Wurtman, R. (1976) A prospective study of
sexual development and growth in American girls: the statistics of menarche.
Obstet. Gynecol. Surv. 31: 325–337.
Zacharias, L. W. & Wurtman, R. (1969) Age at menarche: genetic and envi-
ronmental influences. N. Engl. J. Med. 280: 868 – 875.
Downloaded from jn.nutrition.org by guest on June 13, 2013