Effects of maternal micronutrient supplementation on fetal loss and
infant mortality: a cluster-randomized trial in Nepal1–3
Parul Christian, Keith P West Jr, Subarna K Khatry, Steven C Leclerq, Elizabeth K Pradhan, Joanne Katz,
Sharada Ram Shrestha, and Alfred Sommer
ABSTRACT
Background: We previously reported that maternal micronutrient
supplementation in rural Nepal decreased low birth weight by
⬇15%.
Objective: We examined the effect of daily maternal micronutri-
ent supplementation on fetal loss and infant mortality.
Design: The study was a double-blind, cluster-randomized, con-
trolled trial among 4926 pregnant women and their 4130 infants in
rural Nepal. In addition to vitamin A (1000 ␮g retinol equivalents),
the intervention groups received either folic acid (FA; 400 ␮g), FA
⫹ iron (60 mg), FA ⫹ iron ⫹ zinc (30 mg), or multiple micro-
nutrients (MNs; the foregoing plus 10 ␮g vitamin D, 10 mg
vitamin E, 1.6 mg thiamine, 1.8 mg riboflavin, 2.2 mg vitamin B-6,
2.6 ␮g vitamin B-12, 100 mg vitamin C, 64 ␮g vitamin K, 20 mg
niacin, 2 mg Cu, and 100 mg Mg). The control group received
vitamin A only.
Results: None of the supplements reduced fetal loss. Compared
with control infants, infants whose mothers received FA alone or
with iron or iron ⫹ zinc had a consistent pattern of 15–20% lower
3-mo mortality; this pattern was not observed with MNs. The
effect on mortality was restricted to preterm infants, among whom
the relative risks (RRs) were 0.36 (95% CI: 0.18, 0.75) for FA,
0.53 (0.30, 0.92) for FA ⫹ iron, 0.77 (0.45, 1.32) for FA ⫹ iron
⫹ zinc, and 0.70 (0.41, 1.17) for MNs. Among term infants, the
RR for mortality was close to 1 for all supplements except MNs
(RR: 1.74; 95% CI: 1.00, 3.04).
Conclusions: Maternal micronutrient supplementation failed to
reduce overall fetal loss or early infant mortality. Among preterm
infants, FA alone or with iron reduced mortality in the first 3 mo
of life. MNs may increase mortality risk among term infants, but
this effect needs further evaluation. Am J Clin Nutr 2003;78:
1194–1202.
KEY WORDS Micronutrients, pregnancy, supplementation,
fetal loss, infant mortality
INTRODUCTION
Infant mortality reduction remains a major public health goal
throughout the developing world. Postneonatal mortality has
declined globally, in contrast with neonatal mortality, which
constitutes one-half of all infant deaths in many developing
countries (1, 2). In general, populations with high infant mor-
tality also have a high prevalence of low birth weight (⬍ 2.5
kg) (3), a condition that predisposes newborns to increased
neonatal mortality (4-6) and morbidity (7-10).
1194 Am J Clin Nutr 2003;78:1194 –1202. Printed in USA. © 2003 American Society for Clinical Nutrition
There is little causal evidence on the effect of maternal
micronutrient supplementation on pregnancy outcome and in-
fant health and survival in the developing world. Although
prenatal multivitamin and mineral supplements are commonly
consumed in developed countries, this practice is less common
in developing countries, in which existing antenatal iron-folate
programs achieve low coverage and have been ineffective in
reducing maternal anemia (11). In fact, the issue of a lack of
efficacy and safety of routine iron supplementation has been
raised (12). Both low and high hemoglobin concentrations
during pregnancy are associated with adverse outcomes, and
yet it is not known whether iron supplementation decreases or
increases the risk of adverse outcomes (13, 14). Of the 23
iron-intervention studies reviewed by Rasmussen (14), none
were free from possible bias, and many suffered from problems
with design and interpretation. Furthermore, in several studies,
the lowest rates of low birth weight and preterm birth occurred
at hemoglobin concentrations that were below the current cut-
offs for anemia during pregnancy (14). Thus, the efficacy of
iron supplementation during pregnancy needs to be urgently
assessed with the use of endpoints such as low birth weight,
preterm delivery, and infant mortality, and not just anemia.
Folic acid is added to antenatal iron but may have an indepen-
dent role in enhancing birth outcomes. An overview of 5
controlled trials suggested a 41% reduction in the prevalence of
intrauterine growth retardation with folic acid supplementation,
1
Department of International Health, Johns Hopkins Bloomberg School
of Public Health, Baltimore (PC, KPW, SCL, EKP, JK, and AS), and the
Nepal Nutrition Intervention Project-Sarlahi (NNIPS), Society for Preven-
tion of Blindness, Kathmandu, Nepal (SKK and SRS).
2
Supported by the US Agency for International Development (USAID),
the UNICEF Country Office, Kathmandu, Nepal, and the Bill and Melinda
Gates Foundation. The study was carried out under Cooperative Agreement
HRN-A-00-97-00015-00 between the Office of Nutrition, USAID, Wash-
ington, DC, and the Center for Human Nutrition (CHN), Department of
International Health, and the Sight and Life Research Institute, Johns
Hopkins University, Bloomberg School of Public Health, Baltimore. The
study was a joint undertaking between the CHN and the National Society
for the Prevention of Blindness, Kathmandu, Nepal. Supplements were
provided by Roche, Brazil, and were manufactured by NutriCorp Interna-
tional, CE Jamieson & Company Ltd, Windsor, Canada.
3
Address reprint requests to P Christian, Center for Human Nutrition,
615 North Wolfe Street, Room 2041, Baltimore, MD 21205. E-mail:
pchristi@jhsph.edu.
Received December 23, 2002.
Accepted for publication June 10, 2003.
 MICRONUTRIENT SUPPLEMENTATION AND INFANT MORTALITY
although these trials were small and their study designs have
been called into question (15). Trials in Bangladesh (16) and
Peru (17) failed to confirm the improvement in birth weight
found in previous studies of antenatal zinc supplementation
(18, 19). A move toward multiple micronutrient supplementa-
tion is currently taking place. UNICEF has recently developed
such a supplement, which is catered to pregnant women in
developing countries and is being proposed for wide-scale use
and distribution. Yet, the benefits of such a supplement are not
well established.
In a randomized trial of women in a poor, malnourished,
rural population in Nepal, we previously reported the effects on
birth weight of daily consumption of 4 different combinations
of antenatal micronutrient supplements (20). The treatment
arms were as follows: control, folic acid, folic acid ⫹ iron,
folic acid ⫹ iron ⫹zinc, and folic acid ⫹ iron ⫹ zinc ⫹ 11
other vitamins and minerals. Each supplement type also con-
tained vitamin A, which we previously found to reduce the risk
of maternal mortality (21) but to have no effect on neonatal
weight (KP West Jr, unpublished observations, 2003) or infant
mortality (22). Compared with the control group, only the folic
acid ⫹ iron group and the multiple micronutrient supplement
group had lower incidences of low birth weight (16% [relative
risk (RR): 0.84; 95% CI: 0.72, 0.99] and 14% (RR: 0.86; 95%
CI: 0.74, 0.99) lower, respectively); none of the treatments had
any effect on preterm delivery (20).
In this article, we report from the same study on the effect of
these alternative micronutrient supplement regimens on fetal
loss and perinatal, neonatal, and 3-mo mortality. The objective
was to determine whether a micronutrient intervention that
decreased low birth weight by ⬇15% also improved survival
among infants. Although the treatments failed to prevent pre-
term birth, we examined treatment effects on mortality strati-
fied by preterm (⬍ 37 wk of gestation) and term births to
examine possible interactions.
SUBJECTS AND METHODS
Study design and population
The study was a cluster-randomized, double-blind trial that
featured an active control group and was conducted in the rural
plains district of Sarlahi, Nepal, to examine the effect of prenatal
and postnatal maternal supplementation on birth weight, fetal loss,
and early infant mortality. This is the same area of Nepal in which
we previously recorded evidence of vitamin A, iron, and zinc
deficiency among pregnant women. The study area, comprising
30 village development communities with a total population of
⬇200 000, was divided into 426 smaller communities called “sec-
tors,” which served as the units of randomization. Randomization
was done in blocks of 5 within each village development com-
munity by the senior study investigators, who drew numbered
chips from a hat. At the outset, women who were sterilized,
widowed, menopausal, or breastfeeding an infant ⬍ 9 mo of age
were considered to have a low risk of becoming pregnant and
were excluded from the study. The remaining women were visited
once every 5 wk by 426 locally hired female workers (called
“sector distributors”), who ascertained pregnancies by offering a
human chorionic gonadotropin–based urine test to those who
reported having not menstruated in the previous 30 d. A positive
urine test and informed consent resulted in a woman being en-
1195
rolled into the study to receive supplements. Enrollment of preg-
nant women took place from January 1999 through February
2000. A baseline interview was conducted to obtain the date of the
last menstrual period and diet, morbidity, and work histories in the
preceding week and to collect data on socioeconomic status
and reproductive history. Gestational age was calculated by using
the date of the last menstrual period, which was verified against
the prospectively collected data on menstrual history as well as the
date of the positive pregnancy test. Anthropometric assessment,
including measurements of weight, height, and midupper arm
circumference, was also performed at baseline. Assessment at
birth of all liveborn infants included anthropometric measure-
ments (within 72 h) and a history of labor and delivery and
symptoms in the newborn immediately after birth.
Ethical approval for the study was provided by the National
Health Research Council of the Ministry of Health, Kathmandu,
Nepal, and the Committee for Human Research at the Johns
Hopkins Bloomberg School of Public Health, Baltimore. A Data
Safety and Monitoring Committee meeting took place midway
through the study, and continuation of the study was approved.
Intervention
The sectors were randomly assigned to 1 of 5 treatment arms.
All the women in the study received vitamin A because it reduced
pregnancy-related mortality by 40% (21) but did not affect fetal
loss or infant mortality (22). The 5 treatment arms were as fol-
lows: control, vitamin A (1000 ␮g retinol equivalents); FA, vita-
min A ⫹ folic acid (400 ␮g); FAFe, vitamin A ⫹ folic acid ⫹ iron
(60 mg); FAFeZn, vitamin A ⫹ folic acid ⫹ iron ⫹ zinc (30 mg);
MN, vitamin A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients
(10 ␮g vitamin D, 10 mg vitamin E, 1.6 mg thiamine, 1.8 mg
riboflavin, 20 mg niacin, 2.2 mg vitamin B-6, 2.6 ␮g vitamin
B-12, 100 mg vitamin C, 65 ␮g vitamin K, 2.0 mg Cu, 100 mg
Mg).
The supplements, which were of identical shape, size, and
color, arrived in Nepal in opaque, sealed, and labeled bottles
coded 1–5. The code allocation was kept locked at the Johns
Hopkins University, Baltimore. The investigators, field staff, and
participants were blinded to the codes throughout the study.
At the outset of the study, each pregnant woman received a
small bottle containing 15 supplement caplets, with instruc-
tions from the staff to take 1 supplement/d. The women were
subsequently visited twice each week by the sector distributors,
who replenished the caplets, monitored their consumption, and
collected data on pregnancies and their outcomes. The supple-
ments were to be taken daily from the time of pregnancy
detection through 3 mo postpartum in the case of a live birth
and through ⱖ 5 wk after a miscarriage or stillbirth.
A sample size estimate of 1000 pregnant women per group
was based on detecting an increase in birth weight of ⱖ 75 g
with 80% power after a 10% fetal loss, a 15% loss to follow-up,
and a 20% design effect estimated from our previous study (21,
22; KP West Jr unpublished observations, 2003) were ac-
counted for. With this sample size and an assumption that the
fetal loss rate and the infant mortality rate among the control
subjects were 10% and 75 deaths/1000 live births, respectively
(22), a reduction in fetal loss of ⱖ 34% and a reduction in
infant mortality of ⱖ 45% would be detected with a type 1
error of 5% and 80% power.
During the first 3 mo of life, the vital status of the infants
was assessed every week. For all the liveborn infants and their
1196 CHRISTIAN ET AL
mothers, a 6-mo postpartum home visit was conducted to
assess vital status. The study was completed in April 2001
when the last infant born in the study had been followed
through 6 mo of life. The vital status of all the infants was also
assessed at that time. At the end of the study, ⬇70% of the
infants across all treatment groups had been followed for
ⱖ 364 d. Miscarriage was defined as a pregnancy that ended in
a fetal loss before 28 wk of gestation. We defined perinatal
mortality to include stillbirths (gestational age of ⱖ 28 wk) and
deaths through the first 7 d of life. This definition also included
stillbirths (n ⫽ 8) that accompanied a liveborn infant among
twin births. Neonatal mortality was defined as deaths among
live births through 28 d of life.
Data from verbal autopsy interviews and infant morbidity as-
sessments conducted on the day of birth were used to assign the
cause of death. A verbal autopsy was usually performed within a
week after an infant death was reported. Thirty well-trained and
experienced workers, who had previously conducted interviews
by using the same verbal autopsy instrument for ⬇800 infant
deaths in our previous study, were responsible for administering
the verbal autopsy interviews in the present study. Causes of death
were elicited by using a 2-step procedure. First, 2 physicians
conducted an independent review of verbal autopsy interviews
and assigned a cause of death. Differences in the assigned causes
were reconciled by discussion between the reviewers. We were
unable to assign causes for 84 (of 174) neonatal deaths by using
this process. The number of cases for which we did not have a
cause of death did not differ by treatment group. For 59 of these
84 deaths, illness symptoms at birth were further used to assign a
cause of death on the basis of published algorithms (23). These
causes of death were prematurity or severe malnutrition and birth
asphyxia. Death was attributed to prematurity or severe malnutri-
tion if the gestational age was ⬍ 32 wk or the birth weight was
⬍ 1.5 kg. Death was attributed to birth asphyxia if 1) the infant
died within 7 d and did not cry or cried weakly at birth and was
not able to breathe after birth, or the infant had convulsions or was
unable to suckle; 2) the infant died within 7 d and either had
convulsions or had a breech presentation, or the woman had
prolonged labor (⬎ 24 h) or obstructed labor; or 3) the infant died
within the first day of life and had a blue body or pale extremities.
Cause-specific neonatal mortality rates were calculated by treat-
ment group.
Health care
All the pregnant women in the study received counseling on
antenatal care and nutrition at the time of enrollment. The
women were encouraged to visit health posts and to take iron
supplements during pregnancy. The policy in Nepal is for all
pregnant women visiting health posts to receive iron supple-
ments, although coverage and adherence are poor (24). Be-
cause we previously found that hookworm was one of the
strongest etiologic factors for anemia (25), we provided de-
worming medicine to all the pregnant women in the second and
third trimesters of pregnancy. The women received a tetanus
toxoid vaccination twice during pregnancy, a safe birthing kit
for home-based delivery from the United Nations Children’s
Fund, and a flannel blanket for their newborns. In a 20%
subsample, from whom venous blood was drawn at baseline
and in the third trimester to measure biochemical indicators of
nutritional status, all severely anemic women (hemoglobin
⬍ 70 g/L) at both times received 90 d of treatment with 120 mg
Fe and 400 ␮g folic acid.
Statistical analysis
Characteristics of the pregnant women and their compliance
with supplementation, which was defined as the proportion of all
eligible doses that were consumed, were compared across treat-
ment groups. All analyses were done on an intention-to-treat basis.
Multiple births in the study (n ⫽ 42) were included in the analysis
because their inclusion did not alter the study results.
Kaplan-Meier survival analysis (26) was used to estimate
and compare differences in the survival probabilities of live-
born infants from birth through 364 d of life by treatment
group. Infants were censored from the survival analysis at the
time when their vital status was last known. Deaths and fol-
low-up time for survivors beyond 364 d were excluded from
the analysis. RRs and 95% CIs for fetal loss and perinatal,
neonatal, and 0–3-mo mortality rates were calculated by using
a generalized estimating equations (27) binomial regression
model with log link and exchangeable correlation to adjust for
the design effect and with the mortality rate among the control
(vitamin A) subjects as the reference category. The vital status
of 8 infants who were lost to follow-up remained unknown, and
they were excluded from the analysis. The same analyses were
repeated for each of the 3 mortality outcomes and for birth
asphyxia after inclusion of an interaction term between treat-
ment group and preterm birth (⬍ 37 wk of gestation) in the
model. The interaction terms were tested at the 5% significance
level. All analyses were performed by using SAS version 6
(SAS Institute Inc, Cary, NC).
RESULTS
Among a total of 36 083 married women of reproductive
age, 14 185 were eligible for inclusion in the routine surveil-
lance for new pregnancies (Figure 1). Mean maternal weight,
height, and body mass index (in kg/m2) were ⬇44 kg, 150 cm,
and 19, respectively, and did not differ by treatment group (20).
A total of 4998 pregnant women (⬇1000 per treatment arm)
were enrolled in the study over a surveillance period of 1 y. Of
these, 72 apparent pregnancies were excluded, including those
that were false positives (n ⫽ 6), had unknown outcomes (n ⫽
3), and ended as induced abortions (n ⫽ 63). A total of 4130
infants were born to the remaining 4926 pregnant women, of
whom all but 8 were followed until the end of the study in April
2001, which represented complete follow up of ⱖ 6 mo. In all,
230 infants died by 6 mo of age, and an additional 26 deaths
occurred by the end of the first year.
Baseline characteristics did not differ significantly by treat-
ment group (Table 1). Nearly one-half of all the mothers were
enrolled before 9 wk of gestation, and only 8–10% were
enrolled after 16 wk of gestation. The median compliance with
supplementation was 86–88% during pregnancy and 70–80%
in the first 12 wk postpartum (data not shown). The level of
compliance did not differ across treatment groups during preg-
nancy but did differ slightly between the groups during the
postpartum period: the median compliance in the control group
was slightly lower (71%) than that in the other groups (77–
80%). As reported previously (20), the RRs for low birth
weight (⬍ 2500 g) were 1.00 (95% CI: 0.88, 1.15), 0.84 (95%
CI: 0.72, 0.99), 0.96 (95% CI: 0.83, 1.11), and 0.86 (95% CI:
 MICRONUTRIENT SUPPLEMENTATION AND INFANT MORTALITY 1197

FIGURE 1. Study design, participation, and follow-up. Ineligible women consisted of those who were currently pregnant, were breastfeeding an infant aged
⬍ 9 mo, had been sterilized, or were menopausal or whose husband had died. Excluded women consisted of those who had a false-positive pregnancy test result,
had unknown outcomes, or had induced abortions. The groups and the supplements that they received were as follows: control, vitamin A; FA, vitamin A ⫹ folic
acid; FAFe, vitamin A ⫹ folic acid ⫹ iron; FAFeZn, vitamin A ⫹ folic acid ⫹ iron ⫹ zinc; MN, vitamin A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients.

0.74, 0.99) for the FA, FAFe, FAFeZn, and MN groups. The
rates of miscarriage did not differ by treatment group and
ranged between 12% and 15% (data not shown).
Kaplan-Meier analysis from birth through the first year of
life showed consistently higher survival among the infants
whose mothers received folic acid with or without added iron
or iron ⫹ zinc than among the infants whose mothers were in
the control group (Figure 2). However, neither the overall nor
the individual treatment-control differences were significant.
Divergence of the mortality curves was most apparent through
the first 60 d of life, after which the curves remained parallel.
The survival probability of the infants in the MN group was not
different, and was possibly slightly worse, than that of the
infants in the control group. Estimates of RRs and 95% CIs for
perinatal, neonatal, and 3-mo mortality among the infants are
provided by treatment group in Table 2. The CIs around the
RR estimates, which ranged from 0.78 to 0.89 for the FA,
FAFe, and FAFeZn groups, included 1. The highest infant
mortality in each period of life was observed in the MN group.
Mortality rates from 3–12 mo were low but showed a pattern in
relation to maternal supplementation that was similar to that
observed in the 0–3-mo period (data not shown).
Stratified analyses showed an interaction (overall P ⬍ 0.01)
between supplement effects and preterm delivery on 0–3-mo
infant mortality (Figure 3); similar interactions were also ob-
served with perinatal, neonatal, and overall infant mortality
rates as the outcomes (data not shown). Among the infants born
to mothers in the control group, the 3-mo mortality rate for the
preterm infants was higher (RR: 6.3; 95% CI: 3.4, 12.1) than
that for the term infants. Among the preterm births, supple-
mentation was associated with protective RRs of 0.36 (95% CI:
0.18, 0.75) and 0.53 (95% CI: 30, 0.92) for the FA and FAFe
groups. The protective RR estimates for the FAFeZn and MN
groups had 95% CIs that included 1.0. Among the term births,
there was no evidence of improved infant survival in any of the
supplement groups. The term infants whose mothers were in
the MN group appeared to have a higher risk of dying than did
those whose mothers were in the control group (RR: 1.74; 95%
CI: 1.00, 3.04).
The rate of neonatal death attributed to birth asphyxia was
lower in the FA, FAFe, and FAFeZn groups than in the control
group (Table 3). However, only the difference between the
FAFeZn group (RR: 0.23; 95% CI: 0.04, 0.81) and the control
group was significant.
DISCUSSION
We examined the survival of infants in a randomized clinical
trial in which prenatal micronutrient supplementation resulted
in small but significant decrements in the incidence of low birth
weight but not of preterm delivery (20). Although our analysis
showed consistently higher survival rates among the infants
1198 CHRISTIAN ET AL

TABLE 1
Characteristics of the pregnant women by treatment group1

C (n ⫽ 1037) FA (n ⫽ 929) FAFe (n ⫽ 940) FAFeZn (n ⫽ 982) MN (n ⫽ 1038)

Age at baseline
ⱕ 19 y 320 (31.1) 292 (31.6) 275 (29.4) 297 (30.2) 289 (27.9)
20–24 y 308 (29.9) 297 (32.1) 327 (34.9) 330 (33.6) 351 (33.9)
25–29 y 224 (21.8) 186 (20.1) 195 (20.8) 197 (20.1) 217 (21.0)
30–35 y 128 (12.4) 108 (11.7) 111 (11.9) 123 (12.5) 141 (13.6)
⬎ 35 y 50 (4.8) 42 (4.5) 28 (3.0) 35 (3.6) 37 (3.6)
Socioeconomic status
Literate 220 (21.6) 177 (19.2) 194 (20.8) 198 (20.4) 188 (18.3)
Low caste 130 (12.8) 112 (12.1) 96 (10.3) 135 (13.9) 138 (13.4)
Parity
0 275 (27.0) 255 (27.6) 237 (25.4) 252 (25.9) 265 (25.8)
1–2 371 (36.4) 338 (36.6) 380 (40.7) 381 (39.2) 388 (37.7)
3–4 216 (21.2) 194 (21.0) 197 (21.1) 208 (21.4) 222 (21.6)
ⱖ5 156 (15.3) 136 (14.7) 120 (12.8) 130 (13.4) 154 (15.0)
Gestational age at enrollment
⬍ 9 wk 503 (49.6) 439 (48.4) 458 (49.5) 465 (48.3) 463 (45.5)
9–16 wk 431 (42.6) 394 (43.4) 378 (40.9) 418 (43.4) 463 (45.5)
ⱖ 17 wk 79 (7.8) 74 (8.1) 89 (9.7) 79 (8.2) 91 (9.0)
Previous miscarriage2 95 (12.6) 77 (11.3) 86 (12.0) 91 (12.1) 95 (12.1)
Previous child death2 301 (39.8) 262 (38.3) 244 (34.1) 252 (33.6) 289 (36.9)
1
n; percentage in parentheses. Because of missing data, the values for each variable may not add up to the total n. C, control group (received vitamin
A only); FA, group who received vitamin A ⫹ folic acid; FAFe, group who received vitamin A ⫹ folic acid ⫹ iron; FAFeZn, group who received vitamin
A ⫹ folic acid ⫹ iron ⫹ zinc; MN, group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients. There were no significant differences
between the treatment groups.
2
Among women with a previous pregnancy.

whose mothers received folic acid with or without iron or iron tions in mortality of ⬇20% that were observed. In addition,
⫹ zinc than among the infants whose mothers were in the multiple micronutrient supplementation, which was noted to
control group, we had low power (⬇20%) to detect the reduc- exert the largest effect on birth weight (increase of 64 g; 95%

FIGURE 2. Kaplan-Meier curves depicting the survival of infants through the first year of life by treatment group. C, control group (received vitamin
A only); FA, group who received vitamin A ⫹ folic acid; FAFe, group who received vitamin A ⫹ folic acid ⫹ iron; FAFeZn, group who received vitamin
A ⫹ folic acid ⫹ iron ⫹ zinc; MN, group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients. There were no significant differences
between the groups.
 MICRONUTRIENT SUPPLEMENTATION AND INFANT MORTALITY 1199
TABLE 2 CI: 12, 115) (20), surprisingly resulted in a survival pattern that
Rates and relative risks (RRs) of perinatal, neonatal, and infant mortality did not differ significantly from that observed in the control
by treatment group1 group. Further analysis showed a significant interaction be-
n Rate/1000 live births RR (95% CI) tween treatment and preterm birth on mortality; the apparent
2
overall reduction in mortality with folic acid and folic acid ⫹
Perinatal deaths
iron was limited to the preterm infants only. Zinc appeared to
C 70/916 76.4 1.00
FA 56/814 68.8 0.89 (0.63, 1.26)
attenuate the mortality reduction observed with folic acid or
FAFe 50/801 62.4 0.80 (0.55, 1.17) folic acid ⫹ iron, although the CIs in all 3 groups overlapped.
FAFeZn 55/858 64.1 0.83 (0.57, 1.21) A similar attenuating effect was also evident among the pre-
MN 80/919 87.1 1.14 (0.82, 1.56) term infants in the MN group.
Neonatal deaths3 It is unlikely that these results were biased, because we had
C 40/876 45.7 1.00 an extremely low rate of loss to follow-up in the study. As-
FA 28/777 36.0 0.79 (0.50, 1.26) suming that any posited survival benefit could not have ex-
FAFe 28/772 36.3 0.80 (0.50, 1.27) tended beyond the supplementation period, we examined mor-
FAFeZn 31/827 37.5 0.82 (0.50, 1.34)
tality differences by treatment allocation among the infants up
MN 47/870 54.0 1.19 (0.77, 1.83)
to 3 mo of age. In addition, we observed no evidence of a
Infant deaths (0–3 mo)4
C 49/876 55.9 1.00 reversal of treatment effects beyond 3 mo of age. There was an
FA 34/777 43.8 0.78 (0.52, 1.17) apparent sustained (but not additional) beneficial effect of
FAFe 34/772 44.0 0.79 (0.52, 1.20) maternal supplementation on survival from 3 to 12 mo of age,
FAFeZn 40/827 48.4 0.87 (0.57, 1.31) but a much smaller proportion of deaths occur in this time
MN 52/870 59.8 1.07 (0.75, 1.58) period than in the period from 0 to 3 mo of age.
1
C, control group (received vitamin A only); FA, group who received Few studies have examined the effect of maternal nutritional
vitamin A ⫹ folic acid; FAFe, group who received vitamin A ⫹ folic acid ⫹ interventions on fetal loss and infant mortality. One random-
iron; FAFeZn, group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc; MN, ized controlled trial in the Gambia found that significant re-
group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronu- ductions in low birth weight and perinatal mortality were
trients. There were no significant differences between the treatment groups. associated with consumption during pregnancy of food supple-
2
Stillbirths and deaths among liveborn infants in the first 7 d of life/live ments that provided high amounts of calories and protein (28).
births and stillbirths. In HIV-1–infected Tanzanian women, multivitamins (20 mg
3
Deaths from 0 to 28 d of life/live births.
4
thiamine, 20 mg riboflavin, 25 mg vitamin B-6, 100 mg niacin,
Deaths from 0 to 90 d of life/live births.

FIGURE 3. Treatment effects of maternal supplementation on the risk of 0–3-mo mortality in infants stratified by preterm and term birth. RR, relative
risk; FA, group who received vitamin A ⫹ folic acid; FAFe, group who received vitamin A ⫹ folic acid ⫹ iron; FAFeZn, group who received vitamin
A ⫹ folic acid ⫹ iron ⫹ zinc; MN, group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients. The overall interaction was
significant, P ⬍ 0.01 [likelihood ratio test (chi-square with 5 df)]. Differences between the preterm and term infants (interaction effects) were significant
for the FA (P ⫽ 0.001) and MN (P ⫽ 0.015) groups. The main effect for the FAFe group was significant (P ⫽ 0.03).
1200 CHRISTIAN ET AL

TABLE 3
Neonatal deaths attributed to various causes by treatment group1

Cause of death C (n ⫽ 876) FA (n ⫽ 777) FAFe (n ⫽ 772) FAFeZn (n ⫽ 827) MN (n ⫽ 870)

Prematurity or severe malnutrition 12 (13.7) 7 (9.0) 8 (10.4) 17 (20.6) 8 (9.2)

Birth asphyxia 14 (16.0) 7 (9.0) 7 (9.1) 3 (3.6) 17 (19.5)
Infection2 10 (11.4) 9 (11.6) 7 (9.1) 3 (3.6) 14 (16.1)
Congenital abnormality 0 0 0 2 (2.4) 2 (2.3)
Unknown, uncertain, or sudden3 4 (4.6) 5 (6.4) 6 (7.8) 6 (7.2) 6 (6.8)
1
n; mortality rate/1000 live births in parentheses. C, control group (received vitamin A only); FA, group who received vitamin A ⫹ folic acid; FAFe,
group who received vitamin A ⫹ folic acid ⫹ iron; FAFeZn, group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc; MN, group who received vitamin
A ⫹ folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients.
2
Includes deaths due to neonatal tetanus, sepsis, pneumonia, diarrhea, or dysentery.
3
The cause of death was unknown, was unable to be specified by using the physician review, or was described by the respondents as being sudden.

50 ␮g vitamin B-12, 500 mg vitamin C, 30 mg vitamin E, and
0.8 mg folic acid) given in doses that were 2–10 times the
recommended dietary allowance for pregnancy led to a rate of
fetal death that was 40% lower than that in the placebo group
(29). However, the implications of this for uninfected popula-
tions are unclear.
It is essential to consider why maternal multiple micronutrient
supplementation failed to reduce mortality to an extent similar to
that observed with folic acid alone or with iron. One explanation
may relate to the disproportionately higher rate of high birth
weight (ⱖ 3.3 kg) among the infants whose mothers received
multiple micronutrients than among those whose mothers were in
the control group (20), an effect that was predictably evident
among the term infants only. Specifically, among term births,
supplementation with multiple micronutrients (but not with other
combinations of micronutrients) led to a significantly higher risk
of high birth weight (RR: 1.71; 95% CI: 1.10, 2.65) than did
supplementation with vitamin A alone (ie, the control treatment),
and high birth weight was associated with an increased risk of
symptoms of birth asphyxia (RR: 1.49; 95% CI: 1.04, 2.13). On
examination, we found that the risk of symptoms of birth asphyxia
was higher among the term infants whose mothers received mul-
tiple micronutrients than among those whose mothers were in the
control group (Figure 4). Thus, while multiple micronutrient
supplementation improved survival among the preterm infants
(RR: 0.70; 95% CI: 0.41, 1.17), it also appeared to increase high

FIGURE 4. Treatment effects of maternal supplementation on the risk of symptoms of birth asphyxia stratified by preterm and term birth. Symptoms of birth
asphyxia were as follows: 1) the infant did not cry or cried weakly at birth and was not able to breathe after birth, or the infant had convulsions or was unable
to suckle; 2) the infant died within 7 d and either had convulsions or a breech presentation, or the woman had prolonged labor (⬎ 24 h) or obstructed labor; or
3) the infant died within the first day of birth and had a blue body or pale extremities. RR, relative risk; FA, group who received vitamin A ⫹ folic acid; FAFe,
group who received vitamin A ⫹ folic acid ⫹ iron; FAFeZn, group who received vitamin A ⫹ folic acid ⫹ iron ⫹ zinc; MN, group who received vitamin A ⫹
folic acid ⫹ iron ⫹ zinc ⫹ other micronutrients. The overall interaction was significant, P ⬍ 0.01 [log likelihood ratio test (chi-square with 5 df). Differences
between the preterm and term infants (interaction effects) were significant for the FAFe (P ⫽ 0.03) and MN (P ⫽ 0.05) groups.
 MICRONUTRIENT SUPPLEMENTATION AND INFANT MORTALITY
FIGURE 5. Plausible explanatory pathway for the observed lack of an
overall effect of multiple micronutrient supplementation on infant mortality.
birth weight and the associated risk of birth asphyxia, a known
cause of perinatal mortality (30), among the term infants. These
countervailing influences may have cancelled one another out and
thus resulted in multiple micronutrients seemingly having no
overall effect on mortality (Figure 5). Other mechanisms that
might also explain the lack of mortality benefit due to multiple
micronutrient supplementation may exist but remain to be eluci-
dated.
These results, coupled with the observation that folic acid
appeared to reduce mortality without improving birth weight
(20), suggest that improvement in mortality observed with
maternal micronutrient supplementation may not always be
mediated through increased birth weight. Although birth
weight is commonly used as a proxy measure for potential
infant health and survival, excess perinatal and neonatal mor-
tality has been shown to operate through mechanisms other
than birth weight (31–34). Indeed, health benefits have been
previously described in infants after maternal micronutrient
supplementation that did not result in increased birth weight. In
a randomized clinical trial in Niger, supplementation with 100
mg elemental Fe had no effect on birth weight (35), and yet
there was some evidence that fetal loss and neonatal death were
reduced (36). In Bangladesh, the incidences of acute diarrhea,
dysentery, and impetigo were significantly lower in low-birth-
weight infants whose mothers received zinc than in those
whose mothers received placebo (37), even though birth weight
itself did not differ significantly between the 2 groups of
infants (16). Small size due to preterm delivery may be more
strongly associated with perinatal mortality than is low birth
weight per se (38, 39). In the present study, maternal supple-
mentation with folic acid (alone or with iron) improved the
survival of high-risk preterm infants even though the risk of
preterm delivery itself remained unchanged (20).
Our results are likely to be applicable to poor, rural populations
in south Asia, where the prevalences of maternal protein-energy
malnutrition and micronutrient deficiencies are high and where the
prevalence of HIV-1 infection is low. Currently, there is a move
toward using multiple micronutrient supplements in pregnant
women to reduce adverse pregnancy outcomes, including low
birth weight and infant mortality, in the developing world. How-
ever, further understanding of the mechanisms involved and of the
potential interactions that may exist among nutrients and that
could produce potentially adverse outcomes is needed. The ratio-
nale for adding other micronutrients to a maternal folic acid ⫹
iron supplement to improve pregnancy outcomes should receive
careful examination. This study clearly needs to be replicated in
different populations to appropriately inform policies and to de-
1201
velop effective maternal nutrition programs in the developing
world.
Apart from the authors, the following members of the Nepal study team
helped in the implementation of the study: Tirtha Raj Shakya, Rabindra
Shrestha, Uma Shankar Sah, Arun Bhetwal, Gokarna Subedi, Dhrub
Khadka, Jaibar Var Shrestha. Sanu Maiya Dali and Ramesh Adhikari
served as consultants on the project, Gwendolyn Clemens performed
computer programming and data management, Seema Rai and Sunita Pant
supervised data entry and cleaning, and Lee Wu helped with the statistical
analyses.
PC was the principal investigator, designed the study hypothesis and
protocol, directed the implementation of the study, conducted the analyses,
and is the guarantor for the study. KPW provided help during the concep-
tual phase of the study, helped with the study procedures, and assisted in
interpreting the results and writing the manuscript. SKK directed the field
implementation, helped with the development of forms, and conducted the
infant verbal autopsy review. SCL supervised the field procedures, data
collection, and quality control and helped to design forms and procedures
and to edit the manuscript. EKP helped to develop forms, manage and enter
data, analyze data, and edit the manuscript. JK helped in the study design,
data management procedures, data analysis, and the writing of the manu-
script. SRS supervised field procedures, community preparation and advo-
cacy, and translations of forms and manuals. AS contributed to data
interpretation and manuscript editing. None of the authors had any conflicts
of interest.
REFERENCES
1. Costello A, Manandhar D. Current state of the health of newborn
infants in developing countries. In: Costello A, Manandhar D, eds.
Improving newborn infant health in developing countries. London:
Imperial College Press, 2000:3–14.
2. Osrin D, de L Costello AM. Maternal nutrition and foetal growth:
practical issues in international health. Semin Neonatol 2000;5:209 –
19.
3. Puffer RR, Serrano CV. Patterns of birthweights. Washington, DC:
Pan American Health Organization, 1987.
4. McCormick MC. The contribution of low birth weight to infant mor-
tality and childhood morbidity. N Engl J Med 1985;312:82–90.
5. Victora C, Barros F, Vaughan J, Teixeria A. Birth weight and infant
mortality: a longitudinal study of 5914 Brazilian children. Int J Epi-
demiol 1987;16:239 – 45.
6. McIntire D, Bloom S, Casey B, Leveno K. Birth weight in relation to
morbidity and mortality among newborn infants. N Engl J Med 1999;
340:1234 – 8.
7. Victora CG, Smith PG, Vaughan JP, Nobre LC, Lombardi C, Teixeria
AMB. Influence of birthweight on mortality from infectious diseases:
a case-control study. Pediatrics 1988;816:807–11.
8. Lira PIC, Ashworth A, Morris SS. Low birth weight and morbidity
from diarrhea and respiratory infection in northeast Brazil. J Pediatr
1996;128:497–504.
9. Ashworth A. Effects of intrauterine growth retardation on mortality
and morbidity in infants and young children. Eur J Clin Nutr 1998;
52:S34 – 42.
10. Read JS, Clemens JD. Low birth weight as a determinant of infectious
disease mortality: a methodological evaluation. Paediatr Perinat Epi-
demiol 1995;9:351– 62.
11. Yip R. Iron supplementation during pregnancy: is it effective? Am J
Clin Nutr 1996;63:853–5.
12. Rush D. Nutrition and maternal mortality in the developing world.
Am J Clin Nutr 2000;72(suppl):212S– 40S.
13. Yip R. Significance of an abnormally low or high hemoglobin con-
centration during pregnancy: special consideration of iron nutrition.
Am J Clin Nutr 2000;72(suppl):272S–9S.
14. Rasmussen KM. Is there a causal relationship between iron deficiency
or iron deficiency anaemia and weight at birth, length of gestation and
perinatal mortality? J Nutr 2001;131:590S– 603S.
15. de Onis M, Villar J, Gulmezoglu M. Nutritional interventions to
prevent intrauterine growth retardation: evidence from randomized
controlled trials. Eur J Clin Nutr 1988;52:S83–93.
1202 CHRISTIAN ET AL
16. Osendarp SJM, van Raaij JMA, Arifeen SE, Wahed MA, Baqui AH,
Fuchs GJ. A randomized, placebo-controlled trial of the effect of zinc
supplementation during pregnancy on pregnancy outcome in Bang-
ladeshi urban poor. Am J Clin Nutr 2000;71:114 –9.
17. Caulfield LE, Zavaleta N, Figueroa A, Leon Z. Maternal zinc supple-
mentation does not affect size at birth and pregnancy duration in Peru.
J Nutr 1999;129:1563– 8.
18. Garg HK, Singhla KC, Arshad Z. A study of the effect of oral zinc
supplementation during pregnancy on pregnancy outcome. Indian
J Physiol Pharmacol 1993;37:276 – 84.
19. Goldenberg R, Tamura T, Neggers Y, et al. The effect of zinc sup-
plementation on pregnancy outcome. JAMA 1995;274:463– 8.
20. Christian P, Khatry SK, Katz J, et al. Effects of alternative maternal
micronutrient supplements on low birth weight and anaemia in rural
Nepal. A double-masked randomized community trial. BMJ 2003;326:
571– 6.
21. West KP Jr, Katz J, Khatry SK, et al. Double blind, cluster randomised
trial of low dose supplementation with vitamin A or beta carotene on
mortality related to pregnancy in Nepal. The NNIPS-2 Study Group.
BMJ 1999;318:570 –5.
22. Katz J, West KP Jr, Khatry SK, et al. Low-dose vitamin A or ␤-car-
otene supplementation does not reduce early infant mortality: a dou-
ble-masked, randomized, controlled community trial in Nepal. Am J
Clin Nutr 2000;71:1570 – 6.
23. Kalter HD, Hossain M, Burnham G, et al. Validation of caregiver
interviews to diagnose common causes of severe neonatal illness.
Paediatr Perinat Epidemiol 1999;13:99 –113.
24. Nepal Micronutrient Status Survey. Kathmandu, Nepal: Ministry of
Health, Child Health Division, HMG/N, New ERA, Micronutrient
Initiative, UNICEF Nepal, and WHO, 1998.
25. Dreyfuss ML, Stoltzfus RJ, Shrestha JB, et al. Hookworms, malaria,
and vitamin A deficiency contribute to anemia and iron deficiency
among pregnant women in the plains of Nepal. J Nutr 2000;130:2527–
36.
26. Kahn HA, Sempos CT. Statistical methods in epidemiology. New
York: Oxford University Press, 1989.
27. Liang KY, Zeger SL. Longitudinal data analysis using generalized
linear models. Biometrika 1986;73:13–22.
28. Ceesay SM, Prentice AM, Cole TJ, et al. Effects on birth weight and
perinatal mortality of maternal dietary supplements in rural Gambia: 5
year randomised controlled trial. BMJ 1997;315:786 –90.
29. Fawzi WW, Msamanga GI, Spiegelman D. Randomised trial of effects
of vitamin supplements on pregnancy outcomes and T cell counts in
HIV-1-infected women in Tanzania. Lancet 1998;351:1477– 82.
30. Ellis M, Manandhar DS, Manandhar N, Wyatt J, Bolam AJ, de L
Costello A. Stillbirths and neonatal encephalopathy in Kathmandu,
Nepal: an estimate of the contribution of birth asphyxia to perinatal
mortality in a low-income urban population. Paediatr Perinat Epide-
miol 2000;14:39 –52.
31. Garner P, Kramer MS, Chalmers I. Might efforts to increase birth
weight in undernourished women do more harm than good? Lancet
1992;340:1021–2.
32. Chalmers I. Perinatal health: the search for indices. Lancet 1979;2:
1063–5.
33. Forssas E, Gissler M, Sihvonen M, Hemminki E. Maternal predictors
of perinatal mortality: the role of birth weight. Int J Epidemiol 1999;
28:475– 8.
34. Villar J, de Onis M, Kestler E, Bolanos F, Cerezo R, Bernedes H. The
differential morbidity of the intrauterine growth retardation syndrome.
Am J Obstet Gynecol 1990;163:151–7.
35. Preziosi P, Prual A, Golan P, Daouda H, Boureima H, Hercberg S.
Effect of iron supplementation on the iron status of pregnant women:
consequences for newborns. Am J Clin Nutr 1997;66:1178 – 82.
36. Christian P. Antenatal iron supplementation as a child survival strat-
egy. Am J Clin Nutr 1998;68:404 (letter).
37. Osendarp SJM, van Raaij JMA, Darmstadt GL, Baqui AH, Hautvast
JGAJ, Fuchs GJ. Zinc supplementation during pregnancy and effects
on growth and morbidity in low birthweight infants: a randomised
placebo controlled trial. Lancet 2001;357:1080 –5.
38. Wilcox AJ, Skjaerven R. Birth weight and perinatal mortality: the
effect of gestational age. Am J Public Health 1992;82:378 – 82.
39. Wilcox AJ. On the importance—and the unimportance— of birth-
weight. Int J Epidemiol 2001;30:1233– 41.