![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
*
From the Department of Economics, University of Houston, Houston, Texas 77204-5882;
International Food Policy Research Institute, 2033 K Street NW, Washington, D.C. 20006 and
**
International Nutrition Foundation, Box 500, Charles Street Station, Boston, Massachusetts 02114
2To whom correspondence should be addressed.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
KEY WORDS: • bioavailable iron • iron deficiency anemia • socioeconomic factors • longitudinal data • random effects models
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). IDA
hinders normal human functions in all age groups. For example, IDA
lowered labor productivity of Indonesian rubber tappers (Basta et al. 1979
) and Sri Lankan tea pickers (Gardner et al. 1977
), adversely affected birth outcomes (Bhargava 2000
) and impaired the cognitive development of children
(Lozoff 1988
, Pollitt 1993
). The
productivity loss and human costs associated with IDA are enormous
(UNICEF/UNU/WHO/MI 1999
).
Poor diet quality and low bioavailability of dietary iron are important
factors that contribute to IDA (Tatala et al. 1998
);
iron loss due to parasitic infection and menstrual bleeding and
pregnancy can exacerbate IDA in women. For the design of effective
policies, it is essential to adopt a multifactorial approach to
analyzing the proximate determinants of IDA under actual living
conditions. Thus, for example, the success of iron fortification
programs is likely to vary with the intakes of meat, fish, poultry
(MFP) and vitamins A and C, which enhance nonheme iron absorption, and
of phytates and tannins, which inhibit it (Garcia-Casal et al. 1998
, Hallberg et al. 1989 and 1997
,
Monsen et al. 1978
).
The algorithms developed for calculating iron bioavailability in the
presence of enhancers such as meat and vitamin C were recently extended
to incorporate the inhibitory effects of phytates, which chelate iron,
thereby reducing its absorption (Tseng et al. 1997
).
Although iron intake by the poor in Bangladesh comes from staple foods
such as rice, contamination iron from pots, small quantities from green
and yellow vegetables and animal products, the phytate content of the
meal is typically high. Thus, algorithms for calculating iron
bioavailability and taking into account the enhancers and inhibitors of
iron absorption at each meal are specially useful for subjects in
developing countries. However, the inhibitory effects of phytate
intakes have been customarily calculated from data on healthy subjects
with 
500 mg of iron stores (Brune et al. 1992
,
Hallberg et al. 1989
). By contrast, iron absorption
rates in the presence of enhancers were tabulated for body stores 0,
250 and 500 mg by Monsen et al. (1978
); alternative
assumptions on body iron stores lead to different estimates of
absorbable iron.
Hemoglobin concentration (Hb) is a widely used measure for assessing
IDA (Khusun et al. 1999
, UNICEF/WHO 1999
). Although measures such as serum ferritin and erythrocyte
protoporphyrin provide additional insights into iron status, there are
logistics and other difficulties in transporting venous blood samples
from a large number of subjects living in remote areas of developing
countries. Thus, modeling the proximate determinants of Hb measured in
the field through a fingerprick can be valuable for deciding on the
allocation of resources among instruments such as government policies
that encourage MFP production, the provision of elemental iron tablets,
and so on.
Thus far, only a few studies in the nutrition literature have attempted
to quantify the effects of dietary intakes on serum iron measures
(Doyle et al. 1999
, Du et al. 2000
).
Also, most previous analyses did not incorporate other confounding
factors and did not address in detail the methodological issues
surrounding the bioavailability of iron. Dietary intakes and
supplements, morbidity from infectious disease and genetic factors are
likely to affect Hb status. Therefore, we developed and estimated
comprehensive longitudinal models for the proximate determinants of the
Hb status of Bangladeshi women. Because the success of interventions
will be influenced by the way in which dietary intakes change with
incomes, we also estimated a model for assessing the impact of
increases in household incomes on iron intake from MFP (FeMFP) and for
iron intake from all animal sources.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The study was conducted in 16 villages in Jessore, 10 in Saturia and 18
in Mymensingh, Bangladesh (Bouis et al. 1998
). One of
the purposes of the research was to investigate the nutritional impact
of the adoption of agronomically improved vegetables in Saturia and
fish pond management strategies in Jessore and Mymensingh. At each
site, the new technologies were introduced through nongovernment
organization programs that provided credit and training to women. The
selection of households was done in a complex manner to ensure that the
selected households were representative of the three sites. Initially,
there were 990 households, of which 33 did not participate in all four
survey rounds. The sample size was further reduced because some women
were not available for Hb measurements during one or more survey
rounds. A comparison of the households in the census sample with
nationally representative surveys for Bangladesh (Rahman et al. 1996
) indicated that households in our sample were
representative of households in the threes sites and not atypical of
rural Bangladesh.
The study design was approved in 1996 by a human subjects committee of the Bangladesh Medical Research Council in Dhaka. The surveys began in June 1996, and the fourth survey round was completed by September 1997. Because Hb was measured in survey rounds 2, 3 and 4, we analyzed the data from these rounds. Overall, there were 664 women in the age group 15–49 y (one from each household) for whom three observations separated by 4-mo intervals were available. Because observations on nutritional, anthropometric and other variables were missing for some women in the three survey rounds, models for Hb were estimated using the complete data on 514 women. The models for women’s intake of FeMFP and for iron intake from all animal sources were estimated using the data on 514 women. For the simple autoregressive models estimating between- and within-subject variations in Hb, data on 664 women in three survey rounds were used; Hb data on a subset of 71 women on 4 consecutive d in the fourth survey round were also analyzed.
Economic, demographic and morbidity variables.
Background information was compiled on household members’ age, relationship with the head of the household, occupation, education and other items. Detailed information was gathered in the four survey rounds on economic variables such as household assets, incomes, food and nonfood expenditures, wages and so on; a variable was constructed for the average per capita monthly expenditures for each round. In poor countries, expenditure data are more reliable measures of economic well-being than are incomes. The reproductive history of each woman was investigated, and the current pregnancy and lactation status was recorded. In each survey round, the women were questioned about symptoms such as fever, cough, diarrhea and any diseases in the past 2 wk; women with mucus and blood in the diarrhea were questioned. Work loss due to chronic illness during the year was also assessed.
Anthropometry and hematology.
Weight and arm circumferences of the women were measured in each survey round; height was measured at the start of the study. Spring scales accurate to 0.25 kg were used to measure the subjects’ weight in light clothing. An adjustable wooden measuring board was used to measure height to the nearest 0.1 cm with the woman standing in upright position. A paper tape was used to measure the mid upper arm circumference (MUAC). Hb status was measured in the three survey rounds with a fingerprick sample of capillary blood obtained by a physician and analyzed immediately using a portable photometer (HemoCue; HemoCue Inc., Mission Viejo, CA).
Nutritional intakes and bioavailable iron.
In each of the three survey rounds, food intakes were measured using
the 24-h recall method for the four meals consumed (i.e., breakfast,
lunch, dinner and snacks). The person primarily responsible for
preparing the meals was questioned about the recipes, ingredients and
amounts of dishes consumed by household members and guests. The
women’s intakes of 40 nutrients at each meal were estimated using the
food composition database of Calloway et al. (1994
) for
six countries.
For estimating the bioavailable iron, taking into account nutrient
interactions at each meal, we focused on the intakes of dietary iron,
iron from animal sources, ascorbic acid, phytates and tannins from tea.
Because most women did not consume any snacks, the data on snacks were
aggregated with nutrient intakes at lunch. Moreover, because tea
consumption was negligible in this population, the inhibitory effects
of phytates were the prime concern. To calculate the bioavailable iron,
algorithms (Monsen et al. 1978
, Monsen and Balintfy 1982
, Tseng et al. 1997
) were suitably
extended. The enhancing effects of MFP and ascorbic acid were
programmed to calculate bioavailable iron at each meal under the
assumptions that the women’s iron body stores were 0, 250 and 500 mg.
In comparison with normal subjects with iron stores of 500 mg, the
absorption rates for heme and nonheme iron were 25 and 50% higher for
subjects with iron stores of 250 and 0 mg, respectively (Monsen et al. 1978
, Table 1
). The mathematical formulas for body stores of 0 and 250 mg have not
been used in previous studies and may be more realistic for
undernourished populations.
|
) for
calculating the inhibitory effects of phytate intakes assuming an iron
store of 500 mg contained a mathematical error. The correct formulas
for the 0-, 250- and 500-mg iron stores in the presence of enhancers
and inhibitors of iron absorption in the meal are presented as eqs. 3–5
. A Fortran program for meal-by-meal calculations is available from
the first author; the program also sums the bioavailable iron (and
other nutrients) at three or more meals to produce figures for the 24-h
period. Computer programs for statistical packages, such as those
developed by Tseng et al. (1997)
for SAS (1996), can be
easily modified on the basis of these mathematical formulas.
In the following equations, FeTOT indicates the total iron intake, and
FeBIO indicates the bioavailable iron. Heme iron was assumed to
constitute 40% of FeMFP (Monsen et al. 1978
). Using the
notation of Monsen and Balintfy (1982
), let the
enhancing factor (EF) for a meal be given by
 | (1) |
where M, F and P are the edible quantities of MFP (in g), respectively, and AA is the intake of ascorbic acid (in mg). If EF is >75, then EF was assumed to be 75.
To take account of the inhibitory effects of phytates, using the data
of Hallberg et al. (1989
), Tseng et al. (1987) calculated the "correction term" (CT) (0 
CT
1) that gives the proportion of FeBIO. However, Tseng et al. (1997
) defined CT incorrectly when phytate intake was
2.88 mg, because in the data of Hallberg et al. (1989
), phytate intakes were either 0 or >2.88 mg. Let PHY be
the total phytate intake (in mg) during the meal. Then, for PHY of
2.88 mg, we defined CT as 1 (i.e., we assumed that there were
no inhibitory effects of phytate intake for such small values). By
contrast, the algorithm of Tseng et al. (1997
) would
inadvertently imply that PHY intakes in the interval of 0–2.88 mg
increase iron absorption. For other values of PHY, CT was defined by
 | (2) |
where log10 is logarithm to the base 10.
Thus, high values of PHY intake reduce FeBIO in a curvilinear fashion
(Brune et al. 1992
, Tseng et al. 1997
).
Assuming that body iron stores were 0, 250 and 500 mg, the FeBIO can be
calculated, respectively, from the following three equations
(Monsen et al. 1978
; Table 1
):
 | (3) |
 | (4) |
 | (5) |
where logn is natural logarithm. Note that because the absorption of heme iron was assumed to be not affected by the presence of other nutrients in the meal, the formulas for calculating FeBIO simultaneously take into account the effects of enhancers and inhibitors on nonheme iron absorption (i.e., the order in which the adjustment is made for enhancers and inhibitors is immaterial). Also, the inhibitory effects of phytate intake in eqs. 3–5 assumed body iron stores of 500 mg; this issue is further discussed later.
A model for the proximate determinants of Hb concentration.
The Hb status of a woman is influenced by previous nutritional intakes, sicknesses, pregnancy and lactation status, dietary supplements and genetic factors. The intake of heme iron and interactions between nonheme iron and ascorbic acid, vitamin A, beta-carotene and phytates in the meal affect iron absorption and, ultimately, the Hb level. In developing countries, economic constraints on household food consumption adversely affect the quality of diet, thereby hindering iron absorption. Determination of within-subject variability in 24-h recall data at a given time requires two or three random repetitions within a limited time period. In our study, we obtained useful information on within-subject variability during the entire study by using the women’s dietary intakes in each of the three survey rounds.
Longitudinal studies that measure subjects’ dietary intakes over more
extended periods of time would be prohibitively expensive. It is
therefore important to introduce anthropometric variables such as the
MUAC, height and weight in models that explain Hb status because they
reflect the history of nutritional intakes and infections. Moreover,
recent sicknesses, especially those involving blood loss, can reduce Hb
(Scrimshaw et al. 1959
). By contrast, iron supplements
can raise Hb within a short time frame (Viteri 1999
).
The model for Hb should account for all of these factors.
The effects of dietary intakes and other factors on Hb status can be
analyzed using longitudinal data by estimating autoregressive
regression models that include previous measurement on Hb as an
explanatory variable. Denoting the ith subject’s Hb in
time period t by Hbit
(i = 1, 2, ... , n;
t = 2, 3), we postulated the autoregressive model
for Bangladeshi women:
 |
 |
 |
 | (6) |
where a0, ... , a9 are the regression coefficients, uit is an error term (see eq. 7 ) and Hbit - 1 is the previous measurement of Hb status, with coefficient a9.
There were several attractive features of the model in eq. 6 for Bangladeshi women’s Hb status using three repeated observations in an 8-mo period. First, the short-run, or immediate, impact of a change in a variable such as the FeBIO on Hb was given by a6; the long-run impact was [a6/(1 - a9)]. Generally, one would expect 0 < a9 < 1, so the long-run effect would be greater than the short-run impact. Moreover, for small estimated values of a9, the 8-mo observation period was sufficiently long for much of the long-term effect to materialize.
Second, the error term uit in eq. 6
can be
decomposed in a random effects fashion as:
 | (7) |
where 
s are women-specific random effects, and
v’s are independently distributed random variables. Because
Hb is influenced by genetic factors (Garner et al. 2000
)
and other unobserved characteristics, one would expect the
between-subject variance (of i) to be an
important parameter in the model that explains differences in Hb
status.
Third, Hb measurements using capillary blood often exhibit large
within-subject variation that can obscure the effects of
explanatory variables (Liu et al. 1976
, Morris et al. 1999
). In contrast with a single observation on the
subjects, however, maximum likelihood estimates computed using repeated
observations in a random effects framework can alleviate some of these
problems. Fourth, interesting hypotheses can be tested using the model
given in eq. 6
. For example, economic factors are likely to affect diet
quality of the Bangladeshi women; diet quality in turn would determine
the quantity of FeBIO that would be expected to predict Hb status.
Autoregressive models for iron intake from MFP and iron from all animal sources.
To investigate the effects of economic variables on Hb status, we
postulated an autoregressive model for the intake of FeMFP; such models
are consistent with theories of "habit persistence" in diets that
are important for analyzing dietary patterns, especially in traditional
societies (Bhargava 1991
, Gorman 1967
).
The model can be written as:
 |
 |
 |
 | (8) |
Note that household size (and its square) were included in the
model in eq. 8
because women in poor households with a large number of
children were likely to consume inadequate quantities of FeMFP. The
household per capita total expenditure, which changed with the survey
rounds, was potentially important for explaining women’s intake of
FeMFP. Women’s height and weight were introduced into the model
because they reflected the energy requirements (James and Schofield 1990
). In eq. 8
, height and weight were combined as
the body mass index (BMI; in kg/m2) due to the
results of a statistical test (Bhargava 1994
). In rural
populations subsisting in poverty, BMI can be a measure of chronic
undernutrition and so associated with a variety of adverse outcomes
(James and Ralph 1994
).
Econometric procedure.
Because there were only three time observations available for the
women, statistical estimation was based on the assumptions that the
number of women was large but the number of survey rounds was fixed.
Thus, initial observations on the dependent variables were treated as
endogenous variables (correlated with the errors, Bhargava and Sargan 1983
). The errors in eqs. 6 and 8
were assumed
independent across women but correlated over time with a positive
definite variance-covariance matrix. The random effects
decomposition for the u values given in eq. 7
was a
special case of this model.
The joint determination of the three observations on Hb (or FeMFP)
implied that econometric techniques used for simultaneous equations
were likely to be useful in this application. Details of the maximum
likelihood estimation method are presented elsewhere (Bhargava and Sargan 1983
). Here, we note that the profile
log-likelihood functions of the models in eqs. 6 and 8
were
optimized using numerical scheme E04 JBF from the Numerical Algorithm Group (1989
); asymptotic standard errors of the
parameters were obtained by approximating second derivatives of the
function at the maximum. The maximized values of the logarithm of the
likelihood function were used to test hypotheses regarding the
coefficients of the variables included in the models for Hb and FeMFP.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The sample means of selected variables in three survey rounds for 514
Bangladeshi women are given in Table 1
. Striking features of the intake
data were the low FeMFP intakes and the high phytate intakes. For
example, in the first survey round, the average daily FeTOT intake was
6.93 ± 3.00 mg, of which only 0.33 ± 0.47 mg was FeMFP.
Further, assuming 0-mg iron body stores and taking into account
enhancers of iron absorption, the average daily FeBIO was 0.85 ± 0.54 mg. However, this was reduced to 0.20 ± 0.13 mg when the
effect of phytate intake was incorporated; the suitability of the
algorithm for incorporating inhibitory effects of phytate intakes in
undernourished populations is addressed in Discussion. Similarly,
assuming 250-mg iron stores, the average daily FeBIO fell from 0.56
± 0.33 to 0.14 ± 0.09 mg when phytate intakes were taken
into account. These intakes are well below the "safe" level of 2 mg
recommended for a woman who weighs 50 kg (FAO/WHO 1988
).
The average daily intake of FeMFP by 514 women at percentiles 1, 5, 10, 15, 25, 50, 75 and 90 was 0.009, 0.030, 0.058, 0.080, 0.111, 0.209, 0.338 and 0.513 mg, respectively. The corresponding values for Hb at these percentiles were 81.4, 97.0, 103.5, 106.1, 111.7, 118.0, 125.0 and 130.3 g/L, respectively. Using 120 g Hb/L as a cutoff point for IDA, >55% of the women would be classified as having IDA. For the 71 women who were pregnant during one or more of the three survey rounds, the respective Hb measurements were 76.7, 88.7, 99.3, 101.9, 104.0, 111.3, 116.0 and 122.8 g/L; hemodilution and iron cost of pregnancy are likely to diminish the Hb status of pregnant women.
Autocorrelations and between- and within-subject variations in Hb concentration.
Table 2
presents the results for a simple autoregressive model for the natural
logarithm of Hb; the constant term was the only additional explanatory
variable in eq. 6
. The point estimate of the coefficient of the lagged
dependent variable using the data in three survey rounds was 0.18
(SE = 0.058). This was smaller than the corresponding
estimate 0.248 (SE = 0.112) for the 4-d observations; both
estimates were statistically significant. The difference in estimates
seemed reasonable because nutritional and other factors such as
women’s pregnancy status and anthropometric indicators changed over
the three survey rounds.
|
) may also have contributed to the between-women
differences. Last, the within-subject variance was large,
especially in comparison with studies in which simple random effects
models were estimated for venous blood samples (Morris et al. 1999
). However, the autocorrelations were not estimated in
previous studies. Results from the autoregressive model explaining Bangladeshi women’s Hb concentration by demographic and nutritional variables.
The results from estimation of the autoregressive model given in eq. 6
for women’s Hb status are presented in Table 3
, where body iron stores were assumed to be 0 mg (similar results were
obtained when iron stores were assumed to be 250 and 500 mg and hence
are not reported). Table 3
presents results for the cases where, first,
only the enhancers of iron absorption were taken into account and,
second, the inhibitory effects of phytates in the meal were also
incorporated in calculating FeBIO. The subjects’ age, height, MUAC,
weight, FeBIO and the previous measurement of Hb were transformed into
natural logarithms. The logarithmic transformation reduced internal
variation in the data (Nelson et al. 1989
). The
estimated coefficients of the variables in logarithms were
short-run "elasticities" (percentage change in the dependent
variable resulting from a 1% change in an independent variable). The
long-run elasticity with respect to an explanatory variable can be
obtained by dividing the short-run elasticity by (1 - a9), where
a9 was the coefficient of Hb in the
previous time period.
|
The women’s height and MUAC were positively associated with Hb status; these coefficients were statistically significant. Height is a good indicator of early nutrition, whereas MUAC can approximate lean body mass. It was not surprising that these two variables were positively associated with Hb. By contrast, weight was negatively associated, which may have been due to the presence of pregnant women in the sample; the indicator variable did not account for pregnancy trimesters.
In the first column, where only the enhancers of iron absorption were used for calculating the FeBIO, the coefficient of FeBIO was positive and significant at the 5% level. Moreover, this coefficient showed a 20% increase in the second column where the inhibitory effects of phytate intakes were also taken into account. The maximized value of the logarithm of the likelihood function was higher when the inhibitory effects of phytate intake were incorporated.
The indicator variable for woman with bloody diarrhea was negative but
not significantly associated with Hb status; parity and the average
birth interval were also insignificant. However, the indicator variable
for women taking iron tablets during the survey round was positive and
significant. In Bangladesh, iron tablets were often sold together with
birth control pills. The results reported in Table 3
set the indicator
variable for iron tablets to 1 only when the women answered the
specific question regarding iron tablets intake. Last, maximized values
of the logarithms of the likelihood functions were slightly higher when
it was assumed that the women had 0-mg body iron stores than in the
cases where the iron stores were assumed to be 250 or 500 mg. Although
this was not direct evidence for the actual iron stores, the results
suggested that the average woman probably had iron stores lower than
250 mg.
Results from autoregressive models explaining Bangladeshi women’s iron intake from MFP and from all animal sources by economic, demographic and nutritional variables.
The results from estimation of the model in eq. 8
for intake of FeMFP
and iron from all animal sources are given in Table 4
. Because some women did not consume any MFP in one or more survey
rounds but consumed some milk and animal products, the two sets of
results give a better description of the food consumption patterns. The
zero intakes of iron by some women were set to 0.01 mg before the
logarithmic transformation; a sensitivity analysis was performed to
investigate whether this assumption affected the results, but it did
not. The interesting aspects of the results were that 1) age
had a large negative coefficient that was statistically significant in
both models; older women consumed lower quantities of animal products.
This finding was also consistent with the slight decline in Hb with age
in Table 3
. 2) There was a nonlinear effect of household
size on intake of FeMFP and iron intake from all animal sources. This
was an expected finding because larger households typically consist of
a greater number of children growing up in poverty; the diet quality in
such households is generally poor. 3) The coefficients of
monthly average per capita expenditure were large and significant; a
1% increase in households’ monthly expenditure was associated with
0.65 and 0.71% increases in FeMFP intake and iron intake from all
animal sources, respectively. The corresponding long-run income
elasticities were 0.70 and 0.87. These results suggested that women’s
iron intakes improved significantly with increases in household
incomes.
|
i) in eq. 8
and
BMI was 0; the hypothesis was accepted for FeMFP (likelihood ratio
statistic, 0.30, df, 3; P = 0.999). Similar
results were obtained for iron intake from all animal sources. 5) The coefficients of lagged dependent variables were 0.077 and 0.130, respectively, in the models for intake of FeMFP and iron intake from all animal sources; both coefficients were statistically significant. The ratios of between to within variance were not statistically significant in either model. This was in part due to the large within-subject variation in iron intake from animal sources.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). Using the recent longitudinal survey
from Bangladesh, the intake of FeBIO were calculated under alternative
assumptions of women’s iron stores. For example, the mean intake of
FeBIO in survey round 2 was 0.20 mg/d, taking into account the
meal-by-meal nutrient interactions and assuming negligible body iron
stores; mean intakes of FeBIO were 0.14 and 0.10 mg/d, respectively,
when the body stores were assumed to be 250 and 500 mg. Furthermore,
high phytate content of the meals reduced FeBIO intake from 0.85 to
0.20 mg/d with the assumption of zero iron stores.
The inhibitory effects of phytates were calculated using algorithms
based on data on healthy populations with iron stores of 500 mg
(Hallberg et al. 1989
, Tseng et al. 1997
). It is plausible that inhibitory effects of phytates and
tannins differ in populations with lower iron stores. For example,
Murphy et al. (1992
) adjusted nonheme iron intakes of
toddlers in Egypt, Kenya and Mexico for the presence of tannins in the
meal, although the data were not available by meals. The authors
estimated that average FeMFP intakes of toddlers in Egypt, Kenya and
Mexico were 0.64, 0.05 and 0.37 mg/d, respectively (Murphy et al. 1992
, Table 4
). However, the corresponding FeBIO intakes
incorporating the effects of tannins in the three populations were
0.49, 0.61 and 0.37 (mg/d), respectively. It is perhaps surprising that
despite the lowest average intake (0.05 mg/d) of FeMFP in Kenya, these
toddlers had the highest intake (0.61 mg/d) of FeBIO. It would seem
critical to conduct studies in undernourished populations to derive
suitable algorithms for iron absorption from mixed diets.
The second objective of the study was to develop comprehensive
longitudinal models for the Hb status of Bangladeshi women and to model
the socioeconomic determinants of the intake of FeMFP. From the results
in Table 3
for Hb status, the short- and long-run elasticities of
Hb with regard to FeBIO were 0.015 and 0.018, respectively. Using the
sample means in round 2 in Table 1
, a doubling of FeBIO to 0.40 mg/d
would result in a long-run increase of 3.6% in Hb. Because the
sample mean of Hb was 117.3 ± 11.6 g/L, the average woman would
then not be classified as having IDA. Moreover, a 10-fold increase of
FeBIO from 0.20 mg to the "safe" level of 2.0 mg (FAO/WHO 1988
) would be associated with a 18% increase in Hb, which
would greatly reduce the prevalence of IDA. The magnitude of such
increases is likely to be underestimated because within-subject
variation in iron intakes results in measurement errors that bias the
estimated coefficients downward (Liu et al. 1976
). At
any rate, there is evidence that even minor improvements in Hb have
beneficial effects on child-bearing, time allocation, morbidity
from infections, work productivity and cognition
(UNICEF/UNU/WHO/MI 1999
).
IDA among the poor in part results from high prices of animal products,
which reduces the intake of heme iron. Also, prices of fresh vegetables
and fruits show large seasonal fluctuations, making them less
affordable outside the harvest season. Thus, the enhancing effects of
vitamin C for iron absorption may be applicable primarily to the
well-off groups and not to those most likely to have IDA. Cereal
staples, such as rice, in Bangladesh have low iron and a high phytate
content. Taken together, these factors suggest that to reduce IDA, one
would need to evaluate the cost-effectiveness of instruments such
as iron fortification of rice, increased consumption of MFP and easier
access to iron tablets. Because women in our sample who took iron
tablets had 2.5% higher Hb levels and the costs of daily iron
supplementation were $0.25 per woman per month (Levin et al. 1993
), women, especially those of child-bearing age, would
be good candidates for daily or weekly iron supplementation
(Allen 2000
, Beard 1998
).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
3 Abbreviations used: FeBIO, bioavailable iron; FeMFP, iron from meat, fish and poultry; FeTOT, total iron; Hb,
hemoglobin; IDA, iron deficiency anemia; MFP, meat, fish and poultry; MUAC, mid upper arm circumference.
Manuscript received February 28, 2000. Initial review completed May 12, 2000. Revision accepted November 28, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1.
Allen L. H. Anemia and iron deficiency: effects on pregnancy outcome. Am. J. Clin. Nutr. 2000;71:1280S-1284S
2.
Basta S. S., Soekirman M. S., Karyadi D., Scrimshaw N. S. Iron deficiency anemia and the productivity of adult males in Indonesia. Am. J. Clin. Nutr. 1979;32:916-925
3. Beard J. L. Weekly iron intervention: the case for intermittent iron supplementation. Am. J. Clin. Nutr. 1998;68:209-212[Medline]
4. Bhargava A. Estimating short and long run income elasticities of foods and nutrients for rural south India. J. R. Stat. Soc. A 1991;154:417-432
5. Bhargava A. Modelling the health of Filipino children. J. R. Stat. Soc. A 1994;157:417-432
6. Bhargava A. Modeling the effects of maternal nutritional status and socioeconomic variables on the anthropometric and psychological indicators of Kenyan infants from age 0–6 months. Am. J. Phys. Anthropol. 2000;111:89-104[Medline]
7. Bhargava A., Sargan J. D. Estimating dynamic random effects models from panel data covering short time periods. Econometrica 1983;51:1635-1660
8. Bouis H. E., Briere B., Guitierrez L., Hallman K., Hassan N., Hels O., Quabili W., Quisumbing A., Thilsted S., Zihad Z. H., Zohir S. Commercial Vegetable and Polyculture Fish Production in Bangladesh: Their Impacts on Income, Household Resource Allocation, and Nutrition 1998 International Food Policy Research Institute Washington, D.C.
9. Brune M., Rossander-Hulten, Hallberg L., Gleerup A., Sandberg A.-S. Iron absorption from bread in humans: inhibiting effects of cereal fiber, phytate, and inositol phosphates with different numbers of phosphate groups. J. Nutr. 1992;122:442-449
10. Calloway D. H., Murphy S. P., Bunch S. User’s Guide to the International Minilist Nutrient Data Base 1994 Department of Nutritional Sciences University of California, Berkeley, CA.
11. Doyle W., Crawley H., Robert H., Bates C. J. Iron deficiency in older people: interactions between food and nutrient intakes with biochemical measures of iron: further analysis of the National Diet and Nutrition Survey of people aged 65 years or over. Eur. J. Clin. Nutr. 1999;53:552-559[Medline]
12.
Du S., Zhai F., Wang Y., Popkin B. M. Current methods of estimating dietary iron bioavailability do not work in China. J. Nutr. 2000;130:193-198
13. FAO/WHO Requirements of Vitamin A, Iron, Folate and B12: Report of a Joint FAO/WHO Expert Consultation 1988 Food and Agriculture Organization Rome, Italy
14.
Garcia-Casal M. N., Layrisse M., Solano L., Baron M. A., Arguello F., Llovera D., Leets I., Tropper E. Vitamin A and ß-carotene can improve nonheme iron absorption from rice, wheat and corn by humans. J. Nutr. 1998;128:646-650
15.
Gardner G. W., Edgerton V. R., Senewratne B., Barnard J. J., Chiro Y. Physical work capacity and metabolic stress in subjects with iron deficiency anemia. Am. J. Clin. Nutr 1977;30:910-917
16.
Garner C., Tatu T., Reittie J. E., Littlewood T., Darley J., Cervino S., Farrall M., Kelly P., Spector T. D., Thein S. L. Genetic influences on F cells and other hematologic variables: a twin heritability study. Blood 2000;95:342-346
17. Gorman W. M. Tastes, habits and choices. Int. Econ. Rev. 1967;8:218-222
18.
Hallberg L., Brune M., Rossander L. Iron absorption in man: ascorbic acid and dose-dependent inhibition by phytate. Am. J. Clin. Nutr. 1989;49:140-144
19.
Hallberg L., Hulten L., Gramatkovski E. Iron absorption from the whole diet in men: how effective is the regulation of iron absorption?. Am. J. Clin. Nutr 1997;66:347-356
20. James, W.P.T. & Ralph, A. (1994) The functional significance of low body mass index. Eur. J. Clin. Nutr. 48: (suppl. 3).
21. James W.P.T., Schofield E. C. Human Energy Requirements 1990 Oxford University Press Oxford, U.K.
22.
Khusun H., Yip R., Schultink W., Dillion D.H.S. World Health Organization hemoglobin cut-off points for the detection of anemia are valid for an Indonesian population. J. Nutr. 1999;129:1669-1674
23. Levin H., Pollitt E., Galloway R., McGuire J. Micronutrient deficiency disorders. Jamison D. Mosley W. Measham A. Bobadila J. eds. Disease Priority in Developing Countries 1993 Oxford University Press Oxford, U.K.
24. Liu K., Dyer J., Mc Keever J., McKeever P. Statistical methods to assess and minimize the role of intra-individual variability in obscuring the relationship between dietary lipids and serum cholesterol. J. Chron. Dis. 1976;31:399-418
25. Lozoff B. Behavioral changes in iron deficiency. Adv. Pediatr. 1988;35:331-360[Medline]
26. Monsen E. R., Balintfy J. L. Calculating dietary iron bioavailability: refinement and computerization. J. Am. Diet. Assoc. 1982;80:307-311[Medline]
27.
Monsen E. R., Hallberg L., Layrisse M., Hegsted D. M., Cook J. M., Mertz W., Finch C. A. Estimation of available dietary iron. Am. J. Clin. Nutr. 1978;31:134-141
28.
Morris S., Ruel M. T., Cohen R. J., Dewey K. G., de la Briere B., Hassan M. N. Precision, accuracy, and reliability of hemoglobin assessment with use of capillary blood. Am. J. Clin. Nutr. 1999;69:1243-1248
29.
Murphy S. P., Beaton G. H., Calloway D. H. Estimated mineral intakes of toddlers: predicted prevalence of inadequacy in village populations in Egypt, Kenya, and Mexico. Am. J. Clin. Nutr. 1992;56:565-572
30.
Nelson M., Black A. E., Morris J. A., Cole T. J. Between-and-within subject variation in nutrient intake from infancy to old age: estimating the number of days to rank dietary intakes with desired precision. Am. J. Clin. Nutr. 1989;50:155-167
31. Numerical Algorithm Group Numerical Algorithm Group Library, Mark 13 1989 Oxford University Oxford, U.K.
32. Pollitt E. Iron deficiency and cognitive function. Annu. Rev. Nutr. 1993;13:521-537[Medline]
33. Rahman H. Z., Hossain M., Sen B. 1987–1994: Dynamics of Rural Poverty in Bangladesh 1996 Bangladesh Institute of Development Studies Dhaka, Bangladesh
34. SAS Institute Statistical Analysis Software Version 6.12 1999 SAS Institute, Inc Cary, NC.
35. Scrimshaw N. S., Taylor C. E., Gordon J. E. Interactions of nutrition and infection. Am. J. Med. Sci. 1959;237:367-403[Medline]
36.
Stoltzfus R., Chwaya H. M., Tielsch J. M., Schulze K. J., Albonico M., Savioli L. Epidemiology of iron deficiency anemia in Zanzibari schoolchildren: the importance of hookworms. Am. J. Clin. Nutr. 1997;65:153-159
37. Tatala S., Svanberg U., Mduma B. Low dietary iron availability is a major cause of anemia: a nutrition survey in the Lindi district of Tanzania. Am. J. Clin. Nutr. 1998;68:171-178[Abstract]
38.
Tseng M., Chakraborty H., Robinson D. T., Mendez M., Kohlmeier L. Adjustment of iron intake for dietary enhancers and inhibitors in population studies: bioavailable iron in rural and urban residing Russian women and children. J. Nutr. 1997;127:1456-1468
39. UNICEF/UNU/WHO/MI Preventing Iron Deficiency in Women and Children: Background and Consensus on Key Technical Issues and Resource for Advocacy, Planning and Implementing National Programmes 1999 International Nutrition Foundation Boston, MA.
40. UNICEF/WHO Prevention and Control of Iron Deficiency Anaemia in Women and Children 1999 World Health Organization Geneva, Switzerland.
41.
Viteri F.E., Ali F., Tujague J. Long-term iron supplementation improves and sustains nonpregnant women’s iron status as well or better than currently recommended short-term daily supplementation. J Nutr 1999;129:2013-2020
This article has been cited by other articles:
|
|
 |
|
  S. C. Rodriguez, C. Hotz, and J. A. Rivera Bioavailable Dietary Iron Is Associated with Hemoglobin Concentration in Mexican Preschool Children J. Nutr., October 1, 2007; 137(10): 2304 - 2310. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. Beard, L. E. Murray-Kolb, J. D. Haas, and F. Lawrence Iron Absorption Prediction Equations Lack Agreement and Underestimate Iron Absorption J. Nutr., July 1, 2007; 137(7): 1741 - 1746. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Bhargava and A. Amialchuk Added Sugars Displaced the Use of Vital Nutrients in the National Food Stamp Program Survey J. Nutr., February 1, 2007; 137(2): 453 - 460. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Tuntipopipat, K. Judprasong, C. Zeder, E. Wasantwisut, P. Winichagoon, S. Charoenkiatkul, R. Hurrell, and T. Walczyk Chili, but Not Turmeric, Inhibits Iron Absorption in Young Women from an Iron-Fortified Composite Meal J. Nutr., December 1, 2006; 136(12): 2970 - 2974. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Moretti, M. B Zimmermann, S. Muthayya, P. Thankachan, T.-C. Lee, A. V Kurpad, and R. F Hurrell Extruded rice fortified with micronized ground ferric pyrophosphate reduces iron deficiency in Indian schoolchildren: a double-blind randomized controlled trial. Am. J. Clinical Nutrition, October 1, 2006; 84(4): 822 - 829. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Hunt Reversing Productivity Losses from Iron Deficiency: The Economic Case J. Nutr., April 1, 2002; 132(4): 794S - 801. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. E. Bouis Plant Breeding: A New Tool for Fighting Micronutrient Malnutrition J. Nutr., March 1, 2002; 132(3): 491S - 494. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
