![[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}
|
|
|
|
|
||
2,**
*
Section for International Maternal and Child Health, Department of Women’s and Children’s Health, Uppsala University, Uppsala, Sweden;
Epidemiology, Department of Public Health and Clinical Medicine, Umeå University, Umeå, Sweden;
**
Nutrition Academy, Ministry of Health, Yogyakarta, Indonesia and
Community Health and Nutrition Research Laboratory, Faculty of Medicine, University of Gadjah Mada, Yogyakarta, Indonesia
2To whom correspondence should be addressed at Section for International Maternal and Child Health, Entrance 11, Uppsala University, 751 85 Uppsala, Sweden. E-mail: viveka.persson{at}kbh.uu.se
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
0.6 for the
micronutrients. To estimate true individual average intake with a
precision of ±20%, six replicate recalls were sufficient for energy,
carbohydrates, vitamin A, iron and vitamin C. In conclusion, mean
intake of several nutrients can be reliably measured with the 24-h
recall method, using a limited number of days. The nutrient of
interest, the primary objectives and method of analyses should all be
taken into account when planning sample size and number of replicates.
KEY WORDS: • diet recall • pregnancy • reliability • variance • Indonesia • humans
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, FAO/WHO/UNU 1985
), activity (Banerjee et al. 1971
), appetite
(Coons 1933
) and self-selected diet (Dickens and Trethowan 1971
). Any dietary intake measurement is specific
to the stage of pregnancy.
Information on variation in nutrient intake (intraindividual and
interindividual) is required to guide decisions on the number of
replicate measurements needed and on sample sizes. If too few replicate
measures are taken or the sample size is too small, the statistical
precision of intakes can be jeopardized and measures of diet-health
outcome associations, such as correlations and relative risk, may be
attenuated (Freudenheim et al. 1989
, Liu et al. 1978
, Sempos et al. 1985
, Walker and Blettner 1985
). In contrast, taking too many replications or
too large a sample wastes resources and disturbs respondents without
serving any purpose.
Unfortunately, data assessing the precision of dietary intake methods
during pregnancy are scarce and most come from the Western World
(Nelson et al. 1989
, Osofsky 1975
,
Rush and Kristal 1982
). To our knowledge, only one study
has reported patterns of dietary variability in women from a developing
country (Launer et al. 1991
). Food intake in that study
was directly weighed for three consecutive d/mo from mo 6–9 in
pregnancy among 743 Indonesian women.
The primary goal of the present study was to describe the intraindividual (within) and interindividual (between) variability in energy and nutrient intake in each trimester of pregnancy among women in a developing country. Second, because the weighing method presents difficulties in large studies, we wanted to elucidate how reliably the more practical 24-h recall method could be used in a developing country. Finally, we used our estimates of variance to show the implications of using different number of days for estimating true average intake as well as relationships between dietary intake and health outcomes.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The study was conducted in Purworejo District, Central Java, Indonesia, which consists of 16 subdistricts and 494 villages. The total population is 750,000. Since 1994, the Faculty of Medicine at Gadjah Mada University has been operating a Community Health and Nutrition Research Laboratory in this area to support the Ministry of Health of Indonesia in developing and implementing a community health and nutrition surveillance program.
A two-stage cluster sampling method was used to select a 10%
sample of households representative of the district. The sampling frame
for the first stage consisted of a 20% sample of the enumeration areas
or "wilcah," developed by the Central Bureau of Statistics for the
1990 census. In the second stage, the same number (101) of households
was systematically sampled from each wilcah (Wilopo and team Community Health and Nutrition Laboratory 1997
). From 1994 to
1998, each household was visited every 3rd mo for data collection. In
addition, to identify new pregnancies, households with women likely to
become pregnant (i.e., married women of reproductive age) were visited
monthly. The nutritional status of women of reproductive age in the
area is described in detail elsewhere (Nurdiati et al. 1998
, Winkvist et al. 2000
).
Between April 1996 and October 1998, a cohort of 846 women in early pregnancy was recruited for a study on nutritional status during reproduction. Within this framework, dietary data were collected among a subsample of 493 women. The remaining 353 women were not included for the following reasons: refusal (n = 42); abortions, stillbirths or death (n = 32); being deaf, too shy or mentally ill (n = 23); migration (n = 21); or because of difficulties with the fieldwork during the economic crisis or because they were recruited before dietary assessments were started (n = 235). We also excluded women who did not have six complete 24-h recalls per trimester (n = 28, 43 and 67 in trimesters 1, 2 and 3, respectively). However, these women were only excluded from analyses for those trimesters where fewer than six recalls were completed. This caused the total study sample to drop from 493 to 451. Of these 451 women, 122, 406 and 356 women in trimesters 1, 2 and 3, respectively, were included in the dietary analysis, referred to as the study sample. Hence, 451 women with data on at least one trimester were included.
Dietary intake.
In each trimester, six 24-h recalls were used to estimate the dietary
intake of the individual women. This number was based on the results of
Launer et al. (1991
) on random variation and was
expected to be adequate to estimate intake of vitamin A and energy
within an error of ±20%. These replications were randomly distributed
among the five different days of the Javanese calendar on
nonconsecutive days. The time lapses (mean ± SD)
between the first and the sixth interview in the first, second and
third trimesters were 30 ± 5, 37 ± 12 and 35 ± 8 d, respectively. Detailed descriptions of all foods, beverages and
vitamin and mineral supplements consumed between 00:00 and midnight the
previous day, as well as cooking methods, were recorded. Exact recipes
were collected for each woman’s intake on each day. In total,
information on >11,000 recipes was collected. Quantities were
estimated using seven household utensils and 
20 types of food models
(e.g., fish, tomato, banana and meat). The average weight of each type
of food equivalent to the portion sizes and fitting into the household
utensils was estimated to the nearest gram. To calculate nutrient
intakes, the Indonesian nutrient composition database was used in most
cases (Department of Health, Indonesia, 1995
,
Hardiansyah and Briawan 1990
, Mahmud et al. 1990
); for example, 92% of all energy and fat values came from
these sources. Vitamin A values were predominantly taken from the food
composition table of de Pee and Bloem (1999
). The
standard conversion factor of 1 µg retinol = 6 µg of
ß-carotene was used (FAO 1988
).
The data were collected by 22 trained female interviewers, and each respondent was interviewed at home by the same interviewer for all measurements. The interview lasted 1–2 h. Data form editing was conducted in the field within a few days, and supervisors checked all interviewers periodically. Ethical approval was received from the research ethics committees of the medical faculties of Gadjah Mada University, Yogyakarta, Indonesia, and Umeå University, Umeå, Sweden.
Statistical methods.
Data were analyzed with SPSS/PC statistical software (Version 8.0). The distribution of each nutrient (the mean for each trimester) was tested for normality before analysis. Results for slightly skewed variables were transformed to logarithmic scale.
To address the first study goal, variance components are reported as
absolute values, ratios and coefficients of intraindividual (within)
variation (CVw)
.3
The Variance Components procedure in SPSS was used to calculate the
absolute values, and eq. 1
shows how CVw was calculated:
 | (1) |
where sw is the square root of the
estimated intraindividual variance. To evaluate another source of
variation, i.e., the trend in energy intake between each measurement
occasion of each trimester (expressed as change per day), we applied a
multilevel modeling technique (MlwiN 1.02). The first level was
measurement occasion, and the second level was each woman. The model
itself is a random regression model (Goldstein 1995
):
 |
where ß0 and ß1 are assumed to vary over the study population, and where Xij represents time between last menstrual period and measurement occasion.
To address our second goal, the reliability of 24-h recall, we used the
intraclass correlation coefficient (
coefficient,
rxx) as a function of number of recalls.
This is a measure derived by Cronbach et al. (1972
) that
indicates the degree of agreement among repeated measures of some
variable, in the present case, dietary intake. The Reliability Analysis
procedure in SPSS was used for this purpose.
For the third goal, values of CVw were used to
illustrate the required number of recalls per individual for the
various nutrients (Willett 1990
):
 | (2) |
where n is the number of replicate days required, and
Z is the normal deviate for the
percentage of times a confidence interval should cover the "true"
average mean intake of an individual (e.g.,
Z = 1.96 for 95% confidence).
Finally, D is half the length of the interval, as a percentage of the
mean.
In addition, the values of variance (sb2 and
sw2) were used to illustrate the error in a
regression analysis designed to look at the relation between
dietary intake and some hypothetical outcome, with observed dietary
intake as an independent variable. The error is measured by the ratio
of the observed-to-true slope coefficients (Beaton et al. 1979
), using eq. 3
:
 | (3) |
where b0 is observed regression coefficient, bt is true regression coefficient, r is number of replications per individual, sb2 is estimated interindividual variance component and sw2 is estimated intraindividual variance component. Independent sample t tests were used to determine whether those excluded from dietary analyses (n = 395) were different from the study sample (n = 451) with respect to socioeconomic status and nutritional status.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The age of the study sample was 28.8 ± 5.4 y, and the mean
parity was 1.6 ± 1.4. The mean height and mid-upper arm
circumference (MUAC) at 2 mo of pregnancy were 150.0 ± 4.9 and
25.1 ± 2.9 cm, respectively. Most (90%) lived in the rural
areas, and 46% had 7 y of schooling.
Representativeness of the study sample.
The sample of 451 women from the pregnancy cohort included in the dietary intake analyses did not differ significantly from those excluded (n = 395) with respect to age, parity, height, MUAC at 2 mo of pregnancy, altitude of residence, education, type of toilet or water source (P > 0.05). However, a greater proportion of those included in the analyses lived in the urban areas (10% versus 4.5%, P < 0.01), and fewer worked with agriculture (38% versus 47%, P < 0.05).
The mean height and age of the total sample of women of reproductive
age (n = 13,094) in Purworejo were 149.1 ± 5.1 cm
and 30.4 ± 9.7 y, respectively, of which both were different
from the study sample (P < 0.05). Forty-six
percent of these women had 7 y of schooling, and 14% worked with
agriculture, of which both were not different from the study sample.
However, a greater proportion, 14% compared with 10%, lived in the
urban areas (P < 0.01) (Nurdiati et al. 1998
).
Variance component analyses.
Intravariance and intervariance components for the nutrient intake of
the Indonesian women are shown by trimester in Table 1
. In all three trimesters, the ratio of intraindividual to
interindividual variation was <1.0 for energy and carbohydrates. Also
in all trimesters, the ratios were greatest for the micronutrients
(iron, vitamins A and C, thiamin and calcium). For most nutrients, the
ratios were lowest in the first trimester.
|
. For most nutrients, the CVw was highest in the
first trimester and lowest in the third. However, the differences in
CVw between the trimesters were not large. We
also evaluated those women with 18 complete 24-h recalls (n
= 84) to determine whether fluctuations in
CVw among these women were the same as the
fluctuations in CVw for the entire study sample.
The differences between the two samples were always <10% in all three
trimesters (data not shown).
|
Reliability of 24-h recalls.
Table 3
indicates the coefficient as a function of the number of 24-h
recalls in trimester 2. The reliabilities for two recalls are low for
the micronutrients but respectable for the macronutrients. When six
recalls are used, the coefficients for all nutrients fall above
0.55.
|
The number of 24-h recalls needed to estimate true average intakes of
individuals is presented in Table 2
. The number of replicates needed to
estimate true average intake within an error of ±10% would be beyond
the scope of most surveys for all nutrients. However, if an error range
of ±20% is accepted, six replicates would be sufficient to estimate
energy, carbohydrates, iron and vitamins A and C.
Dietary intake in regression analyses.
Relationships between dietary intake and health outcomes are often
evaluated with regression analyses. The ratio of the observed to true
regression slope as a function of the number of replicates, with the
different nutrients as the independent variable, is shown in
Table 4
. The results are based on the second trimester, but similar results
were found for the other two trimesters. In contrast to
CVw, the ratio of the observed to true regression
coefficient is a direct function of the ratio of intravariation to
intervariation. The attenuation of the true regression coefficient with
decreased replicates was greatest for the micronutrients. The
regression coefficient for energy and carbohydrates did not change
substantially by reducing the number of days from 6 to 4.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A major finding was that the intravariance-to-intervariance ratios were <1.0 for energy and carbohydrates in all three trimesters but were >1.0 for all other nutrients, except for protein in the first trimester. The largest variance ratios were found for the micronutrients. Hence, micronutrients are measured with greater error, meaning it will be more difficult to discover associations between micronutrient intake and certain outcomes than is the case for macronutrients. However, this study also showed that to estimate the mean intake of individuals with a precision of ±20% of the true intake, the current number of replicates (six) per individual would be sufficient for energy, carbohydrates, iron and vitamins A and C.
For most nutrients, a tendency toward larger CVw
values in the first trimester was seen, perhaps because for the
majority of nutrients, the intraindividual variation was largest
during the first trimester. This may in turn have been because
many women experienced nausea at that time. In addition, the nutrient intake was smallest that trimester, as shown in Table 1
.
In this study, we only accounted for intravariation and intervariation in nutrient intake. We could not examine the extent to which the estimate of usual intake contained other errors, such as nutrient content of foods, bioavailability of nutrients, nutrient interactions or errors in recording. We do believe that the food composition data we used were more accurate and complete than those used in most studies from Indonesia, because our project included extensive efforts to obtain details on recipes and to conduct additional examinations on the composition of foods that have not been analyzed previously.
Intraindividual variance has in earlier studies been shown to consist
of several different components; these include method of data
collection (Tarasuk and Beaton 1991
), sequence of
observation (Beaton et al. 1979
, Hartman et al. 1990
), day-of-the-week effect (Beaton et al. 1979
, Hankin et al. 1967
, Hartman et al. 1990
, McGee et al. 1982
) and seasonal
differences (Hartman et al. 1990
).
Intraindividual variance may also vary with changes in appetite, physical activity and intake that occur during pregnancy. We found an average increase of 45.23, 12.62 and 6.14 kJ/d during each trimester, although individual variation around this mean increase was large. Thus, there was no indication of fatigue in repeated recalls. Also, for the other nutrients, no trend of declining intake was seen according to sequence of the recall (data not shown).
To minimize the days of the week effect, we included all 5 d in
the Javanese calendar, accounting for market days when dietary habits
may change. The variation due to interview occasion was partitioned
from the total intravariance but was found to be <5% of the total
intravariance. The effect of season was not investigated in this study,
although the effect on macronutrient intake in this area is likely to
be small. The economic crisis in Indonesia, which started approximately
in October 1997, could also have contributed to variation in intake.
However, we compared the variance ratios of women who completed their
pregnancy before the crisis (50%) with those after the crisis and
found them to be quite similar.
To our knowledge, the only other study that has evaluated variance
components of pregnant women from the developing world is that of
Launer and colleagues (1991
) from East Java, Indonesia.
Their results are derived from mo 6–9 in pregnancy among 743 women,
whereas we present one estimate per trimester among 451 women. Launer
and colleagues used a 3-consecutive-d food weighing method every month,
whereas we used six nonconsecutive 24-h recalls within 1 mo in each
trimester. Although error may be greater (likely toward underreporting)
in measuring individual intake with the 24-h recall method than with
the direct weighing method (Block 1982
), recall is
cheaper and simpler, making it more suitable for larger studies. A
recent study (Harrison et al. 2000
) suggests that
underreporting was less of a problem in one developing country, Egypt.
It may also be easier with the 24-h recall method to assess
nonconsecutive days, which gives a better estimate of the variance
ratio (Block 1982
, Tarasuk and Beaton 1991
). The 24-h recall method is considered suitable for
measuring change over time (Tarasuk and Beaton 1992
) and
does not require literate respondents.
Launer and coworkers (1991
) suggest that "whether or
not the ratio of intra- to inter-individual variation derived from
recall data would differ from those reported here depends on the type
of error." For example, if errors in the recall data cause only a
systematic shift away from the true mean, this may not affect the
variance components or their ratio. However, other types of error could
inflate or deflate the ratios.
Because the two studies not only used different methods but were conducted in different areas of Central Java 14 y apart, a comparison of results should be made with caution. Most likely, the socioeconomic situation has improved, resulting in the higher energy and fat intakes seen among our women. For protein and vitamin A, the variance ratios are comparable (ranging from 0.82 to 1.50 and 3.12 to 3.44, respectively, for the three trimesters in our study, compared with 1.28 and 3.8, in their study), as were the CVw’s for energy (0.20–0.26, compared with 0.24 in their study). However, for energy and fat, our ratios were lower (0.57–0.78 and 1.30–1.50, respectively, compared with 1.35 and 3.24 in theirs). In contrast, the CVw for protein and vitamin A in our study was slightly higher (0.31–0.35 and 0.21–0.24, respectively, compared with 0.25 and 0.19 in their study).
The studies available from Western countries on pregnant women are
generally smaller (n = 60–225) and less complete
samples than these two from Indonesia. Two studies used the 24-h recall
method, either up to four repeated recalls taken during the second
trimester (Rush and Kristal 1982
) or up to four recalls
before delivery without specifying time points (Osofsky 1975
). A third one used weighed diet records (4 d at
approximately monthly intervals from week 12–16) (Nelson et al. 1989
). Their intravariance-to-intervariance ratios for energy
(1.14–1.4), carbohydrates (1.18–1.2) (Nelson et al. 1989
, Osofsky 1975
, Rush and Kristal 1982
) and vitamin A (4.9) (Nelson et al. 1989
)
were all higher than ours. For protein, they were either similar (1.38)
(Rush and Kristal 1982
) or higher (1.7–1.9)
(Nelson et al. 1989
, Osofsky 1975
). For
vitamin C (1.2) (Nelson et al. 1989) and calcium, they were
lower (0.98–1.8) (Nelson et al. 1989
, Rush and Kristal 1982
). The reasons behind the mostly lower variance
component ratios seen in our study could be that in developing
countries, diets tend to be more monotonous and what people eat is more
closely linked to income. In both cases, this would increase
interindividual variation in relation to intraindividual variation.
Implications regarding the design of future studies.
Our population-based study using the 24-h recall method has generated results similar to a population-based study conducted in the same country using the more accurate weighing method. This has several implications for future research.
First, it suggests that the usual mean intake of several nutrients
among pregnant women can be reliably measured using the more practical
24-h recall method. The reliability analysis (Table 3)
indicates that
it is possible to obtain good agreement with two or three repeated
recalls for the macronutrients. However, for vitamin A and vitamin C,
six replications are not sufficient to obtain an "acceptable" value
of 0.7 for the coefficient.
Second, our data on attenuation of the simple regression coefficients of the macronutrients suggest that it should be possible to use dietary intake data from 24-h recalls during pregnancy when evaluating the effect of diet on different outcomes of pregnancy. However, the regression coefficients of the micronutrients showed a much larger attenuation. Thus, for example, an association between the mother’s vitamin A intake and her breast milk vitamin A content would be difficult to find even with six replicate 24-h recalls, because the observed regression coefficient would be attenuated almost 35%.
Third, in our analyses, we evaluated the effect of using different
numbers of replications on CVw, regression
coefficients and coefficients. When the primary objective of a
study is to detect differences between groups of individuals, the
choices also include increasing sample size. The choice between
increasing sample size or increasing number of replicates would depend
on factors such as costs of repeated dietary sampling relative to the
cost of recruiting additional subjects as well as availability of
subjects. However, Freudenheim et al. (1989
) showed that
for weaker underlying associations, nondifferential misclassification
due to intraindividual variability induced bias in the odds ratio
(toward unity) would persist, even with a larger sample size.
Fourth, our variance component ratios for macronutrients were generally
lower than those reported for pregnant women in the industrialized
world (Nelson et al. 1989
, Osofsky 1975
,
Rush and Kristal 1982
). Thus, it may not be appropriate
to generalize findings on intraindividual-to-interindividual variance
derived from the Western world to populations in low-income
countries.
Finally, when using the 24-h recall method, the nutrient of interest, the primary objectives of the study and the methods of analysis to be used should all be taken into account when planning the sample size and number of replicate measures.
|
FOOTNOTES |
|---|
3 Abbreviations used: b0, observed regression coefficient; bt, true regression coefficient; CVw, coefficients of intraindividual (within) variation; MUAC, mid-upper arm circumference; Za, the normal deviate; r, number of replications per individual; sb2, estimated interindividual variation component; sw,2 estimated intraindividual variation component; sw, square root of the estimated intraindividual variation.
Manuscript received June 9, 2000. Initial review completed August 5, 2000. Revision accepted November 2, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
1. Banerjee B., Khew K., Saha N. A comparative study of energy expenditure in some common daily activities of non-pregnant and pregnant Chinese, Malay and Indian women. J. Obstet. Gynaecol. Br. Commonw. 1971;78:113-116[Medline]
2.
Beaton G. H., Milner J., Corey P., McGuire V., Cousins M., Stewart E., de Ramos M., Hewitt D., Grambsch P. V., Kassim N., Little J. A. Sources of variance in 24-h dietary recall data: Implications for nutrition study design and interpretation. Am. J. Clin. Nutr. 1979;32:2546-2549
3.
Block G. A review of validations of dietary assessment methods. Am. J. Epidemiol. 1982;115:492-505
4. Coons C. Dietary habits during pregnancy. J. Am. Diet. Assoc. 1933;9:95-106
5. Cronbach L., Aleser A., Nanda H., Rajarathan N. The Dependability of Behavioral Measurements 1972 Wiley New York.
6. De Pee S., Bloem M. 24-VASQ Method for Estimating Vitamin A Intake (24-Hour Vitamin A Semi-Quantitative) 1999 Helen Keller International Asia Pacific Regional Office Jakarta, Indonesia
7. Department of Health Departemen Kesehatan RI Direktorat Jenderal Pembinaan Kesehatan Masyarakat Pusat Penelitian dan Pengembangan Gizi [Daftar Komposisi zat gizi Pangan Indonesia.] 1995 Jakarta Indonesia
8. Dickens G., Trethowan W. Cravings and aversions during pregnancy. J. Psychosom. Res. 1971;15:259-268[Medline]
9. FAO/COLLAB> Requirements of Vitamin A, Iron, Folate, and Vitamin B-12. Food and Agricultural Organization Food and Nutrition Series No. 23 1988 FAO Rome, Italy
10. FAO/WHO/UNU/COLLAB> Energy and Protein Requirements. World Health Organization Technical Report Series No. 724 1985 WHO Geneva, Switzerland
11.
Freudenheim J. L., Johnson N. E., Wardrop R. L. Nutrient misclassification: Bias in the odds ratio and loss of power in the Mantel test for trend. Int. J. Epidemiol. 1989;18:232-238
12. Goldstein H. Multilevel Statistics Models 2nd ed. 1995 Arnold London, UK.
13. Hankin J., Reynolds W., Margen S. A short dietary method for epidemiologic studies. Am. J. Clin. Nutr. 1967;20:935-945[Abstract]
14. Hardiansyah, Briawan D. Penilaian dan Perencanaan Konsumsi Pangan (Judgement and Planning for Food Consumption) 1990 Jurusan gizi Masyarakat dan Sumber daya Keluarga Fakultas Pertanian, Institut Pertanian Bogor, Indonesia
15.
Harrison G. G., Galal O. M., Ibrahim N., Khorshid A., Stormer A., Leslie J., Saleh N. T. Underreporting of food intake by dietary recall is not universal: A comparison of data from Egyptian and American women. J. Nutr. 2000;130:2049-2054
16.
Hartman A. M., Brown C. C., Palmgren J. Variability in nutrient and food intakes among older middle-aged men: Implications for design of epidemiologic and validation studies using food recording. Am. J. Epidemiol. 1990;132:999-1012
17. Launer L., Kardjati S., Kusin J., Reed G. Patterns of variability in the nutrient intake of nutritionally vulnerable pregnant women. Eur. J. Clin. Nutr. 1991;45:131-138[Medline]
18. Liu K., Stamler J., Dyer A., McKeever 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. 1978;31:399-418[Medline]
19. Mahmud M., Slamet D., Apriyantono R., Hermana Komposisi zat gizi Pangan Indonesia 1990 Departemen Kesehatan RI Direktorat Bina Gizi Masyarakat dan Pusat Penelitian dan Pengembangan Gizi Jakarta, Indonesia.
20.
McGee D., Rhoads G., Hankin J., Yano K., Tillotson J. Within-person variability of nutrient intake in a group of Hawaiian men of Japanese ancestry. Am. J. Clin. Nutr. 1982;36:657-663
21.
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 required to rank dietary intakes with desired precision. Am. J. Clin. Nutr. 1989;50:155-167
22. Nurdiati D., Hakimi M., Wahab A., Winkvist A. Concurrent prevalence of chronic energy deficiency and obesity among women in Purworejo, Central Java, Indonesia. Food Nutr. Bull. 1998;19:321-333
23. Osofsky H. Relationship between prenatal medical and nutritional measures, pregnancy outcome and early infant development in an urban setting. I. The role of nutritional intake. Am. J. Obstet. Gynecol. 1975;123:682-690[Medline]
24.
Rush D., Kristal A. R. Methodologic studies during pregnancy: The reliability of the 24-hour dietary recall. Am. J. Clin. Nutr. 1982;35:1259-1268
25.
Sempos C. T., Johnson N. E., Smith E. L., Gilligan C. Effects of intraindividual and interindividual variation in repeated dietary records. Am. J. Epidemiol. 1985;121:120-130
26.
Tarasuk V., Beaton G. H. The nature and individuality of within-subject variation in energy intake. Am. J. Clin. Nutr. 1991;54:464-470
27.
Tarasuk V., Beaton G. H. Statistical estimation of dietary parameters: Implications of patterns in within-subject variation: A case study of sampling strategies. Am. J. Clin. Nutr. 1992;55:22-27
28. Thompson F., Byers T. Dietary assessment resource manual. J. Nutr. 1994;124:2245S-2317S
29.
Walker A., Blettner M. Comparing imperfect measures of exposure. Am. J. Epidemiol. 1985;121:783-790
30. Willett W. Nutritional Epidemiology 1990 Oxford University Press, Inc New York.
31. Wilopo S., Team, C.H.N.-R.L. Key issues on the research design, data collection and management 1997 Community Health and Nutrition Research Laboratory, Faculty of Medicine, Gadjah Mada University, Yogyakarta, Indonesia
32. Winkvist A., Nurdiati D., Stenlund H., Hakimi M. Predicting over- and undernutrition among women of reproductive age: A population-based study in Central Java, Indonesia. Public Health Nutr 2000;3:193-200[Medline]
This article has been cited by other articles:
|
|
 |
|
  I. Huybrechts, D. De Bacquer, B. Cox, E. H. Temme, H. Van Oyen, G. De Backer, and S. De Henauw Variation in energy and nutrient intakes among pre-school children: implications for study design Eur J Public Health, October 1, 2008; 18(5): 509 - 516. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Jahns, A. Carriquiry, L. Arab, T. A. Mroz, and B. M. Popkin Within- and Between-Person Variation in Nutrient Intakes of Russian and U.S. Children Differs by Sex and Age J. Nutr., November 1, 2004; 134(11): 3114 - 3120. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Winkvist, H. Stenlund, M. Hakimi, D. S Nurdiati, and M. J Dibley Weight-gain patterns from prepregnancy until delivery among women in Central Java, Indonesia Am. J. Clinical Nutrition, June 1, 2002; 75(6): 1072 - 1077. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Nyambose, K. G. Koski, and K. L. Tucker High Intra/Interindividual Variance Ratios for Energy and Nutrient Intakes of Pregnant Women in Rural Malawi Show That Many Days Are Required to Estimate Usual Intake J. Nutr., June 1, 2002; 132(6): 1313 - 1318. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
