Estimating micronutrient intakes from Household

Consumption and Expenditures Surveys (HCES):

An example from Bangladesh

Odilia I. Bermudez, Keith Lividini, Marc-Francois Smitz, and John L. Fiedler

Abstract population and observed wide differences among divi-

sions. However, the methodological approaches reported
Background. Globally, there is a scarcity of national here, although feasible and promising, need to be vali-
food consumption data that could help to assess food dated with other dietary intake methods.
patterns and nutrient intakes of population groups.
Estimates of food consumption patterns and apparent
intakes of energy and nutrients could be obtained from Key words: Bangladesh, food consumption, HCES
national Household Consumption and Expenditures Household Consumption and Expenditures Surveys,
Surveys (HCES). micronutrient intakes
Objective. To use the HCES conducted in Bangladesh
in 2005 (HIES2005) to estimate apparent intakes of
vitamin A, iron, and zinc. Introduction
Methods. Food acquisition data from HIES2005,
which surveyed 10,080 households, were transformed Technical and financial constraints limit the acquisi-
into standard measurement units. Intrahousehold food tion of nutrient intake data from direct assessments of
and nutrient distribution was estimated with Adult dietary intakes in developing countries. To overcome
Male Equivalent (AME) units. Adequacy of intake was this limitation, alternative sources of information about
assessed by comparing individual nutrient intakes with food and nutrient intake patterns of population groups
requirements and was then aggregated by households. are commonly used, but they may be inaccurate. New
Results. The weighted mean energy intake for the and less expensive techniques for direct measurement
population was 2,151 kcal/person/day, with a range of dietary intakes of large population groups are desir-
among divisions from 1,950 in Barisal to 2,195 in Dhaka able, as is the development of alternative strategies,
division. The apparent intakes of vitamin A and iron including the possibility of using food expenditure data
were insufficient to satisfy the recommended intakes for for estimating food and nutrient intakes [1].
more than 80% of the population in Bangladesh, while Bangladesh is one of those countries with limited
apparent intakes of zinc, adjusted by bioavailability, national measurements of nutrient intakes, but with
satisfied the requirements of approximately 60% of the high-quality data from Household Consumption and
population. Expenditures Surveys (HCES), such as the one con-
Conclusions. Using the HIES2005, we were able to ducted in 2005 [2, 3]. In this study, we developed esti-
produce estimates of apparent food consumption and mates of food consumption and specific micronutrient
intakes of some key micronutrients for the Bangladeshi intakes from food acquisition data generated from the
2005 survey. We report the analysis of the contents
of vitamin A, iron, and zinc in diets consumed at the
Odilia I. Bermudez is affiliated with Tufts University household level.
School of Medicine, Boston, Massachusetts, USA; Keith Livi-
dini and John L. Fiedler are affiliated with HarvestPlus/Inter-
national Food Policy Research Institute (IFPRI), Washington,
DC; Marc-Francois Smitz is an independent consultant based Methods
in Washington, DC.
Please direct queries to the corresponding author: Odilia
I. Bermudez, Department of Public Health and Community Study population
Medicine, Tufts University School of Medicine, 136 Harrison
Ave., Boston, MA 02111, USA; e-mail: Odilia.Bermudez@ Data for this study were obtained from the 2005
tufts.edu. HCES conducted in Bangladesh (HIES2005). This

S208 Food and Nutrition Bulletin, vol. 33, no. 3 (supplement) © 2012, The United Nations University.
Estimating micronutrient intakes from HCES in Bangladesh S209

was a national survey using a two-stage stratified these vegetables were listed together in one line and we
sampling design [2]. The HIES2005 examined 10,080 have no information about the frequency of consump-
households, of which 6,400 were rural and 3,680 were tion of those foods by the Bangladeshi population. For
urban [3]. The households contained a total of 48,969 mixed dishes eaten away from home, we developed
members. We obtained exempt status from the Insti- recipes based on published papers [7, 11, 12], consul-
tutional Review Board at Tufts Medical Center/Tufts tation with Bangladeshi colleagues, and descriptions
University, as this study consisted of secondary analysis found on the Internet.
of a publicly available dataset that was free of personal
identifiers. Intrahousehold food distribution

Food data The food acquisition data in the HIES2005 were col-
lected for households as the sampling unit. In order
As part of the HIES2005, a questionnaire with 134 to estimate nutrient intakes for individual household
lines or categories was employed to collect data on members and compare them with nutrient intake rec-
food acquisition that occurred during the previous ommendations, we divided the total household intake
14 days [4]. The food lines consisted of individual or by the number of household members.
multiple foods. Food lines included spices, alcoholic To perform the intrahousehold food allocation, we
and nonalcoholic beverages, and foods eaten away assumed that household members did not receive an
from home, some of which were mixed meals (e.g., equal share of the food available for consumption.
fish and vegetable dishes). Household food acquisition Therefore, instead of working with estimates of per
data were transformed into standard units of weight capita distribution, we calculated Adult Male Equiva-
(grams) or volume (milliliters) and then converted to lent units (AMEs) for each household member, fol-
daily amounts. lowing the steps outlined by Fiedler et al. [1]. In short,
AMEs were estimated as the ratio of the energy require-
Sources of foods ment of a household member of a particular age and
sex to the energy requirement of an adult male 18 to
Food reports from HIES2005 included the sources 30 years of age, with moderate physical activity, as rec-
of acquired foods: purchased foods, foods produced ommended by the Food and Agriculture Organization
by household members, or foods received as gifts or (FAO) [13]. The AMEs are listed in table 1. There are
donations (gifted) or as in-kind wages. A categorical two limitations to these estimations: the HIES2005 did
variable was created for food sources with four catego- not report the age of infants in months, and the iden-
ries: purchased, produced, wage in-kind, and gifted. tification of women as pregnant or lactating was not
available. Therefore, we assigned AMEs for all children
Nutrient database under 1 year of age as that of children 7 to 12 months
of age. For women, no adjustments were made, and all
As we were unable to find a current national (Bangla- women were treated as nonpregnant and nonlactating.
desh) nutrient database, we developed a nutrient data-
base for foods listed in the HIES2005 based on the US Estimation of nutrient intakes and requirements
Department of Agriculture nutrient database version
22 [5], plus information about regional foods obtained We performed the nutrient analysis with the estimates
from published reports [6–10]. Additional information of daily apparent food intakes. Correction factors were
was identified from a nutrient database developed by obtained from the US Department of Agriculture nutri-
HarvestPlus that was used for the analysis of a nutri- ent database [5] to account for nonedible portions and
tion survey conducted in one district of Bangladesh*. cooking losses. In addition to dietary energy, the nutri-
In addition, we consulted Bangladeshi colleagues and ents assessed were iron, zinc, and vitamin A.
others familiar with Bangladesh about the identifica- Household total intakes of energy, vitamin A, iron,
tion of local foods. and zinc were allocated to household members accord-
As indicated earlier, some food lines from the ing to their AMEs. Apparent individual intakes were
HIES2005 questionnaire included more than one compared with the requirements established for vita-
food. In those cases, we estimated the nutrient content min A, iron, and zinc. To assess adequacy of the
for those foods as the average of the foods included in apparent intakes of vitamin A, we used the Estimated
those food lines. For example, we estimated the aver- Average Requirements (EARs) promulgated by the
age nutrient content of spinach, amaranth, and basil, as Institute of Medicine (IOM)[14].
Although the Dietary Reference Intakes (DRIs)
* Arsenault J. Bangladesh food composition table. Docu- established by the IOM were defined using bioavail-
mentation manual. HarvestPlus (Internal report, unpub- ability as one of several factors taken into account for
lished). 2010. nutrients such as iron and zinc, those bioavailability
S210 O. I. Bermudez et al.

TABLE 1. Adult Male Equivalent (AME) consumption units TABLE 2. Intake of dietary energy according to division, from
applied to household members from the 2005 Bangladesh the 2005 Bangladesh Household Income and Expenditure
Household Income and Expenditure Survey Surveya
Males Females Population Energy intake
Age range Energy Energy (expanded sample) (kcal)
(yr) (kcal/day) AME (kcal/day) AME Households Individuals
< 1.0 661 0.22 661 0.22 Division (no.) (no.) Mean SE
1.0–1.9 950 0.31 850 0.28 National 28,644,938 138,817,749 2,151 3
2.0–2.9 1,125 0.37 1,050 0.34 Barisal 1,789,213 8,905,211 1,950 11
3.0–3.9 1,250 0.41 1,150 0.38 Chittagong 4,932,030 26,727,815 2,111 8
4.0–4.9 1,350 0.44 1,250 0.41 Dhaka 9,521,146 44,714,603 2,195 7
5.0–5.9 1,475 0.48 1,325 0.43 Khulna 3,456,250 16,292,108 2,126 9
6.0–6.9 1,575 0.52 1,425 0.47 Rajshahi 7,368,438 33,379,209 2,181 7
7.0–7.9 1,700 0.56 1,550 0.51 Sylhet 1,577,863 8,798,802 2,177 15
8.0–8.9 1,825 0.60 1,700 0.56 a. All analyses were performed with weighted data.
9.0–9.9 1,975 0.65 1,850 0.61
10.0–10.9 2,150 0.70 2,000 0.66 that reported four or fewer food items over the 14-day
11.0–11.9 2,350 0.77 2,150 0.70 period used for data collection or those in which
12.0–12.9 2,550 0.84 2,275 0.75 members had a mean intake of less than 500 kcal/day.
13.0–13.9 2,775 0.91 2,375 0.78 To further correct for stockpiling or for overreporting,
14.0–14.9 3,000 0.98 2,450 0.80 we identified households with mean individual intakes
15.0–15.9 3,175 1.04 2,500 0.82 above 6,000 kcal. After applying these steps, we identi-
16.0–16.9 3,325 1.09 2,500 0.82 fied only eight households with implausible data, which
17.0–17.9 3,400 1.11 2,500 0.82 were excluded from our analysis.
18.0–29.9 3,050 1.00 2,400 0.79 We applied the cutpoint method [14, 16] to evaluate
30.0–59.9 2,950 0.97 2,350 0.77 the adequacy of intakes of vitamin A and zinc. With
≥ 60.0 2,450 0.80 2,100 0.69 the use of this method, household members with
intakes below the recommendations for their age and
sex were classified as having inadequate intakes of that
factors were based on the “typical American” diet, micronutrient.
which is higher in animal sources of iron and zinc and For iron, the distribution of requirements is not
lower in phytates and other food components than normally distributed for women of reproductive years
those observed in countries like Bangladesh. There- and young children [14]. Therefore, the full probability
fore, we set different levels of bioavailability for those method [14] was required to estimate the percentage of
nutrients. For iron, an assumption of 5% bioavailabil- inadequate intakes for these subgroups, as well as for
ity was used to adjust intakes; values of physiological the other age and sex groups. Individuals were assigned
requirements were obtained from the IOM DRIs [14]. a probability of inadequacy based on which range their
The International Zinc Nutrition Consultative Group intakes fell into. The number and percentage of inad-
(IZiNCG) has estimated the bioavailability for differ- equate intakes for each subgroup were then determined
ent age and sex groups based on different diets with by summing the probabilities over that group.
varying amounts of phytate content [15]. We made our Estimates of the average apparent intakes of energy,
assumption of bioavailability based on an unrefined, vitamin A, iron, and zinc, as well as the percentages of
cereal-based diet reported by IZiNCG and used the individuals with inadequate intakes, were calculated
physiological requirements published by IZiNCG [15] according to age group, area of residence (urban or
as the standards with which to compare our adjusted rural), and country divisions. We used the statistical
intakes of zinc. package PASW-SPSS, version 18, for our analyses. For
all analyses, we weighted the data using the expansion
Statistical analysis factors provided in the HIES data sets. Here we present
only results aggregated by Bangladesh’s six geopo-
The plausibility of our apparent food and nutrient litical divisions: Barisal, Chittagong, Dhaka, Khulna,
intakes was assessed with exploratory analyses, includ- Rajshahi, and Sylhet.
ing simple counts of food items by household, fre-
quency tables, simple identification of extreme values,
plot charts, and others. Obvious stockpiled quantities Results
were also detected and labeled as outliers. We also
assessed issues of underreporting, flagging households Sixty-four percent of the Bangladeshi households
Estimating micronutrient intakes from HCES in Bangladesh S211

included in the HIES2005 were from TABLE 3. Intakes of vitamin A, iron, and zinc by household members, accord-
rural areas. The average household ing to division, from the 2005 Bangladesh Household Income and Expenditure
size was 4.8 members, and was higher Survey a
for rural (4.9) than for urban (4.7) Vitamin A (µg) Iron (mg)b Zinc (mg)b
households. Division Mean SE Mean SE Mean SE
National 286 1 0.72 0.002 2.06 0.005
Intake of dietary energy
Barisal 301 3 0.69 0.005 1.75 0.011
The population studied in the HIES2005 Chittagong 285 2 0.75 0.004 2.05 0.009
was distributed across the six divisions Dhaka 285 2 0.78 0.004 2.18 0.009
of Bangladesh, with the highest propor- Khulna 354 4 0.71 0.005 1.95 0.012
tion of households concentrated in the Rajshahi 266 2 0.62 0.003 2.04 0.010
division of Dhaka (table 2). As reflected Sylhet 235 3 0.68 0.006 1.98 0.018
in this table, the apparent daily intake a. All analyses were performed with weighted data.
of energy fluctuated from 1,950 kcal b. After adjustment for bioavailability: iron, 5%; zinc, physiological requirements based
among households in the Barisal divi- on an unrefined, cereal-based diet reported by IZiNCG [15].
sion to 2,195 kcal in Dhaka division,
with 2,151 kcal as the national average.
80
Intakes of vitamin A, iron, and zinc
70

The mean daily intakes and standard errors of vita- 60

min A, iron, and zinc are presented in table 3. The
% adequacy

50
lowest mean intake of vitamin A, 235 µg of retinol
40
activity equivalents (RAE), was observed among the
population from Sylhet division, while the highest 30

mean intake, 354 µg of RAE, occurred in Khulna divi- 20

sion. Iron intakes, adjusted for a bioavailability of 5%,
10
ranged between 0.6 mg in Rajshahi division to 0.8 mg
in Dhaka division. For zinc, also adjusted by bioavail- 0
Sylhet Rajshahi Khulna Dhaka Chittagong Barisal National
ability, intakes ranged from 1.8 mg in Barisal to 2.2 mg
in Dhaka. FIG. 1. Weighted mean intake of vitamin A as percentage of
the Estimated Average Requirement (EAR), by division, from
the 2005 Bangladesh Household Income and Expenditure
Adequacy of intake of vitamin A, iron, and zinc Survey. All analyses were performed with weighted data
The adequacy of daily intakes of vitamin A, iron, and
zinc was determined, by individual, for national and
division aggregates, as represented in figure 1. The 80
apparent intake levels of vitamin A were equivalent to
70
61% adequacy when compared with the requirements
for this nutrient. The lowest proportion of adequacy 60
of vitamin A intake was observed in Sylhet division
% adequacy

50
(51%), while residents of Khulna division appeared to
40
obtain 74% of their required vitamin A.
Apparent intakes of iron covered 67% of the require- 30

ments for the Bangladeshi population, with the divi- 20

sions of Dhaka (73%) and Chittagong (70%) the only 10
ones above the national average (fig. 2). The mean
intake of zinc, adjusted by bioavailability, exceeded 0
Sylhet Rajshahi Khulna Dhaka Chittagong Barisal National
the requirements in all divisions and at the national
level (fig. 3). FIG. 2. Weighted mean intake of iron, adjusted for bioavail-
ability, as percentage of the physiological requirements by
division, from the 2005 Bangladesh Household Income
Population groups at risk for inadequate intakes of and Expenditure Survey All analyses were performed with
micronutrients weighted data. Values of physiological requirements for iron,
with an assumption of 5% bioavailability, were obtained from
We assessed the extent of inadequate intakes of vitamin the Dietary Reference Intakes Handbook produced by the
A, iron, and zinc among the Bangladeshi population in Institute of Medicine [15].
 S212 O. I. Bermudez et al.

140
were consistent with previously reported research. Vita-
120 min A has been considered one of the most limiting
micronutrients in diets of the Bangladeshi population.
100
According to the national nutrition surveys of 1975/76
% adequacy

80 and 1981/82, vitamin A intake was highly inadequate,

with just 36% of the recommended level reported in
60
1975/76 and 39% in 1981/82 [18, 19]. Anemia, often
40 associated with low iron intake, is a prevalent problem
in Bangladesh. The national prevalence of anemia
20
was estimated at 74% in 1995/96 [20], which has been
0
Sylhet Rajshahi Khulna Dhaka Chittagong Barisal National
partly explained by a low consumption of iron-rich
food sources [6].
FIG. 3. Weighted mean intake of zinc, adjusted for bioavail- Although the use of nondietary surveys to estimate
ability, as percentage of the physiological requirements, by food consumption is possible, it is important to take
division, from the 2005 Bangladesh Household Income into consideration certain limitations. HCES usually
and Expenditure Survey. All analyses were performed with collect data for nondietary purposes, with limited
weighted data. Estimations of physiological requirements of
zinc were based on an unrefined, cereal-based diet reported
information about food preparation, nonedible com-
by IZiNCG [16]. ponents, eating practices, and intrahousehold food
distribution. The food lists are constructed primarily
using economic, not dietary, considerations, which
order to identify the groups at risk for micronutrient may limit information on certain foods potentially rich
deficiencies. A high risk of inadequate intake of vitamin in specific nutrients, but of limited interest in terms
A was observed for the total (national) Bangladeshi of food purchases or production (e.g., wild plants or
population (table 4). Over 80% of households had a small game). Moreover, HCES omit the collection of
low apparent intake of vitamin A. Sylhet division had data about use of nutrient supplements, as was the case
the highest proportion of households with insufficient for the HIES2005.
apparent intake of vitamin A. We applied AMEs for the intrahousehold distri-
The probability of inadequate intake of iron was bution of food intakes among individual household
also very high, according to our estimates based on the members, with certain limitations due to the lack of
HIES2005. As observed in table 4, at the national level details about age of infants and pregnancy status of
87% of the population was identified as having a high women of reproductive age. With those details, it would
probability of inadequate iron intake. be possible to refine estimates, as reported by other
Even though apparent intakes of zinc exceeded aver- researchers [21].
age requirements, as noted above (fig. 3), 41% of the We acknowledge the importance of using more
total population was at risk for inadequate intake of direct methods (e.g., food records and 24-hour recalls)
this mineral (table 4). Moreover, we found inadequate for assessing dietary and nutrient intakes. However,
apparent intakes of bioavailable zinc for more than technical and economical limitations prevent many
40% of the populations of the divisions of Sylhet (41%), developing countries from periodically conducting
Rajshahi (44%), Khulna (50%), and Barisal (53%). these dietary surveys. Because of this, efforts should
continue to improve the estimation of food consump-
tion and nutrient intake data from HCES such as the
Discussion HIES2005. The methodological approaches reported
here, although feasible and promising, need to be
Using the food acquisition data from Bangladesh validated with other food and nutrient intake methods.
HIES2005, we found that it is possible to obtain esti- Nutrient analyses, such as those reported here, are
mates of energy and specific micronutrient intakes by valuable, as the results could guide the identification
using HCES. A particular strength of the data was the of vulnerable populations that could be targeted for
inclusion of foods eaten away from home, which gave micronutrient fortification, supplementation programs,
a more complete picture of apparent food consumption or other types of intervention.
in Bangladesh.
The apparent energy consumption obtained from
our analysis was consistent with other reports. Accord- Acknowledgments
ing to information collected by the World Food Pro-
gramme for Bangladesh, the average intake has been Partial support was obtained from the Nutrition and
estimated at 2,122 kcal [17], which is remarkably close Economic Research Support for Grand Challenge No.
to our estimate of 2,151 kcal. 9 Projects Grant OPP52013 to HarvestPlus from the
For micronutrients, we also found that our results Bill & Melinda Gates Foundation.
Estimating micronutrient intakes from HCES in Bangladesh S213

References
1. Fiedler JL, Lividini K, Bermudez OI, Smitz M-F. House- 19:562–571.
hold consumption and expenditures surveys (HCES): a 12. Roos N, Islamy MM, Shakuntala H, Thilsted SH. Small
primer for food and nutrition analysts. Food Nutr Bull indigenous fish species in Bangladesh: contribution
2012, 33:S170–84. to vitamin A, calcium and iron intakes. J Nutr 2003,
2. Asian Development Bank, Bangladesh Resident Mis- 133:4021S–4026S.
sion. Bangladesh Quarterly Economic Report. Sep- 13. FAO/WHO. Human energy requirements. Report of
tember 2006. URL: http://www.adb.org/Documents/ a Joint FAO/WHO/UNU Expert Consultation. Rome,
Economic_Updates/BAN/2006/sep-2006.pdf Accessed 17–24 October 2001. FAO Food and Nutrition Technical
April 20, 2010. Report series 1. Rome: FAO, 2004.; 2004.
3. Bangladesh Bureau of Statistics. Key findings of HIES 14. Food and Nutrition Board, Standing Committee of
2005. URL: http://www.bangladeshembassy.com.cn/em/ the Scientific Evaluation of Dietary Reference Intakes.
bbs/bbs.asp Accessed April 20, 2010. Dietary reference intakes: applications in dietary
4. Bangladesh Bureau of Statistics. Household income assessment. A Report of the Subcommittees on Inter-
and expenditure survey (HIES); Survey questionnaire, pretation and Uses of Dietary Reference Intakes and
2005. URL: http://surveynetwork.org/home/index. Upper Reference Levels of Nutrients, and the Standing
php?q=activities/catalog/surveys/ihsn/050-2005-001 Committee on the Scientific Evaluation of Dietary Ref-
Accessed March 9, 2010. erence Intakes. Washington, DC: Institute of Medicine,
5. U.S. Department of Agriculture, Agricultural Research National Academy Press; 2001.
Service. USDA national nutrient database for standard 15. Brown KH, Rivera JA, Bhutta Z, Gibson RS, King JC,
reference, Release 22. Nutrient data laboratory home Lonnerdal B, Ruel MT, Sandtrom B, Wasantwisut E,
page, http://www.ars.usda.gov/ba/bhnrc/ndl. Accessed Hotz C. International zinc nutrition consultative group
February 11, 2010. (IZiNCG) technical document #1. Assessment of the
6. Ahmed F, Mahmuda I, Sattar A, Akhtaruzzaman M. risk of zinc deficiency in populations and options for
Anaemia and vitamin A deficiency in poor urban preg- its control. Food Nutr Bull 2004, 25:S99–203.
nant women of Bangladesh. Asia Pac J Clin Nutr 2003, 16. Food and Nutrition Board, Standing Committee of
12:460–466. the Scientific Evaluation of Dietary Reference Intakes.
7. Hels O, Larsen T, Christensen LP, Kidmose U, Hassan Dietary reference intakes for vitamin A, vitamin K,
N, Thilsted SH. Contents of iron, calcium, zinc and arsenic, boron, chromium, copper, iodine, iron, man-
β-carotene in commonly consumed vegetables in Bang- ganese, molybdenum, nickel, silicon, vanadium, and
ladesh. J Food Comp & Analysis 2004, 17:587–595. zinc. Washington, DC: Institute of Medicine, National
8. Hels O, Kidmose U, Larsen T, Hassan N, Tetens I, Academy Press; 2001.
Thilsted S. Estimated nutrient intakes and adequacies 17. World Food Program. Food security at a glance: Bang-
in Bangladesh change when newer values for vitamin ladesh. URL: http://foodsecurityatlas.org/bgd/country/
A, iron and calcium in commonly consumed foods are food-security-at-a-glance. Accessed September 5, 2011.
applied. Int J Food Sci Nutr 2003, 54:457–465. 18. Ahmed F. Vitamin A deficiency in Bangladesh: a review
9. Hels O, Hassan N, Tetens I, Thilsted S. Food consump- and recommendations for improvement. Public Health
tion, energy and nutrient intake and nutritional status in Nutr 1999, 2:1–14.
rural Bangladesh: changes from 1981–1982 to 1995–96. 19. Ahmed F, Azim A, Akhtaruzzaman M. Vitamin A defi-
Eur J Clin Nutr 2003, 57:586–594. ciency in poor, urban, lactating women in Bangladesh:
10. Alam N, Roy SK, Ahmed T, Ahmed AM. Nutritional factors influencing vitamin A status. Public Health Nutr
status, dietary intake, and relevant knowledge of ado- 2003, 6:447–452.
lescent girls in rural Bangladesh. J Health Popul Nutr, 20. Ahmed F. Anaemia in Bangladesh: a review of preva-
28:86–94. lence and aetiology. Public Health Nutr 2000, 3:385–393.
11. Kidmosea U, Yangb RY, Thilsted SH, Christensena 21. Claro RM, Bertazzi-Levy R, Bandoni DH, Mondini L. Per
LP, Brandtd K. Content of carotenoids in commonly capita versus adult-equivalent estimates of calorie avail-
consumed asian vegetables and stability and extract- ability in household budget surveys. Cad Saude Publica
ability during frying. J Food Comp & Analysis 2006, 2010, 26:2188–2195.