Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Comparative Study
, 101 (5), 1038-46

Assessing Global Dietary Habits: A Comparison of National Estimates From the FAO and the Global Dietary Database

Affiliations
Comparative Study

Assessing Global Dietary Habits: A Comparison of National Estimates From the FAO and the Global Dietary Database

Liana C Del Gobbo et al. Am J Clin Nutr.

Abstract

Background: Accurate data on dietary habits are crucial for understanding impacts on disease and informing policy priorities. Nation-specific food balance sheets from the United Nations FAO provided the only available global dietary estimates but with uncertain validity.

Objectives: We investigated how FAO estimates compared with nationally representative, individual-based dietary surveys from the Global Dietary Database (GDD) and developed calibration equations to improve the validity of FAO data to estimate dietary intakes.

Design: FAO estimates were matched to GDD data for 113 countries across the following 9 major dietary metrics for 30 y of data (1980-2009): fruit, vegetables, beans and legumes, nuts and seeds, whole grains, red and processed meats, fish and seafood, milk, and total energy. Both absolute and percentage differences in FAO and GDD mean estimates were evaluated. Linear regression was used to evaluate whether FAO estimates predicted GDD dietary intakes and whether this prediction varied according to age, sex, region, and time. Calibration equations were developed to adjust FAO estimates to approximate national dietary surveys validated by using randomly split data sets.

Results: For most food groups, FAO estimates substantially overestimated individual-based dietary intakes by 74.5% (vegetables) and 270% (whole grains) while underestimating beans and legumes (-50%) and nuts and seeds (-29%) (P < 0.05 for each). In multivariate regressions, these overestimations and underestimations for each dietary factor further varied by age, sex, region, and time (P < 0.001 for each). Split-data set calibration models, which accounted for country-level covariates and other sources of heterogeneity, effectively adjusted FAO estimates to approximate estimates from national survey data (r = 0.47-0.80) with small SEs of prediction (generally 1-5 g/d).

Conclusions: For all food groups and total energy, FAO estimates substantially exceeded or underestimated individual-based national surveys of individual intakes with significant variation depending on age, sex, region, and time. Calibration models effectively adjusted the comprehensive, widely accessible FAO data to facilitate a more-accurate estimation of individual-level dietary intakes nationally and by age and sex.

Keywords: FAO; dietary intakes; food availability; food balance sheets; global diet; nutrition.

Figures

FIGURE 1
FIGURE 1
National GDD means compared with FAO means (95% CIs) (g/d), 1980–2009. FAO national food-supply estimates exceeded individual-based GDD national dietary intake estimates for most food groups, including fruit, vegetables, whole grains, red and processed meats, fish and seafood, milk, and total energy (P < 0.001 for each; most data points above 45 degree lines). FAO estimates significantly underestimated GDD intakes for beans and legumes (P < 0.001) and nuts and seeds (P < 0.05; most data points below 45 degree lines). Sample sizes of each food group are given in Table 2. GDD, Global Dietary Database.
FIGURE 2
FIGURE 2
National GDD means compared with mean differences (FAO-GDD) (g/d), 1980–2009. For food groups underestimated by FAO data (beans and legumes; nuts and seeds), the discrepancy between FAO and GDD means was greater at higher absolute intakes, whereas for total energy, the overestimation was greater at lower absolute intakes. Sample sizes of each food group are given in Table 2. GDD, Global Dietary Database.
FIGURE 3
FIGURE 3
Observed compared with predicted GDD means from split–data set calibration modeling. For each food group, the β (±SE) for each term included in the models for the age-sex data set is given in Supplemental Table 4. The MSE represents the average of squares of the difference between observed and predicted GDD means in the validation data set. Calibration models effectively adjusted FAO estimates to approximate estimates from national survey data with generally small MSEs. Filled navy blue circles denote North America, Western Europe, Australia, and New Zealand. Open red circles denote Southeast Asia and Asia Pacific. Orange triangles denote Central Asia and Eastern and Central Europe. Light blue squares denote Latin America, the Caribbean, and Oceania. Green diamonds denote Sub-Saharan Africa, North Africa, and the Middle East. GDD, Global Dietary Database; MSE, mean squared error.

Similar articles

See all similar articles

Cited by 21 articles

See all "Cited by" articles

References

    1. Sasaki S, Kesteloot H. Value of Food and Agricultural Organization data on food-balance sheets as a data source for dietary fat intake in epidemiologic studies. Am J Clin Nutr 1992;56:716–23. - PubMed
    1. Helsing E. Traditional diets and disease patterns of the Mediterranean, circa 1960. Am J Clin Nutr 1995;61(Suppl):1329S–37S. - PubMed
    1. Food and Agricultural Organization (FAO). FAOSTAT database, food balance sheets. Version current 2013 [cited 20 Oct 2013]. Available from: http://faostat.fao.org/site/354/default.aspx.
    1. Jacobs K, , Sumner D. The food balance sheets of the Food and Agriculture Organization: a review of potential ways to broaden the appropriate uses of the data. 2002. Review sponsored by the FAO [cited 20 Oct 2013]. Available from: http://182.160.112.236/agridrupal/sites/default/files/FBS_Review_of_Potential_Ways_to_Broaden_the_Uses_of_Data.pdf.
    1. Claro RM, Jaime PC, Lock K, Fisberg RM, Monteiro CA. Discrepancies among ecological, household, and individual data on fruits and vegetables consumption in Brazil. Cad Saude Publica 2010;26:2168–76. - PubMed

Publication types

Feedback