Food composition is the determination of what is in the foods we eat and is the critical bridge between nutrition, health promotion, disease prevention, and food production. Compilation of data into usable databases is essential to the development of dietary guidance for individuals and populations and plays an important role in food production, commerce, research and development, and policy setting. This article describes the history of food composition, important considerations in developing databases and tables, and the role of national and regional authorities, for example, the US Department of Agriculture and EuroFIR, and international organizations, for example, the Food and Agriculture Organization.

This manuscript describes progress by the Nutrient Data Laboratory (NDL) in monitoring sodium in processed and prepared foods frequently consumed by the U.S. population. Recent concerns by the U.S. public health community about the intake of sodium by Americans have led ARS, USDA to develop a plan to monitor the levels of sodium in highly consumed commercial packaged and restaurant foods. NDL scientists worked with USDA's Food Surveys Research Group (FSRG) to identify 125 Sentinel Foods, to serve as indicators for assessment of change in the sodium content in the food supply. For each food (e.g., cheese pizza) NDL used market share data to identify the predominant brands and types of foods (e.g., frozen cheese pizza, restaurant pizza) to be monitored for changes in the sodium level over time. Periodically, nutrient values for frequently consumed foods will be updated by chemical analysis or label checks. Estimates will be compared to existing values in the National Nutrient Database for Standard Reference (SR). Since 2010, about 140 foods have been sampled and analyzed by NDL contractors. NDL will continue to generate new sodium data which will be disseminated in the successive releases of the SR. Accurate and current data for sodium in processed foods will support the assessment of changes in sodium in foods as well as the assessment of sodium intake by the U.S. population in the years ahead.

In 2012, USDA prepared Release 3.1 of the “USDA Database for the Flavonoid Content of Selected Foods”, which contained data on 508 food items collected from 308 published sources. Flavonoids are secondary metabolites produced by plants in response to various environmental stresses such as climate and ultraviolet radiation. Other sources of variability include cultivar, growing location, agricultural practices, processing techniques and preparation methods, as well as analytical variability. The objective of this manuscript is to examine and report on variability in the flavonoid content of foods. While the required information needed to assess variability is not available for all foods, data for a number of foods was analyzed using analysis of variance for cultivar, location and other factors. For orange juice, data for 247 samples representing 109 mean values, i.e., different cultivars and location, were analyzed. The overall range for hesperetin was 1-39mg/100g; samples from the United States 5-30mg/100g; samples from Brazil 5-25mg/100g; and samples from Corsica (France) 12-26mg/100g. For strawberries, data for 148 samples representing 98 mean values were analyzed. The overall range for pelargonidin was 8-58mg/100g. The range of quercetin values in raw yellow onions is 0-91mg/100g, and represents 96 mean values for 402 samples from the U.S., Japan, Spain, and other countries. The flavonoid content of foods is extremely variable and is influenced by both location and cultivar, which account for 25 to 33 percent of the variability as well as by numerous other factors, which were not examined in this analysis.

The National Food and Nutrient Analysis Program (NFNAP) was implemented in 1997 to update and improve the quality of food composition data maintained by the United States Department of Agriculture (USDA). NFNAP was designed to sample and analyze frequently consumed foods in the U.S. food supply using statistically rigorous sampling plans, established sample handling procedures, and qualified analytical laboratories. Methods for careful handling of food samples from acquisition to analysis were developed to ensure the integrity of the samples and subsequent generation of accurate nutrient values. The infrastructure of NFNAP, under which over 1500 foods have been sampled, mandates tested sample handling protocols for a wide variety of foods. The majority of these foods were categorized into several major areas: 1) frozen foods; 2) fresh produce and/or highly perishable foods requiring refrigeration; 3) fast foods and prepared foods; 4) shelf-stable foods; 5) specialized study and non-retail (point of production) foods; and 6) foods from remote areas (e.g. American Indian reservations). This paper describes the sample handling approaches, from the collection and receipt of the food items to the preparation of the analytical samples, with emphasis on the strategies developed for those foods. It provides a foundation for developing sample handling protocols of foods to be analyzed under NFNAP and for other researchers working on similar projects.

Key features of USDA's Nutrient Databank System (NDBS) allow processing of food composition data from diverse sources, including USDA's National Food and Nutrient Analysis Program, the food industry, scientific literature, and food labels. The Nutrient Data Laboratory (NDL) designed the NDBS as a three-tiered ("Initial", "Aggregation", and "Compiled") data management system to facilitate handling of data. Raw data and documentation (data source, sample description, sample handling, and analytical methods) are migrated into the Initial module. NDL scientists compare new data with old values and decide how to combine the initial data into aggregated data. In the Aggregation module, data can be grouped and weighted by parameters such as study, source, and market share. Depending on the type of data, various statistical algorithms are used to generate statistics, such as mean, standard error, number of data points, and error bounds. In the Compiled module, food names are finalized and common measures selected. Nutrient profiles are developed and missing nutrients/food components are imputed according to standardized scientific principles. A formulation application employing linear programming techniques, estimate, formulations for commercial foods and nutrient profiles based on the nutrient content of ingredients and target values derived from label information. A recipe application calculates nutrient profiles based on ingredients and their known proportions, allowing for the application of food yield and nutrient retention factors. The NDBS automatically documents how each value was derived and incorporates quality control checks at all levels. Prior to release, the completed nutrient profiles are reviewed by NDL scientists and, if approved, disseminated. The NDBS brings together a number of stand-alone modules and applications into one integrated system allowing the management of ∼7500 food items for up to 140 nutrients/food components. Data points and documentation are managed and maintained in one place, providing an "audit trail" for each data point. The NDBS contains algorithms to assign confidence codes using NDL's data quality evaluation system. The NDBS permits the annual release of reliable data for a comprehensive set of nutrients/food components for a wide variety of foods on NDL's Web site: http://www.ars.usda.gov/nutrientdata. Through these releases, NDL provides food composition data for researchers, diet and health professionals, and consumers, including the "What We Eat in America" component of the National Health and Nutrition Examination Survey (NHANES).

The role of fluoride in reducing the risk of dental caries, especially among children, is well recognized and is the basis for current intake recommendations. The US Department of Agriculture, Nutrient Data Laboratory conducted a comprehensive study of the fluoride content of US drinking water, as part of the US National Fluoride Database and Intake Assessment Study, a collaborative effort with the University of Minnesota, the University of Iowa College of Dentistry, and Virginia Polytechnic Institute and State University. The sampling method involved: serpentine ordering of the US population by census region, division, and county; dividing the population into 72 equal population size zones; and randomly selecting one county per zone and two residences per county. Participants were recruited by phone to provide two tap water samples, 3–4 months apart; samples (n=288) were analyzed by the direct read method. Well water averaged <20mcgfluoride/100g, municipal water averaged 100–110mcgfluoride/100g, and the national average across sources was 71mcgfluoride/100g. These nationally representative data for drinking water will support public health research on the impact of fluoride on bones and teeth and will provide a foundation for assessment tools in the dental and medical communities.

Data in the food composition analytical literature were reviewed. They were then aggregated and assembled into a provisional database for the major flavonoids in brewed tea. As levels of fermentation increased from green to oolong to black tea, the major flavan-3-ol profiles changed. Total catechins were 13.6 g/100 g in green and 4.2 g/100 g dry weight in black tea. A discussion of methods to calculate the flavonoid content in tea is presented.

Assessment of vitamin K (VK) dietary intakes has been limited by incomplete VK food composition data for the US food supply. The phylloquinone (VK-1 or vitamin K1) concentrations of a variety of geographically representative vegetables (n=218) were determined by reversed-phase high performance liquid chromatography with fluorescent detection. Green leafy and flower vegetables including broccoli, broccoli raab, spinach, and certain lettuces, contained >100μg phylloquinone/100g vegetable. In contrast, raw tubers and roots contained <10μg phylloquinone/100g vegetable. Iceberg lettuce, a primary dietary source of phylloquinone, contained 24.1μg phylloquinone/100g vegetable, which is less than previously listed in nutrient databases. Potential factors affecting phylloquinone concentrations include processing and varietal type of leafy vegetables.

The USDA Special Interest Database for flavonoid content of selected foods contains 26 most abundant compounds within 5 predominant subclasses of flavonoids–flavonols, flavones, flavanones, flavan-3-ols, and anthocyanidins. All the data were evaluated for 5 quality evaluation categories (sampling plan, sample handling, analytical method, analytical quality control and number of samples), using the data quality evaluation system developed by the USDA scientists. Confidence Codes (A–through D) were then assigned to every value. The database contains acceptable values for 225 selected foods. Only 97 sources out of approximately 475 collected included acceptable analytical data. The overall quality of data was good with 64% of the observations receiving A or B confidence codes; the flavan-3-ols subclass received better ratings than other subclasses. While this is the first comprehensive database for flavonoids in foods, the majority of data came from Europe and countries other than the US. Due to the observed variability in the values it will be important to have data for US foods. The evaluation of data quality helps set priorities and further identifies the foods to be analyzed as well as areas to improve data quality. Furthermore, release of data quality confidence codes with data provides necessary information to investigators to assess the impact of flavonoid intake on risk of various chronic diseases.

The purpose of this study was to determine the contents of three forms of vitamin K [phylloquinone, dihydrophylloquinone, and menaquinone-4 (MK-4)] in representative samples (including different samples within the same food category) of meat (n = 128), dairy and eggs (n = 24), and fast foods (n = 169) common to the U.S. diet. The findings of our analysis indicate that no single food item in these categories is a rich dietary source of any one form of vitamin K. However, these foods are often consumed in large quantities; hence, they may be of importance in overall contribution to total vitamin K intake. The presence of MK-4 in meat, eggs, and dairy foods could be important as physiologic functions unique to MK-4 are identified.

Analysis of total folate concentration measured by microbiological assay in a variety of foods submitted in a routine manner to experienced laboratories that regularly perform folate analysis on fee-for-service basis was evaluated. Homogenates of fresh strawberries, frozen spinach, orange juice, frozen meat and vegetable pizza, dry macaroni, and dried pinto beans were prepared and stored under conditions previously determined to maintain stability of folate content. An aliquot of each composite and of 3 certified reference materials were sent on each of 4 occasions to 4 laboratories. Results for macaroni and pizza, the only folic acid-fortified foods, had considerably lower between-laboratory variation (CV(B)) with CV(B) of 9-11% versus >45% for other foods. Mean total folate ranged from 14 to 279 microg/100 g for a mixed vegetable reference material, from 5 to 70 microg/100 g for strawberries, and from 28 to 81 microg/100 g for wholemeal flour. Only 1 laboratory reported using a tri-enzyme extraction, and all laboratories used folic acid fortified foods as internal control materials. Users of commercial total folate analysis should understand the uncertainty in values determined by microbiological assay, particularly for foods containing primarily naturally occurring folate, which may not be apparent when replicate samples are not submitted for analysis.

The stability of 5-methyltetrahydrofolate (5MTHF) in homogenized fresh fruits and vegetables representing samples for the USDA National Food and Nutrient Analysis Program was evaluated. Samples were homogenized in liquid nitrogen and 5MTHF was measured after 0, 2, 7, 30 days and then at 3-month intervals for a total of 12 months storage at −60 ± 5 °C, utilizing extraction by a tri-enzyme treatment, purification by strong anion-exchange solid-phase extraction, and quantification by reverse-phase HPLC. Method validation included analysis of a reference material and interlaboratory analysis of selected samples by HPLC and LC-MS. A canned spinach composite was assayed in each analytical batch to monitor inter-assay precision. No change in 5MTHF content was detected in any of the samples after 12 months. Concentrations ranged from <10 μg/100 g in bananas to >100μg/100 g in spinach. Relative standard deviations were generally <7% within assay and <11% between assays.

Proanthocyanidins (PAs) have been shown to have potential health benefits. However, no data exist concerning their dietary intake. Therefore, PAs in common and infant foods from the U.S. were analyzed. On the bases of our data and those from the USDA's Continuing Survey of Food Intakes by Individuals (CSFII) of 1994-1996, the mean daily intake of PAs in the U.S. population (>2 y old) was estimated to be 57.7 mg/person. Monomers, dimers, trimers, and those above trimers contribute 7.1, 11.2, 7.8, and 73.9% of total PAs, respectively. The major sources of PAs in the American diet are apples (32.0%), followed by chocolate (17.9%) and grapes (17.8%). The 2- to 5-y-old age group (68.2 mg/person) and men >60 y old (70.8 mg/person) consume more PAs daily than other groups because they consume more fruit. The daily intake of PAs for 4- to 6-mo-old and 6- to 10-mo-old infants was estimated to be 1.3 mg and 26.9 mg, respectively, based on the recommendations of the American Academy of Pediatrics. This study supports the concept that PAs account for a major fraction of the total flavonoids ingested in Western diets.

Both lipophilic and hydrophilic antioxidant capacities were determined using the oxygen radical absorbance capacity (ORAC(FL)) assay with fluorescein as the fluorescent probe and 2,2'-azobis(2-amidinopropane) dihydrochloride as a peroxyl radical generator on over 100 different kinds of foods, including fruits, vegetables, nuts, dried fruits, spices, cereals, infant, and other foods. Most of the foods were collected from four different regions and during two different seasons in U.S. markets. Total phenolics of each sample were also measured using the Folin-Ciocalteu reagent. Hydrophilic ORAC(FL) values (H-ORAC(FL)) ranged from 0.87 to 2641 micromol of Trolox equivalents (TE)/g among all of the foods, whereas lipophilic ORAC(FL) values (L-ORAC(FL)) ranged from 0.07 to 1611 micromol of TE/g. Generally, L-ORAC(FL) values were <10% of the H-ORAC(FL) values except for a very few samples. Total antioxidant capacity was calculated by combining L-ORAC(FL) and H-ORAC(FL). Differences of ORAC(FL) values in fruits and vegetables from different seasons and regions were relatively large for some foods but could not be analyzed in detail because of the sampling scheme. Two different processing methods, cooking and peeling, were used on selected foods to evaluate the impact of processing on ORAC(FL). The data demonstrated that processing can have significant effects on ORAC(FL). Considering all of the foods analyzed, the relationship between TP and H-ORAC(FL) showed a very weak correlation. Total hydrophilic and lipophilic antioxidant capacity intakes were calculated to be 5558 and 166 micromol of TE/day, respectively, on the basis of data from the USDA Continuing Survey of Food Intakes by Individuals (1994-1996).

For the first time, a database of the antioxidant capacities of both the lipophilic and hydrophilic components of foods has been developed using the modified oxygen radical absorbance capacity (ORACFL) assay and a peroxyl radical generator. For lipophilic components, randomly methylated β-cyclodextrin was used as a solubility enhancer. Four representative samples were extracted directly with the hydrophilic solvent (acetone:water:acetic acid, 70:29.5:0.5). Their ORACFL values were similar to that obtained for hydrophilic ORACFL (H-ORACFL) following lipophilic extraction with hexane:dichloromethane (1:1). Lipophilic ORAC values (L-ORACFL) were relatively low compared to H-ORACFL, ranging from 0.11±0.06 to 154.70±3.58μmol TE/g of fresh or dry weight, whereas H-ORACFL ranged from 1.23±0.17 to 175.24±10.36μmol TE/g of fresh or dry weight. Total antioxidant capacity (TAC) was calculated as the sum of the lipophlic and hydrophilic ORACFL values. L-ORACFL as a percentage of TAC ranged from 0.27% to 63.70%. Sampling time during the year significantly influenced lipophilic and/or hydrophilic ORACFL values in some food samples. In order to get an accurate total antioxidant capacity of a given food sample, both lipophilic and hydrophilic fractions need to be measured. Food processing, such as cooking or peeling, need to be considered as additional factors which can introduce variation in antioxidant capacity measurements of foods.