Frequently Asked Questions

General:

1. What is the role of U.S. EPA's Office of Pesticide Programs (OPP) in food safety and why did it develop the Food Commodity Intake Database (FCID)?

2. How is national food consumption surveyed in the U.S.?

Food Commodity Intake Database (FCID):

3. Why was U.S. EPA's FCID developed?

4. How was FCID developed and what information is included in the FCID database?

5. I've heard the acronyms CSFII, NHANES, WWEIA, FCID, and FNDDS? What are they, how are they interconnected, and how was each used (or not) in creating the data on the JIFSAN website?

6. If FCID just combines the WWEIA consumption estimates from NHANES and the FNDDS food composition data from USDA, why can't I just do that myself? What additional value added and information does the FCID provide?

7. I hear continuing reference to the FCID "recipe files" or "100 gram files". What are they, how are they used, and where can I get them?

8. What limitations does FCID have when assessing dietary exposure and commodity consumption?

JIFSAN WWEIA FCID Resources:

9. How is FCID used by EPA? What capabilities are available on the JIFSAN foodrisk.org website?

10. How do I use the JIFSAN recipe search tool and commodity consumption calculator?

11. How do we know that the results reported in the JIFSAN calculator are correct? What kind of review was done on the results?

12. I'd like to look at regional and seasonal consumption data (e.g., apple, fresh with peel in the Northeast during the autumn). I used to be able to do that with CSFII, but I don't see that option with the JIFSAN calculator (although age, sex, and race/ethnicity are available). How can I drill down into the data and get this information?

13. I am interested in using the JIFSAN data to explore nutrition issues and develop nutritional intake estimates for the U.S. population based on the consumption data provided there. How can I do this?

14. The JIFSAN consumption calculator only provides fixed percentiles at intervals of 5 percentile points. I'd like to obtain consumption estimates at additional percentiles than those presented in the JIFSAN consumption calculator. In addition, I'd like a greater number of decimal points in the answer and I'd like to estimate the standard error of the mean. How do I do this?

15. I've been told by my statistician that I need to have (and use) the sampling weights, the PSUs (Primary Sampling Units), and the strata to properly use NHANES/WWEIA survey data. Does the JIFSAN calculator use these? If not, why does my statistician say these are necessary for proper use and interpretation of this data?

Additional Information:

16.Where can I go for additional information on EPA dietary exposure assessment resources?

17. I looked at the WWEIA-FCID 2003-2008 files and do not know what the field names and codes in the files represent. Where can I get additional information on the list of variables and data fields available in WWEIA-FCID 2003-2008?

General

1. What is the role of U.S. EPA's Office of Pesticide Programs (OPP) in food safety and why did it develop the Food Commodity Intake Database (FCID)?

EPA is responsible for the registration of pesticides under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and Food Quality Protection Act of 1996. Under FIFRA and FQPA, EPA evaluates the potential health effects of pesticides using a scientific process called risk assessment, which requires EPA to assess the potential human health risks of pesticide exposure through food and drinking water consumption. This requires EPA to perform dietary risk assessments that incorporate information on U.S. food consumption and monitoring data on pesticide residues in and on foods. As such, data on U.S. food consumption plays a critical role in EPA's risk assessment of pesticides. The Food Commodity Intake Database (FCID) was developed to permit available food consumption data to be used in OPP's dietary exposure evaluation for pesticides. For further information on EPA/OPP's risk assessment process, please visit: www.epa.gov/pesticides/about/overview_risk_assess.htm. A more complete and technically-focused description of the human heath risk assessment process used by OPP – including how dietary consumption data is used to evaluate exposure and risk and probabilistic approaches used in developing these estimates – is available in a downloadable pamphlet from Purdue University entitled Pesticides and Human Health Risk Assessment: Policies, Processes, and Procedures.

2. How is national food consumption surveyed in the U.S.?

For purposes of dietary risk assessment, OPP developed its current exposure assessment methodology using USDA's Continuing Survey of Food Intakes by Individuals 1994-96, 1998 (CSFII 1994-96/1998). CSFII 1994-96/1998 was statistically designed to be a nationally-representative of U.S. food consumption and was the 10th nationwide USDA food consumption survey (USDA, 1997). The 1998 survey period of CSFII focused on children ages 0 through 9 and was a supplement to the CSFII 1994-96. It used the same sample design as the CSFII 1994-96 and was intended to be merged with CSFII 1994-96 to increase the sample size for children.

Following CSFII 1994-96/1998, USDA partnered with the Department of Health and Human Services (DHHS) and integrated CSFII with DHHS' National Health and Nutrition Examination Survey (NHANES) in 2002. Under this partnership, USDA's food consumption survey was renamed "What We Eat In America" (WWEIA) and is conducted on a two-year cycle on a continuous basis as part of NHANES. While data is collected through NHANES, USDA maintains responsibility for development of dietary data collection methods, maintenance of food and nutrient databases, and data review and processing. For further information on WWEIA survey design and data collection methods, please visit: www.ars.usda.gov/Services/docs.htm?docid=13793.

Both CSFII and WWEIA use similar data collection methods and generally capture similar information on U.S. food consumption. Specifically, both CSFII and WWEIA collect two non-consecutive days of dietary recall data using a USDA dietary data collection method called the Automated Multiple-Pass Method. While both surveys collect similar data and have a similar data structure, there are some key differences between the surveys. One key difference for OPP dietary exposure assessment purposes is that CSFII provides data related to the region and season. This regional and seasonal data is no longer publically available through NHANES in order to protect the confidentiality of respondents who also undergo a comprehensive medical examination and provide detailed medical history and health information through a series of interviews. Another importance difference between the surveys is that the 1998 survey period of CSFII specifically oversampled young children at the request of EPA as part of its Supplemental Children's Survey (SCS). While children have not been oversampled through NHANES, multiple survey cycles can be combined to increase the sample size of infant and child age groups.

Food Commodity Intake Database (FCID)

3. Why was FCID developed?

When conducting dietary risk assessments, OPP is most interested in consumption of food commodities in the form of ingredients such as beef, wheat flour, tomato sauce, soybean oil, etc. rather than foods "as eaten" (e.g., lasagna). This is because substantially more pesticide residue concentration data – from EPA-required field trials and/or from the USDA's Pesticide Data Program - is available on pesticide residues on food commodities (aka ingredients) than on foods per se. While the CSFII and WWEIA surveys provide extensive, statistically representative information on food consumption, information on food commodity consumption (i.e., foods expressed in terms of ingredients) is not present in CSFII or WWEIA. As a result, the FCID was developed for use by EPA and other organizations when conducting exposure assessments which are most appropriately done on an "ingredient" or "food commodity" basis. WWEIA-FCID 2003-08 is intended to complement the CSFII and NHANES/WWEIA databases in that it provides estimates of food consumption as commodities as opposed to foods "as eaten" which can in some exposure and other situations be of more utility.

4. How was FCID developed and what information is included in the FCID database?

FCID was originally developed using USDA's CSFII (1994-96/1998) and now contains intake data and related information on over 500 commodities listed in the Environmental Protection Agency's (EPA) Food Commodity Vocabulary. The FCID database is the result of cooperative efforts by USDA/ARS and EPA/OPP and translates food consumption as reported eaten in WWEIA (1999-08 survey cycles) and CSFII (1994-96/1998) surveys into consumption of U.S. EPA-defined food commodities. The FCID consists of a number of component database parts which can be found on the JIFSAN website. Descriptions of each database component can be found under the "Database Contents" tab, but the major data tables making up the FCID are listed below:

  • FCID Recipe Database: used to translate WWEIA food consumption to consumption of agricultural food commodities.
  • Commodity Specific Food Consumption: contains commodity-specific consumption by WWEIA respondent sequence number (SEQN) and day code (DAYCODE).
  • Commodity-Specific Food Consumption – Detailed: contains commodity specific consumption by WWEIA respondent sequence number (SEQN), day code (DAYCODE), food form (FF), cooking method (CM), and cooked status (CS).
  • WWEIA Demographic Data: Select demographic data and bodyweight data from WWEIA, including statistical survey weights, PSUs, and strata.

5. I've heard the acronyms CSFII, NHANES, WWEIA, FCID, and FNDDS? What are they, how are they interconnected, and how was each used (or not) in creating the data on the JIFSAN website?

  • Continuing Survey of Food Intake by Individuals (CSFII) – USDA survey designed to measure the types and amounts of foods consumed from a nationally representative sample of children and adults in the United States. EPA/OPP's dietary exposure assessment methods, including the development of FCID, were originally based on CSFII (1994-96/1998).
  • National Health and Nutrition Examination Survey (NHANES) – A series of studies conducted by DHHS that are designed to assess the health and nutritional status of a nationally representative sample of children and adults in the United States. Beginning in 1999, the CSFII was integrated into the NHANES survey and was renamed the "What We Eat In America" component.
  • What We Eat In America (WWEIA) – WWEIA represents the integration of CSFII with NHANES. NHANES is responsible for data collection and USDA is responsible development of dietary data collection methods, maintenance of food and nutrient databases, and data review and processing. EPA/OPP's dietary exposure assessment methods have been updated to reflect dietary consumption reported in WWEIA 2003-08.
  • Food Commodity Intake Database (FCID) – EPA database developed in collaboration with USDA. The recipes within the FCID translates food consumption as reported eaten in WWEIA (1999-08 survey cycles) and CSFII (1994-96, 1998) surveys into consumption of U.S. EPA-defined food commodities
  • Food and Nutrient Database for Dietary Studies (FNDDS) – A USDA database of foods, their nutrient values, and weights for typical food portions. FNDDS can be used to analyze dietary consumption data from WWEIA and also incorporates food composition data from USDA's National Nutrient Database for Standard Reference. When updating FCID for analysis of NHANES/WWEIA, FNDDS was utilized to identify the components of new foods reported consumed by WWEIA survey respondents. The FNDDS is often used to estimate nutritional aspects of the American diet, and for some foods does not always need to break foods reported "as eaten" down to a US EPA food commodity level. For example, FNDDS recipes may include pasta and cheese as a part of a recipe for lasagna which is a sufficient breakdown of ingredients for nutritional assessment purposes, but not for dietary dietary exposure purposes when pesticide residue information is not available for pasta and cheese but instead is available for wheat flour and milk. Thus, FCID recipes further convert these pasta and cheese to the US EPA food commodities of wheat flour and various components of milk which are US EPA food commodities and thus can be used to assess dietary exposure.

6. If FCID just combines the WWEIA consumption estimates from NHANES and the FNDDS food composition data from USDA, why can't I just do that myself? What additional value-added and information does the FCID provide?

FNDDS and FCID provide similar information on the composition of WWEIA foods. In many cases, however, FNDDS does not sufficiently break food components down into the individual commodities that are needed by EPA/OPP when conducting dietary risk assessments. For example, FNDDS lists wheat tortillas as a food component, whereas EPA needs commodity-level information on the components of tortillas (e.g., wheat flour, soybean oil, cottonseed oil, etc.). In addition, FNDDS does not provide ancillary information such as cooked status, cooking methods, and food form which can be useful for dietary exposure assessments of pesticide residues and food contaminants since residues may be affected by cooking. As such, FCID complements FNDDS and provides more specific information on food commodities defined in EPA's Food Commodity Vocabulary.

7. I hear continuing reference to the FCID "recipe files" or "100 gram files". What are they, how are they used, and where can I get them?

The FCID recipes are what are used to convert food consumption reported in CSFII (1994-96, 1998) or WWEIA (2003-08) into intake of food commodities defined in EPA's Food Commodity Vocabulary. The FCID recipes are available for download on the JIFSAN website in .csv format (fcid.foodrisk.org/dbc/) and are both fully searchable using a JIFSAN-developed web application (fcid.foodrisk.org/recipes/) and fully printable after conversion of each recipe to a .csv spread-sheet-like file using the web button labeled "Convert Results to CSV" on the upper right hand side of the webpage. Here is a screenshot of the recipe page showing the beginnings of a search for recipes for meatless lasagna, with the user subsequently selecting one of the three meatless lasagna choices and clicking the "Generate Recipe Report" to obtain the relevant recipe/100 gram file:

The JIFSAN recipe database can also be "reverse-searched" by commodity using the "Commodities" tab instead of the "Recipes" tab such that all CSFII or WWEIA foods containing a given EPA commodity can be listed. For this latter search, clicking on the "Generate Recipe Report" at the bottom of the page will generate a list of all the foods that contain the user-selected commodity (ingredient).

FCID recipes are expressed as grams of commodity per 100 grams of the food reported consumed in CSFII or WWEIA (i.e., equivalent to percent). They are used by EPA to convert all foods reported "as eaten" into their agricultural commodity equivalents. For example, what was reported by a WWEIA survey respondent as a 1/8 slice of a 12" pepperoni pizza (e.g., PIZZA W/ PEPPERONI, NS AS TO TYPE OF CRUST, WWEIA Food Code: 58106540) would be converted in FCID to gram amounts of wheat flour, beef and pork, various components of milk (reflecting the cheese), tomato puree, soybean oil, etc. for that respondent. All of the demographic information associated with that respondent as collected by CSFII or NHANES such as socio-economic status, age, race/ethnicity, etc. and other information such as body weight and other anthropometric measurements is also retained with that record. FCID also contains additional information with respect to the cooked status (yes/no), cooking method (baked, broiled, fried, etc.) and food form (fresh, frozen, canned, pickled, etc.) of each of the ingredients, information that is not available in CSFII or WWEIA.

8. What limitations does FCID have when assessing dietary exposure and commodity consumption?

FCID was developed primarily to assist OPP in the estimation of food commodity intake for use in dietary exposure and risk estimates for pesticide residues in food. When using FCID to estimate food commodity consumption, it is important to consider a variety of limitations and caveats that may exist. For example:

  • Many of the recipes in the current version of FCID were originally developed for and correspond to the USDA's CSFII 1994-96, 1998 which was developed jointly by USDA and EPA. These recipes have not been updated in the current FCID to account for what may be changes in commercial food products over the intervening period. For example, it may be that current commercial frozen lasagna may have changed (e.g., different ingredients and/or different amounts of ingredients) and is now different from the original recipes were developed for the 1994-96, 1998 CSFII. While USDA updates their FNDDS database for each cycle of the NHANES/WWEIA survey, OPP did NOT update the recipes present in FCID to reflect these changes. Thus, many of the ingredients and amounts listed in FCID for commercial food products are based on older commercial formulations and may not reflect the most current information regarding the nature and amounts of ingredients that these products contain.
  • For new or modified foods reported in WWEIA that were not present in CSFII, 1994-96, 1998, OPP did develop new recipes, with ingredient amounts in recipes for commercial products estimated based on available older recipes , professional judgment, and –where necessary - mathematical algorithms. For example, if lasagna with chicken or turkey were reported in WWEIA but not reported in the older CSFII survey, an older CSFII recipe for lasagna with beef would be modified by changing the grams of beef to grams of chicken and turkey. In all cases, mass-balance was maintained. Ingredients and amounts for other food mixtures were based on representative recipes, usually from popular cookbooks but also from regional or specialty cookbooks when necessary. The recipes were generic and did not reflect and were not specific for each sample person.
  • When developing recipes, ingredients contained in food items in the CSFII/WWEIA surveys were matched to food commodities defined in EPA's Food Commodity Vocabulary. While most ingredients were matched to commodities, EPA's Food Commodity Vocabulary focuses on agricultural commodities that can be treated with pesticides and does not include ingredients that tend to be highly processed or produced in a controlled manufacturing environment. These types of ingredients include salt, yeast, artificial sweeteners, and food additives.
  • For some food commodities, there are very few survey respondents who report consuming that commodity. In those instances, the uncertainty around estimated consumption can be quite large. The database user should keep in mind that statistical estimates based on a small number of survey respondents may be less statistically reliable than estimates based on larger numbers of respondents. Additional information regarding sample size and associated statistical caveats and limitations can be found under the button labeled "Caveats regarding the Use of the Percentile Calculator" located in the upper right hand corner under the "Commodity Weight Percentiles" screen.
  • Specific oil ingredients in food reported consumed in CSFII and WWEIA could not be determined when the oil type was not specified. As a result, assumptions were made about the proportion of different types of oil, based on market share information from USDA's Economic Research Service and Foreign Agricultural Service. Similarly, estimates for sugar were derived from market share information on sugar from sugarcane and sugar beets.

Given the above considerations and limitations, FCID is perhaps best suited for (and accurate in) estimation of consumption of fruits and vegetables that are consumed as fresh, frozen, or canned forms as well as of meats, poultry, seafood, and dairy products. FCID is probably less suited for estimating consumption of ingredients in highly processed foods or foods that are consumed mainly as ingredients in commercially-prepared foods for which recipes may have changed or for which ingredient amounts are estimated. For purposes of pesticide exposure and risk assessments, this limitation is not likely to be of any practical or significant importance since most highly-processed /commercially-prepared foods are expected to contain only minimal pesticide residues, particularly when compared to their fresh fruit and vegetable counterparts. If primary interest resides in estimating nutrient intakes, intakes of food ingredients that are not expressed on an EPA food commodity basis (e.g., pasta instead of wheat flour), or for intake of fats and oils, other data sources such as USDA's Food Nutrient Database for Dietary Studies (FNDDS), may be more appropriate.

JIFSAN WWEIA FCID Resources

9. How is FCID used by EPA? What capabilities are available on the JIFSAN foodrisk.org website?

The FCID database and consumption files have been incorporated into EPA/OPP's primary dietary exposure model, Dietary Exposure Evaluation Model (DEEM), and have also been used to derive the fruit, vegetable, meat/poultry, dairy, grain, and seafood consumption estimates which are reported in EPA's 2011 Exposure Factors Handbook. Thus, consumption estimates developed using the raw FCID data available in .csv format on the JIFSAN website, using the JIFSAN consumption calculator, or using the DEEM dietary exposure model will be equivalent (to numerical round-off error).

The WWEIA-FCID 2003-08 data that is available on the JIFSAN foodrisk.org website and the web-based recipe and consumption tools that are present there provide the user with the following capabilities and features:

  • A point-and-click user interface that makes the underlying WWEIA data more accessible, as well as raw data files in .csv format which can be imported into a variety of database and statistical software programs for further analysis and more detailed reporting.
  • The ability to determine which food commodities (ingredients) are present in a given food or, the reverse, which foods contain a given food commodity, and their respective amounts or proportions. In addition, the cooked status (yes/no), cooking method (baked, boiled, fried, etc.) and food form (fresh, frozen, dried, canned, etc.) are also provided.
  • The ability to "match" U.S. EPA FCID (crop group) codes for agricultural commodities and their crop group codes to their associated Codex codes and descriptions.
  • The ability to generate estimates of mean and various percentile consumption values of a given food commodity for the total population and various user-defined age and racial/ethnic subgroups. This can be done on both a per capita and "eaters only" basis and can be output in both an absolute gram consumption and gram consumption/kg b.w. basis.

10. How do I use the JIFSAN recipe search tool and consumption calculator?

JIFSAN has developed a recipe search tool and consumption calculator that can be used to search and analyze WWEIA-FCID 2003-08 on the web. The recipe search tool enables users to search recipes by CSFII/WWEIA food description or 8-digit CSFII/WWEIA Food Code; by the FCID food commodity description or 10-digit FCID food commodity code; and by the EPA-defined crop groups. An example of the use of this tool for recipe determination purposes is provided in the FAQ "I hear continuing reference to the FCID "recipe files" or "100 gram files". What are they, how are they used, and where can I get them?"

The consumption calculator enables customized analysis of WWEIA-FCID 2003-08, based on FCID commodity and demographic information captured in WWEIA, including age, gender, and race/ethnicity. An example analysis of the JIFSAN consumption calculator is shown below. As shown, a user may search FCID commodity based on either description (or code) or crop group and then select specific commodities that are of interest (e.g., "Apple, fruit with peel"). After selecting commodities of interest, additional age, gender, and race/ethnicity filters can be applied to customize the demographic population that is the focus of the analysis.

After selecting the criteria, the user can click the "Generate Percentiles" button at the bottom of the screen to generated weighted summary statistics that include the number of respondents, mean consumption, and a range of percentiles. The summary statistics are reported on both a commodity mass basis (i.e., total grams of commodity consumed) and commodity mass per body mass basis (i.e., total grams of commodity consumed divided by respondent bodyweight). Summary statistics are further stratified and provided for "Commodity Eaters Only" (i.e., only respondents who reported commodity consumption on that day) and per capita consumption for the "Total Filtered Population" (i.e., both commodity eaters and non-commodity eaters that meet the demographic filter criteria). Example output from the search above is provided here:

Note that statistical estimates based on a small number of survey respondents may be less statistically reliable than estimates based on larger numbers of respondents. Further information on sample size considerations can be accessed by clicking the "Caveats re: Use of the Percentile Calculator" button at the top right corner of the consumption calculator page.

11. How do we know that the results reported in the JIFSAN calculator are correct? What kind of review was done on the results?

The JIFSAN consumption calculator has been developed in close collaboration with EPA/OPP. EPA/OPP has done extensive testing of the calculator and closely reviewed the results to ensure that it provides accurate intake estimates. As part of this review process, EPA/OPP made comparisons between the JIFSAN calculator and EPA/OPP's Dietary Exposure Evaluation Model (DEEM). These comparisons confirm that the JIFSAN calculator provides accurate summary statistics that can be customized to evaluate intake in different population groups based on age, gender, and race/ethnicity. An example of QA/QC spreadsheet that was prepared to evaluate and compare the consumption estimates for 30 commodities among 0-9 years olds is presented here.

12. I'd like to look at regional and seasonal consumption data (e.g., apple, fresh with peel in the Northeast during the autumn). I used to be able to do that with CSFII, but I don't see that option with the JIFSAN calculator (although age, sex, and race/ethnicity are available). How can drill down into the data and get this information?

As described above, regional and seasonal information is no longer available in the WWEIA consumption data that is made publically available by the CDC's National Center for Health Statistics (NCHS). As such, it is not possible to use WWEIA-FCID 2003-08 or the JIFSAN calculator to evaluate regional or seasonal differences in consumption.

13. I am interested in using the JIFSAN data to explore nutrition issues and develop nutritional intake estimates for the U.S. population based on the consumption data provided there. How can I do this?

FNDDS is a more appropriate data source for evaluating nutritional intake in the U.S. population and provides comprehensive nutritional information on the foods reported consumed in WWEIA. For further information on FNDDS, please visit: www.ars.usda.gov/Services/docs.htm?docid=12072#SR.

14. The JIFSAN consumption calculator only provided fixed percentiles at intervals of 5 percentile points. I'd like to obtain consumption estimates at additional percentiles than those presented in the JIFSAN consumption calculator. In addition, I'd like a greater number of decimal points in the answer, and I'd like to estimate the standard error of the mean. How do I do this?

The JIFSAN consumption calculator provides standardized statistical output including means and selected percentile estimates. Any additional analysis, such as calculating additional percentiles and standard errors are best performed using dedicated statistical software such as SAS, R, Stata, or another statistical software program that is capable of properly handling complex, multi-stage survey designs If you wish to conduct any additional analysis using these or similar softwares, the WWEIA-FCID 2003-08 raw data in .csv format can be downloaded from the JIFSAN website (fcid.foodrisk.org/dbc/).

15. I've been told by my statistician that I need to have (and use) the sampling weights, the PSUs (Primary Sampling Units), and the strata to properly use NHANES/WWEIA survey data. Does the JIFSAN calculator use these? If not, why does my statistician say these are necessary for proper use and interpretation of this data?

The NHANES sampling weights, PSUs, and strata can be important to consider when performing analysis of WWEIA-FCID 2003-08 data. The JIFSAN calculator provides estimates of the mean and a range of percentiles. Estimation of these summary statistics (i.e., mean and percentiles) requires appropriate use of the NHANES survey weights in the analysis (which the JIFSAN calculator does), but the PSUs and strata are NOT required. While the PSUs and strata are not needed for the summary statistics reported by the JIFSAN calculator, they MUST be used when calculating standard error or for making statistical comparisons between population subgroups. Thus, the JIFSAN calculator makes use of the necessary survey design parameters for the statistics that it does provide.

Additional Information

16. Where can I obtain additional information on EPA dietary exposure assessment resources?

17. I looked at the WWEIA-FCID 2003-2008 files and do not know what the field names and codes in the files represent. Where can I get additional information on the list of variables and data fields available in WWEIA-FCID 2003-2008?

Data users who are interested in additional information on the content of WWEIA-FCID 2003-2008 are encouraged to the WWEIA-FCID 2003-2008 Background Codebook. The codebook provides an introduction to and listing of the variables and fields available in WWEIA-FCID 2003-2008 and is intended to assist the new data user in understanding the structure and function of the available files. Much of the information in the Background Codebook is derived or originated from the NHANES/WWEIA survey. As such, data users are encouraged to read and be familiar with the background, statistical, and survey information provided by both NCHS (including their website tutorials) and the USDA with respect to this data.