Lois C. Steinfeldt

1
Strategies for Managing Complex Coding
Situations Food Coding in a National Survey
Lois C. Steinfeldt Linda A. Ingwersen Randy P.
LaComb Food Surveys Research Group Beltsville
Human Nutrition Research Center Agricultural
Research Service U. S. Department of Agriculture
Association for Survey Computing Third
International Conference September 23, 1999
2
Continuing Survey of Food Intakes by Individuals
10th nationwide food survey
16,000 individuals
2 days
24-hour dietary recall
76 for 2 days
80 for Day 1
3
1994-96 What We Eat in America
16,000 Individuals
31,000 Days
440,000 foods reported
4
1994-96 What We Eat in America
Food and nutrient intakes for individuals of all
ages in a nationally representative sample
5
1994-96 What We Eat in America
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday

• Food and nutrient intakes representing all
  seasons and across all days of the week

6
Uses of the CSFII Data

• Monitor the nutritional adequacy of American
  diets
• Measure the impact of food fortification on
  nutrient intakes
• Develop dietary guidance and related programs
• Estimate exposure of population groups to
  food contaminants

7
Uses of the CSFII Data

• Evaluate the nutritional impact of food
  assistance programs
• Evaluate the impact of food labeling on
  nutrient intakes
• Assess the demand for agricultural products

8
Food Intake Survey Processing
Descriptions of foods and amounts eaten as
reported
1 cup of whole milk
Food Coding, Recipe and Nutrient Databases
11111000 Milk, cows, fluid, whole
8.15 gm of fat 291.34 mg of calcium
Survey results
9
Coding process
1) Translation 2) Categorization 3) Process
improvement learning
10
Food Coding -Text translation
PBJ
translates to
Peanut butter and jelly sandwich
11
Food Coding - Categorization

• Coding guidelines incorporated into the Food
  coding database as part of the food description

Peas, green, from fresh, fat not added in
cooking Peas, green, from frozen, fat added in
cooking Peas, green, not specified as to form,
not specified as to fat added in cooking
Milk, not further specified Milk, whole, Milk, 2
12
Food Coding - Categorization

• General Coding guideline
• no specific information available
• more detailed information is available

13
Food Coding - Categorization

• Example Peanut butter and jelly sandwich
• No specific information about type of peanut
  butter, jelly, or bread - coder selects the
  default sandwich in the codebook

Peanut butter and jelly sandwich
14
Food Coding - Categorization

• Example reduced fat peanut butter with
  marmalade on whole wheat bread
• Specific information about any of the components
  - coder will select 3 separate foods

Bread, whole wheat, 100 Peanut butter, reduced
fat Marmalade, all flavors
15
Food Coding - Databases
16
Food Coding

• over 7,300 foods
• Exact text matches rare
• Exact food matches are common

17
Food Coding
Codebook
Respondent reports

• Database contains
• Apple, raw
• Apple, cooked

• Food coder selects Apple, raw
• using categorization (guidelines)
• for apples, raw is assumed if cooked is not
  specified

18
Food Coding

• Codebook foods were modified for selected
  characteristics
• type of fat used in preparation
• type of milk used in preparation
• amount of liquid used in dilution

• Created 3,368 Modified foods

19
Food Coding
Respondent reports

• Database contains
• Peas cooked in margarine

• Food coder changes margarine to butter creating a
  modified food
• Peas cooked in butter

20
Food Coding

• Foods new to the market
• Cereals
• Snack foods
• Frozen and prepared foods
• Unusual foods
• Regional and ethnic specialties
• Mixtures of common foods

21
Food Coding
Respondent reports Chocolate Marshmallow Bears
cereal

• Food coder creates a new unknown food
• Food coder enters Chocolate Marshmallow Bears
  cereal
• Food coder selects an existing unknown food

22
Food Coding Management

• Separate databases

• Status codes kept track of Modified and Unknown
  Foods
• Feedback to coders on problems
• choice of Codebook Food
• creation or use of Modified Food
• creation or use of Unknown Food

23
Type of Foods Reported
24
Unique Food Codes
25
Average Number of Foods Per Code
26
Conclusions

• 1994-96 CSFII released within 1 year of
  completion of interviewing
• less than half the time of previous surveys
• processing unknown foods had been a major reason
  for previous delays

27
Conclusions

• Shared use of database
• reduced overall number of reports
• improved tracking of unknowns and resolutions
• resolve unknowns sooner
• update food coding database 1000 changes in
  94-96
• improve coding guidelines

28
Food Surveys Research Group home page internet
address
http//www.barc.usda.gov/bhnrc/foodsurvey/home.htm