Dietary Screener in the 2009 CHIS: Scoring Procedures

• How Analytical Scoring Procedures Were Developed
• Conversion of frequency category to times per day
• Exclusion of extreme values
• Estimation of median sex-age portion sizes
• Creation of food-level variables
• Creation of aggregate variables
• Variance adjustment factors

How Analytical Scoring Procedures Were Developed

Scoring procedures were developed to convert the individual respondent's screener responses to estimates of intake. We created variables for amount consumed of each food group queried, and for amount consumed of total fruits and vegetables (cup equivalents) and added sugars (teaspoons). To do so, we used 24-hour dietary recall data from the National Center for Health Statistics (NCHS) National Health and Nutrition Examination Survey (NHANES) 2003-2006.

[Return to top]

Conversion of frequency category to times per day

Data collected on the 2009 California Health Interview Survey (CHIS) Dietary Screener consist of frequency response and time unit: times per day, week, or month. We converted these responses to daily frequency values, denoted as N_FGk and shown below.

[Return to top]

Exclusion of extreme values

We examined each variable’s distribution of values for times per day. For each dietary variable, values discontinuous with the overall CHIS distribution were excluded. The maximum daily acceptable value and the number of cases excluded are shown below.

[Return to top]

Estimation of median sex-age portion sizes

We estimated the median sex- and age-specific portion sizes for each food from 24-hour recall data in NHANES 2003-2006. The units used were specific to each food group: cup equivalents of fruits and vegetables for fruit, fried potatoes, and other vegetables, and teaspoons of added sugars for soda, energy/fruit drinks, coffee/tea, cookies/cake/pie/brownies, and ice cream. These portion sizes are denoted by P_k.

For fruits and vegetables, cup equivalents are defined in the US Department of Agriculture’s (USDA) MyPyramid Equivalents Database (MPED; version 2.0 for USDA Survey Foods, 2003-2004) and cited in the US Dietary Guidelines for Americans 2010 as:

• 1 cup raw or cooked vegetable or fruit
• 1/2 cup dried vegetable or fruit
• 1 cup vegetable or fruit juice
• 2 cups leafy salad greens.

Median portion sizes for fruits and vegetables are given as cup equivalents of fruits and vegetables:

For added sugars, median portion sizes are given as teaspoons of added sugars:

[Return to top]

Creation of food-level variables

Food-level variables were created for each food queried by multiplying the individual's standardized daily frequency of intake by his/her sex-age specific portion size, denoted by:

Estimated amount consumed daily of Food Group_k = N_FGkP₁

The following food-level variables are available:

[Return to top]

Creation of aggregate variables

A. Fruit and Vegetable Intake

For estimates of fruits and vegetables aggregated across all food questions:

E(Fruits and Veg^1/2) = b₀ + b₁(N_FG1P₁ + N_FG2P₂ + N_FG3P₃)^1/2

The cup equivalents of fruits and vegetables variable was square-root-transformed to approximate normality; N_FGk is the usual number of times per day an individual consumed food group k; k indexes the three fruit and vegetable food groups. P_k is the median portion size of group k. We calculated weighted least-squares estimates of the regression coefficients b₀ and b₁ on the adults (aged 18 and above) in the NHANES 2003-2006 sample, stratifying by sex and excluding extreme exposure values.

Estimation of b₀ and b₁: For cup equivalents of fruits and vegetables excluding dried beans and French fries, the sex-specific estimates of the parameters are:

B. Added Sugars Intake

For teaspoons of added sugars:

E(Added sugars^1/3) = b₀ + b₁ (N_FG1P₁ + N_FG2P₂ + ... + N_FG5P₅)^1/3

The teaspoons of added sugars variable was cube-root-transformed to approximate normality; NFGk is the usual number of times per day an individual consumed food group k; k indexes the five added sugars food groups. P_k is the median portion size of group k. We calculated weighted least-squares estimates of the regression coefficients b₀ and b₁ on the adults (aged 18 and above) in the NHANES 2003-2006 sample, stratifying by sex and excluding extreme exposure values.

Estimation of b₀ and b₁: For teaspoons of added sugars, the sex-specific estimates of the parameters are:

The following aggregated variables were created:

[Return to top]

Variance adjustment factors

In addition to these variables, we applied variance adjustment factors to create variance-adjusted versions of each for use in many analytical situations.

[Return to top]