Transcript FFQ - WHI

Use of Dietary/Nutrient Data in
the WHI
Marian L. Neuhouser, PhD, RD
WHI Data Training Workshop
May 6, 2009
[email protected] or [email protected]
Presentation Outline & Objectives
1. Review dietary assessment procedures
in WHI
2. Present types of dietary data available for
analysis
3. Special examples for using nutrient data
4. Using computed variables from the FFQ
line items
5. Summary
6. Q and A
How Was Diet Measured
(DM)?
Study Start
Food Frequency
Questionnaire
All
Year 1
Food Frequency
Questionnaire
All
Year 2Closeout
Food Frequency
Questionnaire
Rotating 1/3
of participants
Year 3,6,9
4-day food records 4-day food records Repeat (x2)24 hr Recalls
4.6% of ppts
4.6% of participants
24 hour recalls
1% of participants
Dietary
How Was Diet Measured
(OS)?
Study Start
Food Frequency
Questionnaire
All
Year 3
Food Frequency
Questionnaire
All
Note: CT participants not in the DM only have SV (baseline) FFQ)
Dietary
Which dietary data are available?
Only FFQ data have been released to
www.whiops.org
Food record data have been used in
selected publications (including some in
process) but these data are not posted.
The focus of this presentation will be on the
FFQ data.
FFQ Data Availability & Analysis
1. Analysis by nutrients
2. Analysis by food groups (computed
variables)
3. Analysis by individual or grouped line
items, adjustment questions and
summary questions
Potentially useful for “index” or dietary
pattern type analyses (computed
variables not yet made)
FFQ analysis by nutrients and use
of nutrient data
• Standard analysis of FFQ data yields
nutrients per person per day
• 135 nutrients/important compounds:
Fat, fatty acids protein, carbohydrate, vitamins,
minerals, carotenoids, genestein, daidzein,
sugars, fiber, vegetable protein
• Most WHI analyses have used the
standard nutrient data
WHI FFQ Structure and Format
• The WHI FFQ has three parts
• 19 Adjustment Questions
– Food preparation, fats added in
cooking/table
• 122 line items of foods or food groups
– Frequency and portion size
– 122 line items = > 350 foods
• 4 Summary Questions
– Overall usual intake of fruits and
vegetables, fats added to foods in
cooking
– Used to reduce measurement bias of
over-reporting long lists of foods
Processing the WHI FFQ and
creation of nutrient output
•
Step 1: FFQ are passed thru a marksense scanner. Responses are converted
into ASCII data.
•
Step 2: ASCII data checked for errors (ie,
missing)
•
Step 3: Calculating nutrient intake
Processing the WHI FFQ and
creation of nutrient output
1. Processing system estimates annual
medium portion sizes by multiplying
frequency * portion size
Example: Medium portion of eggs once a
week = 52 annual medium portions.
Small portion of eggs once a week = 26
annual medium portions [s = .5(med)]
2. Annual medium servings are multiplied by
the nutrient string for the food code (135
possible nutrients)
3. All nutrient results for all foods are
summed and divided by 365 = average
daily nutrient intake of each nutrient
Processing the WHI FFQ and
creation of nutrient output
Note that many line items have multiple
foods
Example: peaches, nectarines, plums
rice, grains and plain noodles
Each food is proportionately represented on
the line item. Proportional weights are
based on national consumption data and
expert opinion.
The processing stream occurs for each
food. USDA (NCC) may have several food
codes for each item. Careful assignments
are made by nutritional epidemiologists.
FFQ analysis by nutrients: caveats
– know what you are analyzing
Example 1: Folic Acid/Folate
Which variable to use?
•
•
•
•
Dietary Folacin
Dietary Folate Equivalents
Dietary Synthetic Folate
Dietary Natural Folate
FFQ Analysis by Nutrients
Dietary Folacin
“Old” folate values that represent the
pre-1998 food fortification values.
Approximately equivalent to natural
folate PLUS foods already fortified
pre-1998 (e.g., some cereals)
Dietary Synthetic Folic Acid
The enriched/fortified portion (post
1997-8)
FFQ Analysis by Nutrients
Dietary Natural Folate
Folate from natural sources only,
such as fruit, vegetables, liver, egg
Dietary Folate Equivalents
Combines synthetic folic acid and
natural folate but takes into account
the lower bioavailability of natural
folate
DFE = [µg of food folate + (1.7 x µg folic acid)]
FFQ Analysis by Nutrients
Example 2: Vitamin E
Dietary Alpha-Tocopherol
Dietary Beta-Tocopherol
Dietary Delta-Tocopherol
Dietary Gamma-Tocopherol
Dietary Natural Alpha-Tocopherol
Dietary Synthetic Alpha-Tocopherol
Dietary Total Alpha-Tocopherol
Equivalents
Dietary Vitamin E
FFQ Analysis by Nutrients
Example 2: Vitamin E
Dietary Alpha-Tocopherol
Dietary Beta-Tocopherol
Dietary Delta-Tocopherol
Dietary Gamma-Tocopherol
Dietary Natural Alpha-Tocopherol
Dietary Synthetic Alpha-Tocopherol
Dietary Total Alpha-Tocopherol
Equivalents
Dietary Vitamin E (IU)
Use others at your own discretion
FFQ Analysis by Nutrients
Example 3: Trans Fats
• Many investigators interested in trans
fats due to the purported associations
with disease risk
• Caution: not all trans fats are created
equal
• Some trans fats are naturally
occurring, some are created during
food processing (but not
hydrogenation) and some created by
hydrogenation
FFQ Analysis by Nutrients
Example 3: Trans Fats
Which trans fat variable(s) should be
used?
Dietary Trans Fatty Acid16:1T
Dietary Trans Fatty Acid18:1T
Dietary Trans Fatty Acid18:2T
Dietary Total TransFattyAcid
Be very cautious about trans16:1T and
total trans fats
FFQ Analysis by Food Groups
Food groups are constructed variables
made by grouping and summing the line
items from the FFQ
Computed variables are available for
the following food groups:
Fish
Red Meat
Poultry
Grains
Whole Grains
Nuts
Soy
Milk
Dairy
These have all been created from the line
items and are now posted.
Are there other uses of the line items?
New Food Group Example:
Cruciferous Vegetables
1. There is no line item called “cruciferous
vegetables” so it must be constructed
from existing line items
2. Find line items with cruciferous
vegetables
3. Summing across yields medium servings
per participant per day
So far so good – these are all in the Brassica or Cruciferae Family
But here is a problem since spinach is not in the Brassica family, but the
other greens listed are Brassica.
Other uses of the FFQ line items
Dietary Patterns or Indices
1. Diet Quality Index
– Reflects adherence to the 8 Diet and Health
recommendations of the National Academy of
Sciences
– Estimates of nutrient intakes are scored
–
–
Neuhouser et al J Am Diet Assoc 2000 100: 576-579
Neuhouser et al Pub Health Nutr 2003 6: 703-709
2. High-Fat Snacks/Desserts Index
•
Potischman et al Cancer Causes Control 2002; 13:937-946
3. Mediterranean Diet
4. Healthy Eating Index
5. MPEDs (My Pyramid Equivalents)
We don’t have these food groups/indices set-up as
constructed variables. Some manuscript writing
groups are doing on their own
FFQ Adjustment Questions (newly
posted)
19 Adjustment Questions at start of FFQ
– Types of cereal
– Types of milk on cereal and in
beverages
– Food preparation, fats added in
cooking/table
– Types of cookies and pastries
These questions were designed to more
precisely adjust the fat content of the line
items since fat reduction was a primary
goal of the WHI-DM
One response option permitted for this adjustment question
The choice selected in the adjustment question is applied
to the response for this question. Thus, is someone
marked “regular” then the nutrient string for regular
ground beef is used for this question. If someone marked
“extra lean” then the “extra lean” nutrient string is used
for this question.
And so on …. 5 possible choices. Participants permitted
to mark one or two responses.
The adjustment question is used for this line item.
If only one option is marked, then the marked
option used as 100% of the nutrient string. If two
options are marked, ½ of the nutrient string
comes from each option. For example, ½ from
“highly fortified cereals” and ½ from “granola”.
FFQ Adjustment Questions:
Note: the adjustment questions were not
originally designed to be used by
themselves.
Therefore, we cannot provide any data on
the validity or reliability of using the
adjustment questions alone (ie – “did you
eat popcorn?” or “What type of salad
dressing did you use?”
Caveat emptor!
FFQ Summary Questions
Four summary questions at the end of the FFQ on:
Fat used to deep fry or sauté
Fat added in cooking
Vegetable intake
Fruit intake
Note that these questions are frequency, not “what
kinds of fat” so they are different than the
adjustment questions
Used to adjust the frequency of fruit or vegetables
since respondents may tend to overreport when
long lists are provided
The fruit and vegetable summary questions have
previously been used to create the fruit and
vegetable variables already in the database.
Summary
The FFQ data are posted on whiops.org;
recall and food record data are not posted.
The nutrient data are the primary data
generated from the FFQ. The output yields
average daily intake of 135 nutrients per
person per day.
When using nutrient data be careful about
which nutrients to use. Know what you are
analyzing. Examples: folate, trans fats and
vitamin E variables.
Summary
The individual 122 line items, 19 adjustment
questions and 4 summary questions are
also now posted. No known
validity/reliability of using the adjustment
questions by themselves.
The 122 line items represent about 350
foods, each with a proportional weight.
Don’t deconstruct the line items under the
assumption that you know the weights.
Selected constructed variables from the line
items are now posted.
Summary
We did not cover:
Measurement error in the FFQ
Use of calibrated estimates of energy,
protein or percent energy from protein
Adequacy of the actual nutrient values in
the database. Imputation of various
nutrients.
Questions?
• Marian Neuhouser, PhD, RD
– [email protected] or
[email protected]
• Lesley Tinker, PhD, RD
– [email protected]