Food Frequency Questionnaires - FAQs

When was the orange FFQ analysis program created?

The Cancer Epidemiology Centre of the Cancer Council Victoria released the software Food Frequency Questionnaire Analysis version 3.0.6.1 in 2006.

What does the FFQ show me?

In most epidemiological applications the aim of the FFQ is to rank people according to intake. It will show the intake of macronutrients and a variety of micronutrients. Please download attached pdf of the User's guide for the FFQ which contains some examples of the dietary output and an excel file with some further examples of raw data and nutrient analysis. 

What will I receive with my FFQ analysis?

You'll receive 5 electronic excel files, which include:

1. raw data
2. Nutrients computed from food without alcoholic beverages (including carotenoids & fatty acids, glycaemic index & glycaemic load),
3. Nutrients from alcoholic beverages,
4. Nutrients from food intakes, and
5. An error report.

Can we change the FFQ questions?

The current analysis program has been developed in 2006 to analyse the orange version of the FFQ. We aim to produce a FFQ that's scientifically calibrated and a useful tool that gives you results. Our FFQ has been calibrated and validated for the foods and time period specified. We are, however, interested in hearing about your specific needs for future versions.

Can I select only certain questions from the FFQ?

In most instances isolating micronutrients will not give an accurate indication of the participant's intake as these nutrients may be dependent on other micronutrients in the diet. For example, the omega 3 intake is usually considered in relation to the omega 6 intake, as the absorption in the body will depend on the ratio of concentration of each in the diet.

Can I use the FFQ to measure a month or 3-month of dietary intake?

Our FFQs have been designed to measure usual eating habits over a 12-month period. They have also been validated on intake for a 12-month period. They may not be sensitive to small changes in intake, and thus may not be appropriate for measuring short-term dietary change in response to interventions.

By confining the FFQ to a 3-month period, there may be inaccuracies due to several factors:

• The availability of fruit and vegetables vary due to their season, which may exclude some items depending on which time of the year the FFQ is completed.
• Alcohol consumption may be overestimated if the FFQ is completed at a time when there have been more occasions where alcohol was consumed, compared to the rest of the year.
• Periods of fasting or religious events (this may or may not be relevant to the study population)
• Holidays (may affect the amounts and types of foods eaten during this period).

What if the participant's answer doesn't correspond to the FFQ timeframe?

As the FFQ is asking about food intake over the previous 12 months, if a participant consumes a food item, for example 10 times per year, this would be considered as ‘less than once a month'. If a participant were to consume a food everyday for only a month, approximately 30 times a year. These need to be divided over the 12-month period, and the answer would be ‘1 to 3 times per month'.

How are portion sizes represented?

The Food Model Booklet (FMB) is an updated version of ACCV's 1994 Health 2000 Food Model Booklet that we've used in our recent Dietary Calibration Study (DCS). As it proved extremely effective in the DCS we have decided to also make it available for other studies.

Selection of the number of images/drawings were based on our 2 main criteria:

1. to develop a visual reference guide to complement both our dietary questionnaire and the 24-hr recall program by FoodWorks and
2. to provide enough portion size references while still being easy to access by participants (i.e. user friendly).

The FFQ is missing portion size information, what do I do?

A portion size factor is determined based on an average of the responses on the portion size questions as selected by the participant. If the participant has not selected a portion (for example, a vegetarian, who has answered ‘I never ate steak' to the meat question) then the average will be determined by the remaining portions, which have been selected.

If no portion size images are selected nutrients will be determined with the assumption of the standard portion size. This will be reflected in a serious warning message in the error file.

For missing or less portion size questions, it is the user's decision to use or not use that information.

Can we photocopy pages from the Food Model Booklet?

The Intellectual Property in all versions of the FFQ remains the exclusive property of The Cancer Epidemiology Centre, and cannot be reproduced without consent.

How long does it take to process the FFQ?

The standard timeframe for FFQ is within 10 business days from the date of receiving the completed FFQs.

What happens when participants under-report their energy intake?

The investigator should decide on his/her own cut-off values. For example, in the Melbourne Collaborative Cohort Study, we considered outliers at the top and bottom of the 1% sex specific energy distributions to be extreme and have excluded them from analysis.

Other studies have used an energy intake/BMR ratio (EI:BMR) as a guideline. An EI:BMR of less than 1.15 was deemed to be below physiological limits and used as the cut point in another study (Bingham et al., 1997).

Does the FFQ come in a web-based version?

The CCV is currently in the process of developing a web-based FFQ.

What do the error messages mean?

1. Serious messages indicate the following errors ; For example: "(Q1) 'I don''t eat fruit' and fruit intake reported in (Q15)"; "portion size factor(Q11-14) is based on only 1 food or has defaulted to 1"; "consumption of alcoholic beverages indicated in (Q16) but number of glasses per day (Q17) is missing"; (Q11-14) no portion size images have been selected'
   
2. Intermediate messages indicate the following errors; For example: "cheese type (Q10) is missing or 'I don''t eat cheese' but (Q15) indicates cheese consumed"; "portion size factor(Q11-14) is based on only 2 foods"; "Milk type (Q3) is missing or 'none' but (Q4) indicates milk consumed"; "Bread type (Q5) is missing or 'I don''t eat bread' but more than 1 slice of bread/day consumed (Q6)"; "Bread type selected in (Q5) but bread amount missing in (Q6)"; "Bread type (Q5) is missing or 'I don''t eat bread' but more than 1 slice of bread/day consumed (Q6)";
   
3. Trivial messages indicate the following errors; For example: "portion size factor(Q11-14) is based on only 3 foods"; "missing maximum number of glasses (Q18)"; "Milk type selected in (Q3) but 'none' used in (Q4)"; "cheese consumption indicated in (Q10) but frequency 'never' in (Q15)"; "(Q1) reports eating fruit and no fruit selected in (Q15)"; "(Q2) reports eating 1 or more vegetable per day and no vegetables selected in (Q15)";

Why have some FFQs been rejected?

The analysis program automatically rejects questionnaires where a page of data is missing and reports this in the accompanying error report file.

It is up to you to decide how to manage this data (i.e. recontact the participant to complete the questionnaire). To speed up the data analysis this is best decided before the FFQs are sent for processing.

If any question has been left unanswered or is inconsistent with previous responses (i.e. ambiguous) then these responses or lack of responses are also identified and reported in the error file.  By manipulating this file (sorting by ID) individual records can be identified and missing data quantified. Based on this information arbitrary decisions can be made as to whether or not to include these records in analyses. The error file also qualifies (e.g. trivial) and locates (e.g. Question no. 11) each type of error enabling further review by your research team.

Do we need to send all the completed FFQs back together in one batch for analysis?

We're happy to receive the completed FFQs either in small batches or all together - which ever you prefer.

How are the FFQs labelled?

For confidentiality purposes all FFQs provided will contain a barcode. To create these barcodes we require your study name (acronyms are most often used, and this part remains constant) and a numbering system, which usually corresponds to the number of FFQs ordered. The alphanumeric barcode can be up to 10 figures (e.g. TCCV0001 - TCCV1500 for an order of 1,500).

How is the nutritional composition of foods calculated?

Weight of a standard portion (determined by the respondents portion size) x Frequency of consumption = total intake per day

How are the nutritional composition of fruit and vegetables calculated?

Weight of standard portion x frequency of consumption x the consumption of this fruit or vegetable relative to all other reported fruit and vegetables = total intake per day

How is alcohol consumption per day calculated in the FFQ?

Question 16 - indicates the types of alcoholic beverages consumed

Question 17 - provides you with the number of drinks per day. Alcohol g/day in nutrient analysis is determined from both question 16 and 17.

Question 18 - provides information on maximum number of glasses consumed in any 24 hour period and has been designed to help identify any 'binge' behaviour.

What does the FFQ actually measure?

The current FFQ comprises a food list of 74 items with 10 frequency response options ranging from 'never' to '3 or more times per day''. . It also contains 4 sets of photographs of scaled portions (used to calculate a portion size calibrator), questions on the overall frequency of consumption of fruit and vegetables, and some foods, which do not fit easily into the frequent format.  The 74 food items are grouped into 4 categories: 1) cereals, sweets and snacks; 2) dairy products, meat and fish; 3) fruit; and 4) vegetables, and separate questions for alcohol. The raw data will provide the following information:

Water Protein Starch & Dextrins Sodium Phosphorus Zinc Thiamin Niacin Alcohol Fat (total)

Kilojoules Carbohydrate (total) Dietary Fibre Potassium Magnesium Retinol Equivalent Beta-Carotene Equivalent Niacin Equivalent Saturated Fatty Acids (total) Polyunsaturated Acids (total)

Calories Sugars Cholesterol Calcium Iron Retinol Riboflavin Vitamin C Monounsaturated fatty Acids (total)

 

Plus new additional output:

Vitamin E, Folate	British data  [2]
Carotenoids	US data [3]
Fatty acids	Australian data [5]
Glycaemic Index, Glycemic load	International tables [4]

The frequency scale uses the following variables:

Recorded response	Frequency	Daily equivalent frequency
1	never	0
2	less than once per month	0.02
3	1-3 times per month	0.07
4	once per week	0.14
5	twice per week	0.28
6	3-4 times per week	0.5
7	5-6 times per week	0.78
8	once per day	1
9	twice per day	2
10	3 or more times per day	3

 

To determine a daily food intake the formula would be:

Weight of a standard portion (determined by the respondents portion size) x Frequency of consumption = total intake per day

Has the FFQ been validated?

Yes. See relevant studies below.