Indication
FFQs are one of the most commonly used retrospective methods in nutritional epidemiology and have become a key research tool in examining the relationship between dietary intake and disease risk. They have been used in a wide range of dietary studies including cross-sectional surveys, case-control studies and cohort studies. They may be a particularly useful method to measure specific dietary behaviours and the intake of particular food groups (e.g. fruit and vegetables) or selected micronutrients which occur in a limited number of foods (e.g. calcium). Due to their ease of administration and relatively low respondent burden FFQs have been used extensively in large-scale cohort (prospective) studies, including for example, the Whitehall II study (Mosdol et al, 2007), the European Prospective Investigation into Cancer and Nutrition (EPIC) - for which different questionnaires were developed and validated for use in the various participating populations – (Riboli et al, 2002) and the UK Women’s Cohort Study (Greenwood et al, 2000). However, as the duration of cohort studies can last several decades it is likely that the questionnaire becomes ‘out of date’ as new foods enter the market over the course of the study period and dietary patterns change. It is important to consider the length of time over which an FFQ may need to be used at the time of its design. The FFQ may need to be revised if it is to be used at several time points throughout the duration of the study. This can make comparisons between time points more difficult.
FFQs can be adapted for a particular purpose e.g. to assess calcium intake or habitual consumption of oily fish, but in many cases a comprehensive food list is used so that intakes of all nutrients and total energy intake may be determined. Total energy intake is required for energy-adjustment and for assessment of mis-reporting.
There is currently a debate about the relative merits of using FFQs in large-scale prospective studies investigating diet and disease relationships (Kristal et al, 2005; Bingham et al, 2006) and whether FFQs are sensitive enough to detect important diet-disease relationships (Freedman et al, 2006). In particular, the association between dietary fat consumption and breast cancer is controversial (Thiebaut et al, 2008) and it has been questioned as to whether FFQs have too many limitations (Wirfalt et al, 1998 Bingham et al, 2003; Crozier et al, 2008). Results from studies using biomarkers of specific nutrients as the reference method for assessment of dietary intake suggest that the measurement error associated with FFQs is larger than was previously estimated (Day et al, 2001; Molag et al, 2007). Others have found that the FFQ performs well in comparison with diet records (Brunner et al, 2001). The place of FFQs has been defended by other researchers (e.g. Willett, 1998; Willett, 2001). It is important to remember that not all FFQs are the same and some may be better than others in terms of assessing food and nutrient intakes. For any study, the advantages and disadvanteges of using FFQs compared to other dietary assessment methods should be carefully considered (McNeill et al, 2009).
Recently, several researchers (Chambers et al, 2000; Johnson-Kozlow et al, 2006; Matt et al, 2006) have explored the use of cognitive interviewing techniques to increase the validity of self-report data. Respondents are encouraged to ‘think-aloud’ and verbalize their thought processes as they comprehend questions and retrieve information from long-term memory to answer questions on the FFQ. The aim is to improve the validity of data obtained from food frequency instruments by examining the cognitive strategies people use to formulate answers to FFQs and identify difficulties in formulating answers to specific questions.
The Diet History Questionnaire (DHQ) is an example of an FFQ which has been developed by investigators at the U.S National Cancer Institute based on cognitive research findings (Subar et al, 2001; Thompson et al, 2002). Cognitive interviewing helped highlight several issues in FFQs related to respondent comprehension, intake of seasonal foods, and ordering of food items which were then addressed in the DHQ. Attention was also paid to developing a food list which was nationally representative and reflected changes in the food supply such as the increasing presence of low-fat choices.
Study findings suggest that individuals have difficulty estimating portion sizes of foods and may skip questions related to portion size after completing questions on frequency of consumption (Subar et al, 1995). Qualitative study methods have also highlighted that several factors can influence individuals’ perception of portion size, including; the type of food being considered, the role of a given food item in the meal (i.e. as a main or a side dish) and personal preference for the food (Vuckovic et al, 2000). Cognitive research has also indicated that the level of grouping of foods can affect the recall of food intake. Results suggest respondents find it easier to respond to items when disaggregated for example, disaggregating peanut butter from peanuts, biscuits or cookies from cakes and beef from pork, lamb and ham (Thompson et al, 2002; Kozlow et al, 2006), but this needs to be balanced with the disadvantages of longer food lists and the likelihood of over-estimation of intake and additional respondent burden.