March 2017 Question 2
Participants in this study were sent self-administered food
frequency questionnaires every four years to assess their own diet. For this
particular study, do you believe this is the best option to achieve accurate
information over this longitudinal study? Can you think of better means to
recording accurate diet information?
Supplemental Article:
W.C. Willett, L Sampson, M.L. Browne, et al. The use of a
self-administered questionnaire to assess diet four years in the past Am
J Epidemiol, 127 (1) (1988), pp. 188-199.
posted by Nikki @ 9:54 AM
27 comments
27 Comments:
In my opinion, after reading the methodology for this present study, as well as the supplementary article, I do think this was a good option. We all know how hard it is to get an accurate assessment or picture of dietary intake and I think in this case, with its many participants and its longitudinal design, the use of a FFQ was appropriate. FFQs ask about of the frequency of consumption of certain types of foods or food groups over a specified period of time and therefore, they present a good picture of historical or typical dietary intake. Additionally, a significant strength is that they can be mailed and done individually without a dietitian’s or other clinician’s guidance. However, since they do focus on food groups and food types (i.e. fruit, vegetables, sweets, baked goods, breads/cereals), it may be challenging to estimate average total calorie intake as not as many details about specific foods/products are reported. Another issue with FFQs involves the estimation of portion sizes – respondents may not know what the described portion size looks like and therefore may over or underestimate it. But like all dietary intake assessment tools, there are both pros and cons. In this case, I think FFQ is a feasible and reliable tool to use in the scope of this research study and its purpose.
Obviously, a tool that captures a shorter or more limited period of intake, such as a 24-hour recall, would not be a good option to estimate average usual intake because it does not allow for a very accurate “typical” representation of food and beverages consumed. Additionally, with this many participants and the scope of being a longitudinal study, any assessment tool that required administration or interviewing techniques, while may be more accurate, would most likely be unfeasible, both in terms of time and expense.
While a FFQ is not recognized as the strongest diet information collection tool, I understand why researchers chose to utilize it for this study. I think the FFQ does a good job at capturing long-term diet history information which was the goal of the study. I also liked that they administered diet history collection every 4 years in order to get on-going, updated information. Perhaps competing 24-hour diet recalls every year or two would have collected a more accurate picture of participants diet, but it would have been more difficult for researchers to coordinate.
I do not believe that a self reported FFQ While a self-administered FFQ is the best option to achieve accurate information over any longitudinal study, but do think it was the best option to complete a study of this size. Any self reported data is very prone to bias and selective reporting by the participant. A one-on-one interview with the participants would have been better so follow-up questions could be asked. Interviews would also allow researchers to jog the participant’s memory and clarify portion sizes in ways that a survey could not. However, this would have been unrealistic in this study and the sample size would have suffered from it. Since the researchers had a relatively large number of participants, the results may have evened out and may have produce fairly reliable results regardless of bias.
When reading this study, I was impressed at how many respondents the study received. However, there was a large difference in the number of respondents based on sexual orientation; there were far more heterosexual respondents than either lesbian or bisexual respondents (98,317 versus 926 and 415, respectively). This is a bit concerning, as the heterosexual group has approximately 100 times more participants than the lesbian group and approximately 200 times more than the bisexual group; I would prefer to see respondent groups of similar sizes.
I do not think that self-administered questionnaires are ever the most accurate way to gather information for research purposes. However, I do believe that the number of respondents would decrease drastically if the participants were made to appear in person in order to participate in the study. As food frequency questionnaires rely on the memory of participants, a few days or week-long dietary log may be more accurate in assessing each participants’ dietary habits, although it would require more effort on the participants’ end and this may decrease participant numbers.
I think a food frequency questionnaire is an appropriate dietary assessment tool for this study based on the number of participants and the study’s objective to assess dietary quality.
When reading this study, I was impressed at how many respondents the study received. However, there was a large difference in the number of respondents based on sexual orientation; there were far more heterosexual respondents than either lesbian or bisexual respondents (98,317 versus 926 and 415, respectively). This is a bit concerning, as the heterosexual group has approximately 100 times more participants than the lesbian group and approximately 200 times more than the bisexual group; I would prefer to see respondent groups of similar sizes.
I do not think that self-administered questionnaires are ever the most accurate way to gather information for research purposes. However, I do believe that the number of respondents would decrease drastically if the participants were made to appear in person in order to participate in the study. As food frequency questionnaires rely on the memory of participants, a few days or week-long dietary log may be more accurate in assessing each participants’ dietary habits, although it would require more effort on the participants’ end and this may decrease participant numbers.
I think a food frequency questionnaire is an appropriate dietary assessment tool for this study based on the number of participants and the study’s objective to assess dietary quality.
I agree with the other comments that while FFQ will not provide the greatest, most accurate results, they are a means to an end. Erin makes a good point that the number of respondents would have most likely decreased if they were asked to do person-to-person interviews or dietary recalls. Many have commented that 24 hour recalls might have given a more accurate description, but this might not be the case in a longitudinal study. As we know, a 24 hour recall captures one day--it is not always representative of an overall diet. While sometimes it has better measurement that a FFQ, people often over and under estimate. The FFQ gives overall portion size estimations that are not perfect, but is easier for participants to use/choose rather than blindly assuming how much they ate. Again, the FFQ covers a longer period of time, which in turn is a better representation of diet quality.
I am not a fan of FFQ but I understand the convenience and realistic application of them. I do not trust that the majority of those that fill the questionnaires out are truly accurate or honest with their responses. I also find them to be very long and I could foresee applicants losing patience and in turn answering hastily without considerate thought. Although I have my doubts, the article mentions that the follow up rate with the FFQ used as a tool is very successful.
I agree with Jamey that the 24 hour recall may not be a better representation of dietary habits. I've seen it multiple times when working with a weight management program that patients are embarrassed to honestly report what they consume. Also, it is difficult to recall every single thing one ate 24 hours later. That chip here or m&m there are often forgotten. I personally cannot think of a better option than the FFQ even though I am not a fan.
I wouldn’t say that I love the FFQ formatting, but I believe it is the best way to get information that spans a large amount of time for such a high number of participants. What was interesting to me was actually the implementation of the FFQ and the time periods in between. From my understanding, the participants submitted an FFQ every 4 years, while the an FFQ normally asks about choices made across a month time span to estimate intake for a year, correct? If an individual experienced large changes in their diet during year 2, but this didn’t continue to year 4, we would miss out on certain data points that may give insight to dietary patterns. Would a measure every other year or even yearly help to give a greater understanding of changes in dietary pattern, and with further questioning, maybe the reasons behind the changes? The current format is likely the most effective for this study design, but an alteration in the application may have given different kinds of information that might be interesting to see.
I do not believe the FFQ is always the best survey to use. However, I do believe that for this study the researchers chose a good tool for acquiring information. I also think that the longitudinal design of the study was a very good idea. However, changing the years of assessment from 1 to 2 years would provide more accurate information and could allow the researchers to assess for other factors like diet trends throughout the years.
Elyse, I agree with your comment. Through my own experience I have learned, as you mentioned, longer surveys lead to lack of participation and incomplete surveys. For my thesis I administered a few surveys and together they were very lengthy. There were many participants who started the surveys and did not finish them therefore, I was not able to use their responses. I also think it is important to consider that the way one person reads a question can be interpreted differently to others, leaving room for bias. Data was also self-reported leaving more room for potential bias.
I feel that FFQs can be very beneficial in certain circumstances. However, I feel that they may not have been the best tool to use for this particular study. Overall, the study had the FFQ completed every 4 years. This long period of time may tend to "average" the types of foods consumed, instead of portraying a daily usual intake. This may lead to a skewed perception that diet quality is better in one group versus the other. Perhaps a one-week food log or several random three-day food logs would provide more insightful snapshots of the participants' eating habits.
Jenni and Moriah bring up valid points regarding self-reported information, especially portion sizes. There is always room for potential bias and human error to occur. However, this is more convenient for participants and may help to maintain participant retention.
Anytime people elect to participate in a research study, it is important to note there may be subtle biases in the data collected. However, the benefit of completing a self-FFQ is that the survey is anonymous versus an interview where the is another person present. This can allow for greater honesty from the subject. The only concern with this longitudinal study is the FFQ occurring every 4 years. I am curious as to the researchers justification for the 4-year survey administration for I could see pro and cons to this time frame. A benefit may be you are able to get see a persons food trends change subtly; however I question if 4 years is not enough time to see great change in food habits. Contrariwise, completing the FFQ every 6-8 years may better differentiate between eating patterns as one ages, but would have less specificity.
The study addressed the limitations of using a FFQ in comparison to a food diary in terms of the accurately of the data collected and although food diaries/logs may be more accurate in capturing the “exact” dietary patterns for the specific days recorded, logs are still prone to self-reported bias and error. Overall, given the population size and the longitudinal design of the study, I believe that utilizing FFQ was the best choice for this study.
Participants in this study were sent self-administered food frequency questionnaires every four years to assess their own diet. For this particular study, do you believe this is the best option to achieve accurate information over this longitudinal study? Can you think of better means to recording accurate diet information? For this study I feel that FFQs were most likely the best option to achieve the information, however I don’t know how accurate the information was. In this case I believe FFQs are the best way to get longitudinal information, and aren’t as laborious as face to face interviews. FFQs can collect a good amount of data in a short period of time. One thing I do not like about FFQs is that they are somewhat long to complete, leaving opportunity for individuals to just start randomly putting things down due to length and simply trying to get through the FFQ or they might be embarrassed of their diet and put false information down. This is the reason I did not use an FFQ in my study, however, I sometimes wonder if I used a FFQ I would have gotten better information on food trends within my population.
Elyse, I appreciate the insight you bring with the experience you have with your job and administering FFQs to your clients. I imagine that some induvial may be shy to share the truth about their diet and may feel inclined to stretch the truth on certain things. Also, I like how you pointed out its just simply impossible to be 100% accurate on a FFQ, it’s very difficult to remember everything you eat and how often for most individuals!
I believe using an FFQ for this type of study is an acceptable choice of diet assessment tool. 24 hour diet recalls would likely not capture an overview of typical food choices. A food diary would be burdensome on participants to complete and for researchers to analyze, especially for a large sample size. The burden of a food diary could decrease participant adherence and accurate reporting of dietary choices. Interviews would be very time consuming and labor intensive, and would likely involve qualitative data that would then need to be coded into quantitative data. The FFQ captures a snapshot of typical food choices and is easily quantifiable, which is desirable for a large sample size. An electronic option for the FFQ completion may be more convenient than a mailed survey, if feasible.
Erin’s comment about the differences in number of respondents in each group is an important consideration. Even though sample sizes in each group are still adequate, I wonder if these disproportionate response rates in each group would skew any data collected. Moriah makes a valid point about the possibility for variance in individual interpretation of FFQ questions. In any type of survey, there is a risk of variances in interpretation that could lead to inaccurate responses. Moriah suggested giving out the FFQ every 1-2 years to assess diet trends more closely over time, while Noel suggested that 6-8 years would better represent changes in diet behaviors. While I think they both make excellent points, after considering the frequency of survey administration, I believe surveying every 4 years seems adequate. I doubt that regular diet patterns change very much in 1-2 years, but think that 6-8 years is too long to go between assessments.
With this sample size (517,000 participants) and the length of this study (22 years), I think the Food Frequency Questionnaire was the best option for collecting data. (The questionnaires were sent every 4 years. I wonder why they chose that time frame.) If the participants were to provide a food recall for a couple days, the participants may underestimate serving sizes, and the days of the data collection may have been unusual. For example, maybe the data was collected during a week when the participant was sick or during a holiday. Ideally, I do think that a seven-day food recall (given that it is an average week for the participant) may give really accurate information, I also think this would be difficulty to analyze for the researchers. It could also be hard to assess things like sodium intake just using the Food Frequency Questionnaire if the participants are simply checking food groups like vegetables and not knowing if these vegetables are fresh, frozen, or canned. Furthermore, are they regular or reduced sodium? Are they steamed fresh or covered in cheese sauce? With all of these unknowns, I think the most realistic evaluation would be using the Food Frequency Questionnaire or a seven-day food analysis if this would not decrease the response rate.
The FFQ seems to be a reasonable choice as a survey tool for this population and time duration. This tool is limited in the type of data it can collect, which resulted in several gaps and areas for further study among the articles conclusion. Trends were introduced of both the lesbian subgroup having a high quality diet on average and higher weight on average. An alternative tool such as intermittent diet recalls would have allowed for total kcal records and opened researchers to the possibility of hypothesizing causes for reported trends.
I find food frequency questions to be a useful to for diet assessment; however, in working on my thesis, I have come across a few studies that have commented on the use of food frequency questionnaires as the sole source of dietary info. Keep in mind, these studies were far shorter in duration, but nevertheless, I think the critique still holds. To paraphrase the studies, they often note that cross-validation of data collection tools are important when it comes to dietary behaviors. By this, they generally mean having another manner of assessing dietary behavior to be able to identify instances when self-reported dietary behavior differs from actual dietary behavior. If I remember correctly, one study suggested that coupling a food frequency questionnaire with a 24-hour diet recall is one way of achieving this--just as Justine mentioned. Granted, coordinating and getting properly completed 24-hour diet recalls in a sample this large would be complicated in-and-of itself; nevertheless, perhaps asking a subset of each group in the study to complete an in-person 24-hour diet recall would help better validate the dietary behavior data.
As many people have already expressed, I too am not a huge fan of the FFQ, but I believe this is an accurate way to evaluate the subject’s diet quality. Because the FFQ was mailed to the participants every four years, it is more logical to provide a questionnaire that asks about frequency instead of extremely detailed information. If this was a smaller sample size or shorter study, a FFQ may not have been the best route to take and I would suggest something more detailed such as an interview, dietary recall, or food journal. These methods would simply be too much for the extent of the study conducted.
I think we can all agree that no dietary assessment tool is ideal and that each has its pros and cons. When choosing a tool for a study or to use with a client, we just need to consider our purpose and goals and then consider how we can deal with and handle the trade offs that come with using that particular tool. For me, our discussion here really reiterated how we must choose a method that will get participation, ideally with a high level of accuracy in reporting. Without participation and completion we have no data. Do you second year interns have any tips for communicating with clients or patients about the importance of accurate and complete reporting and recording regarding dietary assessment/intake tools?
We have all learned that Food Frequency Questionnaires are not always the best option for collecting nutrition related data. It has been questioned if FFQs are actually valid due to depending on participant’s memory and people trying to seem healthier than they actually are. FFQs represent a snapshot of habitual intake and are most commonly used in epidemiological studies. Therefore, I think that an FFQ is acceptable for this study. Since this was a longitudinal study, it would be more inconvenient for the participants to use a diet recall or food record for several years, whereas an FFQ does not take up nearly as much time. Although preparation methods and serving sizes are difficult to account for in an FFQ, I think that overall this is the most effective way to collect data for this population. Erin mentioned that if diet recalls or food records were required, the number of respondents would decrease and I agree. I think it would be even more efficient to have a mailed and electronic version of the FFQ to increase response number even greater.
I also agree that self-administered food frequency questionnaires are not the favorite option to achieve accurate information, but does the job for this longitudinal study. The supplemental article’s abstract says that the FFQ provides “useful estimates,” but obviously something a little more precise would be ideal. In the main article, I read that FFQ were seen in a positive light because other research done within this population focused on yes or no questions regarding intake or only looked at certain aspects of the diet, such as fruit and vegetable intake. At least with the FFQ, we are getting a more encompassing picture! As others have mentioned, when considering other options, we can start to pick out flaws in them as well. It is really about trying to figure out what works best after weighing out certain factors. I like Holly’s idea of random three-day food recalls, or Alyssa’s seven-day food analysis, which may combat some of the concerns we’ve listed regarding FFQs.
We are all pretty much on the same page that FFQ aren't the best, but the have their use and purpose. I am interested to see what the future holds. There are positive and negatives for all questionnaires and surveys, so I wonder if/when professionals in our field will develop ones that are more valid, less bias, user-friendly and so on.
For my response, I was interested in diving more into FFQs for some reason. :) I believe that we simply have to see them as they are and match our desired study outcomes to what they can provide us, not getting too lofty in our goals for the results (or pick a different tool!). Jenni hit on the fact that FFQs are prone to both under- and over-estimation of intake. Taking this a step further, I remember one of my undergrad professors saying how much difficulty she would have filling an FFQ out due to the fact that her intake varies substantially season to season. The example she gave related to summer- during the summer months, she has a garden and grows many vegetables, including tomatoes. Apparently she is fairly picky about the tomatoes she eats, so really only eats her homegrown ones and rarely has a tomato the rest of the year. We also have to be aware of the high prevalence of bias/selective reporting that comes with self-administered tools, as Reilly highlighted. Erin addressed participants’ memory- I would have trouble remembering some of the more specific foods that I ate (different from my “go-to” foods), as well as the frequency and amount. Additionally, I want to address Lauren’s question posted to the group- she showed concern in the fact that with a data collection of every four years, it would be difficult to receive an accurate dietary pattern as I would think many people’s diet could change in two to four years! I would say yes, a yearly FFQ would be a better choice in this case.
Some commented on doing random three-day food records instead of the FFQ to increase accuracy of reported intake. I read an article the other day talking about doing both a FFQ and three-day food record simultaneously. Just as when utilizing other tools, it is important to realize how our results may be skewed. The FFQ was found to overestimate the macros of PRO and fat compared to the three-day food record. I would also suggest for a dietitian to review the three-day food record with the participant (and possibly even the FFQ if they have questions), hashing out portion sizes and trying their best to get the most accurate picture of intake.
Jamie, to follow on your idea of developing new user-friendly survey or questionnaire, I believe the use of technology will be a huge contributor to this. There are already a large amounts of apps utilized by people on their phones in order to track what they eat. It’s only a matter of time before the health field catches on and creates a well-regulated and certified app for people to utilize that isn’t time consuming, easily navigated, and accurate. They could be programmed to aid in specific diets, send random food logs to their health care provider to be monitored, and would be a great tool for recording intake. Though like all questionnaires and surveys, it would only be as accurate and honest as the user allows.
Jenni,
To communicate with a patient or client, they must feel safe and yet maintain the professional atmosphere which preserves the importance of the interview. This is difficult but best modeled by professionals in the field, when gaining experience I preferred to observe my preceptors and note not only their words but their mannerisms in interacting with patients to gain the optimum results.
Justine and Michael, I think you both bring up some good points about the limitations of a FFQ in data collection. I agree, that utilizing a 24-hour recall in conjunction with the FFQ would be a great way to assess differences in self-reported data and help maximize the accuracy of dietary data gathered.
Post a Comment
<< Home