What diet recall methods are used in nutrition research?

I will be giving a talk for the Canadian Nutrition Society about tips and tricks for interpreting peer-reviewed literature for clinicians. So hang on to your seats readers! This is the start of a my next little mini-series of blog posts.

Now, its possible that you think this sounds like a super boring topic and are planning to skip these next few posts, but I have to tell you that I think this is a really important topic. Before I did my PhD, I never read the methods section of papers because none of it made sense to me. But now that I have my PhD, I START reading articles at the methods section. And only if I agree with what they did, will I read the results, the discussion and maybe the intro.

So these next few posts are going to get into the pieces I am looking for in the methods and provide some context of what I am looking for in this section and why.

Step 1: What methods did the researchers use to conduct the dietary assessment as it validated?

When authors make claims that what someone is eating (or not eating) is associated with a health outcome, I want to know the researchers did a good job of assessing what that person actually eats. I know it seems basic, but if you start reading the methods sections, you might be surprised.

According to this 2021 article, the methods used to assess diet in research are similar to those used in clinical practice. Here is a quick summary of different techniques we use in clinical practice. Research methods include:

1. Food records
2. 24 hour recalls
3. Food frequency questionnaires (FFQ)
4. Screeners

Each one has pros and cons. The National Cancer Institute created a really nice table for comparing the pros and cons of each method. Check it out: here.

Food Records

Food records, which in clinic I often think of as food diaries, are when we ask participants to write down everything they eat and drink over a few days. Evidence suggests that making people do this for more 3-4 days decreases the accuracy of these reports. This is because the participant burden with this method is very high. Appropriate use of food records entails ensuring participants receive training about how to weigh or measure the amount of consumed food. I have often seen this method used as the “gold standard” comparison method for validating other tools.

24 hour recalls

Asks about someone’s intake over the last 24 hours. In research settings, it is ideal to administer multiple 24 hour recalls over non-consecutive, random days. One of the most common tools used to conduct 24 hour recalls is the ASA24, which stands for the automated self administered 24 hour dietary assessment tool. I have used this tool in my research and it is favorite.

The pros of 24 hour recalls is that real-time reporting limits missing data or memory issues. The con is that it doesn’t capture seasonal variability and can have a harder time picking up day to day fluctuations in micronutrient intake. The limitations of this tool can be overcome by conducting a 24 hour recall more than once.

Food Frequency Questionnaires (FFQ)

Food frequency questionnaires ask about how often someone is eating specific foods. Typically, researchers combine foods with similar nutrient profiles into a category to help estimate nutrient intake. Researchers can use FFQs to assess intake or changes in intake over time. They are also are less precise than other methods, and cannot measure absolute nutrient intakes. FFQs can be harder for participants to complete.

I have also used FFQs in my research. The tool I used was the DHQ. You can do a demo on yourself by checking out this website. FFQs often have more than 100 questions and are time consuming. Depending on the type you choose you can ask people how many times they ate [enter food category] in the last month or year.

Screeners

Screening tools are similar to FFQs but are more narrowly focused. For example, a screener can be used to estimate how much calcium or fruits and vegetables someone consumes. They are often quicker and easier to implement than an FFQ, though continue to rely on participant memory. They also need to developed and validated and should be population specific.

How accurate are the different methods?

Researchers have used biomarkers to test the accuracy of diet recall methods. While there aren’t recovery biomarkers for all nutrients, there are recovery biomarkers for:

1. Energy – e.g doubly-labeled water
2. Protein – e.g. 24 hour urine nitrogen excretion
3. Sodium – e.g. 24 hour urine sodium excretion
4. Potassium – e.g. 24 hour urine potassium excretion.

The 24-hour recall is considered the best tool for estimating energy and nutrient intake. However, all methods tend to be associated with nutrient under-reporting. Experiments using doubly-labeled water found that participants under-reported energy intake by as much as 10-50%. However, even biomarkers can be inaccurate and tend to be too expensive and technical to use in most research studies.

Why are their errors in nutrient reporting?

Random errors occur because it is normal for humans to eat differently on different days. Repeat measures can help correct for these types of errors.

Systematic errors are harder to correct for. Systematic errors occur because of:

• Social desirability – there is a general tendency for people to over-report consuming foods perceived as healthy and under-report consuming foods perceived as unhealthy
• Patient Ability – for example some people are better at estimating accurately how much they eat, while some people are worse at it. Food recall methods also tend to rely on patient memory and how the interviewer asks the question can change the way a patient responds.
• The accuracy of food nutrient databases – all methods rely on food nutrient databases to generate the nutrient content data. If the nutrient databases are incorrect, the generated reports will also be incorrect.

Warning: When to stop reading the paper

If you come across an observational study that fails to describe how they collected the diet assessment data or hasn’t used one of these tools above, its probably time for the alarm bells in your head to go off.
If the information collected from the diet recall wasn’t done systematically then there could be errors in the diet information. This increases the possibility that what the researchers found isn’t true.

Take Aways

Researchers and clinicians use similar methods to assess dietary intake. The ones I have used/seen used in my practice are:

• 24 hour recalls
• Food records, typically 3-day food records that are requested to be completed for an initial assessment
• Clinician generated screeners (for example if someone has high phosphorus – dietitians might chart how many times per week someone consumes colas, processed meats and dairy products). I think we more frequently refer to them as food frequency questionnaires, but given the specific nature of the questions, these are more closely aligned with a nutrition screener.

Consider that a well-done 24 hour recall, which is not the same as asking someone what they eat in a typical day, is more accurate than requesting a patient complete a 3-day food diary.

Want to know how to conduct a 24 hour recall really well? Use the ASA24 demo site and do one on yourself! Notice how the questions are phrased and the order they are asked in. Notice the level of detail the tool gets people to recall. If you can start using some of these techniques, then you are on your way to getting a good idea of what your patient’s are actually eating!

And if the paper doesn’t describe how the diet information was collected, you have two choices, email the authors to ask or stop reading the paper. Because as my kids would say, that paper is “sus”.