What to ask in a self-administered dietary assessment website: The role of professional judgement

Evaluation of the dietary intake data coding process in a clinical setting: Implications for research practice

BACKGROUND

High quality dietary intake data is required to support evidence of diet-disease relationships exposed in clinical research. Source data verification may be a useful quality assurance method in this setting. The present pilot study aimed to apply source data verification to evaluate the quality of the data coding process for dietary intake in a clinical trial and to explore potential barriers to data quality in this setting.

METHODS

Using a sample of 20 cases from a clinical trial, source data verification was conducted between three sets of data derived documents: transcripts of audio-recorded diet history interviews, matched paper-based diet history forms and outputs from nutrition analysis software. The number of cases and rates of discrepancies between documents were calculated. A total of five in-depth interviews with dietitians collecting and coding dietary data were thematically analysed.

RESULTS

Some 2024 discrepancies were identified. The highest discrepancy rate was 57.49%, and occurred between diet history interviews and nutrition analysis software outputs. Sources of the discrepancies included both quantities and frequencies of food intake. The highest discrepancy rate was for the food group "vegetable products and dishes". In-depth interviews implicated recall bias of trial participants as a cause of discrepancies, but dietitians also acknowledged a possible subconscious influence of having to code reported foods into nutrition analysis software programs.

CONCLUSION

The accuracy of dietary intake data appeared to depend on the level of detailed food data required. More support for participants on reporting consumption, and incorporating supportive tools to guide estimates of food quantities may facilitate a more consistent coding process and improve data quality. This pilot study offers a novel method and an overview of dietary intake data coding measurement errors. These findings may warrant further investigation in a larger sample.

Dietary Patterns and Cardiovascular Disease: Insights and Challenges for Considering Food Groups and Nutrient Sources

PURPOSE OF REVIEW

The relationship between dietary patterns and cardiovascular disease has been the subject of much research, but an important methodological consideration is the interdependence between the nutrient composition of foods and the recognition of healthy dietary patterns. This review considers some of the challenges in researching dietary patterns with implications for translation to public health promotions.

RECENT FINDINGS

A number of statistical methods have emerged for analysing dietary patterns using population dietary data. There are limitations in the assumptions underpinning food categorisation, but this research is able to consistently identify foods and dietary patterns that are positively related to health. Aligned to this activity is the ongoing development of food composition databases which has its own limitations such as keeping up to date with changing foods and newly identified components, sampling of foods, and developments in chemical analytical methods. Finally, dietary patterns form the basis for current dietary guidelines and related public health-oriented programs, but the issues raised for research (e.g. food categorisation and cuisine influences on dietary patterns) can also translate to these settings. The study of dietary patterns in cardiovascular disease prevention presents with a number of methodological challenges relating to the way food groups are formed and the limitations of food composition databases. Added to this are new considerations for the environmental impact of recommended dietary patterns. Future research across the entire knowledge chain should target more accurate methods in a number of analytical areas.

A Tool to Measure Young Adults' Food Intake: Design and Development of an Australian Database of Foods for the Eat and Track Smartphone App

Background

Dietary assessment is reliant on the collection of accurate food and beverage consumption data. Technology has been harnessed to standardize recording and provide automatic nutritional analysis to reduce cost and researcher burden.

Objective

To better assess the diet of young adults, especially relating to the contribution of foods prepared outside the home, a database was needed to support a mobile phone data collection app. The app also required usability testing to assure ease of entry of foods and beverages. This paper describes the development of the Eat and Track app (EaT app) and the database underpinning it.

Methods

The Australian Food and Nutrient Database 2011-13, consisting of 5740 food items was modified. Four steps were undertaken: (1) foods not consumed by young adults were removed, (2) nutritionally similar foods were merged, (3) foods available from the 30 largest ready-to-eat food chains in Australia were added, and (4) long generic food names were shortened and simplified. This database was used to underpin the EaT app. Qualitative, iterative usability testing of the EaT app was conducted in three phases using the “Think Aloud” method. Responses were sorted and coded using content analysis. The System Usability Scale (SUS) was administered to measure the EaT app’s perceived usability.

Results

In total, 1694 (29.51%) foods were removed from the Australian Food and Nutrient Database, including 608 (35.89%) ingredients, 81 (4.78%) foods already captured in the fast food chain information, 52 (3.07%) indigenous foods, 25 (1.48%) nutrients/dietary supplements, and 16 (0.94%) child-specific foods. The remaining 912 (53.84%) foods removed were not consumed by young adults in previous surveys or were “not defined” in the Australian Food and Nutrient Database. Another 220 (3.83%) nutritionally similar foods were combined. The final database consisted of 6274 foods. Fifteen participants completed usability testing. Issues identified by participants fell under six themes: keywords for searching, history list of entered foods, amounts and units, the keypad, food names, and search function. Suggestions for improvement were collected, incorporated, and tested in each iteration of the app. The SUS of the final version of the EaT app was rated 69.

Conclusions

A food and beverage database has been developed to underpin the EaT app, enabling data collection on the eating-out habits of 18- to 30-year-old Australians. The development process has resulted in a database with commonly used food names, extensive coverage of foods from ready-to-eat chains, and commonly eaten portion sizes. Feedback from app usability testing led to enhanced keyword searching and the addition of functions to enhance usability such as adding brief instructional screens. There is potential for the features of the EaT app to facilitate the collection of more accurate dietary intake data. The database and the app will be valuable dietary assessment resources for researchers.

Development of a Web-Based 24-h Dietary Recall for a French-Canadian Population

Twenty-four-hour dietary recalls can provide high-quality dietary intake data, but are considered expensive, as they rely on trained professionals for both their administration and coding. The objective of this study was to develop an automated, self-administered web-based 24-h recall (R24W) for a French-Canadian population. The development of R24W was inspired by the United States Department of Agriculture (USDA) Automated Multiple-Pass Method. Questions about the context of meals/snacks were included. Toppings, sauces and spices frequently added to each food/dish were suggested systematically. A list of frequently forgotten food was also suggested. An interactive summary allows the respondent to track the progress of the questionnaire and to modify or remove food as needed. The R24W prototype was pre-tested for usability and functionality in a convenience sample of 29 subjects between the ages of 23 and 65 years, who had to complete one recall, as well as a satisfaction questionnaire. R24W includes a list of 2865 food items, distributed into 16 categories and 98 subcategories. A total of 687 recipes were created for mixed dishes, including 336 ethnic recipes. Pictures of food items illustrate up to eight servings per food item. The pre-test demonstrated that R24W is easy to complete and to understand. This new dietary assessment tool is a simple and inexpensive tool that will facilitate diet assessment of individuals in large-scale studies, but validation studies are needed prior to the utilization of the R24W.

First-Stage Development and Validation of a Web-Based Automated Dietary Modeling Tool: Using Constraint Optimization Techniques to Streamline Food Group and Macronutrient Focused Dietary Prescriptions for Clinical Trials

BACKGROUND

Standardizing the background diet of participants during a dietary randomized controlled trial is vital to trial outcomes. For this process, dietary modeling based on food groups and their target servings is employed via a dietary prescription before an intervention, often using a manual process. Partial automation has employed the use of linear programming. Validity of the modeling approach is critical to allow trial outcomes to be translated to practice.

OBJECTIVE

This paper describes the first-stage development of a tool to automatically perform dietary modeling using food group and macronutrient requirements as a test case. The Dietary Modeling Tool (DMT) was then compared with existing approaches to dietary modeling (manual and partially automated), which were previously available to dietitians working within a dietary intervention trial.

METHODS

Constraint optimization techniques were implemented to determine whether nonlinear constraints are best suited to the development of the automated dietary modeling tool using food composition and food consumption data. Dietary models were produced and compared with a manual Microsoft Excel calculator, a partially automated Excel Solver approach, and the automated DMT that was developed.

RESULTS

The web-based DMT was produced using nonlinear constraint optimization, incorporating estimated energy requirement calculations, nutrition guidance systems, and the flexibility to amend food group targets for individuals. Percentage differences between modeling tools revealed similar results for the macronutrients. Polyunsaturated fatty acids and monounsaturated fatty acids showed greater variation between tools (practically equating to a 2-teaspoon difference), although it was not considered clinically significant when the whole diet, as opposed to targeted nutrients or energy requirements, were being addressed.

CONCLUSIONS

Automated modeling tools can streamline the modeling process for dietary intervention trials ensuring consistency of the background diets, although appropriate constraints must be used in their development to achieve desired results. The DMT was found to be a valid automated tool producing similar results to tools with less automation. The results of this study suggest interchangeability of the modeling approaches used, although implementation should reflect the requirements of the dietary intervention trial in which it is used.

Challenges in converting an interviewer-administered food probe database to self-administration in the National Cancer Institute Automated Self-administered 24-Hour Recall (ASA24)

The National Cancer Institute (NCI) is developing an automated, self-administered 24-hour dietary recall (ASA24) application to collect and code dietary intake data. The goal of the ASA24 development is to create a web-based dietary interview based on the US Department of Agriculture (USDA) Automated Multiple Pass Method (AMPM) instrument currently used in the National Health and Nutrition Examination Survey (NHANES). The ASA24 food list, detail probes, and portion probes were drawn from the AMPM instrument; portion-size pictures from Baylor College of Medicine's Food Intake Recording Software System (FIRSSt) were added; and the food code/portion code assignments were linked to the USDA Food and Nutrient Database for Dietary Studies (FNDDS). The requirements that the interview be self-administered and fully auto-coded presented several challenges as the AMPM probes and responses were linked with the FNDDS food codes and portion pictures. This linking was accomplished through a "food pathway," or the sequence of steps that leads from a respondent's initial food selection, through the AMPM probes and portion pictures, to the point at which a food code and gram weight portion size are assigned. The ASA24 interview database that accomplishes this contains more than 1,100 food probes and more than 2 million food pathways and will include about 10,000 pictures of individual foods depicting up to 8 portion sizes per food. The ASA24 will make the administration of multiple days of recalls in large-scale studies economical and feasible.

Computerized dietary assessments compare well with interviewer administered diet histories for patients with type 2 diabetes mellitus in the primary healthcare setting

OBJECTIVE

To test repeatability and relative validity of a computerized and interviewer administered assessment.

METHODS

Using a context-based case-control trial, 41 adults with type 2 diabetes mellitus were randomized into four groups to complete dietary assessments (computerized or interviewer administered) at 0, 2 and 8 weeks and food records at 0 and 2 weeks. Repeatability of reported energy, total fat, saturated, polyunsaturated and monounsaturated fatty acids between the computerized and interviewer administered methods were assessed using repeated measures ANOVA. Paired t-tests and Pearson's correlations determined relative validity of the assessments.

RESULTS

Thirty-one patients completed all visits. Statistically significant differences were found between computerized and interviewer administered data for total fat (p=0.048) and saturated fatty acids (p=0.019) between 0 and 2 weeks. Computerized assessments correlated better with food records (r=0.16-0.52) compared with interviewer administered assessments (r=-0.02 to 0.51).

CONCLUSION

Computerized assessments saw a learning effect with repeated use indicating that users were becoming more familiar with the website with repeated use. Relative validity suggests that the website may capture more foods though this requires further investigation.

PRACTICE IMPLICATIONS

By allowing patients to self-report their intakes on a computer, dietitians will have the ability to spend increased time with their patients counseling them toward change.