Development and Reliability of a Short Food Frequency Questionnaire to Assess Intake of Non-Essential Energy-Dense Nutritionally-Deficient (NEEDNT) Food Items

Abstract

Background: Obesity is a large and growing problem in the world today due to its association with non-communicable disease. Dietary energy density is an important contributor to obesity. People who are overweight are more likely than people of a healthy weight to be consuming an energy dense diet. Energy dense food often has a poor nutritional content highlighting the importance of nutrient as well as energy density. Energy dense nutrient poor food is not required in the diet to maintain good health. There is currently no quick, cost effective tool to measure individual intakes of energy dense nutrient poor food. The NEEDNT (Non-Essential, Energy-Dense, Nutritionally-Deficient) Food List was developed as a clinical tool to aid in weight management. It is a comprehensive list containing energy dense nutrient poor foods that should be avoided, with lower energy dense alternatives provided where possible. A Food Frequency Questionnaire (FFQ) based on this list could be used in research and clinical settings to assess individual intakes of energy dense nutrient poor foods. Objective: The aim of this study was to develop and assess the reliability of a short FFQ based on the newly developed NEEDNT Food List. Design: This was a cross sectional test re-test observational study. Participants completed the Non-Essential Energy-Dense Nutritionally-Deficient Food Frequency Questionnaire (NEEDNT-FFQ) on two separate occasions, 7-10 days apart. Participants were required to be obese (BMI above 30kg/m²) and 18-65years of age. Results: 13 men and 41 women completed the study. Participants had a BMI range of 30.1-54 kg/m², and age range of 21-65 years. Eight out of 48 FFQ items were consumed in the highest frequency category of 3+ times per day, with sugar, butter and whole milk being the three highest. BMI was significantly positively associated with total score in NEEDNT-FFQ time one, even after adjustment for age and education. The intraclass correlation coefficient (ICC) for total score between NEEDNT-FFQ time one and two was 0.83, and the Spearman’s signed-rank correlation coefficient (SCC) was 0.77. The 48 food items in the NEEDNT-FFQ, ICCs ranged from -0.01 for ‘fruit flavoured roll ups, sticks and straps’, to 0.97 for ‘regular luncheon sausage’, and SCCs ranged from -0.02 for ‘fruit flavoured roll ups, sticks and straps’ to 0.94 for ‘alcoholic drinks’. All SCCs were statistically significant except ‘fruit flavoured roll ups, sticks and straps’. ‘Glucose’ could not be correlated as all participants consumed this item at a frequency of never or less than once per month. Cronbach’s alpha scores (internal consistency) were 0.82 and 0.85 for NEEDNT-FFQ time one and two respectively. The number of items correctly classified ranged from 50.0-100.0% (median 75.0%); correctly and adjacently classified ranged from 81.5-100.0% (median 98.1%), and beyond adjacent classification ranged from 0-18.5% (median 1.9%). Conclusion: The newly developed NEEDNT-FFQ is a reliable tool to assess energy dense nutrient poor food intake in obese New Zealand adults. Once validated, this tool will be valuable in both research and clinical settings.