# Development, Reproducibility, and Validity of a Semiquantitative Food Frequency Questionnaire for Use among the Adult Population in Reunion Island

## TL;DR

Researchers developed and tested a food frequency questionnaire for Reunion Island adults, finding it reliable and valid for assessing dietary habits.

## Contribution

The study introduces a locally adapted food frequency questionnaire for Reunion Island with validated reproducibility and validity.

## Key findings

- The CARI FFQ showed moderate to good validity for ranking food and nutrient intakes.
- Reproducibility was good, with low rates of gross misclassification.
- Most participants were correctly classified for nutrients and food groups.

## Abstract

To date, there is no specific food frequency questionnaire (FFQ) to assess dietary habits in any of the French overseas regions.

This study aimed to describe the development of a culture and context-specific semiquantitative FFQ for use among the adult population in Reunion Island, to assess its reproducibility and its relative validity compared with 24-h dietary recalls (DRs).

The CARI (observatoire des Comportements Alimentaires à la RéunIon) FFQ was adapted from the NutriNet-Santé FFQ, with a revised list of 181 food and beverage items reflecting local dietary practices and nutritional concerns. To assess its reproducibility, the FFQ was administered twice, 4 wk apart, to a purposive sample of 108 adults in Reunion Island. During the same period, participants also completed a weekly DR to evaluate the FFQ’s validity. Reproducibility and validity were assessed for 18 food groups, energy, and 38 nutrients using correlations, cross-classification, and weighted κ.

Regarding reproducibility, we found a median rank correlation of 0.56 (nutrients) and 0.64 (food groups). Most participants were correctly classified with a median of 78% (nutrients) and 83% (food group), whereas gross misclassification was low (2.8% and 0.9%, respectively). We found a median weighted κ of 0.44 (nutrients) and 0.47 (food groups). Regarding validity, we found a median crude rank correlation of 0.51 (nutrients) and 0.43 (food groups). Most participants were correctly classified with a median agreement of 71% for nutrients and of 68% for food groups, whereas gross misclassification was low (1.9% and 0.9%, respectively). We found a median weighted κ of 0.32 (nutrients) and 0.27 (food groups).

The CARI FFQ has a moderate to good level of validity for ranking food and nutrient intakes, and a good level of reproducibility, supporting its use for dietary assessment in Reunion Island.