Кардиоваскулярная терапия и профилактика (Mar 2022)

Development, validation and assessment of reproducibility of a modern version of semi-quantitative food frequency questionnaire for the adult population

  • N. S. Karamnova,
  • S. A. Maksimov,
  • S. A. Shalnova,
  • O. B. Shvabskaia,
  • O. V. Izmailova,
  • E. I. Ivanova,
  • O. V. Molchanova,
  • A. V. Kontsevaya,
  • O. M. Drapkina

DOI
https://doi.org/10.15829/1728-8800-2022-3169
Journal volume & issue
Vol. 21, no. 3

Abstract

Read online

The changing structure of food consumption requires updating the nutrition assessment tool, taking into account modern dietary patterns.Aim. To develop, evaluate the validity and reproducibility of semiquantitative food frequency questionnaire (FFQ) for the adult population.Material and methods. The FFQ questionnaire was developed as part of the Russian part of the international Prospective Urban and Rural Epidemiological (PURE) study. To compare the quantitative parameters of the energy and nutritional value of diet, the validation used data on the actual nutrition of 294 men and women aged 25-65 years, collected by 2 methods: 24-hour dietary recall (24hDR) — 4 surveys per year (1 time per season) and the developed FFQ — 2 surveys: initially (FFQ1) and after 12 months (FFQ2).Results. The minimum mean values for the vast majority of nutrients were recorded by the 24hDR method, while higher values — by FFQ2, and maximum values — by FFQ1. Pearson’s correlation coefficients ranged from 0,19 (retinol equivalent) to 0,53 (cholesterol), showing a moderate relationship between the two methods. After crossclassification, the proportion of subjects assigned to the same quartile of 24hDR and FFQ2 nutrient profile ranged from 26,9% (β-carotene) to 43,5% (cholesterol), and those assigned to opposite quartiles ranged from 3,1% (cholesterol) to 11,2% (retinol equivalent), indicating good agreement between these methods. The Bland-Altman plots showed slightly overestimated FFQ2 scores for total carbohydrate, protein, and dietary calories. The Pearson correlation between FFQ1 and FFQ2 values ranged from 0,46 to 0,82. The intraclass correlation showed the questionnaire reproducibility coefficients lower than the Pearson correlation. However, most of them remained at a value of >0,60.Conclusion. The results showed predominantly moderate validity and good reproducibility of the modern FFQ version, which allows it to be used to assess nutrition in the adult population with estimation of the energy and nutritional value of diet in the protocols of Russian epidemiological, preventive and clinical studies.

Keywords