Development and Validation of a Questionnaire to Assess the Level of Salt Intake in the Adult Population of the Russian Federation Using Machine Learning Methods

Abstract

Purpose: To develop and validate a tool to assess salt intake in the adult population of the Russian Federation.Material and Methods: Respondents filled out food diaries, where the type of food intake, its volume, and the fact of additional salting of the dish were taken into account. R language, version 4.2.1, RStudio development environment (packages ggplot2, ggpubr, dplyr, tidyverse, gtsummary, rstatix) were used for statistical processing of the obtained data.Results: A total of 271 respondents were included in the study, with a median age of 52 [20; 70] years. It was found that the main factors for high sodium intake were pre-salting, consumption of salty foods, lower intake in confectionery, low salt intake was characterized by higher consumption of dairy products. Cohen’s consistency test was κ = 0.48 95 % CI (0.08; 0.08), Cronbach’s alpha values α = 0.8. At a threshold score of ≥12 points on the questionnaire, the questionnaire had a sensitivity of 85 % compared with the median score from the 3-day food diary data. At a threshold score < 12 points, the questionnaire has a specificity of 74 % compared with the median score from a 3-day food diary.Conclusion: The SOLE questionnaire can be used to determine the estimated level of salt intake by the population, but for wider application in the territory of the Russian Federation additional validation by regions is required.

Keywords

• salt consumption
• food questionnaire
• sodium consumption
• machine learning