Error estimation of energy intake measurement: validation of the Brazilian version of GloboDiet software

Author(s)

Jessica Levy

• Document type: Doctoral Thesis
• Press: São Paulo.
• Institution: Universidade de São Paulo (USP). Faculdade de Saúde Pública (FSP/CIR)
• Defense date: 2022-08-12

Examining board members

Dirce Maria Lobo Marchioni; Marina Campos Araujo; Severina Carla Vieira Cunha Lima; Maria Laura da Costa Louzada

• Advisor: Dirce Maria Lobo Marchioni

Abstract

Introduction: All existing methods for evaluating diet are subject to measurement errors. One of the main and most recurrent errors in diet measurement is the implausible reporting of energy intake. Although several procedures are used to estimate the underreporting of energy intake (EI), the most accurate is doubly labeled water (DLW), which measures total energy expenditure in individuals outside of confinement and is used as the gold standard to validate the reported EI. Objective: Validate the Brazilian version of the GloboDiet software, computerized R24h, regarding energy intake.Methodology: The study was conducted at the Clinical and Epidemiological Research Center at the University Hospital at University of São Paulo that hosts the Longitudinal Study of Adult Health (ELSA-Brasil) in São Paulo with 101 participants. Two 24-hour food recalls (R24h) were collected on non-consecutive days and the data entry was performed on GloboDiet, simultaneously with the interview. GloboDiet is a standardized and computerized methodology that allows conducting the R24h interview, guided by a standardized script to reduce errors and ensure high standardization in the collection. Participants ingested doses of DLW to assess total energy expenditure (TEE); answered a general characterization questionnaire, which included questions about individual characteristics such as lifestyle habits and socioeconomic status, and the level of physical activity was classified according to the short version of the International Physical Activity Questionnaire (IPAQ). Anthropometric measurements of weight, heightand waist circumference in duplicate were performed with the participants before ingesting the DLW doses and on the last day of collection. The implausible report was calculated with the DLW, but also with the predictive equations proposed by Goldberg (1991), Black (2000) and McCrory (2002). Results: The EI average was 2117Kcal while the TEE average was 2577Kcal. The underestimation of EI evaluated by GloboDiet in relation to DLW was 16.1% for the total population; we observed greater underestimation of intake in men (16.6%), in the elderly (20.2%), in overweight people (22.5%) and with moderate to vigorous physical activity (19.7%). The mean difference between EI and TEE was 460kcal at the population level and the limits of agreement were wide (upper limit of agreement of 1875kcal and lower limit of agreement of -2795kcal), with high dispersion among individuals. The EI of the first measurement of the R24h was higher when compared to the EI of the second measurement of the R24h in the total population, this difference being significant. We also note that the EI of the first measurement of the R24h is closer to the TEE obtained by the ADM. Regarding the classification of participants regarding the report of the EI compared to the DLW, 88.9% of the participants were considered as plausible reporters, with 9.1% being under-reporters and 2% over-reporters. Using the predictive equations to classify the participants in terms of the EI report, 87.9% were plausible reporters, 10.1% under-reporters and 2% over-reporters with the Goldberg equation. Using Black\'s equation, 83.8% were plausible reporters against 15.2% underreporters and 1% over-reporters. In the case of McCrory\'s equation (2002), 97% were considered plausible reporters, 1% under-reporters and 2% over-reporters. Conclusion: The Brazilian version of the GloboDiet software is an accurate method to assess EI at the population level. (AU)

• FAPESP's process: 18/08319-0 - Estimation of the error of energy intake measurement: validation of the Brazilian version of GloboDiet software
• Grantee: Jéssica Levy
• Support Opportunities: Scholarships in Brazil - Doctorate (Direct)