Assessing adherence to healthy and sustainable diets: a measurement performance comparison of indices based on EAT-Lancet diet in French adults
Résumé
Measuring adherence to EAT-Lancet recommendations for healthy and sustainable diets is challenging, leading to diverse methods and a lack of consensus on standardized metrics. Available indices vary mainly in scoring systems, food components, units, energy adjustments, and cut-off points. This study aimed to evaluate and compare the measurement performance of six dietary indices for assessing adherence to the EAT-Lancet reference diet (WISH, PHDI, ELD-I, ELI, HSDI and ELDS). Food consumption data of 1,723 adults were obtained from the French Third Individual and National Study on Food Consumption Survey (INCA3, 2014-2015). Sociodemographic, nutritional and environmental data were used to assess the validity and reliability of dietary indices. Results showed that indices assessing their food components with quantitative scoring (WISH, PHDI and ELD-I) captured dietary variability, were less dependent on energy intake (negligible correlation), have higher reliability and converged to a large extent with nutritional indicators. While binary indices (HSDI and ELDS) showed a stronger correlation with environmental indicators, ELD-I converged with both domains. Higher scores were associated with higher nutritional quality and lower environmental impact, but with unfavorable results for zinc, vitamin B12 and water use. Furthermore, the indices differed between sociodemographic groups, demonstrating concurrent criterion validity. ANCOVA models showed that being women, age, income, education, physical activity, healthy body mass index and non-smoking were associated positively with the indices (p<0.05). A low concordance rate indicated that indices categorized individuals differently. Regarding this, total concordance (i.e., individuals classified in the same quantile) was below 50% and kappa coefficients were <0.20. Moreover, between 23–28% of the participants were classified in adjacent quantile and 1–7.5% in the opposite extreme quantile. Researchers must align study objectives with the applicability, assumptions, and functional significance of chosen indices. Indices employing quantitative scoring provide a comprehensive understanding of the healthiness and sustainability of diets, being advantageous in precision-focused research, such as clinical trials or epidemiological research. Conversely, indices based on binary scoring offer a simplified perspective, being valuable tools for surveys, observational studies, and public health. Recognizing their strengths and limitations is crucial for a comprehensive assessment of diets and understanding their implications.