Development of a sustainable diet index in US adults

Background A transformation towards healthy diets through a sustainable food system is essential to enhance both human and planet health. Development of a valid, multidimensional, quantitative index of a sustainable diet would allow monitoring progress in the US population. We evaluated the content and construct validity of a sustainable diet index for US adults (SDI-US) based on data collected at the individual level. Methods The SDI-US, adapted from the SDI validated in the French population, was developed using data on US adults aged 20 years and older from the National Health and Nutrition Examination Survey, 2007–2018 (n = 25,543). The index consisted of 4 sub-indices, made up of 12 indicators, corresponding to 4 dimensions of sustainable diets (nutritional quality, environmental impacts, affordability (economic), and ready-made product use behaviors (sociocultural)). A higher SDI-US score indicates greater alignment with sustainable diets (range: 4–20). Validation analyses were performed, including the assessment of the relevance of each indicator, correlations between individual indicators, sub-indices, and total SDI-US, differences in scores between sociodemographic subgroups, and associations with selected food groups in dietary guidelines, the alternative Mediterranean diet (aMed) score, and the EAT-Lancet diet score. Results Total SDI-US mean was 13.1 (standard error 0.04). The correlation between SDI-US and sub-indices ranged from 0.39 for the environmental sub-index to 0.61 for the economic sub-index (Pearson Correlation coefficient). The correlation between a modified SDI-US after removing each sub-index and the SDI-US ranged from 0.83 to 0.93. aMed scores and EAT-Lancet diet scores were significantly higher among adults in the highest SDI-US quintile compared to the lowest quintile (aMed: 4.6 vs. 3.2; EAT-Lancet diet score: 9.9 vs. 8.7 p < .0001 for both). Conclusions Overall, content and construct validity of the SDI-US were acceptable. The SDI-US reflected the key features of sustainable diets by integrating four sub-indices, comparable to the SDI-France. The SDI-US can be used to assess alignment with sustainable diets in the US. Continued monitoring of US adults’ diets using the SDI-US could help improve dietary sustainability. Supplementary Information The online version contains supplementary material available at 10.1186/s12937-024-00943-3.

Differences between the SDI-France and SDI-US are summarized in Supplementary Table 1.Briefly, the main differences are in 1) the number of indicators (7 for the SDI-France vs. 12 for the SDI-US with an expanded number of environmental and sociocultural indicators); and 2) the lack of assessment of some measures (e.g., places of food purchase and dietary energy intake) due to data quality or availability.Specifically, while the SDI-France used the PANDiet index to both assess the probability that 24 nutrient intakes met the required level (adequacy) and the probability that 24 nutrient intakes did not exceed the recommended level (moderate) [1], the SDI-US used two indicators, Nutrient-Rich Foods (NRF) 9.3 index and mean nutrient adequacy ratio (MAR), to better represent the dietary diversity and nutrient deficiency level of US adults.Measures related to dietary energy intake were not included in the SDI-US due to substantial underreporting issues in any self-reported dietary assessment method, including the 24-hour dietary recall [3].Unlike other nutrients, underreporting issues may be more critical in total energy intake.Since total energy intake is contained in almost every food and beverage, summing errors in each individual food and beverage would lead to errors in total energy intake to a greater extent than for other nutrients.Thus, it is recommended not to use self-reported energy intake as a true energy intake, but it can be used for adjustment purposes [4].In addition, SDI-US did not include a measure of places of food purchase since this information is only available at the individual food level and thus cannot be combined like a total intake value (e.g., if someone ate 6 foods a day at different places, this person could have 6 different places where foods were obtained).
[3] Archer E., Hand G. A., Blair S. N., Validity of U.S. nutritional surveillance:National Health and Nutrition Examination Survey caloric energy intake data , 1971-2010, PLoS One 8   (10) (2013) e76632.[4] Poslusna K., Ruprich J., de Vries J. H., Jakubikova M., van't Veer P., Misreporting of energy and micronutrient intake estimated by food records and 24 hour recalls, control and adjustment methods in practice, Br J Nutr 101 Suppl 2 (2009) S73-85.Note: mSDI-US-1 was calculated using the Healthy Eating Index-2015 instead of using Nutrient-Rich Foods9.3 index because better nutritional composition does not necessarily translate into an overall healthier diet; mSDI-US-2 was calculated using 5 environmental indicators after excluding one water-related indicator (freshwater use) to consider the possible effect of double counting on water in relation to the other indicators; mSDI-US-3 was calculated by additionally including food security and food price levels to the economic sub-index to better represent food affordability using NHANES 2011-2018 due to data availability; mSDI-US-4 was calculated by additionally including eating together with family or friends to the sociocultural sub-index because higher frequency of ready-to-eat meals may not necessarily indicate negative sociocultural practices related to social exchange or trying diverse recipes if people are eating together using NHANES 2007-2010 due to data availability.

Supplementary Table 1. Comparison between the sustainable diet index-France and the sustainable diet index-US
a The proportions of agreement and the weighted kappa coefficients were obtained from a cross-classification analysis to evaluate the agreement of classification into quintiles between the original SDI-US and the modified SDI-US.The correct classification indicated that participants were classified into the same quintile groups (e.g., original SDI-US quintile 5 and the modified SDI-US quintile 5), and the gross misclassification indicated that participants were classified into the opposite quintile groups (e.g., original SDI-US quintile 5 but the modified SDI-US quintile 1).b The Pearson correlation coefficients of the original SDI-US with the modified SDI-US were estimated using the linear regression model.
SDI-US environmental indicators, included water footprint and nitrogen footprint of foods calculated from a comprehensive database based on a meta-analysis of 1530 publications [2], in addition to carbon footprint and land use.And, SDI-US sociocultural indicators included more specific food practice indicators related to ready-to-eat products use instead of just one indicator in the French original version [1].There is no need for adaptation for the economic indicator (share of food budget).

Table 2 . Demographic characteristics by NHANES cycle, US adults, NHANES 2007-2018 (n=25,543)
Source: Kramer GF, Tyszler M, Veer PV, Blonk H. Decreasing the overall environmental impact of the Dutch diet: how to find healthy and sustainable diets with limited changes.Public Health Nutr 2017;20:1699-709.d "weak data" indicates that data is available, but its quality is not sufficient to use (underreporting of dietary energy intake is a major challenge in any self-reported dietary assessment method).Sample sizes are unweighted and all other estimates are weighted (dietary day 1 weights).b "Other" includes race/Hispanic origin other than non-Hispanic white, non-Hispanic black, and Hispanic, including multiracial.
b Source: Verger EO, Mariotti F, Holmes BA, Paineau D, Huneau JF.Evaluation of a diet quality index based on the probability of adequate nutrient intake (PANDiet) using national French and US dietary surveys.PLoS One 2012;7:e42155.cAbbreviations: NHANES, National Health and Nutrition Examination Survey; GED, general equivalency diploma.a