Associations of Global Burden of Diseases study-derived dietary scores with mortality and chronic disease risk: a comprehensive analysis from the prospective NutriNet-Santé study

International audience

anglais. The Global Burden of Diseases (GBD) network has proposed theoretical minimum risk exposure level (TMREL) for leading risk factors associated with diet that minimize the risk of morbimortality from chronic diseases. TMREL can be applied to develop follow-up or evaluation indicators in individual studies. The validity of these scores can be tested by assessing associations with health outcomes in prospective cohorts. In this study conducted within the NutriNet-Santé cohort, four dietary scores (TMREL-Risk Score, TMREL-Probability of adequacy, TMREL-standardized distance, and TMREL dietary score) using different scoring methods were developed, with higher scores reflecting less healthy diets. Associations of these scores with the risk of type 2 diabetes, cancer, cardiovascular diseases (CVD) and mortality were estimated using multivariable Cox proportional hazards models, adjusted for a wide range of covariates. Counterfactual and marginal structural models were used to infer causality. Analyses were conducted in a sample of up to103,324 participants (78.3% women, mean age of 43.6 years old (y) (SD=14.6)), followed for a median of 8.47 (IQR=14.7) years (2009-2024). The association with dietary scores (for 1SD-increase) varied in magnitude for each health outcome. For mortality, HR varied from 1.12 (95%CI =1.07-1.18, ) to 1.18 (95%CI =1.12-1.24) for TMRELStdis and TMREL-DI, for overall cancer from 1.07 (95%CI =1.03-1.12) to 1.09 (1.04-1.13) for TMREL-RS and TMREL-PA, for CVD from 1.07 (95%CI =1.00-1.16) to 1.12 (95%CI =1.04-1.20) for TMREL-PA and TMREL-RS, and for type 2 diabetes from 1.33 (95%CI =1.23-1.43) to 1.47 (95%CI =1.36-1.59) for TMREL-DI and TMREL-PA. Marginal structural Cox models strengthened all associations compared to classical analyses. Standardized survival curves showed clear associations, especially for the risk of cancer and type 2 diabetes. Dietary scores based on GBD TMREL can serve as key indicators for characterizing diet quality in relation to long-term health, and using different scoring systems helped evaluate the robustness of these associations.