A mediation analysis on the relationship between dietary glycemic load, obesity and cardiovascular risk factors in children

• Published in: International Journal of Obesity Vol. 46; no. 4; pp. 774 - 781
• Main Authors: Suissa, Karine, Benedetti, Andrea, Henderson, Mélanie, Gray-Donald, Katherine, Paradis, Gilles
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 01.04.2022; Nature Publishing Group
• Subjects: 692/499; 692/699/75; Adipose tissue; Adiposity; Adolescent; Blood; Body fat; Body Mass Index; Body size; Body weight; Cardiovascular diseases; Cardiovascular Diseases - epidemiology; Cardiovascular Diseases - etiology; Child; Children; Epidemiology; Glycemic Index; Glycemic Load; Health Promotion and Disease Prevention; Health risks; Heart Disease Risk Factors; High density lipoprotein; Humans; Internal Medicine; Lipids; Mediation Analysis; Medicine; Medicine & Public Health; Metabolic Diseases; Obesity; Obesity - complications; Obesity - epidemiology; Overweight; Overweight - complications; Parents; Public Health; Risk analysis; Risk Factors; Structural models; Youth
• ISSN: 0307-0565; 1476-5497; 1476-5497
• DOI: 10.1038/s41366-021-00958-4

Background/Objective Adiposity may mediate the effect of dietary glycemic load (GL) on lipid profiles in children, as studies have shown an association between dietary GL and adiposity and between adiposity and lipid profiles. Our objective was to evaluate the role of adiposity as a mediator in the association between dietary GL and lipid profiles after 2 years. Subjects/Methods The Quebec Adipose and Lifestyle InvesTigation in Youth study included 630 children, 8–10 years old at recruitment with at least one parent with overweight or obesity with 2-year follow-up. Three baseline 24-h dietary recalls were administered by a dietitian at baseline. Child and parent characteristics were obtained through direct measurement (blood lipids, anthropometrics) or questionnaires (socio-economic characteristics). Indicators of adiposity, including body mass index (BMI) z-score and percent body fat, were the mediators of interest. A conventional approach using the Baron and Kenny method was used. A causal approach using marginal structural models (MSM) was used to estimate the controlled direct effect. Results Mean age at baseline was 9.6 years and 33% were overweight or obese. Both methods revealed that the effect of GL on blood lipids was mediated by adiposity. The weighted MSM did not show evidence of a direct effect (TG: β  =;0.01, 95% CI = −0.01,0.02; HDL: β  = 0.005, 95%CI = −0.002,0.01), whereas the conventional method did for TG but not HDL (TG: β  = 0.04, 95%CI = 0.01,0.07; HDL: β  = −0.01, 95%CI = −0.03,0.01). Conclusion Adiposity contributes substantially to the association between GL and blood lipids. The choice of approach for mediation analysis should be based on the fulfilment of conditions of each method.