Assessment of compliance with dietary interventions is necessary to understand the observed magnitude of the health effects of the diet per se. To avoid reporting bias, different dietary biomarkers (DBs) could be used instead of self-reported data. However, few studies investigated a combination of DBs to assess compliance and its influence on cardiometabolic risk factors. The objectives of this study were to use a combination of DBs to assess compliance and to investigate how a healthy Nordic diet (ND) influences cardiometabolic risk factors in participants with high apparent compliance compared with the whole study population. From a recently conducted isocaloric randomized trial, SYSDIET (Systems Biology in Controlled Dietary Interventions and Cohort Studies), in 166 individuals with metabolic syndrome, several DBs were assessed to reflect different key components of the ND: canola oil (serum phospholipid α-linolenic acid), fatty fish [eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA)], vegetables (plasma β-carotene), and whole grains (plasma alkylresorcinols). High-fat dairy intake (expectedly low in the ND) was reflected by serum pentadecanoic acid. All participants with biomarker data (n = 154) were included in the analyses. Biomarkers were combined by using a biomarker rank score (DB score) and principal component analysis (PCA). The DB score was then used to assess compliance. During the intervention, median concentrations of alkylresorcinols, α-linolenic acid, EPA, and DHA were >25% higher in the ND individuals than in the controls (P < 0.05), whereas median concentrations of pentadecanoic acid were 14% higher in controls (P < 0.05). Median DB score was 57% higher in the ND than in controls (P < 0.001) during the intervention, and participants were ranked similarly by DB score and PCA score. Overall, estimates of group difference in cardiometabolic effects generally appeared to be greater among compliant participants than in the whole study population (e.g., estimates of treatment effects on blood pressure and lipoproteins were ∼1.5- to 2-fold greater in the most compliant participants), suggesting that poor compliance attenuated the dietary effects. With adequate consideration of their limitations, DB combinations (e.g., DB score) could be useful for assessing compliance in intervention studies investigating cardiometabolic effects of healthy dietary patterns. The study was registered at clinicaltrials.gov as NCT00992641.
Journal of Nutrition, 2014, Vol 144, Issue 10, p. 1642-1649