PMC

Patterns of modelled changes in fatty acids which occur with the increase in eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in (a) plasma and (b) blood cell fatty acid pools. Linear combinations of the co-efficients from the multinomial linear regression models were calculated for each fatty acid important for the change in EPA + DHA. As a small regression coefficient for the fatty acid of interest reflected a large change in the outcomes EPA + DHA and vice versa the data are the reciprocal of the co-efficients in order to portray the magnitude of effect, and this is presented in arbitrary units. Fatty acids are grouped according to class to depict the patterns of change in fatty acids for each pool.

Mean change in key fatty acids in response to 12 months supplementation of EPA + DHA equivalent to 0, 1, 2 or 4 portions of fish per week in different plasma and blood cell pools. The observed mean ± SE change from baseline in the fatty acids identified as important in relation to change in EPA + DHA in the multinomial regression models in (a) Plasma PC; (b) Plasma CE; (c) Plasma TAG; (d) RBC; (e) PLAT; (f) MNC. The effect of dose was tested by linear regression models for each fatty acid. Each model was tested with and without age and sex which were included as covariates if significant. Change in dietary SFA, MUFA, n-3 PUFA or n-6 PUFA where relevant were also tested to determine if change in diet influenced the change in fatty acids with EPA + DHA dose. Significant effects of dose detected in these models are shown as: * p < 0.05; ** p < 0.01; *** p < 0.0001; ^† Effect of dose (p = 0.05) only when taking into account the change in dietary n-6 PUFA from baseline to 12 months; ^‡ Effect of dose (p < 0.05) is no longer significant when taking into account the effect of change in dietary MUFA.