Premium
Assessing The Impact of The Diet on Cardiometabolic Outcomes: Are Two Consecutive Measures Post‐Intervention Really Necessary?
Author(s) -
Allaire Janie,
Talbot Denis,
Couture Patrick,
Tchernof André,
Jones Peter JH,
Etherton Penny K,
West Sheila,
Connelly Philip W,
Jenkins David JA,
Lamarche Benoît
Publication year - 2017
Publication title -
the faseb journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.709
H-Index - 277
eISSN - 1530-6860
pISSN - 0892-6638
DOI - 10.1096/fasebj.31.1_supplement.lb466
Subject(s) - context (archaeology) , medicine , sample size determination , crossover study , statistical power , psychological intervention , randomized controlled trial , biomarker , statistics , mathematics , paleontology , biochemistry , chemistry , alternative medicine , pathology , psychiatry , biology , placebo
Background Important day‐to‐day variations in the blood concentrations of risk factors may lessen our ability to detect significant effects of any given nutritional intervention on cardiometabolic risk. In theory, relying on several consecutive measurements of a biomarker outcome should decrease variability and hence increase statistical power. However, there is very little empirical evidence supporting such a theory in nutrition research. The purpose of this study was to examine how using the mean of two consecutive measures vs. only one measure post treatment influences the capacity to detect a change in often‐used cardiometabolic risk factors in response to nutritional interventions. Methods We used data from two randomized double‐blind crossover trials that have assessed the impact of docosahexaenoic acid (DHA) on cardiometabolic risk factors, the first one in the context of a supplementation study and the second one in the context of fully‐controlled feeding conditions. We used equations and simulations to compare the effect of using the mean of two consecutive measures vs. only one measure post treatment on the capacity to detect a change in often‐used cardiometabolic risk factors in response to nutritional interventions. Results We found that the reduction in sample size or the gain in statistical power attributed to using the mean of two consecutive measures vs. a single measure is marginal for serum total cholesterol (C), non‐HDL‐C, HDL‐C, LDL‐C, apolipoprotein B100, triglycerides (TG) and C‐reactive protein, even when attempting to assess small effect sizes. For example, the difference in the required sample size to detect a medium effect size (e.g. 0.5) for the change in TG in response to supplemented DHA between using one vs. the mean of two consecutive TG measurements post treatment is 4 subjects (71 vs. 67 subjects respectively). For LDL‐C, this difference is 2 subjects (67 vs. 65 subjects). Conclusions These data indicate that the reduction in sample size and hence the gain in statistical power attributed to using the mean of two measures taken on consecutive days vs. a single measure is small and unjustified to detect a change in cardiometabolic risk factors in response to DHA in the context of crossover nutritional studies. Support or Funding Information JA is a recipient of PhD Scholarships from the Canadian Institutes of Health Research (CIHR) and Fonds de recherche du Québec – Santé (FRQ‐S).