This dataset comprises data extracted 29 publications which were included in a systematic review and meta-analysis of randomised controlled trials which compared whole grain versus refined grain dietary intake on cardiovascular disease risk factors. There were 80 cardiovascular disease risk factor outcome measures considered in the review. Data were extracted by one investigator and checked for accuracy twice. Study characteristics were extracted, including study design, sample population, and details of the intervention and control conditions. Outcome data were extracted for the intervention and comparator groups were baseline, change, and follow-up variables, variable units, and statistical significance over time and between groups. Adverse events were also recorded. The 29 included studies measured and reported on 40 outcomes. Extracted outcomes can be grouped as haemodynamics (reported by 12 studies), body composition (15 studies), blood lipids (18 studies), glycaemic and/or insulin markers (19 studies), inflammatory markers (21 studies), oxidative stress markers (6 studies), and cardiovascular comorbidity incidence (2 studies). Rigorous analysis, including meta-regression, may be carried out using this data to evaluate the cause and effect relationship of whole grains on cardiovascular risk by considering confounding variables on pooled outcomes.

A systematic search strategy was implemented across five electronic databases on the 8^th of March 2019 (Marshall et al, 2019, unpublished data). The search aimed to identify randomised controlled trials which compared whole grain test products against refined grain test products for ≥8-weeks duration (or ≥2-weeks duration if the study reported inflammatory or oxidative stress markers). Studies were required to provide the test products to participants and control for background diet, and samples may have been health or diagnosed with cardiovascular disease, metabolic disease, or type 2 diabetes mellitus. Studies were required to report at least one of 80 outcomes which contribute to cardiovascular disease risk. All identified records were screened for eligibility by two investigators independently. Studies which were considered eligible underwent data extraction by one investigator and checked for accuracy twice by a second investigator. Study characteristics were extracted, including study design, sample population, and details of the intervention and control conditions. Outcome data were extracted for the intervention and comparator groups were baseline, change, and follow-up variables, variable units, and statistical significance over time and between groups. Adverse events were also recorded.

Data extracted were study and participant characteristics, baseline, follow-up, change in outcome, and p-value for between group comparisons. Data were reported as mean (standard deviation) unless specified otherwise, such as median, standard error (S.E.), 95% CI, or range. In addition to p-values comparing groups, p-values for change from baseline were extracted. Units for all data were reported in the cell for each variable, unless specified in the header (e.g. HbA1c always reported as %). If a study did not report on an outcome, a period was placed in the cell. Any significant finding (p<0.05) reported by a study was bolded in the cell. Data were extracted and entered into an Microsoft Excel worksheet according to the type of data: characteristics, inflammatory markers, oxidative stress markers, glycaemic and insulin markers, blood lipids, haemodynamics and comorbidities, body composition, adverse events.