This dataset comprises data extracted from 86 publications included in an umbrella review which compared the effect of colour-associated bioactive pigments found in fruit and vegetables (carotenoids, flavonoids, betalains and chlorophyll) on human health outcomes relevant to public health.  Meta-analysed data from 83 systematic literature reviews were available for 17 different bioactive pigments spanning all colours of fruit and vegetables except green, with additional data from two randomised controlled trials and one cohort study for chlorophyll. This dataset represents 2,847 original research studies and data from over 37 million participants. There were 449 meta-analysed health outcomes extracted from the 83 systematic literature reviews. Extracted health outcomes were categorised according to pigment, comparator type, broad health outcome, study design, age group, source of pigment, risk of bias, and confidence in the estimated effect.  Data extracted were dose, intervention duration, sample size, number of original studies, effect estimate, 95% confidence intervals, I2 statistics, publication bias and p-value. Rigorous analysis, including estimations of a common effect size, stratification of the evidence, study level sensitivity analyses and reporting on heterogeneity and potential biases, may be carried out using these data, to further evaluate the effect of colour-associated bioactive pigments in fruit and vegetables on human health.

A systematic search strategy was implemented across four electronic databases on the 29th October 2021 [1]. The search aimed to identify systematic literature reviews of randomised controlled trials or cohort studies with meta-analyses which compared the effect of colour-associated bioactive pigments found in fruit and vegetables, compared to placebo or low intakes, on human health outcomes relevant to population health. Studies were required to report: (i) colour pigments consumed by humans through whole fruit or vegetables, extract or supplements derived from fruit or vegetables, and (ii) health-related outcomes relevant to population health including the prevention of disease and optimisation of disease risk factors, general wellbeing, function (cognitive function, physical function and exercise performance), growth and development in children, maternal and neonatal health. All identified records were screened for eligibility by two independent researchers. Studies which were considered eligible underwent data extraction by one researcher and checked for accuracy twice by a second investigator.  Study, participant and outcome data were extracted, including bioactive pigment name/colour, intervention and comparator conditions (type, duration and dose), number of meta-analysed studies/intervention groups, model, meta-analysed outcome, sample size (intervention/case, comparator/control, and total), effect size and statistical significance, risk of bias, heterogeneity, publication bias and the Grading of Recommendations Assessment, Development and Evaluation (GRADE) quality rating (if reported) [2]. During data extraction, included studies were assessed for methodological quality using the Oxford University, Centre for Evidence-Based Medicine (CEBM) critical appraisal tool for systematic reviews, RCTs or prognostic studies [3].

Meta-analysed data from 83 systematic literature reviews was available for 17 different bioactive pigments spanning all colours of fruit and vegetables except green (i.e., chlorophyll), with additional data from two single randomised controlled trials and one cohort study available for chlorophyll. No data were found for betalains or its sub-classes; however, the betalains colours of red, violet, orange and yellow are represented by the included carotenoids and flavonoids. The 83 included systematic literature reviews measured and reported on 449 eligible meta-analysed health outcomes. Extracted health outcomes were grouped as cancer (reported by 192 meta-analyses), cardiovascular disease (135 meta-analyses), exercise (28 meta-analyses), mortality (27 meta-analyses), type 2 diabetes (24 meta-analyses), obesity (13 meta-analyses), bone health (9 meta-analyses), eye health (9 meta-analyses), the nervous system (5 meta-analyses), pregnancy health (4 meta-analyses), cognitive function (2 meta-analyses) and the respiratory system (1 meta-analysis).  Each single randomised controlled trial and cohort study for chlorophyll examined a unique health outcome: cancer (1 cohort study), cardiovascular disease (1 randomised controlled trial) and allergy (1 randomised controlled trial).

Data were extracted and entered into a Microsoft Excel worksheet according to the type of data. Extracted data were presented in four worksheets: (i) systematic literature review, meta-analysis and CEBM quality assessment data, (ii) systematic literature review GRADE assessment data, (iii) original research data, and (iv) reference list. As few original authors applied GRADE, authors (M.B and S.M) completed GRADE assessments for each extracted meta-analysis using information provided in the relevant systematic literature reviews or collated from individual randomised controlled trials or cohort studies reported by the systematic literature review. GRADE was not applied to outcomes reported by single randomised controlled trials or cohort studies due to insufficient number of studies. Any significant finding (p<0.05) reported by a study was bolded in the cell.

References

1. Blumfield, M.; Mayr, H.L.; De Vlieger, N.; Abbott, K.; Starck, C.; Fayet-Moore, F.; Marshall, S. Should we ‘Eat a Rainbow’? An umbrella review of the health effects of colorful bioactive pigments in fruits and vegetables. Molecules 2022, [In Press}.

2. Handbook for Grading the Quality of Evidence and the Strength of Recommendations Using the GRADE approach. 2013, Accessed 1st September 2021. https://gdt.gradepro.org/app/.

3.  Critial Appraisal Tools; Centre for Evidence-based Medicine, Unviersity of Oxford. Accessed 1st September 2021. https://www.cebm.ox.ac.uk/resources/ebm-tools/critical-appraisal-tools