Effect of increasing fruit and vegetable intake by dietary intervention on nutritional biomarkers and attitudes to dietary change: a randomised trial

Participants from the NE of Scotland were recruited through the Human Nutrition Unit (HNU) at the Rowett Institute of Nutrition and Health (RINH), the University of Aberdeen. Participants were recruited through the RINH recruitment database and by advertising the study in the local media. Inclusion criteria were for healthy men and women with a habitually low fruit and vegetable intake, non-smoking, non-medicating, with self-reported normal bowel function, aged 38–60 years and with a BMI between 18 and 39 kg/m². Participants were excluded if they reported consuming more than three portions of fruit and vegetables per day as assessed by a food frequency questionnaire (The Scottish Collaborative Group FFQ version 6.6; http://​www.​foodfrequency.​org), smoked, had a chronic medical condition (including inflammatory bowel disease), had a known allergy to plant products and/or took medication (including contraceptive tablets, hormone replacement therapy or thyroid drugs). This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects was approved by the Ethics Committee of the Rowett Institute of Nutrition and Health, University of Aberdeen (study code: 09/003) following review by the North East of Scotland Research Ethics Committee. Written informed consent was obtained for all subjects. Participants were informed by way of an information sheet as to the purpose and risks of the study. This trial is registered at Controlled-Trials.com; registration ISRCTN71368072.

Fifty-seven volunteers were assessed for entry into the study (Fig. 1). Six of these did not meet the inclusion criteria or declined to participate further. A total of fifty-one participants were randomly assigned (by sex, age and BMI) by a statistician to the study groups (n = 26 in the control group and n = 25 in the treatment group). Participants were paired with others of a similar sex, age and BMI, and treatment assigned randomly within each pair. It was not possible to use an optimal pairing across all volunteers, as it was necessary to start some participants onto study before all had enrolled across both phases. Participants in the control group were asked to maintain their habitual diet that included 3 or fewer portions of fruits and vegetables each day. Participants in the intervention group were asked to consume approx. 480 g of commonly available and taxonomically diverse fruits, vegetables and fruit juices (max. 300 ml per day) in addition to their normal intake. All fruit, vegetables and juice were provided (twice weekly) to the participants. The same fruits and vegetables were provided to all volunteers unless a participant declared a strong dislike for a particular item, in which case it was replaced. Participants were instructed to eat all of the foods provided, ensuring equivalent consumption of fruits and vegetables. Advice on preparation, storing and cooking methods was provided at the start of the intervention and throughout the study. Written instructions on preparation, including options on common preparation practices, was supported by twice-weekly personal contact with participants during fruit and vegetable collection. In addition, all participants were offered telephone or e-mail support throughout the study. The fruit and vegetables included apples, oranges, soft fruit and berries, broccoli, Brussels sprouts, cabbage, carrots, cauliflower, celery, cucumber, mixed lettuce, spinach, sweet pepper and tomatoes. Participants were supplied with a record sheet to record what fruit and vegetables were eaten daily, and if they were unable to eat all the foodstuffs each day. The study was carried out over a 20-week period and included a 2-week run-in period where participants consumed their habitual diet, a 12-week intervention and a 6-week washout period where fruit and vegetable provision was withdrawn and participants were allowed to eat as they wished. A washout time point was included to determine the efficacy of the nutritional and functional biomarkers of fruit and vegetable intake.

Blood sampling was carried out on six occasions; pre-run-in (time −2), baseline (time 0), intervention (weeks 4, 8 and 12) and post-washout (week 18). The study was carried out between March and August over a period of 2 years (2011 and 2012) to control, as much as possible, for seasonal variation. Forty-five participants completed the study (Fig. 1).

Dietary intake (energy, micro- and macronutrient) and compliance was assessed using 4-day weighed food diaries before (pre-run-in), during (weeks 4 and 12) and after (post-washout) the intervention. Diaries were analysed using WinDiets 2005. Compliance during the intervention was assessed by calculating the total number of portions of fruit and vegetables each participant reported consuming over the 4 days for each of the time points. Portion sizes were taken from the National Health Service (NHS) “Livewell” website [29], which states that an adult portion is 80 g for fruit and vegetables, 30 g for dried fruit and 150 ml of fruit and/or vegetable juice. To calculate the number of portions, the weight of each fruit, vegetable and fruit juice serving recorded was divided by the adult portion size for that item. Using this method, the total number of portions reported as consumed for each 4-day food diary was calculated and divided by 4 to give the average number of portions consumed per day. As 150 ml of fruit or vegetable juice counts only as a maximum of one portion per day, any juice consumed above this amount was recorded as a proportion of an additional portion. Compliance to the intervention was further assessed by participants recording all foodstuffs not consumed on a weekly checklist of all items provided.

At each visit, fasted blood (12 h) was collected from the antecubital vein into vacutainers containing EDTA as anticoagulant. EDTA-treated whole blood was incubated with ascorbic acid (1% w/v) for 1.5 h in the dark, snap frozen in liquid nitrogen and stored at −80 °C for folate analysis. All remaining EDTA blood was centrifuged at 2400×g for 15 min at 4 °C and the plasma aliquoted, snap frozen and stored as above. Lymphocytes were isolated from the buffy coat by density gradient centrifugation [30].

Plasma and whole-blood folate and plasma B12 were measured by radioassay [31]. Plasma homocysteine was measured by gas chromatography [31]. Plasma vitamins B2, C, carotenoids (α- and β-carotene, β-cryptoxanthin, lycopene, lutein/zeaxanthin) and tocopherols (α- and γ) were measured by HPLC [32, 33]. Plasma total cholesterol, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, triglycerides (TG) and glucose were measured using a Konelab 20 Clinical Chemistry Analyser (Thermo Scientific, Passau, Germany). Plasma antioxidant capacity was measured using the trolox equivalent antioxidant capacity (TEAC), hydroxyl radical antioxidant capacity (HORAC), ferric reducing antioxidant power (FRAP) and total phenol assays in EDTA-treated plasma spectrophotometrically [34‐37]. Antioxidant enzyme activity in EDTA-treated red blood cell membranes [glutathione peroxidase, catalase, superoxide dismutase (SOD) and glutathione reductase] were measured as described previously [38‐41]. Endogenous DNA strand breakage was measured in lymphocytes isolated from EDTA-treated whole blood using single cell gel electrophoresis [30]. Ability to resist an oxidative stress was measured in lymphocytes exposed to hydrogen peroxide (H₂O₂; 200 μm for 5 min on ice) before analyses of induced DNA strand breakage [30]. All analyses were carried out on blinded samples.

Blood pressure (on average three consecutive readings) was measured at each visit to the HNU with a sphygmomanometer (OMRON705CP) according to guidelines from the British Hypertension Society. Arterial stiffness was assessed by pulse contour analysis (Pulse Trace PCA, Micromedical Ltd).

Attitudes towards consuming fruit and vegetables were explored using a questionnaire designed for the study. The questionnaire included both open and closed questions about barriers to healthy eating, including cost, taste, effort of preparation and cooking, availability, and family acceptance that may inhibit fruit and vegetable intake [42, 43].

The questionnaire was pre-tested for interpretation and ease of completion using cognitive interviewing techniques and modified accordingly. All participants were contacted 10–12 months after the intervention and sent the questionnaire by post to complete. A reminder letter and second questionnaire were sent to non-responders 2 weeks after the initial mailing.

Study sample size was determined for the primary outcome of plasma vitamin C. Based upon a previous human study [44], an SD of 30% was used for calculating sample size. An expected decrease/increase of 30% in vitamin C, with an SD of 30%, power of 80% and a significance level of 5%, yields the need for 20 participants in each group to detect significant differences between treatments. Calculations were based upon tests comparing 2 means with 2-sided equality.

To test for differences in dietary intake and mean number of portions of fruit and vegetables reported as consumed between treatment groups with time (compliance), independent sample t-tests were conducted at baseline (time 0), throughout the intervention (weeks 4 and 12) and post-washout (week 18). For analysis of data collected by questionnaire (sustaining fruit and vegetable intake), for questions using a 5-point Likert scale, the points were collapsed to 3 for comparison (agree, neither agree nor disagree, disagree), collated and descriptively analysed, giving frequencies and percentages applied to the quantitative questions. The responses to the open-ended questions were grouped by 2 researchers independently and results compared [45].

Fifty-seven people were screened prior to the intervention. One did not meet the inclusion criteria (low BMI) and five declined to take part in the study (undisclosed reasons). The remaining 51 participants were stratified by sex, BMI and age and randomised to two groups, control or intervention. Of these, three withdrew before week 2, two withdrew before week 3 and one withdrew before week 4. Forty-five participants completed the study, with 21 (9 men, 12 women) in the intervention group and 24 (10 men and 14 women) in the control group (Fig. 1).

The age of the participants ranged from 39 to 58 years with a mean age of 48 years. Mean BMI was 26.48 ± 3.66 kg/m² with no underweight participants, 31% normal weight, 56% overweight and 13% obese participants. More women than men were recruited (26 versus 19) and the number of participants that completed the study in each group was slightly unbalanced (24 control participants versus 21 in the intervention group). There was no significant difference between the groups with regard to age and BMI at the start of the study (Table 1) and no difference in calculated macro- and micronutrient intake (Table 2). Similarly, the vast majority of plasma and whole-blood nutrients and functional biomarkers were similar between the two treatment groups (Tables 3, 4, 5). However, participants in the intervention group had significantly lower plasma γ-tocopherol and HORAC levels at time 0, compared with subjects in the control group.

Fruit, fruit juice and vegetable consumption, estimated using 4-day weighed intake data, were similar between the treatment groups prior to the start of the study (mean 3.02 and 2.63 portions per day for the control and intervention group, respectively; P > 0.05) and remained essentially unchanged in the control group throughout the intervention (Fig. 2). Participants in the intervention group were asked to eat all of the additional juice, fruits and vegetables provided to them, ensuring equal consumption of fruits and vegetables across the study. Intake of fruit, juice and vegetables increased significantly (P < 0.001) by approximately five portions in the intervention group to a maximum of 8.4 portions per day after 12 weeks (Fig. 2). Post-washout, the intake of fruit, vegetables and juices in the intervention group returned to almost pre-intervention levels (3.33 portions) and was not significantly different from baseline intake (P > 0.05).

Intervention had no significant effect on participant weight, total energy, total protein, total carbohydrate or total fibre intake (Table 2). While consuming up to eight portions of fruits and vegetables daily did not influence total fat intake, PUFA intake was significantly lower in the intervention group at the end of the 12-week intervention (approx. 25%), but returned to pre-study levels during the washout (Table 2). Similarly, total cholesterol intake was lowered by eating additional fruits, vegetables and juices (Table 2), and this effect was sustained post-washout (approx. 20%). Total sugar and NMES intake increased in the intervention group after 12 weeks on the study (approx. 40 and 30%, respectively). Total starch intake decreased (approx. 20%) and NSP intake increased (approx. 25%) in the intervention group (Table 2). None of these changes were maintained after the 6-week washout (Table 2). Intake of the antioxidants vitamin C and β-carotene increased significantly in participants eating additional fruits and vegetables after 12 weeks on study (3.5- and 3.8-fold, respectively), as did the intake of folate (approx. 50%). These changes were not maintained post-intervention (Table 2). Retinol intake was slightly lower in the intervention group 12 weeks into the study (approx. 13%), but returned to pre-intervention levels during the washout phase (Table 2). Vitamin B12 and α-tocopherol intake were not altered by intervention (Table 2). Potassium intake was increased in response to increasing fruit, vegetable and juice consumption (approx. 20%), but this was not maintained (Table 2). Intake of iron, manganese, magnesium, phosphorous, calcium and zinc were unaffected by intervention (Table 2).

Providing volunteers an additional 480 g of fruit and vegetables and 300 ml of fruit juice for 12 weeks significantly (and in some cases progressively) increased the circulating concentrations of folate (approx. 15% plasma and whole blood; Fig. 3), vitamin C (approx. 35%), and the carotenoids, α-carotene (approx. 50%), β-carotene (approx. 70%) and lutein/zeaxanthin (approx. 70%, Fig. 4). No differences in plasma retinol, α- and γ-tocopherol and plasma B2 or B12 were observed (Table 3). Plasma lycopene declined significantly (approx. 10%) in the treatment group after 4 weeks on the intervention and remained lower at week 12. At the end of the washout period, plasma α-carotene, β-carotene, lycopene and lutein/zeaxanthin concentrations had returned to baseline, with no significant differences measured between the treatment groups. Conversely, plasma and whole-blood folate and plasma vitamin C levels remained slightly but significantly elevated in the intervention group (Figs. 3, 4).

The effect of intervention on several plasma and intracellular markers of antioxidant capacity and resistance to oxidative stress was measured. Substantially increasing intake of fruits, fruit juices and vegetables (up to a total of 8.4 portions) in general had no significant effect on plasma antioxidant capacity (TEAC, HORAC and FRAP), nor on the ability of isolated lymphocytes to resist DNA damage in response to an oxidative stress ex vivo (Table 4). Moreover, the intervention had no significant effect on the endogenous enzymatic antioxidant defence system, with no significant changes in the activity of erythrocyte catalase, glutathione peroxidase and superoxide dismutase (Table 4).

Despite significantly increasing circulating folate levels, increasing fruit, fruit juice and vegetable intake substantially did not change plasma total homocysteine concentrations (Table 4). Nor did it alter plasma glucose or lipid status [total cholesterol, HDL, LDL, TG, non-esterified fatty acids (NEFA)] or several markers of vascular health including BP, heart rate and vascular tone (Table 5). Moreover, there were no significant changes in body weight in response to intervention (Table 5). However, it should be noted that these markers of vascular health were not the primary end points of this study and accordingly the study was not powered specifically to detect changes in these biomarkers.

All of the respondents were aware of the UK Government dietary recommendation of “5-a-day” guidelines for fruit and vegetable consumption. The majority reported their consumption of fruit and vegetables 12 months after the study as 3–4 portions per day or less. No difference in intake was found between the intervention and control group. 62.5% of those in the intervention group believed that they consumed more fruits and vegetables 12 months post-intervention compared with before the study, while 37.5% said they ate approximately the same amount. The majority of the control group (62.5%) felt they consumed about the same. Of those that said their intake had increased, the main reasons given included (1) they enjoyed eating fruit and vegetables; (2) they felt better after eating fruits and vegetables, or (3) they wanted to eat a better diet. Only one respondent reported that their intake had decreased, and this was reported to be due to cost.

When comparing the amount of fruit and vegetables they ate 12 months post-intervention with during the study, 68.8% of people in the intervention group reported that they consumed less, while 62.5% of the control group reported that they consumed about the same. The main reasons for decreasing their intake included (1) inconvenience (e.g. during the study all fruit, juices and vegetables were delivered to them); (2) there was too much fruit and vegetables to eat during the study; (3) variation in amount eaten seasonally (i.e. they tended to eat more in the summer) and (4) cost.

The majority of respondents in both treatment groups agreed that eating more fruit and vegetables meant that they ate fewer other foods, that fruit and vegetables were filling, that they liked most fruit and vegetables and they did not think that fruit and vegetables were boring. Twelve months after the intervention, fewer participants in the intervention group agreed with the statement ‘I think that eating 5-a-day is easy’ compared with the control group (Table 6).

The majority of both the intervention and control group believed that they should eat more fruit and vegetables (87.5 and 75.0%, respectively).

With regard to perceived barriers to eating fruit and vegetables regularly, 56.3% of the intervention group and 46.7% of the control group believed that there were obstacles to incorporating more fruit and vegetables into the diet. The most common reasons were (1) the cost of fruit (but not vegetables); (2) the short shelf life of fruit and vegetables and (3) having to shop more frequently (Table 7). A lack of knowledge of how to prepare vegetables or time to prepare them was not reported as a barrier. Moreover, convenience, availability, lack of cooking skills, preparation time and not being accustomed to eating fruit and vegetables were not perceived as major obstacles to consuming fruits and vegetables by either group.

Taste was the most commonly reported influence on purchasing fruit and vegetables in both groups. The least common influences were the provenance of the fruits and vegetables and whether it was organic or not (Table 8).