Is variety more important than quantity of fruits and vegetables in relation to cardiovascular disease incidence and mortality? Results from a prospective cohort study

Abstract We aimed to prospectively assess the association between variety and quantity of fruits and vegetables (FV) and cardiovascular diseases (CVD) incidence and mortality due to the limited evidence. Our analysis included 2,918 adults with a follow-up period of 29,559 person-years. An inverse association was detected between fruit intake and the risk of incidence and mortality from CVD. We found no association between diversity scores of fruits, vegetables, and FV with CVD risk. Subjects with high quantity-high variety, high quantity-low variety, and low quantity-high variety of fruits, vegetables, or FV exhibited no difference in CVD risk compared to the subjects with low quantity-low variety intake. Increasing the variety of FV was associated with increases in the intake of β-carotene, lycopene, lutein, vitamin C, selenium, fibre, fat, and protein after adjustment for the quantity and covariates. We detected an inverse association between fruit intake and the incidence and mortality rates of CVD.


Introduction
Fruits and vegetables (FV) are important components of a healthy diet.Low consumption of fruits and vegetables is associated with a higher risk of chronic disease and approximately 5% of excess deaths worldwide (World Health Organization 2021).Variety of FV, in addition to quantity, is also essential for promoting health; various FV contain varied levels of vitamins, minerals, phytochemicals, etc. and are therefore believed to have the most beneficial effect when consumed in combination (Slattery et al. 1997;Leenders et al. 2015).Because of this, the recommendations for FV intake emphasise eating variety as well as quantity (Public Health England 2016).
Health benefits of the FV variety have been investigated in relation to various health outcomes, including cognitive health (Sheng et al. 2022), type 2 diabetes (Cooper et al. 2012), cancer (Büchner et al. 2010;2011;Jeurnink et al. 2012;Leenders et al. 2015), and cardiovascular disease (CVD) (Bhupathiraju and Tucker 2011;Griep et al. 2012;Conrad et al. 2018).CVD is especially of great interest, as it is the leading cause of death worldwide (Mensah et al. 2019).It has been proposed that FV varieties provide synergistic and additive combinations of phytochemicals (Liu 2003).The phytochemicals prevent CVD through anti-inflammatory and antioxidant activity, reduction of platelet aggregation, modulation of cholesterol synthesis and absorption, and reduction of blood pressure (Pagliaro et al. 2015;Bachheti et al. 2022).
The association between the amount and frequency of FV intake and CVD risk has been extensively studied, and a protective effect in terms of CVD incidence (Aune et al. 2017) and mortality (Kazemi et al. 2023) has been demonstrated.For variety, however, only a few studies have been conducted, and the results of those studies are controversial and mostly in favour of the lack of association (Bhupathiraju and Tucker 2011;Griep et al. 2012;Conrad et al. 2018).While analysis of NHANES data indicated an inverse linear relationship between vegetable variety and coronary heart disease (Conrad et al. 2018), two other cohort studies, including the Nurses' Health Study and the Health Professionals Follow-Up Study (Bhupathiraju and Tucker 2011) and a prospective study on 20,000 Dutch participants (Griep et al. 2012) failed to find any significant relationship.
Attention in the research setting and within dietary guidelines has focused more recently on the concept of food synergy and dietary variety, advocating FV variety (Herforth et al. 2019).However, since few studies have evaluated FV variety in relation to CVD incidence and mortality and none of the previous studies have assessed these associations in an East Asian population, the aim of the present study was to assess the relationship between variety and quantity of FV and CVD incidence and mortality in the Tehran Lipid and Glucose Study (TLGS).

Study population and design
This study was conducted using data from TLGS.TLGS is an ongoing study with the aim of determining and preventing risk factors for non-communicable diseases and is updated every three years.The design of the TLGS has been explained in detail previously (Azizi et al. 2018).Briefly, 150,059 participants aged 3-75 residing in district No. 13 of Tehran were enrolled between 1999 and 2001.In the current study, data from 3,055 people aged older than 18 years who had participated in the third wave (2006)(2007)(2008) of the TLGS and had completed dietary data was used.At the start of the study, every participant signed a written informed consent form.Participants having an energy intake of more than or less than 1% or over 99% of the total energy intake (i.e. less than 916 kcal/d and more than 4990 kcal/d in females and less than 1026 kcal/d and more than 5890 kcal/d in males) were excluded from the analyses (n = 58).Participants with a self-reported history of CVD at the time of enrolment were also excluded (n = 69).All participants (n = 2918) were followed until 2018, and those who lost follow-up were removed from analysis (n = 8).This study was conducted according to the Helsinki declaration ethics, written consent was obtained from all participants and the ethical committee of Shahid Beheshti University of Medical Sciences approved the study (Code: IR.SBMU.ENDOCRINE.REC.1402.006).

Dietary assessment
The dietary intake of participants was assessed using a validated 168-item food frequency questionnaire (FFQ) (Esfahani et al. 2010) by trained interviewers at baseline and afterward every three years.Participants were asked about the frequency (daily, weekly, or monthly) and portion size (e.g.cup, spoon, and ounce) of each consumed food item during the previous year.Portion sizes of food items were asked in household measures and were then converted into grams.To estimate the energy and grams of each food, the Iranian Food Composition Table and the food composition tables of the United States Department of Agriculture (USDA) were used.In the present study, the cumulative average of fruit and vegetable intake for each participant was calculated and applied to assess the relationship between the FV and its subgroups and the risk of CVD incidents.

Food and vegetable variety estimation
This study focuses on the consumption of five subtypes of fruits (berries, citrus fruits, grapes, hard fruits, and stone fruits) and eight vegetables (cabbages, fruiting vegetables, grain and pod vegetables, leafy vegetables, mushrooms, onions and garlic, root vegetables, and stalk vegetables) (Supplementary material 1).The vegetable and fruit variety was assessed by the diet diversity score (DDS).DDS is commonly used to measure the variety of diets by calculating the number of different food items consumed within a given time period (Slattery et al. 1997).Five DDS were calculated: one for all FV items, one for all fruit items, one for all vegetable items, one for the eight subtypes of vegetables as given above, and one for the five subtypes of fruits.For each vegetable, fruit, and their combined and subgroups, '1' indicates the intake of at least one cup-equivalent per week, and '0' indicates a lesser amount (Leenders et al. 2015).The total score was calculated by summing up the scores of items or subgroups.The higher the score, the greater the vegetable and fruit variety.Fruit consumption includes fresh, dried, and canned fruits.The consumption of vegetables did not include potatoes.

Ascertainment of cardiovascular events and CVD-mortality
A professional nurse followed the participants annually to update data on cardiovascular events, mortality, and specific causes of death.Then an experienced physician collected patients' medical records or death certificates for additional examination.At the end, the outcome committee of TLGS assessed and confirmed events.CVD diagnosis was coded using the 10th version of the International Classification of Diseases (ICD-10) codes and included fatal or non-fatal myocardial infarction (MI), unstable angina, angiography-proven coronary heart disease (CHD), heart failure, cardiac deaths, fatal or non-fatal stroke, and transitory ischaemic events.

Assessment of covariates
General information included age, sex, history of medications, smoking, etc., physical activity, and anthropometric measures were collected at baseline and updated every 3 years by trained interviewers.Standardised questionnaires were used to collect data on smoking during each follow-up assessment.
Participants were asked if they currently smoked, and they may choose to answer 'yes, every day, ' 'yes, sometimes,' or 'no.'When the first question was answered in the yes, an evaluation was carried out to determine the quantity of cigarettes smoked daily and the frequency of smoking during the previous seven days.A daily smoker is someone who smokes at least once a day; an occasional smoker is someone who smokes occasionally but not every day; an ex-smoker is someone who used to smoke regularly but is no longer a smoker; and someone who has never smoked is someone who has never smoked before or has smoked too.Physical activity was measured by a modifiable activity questionnaire which assess leisure time and occupational activities over the last year (MAQ) (Momenan et al. 2012).Participants were asked to report the frequency and length for each activity; the metabolic equivalent (MET)-hour per week was calculated by dividing the total physical activities by 52.Anthropometric measures were examined using digital scale (Seca 707, Germany) and tape and body mass index (BMI) was calculated.

Statistical analysis
First, we evaluated the individual associations between variety and total quantity in FV intake in relation to CVD risk and mortality.Participants were divided into tertiles of quantity and quantity-adjusted variety of FV intake, with the lowest tertile considered the reference category.The multivariable Cox proportional hazards regression model was used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) for CVD incidence and mortality across the categories.
In the second approach, both the quantity and variety of FV were dichotomised at their medians.Then, participants were categorised into four groups: 1. subjects with high quantity and high variety of FV (HQHV); 2. subjects with high quantity but low variety of FV (HQLV); 3. subjects with low quantity but high variety of FV (LQHV); and 4. subjects with low quantity and low variety of FV (LQLV).The fourth group was considered the reference category, and HRs and 95% CIs for CVD incidence and mortality for other groups were estimated in comparison to this group.Multivariable models were adjusted for age (years, continuous), sex (male, female), body mass index (BMI) (< 18.5, 18.5 to < 25, 25 to < 30, ≥ 30 kg/m 2 ), smoking status (never, former, current), dyslipidemia, diabetes, and hypertension (yes, no), weight changes (kg, continuous), dietary carbohydrates and fat (percentage from energy, contagious), and total energy intake (kcal/d).The person-years for each participant were estimated from baseline to the first CVD incidence, the date of death, the last visit, or the end of follow-up.
In a sensitivity analysis, participants with a follow-up duration of less than 2 years were excluded to address potential reverse causality.
All statistical analyses were performed using SPSS software (version 20.0;IBM Corporation, Armonk, NY, USA).Descriptive variables were reported as means ± standard deviation (SD), and percentages.

Results
A total of 2,918 adults (1,306 men and 1,612 women) who met the inclusion criteria were included in this study.During 29,559 person-years of follow-up (between 2005During 29,559 person-years of follow-up (between -2008During 29,559 person-years of follow-up (between and 2018;;total, 10.1 years;median, 10.6 years), 181 CVD incidences and 22 CVD deaths were documented.At baseline, the mean (SD) age of participants was 39.3 (14.0) years, with 55.2% being female.In the total population, the median intake of fruits and vegetables was 212 and 297 g/d, respectively.The median variety scores of fruits, vegetables, and FV in the total population were 8 (minimum: 6, maximum: 9), 13 (minimum: 11, maximum: 15), and 21 (minimum: 17, maximum: 24), respectively.The most commonly eaten fruits and vegetables were citrus and fruity vegetables, respectively.
People who consumed a higher amount of FV, either in low or high variety, were older, more likely to be female, less likely to smoke, and had a higher BMI (Table 1).Individuals with a low quantity and variety of FV had a lower level of physical activity than the other three groups.People who consumed a higher variety of FV, either low or high quantities, consumed a higher amount of whole grains and legumes (Table 2).
The third tertile of fruit intake was associated with the higher risk of CVD incidence and mortality compared to the lowest tertile (HR = 0.69, 95% CI = 0.48 to 0.98, Table 4).However, no association was observed between the amount of total vegetables, or FV, and the risk of CVD.Furthermore, subjects in the third tertile of pod (HR = 0.45, 95% CI = 0.30 to 0.66), stalk (HR = 0.60, 95% CI = 0.42 to 0.84), and leafy vegetable (HR = 0.50, 95% CI = 0.35 to 0.72) intake had a lower risk of incidence and mortality of CVD than those in the first tertile of intake.
We found no significant association between diversity scores of fruits, vegetables, and FV and CVD incidence and mortality, as shown in Table 5.Furthermore, when considering FV, the HQHV, HQLV, and LQHV groups exhibited no considerable difference in terms of CVD risk compared to the LQLV group.However, for some subtypes of vegetables, including leafy (HR = 0.56, 95% CI = 0.38 to 0.81), pod (HR = 0.60, 95% CI = 0.42 to 0.85) and stalk vegetable (HR = 0.66, 95% CI = 0.43 to 1.00), HQHV was significantly associated with a lower risk of CVD compared to LQLV.For pod (HR = 1.80, 95% CI =  1.21 to 2.67) and stalk (HR = 4.30, 95% CI = 1.46 to 12.70) vegetables, lower amounts despite higher variety were associated with a higher risk of CVD, as demonstrated in Table 6.Our findings remained largely unchanged when we excluded the participants with less than 2 years of follow-up (data not shown).
There was no significant interaction by age (p = 0.08) or sex (p = 0.35).The joint association of the quantity and variety in fruit and vegetable intake and CVD risk was not affected by age groups (Supplementary material 2), as well as sex, diabetes, and hypertension at baseline (Supplementary material 3).Regarding BMI and dyslipidemia, the risk of CVD was significantly lower in HQHV and HQLV compared to the LQLV group in subjects with BMI < 25 and also in subjects without dyslipidemia.

Discussion
In the present population-based cohort study of generally healthy men and women, we have observed a 31% reduction in the incidence and mortality risk of CVD among those who consumed the highest amount of fruit compared to those who consumed the lowest.Additionally, we detected a linear relationship between the quantity of stalk and pod vegetables, but not total vegetables and FV intake, and the incidence and mortality rate of CVD.Our study also revealed that a greater variety in FV consumption accompanied higher intakes of several micronutrients and macronutrients, similar to previous studies (Griep et al. 2012;López-González et al. 2021).However, we found no significant association between the variety of FV consumption and CVD.
Our findings were consistent with those of a population-based prospective cohort study conducted in Japan during a follow-up period of 459,320 person-years, which showed that a higher intake of fruits but not vegetables was significantly associated with a reduced risk of CVD (Takachi et al. 2008).However, available data on this topic is controversial, with some studies reporting no association between either fruits or vegetables and CVD risk (Dauchet et al. 2006;Yu et al. 2014) Nonetheless, a previous systematic review and meta-analysis study demonstrated that higher intake of fruits, vegetables, and FV was associated with decreased risk of CVD and CHD incidence and mortality.In that study, citrus fruits, pure fruit juice, pommes, allium, carrots, cruciferous vegetables, and green leafy vegetables showed higher CVD benefits than other FV (Zurbau et al. 2020).The protective effects of FV against cardiometabolic risk factors might be attributed to their high levels of bioactive compounds like fibre, minerals, vitamins, and phytochemicals and their synergistic actions (Mitrou et al. 2007;Zhang et al. 2013;Turati et al. 2015).Additionally, their low energy density and glycemic index could contribute to this protection (Zurbau et al. 2020).Conflicting evidence on this topic may be due to differences in the types of FV commonly consumed and variations in cooking and preservation techniques across countries (International Food Policy Research Institute 2020).Most dietary guidelines highlight the importance of food synergy and dietary variety (Jacobs et al. 2009;Marshall et al. 2020), which has triggered a change in perspective from quantity to FV variety to obtain health advantages from a range of nutrient content present in different groups of FV.Greater variety in FV consumption has been associated with improved inflammation (Bhupathiraju and Tucker 2011), diversity of gut microbiota (Heiman and Greenway 2016), overall diet quality, nutrient intake, and a healthier lifestyle (Ramsay et al. 2017;López-González et al. 2021).However, despite these benefits, some studies have reported no association between FV variety and chronic diseases or mortality (Jeurnink et al. 2012;Leenders et al. 2015;Conrad et al. 2018).Our findings concerning the FV variety were consistent with the results of two earlier prospective cohort studies.Bhupathiraju et al. conducted a study using the Nurses' Health Study and Health Professionals Follow-Up Study, which included 113,276 subjects followed up for 22-24 years.Their research found that only quantity, not variety, of FV intake was associated with a reduced risk of CHD (Bhupathiraju and Tucker 2011).In another study by Griep et al. that followed 20,069 participants for a duration of 10 years, no significant relationship was reported between the variety of FV and the incidence of CHD and stroke (Griep et al. 2012).The variety of FV intake was also not associated with cardiometabolic risk factors in diabetic patients (Lamb et al. 2017) or the Framingham risk score in subjects free of CVD (Bhupathiraju and Tucker 2011).Similar to our study, dietary intake was evaluated using a FFQ in the  aforementioned studies.However, another research used 24-h dietary recalls to assess dietary intake and reported a significant inverse association between the quantity and variety of vegetable intake and the prevalence of CHD (Conrad et al. 2018).In another 15-year prospective population-based cohort study conducted on women, vegetable variety was evaluated using a single question, 'How many different vegetables do you usually consume each day?' within a FFQ.They observed an inverse association between the diversity of consumed vegetables and the thickness of the common carotid artery intima-media, which is an atherosclerosis risk marker, independent of total vegetable intake (Blekkenhorst et al. 2020).One plausible explanation for the conflicting evidence regarding FV variety and health outcomes may be attributed to variations in the methodology implemented for evaluating dietary intakes.The FFQ tool may not be appropriate for assessing FV variety due to certain limitations.Firstly, not all FV choices, especially those that are specific to certain regions, are typically included in the FFQ.Secondly, the consumption frequency of certain FV varies according to the season; however, in the FFQ, their frequency of consumption was calculated based on the frequencies per day during the last year.As a result of these limitations, variety scores could be underestimated, leading to a lower variety of FV.A better assessment of FV variety could be achieved by using a questionnaire specifically designed for this purpose or by adding questions to the FFQ about less frequently consumed FV during the season when they are commonly available (Ashton et al. 2019).Another suggestion could involve combining the FFQ tool with dietary history for dietary assessment (Jansen et al. 2004).A further possible explanation for the contradictory evidence could be the disparities in the definition of varieties among different studies.Various researchers have gauged the diversity of FV based on a range of guidelines and considering different groups and subgroups of such foods.For instance, while some variety scores take into account potatoes, legumes, and fruit juice, others do not (Marshall et al. 2020).

Strength and limitations
Our study has several strengths, such as the prospective design, sufficient follow-up duration, low attrition rate (<1%), and considering a wide variety of FV (17 fruits and 19 vegetables).Notably, this is the first long-term population-based prospective research conducted in an East Asian population with a high burden of CVD events.Nonetheless, our study has some limitations.First of all, given the limited sample size and number of events, our study may have been underpowered to detect a significant relationship.The limited sample size, in particular, could have had an impact on the outcomes when examining different types of fruits and vegetables separately.Secondly, dietary intake was assessed only at baseline, which may not fully represent the potential changes in dietary intake and habits over time.Furthermore, using FFQ may introduce some bias into the findings due to various factors, including recall bias, estimation errors, and under-or over-reporting.Finally, as this is an observational study, the findings do not imply causality, and the results may still be influenced by residual confounding factors.

Conclusions and future direction
Our study found a reverse association between the quantity of fruit consumed and the incidence and mortality of CVD.However, the results of our study do not support the recommendation of consuming a variety of FV to lower CVD risk.Notably, greater variety in FV intake may still have some beneficial effects, such as increasing the overall amount of FV consumption and ensuring adequate intake of micronutrients and bioactive compounds.To better understand the impact of variety on CVD risk, future prospective cohort studies that collect detailed data on FV variety using a validated method are needed.

Table 1 .
Baseline characteristics of participants based on joint association of the quantity and variety in fruit and vegetables intake.

Table 2 .
dietary intake of participants based on joint association of the quantity and variety in fruit and vegetables intake.
*adjusted for age, sex, and total energy intake.

Table 3 .
nutrient intake of participants across the fruit and vegetables variety tertiles.
*adjusted for age, sex, total energy intake, and quantity of fruit and vegetable.** Percentage of daily energy intake.

Table 4 .
Hazard ratios (95% cI) for developing cVd across tertiles of quantity in fruit and vegetables intake.
a adjusted for sex and age.b further adjusted for BMI, smoking, dyslipidemia, diabetes, hypertension, weight changes, total energy intake, percentage of energy from carbohydrates, and percentage of energy from fat.

Table 5 .
Hazard ratios (95% cI) for developing cVd across tertiles of variety in fruit and vegetables intake.hazards models were used.Variety in each item was adjusted based on quantity of that item.a adjusted for sex and age.b further adjusted for BMI, smoking, dyslipidemia, diabetes, hypertension, weight changes, total energy intake, percentage of energy from carbohydrates, and percentage of energy from fat.

Table 6 .
Hazard ratios (95% cI) for developing cVd based on joint association of the quantity and variety in fruit and vegetables intake.
a adjusted for sex and age.b further adjusted for BMI, smoking, dyslipidemia, diabetes, hypertension, weight changes, total energy intake, percentage of energy from carbohydrates, and percentage of energy from fat.