Relation of Different Fruit and Vegetable Sources With Incident Cardiovascular Outcomes: A Systematic Review and Meta-Analysis of Prospective Cohort Studies.

Background Public health policies reflect concerns that certain fruit sources may not have the intended benefits and that vegetables should be preferred to fruit. We assessed the relation of fruit and vegetable sources with cardiovascular outcomes using a systematic review and meta-analysis of prospective cohort studies. Methods and Results MEDLINE, EMBASE, and Cochrane were searched through June 3, 2019. Two independent reviewers extracted data and assessed study quality (Newcastle-Ottawa Scale). Data were pooled (fixed effects), and heterogeneity (Cochrane-Q and I2) and certainty of the evidence (Grading of Recommendations Assessment, Development, and Evaluation) were assessed. Eighty-one cohorts involving 4 031 896 individuals and 125 112 cardiovascular events were included. Total fruit and vegetables, fruit, and vegetables were associated with decreased cardiovascular disease (risk ratio, 0.93 [95% CI, 0.89-0.96]; 0.91 [0.88-0.95]; and 0.94 [0.90-0.97], respectively), coronary heart disease (0.88 [0.83-0.92]; 0.88 [0.84-0.92]; and 0.92 [0.87-0.96], respectively), and stroke (0.82 [0.77-0.88], 0.82 [0.79-0.85]; and 0.88 [0.83-0.93], respectively) incidence. Total fruit and vegetables, fruit, and vegetables were associated with decreased cardiovascular disease (0.89 [0.85-0.93]; 0.88 [0.86-0.91]; and 0.87 [0.85-0.90], respectively), coronary heart disease (0.81 [0.72-0.92]; 0.86 [0.82-0.90]; and 0.86 [0.83-0.89], respectively), and stroke (0.73 [0.65-0.81]; 0.87 [0.84-0.91]; and 0.94 [0.90-0.99], respectively) mortality. There were greater benefits for citrus, 100% fruit juice, and pommes among fruit sources and allium, carrots, cruciferous, and green leafy among vegetable sources. No sources showed an adverse association. The certainty of the evidence was "very low" to "moderate," with the highest for total fruit and/or vegetables, pommes fruit, and green leafy vegetables. Conclusions Fruits and vegetables are associated with cardiovascular benefit, with some sources associated with greater benefit and none showing an adverse association. Registration URL: https://www.clini​caltr​ials.gov; Unique identifier: NCT03394339.

Grading of Recommendations Assessment, Development, and Evaluation

Newcastle‐Ottawa Scale

Clinical Perspective

What Is New?

Public health policies discourage the consumption of certain fruit sources (eg, 100% fruit juice, dried fruit, and tropical fruit) because of their sugar content and emphasize vegetable consumption before fruit.

We examined the relation of fruit and vegetable sources with cardiovascular disease outcomes.

What Are the Clinical Implications?

In this systematic review and meta‐analysis of 81 unique cohorts, we identified that fruits and vegetables are associated with cardiovascular benefit and no fruit or vegetable sources are associated with cardiovascular harm.

Certain fruit and vegetable sources showed greater associations with cardiovascular benefit, including citrus, 100% fruit juice, and pommes fruit and allium, carrots, and cruciferous and green leafy vegetables.

Increased fruit and vegetable consumption is the cornerstone of dietary guidance for cardiovascular disease (CVD) prevention. Their benefit as part of heart healthy diets is balanced against an increasing concern of their contribution to an excess intake of sugars. , Some influential commentators have even questioned the value of the proverbial “apple a day.” Public health outlets are emphasizing vegetables before fruit intake and discouraging the intake of certain sources of fruit, such as fruit juice and dried, tropical, and canned fruit, some of which have been reflected in health policies. , , , ,

Given the longstanding perceived value of fruit and vegetables in reducing global CVD morbidity and mortality and in light of developing efforts to limit dietary sugars, there is a need to reassess the role of different fruit and vegetable sources in CVD prevention. Whether different fruit and vegetable sources show comparable CVD risk reduction is unclear. Systematic reviews and meta‐analyses of prospective cohort studies have shown evidence of a cardiovascular benefit of broad categories of fruits and vegetables, , , , , , , but the relative contributions of specific fruit and vegetable sources and the certainty of the estimates for these sources are underexplored. We, therefore, conducted a systematic review and meta‐analysis of prospective cohort studies using Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach to assess the role of different fruit and vegetable sources in CVD risk reduction and to quantify the certainty of the evidence to inform public health policy.

METHODS

All supporting data are available within the article and its online supplementary files. We followed the Cochrane Handbook for Systematic Reviews and Interventions and reported results in accordance with Meta‐Analysis of Observational Studies in Epidemiology and Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines. The protocol was registered at Clini​caltr​ials.gov (identifier, NCT03394339).

Search Strategy

We searched MEDLINE, EMBASE, and the Cochrane Library databases through June 3, 2019, using the search strategy presented in Table S1 and restrictions for prospective cohorts. We supplemented the search with manual searches of the references of included studies.

Study Selection

Prospective cohort studies that reported the association of fruit and/or vegetable intake with CVD, coronary heart disease (CHD), or stroke incidence and mortality with a minimum follow‐up time of 1 year in individuals free of disease at baseline were included. Cohorts that presented data on exposures to fruits and vegetables within the context of a dietary index were not included unless fruits and/or vegetables were presented separately from the other components of the diet index.

Data Extraction

Two reviewers (A.Z., F.A.) independently extracted relevant information, including study design, sample size, subject characteristics, exposure, outcomes, assessment method, dose for each quantile, number of events, population, person‐years of follow‐up, duration of follow‐up, covariates adjustments, and risk ratios (RRs; or odds ratios or hazard ratios) with 95% CIs for each quantile of exposure. We contacted authors for missing data. Data on CVD outcomes were extracted for exposures to total fruits and vegetables, fruits, vegetables, and their sources. Potatoes were not included in the present analysis as they are nutritionally classified as a starchy food and are largely omitted in quantifications of exposure to vegetables.

Outcomes

Outcomes were CVD, CHD, and stroke incidence and mortality.

Risk of Bias

Included studies were assessed for risk of bias with the Newcastle‐Ottawa Scale (NOS), which awards up to 9 points based on cohort selection (up to 4 points), outcome ascertainment (up to 3 points), and degree of covariate adjustments (up to 2 points with adjustment for age as the primary confounding variable awarded 1 point and adjustment for ≥7/9 secondary confounding variables, including sex, family history, smoking, markers of adiposity, energy intake, physical activity, presence of diabetes mellitus, hypertension [or related medications], and dyslipidemia [or related medications]). Studies achieving ≥7 points were considered high quality. Disagreements in NOS score between the 2 reviewers were resolved by a third reviewer (J.L.S.).

Statistical Analysis

Review Manager version 5.3 (The Nordic Cochrane Centre, Denmark) and STATA version 13.0 (StataCorp, TX) were used to conduct all analyses. We prespecified in our analysis plan the use of the generic inverse variance method with DerSimonian and Laird random effects models to pool the natural log‐transformed RRs of extreme quantiles, comparing the highest versus the lowest (reference) exposures. On the basis of a deviation from our prespecified analysis plan requested by the statistical reviewer, we present the generic inverse variance with fixed effects models as the primary analysis and the DerSimonian and Laird random effects models as a secondary analysis in the Supplemental Material. Hazard ratios and odds ratios (as cumulative incidence <10%) were considered equivalent to RR. Studies that provided RR on a continuous scale (ie, per dose increment) were scaled to the highest quantile reported for the exposure in the respective cohort as necessary. Test for differences between fruit and vegetable categories were conducted in RevMan, with a test for subgroup differences, with P<0.05 indicating a significant difference between fruit categories or vegetable categories on a given outcome. We also conducted a dose‐response analysis. A random‐effects linear dose‐response was modeled using a generalized least square trend (glst) for estimation of summarized dose‐response data, as per Greenland and Longnecker and Orsini. A 2‐stage multivariate random‐effects method was used to model a nonlinear association using restricted cubic splines with 3 knots. A Wald test was used to evaluate linear and nonlinear dose‐response trends. The median dose of each quantile was used, and when not provided we chose the midpoint of the upper and lower boundaries for each quantile as the assigned dose. For open‐ended lower and upper quantiles, we defined lowest and highest boundary as the same as the adjacent category cutoff. Servings per day were calculated, with one serving defined as 80 g of fruits and/or vegetables and their categories, with the exception of citrus fruit (122 g), fruit juice (125 g), and green leafy vegetables (88 g), or unless otherwise specified. Heterogeneity was assessed by the Cochran Q statistic and quantified by the I2 statistic. An I2≥50% and PQ<0.1 was considered evidence of substantial heterogeneity. , Sensitivity analyses and a priori subgroup analyses were used to explore sources of heterogeneity. We performed sensitivity analyses by systematically removing each study with recalculation of the summary estimates. A priori subgroup analyses were conducted for all comparisons with ≥10 observations. Subgroup analyses included age (less than median versus median or greater), sex (males, females, and mixed), follow‐up years (less than median versus median or greater), number of covariates in extracted model (<8 versus ≥8 covariates), exposure assessment tool (validated Food Frequency Questionnaire [FFQ], unvalidated FFQ, and food record), risk of bias score (<6 versus ≥6), and country of data collection. Wald test in metaregression was used to assess differences within each subgroup. Because of the exploratory intent of our subgroup analyses, we did not prespecify adjustment for the false discovery rate in our prespecified analysis plan. On the basis of a deviation from our prespecified analysis plan requested by the statistical reviewer, we adjust for the false discovery rate in our subgroup analyses using the Holm‐Bonferroni procedure. If ≥10 cohort comparisons were available, then publication bias was assessed by visual inspection of funnel plots for asymmetry and formal testing with the Begg and Egger tests. If publication bias was suspected (P<0.10), the Duval and Tweedie trim and fill method imputed missing study data in attempt to adjust for funnel plot asymmetry.

Grading the Evidence

The GRADE method was used to assess the certainty of the evidence for each comparison on a 4‐point scale, ranging from “very low” to “high.” , , , , , , , , , , , , Because of their inherent limitations, observational studies start at a “low” certainty of evidence that can be downgraded or upgraded based on established criteria. Criteria to downgrade included risk of bias (weight of studies shows high risk of bias by NOS), inconsistency (substantial unexplained heterogeneity, I2>50%, and PQ<0.10), indirectness (presence of factors that limit generalizability based on populations, exposures, and outcomes), imprecision (95% CIs cross minimally important difference of 5% [RR, 0.95–1.05]), and publication bias (significant evidence of small study effects). Criteria to upgrade included a large risk estimate (RR <0.5 or >2 in the absence of plausible confounders), a dose‐response gradient, and attenuation by plausible confounders.

Results

Flow of the Literature

Figure 1 illustrates a flow of the literature. Of 4271 reports, we included a total of 117 publications , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , of 81 unique prospective cohort studies of 4 031 896 individuals and 125 112 cardiovascular events.

Figure 1

Summary of evidence search and selection.

CV indicates cardiovascular.

Study Characteristics

The Table shows the characteristics of the included studies. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Participants were from 69 countries with cohorts distributed worldwide (36 from Europe, 23 from North America, 1 from South America, 17 from Asia, 4 from Australia, and 1 large global cohort including 18 countries worldwide). The median participant age at baseline was 55 (range, 7–90) years with a median follow‐up of 11 (range, 2–37) years. Median (range) intakes in servings per day in the highest quantiles were 7.4 (2.6–10.4) fruits and vegetables, 2.6 (0.29–11.0) fruits, 2.85 (0.74–11.0) vegetables, 0.4 (0.3–0.5) bananas, 0.27 (0.13–0.7) berries, 0.71 (0.22–2.2) citrus fruit, 0.82 (0.4–2.28) fruit juice, 0.95 (0.29–2.0) pommes, 2.37 (2.1–2.65) watermelon, 0.54 (0.07–2) allium vegetables, 9.5 (5–14) carrots, 0.43 (0.1–3.0) cruciferous vegetables, 0.71 (0.25–1.5) green leafy vegetables, and 0.63 (0.29–2.0) tomatoes. Doses were not available for apricots and celery. Dietary intake was assessed by self‐administered validated food frequency questionnaire (54%), interview administered validated FFQ (10%), unvalidated FFQ (19%), or 24‐hour recalls/food records (17%).

Table 1

Table of Study Characteristics

Study	Cohort	Country	Participants (Men:Women)	Age, y	Follow‐Up, y	Dietary Assessment	Exposure	Quantiles	Outcomes	Incidence (Men:Women)
Adriouch, 2018 41	NutriNet‐Sante	France	84 158 (17 931:66 227)	44.1±14.5	4.9±1.6	24‐h recall	Fruit category	3	CVD incidence CHD incidence Stroke incidence	602 309 293
Appleby, 2002 42	Health Food Shoppers	United Kingdom	10 741 (4325:6416)	16–89	18–24	Unvalidated FFQ	Fruit	2	CVD mortality CHD mortality Stroke mortality	1202 (591:611) 605 (347:258) 356 (142:214)
Atkins, 2014 43	British Regional Heart	England	3328 (3328:0)	60–79	11.3	Validated FFQ	Fruit and/or vegetable	2	CVD incidence CVD mortality CHD incidence	582 327 307
Bahadoran, 2017 44	Theran Lipid and Glucose	Iran	2369 (1047:1322)	≥19	6	Validated FFQ	Vegetable categories	3	CVD risk	79
Bazzano, 2002 45	National Health and Nutrition Examination Survey Epidemiologic Follow‐up Study	United States	9608	25–74	19	Unvalidated FFQ	Fruit and vegetable	4	CVD mortality CVD incidence IHD mortality IHD incidence Stroke mortality Stroke incidence	1145 N/A 639 1786 218 888
Belin, 2011 46	WHI‐OS (Women's Health Initiative Observational Study)	United States	93 676 (0:93 676)	50–79	10	Self‐administered validated FFQ	Fruit, vegetable	2	CVD incidence	6006
Bendinelli, 2011 47	EPIC	Italy	29 689 (0:29 689)	50.0±7.9	7.85	Validated FFQ	Fruit, vegetable, categories	4	CHD incidence	144
Berard, 2017 48	MONICA	France	1311	35–64	16–18	Food recall	Fruit, vegetable	5	CVD mortality	41
Bhupathiraju, 2013 49	NHS (Nurses' Health Study) and HPFS (Health Professionals Follow‐Up Study)	United States	113 276 (42 135:71 141)	40–75 (men) 30–55 (women)	22 (men) 24 (women)	Validated FFQ	Fruit and/or vegetable, categories	5	CHD incidence	6189 (3607:2582)
Bingham, 2008 50	EPIC	United Kingdom	11 134	45–75	4	Validated FFQ	Fruit and vegetable	5	IHD risk	678
Blekkenhorst, 2017 51	PLSAW (Perth Longitudinal Study of Aging Women)	Australia	1226 (0:1226)	75.1±2.7	15	Validated FFQ	Vegetable, categories	Per 5–75 g/d	CVD mortality IHD mortality Stroke mortality	238 128 92
Bos, 2014 52	Rotterdam Study	The Netherlands	3570 (1405:2165)	69.4±6.3	12.9	Unvalidated FFQ	Fruit and vegetable	3	Stroke risk	545
Buijsse, 2008 53	Zutphen Elderly Study	The Netherlands	559 (559:0)	65–84	15	Unvalidated FFQ	Vegetable category	Per 1‐SD increase	CVD mortality	197
Buil‐Cosiales, 2016 55	PREDIMED (Prevención con DIeta Mediterránea)	Spain	7216	55–80	6	Validated FFQ	Fruit, vegetable, categories	5	CVD composite score CVD mortality MI incidence Stroke incidence	342 104 118 169
Buil‐Cosiales, 2017 54	SUN (Seguimiento University of Navarra)	Spain	17 007 (6633:10 374)	38	10.3	Validated FFQ	Fruit, vegetable, categories	5	CVD incidence	112
Cassidy, 2012 56	NHS	United States	69 622 (0:69 622)	30–55	14	Validated FFQ	Fruit, vegetable	5	Stroke incidence	1803
Collin, 2019 57	REGARDS (Reasons for Geographic and Racial Differences in Stroke)	United States	13 440 (7972:5469)	≥45	6±1.8	Validated FFQ	Fruit category	12 oz/d	CHD mortality	168
Conrad, 2018 58	NHANES	United States	29 133 (13 926:15 207)	46.3 (95% CI, 45.8–46.7)	6.5	24‐h recall	Vegetable	3	CVD mortality CHD mortality	726 556
Dauchet, 2004 59	PRIME	France, North Ireland	8087 (8087:0)	50–59	5	Interview	Fruit category	3	CHD event	133
Dauchet, 2010 60	PRIME	France, North Ireland	8060 (8060:0)	50–59	10	Interview‐validated FFQ	Fruit and/or vegetable	3	CVD risk Acute coronary syndrome	612 367
Du, 2016 61	China Kadoorie Biobank	China	451 665 (186 086:265 579)	50.5±10.4	N/A, ≈7.14 y	Interview unvalidated FFQ	Fruit	5	Acute coronary event Hemorrhagic stroke event Other CeVD events Ischemic stroke	2551 14 579 11 054 3523
Du, 2017 62	China Kadoorie Biobank	China	462 342 (189 560:272 782)	51±10.5	≈7	Interview unvalidated FFQ	Fruit	4	CVD mortality IHD mortality Ischemic stroke mortality Hemorrhagic stroke mortality	6166 2038 585 2351
Elwood, 2013 63	Carphilly Cohort Study	United Kingdom	2235 (2235:0)	45–59	30	Unvalidated FFQ	Fruit and vegetable	2	CVD incidence	N/A
Eriksen, 2015 64	SABRE (Southhall and Brent Revised)	United Kingdom	2096	40–69	21	Validated FFQ	Fruit, vegetable	2	CVD incidence CHD incidence	571 520
Fitzgerald, 2012 65	Women's Health Study	United States	34 827 (0:34 827)	55 (46–68) (mean [95% CI])	14.6	Validated FFQ	Fruit, vegetable	5	CVD risk	1094
Fraser, 1992 66	Adventis Health Study	United States	26 473 (10 003:16 740)	Men: 51.3±16.0 Women: 53.2±16.6 (mean±SD)	6	Validated FFQ	Fruit	3	CHD mortality CHD event	463 134
Gardener, 2011 67	NOMAS (Northern Manhattan Study)	United States	2568 (924:1644)	69±10 (Mean±SD)	9	Interview validated FFQ	Fruit, vegetable	Continuous	CVD mortality CVD incidence MI incidence Ischemic stroke incidence	314 518 133 171
Gaziano, 1995 68	Massachusetts Health Care Panel Study	United States	1299 (494:805)	≥66	4.75	Unvalidated FFQ	Fruit and vegetable categories	2	CVD mortality	161
Genkinger, 2004 69	Odyssey	United States	6151 (2276:3875)	30–93	13	Validated FFQ	Fruit, vegetable categories	5	CVD mortality	378
Gillman, 1995 70	Framingham Study	United States	832 (832:0)	45–65	18–22	24‐h recall	Fruit and vegetable	5	Stroke mortality Stroke incidence	14 97
Goetz, 2016 71	REGARDS	United States	16 678	≥45	6.0±1.9	Validated FFQ	Fruit categories	5	CHD events	589
Goetz, 2016 72	REGARDS	United States	20 024 (9011:11 013)	≥45	6.5	Validated FFQ	Fruit, vegetable	5	Stroke incidence	524
Gunge, 2017 73	Danish Diet, Cancer and Health Cohort	Denmark	57 053 (25 759:28 809)	50–64	13.6	Validated FFQ	Fruit and vegetable categories	2	MI incidence	2322 (1669:653)
Gunnell, 2013 74	Health and Wellbeing Surveillance System	Australia	14 890 (6114:8776)	45–97	6	Validated FFQ	Fruit and vegetable	2	IHD hospitalization	538
Hansen, 2010 76	Danish Diet, Cancer and Health	Denmark	53 383 (25 065:28 318)	50–64	7.7	Validated FFQ	Fruit, vegetable categories	4	Acute coronary syndrome	1075 (820:255)
Hansen, 2017 75	Danish Diet, Cancer and Health	Denmark	55 338	50–64	13.5	Validated FFQ	Fruit and vegetable categories	2	Stroke incidence	2283
Harriss, 2007 77	Melbourne Collaborative	Australia	40 653 16 673:23 980	40–69	10.4	Validated FFQ	Fruit, vegetable	4	CVD mortality IHD mortality	697 407
Hertog, 1997 78	Caerphilly Prospective Study	South Wales	1900 (1900:0)	45–59	14.6	Validated FFQ	Vegetable categories	4	IHD mortality	131
Hirvonen, 2000 80	Finnish Male Smokers in the ATBC Study	Finland	26 497 (26 497:0)	50–69	6.1	Validated FFQ	Fruit category	4	Cerebral infarction Subarachnoid hemorrhage Intracerebral hemorrhage	736 83 95
Hirvonen, 2001 79	Finnish Male Smokers in the ATBC Study	Finland	25, 373 (25, 373:0)	50–69	6.1	Validated FFQ	Fruit, vegetable, categories	5	CHD mortality MI event	815 1122
Hjartaker, 2015 81	Migrant Study	Norway	9766 (9766:0)	42–73	20.3	Unvalidated FFQ	Fruit and/or vegetable, categories	4	CVD mortality CHD mortality Stroke mortality	4595 2386 1034
Hodgson, 2016 82	Australian Women aged 70–85 y	Australia	1456 0:1456	>70	15	Validated FFQ	Fruit category	3	CVD mortality	235
Holmberg, 2009 83	Swedish National Farm Register	Sweden	1738 (1738:0)	50±6.0	12	Unvalidated FFQ	Fruit and vegetable	2	CHD incidence	138
Iso, 2007 84	Japan Collaborative Cohort	Japan	N/A	40–79	N/A	Validated FFQ	Fruit or vegetable categories	3	IHD mortality CeVD mortality	N/A N/A
Jacques, 2015 85	Framingham Offspring	United States	2880 (1302:1578)	28–62 (mean=54)	14.9	Validated FFQ	Fruit categories	3	CVD incidence CHD incidence	518 261
Johnsen, 2003 86	Danish Diet, Cancer and Health	Denmark	54 506	50–64	3.09	Validated FFQ	Fruit and/or vegetable	5	Stroke incidence	266
Joshipura, 1999 87	NHS and HPFS cohorts	United States	114 279 (38 683:75 596)	30–55 (men) 40–75 (women)	8 (men) 14 (women)	Validated FFQ	Fruit and/or vegetable, categories	5	Ischemic stroke incidence	570 (366:204)
Joshipura, 2009 88	NHS and HPFS cohorts	United States	109 788 (38 918:70 870)	30–55 (men) 40–75 (women)	14–16	Validated FFQ	Fruit and/or vegetable, categories	5	CVD incidence	3892
Keli, 1996 89	Zutphen Elderly	The Netherlands	552 (552:0)	50–69	15	Interview	Fruit, vegetable categories	3	Stroke risk	42
Kim, 2013 90	British Women's Heart and Health Study	United Kingdom	3080 (0:3080)	60–79	7	Unvalidated FFQ	Fruit	2	CVD incidence	329
Knekt, 1994 93	Finnish Mobile Clinic Health	Finland	5133 (2748:2385)	30–69	14	Interview unvalidated FFQ	Fruit, vegetable	3	CHD mortality	244 (186:58)
Knekt, 1996 92	Finnish Mobile Clinic Health	Finland	5133 (2748:2385)	30–69	26	Interview unvalidated FFQ	Fruit or vegetable categories	4	CHD death	473 (324:149)
Knekt, 2000 91	Finnish Mobile Clinic Health	Finland	9208	≥15	28	Interview unvalidated FFQ	Fruit or vegetable categories	5	CeVD incidence	824
Kobylecki, 2015 94	Copenhagen City Heart	Denmark	78 527	20–100	10	Self‐reported unvalidated FFQ	Fruit and vegetable	3	IHD incidence	2823
Kondo, 2019 95	NIPPON DATA80	Japan	9115 (4002:5113)	30–79	29	3‐d food record	Fruit or vegetable	3	CVD mortality	1070
Kvaavik, 2010 96	Health and Lifestyle Survey	United Kingdom	4866 (2509:2377)	43.7±16.3	20	Interview Unvalidated FFQ	Fruit and vegetable	2	CVD mortality	431
Lai, 2015 97	UK Women's Cohort	United Kingdom	30 458 (0:30 458)	35–69	16.7	Validated FFQ	Fruit, categories	5–6	CVD mortality CHD mortality Stroke mortality	286 138 148
Larsson, 2009 98	Finnish Male Smokers in the ATBC Study	Finland	26 556 (26 556:0)	50–69	13.6	Validated FFQ	Fruit, vegetable	5	Stroke incidence	2702
Larsson, 2013 99	Swedish Mammography and Swedish Men Cohorts	Sweden	74 961 40 291:34 670	45–83	10.2	Validated FFQ	Fruit and/or vegetable, categories	5	Stroke incidence	4089
Leenders, 2013 101	EPIC	Europe (10 countries)*	451 151 129 882:321 269	25–70	12.8 (median)	Validated FFQ and 7‐d food record	Fruit and/or vegetable	4	CVD mortality	5125
Leenders, 2014 100	EPIC	Europe (10 countries)*	451 151 (129 882:321 269)	25–70	13	Validated FFQ and 7‐d food record	Fruit and/or vegetable	4	CHD mortality Stroke mortality	2139 1291
Lin, 2007 102	NHS	United States	66 360 (0:66 360)	30–55	12	Validated FFQ	Fruit, vegetable, categories	5	CHD mortality MI event	324 938
Lin, 2017 103	Survey of Health & Living Status of the Elderly	Taiwan	4176	≥50	11	Interview FFQ	Fruit and vegetable	2	CVD mortality	N/A
Liu, 2000 105	Women's Health Study	United States	39 127 (0:39 127)	45–89	5	Validated FFQ	Fruit and/or vegetable	5	CVD incidence MI event	418 126
Liu, 2001 104	The Physician's Health Study	United States	15 520 (15 520:0)	40–84	6	Validated FFQ	Vegetable	5	CHD incidence	1148
Mann, 1997 106	The Oxford Vegetarian Study	United Kingdom	10 802 (4102:6700)	16–79	13.3	Validated FFQ	Fruit, vegetable, categories	3	IHD mortality	64
Manuel, 2015 107	Canadian Community Health Survey	Canada	82 259 (37 483:44 746)	20–83 (men: 48.2; women: 49.4)	8.6	Interview FFQ	Fruit and vegetable	3	Stroke incidence	1551
Miller, 2017 108	PURE (Prospective Urban and Rural Epidemiology)	18 Countries†	135 335	35–70	7.4 (Median)	Validated FFQ	Fruit and/or vegetable	4	CVD events MI Stroke Cardiovascular mortality	4784 N/A N/A N/A
Mink, 2007 109	Iowa Women's Health	United States	34 492 (0:34 492)	55–69	16	Validated FFQ	Fruit or vegetable categories	3	CVD mortality CHD mortality Stroke mortality	2316 1329 469
Mizrahi, 2009 110	Finnish Mobile Clinic Health Examination Survey	Finland	3932	40–74	24	Interview	Fruit, vegetable, categories	4	Stroke risk	625
Mori, 2018 111	Japan Public Health Center Based Prospective Study	Japan	88 184 (40 622:47 562)	45–74	16.9	Validated FFQ	Vegetable categories	5	CHD mortality Stroke mortality	1968 (1192:776) 1470 (856:614)
Mytton, 2018 112	EPIC‐Norfolk	England	22 992 (10 002:12 990)	40–79	16.4	7‐d food record	Fruit and vegetable	5	CVD incidence	4965
Nagura, 2009 113	Japan Collaborative	Japan	59 485 (25 206:34 279)	40–79	12.7	Validated FFQ	Fruit, vegetable	4	CVD mortality CHD mortality Stroke mortality	2243 (1207:1036) 452 (258:194) 1053 (559:494)
Nakamura, 2008 114	Takayama	Japan	29 079 (13 355:15 724)	≥35 (men: 54.0; women: 55.1)	7.33	Validated FFQ	Fruit, vegetable	4	CVD mortality	384 (200:184)
Nechuta, 2010 115	Shanghai Women's Health	China	71 243 (0:71 243)	40–70	9	Interview Validated FFQ	Fruit and vegetable	Daily	CVD mortality	775
Neelakantan, 2018 116	Singapore Chinese Health Study	China	57 078	45–74	17	Validated FFQ	Fruit or vegetable	1 Serving/d	CVD mortality	4871
Ness, 2005 117	Boyd Orr Cohort	United Kingdom (England and Scotland)	4028 (1995:2033)	3.5–11.2	37	Household survey	Fruit, vegetable	4	CHD mortality Stroke mortality	298 83
Nothlings, 2008 118	EPIC	Europe (10 countries)†	10 262	35–70	9	Validated FFQ	Fruit or vegetable	80 g/d	CVD mortality	517
Okuda, 2015 119	NIPPON DATA80	Japan	9112 (4000:5112)	30–79	24	Household survey	Fruit and/or vegetable	4	CVD mortality CHD mortality Stroke mortality	823 165 385
Oude Griep, 2010 120	MORGEN	The Netherlands	19 819	20–59	10.5	Validated FFQ	Fruit, vegetable	4	CHD incidences	245
Oude Griep, 2011 122	MORGEN	The Netherlands	20 069 (8988:11 081)	20–68	10.3	Validated FFQ	Fruit and vegetable	4	Stroke incidence	233
Oude Griep, 2011 121	MORGEN	The Netherlands	20 069 (8989:11 081)	42±11	10.5	Validated FFQ	Fruit or vegetable categories	3–4	CHD incidence	245
Oyebode, 2014 123	HSE (Health Survey for England)	England	65 226 (28 960:36 266)	56.6±14.3	7.7	24‐h recall	Fruit and/or vegetable	4	CVD mortality	1554
Pham, 2007 124	Miyako Study	Japan	9651 (4254:5397)	Men: 56.5±10.63; women: 57.4±10.89 (mean±SD)	13.8	Questionnaire	Fruit, vegetable	2	Stroke mortality	226
Rebello, 2014 125	Singapore Chinese Health Study	China	53 469 (23 501:29 968)	45–7	15	Interview validated FFQ	Fruit and vegetable	5	IHD mortality	1660 (1022:638)
Rissanen, 2003 126	Kuopio Ischaemic Heart Disease Risk Factor	Finland	1950 (1950:0)	42–60	12.8	4‐d food record	Fruit and vegetable	5	CVD mortality	115
Saglimbene, 2017 127	DIET‐HD	Europe and South America	9757	N/A	1.5	Validated FFQ	Fruit, categories	2	CVD mortality	N/A
Sahyoun, 1996 128	Nutrition Status Study	United States	680	60–101	9–12	3‐d food record	Fruit, vegetable, categories	3	CHD mortality	101
Sauvaget, 2003 129	Life Span Study	Japan	39 337 (14 966:23 471)	34–103	16	Validated FFQ	Fruit, vegetable, categories	3	Stroke mortality	1926 (692:1234)
Scheffers, 2019 130	EPIC Netherlands and MORGEN	The Netherlands	34 560 (25 574:8986)	20–69	14.6	Validated FFQ	Fruit and categories	5	CVD incidence CHD incidence Stroke incidence	3801 2135 1135
Sesso, 2003 131	WHS (Women's Health Study)	United States	38 445 (0:38 445)	45–89	6.9	Validated FFQ	Fruit or vegetable categories	4	CVD incidence	729
Sesso, 2003 133	WHS	United States	38 445 (0:38 445)	≥45	7.2	Validated FFQ	Vegetable categories	5	CVD incidence MI incidence Stroke incidence	729 201 247
Sesso, 2007 132	WHS	United States	38 176 (0:38 176)	54.5	10.1	Validated FFQ	Fruit category	4	CVD mortality CVD incidence MI incidence Stroke incidence	204 1004 289 339
Shah, 2018 134	Cooper Center Longitudinal Study	United States	11 376 (8577:2799)	47	18	3‐d food record	Fruit or vegetable	Continuous	CVD mortality	249
Sharma, 2013 135	Multi Ethnic Cohort	United States	174 028 (78 410:95 618)	45–75	7.5	Validated FFQ	Fruit, vegetable	5	Stroke mortality	860 (434:426)
Sharma, 2014 136	Multi Ethnic Cohort	United States	164 617 (72 866:91 751)	45–75	5–8	Validated FFQ	Fruit, vegetable	5	IHD mortality	1951 (1140:811)
Simila, 2013 137	ATBC	Finland	21 955 (21 955:0)	50–69	19	Validated FFQ	Fruit, fruit juices	Daily	CHD risk	4379
Sonestedt, 2015 138	Malmo Diet and Cancer	Sweden	26 445 (10 048:16 397)	44–74	14	Validated FFQ	Fruit, vegetable	5	CVD incidence CHD incidence Stroke incidence	2921 N/A N/A
Sotomayor, 2018 139	Renal Transplant Recipients	The Netherlands	400 (217:183)	52±12	7.2	Unvalidated FFQ	Fruit	3	CVD mortality	49
Steffen, 2003 140	ARIC (Atherosclerosis Risk in Communities)	United States	11 940 (5271:6669)	45–64 (men: 54.4±5.7; women: 54.1±5.7)	11	Interview validated FFQ	Fruit and vegetable	5	CHD incidence Ischemic stroke incidence	535 214
Stefler, 2016 141	HAPIEE (Health, Alcohol and Psychosocial Factors in Eastern Europe)	Poland, Russia, Czech Republic	19 263	57	7.1	Validated FFQ	Fruit and/or vegetable	4	CVD mortality CHD mortality Stroke mortality	438 226 109
Strandhagen, 2000 142	Men Born in 1913	Sweden	730 (730:0)	54	26	Interview unvalidated FFQ	Fruit, vegetable	5	CVD mortality CVD incidence	226 209
Takachi, 2008 143	Japan Public Health Center Based Prospective Study	Japan	77 891 (35 909:41 982)	45–74	5.9	Validated FFQ	Fruit and/or vegetable, categories	4	CVD incidence	1386 (830:556)
Tanaka, 2013 144	Japan Diabetes Complications Study	Japan	1414	40–70	8.1 (Median)	Validated FFQ	Fruit and vegetable	4	CHD incidence Stroke incidence	96 68
Tognon, 2014 145	MONICA	Denmark	1849 (901:948)	30–59	11	Food record	Fruit, vegetable	2	CVD mortality CVD incidence MI mortality MI incidence Stroke mortality Stroke incidence	223 755 64 161 40 167
Tucker, 2005 146	Baltimore Longitudinal Study of Aging	United States	501 (501:0)	34–80	18	7‐d food record	Fruit and/or vegetable	2	CHD mortality	71
Von Ruesten, 2013 147	EPIC	Germany	23 531 (9098:14 433)	35–65	8	Validated FFQ	Fruit, vegetable, categories	Daily	CVD incidence	363
Vormund, 2015 148	MONICA	Switzerland	17 861 (8663:9198)	16–92	21.4 (Mean)	24‐h recall	Fruit, vegetable	Daily	CVD mortality	1385
Wang, 2016 149	Linxian Nutrition Intervention Trials	China	2455 (1105:1340)	40–69	19–26	Unvalidated FFQ	Fruit and/or vegetable, categories	2	CHD mortality Stroke mortality	355 (men) 452 (women)
Watkins, 2000 150	CPS‐11 (Cancer Prevention Study 11)	United States	1 063 023 (453 962:609 061)	≥30	7	Unvalidated FFQ	Vegetable		CHD mortality	13 761 (9156:4605)
Whiteman, 1999 151	OXCHECK	United Kingdom	10 522 (4929:5593)	35–64	9	Unvalidated FFQ	Fruit, green vegetables	2–3	IHD mortality	144
Yamada, 2011 152	Jidni Medical School Cohort	Japan	10 623 (4147:6476)	N/A	10.7	Validated FFQ	Fruit category	5	CVD event MI event Stroke event	758 (270:488) 565 (383:182) 99 (76:23)
Yokoyama, 2000 153	Shibata Study	Japan	2121 (880:1241)	≥40	20	Unvalidated FFQ	Fruit, vegetable	3	Stroke incidence	196 (91:105)
Yoshizaki, 2019 154	Japan Public Health Centre Based Prospective Study	Japan	16 498 (7726:8772)	45–74	14	Validated FFQ	Fruit and/or vegetable	3	CHD incidence Stroke incidence	839 197
Yu, 2014 155	Shanghai Men and Women's Health Study	China	122 635 (55 424:67 211)	40–74	5.4–9.8	Interview validated FFQ	Fruit and/or vegetable, categories	4	CHD incidence	365 (217:148)
Zhang, 2011 156	Shanghai Men and Women's Health Study	China	134 796 (61 436:73 360)	40–74 40–70	4.5 10.2	Interview validated FFQ	Fruit, vegetable, and categories	5	CVD mortality	5393 (1951:3442)
Zhang, 2011 157	MONICA	Finland	36 686 (17 287:19 399)	25–74	13.7	24‐h recall	Fruit, veg	4	Stroke incidence	1478

ATBC, alpha‐tocopherol, beta‐carotene cancer prevention; CeVD indicates cerebrovascular disease; CHD, coronary heart disease; CVD, cardiovascular disease; DIET‐HD, dietary intake, death and hospitalization in adult with early stage kidney disease treated with hemodialysis; EPIC, European Prospective Investigation Into Cancer and Nutrition; FFQ, Food Frequency Questionnaire; IHD, ischemic heart disease; MI, myocardial infarction; MONICA, monitoring of trends and determinants in cardiovascular disease; MORGEN, monitoring project on risk factors for chronic diseases; N/A, Not Available; NHANES, National Health and Nutrition Examination Survey; NIPPON DATA80, National Integrated Project for Prospective Observation of non‐communicable disease and its trends in aged; and OXCHECK, Oxford and Collaborators health check.

The EPIC cohort represented the following countries: France, Germany, Greece, Italy, the Netherlands, Spain, United Kingdom, Sweden, Denmark, and Norway.

The PURE cohort represented the following countries: high income: Canada, Sweden, United Arab Emirates; middle income: Argentina, Brazil, Chile, China, Colombia, Iran, Malaysia, Poland, South Africa, Turkey, Occupied Palestinian Territory; low income: Bangladesh, India, Pakistan, Zimbabwe.

Table S2 lists the variables that were statistically adjusted in the included studies. Age, the prespecified primary confounding variable, was adjusted for in 95% of included studies, of which 55% also adjusted for all 9 of the prespecified secondary confounding variables.

Study Quality

Table S3 summarizes the NOS assessment of included studies. There was a high risk of bias in associations between fruit juice, cruciferous, green leafy, and tomato vegetables, and CHD and stroke mortality and citrus and stroke mortality as >35% of the pooled risk estimate was derived from Iso et al, which was scored 5 on the NOS. The association between apricots and CVD mortality was derived from one study, Saglimbene et al, which was scored 1 on the NOS. Although most studies had scores reduced because of self‐administered ascertainment of exposure, 88% of studies received a total score ≥6, which was considered high quality.

Cardiovascular Disease

CVD Incidence

Figure 2 and Figures S1 through S11 show the relation of total and specific fruit and vegetables with CVD incidence. We found a lower risk associated with the highest versus the lowest intakes of fruits and vegetables (RR, 0.93 [95% CI, 0.89–0.96], no significant heterogeneity), fruits (RR, 0.91 [95% CI, 0.88–0.95], no significant heterogeneity), and vegetables (RR, 0.94 [95% CI, 0.90–0.97], no significant heterogeneity). Figures S12 and S13 summarize the relation of sources of fruit or vegetables with CVD incidence. A significant interaction by fruit source was observed (P<0.001), with significant associations with lower risk limited to citrus (RR, 0.88 [95% CI, 0.80–0.86], no significant heterogeneity) and pommes (RR, 0.76 [95% CI, 0.66–0.88], no significant heterogeneity). We found no significant associations from the highest versus lowest intakes of berries (RR, 1.27 [95% CI, 0.95–1.71], heterogeneity not applicable) and juice (RR, 1.00 [95% CI, 0.93–1.07], no significant heterogeneity) fruit. No interaction by vegetable source was observed (P=0.227).

Figure 2

Relation between intake of fruits and vegetables and total incident cardiovascular disease (CVD) (highest vs lowest level of intake).

Pooled risk estimates are represented by the black diamond, with principal exposures highlighted in bold. Principal exposures (fruits and vegetables, fruits, and vegetables) represent the pooled data of the risk estimates reported for these exposures and were not tabulated by pooling fruit and vegetable varieties. Values of I2≥50% indicate substantial heterogeneity, with significance at P>0.10. The mean important difference of 5% change in relative risk, indicating a clinically relevant association with lower or higher risk, is indicated by the dashed gray lines. CHD indicates coronary heart disease; GRADE, Grading of Recommendations Assessment, Development, and Evaluation; and RR, risk ratio.

CVD Mortality

Figure 3 and Figures S14 through S30 show the relation of total and specific fruit and vegetables with CVD mortality. We found a lower risk associated with the highest versus the lowest intakes of fruits and vegetables (RR, 0.89 [95% CI, 0.85–0.93], substantial heterogeneity [I2=68%, P<0.001]), fruits (RR, 0.88 [95% CI, 0.86–0.91], substantial heterogeneity [I2=79%, P<0.001]), and vegetables (RR, 0.87 [95% CI, 0.85–0.90], substantial heterogeneity [I2=59%, P<0.001]). Figures S31 and S32 summarize the association of sources of fruits or vegetables with CVD mortality. A significant interaction by fruit (P=0.001) and vegetable sources (P<0.001) was observed with significant associations with lower risk limited to pommes fruit (RR, 0.86 [95% CI, 0.80–0.92], no significant heterogeneity) and to allium (RR, 0.33 [95% CI, 0.22–0.49], heterogeneity not applicable), cruciferous (RR, 0.85 [95% CI, 0.82–0.89], no significant heterogeneity), and green leafy (RR, 0.87 [95% CI, 0.81–0.94], substantial heterogeneity [I2=88%, P<0.001]) vegetables. There was a significant increased risk with CVD mortality from the highest versus lowest intake of apricots (RR, 1.84 [95% CI, 1.27–2.67], heterogeneity not applicable). We found no significant associations from the highest versus lowest intakes of bananas (RR, 1.06 [95% CI, 0.87–1.29], heterogeneity not applicable), berries (RR, 0.97 [95% CI, 0.92–1.03], no significant heterogeneity), citrus (RR, 0.95 [95% CI, 0.90–1.02], substantial heterogeneity [I2=62%, P=0.049]), juice (RR, 0.81 [95% CI, 0.58–1.13], heterogeneity not applicable), and grapes (RR, 0.90 [95% CI, 0.81–1.01], substantial heterogeneity [I2=61%, P=0.077) fruit and carrots (RR, 0.92 [95% CI, 0.85–1.01], no significant heterogeneity), celery (RR, 0.91 [95% CI, 0.83–1.01], heterogeneity not applicable), and tomato (RR, 0.98 [95% CI, 0.93–1.04], no significant heterogeneity) vegetables.

Figure 3

Relation between intake of fruits and vegetables and cardiovascular mortality (highest vs lowest level of intake).

Pooled risk estimates are represented by the black diamond, with principal exposures highlighted in bold. Principal exposures (fruits and vegetables, fruits, and vegetables) represent the pooled data of the risk estimates reported for these exposures and were not tabulated by pooling fruit and vegetable varieties. Values of I2≥50% indicate substantial heterogeneity, with significance at P>0.10. The mean important difference of 5% change in relative risk, indicating a clinically relevant association with lower or higher risk, is indicated by the dashed gray lines. CHD indicates coronary heart disease; CVD, cardiovascular disease; GRADE, Grading of Recommendations Assessment, Development, and Evaluation; and RR, risk ratio.

Figures S33 through S55 show the dose‐response analyses for total and specific fruit and vegetables and CVD incidence and mortality. A nonlinear model best fit the data for citrus fruit and incident CVD (P=0.033), with a plateau at 0.5 servings/day, total fruits and vegetables with CVD mortality (P<0.001), with a plateau at 4 daily servings, and fruits and CVD mortality (P=0.003), with a plateau in risk reduction after 2 daily servings. An inverse dose‐response gradient was found for the following associations: total fruits and vegetables (RR, 0.97 [95% CI, 0.96–0.99] per serving/day), fruits (RR, 0.97 [95% CI, 0.95–0.99] per serving/day), pommes (RR, 0.87 [95% CI, 0.75–0.99] per serving/day), and green leafy vegetables (RR, 0.72 [95% CI, 0.56–0.93]) with CVD incidence and total fruits and vegetables (RR, 0.72 [95% CI, 0.56–0.93] per serving/day), fruits (RR, 0.92 [95% CI, 0.89–0.96] per serving/day), and vegetables (RR, 0.94 [95% CI, 0.92–0.97] per serving/day) with CVD mortality.

Coronary Heart Disease

CHD Incidence

Figure 2 and Figures S56 through S69 show the relation of total and specific fruit and vegetables with CHD incidence. We found a lower risk associated with the highest versus the lowest intakes of fruits and vegetables (RR, 0.88 [95% CI, 0.83–0.92], no significant heterogeneity), fruits (RR, 0.88 [95% CI, 0.84–0.92], no significant heterogeneity), and vegetables (RR, 0.92 [95% CI, 0.87–0.96], substantial heterogeneity [I2=53%, P=0.002]). Figures S70 and S71 summarize the relation of sources of fruits or vegetables with CHD incidence. No interaction by fruit source was observed (P=0.375). A significant interaction by vegetable sources was seen (P<0.001) with significant associations with lower risk limited to green leafy vegetables (RR, 0.82 [95% CI, 0.76–0.89], no significant heterogeneity). We found no significant associations from the highest versus lowest intakes of allium (RR, 0.93 [95% CI, 0.80–1.09], no significant heterogeneity), cruciferous (RR, 1.01 [95% CI, 0.95–1.07], no significant heterogeneity), and tomato (RR, 0.80 [95% CI, 0.57–1.13], no significant heterogeneity) vegetables.

CHD Mortality

Figure 3 and Figures S72 through S87 show the relation of total and specific fruit and vegetables with CHD mortality. We found a lower risk associated with the highest versus the lowest intakes of fruits and vegetables (RR, 0.81 [95% CI, 0.72–0.92], no significant heterogeneity), fruits (RR, 0.86 [95% CI, 0.82–0.90], substantial heterogeneity [I2=62%, P<0.001]), and vegetables (RR, 0.86 [95% CI, 0.83–0.89], no significant heterogeneity). Figures S88 and S89 summarize the relation of sources of fruits or vegetables with CHD mortality. No significant interaction was found by fruit sources (P=0.144). A significant interaction by vegetable source was seen (P=0.023), with significant associations with lower risk limited to allium (RR, 0.67 [95% CI, 0.57–0.79], substantial heterogeneity [I2=88%, P<0.001]), cruciferous (RR, 0.91 [95% CI, 0.85–0.98], substantial heterogeneity [I2=88%, P<0.001]), and green leafy (RR, 0.86 [95% CI, 0.78–0.94], no significant heterogeneity) vegetables. We found no significant associations from the highest versus lowest intakes of carrots (RR, 0.76 [95% CI, 0.37–1.58], heterogeneity not applicable), celery (RR, 0.92 [95% CI, 0.80–1.06], heterogeneity not applicable), and tomato (RR, 0.92 [95% CI, 0.82–1.04], no significant heterogeneity) vegetables.

Figures S90 through S116 show the dose‐response analyses for fruit and vegetables and CHD incidence and mortality. A nonlinear model best fit the data for citrus fruit (P=0.005) and green leafy vegetables (P=0.004) and incident CHD and total fruits and vegetables and CHD mortality (P=0.044), with plateaus in risk reductions following 0.5, 0.5, and 3 daily servings, respectively. An inverse dose‐response was found in the associations between total fruits and vegetables (RR, 0.97 [95% CI, 0.96–0.98] per serving/day), fruits (RR, 0.96 [95% CI, 0.93–0.99] per serving/day), vegetables (RR, 0.98 [95% CI, 0.95–0.99] per serving/day), and green leafy vegetables (RR, 0.85 [95% CI, 0.76–0.94] per serving/day) with CHD incidence and fruits (RR, 0.94 [95% CI, 0.90–0.97] per serving/day) and vegetables (RR, 0.89 [95% CI, 0.83–0.96] per serving/day) with CHD mortality.

Stroke

Stroke Incidence

Figure 2 and Figures S117 through S127 show the relation of total and specific fruit and vegetables with stroke incidence. We found a lower risk associated with the highest versus the lowest intakes of fruits and vegetables (RR, 0.82 [95% CI, 0.77–0.88], no significant heterogeneity), fruits (RR, 0.82 [95% CI, 0.79–0.85], no significant heterogeneity), and vegetables (RR, 0.88 [95% CI, 0.83–0.93], substantial heterogeneity [I2=50%, P=0.006]). Figures S128 and S129 summarize the relation of sources of fruits or vegetables with stroke incidence. A significant interaction by fruit (P=0.017) and vegetable sources (P=0.044) was observed with significant associations with lower risk limited to citrus (RR, 0.88 [95% CI, 0.82–0.94], substantial heterogeneity [I2=51%, P=0.04]), juice (RR, 0.82 [95% CI, 0.68–0.99], substantial heterogeneity [I2=73%, P=0.02]), and pommes (RR, 0.89 [95% CI, 0.84–0.95], no significant heterogeneity) fruit and to allium (RR, 0.89 [95% CI, 0.80–0.99], no significant heterogeneity), green leafy (RR, 0.88 [95% CI, 0.79–0.98], no significant heterogeneity), and tomato (RR, 0.20 [95% CI, 0.05–0.82], heterogeneity not applicable) vegetables. We found no significant associations from the highest versus lowest intakes of berries (RR, 1.03 [95% CI, 0.94–1.13], substantial heterogeneity [I2=50%, P=0.078]) fruit and cruciferous (RR, 0.98 [95% CI, 0.91–1.05], substantial heterogeneity [I2=62%, P=0.022]) vegetables.

Stroke Mortality

Figure 3 and Figures S130 through S144 show the relation of total and specific fruits and vegetables with stroke mortality. We found a lower risk associated with the highest versus the lowest intakes of fruits and vegetables (RR, 0.73 [95% CI, 0.65–0.81], no significant heterogeneity), fruits (RR, 0.87 [95% CI, 0.84–0.91], substantial heterogeneity [I2=75%, P<0.001]), and vegetables (RR, 0.94 [95% CI, 0.90–0.99], substantial heterogeneity [I2=62%, P=0.001]). Figures S145 and S146 summarize the relation of sources of fruit or vegetables with stroke mortality. A significant interaction by fruit (P<0.001) and vegetable sources (P<0.001) was observed with significant associations, with lower risk limited to citrus (RR, 0.90 [95% CI, 0.86–0.95], substantial heterogeneity [I2=82%, P<0.001]) and juice (RR, 0.67 [95% CI, 0.60–0.76], no significant heterogeneity) fruit and carrots (RR, 0.54 [95% CI, 0.48–0.61], heterogeneity not applicable) and green leafy (RR, 0.90 [95% CI, 0.83–0.97], substantial heterogeneity [I2=50%, P=0.09]) vegetables. We found no significant associations from the highest versus lowest intakes of bananas (RR, 1.04 [95% CI, 0.70–1.54], heterogeneity not applicable), berries (RR, 0.97 [95% CI, 0.82–1.15], no significant heterogeneity), grapes (RR, 0.74 [95% CI, 0.53–1.02], no significant heterogeneity), and pommes (RR, 0.91 [95% CI, 0.77–1.09], no significant heterogeneity) fruit and allium (RR, 0.99 [95% CI, 0.79–1.24], substantial heterogeneity [I2=96%, P<0.001]), cruciferous (RR, 0.92 [95% CI, 0.85–1.01], no significant heterogeneity), and tomato (RR, 1.03 [95% CI, 0.94–1.12], no significant heterogeneity) vegetables.

Figures S147 through S171 show the dose‐response analyses for fruit and vegetables and stroke mortality and incidence. A nonlinear model best fit the data for citrus fruit (P=0.039) and vegetables (P=0.012) and stroke incidence and fruit (P<0.001) and green leafy (P=0.043) vegetables and stroke mortality, with plateaus in risk reductions following 0.5, 1, 2, and >0.7 daily servings, respectively. An inverse dose‐response gradient was found in the associations between total fruits and vegetables (RR, 0.95 [95% CI, 0.92–0.98] per serving/day), fruits (RR, 0.92 [95% CI, 0.88–0.96] per serving/day), citrus fruit (RR, 0.83 [95% CI, 0.69–0.98] per serving/day), pommes (RR, 0.87 [95% CI, 0.79–0.96] per serving/day), green leafy vegetables (RR, 0.88 [95% CI, 0.79–0.97] per serving/day), and tomatoes (RR, 0.67 [95% CI, 0.52–0.87] per serving/day) with stroke incidence and fruits and vegetables (RR, 0.93 [95% CI, 0.88–0.98] per serving/day), fruits (RR, 0.85 [95% CI, 0.78–0.92] per serving/day), vegetables (RR, 0.93 [95% CI, 0.87–0.99] per serving/day), citrus fruit (RR, 0.67 [95% CI, 0.57–0.80] per serving/day), fruit juice (RR, 0.54 [95% CI, 0.36–0.89] per serving/day), carrots (RR, 0.44 [95% CI, 0.28–0.69] per serving/day), and green leafy vegetables (RR, 0.85 [95% CI, 0.73–0.98] per serving/day) with stroke mortality.

Sensitivity Analyses

The systematic removal of each study did not modify the direction or significance of the association estimates or the evidence for heterogeneity (data not shown).

Subgroup Analyses

Figures S172 through S188 illustrate a priori categorical subgroup analyses. There were no statistically significant subgroup differences. Inverse associations were predominately limited to studies with statistical adjustments of ≥8 potential confounders. Confining analyses to studies using validated exposure assessment techniques did not alter the associations. No effect modification was seen by sex, age, follow‐up duration, NOS, or study location.

Publication Bias

Figures S189 through S205 illustrate publication bias analyses for comparisons with at least 10 observations. Visual inspection and formal analysis with the Begg and Egger test did not show evidence of publication bias in any comparison, except for vegetable intake with CVD (PBegg=0.015, PEgger=0.004), CHD (PBegg=0.018, PEgger=0.004), and stroke (PBegg=0.545, PEgger=0.018) mortality and fruit intake with stroke mortality (PBegg=0.820, PEgger=0.031), which were subsequently unsupported by the trim and fill test.

Grading of Recommendations Assessment, Development, and Evaluation

Figures 2 and 3 and Tables S4 through S9 summarize the GRADE assessments. The certainty of the evidence was rated as “moderate” for 11, “low” for 21, and “very low” for 52 of the exposure‐outcome relationships. Our certainty in the evidence was strongest for the associations of total fruits and vegetables with lower risks of CHD incidence and CHD and stroke mortality; fruits with lower risks of CVD, CHD, and stroke incidence; vegetables with lower risks of CHD mortality and stroke incidence; pommes fruit with lower risks of stroke incidence; and green leafy vegetables with lower risks of CHD incidence. The evidence was rated as “moderate” in each case, because of an upgrade for dose‐response gradient in the absence of any downgrades. The associations for specific types of fruits and vegetables were rated largely as “very low,” because of downgrades for imprecision, risk of bias, indirectness, and/or inconsistency. The fixed effects model improved our certainty in the evidence for fruit and CVD incidence by improving precision of the pooled risk estimate. There were no other marked differences between the random effects and fixed effects models.

DISCUSSION

We conducted a systematic review and meta‐analysis of 81 unique prospective cohorts involving 4 031 896 individuals and 125 112 cardiovascular events to assess the relation of total and specific fruit and vegetable consumption on CVD incidence and mortality outcomes. Pooled analyses of highest versus lowest consumption illustrate a lower risk in CVD, CHD, and stroke incidence or mortality by 7% to 27% from total fruit and vegetable intake, 9% to 18% from fruit intake, and 5% to 14% from vegetable intake. Of the specific fruit sources, highest versus lowest intakes of citrus and pommes fruit showed significant risk reductions in most CVD outcomes, from 9% to 12% and from 10% to 24%, respectively, and fruit juice showed a significant risk reduction in stroke incidence and mortality by 18% and 33%, respectively. Most notably of the vegetable categories, one daily serving of green leafy vegetables was associated with 12% to 18% risk reduction in CVD, CHD, and stroke incidence and CHD mortality. There was a consistent linear dose‐response between fruits and vegetables and CHD, with a maximum daily intake of 7 fruit and 7 vegetable servings showing a risk reduction of ≈20% and ≈30% in CHD incidence and mortality, respectively.

Findings in the Context of Existing Literature

Our findings are consistent with those of previous systematic review and meta‐analyses, which also detected inverse associations between fruits and/or vegetables and CVD mortality and incident outcomes. , , Our analyses were in line with those reported most recently by Aune et al, who observed the lowest risk on CVD, CHD, and stroke from maximum intakes of total fruits and vegetables. This is despite our division of CVD outcomes differing significantly, with the present study distinguishing between mortality and incidence data. Our findings on individual fruits and vegetables were also relatively consistent, highlighting a high versus low intake of citrus and pommes fruit, fruit juice, and green leafy vegetables as protective on CVD outcomes, suggesting they may independently play a valuable role in the diet. Nonetheless, the current study benefited from the inclusion of updated and novel large prospective cohorts, namely, the SUN (Seguimiento University of Navarra) and PURE (Prospective Urban and Rural Epidemiology) cohorts, which combined contributed an additional 152 342 individuals and 4896 events to our analyses.

Numerous mechanisms have been proposed to explain the benefits of fruit and vegetable consumption on the cardiovascular system. Perhaps the most supported hypothesis is through their essential contribution to total dietary fiber, an established modifier of CVD risk factors. ,

Fruits with highlighted benefits in the present review tend to be of low glycemic index, a characteristic with demonstrated CVD risk factor reductions. Their consumption has also been associated with improved weight management and decreased prevalence of obesity, a risk factor attributed to 7% to 44% of CVD incidence, likely because of their low energy density and displacement of high calorie foods in the diet. The relationships between the extensive list of micronutrients offered by fruits and vegetables and CVD risk reduction has also been widely explored. They are a key source of antioxidants in the diet, necessary for eradicating free radicals, and may defend against damaging lipid oxidation. Individual sources may offer distinct benefits, such as green leafy vegetables, which are dense in dietary nitrates, a compound linked to reductions in early prognostic markers of CVD. , , Interestingly, however, we did not observe a benefit from high consumption of berries as the most concentrated fruit source of antioxidants. Several vasoactive minerals, such as potassium, magnesium, and calcium, are also obtained from fruits and vegetables in the diet. , , Although each mechanism may be individually biologically plausible, the complexity of the nutrient combinations cannot be underestimated. A whole food approach is necessary to evaluate their efficacy in CVD risk reduction as it can account for additive and multiplicative mechanisms.

Strengths and Limitations

Our systematic review and meta‐analysis has several strengths. It provides a comprehensive synthesis of the available knowledge on consumption of fruits, vegetables, and their varieties and CVD outcomes of importance to public health and clinical practice. We included a systematic search strategy to ensure all published prospective cohort data were identified and used a priori established approaches to explore the pooled risk estimates, including dose‐response analyses. Finally, the certainty of the evidence was assessed using the GRADE approach with the evidence upgraded in several cases for the presence of a protective inverse dose‐response gradient for the association of total fruits and vegetables, fruits, vegetables, and green leafy vegetables with CVD outcomes.

There are also several limitations of our systematic review and meta‐analysis. Although ≈90% of the included prospective cohort studies were of high quality, residual confounding (measured and unmeasured) cannot be ruled out in observational studies. This issue is addressed in the GRADE assessment, which starts observational studies as “low” certainty. We downgraded the certainty of evidence because of imprecision in 55 of the 84 associations as the upper 95% CI crossed the minimal clinically important difference of a 5% reduction in relative risk, from which evidence of harm could not be excluded in 30 associations. Because of limited number of observations, indirectness was also present in several cases and the lack of reported exposures for different tropical fruit limited our exploration of this fruit category. Another source of uncertainty leading to downgrades in the evidence was the presence of high risk of bias in several of the studies that presented data on specific sources of fruits and vegetables. Last, the evidence was downgraded for inconsistency based on the presence of substantial unexplained heterogeneity in 19 of the 84 associations.

Balancing the strengths and limitations, the certainty of the evidence was rated as “very low” to “low” for most of the exposure‐outcome relationships for the association of fruits and vegetables with cardiovascular outcomes. The highest (“moderate”) rated evidence was for the cardiovascular benefit of total fruits and vegetables, fruits, vegetables, pommes fruit, and green leafy vegetables. The least certainty was for other specific fruit and vegetable sources.

Implications

Addressing the low prevalence of adequate fruit and vegetable consumption remains an important global health target. With average intakes of 1 and 1.7 servings of fruit and vegetables per day, respectively, in developed countries, such as the United States, there is an opportunity to increase intakes to meet the established minimum recommendations of 5 daily servings and realize the cardiovascular benefits. We observed a linear dose relationship between fruits and vegetables and CHD and stroke risk, suggesting an increased cardiovascular benefit with additional servings and that targets beyond “5 a day” should also be considered. Successful strategies for increasing fruit and vegetable intake, nevertheless, are lacking and may benefit from emphasizing a larger variety of sources. Our synthesis highlighted that different sources of fruit, including 100% fruit juice, are associated with comparable CVD risk reduction as that of vegetables. Public health guidance to limit the intake of certain fruit sources because of concerns related to their contribution to sugars may have unintended harm in preventing people from meeting fruit and vegetable targets for CVD risk reduction.

Conclusions

Current evidence supports the role of a variety of fruits and vegetables for CVD prevention. Higher intakes of fruits and/or vegetables are associated with improvements in all CVD outcomes, with fruit associated with the largest risk reductions. Greater benefits may be seen for some fruits, including citrus, pommes, and 100% fruit juice, and vegetables, including allium, cruciferous, and green leafy vegetables, supporting recommendations for emphasizing specific fruit and vegetable sources in dietary guidelines. No fruit and vegetable sources were adversely associated with CVD, including fruit sources of concern, such as 100% fruit juice and dried fruit. Our certainty in the evidence ranges from “very low” to “moderate,” with the least certainty for specific sources of fruits and vegetables and the highest certainty for broad categories. More research of specific food sources of fruits and vegetables is needed to improve our estimates.

Sources of Funding

This work was funded by the Canadian Institutes of Health Research (funding reference number, 129920). The Diet, Digestive Tract, and Disease Centre, funded through the Canada Foundation for Innovation and the Ministry of Research and Innovation's Ontario Research Fund, provided the infrastructure for the conduct of this project. Zurbau was funded by the Banting & Best Diabetes Centre Tamarack Graduate Award. Sievenpiper was funded by a PSI Graham Farquharson Knowledge Translation Fellowship, Diabetes Canada Clinician Scientist Award, Canadian Institutes of Health Research Institute of Nutrition, Metabolism and Diabetes (INMD) and the Canadian Nutrition Society (CNS) New Investigator Partnership Prize, and Banting & Best Diabetes Centre Sun Life Financial New Investigator Award.

Disclosures

Zurbau is a part‐time employee at INQUIS Clinical Research Ltd., a contract research organization. Khan reports he has received research support from the Canadian Institutes of Health Research (CIHR), the International Life Science Institute (ILSI), and National Honey Board. He has been an invited speaker at the Calorie Control Council Annual meeting for which he has received an honorarium. Vuksan has a Canadian (2410556) and American (7326.404) patent on the medical use of viscous fiber blend for reducing blood glucose for treatment of diabetes mellitus, increasing insulin sensitivity, and reduction in systolic blood pressure and blood lipids issued. Kendall has received grants or research support from the Advanced Food Materials Network, Agriculture and Agri‐Foods Canada, Almond Board of California, American Peanut Council, Barilla, CIHR, Canola Council of Canada, International Nut and Dried Fruit Council, International Tree Nut Council Research and Education Foundation, Loblaw Brands Ltd, Pulse Canada, and Unilever. He has received in‐kind research support from the Almond Board of California, American Peanut Council, Barilla, California Walnut Commission, Kellogg Canada, Loblaw Companies, Quaker (Pepsico), Primo, Unico, Unilever, and White Wave Foods/Danone. He has received travel support and/or honoraria from the American Peanut Council, Barilla, California Walnut Commission, Canola Council of Canada, General Mills, International Nut and Dried Fruit Council, International Pasta Organization, Loblaw Brands Ltd, Nutrition Foundation of Italy, Oldways Preservation Trust, Paramount Farms, Peanut Institute, Pulse Canada, Sun‐Maid, Tate & Lyle, Unilever, and White Wave Foods. He has served on the scientific advisory board for the International Tree Nut Council, International Pasta Organization, McCormick Science Institute, and Oldways Preservation Trust. He is a member of the International Carbohydrate Quality Consortium, Executive Board Member of the Diabetes and Nutrition Study Group of the European Association for the Study of Diabetes (EASD), is on the Clinical Practice Guidelines Expert Committee for Nutrition Therapy of the EASD, and is a Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. Jenkins has received research grants from Saskatchewan Pulse Growers, the Agricultural Bioproducts Innovation Program through the Pulse Research Network, the Advanced Foods and Material Network, Loblaw Companies Ltd, Unilever, Barilla, the Almond Board of California, Agriculture and Agri‐food Canada, Pulse Canada, Kellogg's Company, Canada, Quaker Oats, Canada, Procter & Gamble Technical Centre Ltd, Bayer Consumer Care, Springfield, NJ, Pepsi/Quaker, International Nut & Dried Fruit (INC), Soy Foods Association of North America, the Coca‐Cola Company (investigator‐initiated, unrestricted grant), Solae, Haine Celestial, the Sanitarium Company, Orafti, the International Tree Nut Council Nutrition Research and Education Foundation, the Peanut Institute, Soy Nutrition Institute, the Canola and Flax Councils of Canada, the Calorie Control Council, the CIHR, the Canada Foundation for Innovation, and the Ontario Research Fund. He has received in‐kind supplies for trials as a research support from the Almond Board of California, Walnut Council of California, American Peanut Council, Barilla, Unilever, Unico, Primo, Loblaw Companies, Quaker (Pepsico), Pristine Gourmet, Bunge Limited, Kellogg Canada, and White Wave Foods. He has been on the speaker's panel, served on the scientific advisory board, and/or received travel support and/or honoraria from the Almond Board of California, Canadian Agriculture Policy Institute, Loblaw Companies Ltd, the Griffin Hospital (for the development of the NuVal scoring system), the Coca‐Cola Company, EPICURE, Danone, Diet Quality Photo Navigation, Better Therapeutics (FareWell), Verywell, True Health Initiative, Institute of Food Technologists, Soy Nutrition Institute, Herbalife Nutrition Institute, Saskatchewan Pulse Growers, Sanitarium Company, Orafti, the Almond Board of California, the American Peanut Council, the International Tree Nut Council Nutrition Research and Education Foundation, the Peanut Institute, Herbalife International, Pacific Health Laboratories, Nutritional Fundamentals for Health, Barilla, Metagenics, Bayer Consumer Care, Unilever Canada and Netherlands, Solae, Kellogg, Quaker Oats, Procter & Gamble, the Coca‐Cola Company, the Griffin Hospital, Abbott Laboratories, the Canola Council of Canada, Dean Foods, the California Strawberry Commission, Haine Celestial, PepsiCo, the Alpro Foundation, Pioneer Hi‐Bred International, DuPont Nutrition and Health, Spherix Consulting and White Wave Foods, the Advanced Foods and Material Network, the Canola and Flax Councils of Canada, the Nutritional Fundamentals for Health, Agri‐Culture and Agri‐Food Canada, the Canadian Agri‐Food Policy Institute, Pulse Canada, the Saskatchewan Pulse Growers, the Soy Foods Association of North America, the Nutrition Foundation of Italy, Nutra‐Source Diagnostics, the McDougall Program, the Toronto Knowledge Translation Group (St. Michael's Hospital), the Canadian College of Naturopathic Medicine, The Hospital for Sick Children, the Canadian Nutrition Society, the American Society of Nutrition, Arizona State University, Paolo Sorbini Foundation, and the Institute of Nutrition, Metabolism and Diabetes. He received an honorarium from the US Department of Agriculture to present the 2013 W.O. Atwater Memorial Lecture. He received the 2013 Award for Excellence in Research from the International Nut and Dried Fruit Council. He received funding and travel support from the Canadian Society of Endocrinology and Metabolism to produce mini cases for the Canadian Diabetes Association. He is a member of the International Carbohydrate Quality Consortium. His wife, Alexandra L Jenkins, is a director and partner of INQUIS Clinical Research for the Food Industry, his 2 daughters, Wendy Jenkins and Amy Jenkins, have published a vegetarian book that promotes the use of the low glycemic index plant foods advocated here, The Portfolio Diet for Cardiovascular Risk Reduction (Academic Press/Elsevier 2020 ISBN:978‐0‐12‐810510‐8) and and his sister, Caroline Brydson, received funding through a grant from the St. Michael's Hospital Foundation to develop a cookbook for one of his studies. Sievenpiper has received research support from the Canadian Foundation for Innovation, Ontario Research Fund, Province of Ontario Ministry of Research and Innovation and Science, Canadian Institutes of health Research (CIHR), Diabetes Canada, PSI Foundation, Banting and Best Diabetes Centre (BBDC), American Society for Nutrition (ASN), INC International Nut and Dried Fruit Council Foundation, National Dried Fruit Trade Association, National Honey Board, International Life Sciences Institute (ILSI), The Tate and Lyle Nutritional Research Fund at the University of Toronto, The Glycemic Control and Cardiovascular Disease in Type 2 Diabetes Fund at the University of Toronto (a fund established by the Alberta Pulse Growers), and the Nutrition Trialists Fund at the University of Toronto (a fund established by an inaugural donation from the Calorie Control Council). He has received in‐kind food donations to support a randomized controlled trial from the Almond Board of California, California Walnut Commission, American Peanut Council, Barilla, Unilever, Upfield, Unico/Primo, Loblaw Companies, Quaker, Kellogg Canada, WhiteWave Foods, and Nutrartis. He has received travel support, speaker fees and/or honoraria from Diabetes Canada, Dairy Farmers of Canada, FoodMinds LLC, International Sweeteners Association, Nestlé, Pulse Canada, Canadian Society for Endocrinology and Metabolism (CSEM), GI Foundation, Abbott, Biofortis, ASN, Northern Ontario School of Medicine, INC Nutrition Research & Education Foundation, European Food Safety Authority (EFSA), Comité Européen des Fabricants de Sucre (CEFS), and Physicians Committee for Responsible Medicine. He has or has had ad hoc consulting arrangements with Perkins Coie LLP, Tate & Lyle, Wirtschaftliche Vereinigung Zucker e.V., and Inquis Clinical Research. He is a member of the European Fruit Juice Association Scientific Expert Panel and Soy Nutrition Institute (SNI) Scientific Advisory Committee. He is on the Clinical Practice Guidelines Expert Committees of Diabetes Canada, European Association for the study of Diabetes (EASD), Canadian Cardiovascular Society (CCS), and Obesity Canada. He serves or has served as an unpaid scientific advisor for the Food, Nutrition, and Safety Program (FNSP) and the Technical Committee on Carbohydrates of ILSI North America. He is a member of the International Carbohydrate Quality Consortium (ICQC), Executive Board Member of the Diabetes and Nutrition Study Group (DNSG) of the EASD, and Director of the Toronto 3D Knowledge Synthesis and Clinical Trials foundation. His wife is an employee of AB InBev. The remaining authors have no disclosures to report.

Tables S1–S9

Figures S1–S205

Click here for additional data file.