Association of raw fruit and fruit juice consumption with blood pressure: the INTERMAP Study^,

Abstract

Background: Epidemiologic evidence suggests that fruit consumption may lower the risk of cardiovascular diseases through blood pressure (BP)–lowering effects; little is known on the independent effect of raw fruit and fruit juice on BP.

Objective: The objective was to quantify associations of raw fruit and fruit juice consumption with BP by using cross-sectional data from the INTERnational study on MAcro/micronutrients and blood Pressure (INTERMAP) of 4680 men and women aged 40–59 y from Japan, China, the United Kingdom, and the United States.

Design: During 4 visits, 8 BP, four 24-h dietary recalls, and two 24-h urine samples were collected. Country-specific multivariate-controlled linear regression coefficients, including adjustment for urinary sodium excretion, were estimated and pooled, weighted by inverse of their variance.

Results: The average total raw fruit consumption varied from a mean ± SD of 52 ± 65 g/1000 kcal in the United States to 68 ± 70 g/1000 kcal in China. Individual raw fruit intake was not associated with BP in pooled analyses for all countries or in participants from Western countries, although a positive association with diastolic BP was observed in East Asian participants (per 50 g/1000 kcal; 0.37 mm Hg; 95% CI: 0.02, 0.71). Positive relationships with diastolic BP were found for citrus fruit intake in Western consumers (per 25 g/1000 kcal; 0.47 mm Hg; 95% CI: 0.12, 0.81) and for apple intake in East Asian consumers (0.40 mm Hg; 95% CI: 0.03, 0.78). Among East Asian banana consumers, banana intake was inversely associated with diastolic BP (−1.01 mm Hg; 95% CI: −1.88, −0.02). Fruit juice intake, which was negligible in Asia, was not related to BP in Western countries.

Conclusion: Consistent associations were not found between raw fruit and fruit juice consumption of individuals and BP. This observational study was registered at www.clinicaltrials.gov as NCT00005271.

INTRODUCTION

Research reported in the 1970s and 1980s suggests that persons consuming vegetarian diets, which are generally higher in fruit and vegetables, have lower blood pressures (BPs) (1–4). Two intervention studies showed that participants who increased their daily fruit and vegetable intakes significantly reduced their average systolic and diastolic BPs compared with control subjects (5). Correspondingly, cohort studies found inverse associations between fruit and vegetable consumption and BP (6–9) and the risk of developing elevated BP (10) and hypertension (9, 11, 12).

Evidence is limited, however, on the independent associations of fruit intake on BP. Previous cohort studies found inverse relations between total fruit consumption and BP (6–9), BP change (8), or risk of developing hypertension (9–12). In these studies, fruit was defined as all raw and processed fruit (including juice) (6, 7, 10), or the definition of fruit was not reported (8, 9, 11, 12). This is relevant because the nutritional value of fruit is influenced by processing (13). Compared with their raw counterparts, fruit juices are lower in dietary fiber, but may be good sources of micronutrients, eg, vitamin C (14). Recently, prospective cohort studies reported that raw fruit and vegetable intakes were inversely related to ischemic heart disease (15) and stroke (16). Elevated BP is an important risk factor for these major cardiovascular diseases. To the best of our knowledge, no previous studies have examined the association of raw fruit and fruit juice intake with BP.

In the current study, we investigated cross-sectional associations between intake of raw fruit, intake of fruit juice, and BP among 4680 adults from 17 population samples in China, Japan, the United Kingdom, and the United States using high-quality data from 8 BP readings and four 24-h dietary-recall assessments.

SUBJECTS AND METHODS

Population

The INTERnational study on MAcro/micronutrients and blood Pressure (INTERMAP) Study (www.clinicaltrials.gov; NCT00005271) surveyed 4680 men and women aged 40–59 y from 17 population samples in Japan (4 samples), the People's Republic of China (3 samples), the United Kingdom (2 samples), and the United States (8 samples) (17). Participants were randomly selected from community and workforce populations, stratified by age and sex. The mean participation rate was 49% (45% in Japan, 83% in the People's Republic of China, 22% in the United Kingdom, and 44% in the United States). The measurements were conducted between 1996 and 1999 during 2 study visits on consecutive days and 2 additional study visits on consecutive days about 3 wk later. Quality control of nondietary (17) and dietary data (18) was extensive, with local, national, and international checks on completeness and integrity. The study received institutional review board or ethics committee approval at each site. All participants gave written informed consent.

Of 4895 individuals initially surveyed, we excluded individuals who did not attend all 4 study visits (n = 110); whose dietary data were considered unreliable (n = 7); with a total energy intake from any 24-h recall <500 or >5000 kcal/d for women and <500 or >8000 kcal/d for men (n = 37 total); with unavailable urine samples; or with other data incomplete, missing, or indicating protocol violation (n = 61). This resulted in a study population of 4680 participants (2359 men and 2321 women).

Dietary assessment

Dietary intake was assessed by four 24-h dietary recalls. Standardized procedures were reported in detail (18). In summary, at each study visit, trained certified interviewers using a multipass procedure ascertained in depth all foods, beverages, and supplements consumed in the prior 24 h. Information on preparation methods was also collected. In the United States, dietary data were entered directly into a computerized database (Nutrition Data System, version 2.91; University of Minnesota). In the other countries, data were entered onto standard forms, coded, and computerized. Nutrient intake was calculated by using country-specific food tables, standardized across countries by the Nutrition Coordinating Center, University of Minnesota (18, 19). For all participants, Pearson partial correlation coefficients adjusted for sample and sex that compared intakes by 24-h recall and 24-h urinary excretion were 0.51 for total protein intake and urinary urea, 0.42 for sodium, 0.55 for potassium, and 0.42 for sodium/potassium (18).

We included all fruit reported by all participants in all countries. Excluded were avocado, coconut, sugar cane, plantains, and tamarind, because their nutritional value differs substantially from that of most fruit. Fruit nectars, fruit drinks, lemonades, and soft drinks were not considered fruit juice, because these beverages contain minimal fruit content and are often sweetened.

BP measurement

Systolic and diastolic BP (first and fifth Korotkoff sounds, respectively) were measured by trained staff with a random-zero sphygmomanometer. BP was measured twice at each study visit, for a total of 8 measurements. Measurements were carried out on the right arm while the participant was seated, after a rest of ≥5 min and with the bladder emptied, feet flat on the floor, in a quiet room, and with no physical activity, eating, drinking, or smoking in the preceding half hour.

Other lifestyle factors

During 2 visits, body weight and height without shoes were measured and BMI (in kg/m²) was calculated. Urinary sodium and potassium were measured from 2 timed 24-h urine samples, which were obtained from each participant between consecutive clinic visits; 8% of the specimens were split locally and sent blind to the Central Laboratory to estimate technical error. Interviewer-assisted questionnaire data included daily alcohol intake over the preceding 7 d, cigarette smoking, educational level, physical activity, adherence to a special diet, use of antihypertensive and lipid-lowering drugs, and participant's and family history of cardiovascular diseases and diabetes mellitus.

Statistical methods

The analyses were performed by using SAS version 9.3 (SAS Institute Inc). Dietary intake data were adjusted for total energy intake by using the nutrient density method (20). Measurements were averaged across the 4 study visits for dietary and BP variables and across the two 24-h urinary collections.

Partial correlation was used to explore associations of fruit with dietary and urinary variables, adjusted for sample, age, and sex and pooled across countries, weighted by sample size. Western and East Asian participants differ significantly in dietary pattern and other characteristics (21); therefore, the results are presented for countries separately as well as overall. Fruit juice intake, negligible in East Asian countries, was analyzed in relation to BP only for Western participants.

From the mean of the first and second 2 visits, we estimated the reliability of raw fruit and fruit juice intake for individuals using the following formula: 1/[1+(ratio/2)] × 100, where the ratio is within-participant divided by between-participant variance (22, 23). This gives an indication of the effect of day-to-day variability on the associations with BP.

We used multivariable linear regression analyses to examine associations between individual raw fruit and fruit juice consumption (per 50 g/1000 kcal) and BP and across quartiles of raw fruit intake (lowest quartile as reference). Intake of the most commonly consumed raw fruit (per 25 g/1000 kcal) was analyzed in the total population and in the group of consumers. To estimate the overall association, models were fitted by country, and coefficients were pooled, weighted by inverse of their variance (24). Cross-country heterogeneity of the regression coefficients was assessed by chi-square test. Three models were used adjusted extensively for dietary and lifestyle factors sequentially, including a model adjusted also for BMI.

We repeated the analyses for 3 subcohorts, excluding participants with medical diagnoses and other traits that might bias relations between fruit intake and BP: 1) a subcohort excluding participants with a diagnosis of hypertension and users of antihypertensive drugs, 2) a subcohort of nonhypertensive participants (excluding from the foregoing cohort also those with elevated systolic BP (≥140 mm Hg) or diastolic BP (≥90 mm Hg), and 3) a subcohort free of major chronic disease (excluding also those with prevalent cardiovascular diseases and diabetes). Censored normal regression was used to adjust for potential antihypertensive treatment bias (25). Stratified analyses and inclusion of interaction terms showed no evidence for potential effect modification by age, sex, or smoking. Two-tailed probability values <0.05 were considered statistically significant.

RESULTS

Descriptive statistics

The mean (±SD) raw fruit intake ranged from 52 ± 65 g/1000 kcal in the United States to 68 ± 70 g/1000 kcal in China (see Table S1 under “Supplemental data” in the online issue). Apples and pears, citrus fruit, and bananas were the most commonly consumed raw fruit in each country. These fruits accounted for 63% of total fruit intake in the United States (lowest proportion) and for 82% in China and the United Kingdom. The mean (±SD) fruit juice intake, negligible in East Asian samples, was 46 ± 70 g/1000 kcal in the United States and 27 ± 44 g/1000 kcal in the United Kingdom. In both Western countries, this was mainly citrus fruit juice (∼80%).

Participants with higher raw fruit intakes were more often women, tended to be older, were less likely to smoke and drink alcohol, more often used dietary supplements, had lower reported energy intake, and had lower BP than those with lower intakes (Table 1). In contrast with East Asian participants, higher raw fruit consumers in the United Kingdom and United States differed overall in dietary pattern; compared with lower raw fruit consumers, higher raw fruit consumers had higher intakes of vegetables, low-fat dairy products, fiber-rich cereals and grains, and lower intakes of meats.

TABLE 1.

Baseline characteristics of INTERMAP participants from East Asian and Western countries stratified by low and high raw fruit consumption¹

	East Asian countries		Western countries
Variable	Low (n = 992)	High (n = 992)	Low (n = 1348)	High (n = 1348)
Age ( y)	49.1 ± 5.62	49.3 ± 5.5	48.6 ± 5.5	49.6 ± 5.3
Men (%)	62.6	37.2	56.7	44.9
Education (y)	9.4 ± 4.1	9.0 ± 4.0	14.1 ± 3.1	15.0 ± 3.2
Current smokers (%)	42.2	23.4	23.6	10.2
Alcohol intake (g/d)	19.2 ± 26.1	7.8 ± 16.3	9.7 ± 17.0	7.1 ± 12.9
Physically active during leisure time (%)3	56.3	55.5	56.9	63.0
Moderate/heavy physical activity during work and leisure time (h/d)	4.0 ± 4.0	4.0 ± 4.0	3.3 ± 3.3	2.8 ± 2.8
Taking dietary supplements (%)	15.0	15.5	42.4	55.7
BMI (kg/m2)	23.3 ± 3.1	23.3 ± 3.1	29.2 ± 5.8	28.0 ± 5.6
Systolic BP (mm Hg)	119.1 ± 14.4	118.8 ± 16.6	120.4 ± 13.7	117.5 ± 14.2
Diastolic BP (mm Hg)	73.6 ± 10.3	73.3 ± 10.3	74.9 ± 9.9	73.3 ± 9.7
History of cardiovascular disease or diabetes mellitus (%)	10.6	8.6	15.7	13.8
Use of antihypertensive, cardiovascular disease, or diabetes drugs (%)	6.9	8.4	23.3	21.1
Family history of hypertension (%)
Yes	39.8	43.5	64.9	63.7
Unknown	31.0	28.9	24.6	25.7
Urinary sodium (mmol/24 h)	210.4 ± 80.4	211.0 ± 78.2	164.8 ± 61.1	153.9 ± 54.2
Any special diet (%)	4.8	7.4	14.2	23.4
Total energy intake (kcal/d)	2100 ± 518	1974 ± 488	2349 ± 712	2111 ± 641
Dietary intake (g/1000 kcal)
Raw fruit	16 ± 15	106 ± 57	10 ± 11	99 ± 69
Fruit juice	3 ± 12	3 ± 10	36 ± 59	49 ± 72
Dried fruit	0 ± 1	1 ± 2	1 ± 3	1 ± 5
Processed fruit	1 ± 4	1 ± 3	5 ± 11	8 ± 16
Raw and cooked vegetables	141 ± 77	154 ± 68	70 ± 48	93 ± 62
Low-fat dairy products	6 ± 17	8 ± 23	35 ± 74	61 ± 90
Red and processed meat	20 ± 19	19 ± 17	42 ± 28	33 ± 25
Fish and shellfish	33 ± 32	30 ± 31	9 ± 14	11 ± 16
Fiber-rich cereals and grains	56 ± 144	49 ± 113	13 ± 19	21 ± 25
Nuts and seeds	2 ± 4	2 ± 4	3 ± 5	3 ± 5

¹Participants were classified according to low or high raw fruit consumption defined by median intake (46.5 g/1000 kcal among East Asian participants and 32.8 g/1000 kcal among Western participants). BP, blood pressure; INTERMAP, INTERnational study on MAcro/micronutrients and blood Pressure.

²Mean ± SD (all such values).

³Defined as engagement in moderate or heavy physical activity during leisure time.

Reliability estimates for total raw fruit intake (g/d) were 61% in the total population; ranging from 52% (China) to 82% (United Kingdom); for fruit juice intake the estimates were 67% in the total population, 65% in the United States, and 77% in the United Kingdom. This implies that true associations with other variables may be larger than observed associations, eg, 1.64 (1/0.61) times those for raw fruit in the total population.

Correlations between raw fruit intake and nutrients

Overall, raw fruit intake was positively correlated with intake of vitamin C (see Table S2 under “Supplemental data” in the online issue; r = 0.50), dietary fiber (r = 0.48), potassium (r = 0.48), magnesium (r = 0.30), and 24-h urine potassium excretion (r = 0.27). In participants from Western countries, raw fruit was negatively correlated with total fat intake (r = −0.35) and positively with vegetable protein (r = 0.27). In East Asian participants, raw fruit intake was positively associated with total sugar (r = 0.52) and negatively with starch (r = −0.20). Fruit juice intake among those from Western countries was positively correlated with vitamin C (r = 0.61) and total sugar (r = 0.32), but only low order with raw fruit intake (r = 0.09).

Associations of raw fruit and fruit juice with BP

Average systolic BP ranged from 117.2 ± 13.8 mm Hg in Japan to 121.3 ± 17.4 mm Hg in China. Average diastolic BP ranged from 73.2 ± 10.2 mm Hg in China to 77.3 ± 9.9 mm Hg in the United Kingdom. In the total population and in participants from Western countries, there were no significant associations between raw fruit intake and systolic blood or diastolic BP (Table 2). In participants from East Asian countries, raw fruit intake was positively (not inversely) related to diastolic BP (model 3 + BMI: per 50 g/1000 kcal; 0.37 mm Hg; 95% CI: 0.02, 0.71) but not to systolic BP. Fruit juice consumption was not related to systolic or diastolic BP in participants from Western countries. No significant heterogeneity was observed across countries. Generally comparable results were found in analyses of raw fruit intake and BP across quartiles of raw fruit (see Table S3 under “Supplemental data” in the online issue). Censored normal regression to adjust for potential antihypertensive treatment bias yielded comparable associations in the total population for raw fruit intake higher by 50 g/1000 kcal BP differences were 0.05 mm Hg (95% CI: −0.27, 0.70) for systolic and 0.20 mm Hg (95% CI: −0.02, 0.41) for diastolic in the original analyses and −0.02 mm Hg (95% CI: −0.37, 0.33) for systolic and 0.20 mm Hg (95% CI: −0.09, 0.48) for diastolic in the censored normal regression.

TABLE 2.

Estimated mean difference in BP associated with consumption of raw fruit and fruit juice (>50 g/1000 kcal)¹

	Systolic BP			Diastolic BP
Variable	Difference	95% CI	P	Difference	95% CI	P
	mm Hg	mm Hg		mm Hg	mm Hg
Raw fruit consumption
Total population (n = 4680)
Model 1	−0.59	(−0.91, −0.27)	0.0003	−0.12	(−0.34, 0.09)	0.26
Model 2	−0.22	(−0.53, 0.09)	0.17	0.06	(−0.15, 0.28)	0.57
Model 3	−0.06	(−0.38, 0.27)	0.73	0.14	(−0.08, 0.36)	0.22
Model 3 + BMI	0.05	(−0.26, 0.36)	0.74	0.20	(−0.07, 0.41)	0.07
Participants from Western countries
Model 1	−0.77	(−1.16, −0.37)	0.0001	−0.28	(−0.54, −0.01)	0.04
Model 2	−0.33	(−0.72, 0.06)	0.10	−0.07	(−0.33, 0.20)	0.62
Model 3	−0.13	(−0.53, 0.26)	0.51	0.03	(−0.24, 0.31)	0.81
Model 3 + BMI	−0.04	(−0.39, 0.38)	0.98	0.09	(−0.18, 0.36)	0.52
Participants from East Asian countries
Model 1	−0.24	(−0.80, 0.33)	0.41	0.18	(−0.19, 0.54)	0.34
Model 2	−0.02	(−0.55, 0.51)	0.95	0.29	(−0.06, 0.65)	0.10
Model 3	0.10	(−0.46, 0.65)	0.73	0.33	(−0.04, 0.70)	0.08
Model 3 + BMI	0.15	(−0.37, 0.67)	0.56	0.37	(0.02, 0.71)	0.04
Fruit juice consumption
Participants from Western countries
Model 1	−0.29	(−0.68, 0.10)	0.15	0.03	(−0.23, 0.29)	0.82
Model 2	−0.14	(−0.52, 0.23)	0.46	0.06	(−0.20, 0.32)	0.66
Model 3	−0.11	(−0.49, 0.26)	0.55	0.06	(−0.20, 0.32)	0.66
Model 3 + BMI	−0.14	(−0.50, 0.22)	0.45	0.04	(−0.21, 0.30)	0.74

¹Differences in BP and corresponding 95% CIs were obtained from multivariable linear regression analyses by pooling cross-country regression coefficients weighted by sample size. Model 1 was adjusted for age, sex, and sample. Model 2 was adjusted as for model 1 plus intakes of energy (kcal) and alcohol (g/d), smoking status (never, former, or current), years of education (years completed), physical activity during leisure time (a lot, moderate, little, or none), use of dietary supplements (yes or no), adherence to any special diet (yes or no), history of cardiovascular disease or diabetes mellitus (yes or no), family history of cardiovascular disease (yes or no), urinary sodium (mmol/24 h), and use of antihypertensive, cardiovascular disease, or diabetes medication (yes or no). Model 3 was adjusted as for model 2 plus intake (g/1000 kcal) of low-fat dairy products, raw and cooked vegetables, fiber-rich cereals and grains, red and processed meats, nuts and seeds, and fish and shellfish. Cross-country heterogeneity of regression coefficients was assessed by chi-square test: no significant heterogeneity across countries was detected. BP, blood pressure.

Associations of raw fruit and fruit juice with BP in subcohorts

We repeated the regression analyses between raw fruit intake and BP for several subcohorts with characteristics that might bias relations between raw fruit intake and BP, ie, a subcohort excluding participants with diagnosed hypertension and users of antihypertensive drugs, a subcohort of nonhypertensive participants, and a subcohort free of major chronic diseases. In these predefined subcohorts, no clear associations between raw fruit intake and systolic or diastolic BP were found (Table 3). However, relations between raw fruit intake and systolic BP tended to be nonsignificantly inverse for nonhypertensive participants and in those also free of major chronic diseases.

TABLE 3.

Estimated mean difference in BP associated with consumption of raw fruit (>50 g/1000 kcal) in subcohorts¹

		Systolic BP			Diastolic BP
Subcohort and model	No. of subjects	Difference	95% CI	P	Difference	95% CI	P
		mm Hg	mm Hg		mm Hg	mm Hg
Excluding participants with diagnosis of hypertension and users of antihypertensive drugs2
Model 3	3532	−0.13	(−0.45, 0.20)	0.43	0.11	(−0.12, 0.34)	0.37
Model 3 + BMI	3532	−0.04	(−0.35, 0.27)	0.79	0.16	(−0.06, 0.38)	0.16
Nonhypertensive participants23
Model 3	3363	−0.26	(−0.55, 0.02)	0.07	−0.04	(−0.22, 0.21)	0.97
Model 3 + BMI	3363	−0.20	(−0.48, 0.07)	0.15	0.04	(−0.17, 0.25)	0.71
Further exclusion of participants with prevalent cardiovascular diseases and diabetes mellitus
Model 3	3051	−0.28	(−0.58, 0.02)	0.07	0.03	(−0.19, 0.25)	0.78
Model 3 + BMI	3051	−0.22	(−0.51, 0.06)	0.13	0.07	(−0.14, 0.28)	0.51

¹Differences in BP and corresponding 95% CIs were obtained from multivariable linear regression analyses by pooling cross-country regression coefficients weighted by sample size. Model 1 was adjusted for age, sex, and sample. Model 2 was adjusted as for model 1 plus intakes of energy (kcal) and alcohol (g/d), smoking status (never, former, or current), years of education (years completed), physical activity during leisure time (a lot, moderate, little, or none), use of dietary supplements (yes or no), adherence to any special diet (yes or no), history of cardiovascular disease or diabetes mellitus (yes or no), family history of cardiovascular disease (yes or no), urinary sodium (mmol/24 h), and use of antihypertensive, cardiovascular disease, or diabetes medication (yes or no). Model 3 was adjusted as for model 2 plus intake (g/1000 kcal) of low-fat dairy products, raw and cooked vegetables, fiber-rich cereals and grains, red and processed meats, nuts and seeds, and fish and shellfish. BP, blood pressure.

²Additionally adjusted for history of cardiovascular disease or diabetes mellitus.

³Elevated BP is defined as systolic BP ≥140 mm Hg or diastolic BP ≥90 mm Hg.

Associations of specific fruit with BP

The most commonly consumed fruits were analyzed separately. Apples and pears, citrus fruit, and banana were not related to systolic BP (Table 4). Apple and pear intake was positively (not inversely) associated with diastolic BP in East Asian participants (model 3 + BMI: per 25 g/1000 kcal; +0.31 mm Hg; 95% CI: 0.02, 0.60), with cross-country heterogeneity (P = 0.05). In participants from Western countries, citrus fruit intake was positively (not inversely) related to diastolic BP (model 3 + BMI: per 25 g/1000 kcal; +0.33 mm Hg; 95% CI: 0.05, 0.46). Comparable associations were found among consumers (Table 5). In banana consumers from East Asian countries, banana intake was inversely related with diastolic BP (model 3 + BMI: per 25 g/1000 kcal; −1.01 mm Hg; 95% CI: −1.88, −0.15).

TABLE 4.

Estimated mean difference in BP associated with higher consumption (by 25 g/1000 kcal) of apples and pears, citrus fruit, and banana in all participants and for East Asian and Western countries separately¹

			Systolic BP						Diastolic BP
			Not adjusted for BMI			Adjusted for BMI			Not adjusted for BMI			Adjusted for BMI
Variable	No. of subjects	Mean ± SD	Difference	95% CI	P	Difference	95% CI	P	Difference	95% CI	P	Difference	95% CI	P
		mm Hg	mm Hg	mm Hg		mm Hg	mm Hg		mm Hg	mm Hg		mm Hg	mm Hg
Apples and pears
Total	4680	17 ± 33	0.04	(−0.30, 0.37)	0.83	0.08	(−0.25, 0.41)	0.64	0.10	(−0.13, 0.32)	0.40	0.12	(−0.10, 0.34)	0.28
West	2696	13 ± 18	−0.02	(−0.51, 0.48)	0.95	−0.002	(−0.48, 0.48)	0.99	−0.14	(−0.48, 0.20)	0.42	−0.13	(−0.46, 0.34)	0.44
Asia	1984	22 ± 39	0.08	(−0.38, 0.55)	0.72	0.15	(−0.31, 0.60)	0.53	0.28	(−0.02, 0.58)	0.06	0.31	(0.02, 0.60)	0.04
Citrus fruit
Total	4680	14 ± 31	0.22	(−0.09, 0.54)	0.16	0.24	(−0.06, 0.54)	0.12	0.24	(0.02, 0.46)	0.03	0.25	(0.04, 0.46)	0.02
West	2696	11 ± 30	0.23	(−0.18, 0.63)	0.27	0.21	(−0.18, 0.61)	0.29	0.34	(0.06, 0.61)	0.02	0.33	(0.05, 0.46)	0.02
Asia	1984	18 ± 32	0.22	(−0.28, 0.71)	0.39	0.28	(−0.19, 0.76)	0.24	0.10	(−0.25, 0.44)	0.58	0.14	(−0.20, 0.47)	0.43
Banana
Total	4680	9 ± 17	−0.36	(−0.95, 0.23)	0.23	−0.27	(−0.84, 0.30)	0.35	−0.08	(−0.48, 0.32)	0.68	−0.03	(−0.42, 0.36)	0.88
West	2696	11 ± 18	−0.08	(−0.79, 0.63)	0.82	0.01	(−0.68, 0.70)	0.97	0.17	(−0.31, 0.66)	0.48	0.23	(−0.25, 0.70)	0.36
Asia	1984	6 ± 16	−1.00	(−2.06, 0.07)	0.07	−0.91	(−1.94, 0.12)	0.08	−0.61	(−1.30, 0.09)	0.09	−0.53	(−1.20, 0.14)	0.12

¹Differences in BP and corresponding 95% CIs were obtained from multivariable linear regression analyses by pooling cross-country regression coefficients weighted by sample size. Model 1 was adjusted for age, sex, and sample. Model 2 was adjusted as for model 1 plus intakes of energy (kcal) and alcohol (g/d), smoking status (never, former, or current), years of education (years completed), physical activity during leisure time (a lot, moderate, little, or none), use of dietary supplements (yes or no), adherence to any special diet (yes or no), history of cardiovascular disease or diabetes mellitus (yes or no), family history of cardiovascular disease (yes or no), urinary sodium (mmol/24 h), and use of antihypertensive, cardiovascular disease, or diabetes medication (yes or no). Model 3 was adjusted as for model 2 plus intake (g/1000 kcal) of low-fat dairy products, raw and cooked vegetables, fiber-rich cereals and grains, red and processed meats, nuts and seeds, and fish and shellfish. BP, blood pressure.

TABLE 5.

Estimated mean difference in BP associated with higher consumption (by 25 g/1000 kcal) of apples and pears (in consumers only; n = 1945), citrus fruit (n = 1767), and banana (n = 1604)¹

		Systolic BP						Diastolic BP
			Not adjusted for BMI			Adjusted for BMI			Not adjusted for BMI			Adjusted for BMI
Variable	No. of subjects	Mean ± SD	Difference	95% CI	P	Difference	95% CI	P	Difference	95% CI	P	Difference	95% CI	P
		mm Hg	mm Hg	mm Hg		mm Hg	mm Hg		mm Hg	mm Hg		mm Hg	mm Hg
Apples and pears
Total	1945	41 ± 40	0.09	(−0.36, 0.54)	0.70	0.10	(−0.34, 0.54)	0.66	0.25	(−0.04, 0.54)	0.09	0.26	(−0.02, 0.54)	0.07
West	974	37 ± 33	0.06	(−0.61, 0.73)	0.86	0.10	(−0.55, 0.75)	0.76	0.04	(−0.40, 0.49)	0.85	0.07	(−0.37, 0.50)	0.76
Asia	971	46 ± 46	0.11	(−0.50, 0.72)	0.72	0.10	(−0.50, 0.69)	0.75	0.41	(−0.02, 0.79)	0.04	0.40	(0.03, 0.78)	0.03
Citrus fruit
Total	1767	40 ± 42	0.34	(−0.05, 0.74)	0.09	0.32	(−0.06, 0.70)	0.10	0.35	(0.08, 0.62)	0.01	0.34	(0.08, 0.60)	0.01
West	722	41 ± 47	0.48	(−0.04, 1.00)	0.07	0.44	(−0.07, 0.94)	0.09	0.47	(0.12, 0.83)	0.01	0.47	(0.12, 0.81)	0.01
Asia	954	38 ± 37	0.16	(−0.45, 0.76)	0.62	0.16	(−0.42, 0.73)	0.59	0.16	(−0.26, 0.59)	0.45	0.17	(−0.24, 0.57)	0.43
Banana
Total	1604	26 ± 21	−0.32	(−1.12, 0.49)	0.44	−0.28	(−1.06, 0.51)	0.49	−0.36	(−0.91, 0.19)	0.21	−0.33	(−0.87, 0.21)	0.23
West	1146	26 ± 19	0.11	(−0.89, 1.11)	0.83	0.12	(−0.85, 1.09)	0.80	0.10	(−0.60, 0.80)	0.78	0.10	(−0.59, 0.79)	0.77
Asia	458	26 ± 25	−1.13	(−2.50, 0.25)	0.11	−1.04	(−2.38, 0.13)	0.13	−1.07	(−1.96, −0.19)	0.02	−1.01	(−1.88, −0.15)	0.02

¹Differences in BP and corresponding 95% CIs were obtained from multivariable linear regression analyses by pooling cross-country regression coefficients weighted by sample size. Model 1 was adjusted for age, sex, and sample. Model 2 was adjusted as for model 1 plus intakes of energy (kcal) and alcohol (g/d), smoking status (never, former, or current), years of education (years completed), physical activity during leisure time (a lot, moderate, little, or none), use of dietary supplements (yes or no), adherence to any special diet (yes or no), history of cardiovascular disease or diabetes mellitus (yes or no), family history of cardiovascular disease (yes or no), urinary sodium (mmol/24 h), and use of antihypertensive, cardiovascular disease, or diabetes medication (yes or no). Model 3 was adjusted as for model 2 plus intake (g/1000 kcal) of low-fat dairy products, raw and cooked vegetables, fiber-rich cereals and grains, red and processed meats, nuts and seeds, and fish and shellfish. BP, blood pressure.

DISCUSSION

In this cross-sectional, cross-cultural, population-based study of middle-aged individuals, contrary to our original hypothesis, raw fruit intake was not inversely associated with BP. Given the cross-sectional nature of these data and the scarcity of findings from other studies, causal inferences are premature. We are aware of 4 other epidemiologic studies that investigated total fruit intake in relation to BP (6–9). Two observed inverse relations between total fruit intake and baseline systolic BP (7, 8): inverse associations with both systolic and diastolic BP after 4 y of follow-up in one study (6) and an inverse association with diastolic BP after 6 y of follow-up in the other study (9). Our cross-sectional results are not in agreement with these previous findings; overall, raw fruit intake was not related to systolic BP, and a positive association was observed with diastolic BP in East Asian participants.

The high-quality and detailed dietary data derived from 4 interviewer-administered in-depth 24-h dietary recalls collected over 4 visits enabled us to define raw fruit and fruit juice intake exclusive of processed fruit, fruit nectars, fruit-flavored drinks, lemonades, or soft drinks. However, whether fruit juices were actually 100% juice remains questionable. Differences between our findings and previous studies may reflect the use of different methods by others (eg, food-frequency procedures to collect dietary data) (6, 7, 9), misclassification with self-reported BP (6, 7, 9), ambiguous definition of fruit intake (6–9), random variation, or other currently unrecognized factors.

Furthermore, our results do not suggest a relation of fruit juice intake with BP among Western populations. This was the first cross-sectional population study to our knowledge that addressed this question. The most often consumed fruit juice was orange juice. Results from a short-term intervention study of 4 wk among 24 healthy, overweight, middle-aged men suggested that consumption of orange juice decreased diastolic BP (26). When raw fruit is converted to juice, intact cell walls are obliterated, which results in a reduced dietary fiber content, but vitamin C and other phytochemicals remain present (14). A lower dietary fiber content may explain the current finding of no significant association between fruit juice consumption and BP.

Various types of fruit provide different arrays of micronutrients and bioactive compounds that may act differently in relation to BP. Previous population studies found inverse associations between berries and systolic BP (7); strawberries and diastolic BP (7); apples, oranges, prunes, and grapes with systolic and diastolic BP (6); and apples, oranges, and raisins with risk of hypertension (12). Apples and pears, citrus fruit, and bananas were the most commonly consumed fruit in all 4 countries. We found positive associations between apple and pear intake and diastolic BP in East Asian participants and citrus fruit in Western participants. Although methodologic issues, ie, the cross-sectional design of the study and measurement error in dietary assessment, could explain these unexpected results, it has been suggested that intake of fructose—of which fruit is a rich source—may raise BP as a result of increased uric acid production and reduced nitric oxide concentrations (27). A direct association between fructose intake and BP was previously reported among Western participants from the INTERMAP Study (28). However, a recent meta-analysis of 13 controlled feeding trials did not show any adverse effect of fructose supplementation on BP (29).

Banana intake was inversely related to BP among East Asian banana consumers. Bananas are rich sources of dietary fiber (2.6 g/100 g), potassium (358 mg/100 g), and magnesium (27 mg/100 g) (30). Results from intervention studies indicate that supplementation of these nutrients may beneficially influence BP (31–33). These nutrients are highly correlated with each other (34) and their dietary sources, which suggests that any observed beneficial associations between fruit and BP may be due to combined or synergistic effects of multiple components in their natural food matrix (35). Thus, a recent prospective cohort study of 20,000 Dutch men and women, using a food-frequency questionnaire, reported significant inverse associations with incident stroke for white fruit and vegetables, including apples and bananas (36).

Participants with diagnosed elevated BP, cardiovascular disease, or diabetes may have adopted a healthier lifestyle and dietary pattern, including higher fruit intake. Although data for these subcohorts showed inverse relations between raw fruit intake and systolic BP, the results were not significant. In the current study, BP differences associated with raw fruit were small and possibly underestimated, so that significant actual associations may have been missed.

A limitation of the current study was its cross-sectional nature; therefore, making inferences on causality over the long-term is premature. In addition, some participants may have altered their habits toward a healthier lifestyle in response to a prior diagnosis of elevated BP, which attenuated our findings and jeopardized inferences. Furthermore, fruit intake measured by 4 detailed 24-h dietary recalls over a few weeks may not be representative of habitual long-term intake for some individuals, which can result in misclassification of usual fruit consumption at the individual level. These limitations may explain our findings and relate to inconsistencies across studies.

High raw fruit consumers tended to have a generally healthier diet and lifestyle than did low raw fruit consumers. We adjusted for multiple dietary and other lifestyle factors as potential confounders, including timed 24-h urinary sodium excretion—a high-quality marker for dietary sodium. The possibility of residual confounding, however, cannot be ruled out. On the other hand, extensive adjustments may have attenuated small BP differences and led to general null findings. Finally, multiple comparisons may have led to chance findings. However, the analyses were hypothesis driven, and inferences were made in the context of a critical evaluation of previously published results. Before definitive conclusions can be made, further data are needed from dietary intervention studies and large high-quality prospective population studies with detailed dietary data and measured BP.

In conclusion, results from this cross-sectional study showed no consistent relations to BP of raw fruit and fruit juice intakes. Specific types of fruit containing various nutrients may differ in their effect on BP.