Fruit and vegetable intake and risk of frailty in women 60 years old or older

ABSTRACT

Background

Prior research has suggested that the antioxidative and anti-inflammatory potential of fruits and vegetables may ameliorate aging-related frailty.

Objective

We sought to prospectively examine the association between fruit and vegetable intake and incident frailty in older women.

Design

We followed 78,366 nonfrail women aged ≥60 y from the Nurses’ Health Study from 1990 to 2014. In this analysis, the primary exposure was the intake of total fruits and vegetables, assessed with an FFQ administered 6 times during follow-up. Frailty was defined as having ≥3 of the following 5 criteria from the FRAIL scale: fatigue, poor strength, low aerobic capacity, having ≥5 illnesses, and ≥5% weight loss. Cox models adjusted for potential confounders were used to estimate HRs and 95% CIs for the association between fruit and vegetable intake and incident frailty.

Results

In total, 12,434 (15.9%) incident frailty cases were accrued during follow-up. Total fruit and vegetable intakes were associated with a lower risk of frailty (adjusted HR comparing 7+ servings/d compared with <3 servings/d: 0.92; 95% CI: 0.85, 0.99). The inverse association appeared to be stronger for those with physical activity above the median (P-interaction < 0.05). Among physically active women, compared with those who consumed <3 servings/d, the HR for 7+ servings/d was 0.68 (95% CI: 0.57, 0.81).

Conclusion

Higher fruit and vegetable intake was associated with a lower risk of frailty in this cohort of US women aged ≥60 y. Because of limited evidence on intakes of fruits and vegetables and the development of frailty, more data are needed to confirm our results.

Introduction

Frailty in older adults involves declining physical function and increased vulnerability to illness and disability over time (1). Although physical function is a major component of frailty, this syndrome also includes weight loss and other characteristics that compromise the ability to handle stressors such as acute illnesses (2). Frailty has been associated with a poor quality of life, as well as increased morbidity and mortality (3–6).

Dietary patterns that are characterized by high intakes of fruits and vegetables have been shown to be associated with a lower risk of frailty (7–9). In elderly individuals, fruit and vegetable consumption may influence the rate of functional decline through decreasing oxidative stress and inflammation and may ameliorate age-related frailty (10–12). In addition, fruits and vegetables also contain components (e.g., fiber, potassium) that may reduce the risk of major chronic diseases (13).

A few small prospective studies have observed an inverse association between fruit and vegetable intake and frailty (14). However, because their follow-up durations were <5 y, these studies could only examine short-term associations (8, 15). In addition, the results were mixed, with an inverse association for both fruits and vegetables in a combined French and Spanish study (15), an inverse association with fruits but not vegetables in a Spanish study (8), and an inverse association only with vegetables that were not salads, carrots, or potatoes in an African-American cohort (16). On the other hand, a previous investigation from the Nurses’ Health Study (NHS) on diet quality and physical function noted an inverse association with higher intakes of fruits and vegetables (17). Other commonly considered features of geriatric frailty, such as fatigue, unintentional weight loss, and chronic illness (1, 18), may also contribute independently or compound poor physical function. Among participants in the NHS who had diabetes, higher fruit and vegetable intake was associated with a lower risk of frailty (19). However, the association was attenuated upon additional adjustment for physical activity. Therefore, we expanded our analysis to the entire NHS cohort and prospectively examined intake of total fruits and vegetables in relation to development of frailty, based on multiple assessments of diet and a long follow-up.

Methods

Study participants

Women in this analysis were participants of the US NHS. This ongoing cohort began in 1976 with 121,700 nurses aged 30–55 y in 11 states (20). Participants self-reported lifestyle and disease information through a questionnaire every 2 y. In addition, an FFQ was sent in 1980, 1984, 1986, and every 4 y thereafter to collect dietary information.

In this analysis we included women aged ≥60 y in 1992 with valid dietary information (intake between 500 and 3500 kcal/d) from the 1990 FFQ or from the 1986 FFQ if the 1990 FFQ was not available. We used 1992 as baseline for this analysis, as it was the first time frailty characteristics were assessed. Women aged <60 y in 1992 entered follow-up when they reached age 60 in subsequent questionnaire cycles if they also had returned an FFQ in 1990. After excluding those who met the frailty criteria at entry, data from 78,366 women were included for analysis and followed up to 2014 (Supplemental Figure 1). The Harvard School of Public Health and Brigham and Women's Hospital Institutional Review Board approved the protocol of the study.

Dietary assessment

Dietary intake was assessed at baseline and in 1994, 1998, 2002, 2006, and 2010 using a validated self-administered semiquantitative FFQ (21) that contained ∼135 items. It included ∼25 vegetables, 9 whole fruits, and 3 juice items. Standard portion size was provided for each item and 9 frequency choices, ranging from <1 time/mo to ≥6 times/d were available. In this analysis, we considered all vegetables and only whole fruits. In validation studies, the deattenuated correlation coefficients ranged from 0.16 (yellow squash) to 0.53 (string beans) among vegetables and 0.50 (oranges) to 0.74 (apples) among fruits (22) as measured by the FFQ and multiple-week food records. In addition, biomarkers sensitive to fruit and vegetable intake, such as specific carotenoids and folate, were strongly associated with intakes of these nutrients based on our FFQ (23).

Assessment of frailty

Frailty is defined using the FRAIL scale that comprises 5 self-reported criteria: fatigue, poor strength (reduced resistance), low aerobic capacity, history of several major chronic illnesses, and a significant weight loss during the previous year (18). The FRAIL scale correlates well with the Fried frailty phenotype, and it has been adapted and validated to be used in several populations (24–26). In 1992, 1996, 2000, 2004, 2008, and 2012 participants completed the Medical Outcomes Study Short Form–36 Health Survey (SF-36), a 36-item questionnaire addressing 8 health dimensions, including physical and mental components (27). From the SF-36, we assessed the first 3 frailty criteria with the following questions: 1) for fatigue: “Did you have a lot of energy?,” with replies “some of the time” or “none of the time,” or with the question “I could not get going,” with responses “moderate amount” or “all of the time”; 2) for poor strength (low resistance): “In a normal day, is your health a limitation to walk up 1 flight of stairs?” with responses “yes” or “a lot”; and 3) for low aerobic capacity: “In a normal day, is your health a limitation to walk several blocks or several miles?” with responses “yes” or “a lot.” In addition, the illnesses criterion was ascertained from the question “In the last 2 y, have you had any of these physician-diagnosed illnesses?” Participants reporting ≥5 of the following diseases were considered to meet this criterion: cancer, hypertension, type 2 diabetes, angina, myocardial infarction, stroke, congestive heart failure, asthma, arthritis, chronic obstructive lung disease, Parkinson's disease, kidney disease, and depression. Finally, the weight loss criterion was defined as a ≥5% decrease in the weight reported in 2 consecutive follow-up cycles. At the end of each follow-up cycle, incident cases of frailty were defined as participants having ≥3 criteria in the FRAIL scale. Missing response in 3 or more components was assumed as missing on frailty status and excluded. For those with 1 or 2 missing responses, we were able to assess frailty status, considering missing in each characteristic as not having it.

Assessment of health and lifestyle characteristics

Participants’ self-reported height in 1976 was used to calculate updated BMI (kg/m²) with weight reported at each biennial questionnaire. Also assessed every 2 y were smoking status and the quantity of cigarette use, use of brand-specific multivitamins (yes or no), physical activity, use of medication for hypertension, diabetes, hyperlipidemia, and postmenopausal hormones. Recreational physical activity was assessed with 10 activities and was reported as hours per week and assigned a metabolic equivalent score. These scores were then summed over all activities to create a value in metabolic equivalent tasks in hours per week (28).

Statistical analysis

The primary exposure in this analysis was total fruit and vegetable intake. We computed cumulative averages from each FFQ to reduce within-person variation and represent long-term intake (29). For example, the average intake in 1990 and 1994 was used to predict frailty incidence between 1994 and 1996, and the average intake from 1990, 1994, and 1998 were used to predict frailty incidence between 1998 and 2000. Cumulative averages were computed using available valid FFQs. The cumulative average of fruit and vegetable intakes were then classified into predefined categories, with the lowest one as the reference group, and the association with incident frailty was examined using Cox proportional hazard models. Tests of trends were conducted by modeling intake as a continuous variable. We tested the proportional hazard assumption with a likelihood ratio test comparing the main model and a model with interaction terms between age and the exposure and 4-y time periods and the exposure. No violation of the assumption was observed. We also explored potential differences in associations by BMI and physical activity status with an analysis stratified by BMI at 25.0 and physical activity at the median. A test for interaction was performed using the likelihood ratio test comparing regression models with and without interaction terms.

In multivariable analysis, we adjusted for age (in months), energy intake (quintiles), alcohol intake (0, 1 to <5 g/d, 5 to <10 g/d, 10 to <15 g/d, and 15+ g/d), BMI (<23, 23 to <25, 25 to <28, 28 to <30, and 30+), physical activity (quintiles), smoking (never, past, current 1–14 cigarettes/d, 15–24 cigarettes/d, 25+ cigarettes/d), and use of aspirin, diuretics, β-blockers, calcium channel blockers, angiotensin converting enzyme inhibitors, and other antihypertensive medication, statins and other cholesterol lowering drugs, insulin, and oral hypoglycemic medication. As part of the multivariable model, we also adjusted for indicators of socioeconomic status, with highest academic degree (high school or lower, college, graduate degree) and census track income data (continuous). Missing data were categorized into a missing indicator. In addition, to account for other dietary characteristics, we adjusted for diet quality using a modified version of the Alternate Healthy Eating Index–2010 that omitted the fruit and vegetable components (30). Covariates were chosen based on their potential as confounders and predictors of components of the FRAIL score. We also explored the association of subgroups of fruit and vegetable intake with risk of developing frailty. In these analyses specific groups of fruits and vegetables were mutually adjusted for each other.

Although in the main analysis we included only women who did not meet the frailty criteria (i.e., having at least 3 frailty components) at baseline, some had 1 or 2 pre-existing frailty components. Therefore, we repeated the analysis among those without any frailty component at baseline to explore if the association of fruit and vegetable intake on frailty may differ depending on the baseline status. We also used this subset of participants to explore the association between fruit and vegetable intake and risk of developing each criterion of the FRAIL scale. We examined nonlinearity in the association by fitting restricted cubic splines. The test for nonlinearity used the likelihood ratio test, comparing the model with only the linear term to the model with the linear and the cubic spline terms. In addition, we examined potential reverse causation by lag time analysis, discarding the first 8 y of follow-up. Analyses were performed with the SAS software, version 9.4 (SAS Institute Inc.).

Results

In up to 20 y of follow-up, 12,434 women developed frailty (15.9%). Women with higher fruit and vegetable intake tended to be more physically active, were less likely to be a current smoker, and consumed more whole grains and more alcohol but less saturated fat (Table 1). Total fruit and vegetable intake was inversely associated with the risk of developing frailty (Table 2). In an age- and energy-adjusted model, the HR (95% CI) for women who consumed ≥7 servings of fruits and vegetables per day compared with those with <3 servings was HR: 0.55 (95% CI: 0.51, 0.59; P-trend < 0.001), and the multivariable HR for the same comparison of fruits and vegetables per day was HR: 0.92 (95% CI: 0.85, 0.99; P-trend = 0.03). We also observed a significant inverse trend (P = 0.01) for fruit intake. The test for nonlinearity was not significant. Although the inverse association with total fruits and vegetables did not differ by BMI status (Table 3), it appeared to be stronger for those with physical activity above the median (P-interaction < 0.05). Among physically active women who consumed more total fruits and vegetables, when compared with those who consumed <3 servings/d, the HR was 0.75 (95% CI: 0.63, 0.88) for 4 to <5 servings/d, 0.76 (95% CI: 0.64, 0.91) for 5 to <6 servings/d, 0.70 (95% CI: 0.59, 0.84) for 6 to <7 servings/d, and 0.68 (95% CI: 0.57, 0.81) for 7+ servings/d.

TABLE 1.

Lifestyle and dietary characteristics by categories of total fruit and vegetable intake (servings/d) at age 60 y¹

	Fruit and vegetable servings/d (n = 78,366)
	<3	3 to <4	4 to <5	5 to <6	6 to <7	7+
BMI, kg/m2	25.5 ± 5.2	25.5 ± 5.2	25.5 ± 5.1	25.4 ± 5.1	25.2 ± 5.1	25.2 ± 5.0
Physical activity, MET hr/wk	7.7 ± 13.7	8.8 ± 15.1	10.0 ± 14.8	11.3 ± 16.5	12.7 ± 18.7	16.1 ± 24.5
Current smoker, %	7.6	5.4	4.4	3.8	3.3	2.8
Medication use, % yes
Postmenopausal hormone	6.5	7.8	8.2	8.0	8.6	8.9
Diuretics	12.6	15.2	15.3	15.1	15.0	14.5
β-blockers	16.9	19.5	19.1	19.0	19.2	17.4
Calcium channel blockers	12.5	14.0	14.1	13.8	13.5	13.4
Other antihypertensive medication	7.5	7.9	7.5	7.5	6.9	6.9
ACE inhibitors	7.4	9.0	8.8	9.1	9.0	8.4
Statins	34.6	38.5	37.6	37.1	35.5	32.9
Other lipid-lowering medication	4.7	5.0	5.1	4.7	5.2	5.1
Insulin	2.2	2.5	2.2	2.3	2.1	1.9
Oral hypoglycemics	3.4	3.7	3.4	3.0	3.4	2.8
Dietary intake
Energy, kcal/d	1250 ± 433	1429 ± 443	1569 ± 466	1691 ± 473	1804 ± 494	2062 ± 535
Protein,2 g/d	64.1 ± 14.0	66.0 ± 13.0	70.0 ± 12.6	67.7 ± 12.5	68.3 ± 12.3	69.0 ± 12.5
Alcohol, g/d	4.8 ± 10.3	5.3 ± 10.2	5.5 ± 10.1	5.6 ± 10.0	5.7 ± 9.8	5.6 ± 9.8
Saturated fat,2 g/d	20.6 ± 5.7	18.8 ± 4.9	17.9 ± 4.6	17.3 ± 4.5	16.7 ± 4.3	15.3 ± 4.1
Whole grains,2 g/d	33.2 ± 22.4	34.7 ± 20.4	35.3 ± 19.1	35.3 ± 18.3	35.6 ± 17.8	35.8 ± 17.1
Total vegetables and fruits servings/d	2.1 ± 0.7	3.5 ± 0.3	4.5 ± 0.3	5.5 ± 0.3	6.5 ± 0.3	9.1 ± 2.1
Total vegetables, servings/d	1.2 ± 0.6	1.9 ± 0.6	2.4 ± 0.7	3.0 ± 0.8	3.6 ± 1.0	5.1 ± 1.8
Total fruit, servings/d	0.9 ± 0.6	1.6 ± 0.6	2.1 ± 0.7	2.5 ± 0.8	2.9 ± 0.9	3.9 ± 1.5
Cruciferous vegetables, servings/d	0.2 ± 0.1	0.2 ± 0.2	0.3 ± 0.2	0.4 ± 0.3	0.5 ± 0.3	0.7 ± 0.5
Yellow/orange vegetables, servings/d	0.2 ± 0.2	0.3 ± 0.2	0.3 ± 0.3	0.5 ± 0.3	0.6 ± 0.4	0.8 ± 0.6
Tomatoes, servings/d	0.3 ± 0.3	0.5 ± 0.3	0.6 ± 0.4	0.7 ± 0.4	0.8 ± 0.4	1.0 ± 0.6
Leafy vegetables, servings/d	0.2 ± 0.2	0.4 ± 0.3	0.6 ± 0.4	0.7 ± 0.4	0.8 ± 0.4	1.1 ± 0.7
Other vegetables, servings/d	0.3 ± 0.2	0.5 ± 0.3	0.7 ± 0.4	0.8 ± 0.5	1.0 ± 0.5	1.5 ± 0.8
Citrus fruits, servings/d	0.07 ± 0.1	0.1 ± 0.2	0.2 ± 0.3	0.2 ± 0.3	0.3 ± 0.3	0.4 ± 0.5
Apples and pears, servings/d	0.09 ± 0.1	0.2 ± 0.2	0.2 ± 0.2	0.3 ± 0.3	0.3 ± 0.3	0.5 ± 0.4
Berries, servings/d	0.1 ± 0.1	0.2 ± 0.2	0.2 ± 0.3	0.3 ± 0.3	0.4 ± 0.4	0.5 ± 0.5

¹Values presented as means ± SDs unless otherwise indicated. P-trend < 0.05 except for other lipid-lowering medication, P value computed by generalized linear models. ACE, angiotensin-converting enzyme; MET, metabolic equivalent.

²Energy adjusted.

TABLE 2.

Risk of frailty by categories of fruit and vegetable intake¹

	Fruit and vegetable intake servings/d (n = 78,366)
Total fruits and vegetables	<3	3 to <4	4 to <5	5 to <6	6 to <7	7+	P-trend
No. of cases	1620	2199	2362	2248	1628	2377
Person y	118,894	165,204	208,165	202,222	157,829	273,616
Age and energy adjusted	1	0.87 (0.82, 0.93)	0.72 (0.68, 0.77)	0.69 (0.65, 0.74)	0.64 (0.59, 0.69)	0.55 (0.51, 0.59)	<0.001
Multivariable adjusted2	1	1.01 (0.95, 1.08)	0.91 (0.85, 0.97)	0.96 (0.89, 1.03)	0.94 (0.87, 1.01)	0.92 (0.85, 0.99)	0.03
Total vegetable servings/d	<2	2 to <3	3 to <4	4 to <5	5+
No. of cases	3275	3834	2810	1410	1105
Person y	246,374	329,971	266,152	151,245	132,188
Age and energy adjusted	1	0.84 (0.80, 0.88)	0.77 (0.73, 0.81)	0.69 (0.65, 0.74)	0.67 (0.62, 0.72)		<0.001
Multivariable adjusted2	1	0.98 (0.94, 1.03)	0.97 (0.92, 1.03)	0.94 (0.88, 1.01)	0.99 (0.92, 1.07)		0.24
Total fruit servings/d	<1	1 to <2	2 to <3	3 to <4	4+
No. of cases	1257	4173	4150	1939	915
Person y	105,712	340,925	371,258	195,286	112,749
Age and energy adjusted	1	0.86 (0.81, 0.92)	0.73 (0.68, 0.78)	0.64 (0.59, 0.68)	0.56 (0.51, 0.62)		<0.001
Multivariable adjusted2	1	0.97 (0.90, 1.03)	0.94 (0.88, 1.01)	0.92 (0.85, 1.00)	0.91 (0.83, 1.00)		0.01

¹Values presented as HRs (95% CIs) unless otherwise indicated.

²Adjusted for age, smoking, energy intake, BMI, physical activity, postmenopausal hormone use, aspirin, antihypertensive medications, lipid lowering medications, diabetes medication, insulin, highest academic degree, census track income data, alcohol, and a modified Alternate Healthy Eating Index that does not include fruits and vegetables. HRs computed with Cox proportional hazard models.

TABLE 3.

Fruit and vegetable intake and risk of frailty, stratified by BMI and physical activity, (n = 78,366).¹

	Fruit and vegetable servings/d
	< 3	3 to <4	4 to <5	5 to <6	6 to <7	7+	P-trend
BMI,2 <25.0	1	0.98 (0.88, 1.08)	0.88 (0.79, 0.97)	0.94 (0.84, 1.04)	0.97 (0.86, 1.09)	0.89 (0.79, 1.01)	0.11
BMI,2 ≥25.0	1	1.05 (0.96, 1.15)	0.95 (0.87, 1.04)	0.98 (0.89, 1.08)	0.93 (0.84, 1.03)	0.96 (0.86, 1.06)	0.21
Physical activity,3 <median	1	0.98 (0.91, 1.06)	0.87 (0.80, 0.94)	0.90 (0.83, 0.97)	0.89 (0.81, 0.97)	0.86 (0.79, 0.94)	<0.001
Physical activity,3 ≥median	1	0.88 (0.74, 1.05)	0.75 (0.63, 0.88)	0.76 (0.64, 0.91)	0.70 (0.59, 0.84)	0.68 (0.57, 0.81)	<0.001
Total vegetable servings/d	<2	2 to <3	3 to <4	4 to <5	5+
BMI,2 <25.0	1	0.98 (0.91, 1.05)	0.91 (0.89, 1.06)	0.93 (0.84, 1.04)	0.99 (0.87, 1.12)		0.17
BMI,2 ≥25.0	1	1.01 (0.94, 1.08)	0.99 (0.92, 1.07)	0.97 (0.89, 1.07)	1.03 (0.93, 1.14)		0.92
Physical activity,3 <median	1	0.96 (0.91, 1.01	0.94 (0.88, 0.99)	0.90 (0.83, 0.97)	0.85 (0.86, 1.04)		0.02
Physical activity,3 ≥median	1	0.88 (0.79, 0.99)	0.84 (0.74, 0.96)	0.81 (0.69, 0.93)	0.84 (0.72, 0.99)		0.01
Total fruit servings/d	<1	1 to <2	2 to <3	3 to <4	4+
BMI,2 <25.0	1	0.97 (0.87, 1.08)	0.94 (0.85, 1.05)	0.94 (0.83, 1.06)	0.93 (0.81, 1.08)		0.23
BMI,2 ≥25.0	1	0.96 (0.88, 1.05)	0.94 (0.86, 1.03)	0.91 (0.82, 1.01)	0.89 (0.78, 1.01)		0.01
Physical activity,3 <median	1	0.94 (0.87, 1.01)	0.88 (0.81, 0.95)	0.89 (0.81, 0.97)	0.87 (0.77, 0.97)		0.001
Physical activity,3 ≥median	1	0.86 (0.73, 1.02)	0.83 (0.70, 0.98)	0.69 (0.58, 0.83)	0.72 (0.58, 0.89)		<0.001

¹Values presented as multivariable HRs (95% CIs) unless otherwise indicated. Adjusted for age, smoking, energy intake, BMI, physical activity, postmenopausal hormone use, aspirin, antihypertensive medications, lipid lowering medications, diabetes medication, insulin, highest academic degree, census track income data, alcohol, and a modified Alternate Healthy Eating Index that does not include fruits and vegetables. HRs computed with Cox proportional hazard models.

²Case count: BMI <25 = 5336, BMI ≥25 = 7098; P-interaction ≤ 0.05 for total fruits and vegetables and total vegetables.

³Case count: physical activity <median = 9559, physical activity ≥median = 2875; P-interaction ≤ 0.05 total vegetables, total fruit, and total fruits and vegetables.

Among women without any frailty component at baseline, 6288 frailty incidents occurred during follow-up. An inverse association was observed when we compared 4+ servings/d with <1/d of fruit intake (HR: 0.83; 95% CI: 0.72, 0.96) (Supplemental Table 1). When we examined each frailty component, the most consistent finding appeared to be for fatigue, with independent inverse associations for intake of total fruits and vegetables (HR for 7+ compared with <3 servings/d: 0.81; 95% CI: 0.76, 0.86), vegetables (HR for 5+ compared with <2 servings/d: 0.83; 95% CI: 0.78, 0.88), and fruits (HR for 4+ compared with <1 serving/d: 0.82; 95% CI: 76, 0.89) (Supplemental Table 2). When we explored the association of specific types of fruits and vegetables, we observed a clear inverse association with leafy vegetables (P-trend < 0.001), yellow vegetables (P-trend < 0.001) (Supplemental Table 3), and apples and pears (P-trend < 0.001) (Supplemental Table 4).

Results for 8-y lag analysis (10,068 cases) showed some attenuation of the association, but the linear trend remained statistically significant for total fruits (P = 0.006) (Supplemental Table 5).

Discussion

We observed a clear inverse trend between total fruit and vegetable intake and the risk of developing frailty among women aged ≥60 y. The association was similar regardless of BMI status. While this association was observed in women both above and below the physical activity median, it was particularly strong for those above the physical activity median.

Existing data on fruit and vegetable intake and frailty is sparse and composed of relatively small studies. Results were a mix of no association (8, 31) and inverse association separately for fruits, vegetables (15), or specific vegetable groups (16). Most of these studies also lacked adjustment for energy intake or did not control for dietary characteristics (8, 16, 31). In some studies, the measurement of diet was crude and consisted of questions on overall fruit or vegetable intake without information on specific fruits and vegetables (16, 31). In contrast, our FFQ has ∼37 fruit and vegetable items. This comprehensive assessment of intake allowed us to examine the associations of specific fruit and vegetable groups with frailty outcomes.

Antioxidants in fruits and vegetables may limit excessive oxidative stress, which plays a role in inflammation, and in turn may accelerate aging and frailty development (32–34). Excessive oxidation may impair muscle preservation in aging, resulting in sarcopenia that leads to poor strength and reduced physical function (35). Serum carotenoids level was shown to be inversely associated with muscle strength decline (36). Thus, this could also be a pathway through which fruit and vegetable intake may contribute to the development of frailty. In addition, fiber and other phytochemicals may reduce the risk of cardiovascular and other chronic diseases (13), of which poor physical function could be a sequela. Phytochemicals and antioxidant nutrients such as vitamin C may also improve immune function in older adults through cell-mediated and non–cell-mediated pathways (10), thus preserving ambulation ability and physical functioning. Some fruit have a high glycemic index, and different fruits and vegetables vary greatly in polyphenols, vitamins, and sugar content, which may contribute to the different results observed in individual fruit and vegetable groups.

Obesity has been associated with functional decline (37) and the development of frailty (38–40). Increased oxidative stress due to higher levels of circulating inflammatory molecules may reduce the already lower anabolic activities in muscles in aging (41). This may exacerbate functional decline and accelerate progression to disability. In our analysis, obesity was not a clear effect modifier, and the association between total fruit and vegetable intake appeared to be independent of obesity. On the other hand, we noted potential interactions with physical activity, with stronger associations among women who had higher fruit and vegetable intakes and were also physically active. Physical activity has been shown to improve physical functioning and decrease frailty in older people (42, 43). Physical activity may mitigate the reduced anabolic capacity in muscles through increasing activity in the insulin-like growth factor I pathway (44). When this is coupled with antioxidants in fruits and vegetables to reduce the catabolic effects of oxidative stress on muscles in aging, physical activity and a healthy diet thus may be an effective combination to preserve muscle mass and physical functioning. In addition, both physical activity (45) and fruit and vegetable intake (46) are associated with lower risk of chronic diseases, a key determinant of disability (47). Therefore, the combination of high amounts of physical activity and a healthy diet characterized by high fruit and vegetable intake may have additive or synergistic potential in preventing frailty.

The strengths of this study included detailed and repeated dietary information that allowed us to examine different groups of fruits of vegetables as well as consider changes in consumption of these groups. We also had detailed lifestyle information for fine control of confounders. Our thorough analysis included examining the individual components of frailty and potential differential association due to pre-existing frailty characteristics.

There is no standard definition of frailty, but fatigue/exhaustion, weakness, weight loss, and sometimes presence of morbidities are common components (1, 18). In the absence of uniform diagnostic criteria, previous studies used different characteristics to define frailty and also set the criteria differently based on the types of data available. On one hand, for example, Garcia-Equinas et al. (15) assessed the exhaustion component of the Fried (1) definition of frailty with questions from the Center for Epidemiologic Studies Depression Scale and the weakness component based on measured grip strength. On the other hand, Johannessen defined frailty with 8 characteristics such as self-reported weight loss and measured walking speed (31). In this study, we used the FRAIL scale, a definition of frailty based on an international consensus for physical frailty (48), and used self-reported data to determine fulfillment. The varied definition complicates the comparison of results in the existing literature, but nevertheless, our results do support fruit and vegetable intake having an association with the development of frailty. In the present analysis, we adapted the exact questions in the FRAIL scale (18) in the NHS questionnaires. The FRAIL scale included questions based on self-reported information. However, it has been shown to be correlated (r = 0.617, P < 0.001) with the Fried scale (1), the most widely used scale for frailty assessment, which includes both self-reported and performance-based measures.

Frailty can detrimentally affect the ability to prepare food and may therefore limit fruit and vegetable intake in affected individuals. This may explain the weak direct association observed in some fruits and vegetable groups, as our extensive food list contains items such as prunes and canned fruits that may be more highly used by individuals who have some physical impairment. This finding also contributes to the complexity in the study of body weight and mortality. Although the potential for reverse causation in this analysis was reduced due to its prospective nature, reversed causation cannot be totally discounted. While results for the 8-y lag analysis were weaker, a signal for an inverse association for fruits and vegetables was nevertheless observed. As in all observational studies, complete elimination of confounding is difficult. However, we adjusted for a broad range of lifestyle, socioeconomic, and dietary factors in finely categorized groups. In addition, with repeated measurements of lifestyle and diet, we were able to update these potential confounders throughout follow-up. Therefore, any residual confounding should be small. Finally, because several definitions of frailty exist, our results should be confirmed in studies using other definitions, such as the Fried scale or the Rockwood index (49).

In conclusion, in this US cohort of nurses age ≥60 y, fruit and vegetable intake was associated with lower risk of frailty The inverse association was particularly strong when high fruit and vegetable intake was combined with high levels of physical activity. Although additional data are needed to confirm our results, this study adds to the evidence that supports the role of fruit and vegetable intake in decreasing the risk of individual chronic conditions.

Notes

Supported by the Instituto de Salud Carlos III, State Secretary of R + D + I of Spain and FEDER/FSE grants FIS 16/1512, 13/609, and 19/319; the Joint Programming Initiative: A Healthy Diet for a Healthy Life, SALAMANDER project; and NIH grant UM1 CA186107.

Supplemental Tables 1–5 and Supplemental Figure 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.

Abbreviations: NHS, Nurses’ Health Study; SF-36, Short Form–36 Health Survey