Abstract
BACKGROUND:
The link between nut consumption and cardiovascular (CV) mortality remains unclear.
OBJECTIVE:
to examine whether nut consumption is associated with CV mortality and estimate the proportion of reduced risk of CV mortality explained by intermediate factors.
METHODS:
We studied 39,167 women from the Women’s Health Study; 28,034 provided blood samples. Nut consumption was self-reported at baseline and at follow-up using a food frequency questionnaire. Our primary outcome was cardiovascular death, which was ascertained via medical records, confirmed with the national death index and death certificates.
RESULTS:
During a mean follow-up of 19 years, 959 CV deaths occurred. In a multivariable Cox regression model adjusting for age, body mass index, smoking, alcohol use, physical activity, postmenopausal status, marital status, family history of premature myocardial infarction and the alternate healthy eating index score, hazard ratios for CV mortality were 0.93 (0.76–1.14) for nut consumption of 1–3 times/month, 0.84 (0.69–1.01) for nut intake of 1 time/week, and 0.73 (0.61–0.87) for nut consumption of ≥2 times/week when compared to women who did not consume nuts (p = 0.0004). LDL and total cholesterol accounted for about 19%, HbA1c 18% and all mediating factors together accounted for about 6.6% of the lower risk of CV mortality for those who consumed nuts ≥2 times/week. For the secondary outcome of CV events, although the effect was noted to be in the same direction with increasing nut consumption associated with lower risk of CV events, it was not statistically significant (p = 0.07).
CONCLUSION:
This study suggests that nut consumption is inversely associated with cardiovascular mortality in women. Lipids, inflammatory markers and glucose metabolism account for a modest proportion of the lowered CV mortality observed with nut consumption, assuming a causal nut-CV mortality association.
Keywords: Nut consumption, Cardiovascular disease, Cardiovascular mortality, Mediating factors, Lipids
Introduction
Cardiovascular mortality remains a major cause of death in the United States and worldwide.1 A randomized trial found that an intervention with a Mediterranean diet (which included walnuts, hazelnuts, or almonds) led to a reduction in major cardiovascular events as compared to a control diet without nuts.2 Nuts are recommended in the 2015 federal dietary guidelines for prevention of cardiovascular disease and in the American Heart Association’s dietary metrics for ideal cardiovascular health (goals of health promotion for 2020 report).1,3 Nuts are high in unsaturated (polyunsaturated and monounsaturated) fats, vitamins, minerals and antioxidants, and are a source of fiber.2
An inverse association between nut consumption and all-cause mortality has been reported in several large prospective cohorts.4-6 However, fewer studies have examined the association of nut consumption with cardiovascular (CV) mortality. Some have found that nuts were associated with a significantly lower CV mortality, while others have found no such association.5,7-14 Given that there are inconsistent results on the association between total nut consumption and CV risk and mortality, further research is needed to elucidate this phenomenon.
Factors that have been shown to be associated with nut consumption include hyperlipidemia, type-2 diabetes, hypertension, obesity and inflammatory markers.6 In the Nurses Health Study and the Shanghai Women’s Health Study, nut consumption was inversely associated with risk of type-2 diabetes in multivariate adjusted models, however this effect was not observed in the Iowa Women’s Health Study and the Physicians’ Health Study.12,15,16 Observational studies have found that nut consumption leads to a lower oxidative stress, inflammatory markers, insulin resistance, central adiposity, and endothelial dysfunction.12 Inflammatory markers, such as plasma levels of ICAM-1 (intercellular adhesion molecule 1), vascular cell adhesion molecules (VCAM-1), or IL (interleukin)-6 were found to be lower after consumption of nuts in randomized trials.17,18
Nut consumption has also been linked to improvement in blood lipid levels in a dose related manner. In a pooled analysis of 25 intervention trials, those who consumed 20% of their dietary intake from nuts had a reduction in LDL cholesterol by 9.5 mg/dL.19 However, no prior study has estimated the proportion of reduced mortality risk accounted for by each of these factors. Thus, we examined the association between nut consumption and CV mortality and estimated the proportion of reduced risk of CV mortality accounted for by several intermediate factors. An improved understanding of mediating factors and mechanisms of nut consumption on CV mortality would have important public health implications and could lead to the design of diets that maximally reduce chronic disease risk and mortality.20
Methods
Study population
Our participants included women from the Women’s Health Study (WHS), a randomized trial (clinicaltrials.gov: NCT00000479, 1992–2004) to study the benefits and risks of low dose aspirin and Vitamin E in the primary prevention of cardiovascular disease and cancer among female health professionals over the age of 45 years. At the completion of the trial, 89% of these women elected to continue with observational follow-up.21 Our cohort includes women who did not have cardiovascular disease at study entry. Detailed methods of the Women’s Health Study have been published previously.22 The Institutional Review Board at Brigham and Women’s Hospital approved the study protocol and all participants gave written informed consent.
The Women’s Health Study trial began in 1991. From September 1992 to May 1995, each woman received an invitation letter and a questionnaire on demographic data, health habits, and medical history. Eligible and interested women were entered into a 3-month run-in phase, receiving placebos and a questionnaire. At the end of the run-in phase, 39,876 women who indicated continued willingness and eligibility were randomized into the trial, using a 2 × 2 × 2 design, to aspirin, vitamin E, and β-carotene, or placebo. A self-administered food frequency questionnaire (FFQ) was sent to each participant during the run-in phase. A second FFQ was collected from participants at the trial’s end in 2004. The WHS cohort includes 39,876 participants, of which 39,310 completed the food frequency questionnaires during these two periods. We excluded women with pre-existing cardiovascular disease (14), and those with missing data on nut consumption (129). After these exclusions, the remaining 39,167 participants were included in the current analyses.
Ascertainment of nut consumption and covariates of interest
Nut consumption (peanuts and tree nuts) was ascertained from a validated food frequency questionnaire administered during the run-in phase of the study, prior to randomization and then administered again as part of the final risk-factor questionnaire at the end of the trial year. In the FFQ, nut consumption was categorized as how often, on average, the participants had consumed nuts during the previous year, possible answers were: never/almost never, 1 to 3 times per month, once a week, 2 to 4 times a week, 5 to 6 times a week, once a day, 2 to 3 times a day, 4 to 6 times a day, or more than 6 times a day.
Demographic data included the following covariates: age, body mass index (BMI) (kg/m2), marital status, smoking status (never, past, current), alcohol (grams/d), physical activity, education, menopausal status, hormone replacement therapy randomization arm, family history of premature myocardial infarction in a parent and the alternate healthy eating index score, excluding nuts. The scoring criteria of the alternate healthy eating index have been previously described.23 Briefly, the alternate healthy index is a score based on 11 components: six components of the highest ideal intake foods (vegetables, whole grains, fruits, nuts and legumes, long chain omega-3 fatty acids), one component for moderate intake (alcohol), and four components for the lowest ideal intake (sugar sweetened beverages, red and processed meats, trans fats, sodium). Each component is given a score between 0 and 10. The components are summed to obtain a total score, ranging from 0 to 110, with a higher score indicating a healthier diet.
Ascertainment of cardiovascular mortality and incident coronary heart disease
The WHS Endpoint Committee reviewed hospital records for all cases of myocardial infarction, coronary revascularization and angioplasty or cardiovascular death reported after study enrollment. Myocardial infarction was confirmed based on the World Health Organization criteria and elevated cardiac enzymes or suggestive ECG changes and revascularization or angioplasty procedures were confirmed via hospital records. Stroke cases were verified if a patient had new neurological deficits persisting for more than 24 h, often confirmed on imaging (CT scans). Autopsy reports and death certificates were used to confirm death. Details on the ascertainment of CV mortality and incident heart disease in WHS have been previously published.22 Incident cardiovascular disease included myocardial infarction, coronary angioplasty and revascularization procedures.
Blood collection and measurement of biomarkers
Of the 39,167 women in the cohort, 28,034 provided blood samples. Blood samples from participants were collected in EDTA tubes and shipped cooled to a core laboratory, where the samples were centrifuged and frozen at −170° Celsius (vapor phase liquid nitrogen) until the time of analysis. Hemoglobin A1c was measured using turbidometric immunoassay in red blood cells with the Hitachi 911 Analyzer (Roche Diagnostics, Indianopolis, IN). Soluble ICAM was measured with an enzyme-linked immunosorbent assay (R & D Systems, Minneapolis, MN). High sensitivity C-reactive protein was measured using a high-sensitivity ELISA (Abbot Laboratories). Total cholesterol and HDL cholesterol concentrations were measured using a Hitachi 911 autoanalyzer (Roche Diagnostics, Basel, Switzerland), and LDL cholesterol (Genzyme, Cambridge, MA). All assays had been approved for clinical use by the United States Food and Drug Administration. Blood samples were not collected after the second FFQ.
Statistical analysis
We classified each participant into the following categories of nut consumption: never, 1–3 times/month, 1/week and ≥2 times/week. We collapsed the last five categories of nut consumption due to insufficient numbers for stable estimate of effect. For those whose follow-up time ended prior to the second nut consumption recording, the first nut consumption recorded was used in the model. For those whose follow-up time continued until the end of the study, nut consumption and covariates were updated to include data from the second food frequency questionnaire (prior to randomization). We calculated the person-time of follow-up from baseline to the occurrence of CV death or until the censoring date (date of the last follow-up questionnaire). Using multivariate adjusted Cox proportional regression models, we computed the hazard ratios and corresponding 95% confidence intervals using participants in the “never/almost never” group of nut consumption as the reference category. The initial model adjusted for age. The multivariate adjusted for age (continuous), smoking (current smoker yes/no), body mass index (calculated as kg of weight/height in meters squared), alcohol (g/day), physical activity (met-hours/week), postmenopausal status (yes or no), family history of myocardial infarction in parent <60 years-old (yes or no), marital status (categories), and the alternate healthy eating index score excluding nuts (continuous) at baseline. We evaluated confounding using a 10% change in the estimate.
To determine the extent to which the lower risk of CV mortality associated with nut consumption was explained by potential intermediate factors, we used the difference method approach.24 First, we fitted a Cox regression model with the exposure, outcome and covariates of interest, and estimated the beta coefficients. We then fitted another model that included both the exposure, outcome, covariates of interest and a potential mediating factor of interest one at a time and estimated the beta coefficients. The potential mediating factors included hemoglobin A1c, Total cholesterol, HDL cholesterol, LDL cholesterol, ICAM, and hsCRP. We then computed the proportion of the effect of nut consumption on CV mortality that was explained by the potential mediator by determining the magnitude of the change in the beta coefficients for those with ≥2/week of nut consumption compared to those who did not eat nuts with the addition of each intermediate factor of interest. To calculate the proportion of the effect we used the following pre-defined equation: Proportion of effect = [total effect without potential mediator – effect with potential mediator/total effect without potential mediator] × 100%, in which the effect was the beta coefficient estimated by each of the two aforementioned models.
We conducted sensitivity analyses to examine the robustness of our findings. We excluded participants who died within the first two years to address the concern of occult chronic disease. We also excluded women with very low BMI (<18.5) and high BMI (>40). All analyses were conducted using SAS, version 9.4 (SAS Institute, NC), and significance level was set at 0.05.
Results
Characteristics
We studied 39, 167 participants in the WHS with a mean age at randomization of 54.6 ± 7.0 years (range 38.7–89.9 years) in a prospective cohort design. Tables 1a and 1b presents the baseline characteristics and lifestyle factors for the first nut consumption FFQ (during the run-in phase) and second nut consumption FFQ (prior to randomization). Nut consumption was associated with higher educational attainment, energy consumption, alcohol consumption, and higher HDL cholesterol. Those in the higher nut consumption category (≥2 times/week) had overall higher energy intake (2038 kcal/day) as compared to those who did not consume nuts (1602 kcal/day). Women in the higher nut consumption category (≥2 times/week) also had higher fruit and vegetable consumption per day (mean 7.4 ± 4.5 vs 5.7 ± 3.5) and higher whole grain servings (mean 1.7 ± 1.4 vs 1.3 ± 1.2) per day as compared to those who did not consume nuts.
Table 1a.
Baseline characteristics of 39,167 participants in the Women’s Health Study according to frequency of nut consumption.
| Never (n = 18,817) | 1-3/month (n = 8046) | 1/week (n = 8558) | ≥2/week (n = 3746) | |
|---|---|---|---|---|
| Age (y) | 54.3 ± 7.0 | 54.7 ± 7.0 | 54.7 ± 7.0 | 56.0 ± 7.4 |
| BMI(kg/m2) | 26.2 ± 5.2 | 25.9 ± 4.9 | 26.0 ± 5.0 | 25.6 ± 4.9 |
| Married (%) | 70.2 | 72.4 | 73.1 | 69.4 |
| Smoking status | ||||
| Never smoker (%) | 50.5 | 51.8 | 51.2 | 51.3 |
| Ex-smoker (%) | 36.2 | 35.4 | 36.0 | 34.9 |
| Current smoker (%) | 13.3 | 12.7 | 12.8 | 13.8 |
| Physical activity (MET-hours/week) | 14.4 ± 18.5 | 14.2 ± 17.8 | 14.4 ± 17.9 | 15.5 ± 19.0 |
| Highest level of education | ||||
| Associate or BS degree (%) | 82.0 | 79.4 | 78.9 | 77.2 |
| MS or Doctorate/MD degree (%) | 18.0 | 20.6 | 21.1 | 22.8 |
| Post-menopausal (%) | 52.4 | 55.3 | 55.0 | 61.0 |
| Family history of MI (%) | 14.9 | 13.9 | 13.7 | 12.8 |
| Hypertension (%) | 26.0 | 25.6 | 25.5 | 26.0 |
| Hyperlipidemia (%) | 29.3 | 29.1 | 29.6 | 30.3 |
| Diabetes Mellitus (%) | 2.6 | 2.4 | 2.5 | 2.8 |
| Dietary variables | ||||
| aHEI score excluding nuts | 51.3 ± 9.9 | 51.1 ± 9.6 | 51.0 ± 9.5 | 51.8 ± 10.1 |
| Energy (Kcal) | 1602 ± 498 | 1723 ± 510 | 1874 ± 531 | 2038 ± 565 |
| Alcohol (g/day) | 3.6 ± 7.9 | 4.1 ± 7.9 | 4.5 ± 8.6 | 5.5 ± 10.3 |
| Total fat (g) | 56.0 ± 12.2 | 58.2 ± 11.1 | 59.3 ± 10.9 | 62.1 ± 11.2 |
| Saturated fat (g) | 19.4 ± 5.1 | 19.9 ± 4.7 | 19.9 ± 4.5 | 19.8 ± 4.6 |
| Fruit and vegetable (serv/day) | 5.7 ± 3.5 | 6.0 ± 3.2 | 6.6 ± 3.8 | 7.4 ± 4.5 |
| Red meat (serv/day) | 0.7 ± 0.5 | 0.7 ± 0.6 | 0.8 ± 0.7 | 0.9 ± 0.8 |
| Whole grain (serv/day) | 1.3 ± 1.2 | 1.4 ± 1.2 | 1.5 ± 1.2 | 1.7 ± 1.4 |
| Biomarkersa | ||||
| HbA1c | 5.1 ± 0.63 | 5.1 ± 0.62 | 5.1 ± 0.58 | 5.1 ± 0.67 |
| Total cholesterol (mg/dL) | 211 ± 42 | 212 ± 41 | 212 ± 42 | 213 ± 42 |
| HDL cholesterol (mg/dL) | 53.0 ± 14.9 | 54.3 ± 15.0 | 54.1 ± 15.1 | 55.2 ± 15.5 |
| LDL cholesterol (mg/dL) | 124 ± 34 | 125 ± 34 | 124 ± 34 | 125 ± 34 |
| Soluble ICAM-1 (ng/ml) | 355 ± 83 | 354 ± 83 | 356 ± 83 | 353 ± 84 |
| High sensitive CRP (mg/L) | 3.8 ± 5.9 | 3.5 ± 5.3 | 3.7 ± 5.9 | 3.3 ± 4.6 |
The Women’s Health Study cohort includes 39,786 participants, of which 39,310 completed the food frequency questionnaire. Nut consumption was ascertained from the food frequency questionnaire collected during the run-in phase of the study, prior to randomization. Those with pre-existing cardiovascular disease (14) and missing data on nut consumption (129) were excluded, with the remaining 39,167 participants included in this analysis.
BMI = body mass index, MI = myocardial infarction, aHEI = alternate healthy eating index, HbA1c = hemoglobin A1c, HDL = high-density lipoprotein, LDL: low-density lipoprotein, ICAM-1 = intercellular adhesion molecule-1, hs-CRP = high sensitivity c-reactive protein.
Characteristics are shown as follows: mean ± standard deviation for continuous variables and frequencies (percentages) for categorical variables.
Numbers of participants with missing covariates are as follows: BMI (49), smoking status (35), physical activity (1), postmenopausal status (67), family history (674), diabetes mellitus (20), total calories consumed daily (792), and alcohol (10).
Blood samples were collected in a subset of the population (28,034).
Table 1b.
Characteristics of 32,148 Participants in Women’s Health Study According to Frequency of Nut Consumption at the time of the second Food Frequency Questionnaire.a
| Never (n = 2615) | 1-3/month (n = 3169) | 1/week (n = 6326) | ≥2/week (n = 20,038) | |
|---|---|---|---|---|
| Age (y) | 65.1 ± 7.1 | 65.0 ± 7.1 | 64.8 ± 6.8 | 64.8 ± 6.7 |
| BMI(kg/m2) | 27.2 ± 5.9 | 27.4 ± 5.8 | 27.0 ± 5.5 | 26.6 ± 5.2 |
| Smoking status | ||||
| Never smoker (%) | 46.5 | 48.4 | 49.4 | 50.9 |
| Ex-smoker (%) | 43.5 | 42.7 | 42.9 | 43.1 |
| Current smoker (%) | 10.0 | 8.9 | 7.7 | 6.0 |
| Physical activity (MET-hours/week) | 14.4 ± 18.8 | 14.5 ± 17.7 | 16.2 ± 18.0 | 19.4 ± 19.8 |
| Post-menopausal (%) | 94.7 | 94.0 | 94.3 | 94.9 |
| Hypertension (%) | 58.1 | 57.3 | 56.9 | 52.6 |
| Hyperlipidemia (%) | 60.6 | 61.2 | 61.4 | 59.8 |
| Diabetes Mellitus (%) | 7.6 | 7.5 | 7.2 | 7.2 |
| Dietary variables | ||||
| aHEI score excluding nuts | 52.0 ± 11.4 | 52.6 ± 11.0 | 54.4 ± 10.9 | 57.0 ± 11.4 |
| Alcohol (g/day) | 4.4 ± 8.9 | 4.1 ± 8.1 | 4.4 ± 8. | 4.5 ± 8.2 |
Blood samples were not collected at the time of the second food frequency questionnaire.
Nut consumption and risk of CV mortality
During a mean follow-up of 19 years, 959 CV deaths occurred. In a multivariable Cox regression model adjusting for age, body mass index, smoking, alcohol use, physical activity, postmenopausal status, family history of premature myocardial infarction, marital status, and the alternate healthy eating index score, hazard ratios for CV mortality were 0.93 (95% CI: 0.76–1.14) for nut consumption of 1–3 times/month, 0.84 (95% CI: 0.69–1.01) for nut intake of 1 time/week, and 0.73 (95% CI: 0.61–0.87) for nut consumption of ≥2 times/week when compared to women who did not consume nuts, p linear trend 0.0004 (Table 2).
Table 2.
Hazard ratios (95% CI) of cardiovascular mortality according to categories of nut consumption in 39,167 participants of the Women’s Health Study.
| Frequency of total nut consumption |
|||||
|---|---|---|---|---|---|
| Never | 1-3/month | 1/week | ≥2/week | p for trend | |
| Crude model | 1.00 (ref) | 0.89 (0.73–1.08) | 0.76 (0.63–0.91) | 0.61 (0.51–0.72) | <0.0001 |
| Age adjusted model | 1.00 (ref) | 0.86 (0.71–1.05) | 0.74 (0.61–0.89) | 0.59 (0.49–0.70) | <0.0001 |
| Multivariable model 1a | 1.00 (ref) | 0.93 (0.76–1.13) | 0.82 (0.68–0.98) | 0.70 (0.59–0.84) | <0.0001 |
| Multivariable model 2b | 1.00 (ref) | 0.93 (0.76–1.14) | 0.84 (0.69–1.01) | 0.73 (0.61–0.87) | 0.0004 |
The never nut consumption category is the reference category.
PY = person years.
Adjusted for age, body mass index, alcohol (g/day), physical activity, smoking, postmenopausal status (yes or no), and family history of myocardial infarction in parent <60 years-old (yes or no).
Adjusted for the above variables plus marital status and the alternate healthy eating index score excluding nuts.
Additionally, we performed analysis to evaluate the association of nut consumption with incident CV (cardiovascular) events. The effect was noted to be in the same direction with increasing nut consumption associated with lower risk of CV events. Although the age adjusted model was statistically significant (p for trend = 0.005), the multivariable model did not reach statistical significance (p for trend = 0.07) (Suppl Table 1).
Sensitivity analyses
In sensitivity analyses excluding women who died within the first two years, and those who had a BMI ≤18.5 or BMI ≥40, did not materially alter the findings (Table 3). The fully adjusted HRs after excluding women who died in the first two years for CV mortality were 0.92 (0.75–1.13) for nut consumption of 1–3 times/month, 0.85 (0.70–1.03) for once per week, and 0.73 (0.61–0.87) for nut consumption of ≥2 times/week as compared to the group that did not consume nuts (Table 3). The multivariable adjusted HRs excluding those with BMI ≤ 18.5 were 0.92 (0.75–1.13) for nut consumption of 1–3 times/month, 0.83 (0.69–1.00) for once per week, and 0.72 (0.60–0.86) for nut consumption ≥ 2 times/week as compared to the group that did not consume nuts. The HRs excluding those with BMI ≥40 were 0.93 (0.76–1.14) for nut consumption of 1–3 times/month group, 0.84 (0.70–1.02) for once per week, and 0.75 (0.62–0.90) for nut consumption ≥ 2 times/week (Table 3).
Table 3.
Sensitivity analysis: risk ratios (95% CI) of CV mortality according to categories of nut consumption.
| Cardiovascular mortality | |||||
|---|---|---|---|---|---|
| Frequency of nut consumption |
Never | 1-3/month | 1/week | ≥2/week | p for trend |
| Excluding those who died in the first two years (n = 39,082) | |||||
| Multivariable model 1a | 1.00 | 0.92 (0.75–1.13) | 0.83 (0.68–1.00) | 0.69 (0.58–0.83) | <0.0001 |
| Multivariable model 2b | 1.00 | 0.92 (0.75–1.13) | 0.85 (0.70–1.03) | 0.73 (0.61–0.87) | 0.0004 |
| Excluding those with BMI ≤ 18.5 (n = 38,785) | |||||
| Multivariable model 1a | 1.00 | 0.92 (0.75–1.12) | 0.81 (0.67–0.98) | 0.69 (0.58–0.83) | <0.0001 |
| Multivariable model 2b | 1.00 | 0.92 (0.75–1.13) | 0.83 (0.69–1.00) | 0.72 (0.60–0.86) | 0.0003 |
| Excluding those with BMI ≥ 40 (n = 38,430) | |||||
| Multivariable model 1a | 1.00 | 0.93 (0.76–1.14) | 0.83 (0.68–1.00) | 0.72 (0.60–0.86) | 0.0002 |
| Multivariable model 2b | 1.00 | 0.93 (0.76–1.14) | 0.84 (0.70–1.02) | 0.75 (0.62–0.90) | 0.0011 |
The never nut consumption category is the reference category.
BMI = body mass index (kg/m2).
Adjusted for age, BMI, smoking, alcohol (g/day), and physical activity, postmenopausal status (yes or no), and family history of myocardial infarction in parent <60 years-old (yes or no).
Adjusted for the above variables plus marital status, the alternate healthy eating index score excluding nuts.
Proportion of nut consumption – CV death association explained by mediators
The proportion of change of the effect estimate was 19.4% for LDL cholesterol, 18.8% for total cholesterol, 19.8% for HDL cholesterol, 18.3% for hsCRP, 17.6% for HbA1c, and 13.7% for ICAM, assuming a causal relation between nut consumption and CV mortality (for those in the highest nut consumption category (see Table 4). Together, all these intermediate factors accounted for about 6.6% of the proportion of change in estimates.
Table 4.
Mediating factors and the percent of the total effect explained by each factor for those with ≥2/week nut consumption (n = 28,034).a
| Model | Beta coefficient ± standard error | Percent of total effect explained by mediator |
|---|---|---|
| Multivariable modelb for ≥ 2/week nut consumption | −0.313 ± 0.09 | – |
| Multivariable model + HbA1c | −0.368 ± 0.11 | 17.6% |
| Multivariable model + LDL | −0.374 ± 0.11 | 19.4% |
| Multivariable model + HDL | −0.375 ± 0.11 | 19.8% |
| Multivariable model + total cholesterol | −0.372 ± 0.11 | 18.8% |
| Multivariable model + hsCRP | −0.370 ± 0.11 | 18.3% |
| Multivariable model + ICAM | −0.356 ± 0.11 | 13.7% |
| Multivariable model + LDL + HDL + total cholesterol | −0.364 ± 0.11 | 16.2% |
| Multivariable model + all mediating factors | −0.334 ± 0.11 | 6.6% |
LDL = low density lipoprotein, HDL = high density lipoprotein; hsCRP = high sensitivity C reactive protein, ICAM = intercellular adhesion molecule; HbA1c = hemoglobin A1c.
Blood samples were collected in a subset of the population (n = 28,034).
Adjusted for age, body mass index, alcohol (g/day), physical activity, smoking, postmenopausal status (yes or no), and family history of myocardial infarction in parent <60 years-old (yes or no), marital status, alternate healthy eating index score.
Discussion
In this large prospective study, we found that nut consumption was inversely associated with CV mortality in women. To our knowledge, this is the first study to examine the proportion of the nut-CV relation that is explained by potential mediating factors. We found that lipids accounted for 19%, whereas HbA1c 18%, and inflammatory markers, hsCRP and ICAM, accounted for 18% and 14% of the lowered risk of CV mortality with nut consumption. However, all mediating factors together accounted for about 6.6% of the lower risk of CV mortality. The mediating effect for all factors together may be lower than the individual factors if some factors have an interaction with each other. Although increased nut consumption was found to be beneficial, it is recommended that calories from the addition of nuts in the diet displace isocalorically less healthy/unhealthy foods and nutrients to avoid weight gain.
It is plausible that women who had higher nut consumption also exhibited other healthy behaviors such as greater fruit and vegetable and whole grain consumption. However, given the large standard deviation for fruit and vegetable and whole grain consumption in our data, we cannot definitively conclude this. We also noticed an increase in nut consumption during the second FFQ. The Women’s Health study trial began in 1991 with the first run-in phase between 1991 and 1995. The second nut consumption FFQ was obtained near the trial’s end in 2004. Dietary guidelines for Americans were released in 1990 with another set of guidelines in the early 2000s. During this time period of more than 10 years, it is possible that dietary patterns may have changed, which may in part account for the increased nut consumption observed during the second FFQ survey.
Our findings of an inverse association between nut consumption and CV mortality are consistent with prior studies. Other investigators have found that nut consumption was inversely associated with CV mortality. The Nurses’ Health Study, which included 76,464 women with 30 years of follow-up, demonstrated an inverse association between nut consumption and CV mortality, with a HR of 0.75 (0.62–0,84) for ≥5 times/week nut consumption as compared to the never nut consumption group.4 One study which evaluated three large cohorts (the Southern Community Cohort, which consists of African American and European American participants of low socioeconomic status in the southeastern United States, the Shanghai Women’s Health Study cohort and the Shanghai Men’s Health Study cohort) found that nut consumption was inversely associated with CV mortality for all these groups.13 Participants in the Southern Community cohort had a mean age of 51 years for African Americans and 54 years for European Americans, those in the Shanghai Men’s Health Study had a mean age of 55 years, and participants in the Shanghai Women’s Health Study had a mean age of 53 years. With a median follow-up of 5.4 years in the Southern Community Cohort, 6.5 years in the Shanghai Men’s Health Study, and 12.2 years in the Shanghai Women’s Health Study, HRs and 95% CIs were 0.62 (0.46–0.84) for Europeans and 0.77 (0.63–0.92) for Africans in the highest quintile of nut consumption as compared to those in the first quintile.13 However in the Shanghai Men’s and Women’s Health Study cohorts, nut consumption consisted of peanuts only, with a HR and 95% CI of 0.76 (0.67–0.85) for the highest quintile of nut consumption with reference to the first quintile in the fully adjusted model.13 The PREDIMED (PREvención con DIeta MEDiterránea) trial found an inverse association between a Mediterranean diet enriched with nuts and all-cause mortality and CV mortality: HRs (95% CIs) were 0.45 (0.25–0.81) for the >3 servings/week group and 0.42 (0.24–0.74) for 1–3 servings/week of nut consumption in reference to the “never” nut consumption category in the fully adjusted multivariable model.10 Similarly, in the Physicians’ Health Study, in which 20,742 men were followed for an average of 9.6 years, nut consumption was inversely associated with the risk of CV mortality, with multivariable-adjusted HRs (95% CIs) of 1.0 (reference), 0.98 (0.82, 1.17), 0.89 (0.72, 1.11), 0.80 (0.62, 1.03), and 0.74 (0.55, 1.02) for nut consumption of never or <1 serving/mo, 1–3 servings/mo, 1 serving/wk, 2–4 servings/wk, and ≥5 servings/wk, respectively (p trend = 0.015).5
Contrary to our results, some prior studies did not find a significant association of nut consumption with CV: in the Golestan cohort study, which included 49,112 participants ages 40 and older from northeastern Iran, the HR (95% CI) for CV death among subjects reporting ≥ 3 servings per week of nut consumption as compared to no nut consumption was 0.77 (0.58–1.01) after seven years of follow-up.14 The study population in this group was on average 10 years younger than in our study, and the follow-up was only seven years, which may not be enough time to detect CV mortality in a younger population.
There are several postulated biological mechanisms that support a link between nuts and heart disease. Nut consumption has been associated with a lower incidence of hyperlipidemia, type-2 diabetes, hypertension, obesity and inflammatory markers. Several trials have found beneficial effects of nut consumption on lipid profiles.25 In the Nurses’ Health Study and the Shanghai Women’s Health Study, nut consumption was inversely associated with risk of type-2 diabetes in multivariate adjusted models.12,15,16 Nut (tree nut and peanut) consumption is associated with lower oxidative stress, inflammatory markers, insulin resistance, central adiposity, and endothelial dysfunction.12 Davis et al. found that walnut consumption decreased the expression of endothelin-1 (a potent endothelial activator) in an animal model of atherosclerosis.26 It is thought that walnuts may have beneficial effects due to several components: l-arginine, which is a precursor of nitric oxide, alpha-linolenic acid and phenolic antioxidants.20 Trials have shown that nuts may lead to decreased production of proinflammatory cytokines such as IL-6, ICAM-1, c-reactive protein, TNF-alpha, and IL-18, and increased production of anti-inflammatory factors such as adiponectin.17,18,20 This in turn may lessen endothelial dysfunction and decrease the risk of type 2 diabetes and CV events.27 Also, nuts contain nutrients such as unsaturated fatty acids, proteins, fiber, vitamins (including folate, niacin, vitamin E), minerals (such as magnesium, calcium), and phytochemicals (flavonoids, carotenoids), which may have antioxidant and cardio-protective effects.28
In the present study, we found that lipids, inflammatory markers and markers of glucose metabolism accounted for a modest reduction of CV mortality with nut consumption among women. Few small clinical trials have examined the association of nut consumption and these biomarkers. A single group intervention study examined the effect of hazelnut consumption on cardiovascular biomarkers in 21 subjects. Each subject had three diet periods: a control diet for 4 weeks, a hazelnut rich diet, in which nuts comprised 18–20% of the daily intake (4 weeks), and another control period. In this study, the total cholesterol was reduced by 7.8%, LDL cholesterol by 6.2% and HDL cholesterol was increased by 6.1% compared to the control diet.29 Also, in this study, investigators found that nut consumption significantly improved flow-mediated dilation (a marker of endothelial dysfunction) by 56.5%.29 It is plausible that improvement in markers of endothelial function (which were not measured in our study) may explain a significant percentage of the nuts – CV relation.
The current study has limitations. Nut consumption was self-reported, which could have led to misclassification of the exposure. Since participants were not randomly assigned to groups of nut consumption and as with any observational study, residual confounding and confounding due to unmeasured variables cannot be excluded. We did not have data on the specific types of nuts (other than peanuts and other nuts), or the method of preparation (salted, spiced, roasted, sweetened etc) to include these factors in our analysis. Reverse causality is also possible because those with chronic disease might have lower nut consumption. However, we excluded those with heart disease at baseline and in sensitivity analysis, excluded those who died within the first two years of follow-up, which did not change the results. Given that our cohort includes female health professionals who have higher education and socioeconomic status than the general population, the results may not entirely generalize to males or the population at large who may have different dietary patterns. However, this could also minimize residual confounding due to socioeconomic status. Moreover, the metabolic effects of nut consumption are unlikely to be different across populations. Strengths of our study include a large sample size which allows for greater power, a prospective design, long follow-up duration, standardized ascertainment of coronary heart disease events and CV mortality and incorporation of detailed lifestyle and anthropometric variables into the analysis.
Conclusion
This study suggests that nut consumption is inversely associated with cardiovascular mortality in women. A modest proportion of this association may be accounted for by lipids, inflammatory markers and factors of glucose metabolism in adult women, assuming a causal relation between nut consumption and risk of CV death. If confirmed with other studies and populations, this information may lead to diets that maximally reduce the risk of premature death from cardiovascular disease.
Supplementary Material
Acknowledgments
We thank the participants and staff of the Women’s Health Study for their time and valuable contributions. The Women’s Health Study is supported by grants CA-047988, CA-182913, HL-43851, HL-080467, and HL-099355 from the National Institutes of Health, Bethesda, MD, United States.
Footnotes
Conflicts of interest
In the past, Luc Djousse has received an investigator-initiated grant from the California Walnut Commission, United States. All other authors have no declarations of interest.
Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jacl.2021.01.001.
Contributor Information
Tasnim F. Imran, Cardiology Section, Department of Medicine, Warren Alpert School of Medicine at Brown University; VA Boston Healthcare System and Division of Aging, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA.
Eunjung Kim, Division of Preventive Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA.
Julie E. Buring, Division of Preventive Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA; Department of Epidemiology, Harvard T. H Chan School of Public Health, Boston, MA, USA
I-Min Lee, Division of Preventive Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA; Department of Epidemiology, Harvard T. H Chan School of Public Health, Boston, MA, USA.
J. Michael Gaziano, VA Boston Healthcare System and Division of Aging, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA.
Luc Djousse, VA Boston Healthcare System and Division of Aging, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA.
References
- 1.Lloyd-Jones DM, Hong Y, Labarthe D, et al. American heart association strategic planning task force and statistics committee. American heart association strategic planning task force and statistics committee. Defining and setting national goals for cardiovascular health promotion and disease reduction: the American heart association’s strategic impact goal through 2020 and beyond. Circulation. 2010;121:586–613. [DOI] [PubMed] [Google Scholar]
- 2.Estruch R Ros E, Salas-Salvadó J, et al. Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med. 2018;378(25). [DOI] [PubMed] [Google Scholar]
- 3.U.S. Department of Health and Human Services and U.S. Department of Agriculture. 2015 – 2020 Dietary Guidelines for Americans. 8th Edition. 2015. Available at https://health.gov/our-work/food-nutrition/previous-dietary-guidelines/2015. [Google Scholar]
- 4.Bao Y, Han J, Hu FB, et al. Association of nut consumption with total and cause-specific mortality. N Engl J Med. 2013;369:2001–2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hshieh TT, Petrone AB, Gaziano JM, Djoussé L. Nut consumption and risk of mortality in the physicians’ health study Am J Clin Nutr. 2015;101(2):407–412. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bonnaccio M, Di CA, De Curtis A, et al. Nut consumption is inversely associated with both cancer and total mortality in a Mediterranean population: prospective results from the Moli-sani study. Br J Nutr. 2015;114(5):804–811. [DOI] [PubMed] [Google Scholar]
- 7.Ellsworth JL, Kushi LH, Folsom AR. Frequent nut intake and risk of death from coronary heart disease and all causes in postmenopausal women: the Iowa Women’s Health Study. Nutrition, metabolism, and cardiovascular diseases. Nutr Metabol Cardiovasc Dis. 2001;11:372–377. [PubMed] [Google Scholar]
- 8.Fraser GE, Sumbureru D, Pribis P, Neil RF, Frankson MA. Association among health habits, risk factors, and all-cause mortality in a black California population. Epidemiology. 1997:8:168–174. [DOI] [PubMed] [Google Scholar]
- 9.Albert CM, Gaziano JM, Willett WC, Manson JE. Nut consumption and decreased risk of sudden cardiac death in the Physicians’ Health Study. Arch Intern Med. 2002;162:1382–1387. [DOI] [PubMed] [Google Scholar]
- 10.Guasch-Ferre M, Bullo M, Martinez-Gonzalez MA, et al. Frequency of nut consumption and mortality risk in the PREDIMED nutrition intervention trial. BMC Med. 2013;11:164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sun Y, Jiang CQ, Cheng KK, et al. Nut consumption and cardiovascular risk in older Chinese: the guangzhou biobank cohort study. PLoS One. 2015;10:e0137178. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Jiang R, Manson JE, Stampfer MJ, Liu S, Willett WC, Hu FB. Nut and peanut butter consumption and risk of type 2 diabetes in women. J Am Med Assoc. 2002;288:2554–2560. [DOI] [PubMed] [Google Scholar]
- 13.Villegas R, Gao YT, Yang G, et al. Legume and soy food intake and the incidence of type 2 diabetes in the Shanghai Women Health Study. Am J Clin Nutr. 2008:87:162–167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Djoussé L, Rudich T, Gaziano JM. Nut consumption and risk of hypertension in US male physicians. Clin Nutr. 2009:28:10–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Estruch R Martínez-González MA, Corella D, et al. Effects of a Mediterranean-style diet on cardiovascular risk factors. A randomized trial. Ann Intern Med. 2006;145:1–11on behalf of the PREDIMED Study Investigators. [DOI] [PubMed] [Google Scholar]
- 16.Ros E, Núñez I, Pérez-Heras A, et al. A walnut diet improves endothelial function in hypercholesterolemic subjects. Circulation. 2004;109:1609–1614. [DOI] [PubMed] [Google Scholar]
- 17.Sabaté J, Oda K, Ros E. Nut consumption and blood lipid levels: a pooled analysis of 25 intervention trials. Arch Intern Med. 2010;170(9):821–827. [DOI] [PubMed] [Google Scholar]
- 18.Ros E Health benefits of nut consumption. Nutrients. 2010;2(7):652–682. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Shiroma EJ, Cook NR, Manson JE, Buring JE, Rimm ER, Lee IM. Comparison of self-reported and accelerometer-assessed physical activity in older women. PLoS One. 2015;10(12):e0145950. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lee IM, Cook NR, Manson JE, Buring JE, Hennekens CH. Beta-carotene supplementation and incidence of cancer and cardiovascular disease: the Women’s Health Study. J Natl Cancer Inst. 1999;91:2102–2106. [DOI] [PubMed] [Google Scholar]
- 21.Chiuve SE, Fung TT, Rimm EB, et al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142:1009–1018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Vanderweele TJ, Vansteelandt S. Odds ratios for mediation analysis for a dichotomous outcome. Am J Epimediol. 2010;172:1339–1348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Luu HN, Blot WJ, Xiang YB, et al. Prospective evaluation of the association of nut/peanut consumption with total and cause-specific mortality. JAMA Intern Med. 2015;175(5):755–766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Eslamparast T, Sharafkhah M, Poustchi H, et al. Nut consumption and total and cause-specific mortality: results from the Golestan Cohort Study. Int J Epidemiol. 2016;46:75–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sabaté J, Oda K, Ros E. Nut consumption and blood lipids: a pooled analysis of 25 intervention trials. Arch Intern Med. 2010;170:821–827. [DOI] [PubMed] [Google Scholar]
- 26.Davis P, Valacchi G, Pagnin E, et al. Walnuts reduce aortic ET-1 mRNA levels in hamsters fed a high-fat, atherogenic diet. J Nutr. 2006;136:428–432. [DOI] [PubMed] [Google Scholar]
- 27.Salas-Salvado J, Casas-Agustench P, Murphy MM, et al. The effect of nuts on inflammation. Asia Pac J Clin Nutr. 2008;17(Suppl 1):333–336. [PubMed] [Google Scholar]
- 28.Kris-Etherton PM, Hu FB. Ros E, Sabaté J. The role of tree nuts and peanuts in the prevention of coronary heart disease: multiple potential mechanisms. J Nutr. 2008:138:1746S–1751S. [DOI] [PubMed] [Google Scholar]
- 29.Orem A, Yucesan FB, Orem C, et al. Hazenut-enriched diet improves cardiovascular risk biomarkers beyond a lipid lowering effect in hypercholesterolemic subjects. J Clin Lipidol. 2013;7:123–131. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.