Frequency of nut consumption and mortality risk in the PREDIMED nutrition intervention trial

Nuts are an important component of the so-called Mediterranean diet (MedDiet) and a good source of unsaturated fatty acids, fiber, minerals (potassium, calcium and magnesium), vitamins (folate and tocopherols) and other bioactive compounds, such as phytosterols and polyphenols [1]–[4].

There is consistent evidence to suggest that the consumption of nuts has a beneficial effect on cardiovascular health, and this effect is attributable to their unique nutritional composition [5]. A pooled analysis of four large-scale observational studies showed that subjects in the highest nut consumption categories had an approximately 35% reduced risk of incident coronary heart disease (CHD) [6]. The frequency of nut consumption was also related to lower rates of sudden cardiac death in a large cohort of men [7]. Furthermore, epidemiologic studies and clinical trials have shown that frequent nut consumption is associated with a reduced load of cardiovascular disease risk factors, such as dyslipidemia, type 2 diabetes, and metabolic syndrome [4, 6, 8, 9]. In addition, reports from the Iowa Women’s Health study [10], a large Dutch cohort [11], and the US Nurses’ Health Study [12], which assessed populations with relatively low overall nut intake, suggested that frequent nut consumption related inversely to total mortality, albeit the protective effect was weak, with adjusted risk reductions ranging from 5% to 15% [10]–[12]. If an inverse association between nut consumption and all-cause mortality exists, the beneficial effect might be more robust in Mediterranean regions, where nut consumption per person is relatively high compared to other countries [13].

The main aim of the PREDIMED study was to test the efficacy of two Mediterranean diets (one supplemented with extra-virgin olive oil and another with nuts), as compared to a control diet (advice on a low-fat diet), on primary cardiovascular prevention. In contrast, in this current manuscript our aims were only to assess the association between baseline consumption of nuts (that is, the consumption of nuts previous to starting the intervention) and total mortality (instead of cardiovascular events). We have additionally included the repeated measurements for the consumption of nuts during follow-up as another exposure, regardless of the allocated arm of the trial; this is in contrast with the original PREDIMED study, which used an intention-to-treat analysis.

We hypothesized that level of nut consumption would be strongly associated with mortality in the cohort of the PREDIMED (‘PREvención con DIeta MEDiterránea’) study, including older men and women at high cardiovascular risk [14]. To this end, in this cohort we longitudinally examined the association between the frequency of nut consumption at baseline and the risk of mortality at the end of follow-up.

Changes in total nut consumption were +15.95 ± 21.10 g/day (mean ± SD) in the MedDiet supplemented with nuts, -0.80 ± 16.31 g/day in the MedDiet supplemented with extra virgin olive oil and −3.12 ± 13.85 g/day in the control group.

Figure 1 shows the multivariate adjusted HRs for total mortality by frequency of total nut consumption and intervention group. In the three arms of the trial, individuals who consumed nuts >3 servings per week tended to have a lower risk of mortality than those in the reference category. Subjects in the upper category of nut consumption at baseline allocated to the MedDiet with nuts intervention had a significant reduction in total mortality risk of 63% (95% CI −34% to −78%), while those allocated to the MedDiet with EVOO and the control diet had non-significant reductions of 34% (95% CI −64% to 10%) and 16% (95% CI −52% to 44%), respectively. The interaction between baseline total nut consumption and intervention group was significant, P = 0.019).

When we used generalized estimating equations to assess the association between yearly updated measurements of total nut consumption and all-cause mortality we also found a significant inverse association. The fully-adjusted relative risk (RR) was 0.68 (95% CI 0.50 to 0.93) with a significant linear trend test. When we repeated the analysis to evaluate the association between nut intake and cardiovascular mortality and cancer mortality the fully-adjusted relative risk (RR) were 0.76 (95% CI 0.42 to 1.36) and 0.63 (95% CI 0.39 to 1.03), respectively; however the linear trend tests were not significant (data not shown).

In this longitudinal cohort study of individuals at high cardiovascular risk with relatively high nut intake living in a Mediterranean country, the frequency of nut consumption was inversely related to total mortality after 4.8 years of follow-up. Compared to non-consumers, subjects who consumed >3 servings of nuts per week at baseline had a significant 39% lower risk of all-cause mortality. Relative risk reductions were similar for the upper baseline category of non-walnut nuts (34%) or walnut consumption (45%), and when we evaluated the repeated measurements of total nut consumption over time (32%).

Moreover, those in the upper category of baseline nut consumption had a 55% lower risk of cardiovascular mortality and 40% lower risk of cancer mortality compared to those who never consumed nuts. The study subjects participated in the PREDIMED study, a long-term, randomized nutrition intervention trial [14], and those consuming more nuts at baseline and allocated to intervention with a MedDiet supplemented with nuts showed a significantly reduced total mortality risk of 63%.

The inverse association between baseline nut consumption and total mortality was of borderline significance in participants in the upper category of baseline nut consumption in a MedDiet supplemented with EVOO group, while there was no significant association in those allocated to a low-fat control diet, who were advised to reduce intake of all fatty foods, including nuts. Thus, advice against eating nuts throughout the study might have counterbalanced the protective effect of a lifetime intake of these foods. We assume that the baseline assessment can be considered as a good correlate of lifetime habits in this population.

Our findings concur with those of epidemiological studies showing inverse associations between nut consumption and cardiovascular mortality. Four large prospective studies have reported consistent inverse associations between nut consumption and fatal CHD or sudden cardiac death [23]–[26]. In the Adventist Health Study, subjects who consumed nuts >5 times per week had a 48% reduced risk of fatal CHD [23]. The reduction in death from CHD among women who consumed nuts 2 to 4 times/week in the Iowa Women’s Health Study were 57% [24]. However, a later report from the same study with longer follow-up failed to confirm that nut consumption protected from CHD death [10]. In addition, the Nurses’ Health Study observed that women who consumed nuts ≥5 times/week had a 30% reduced risk of fatal CHD [25]. Finally, the Physicians’ Health Study reported a 47% lower risk of sudden cardiac death and 30% lower risk of total CHD death among men who consumed nuts twice a week or more [7].

In our study, a reduced risk of cancer mortality has been observed in individuals that frequently consumed total nuts and walnuts. Few epidemiologic studies have been conducted evaluating the association between nut consumption and cancer. An ecological study showed that prostate cancer mortality was inversely associated with nuts and oilseed consumption [27]. Moreover, findings from prospective studies suggest inverse associations between nut consumption and colorectal or endometrial cancer, especially in women [28]–[31]. Some studies showed inverse associations of nut intake and prostate cancer [32], however the relationship between nuts and cancer incidence and mortality is insufficient and further research is needed [33]. A possible explanation that may account for the inverse relationship between walnuts and cancer mortality but not with other nuts could be that walnuts were richer in free and total polyphenols than all the other nuts [34]. As walnuts are usually consumed raw, and roasting can cause a decline in the efficacy in the antioxidant capacity, it has been shown that raw walnuts, as consumed in the PREDIMED study, had the highest antioxidant efficacy among all the nuts [34]; this could play a beneficial role in the prevention of cancer.

The present results also support those of prior observational studies suggesting that nut consumption protects against mortality. In the Iowa Women’s Health Study, subjects consuming nuts ≥2 times/week had a significant 12% lower mortality risk than those who ate nuts less than once monthly after a 12-year follow-up [10]. A recent study from a large Dutch cohort followed for 10 years reported that men and women in the 75th percentile of nut intake had 8% and 5% lower risks of all-cause mortality, respectively, compared with subjects in the 25th percentile [11]. Additionally, data from the Nurses’ Health Study, where participants were followed-up for nearly 18 years, showed that consuming nuts ≥2 times/week was associated with a 14% reduced risk of all-cause mortality [12]. It is noteworthy that the protection against total mortality afforded by nut consumption in our study was ≥3 orders of magnitude higher than that observed in studies of non-Mediterranean populations. A likely reason is that PREDIMED participants had a rather high self-selected nut intake before entering the study. Thus, 32% of PREDIMED participants consumed nuts >3 times/week, compared with nearly 10% consuming nuts ≥2 times/week in both the Iowa Women’s Health Study [10] and the Nurses’ Health Study [12]. In the Dutch study, participants in the 75th percentile of nut consumption had rather low average daily intakes of 11.1 g for men and 6.2 g for women [11].

The healthy nutritional profile of nuts may account for the inverse association observed between nuts and mortality. Nuts are high in monounsaturated fatty acids, fiber, minerals, vitamins and many bioactive compounds; all these nutrients may partly explain the beneficial effects on health that nuts have been shown to exert [3, 4]. The frequency of nut consumption has been inversely related to several chronic prevalent conditions, such as diabetes, hyperlipidemia, hypertension, obesity, metabolic syndrome, cancer, and CHD, among others [5, 26]. These inverse associations can be influenced by various mechanisms: nuts improve the blood lipid profile [6] and appear to decrease insulin resistance [8], and there is also evidence suggesting that they can modulate inflammation [35], oxidative stress [36], and endothelial function [37]. As a large body of evidence supports the beneficial effects of frequent nut consumption on many health outcomes, it is plausible that nuts protect as well against all-cause mortality.

Our study has limitations. First, given its observational nature, it is not possible to firmly conclude that the inverse relationship between nut consumption and total mortality reflects cause and effect. Second, even though data were adjusted for all possible confounders, there is still the possibility of residual confounding. However, the enhanced protective effect against all-cause mortality observed in frequent nut consumers at baseline who continued eating nuts during follow-up because they were allocated to the nuts intervention arm supports a causal relationship between increasing dietary exposure to nuts and reduced mortality. However, as the study was conducted in an older Mediterranean population at high cardiovascular risk, the results cannot easily be extrapolated to the general population. Nevertheless, it is relevant to assess these associations in individuals at high cardiovascular risk because this population is the most frequently attended by primary care physicians and the segment of population that can obtain higher benefits with diet or lifestyle changes.

There are also strengths to our study, such as a large sample size, relatively long duration of follow-up, and objective and thorough ascertainment of mortality as outcome in this prospective observational assessment.

In summary, this study provides further evidence of the inverse relationship between the frequency of nut consumption and the risk of mortality in a Mediterranean population at high cardiovascular risk with relatively high nut intake.

Hospital Clinic, Institut d’Investigacions Biomediques August Pi i Sunyer, Barcelona, Spain: M Serra-Mir, A Pérez-Heras, C Viñas, R Casas, LS Romero, M Cofán, C Valls-Pedret, A Sala-Vila and M Doménech.

University of Navarra, Primary Care Centres, Pamplona, Spain: E Toledo, A Sánchez-Tainta, I Zazpe, M Marques, E Goñi, B Sanjulian, A Marti, P Buil-Cosiales, M Serrano-Martinez, J Diez-Espino, A Garcia-Arellano and FJ Basterra-Gortari.

University Rovira i Virgili, Reus, Spain: R Gonzalez, C Molina, F Marquez, N Babio, P Martinez, N Ibarrola-Jurado, R Balanza, A Díaz-López, M Juanola-Falgarona, M Sorlí, J Garcia Roselló, F Martin, R Tort, A Isach, B Costa, JJ Cabré and J Fernandez-Ballart.

Institut de Recerca Hospital del Mar, Barcelona, Spain: MI Covas, H Schröder, S Tello, R de la Torre, MA Muñoz and J Vila.

University Hospital of Alava, Vitoria, Spain: I Salaverría, S Castro, E Sanz, F Ricarte and J Rekondo.

University of Málaga, Málaga, Spain: R Benítez-Pont, M Bianchi-Alba, J Fernández-Crehuet and E Gómez-Gracia.

Department of Family Medicine, Primary Care Division of Sevilla, Sevilla, Spain: FJ García, M Ortega-Calvo, P Román, JM Santos and Y Corchado.

University of Las Palmas de Gran Canaria, Las Palmas, Spain: J Álvarez-Pérez, E Díez-Benítez, I Bautista-Castaño and A Sánchez-Villegas.

University of Valencia, Department of Preventive Medicine, Spain: C Ortega-Azorin, EM Asensio-Márquez, P Guillem-Saiz, JI Gonzalez and O Portoles.