Dietary fat intake and risk of cardiovascular disease and all-cause

mortality in a population at high risk of cardiovascular disease1

Marta Guasch-Ferré,2,3 Nancy Babio,2,3 Miguel A Martínez-González,3,4 Dolores Corella,3,5 Emilio Ros,3,6
Sandra Martín-Peláez,3,8 Ramon Estruch,3,7 Fernando Arós,3,9 Enrique Gómez-Gracia,3,10 Miquel Fiol,3,11
José M Santos-Lozano,3,12 Lluís Serra-Majem,3,13 Mònica Bulló,2,3 Estefanía Toledo,3,4 Rocío Barragán,3,5
Montserrat Fitó,3,8 Alfredo Gea,3,4 and Jordi Salas-Salvadó2,3* for the PREDIMED Study Investigators
2
Human Nutrition Unit, University Hospital of Sant Joan de Reus, Department of Biochemistry and Biotechnology, Faculty of Medicine and Health Sciences,

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

IISPV, Rovira i Virgili University, Reus, Spain; 3Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición (CIBEROBN),
Institute of Health Carlos III, Madrid, Spain; 4Department of Preventive Medicine and Public Health, University of Navarra IDISNA (Instituto de
Investigación Sanitaria de Navarra), Pamplona, Spain; 5Department of Preventive Medicine, University of Valencia, Valencia, Spain; 6Lipid Clinic,
Endocrinology and Nutrition Service and 7Department of Internal Medicine, August Pi i Sunyer Institute of Biomedical Research (IDIBAPS), Hospital
Clinic, University of Barcelona, Barcelona, Spain; 8Cardiovascular Risk and Nutrition (Regicor Study Group), Hospital del Mar Medical Research Institute
(IMIM), Barcelona, Spain; 9Department of Cardiology, University Hospital Araba, Vitoria, Spain; 10Department of Preventive Medicine, University of
Malaga, Malaga, Spain; 11Palma Institut of Health Research (IdISPa), Palma de Mallorca, Spain; 12Department of Family Medicine, Primary Care Division
of Seville, San Pablo Health Center, Seville, Spain; and 13Research Institute of Biomedical and Health Sciences, University of Las Palmas de Gran Canaria,
Las Palmas, Spain

ABSTRACT with MUFAs and PUFAs or of trans fat with MUFAs was inversely
Background: Dietary fat quality and fat replacement are more associated with CVD. This trial was registered at www.controlled-
important for cardiovascular disease (CVD) prevention than is total trials.com as ISRCTN 35739639. Am J Clin Nutr 2015;102:1563–73.
dietary fat intake.
Objective: The aim was to evaluate the association between total
fat intake and fat subtypes with the risk of CVD (myocardial infarc- 1
Supported in part by the Spanish Ministry of Health [Instituto de Salud
tion, stroke, or death from cardiovascular causes) and cardiovascular Carlos III (ISCIII)], PI1001407, Thematic Network G03/140, RD06/0045,
and all-cause death. We also examined the hypothetical effect of the FEDER (FondoEuropeo de Desarrollo Regional), and the Centre Català de la
isocaloric substitution of one macronutrient for another. Nutrició de l’Institut d’Estudis Catalans. Supported by the official funding
Design: We prospectively studied 7038 participants at high CVD risk agency for biomedical research of the Spanish government, Instituto de
Salud Carlos III (ISCIII), through grants provided to research networks
from the PREvención con DIeta MEDiterránea (PREDIMED)
specifically developed for the trial (RTIC G03/140, to RE; RTIC RD 06/0045,
study. The trial was conducted from 2003 to 2010, but the present
to MAM-G; and through Centro de Investigación Biomédica en Red de
analysis was based on an expanded follow-up until 2012. At Fisiopatología de la Obesidad y Nutrición (CIBEROBN); and by grants
baseline and yearly thereafter, total and specific fat subtypes from Centro Nacional de Investigaciones Cardiovasculares (CNIC 06/
were repeatedly measured by using validated food-frequency 2007), Fondo de Investigación Sanitaria–Fondo Europeo de Desarrollo Re-
questionnaires. Time-dependent Cox proportional hazards models gional (PI04-2239, PI 05/2584, CP06/00100, PI07/0240, PI07/1138, PI07/
were used. 0954, PI 07/0473, PI10/01407, PI10/02658, PI11/01647, and P11/02505;
Results: After 6 y of follow-up, we documented 336 CVD cases and PI13/00462), Ministerio de Ciencia e Innovación (AGL-2009-13906-C02
414 total deaths. HRs (95% CIs) for CVD for those in the highest and AGL2010-22319-C03), Fundación Mapfre 2010, Consejería de Salud
de la Junta de Andalucía (PI0105/2007), the Public Health Division of
quintile of total fat, monounsaturated fatty acid (MUFA), and poly-
the Department of Health of the Autonomous Government of Catalonia,
unsaturated fatty acid (PUFA) intake compared with those in the Generalitat Valenciana (ACOMP06109, GVA-COMP2010-181, GVA-
lowest quintile were 0.58 (0.39, 0.86), 0.50 (0.31, 0.81), and 0.68 COMP2011-151, CS2010-AP-111, and CS2011-AP-042), and the Regional
(0.48, 0.96), respectively. In the comparison between extreme Government of Navarra (P27/2011). The Fundación Patrimonio Comunal
quintiles, higher saturated fatty acid (SFA) and trans-fat intakes were Olivarero and Hojiblanca SA (Málaga, Spain), the California Walnut
associated with 81% (HR: 1.81; 95% CI: 1.05, 3.13) and 67% (HR: Commission (Sacramento, California), Borges SA (Reus, Spain), and
1.67; 95% CI: 1.09, 2.57) higher risk of CVD. Inverse associations with Morella Nuts SA (Reus, Spain) donated the olive oil, walnuts, almonds,
all-cause death were also observed for PUFA and MUFA intakes. Iso- and hazelnuts, respectively, used in the study. CIBEROBN (Centro de
Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nu-
caloric replacements of SFAs with MUFAs and PUFAs or trans fat with
trición) and CIBERDEM (Centro de Investigación Biomédica en Red de
MUFAs were associated with a lower risk of CVD. SFAs from pastries Diabetes y Enfermedades Metabólicas Asociadas) are initiatives of
and processed foods were associated with a higher risk of CVD. ISCIII, Spain.
Conclusions: Intakes of MUFAs and PUFAs were associated with *To whom correspondence should be addressed. E-mail: jordi.salas@urv.cat.
a lower risk of CVD and death, whereas SFA and trans-fat intakes Received May 28, 2015. Accepted for publication October 5, 2015.
were associated with a higher risk of CVD. The replacement of SFAs First published online November 11, 2015; doi: 10.3945/ajcn.115.116046.

Am J Clin Nutr 2015;102:1563–73. Printed in USA. Ó 2015 American Society for Nutrition 1563
1564 GUASCH-FERRÉ ET AL.

Keywords: dietary fat, fat subtypes, saturated fat, cardiovascular fatty acids and the risk of CVD, cardiovascular death, and
disease, all-cause death, PREDIMED study all-cause death in a population at high CVD risk from the
PREDIMED study. We also estimated the associations of hy-
pothetic isocaloric substitutions of several dietary components
INTRODUCTION and fat subtypes on the risk of CVD and death. Because more
The reduction in the intake of dietary fat has traditionally been importance has been placed on the differences in the biological
recommended by health institutions (1). However, dietary fat effects of fatty acids derived from different food sources (10),
quality has been recognized as being even more important than the association of SFA intake from different food sources and
the total amount of fat for the prevention of cardiovascular the risk of CVD and death was also explored.
disease (CVD)14 and death (2, 3). In general, trans-fat intake
from partially hydrogenated vegetable oils is associated with
METHODS
adverse effects on health, and its consumption has been rec-
ommended to be reduced to a minimum (4). In addition, CVD Study population
risk can also be reduced by decreasing the intake of SFAs and
replacing them with a combination of PUFAs and MUFAs (2). We used data from the PREDIMED trial as an observa-
The findings from the PREvención con DIeta MEDiterránea tional prospective cohort study. The design and protocol of the

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

(PREDIMED) study, a randomized primary-prevention nu- PREDIMED study (http://www.predimed.es) have been detailed
trition trial, in individuals at high CVD risk (5), showed that elsewhere (5, 11). The PREDIMED study was a large, multi-
Mediterranean diets (MedDiets), which were high in MUFAs center, parallel-group randomized trial for the primary prevention
and PUFAs [because they were supplemented with extra- of CVD. The participants were recruited from October 2003 until
virgin olive oil (EVOO) or nuts] and low in SFAs and trans June 2009. Although the trial was completed in December 2010,
fat, were effective for the prevention of clinical events of CVD the endpoints for the present analysis were based on an extended
compared with a low-fat control diet (5). follow-up until June 2012 with the use of the same methods as
Because of the controversy in the findings of multiple studies, those used during the trial to obtain updated information on CVD
the research in the field has questioned if there is really a positive clinical events and death. Participants were men (aged 55–80 y)
association between SFAs and CVD, as traditionally speculated and women (aged 60–80 y) who were free of CVD at baseline
(6). Moreover, the results on the relation between the intake of but who were at high CVD risk because they had either type 2
other fat subtypes, CVD, and death are still inconsistent (4, 6–8). diabetes or at least 3 of the following CVD risk factors: current
Previous studies were conducted in apparently healthy in- smoking, hypertension, hypercholesterolemia, low HDL choles-
dividuals, and data on the role of fat subtypes and the risk of terol, overweight/obesity, or family history of premature coronary
CVD in high-risk patients are limited. For instance, although heart disease. Exclusion criteria were the presence of any severe
MUFA intake is reported to be beneficial for preventing several chronic illness, alcohol or drug abuse, a BMI (in kg/m2) $40, and
CVD risk factors (e.g., high HDL cholesterol, low blood pres- allergy or intolerance to olive oil or nuts (5). Participants were
sure, and improved inflammatory status), there is not enough randomly assigned to receive: a MedDiet supplemented with
available evidence supporting a reduction in the risk of clinical EVOO oil (MedDiet+EVOO), a MedDiet supplemented with
events of CVD or death by MUFA consumption (3, 4). mixed nuts (MedDiet+nuts), or advice on a low-fat diet (control
Recently, researchers have focused not only on total SFA group). Several companies donated olive oil and nuts to the
intake but also on the subtypes of SFAs, their food sources, what participants of both MedDiet interventions during the study.
dietary patterns these foods are part of, and by which foods they The trial was registered at http://www.controlled-trials.com
are replaced. In fact, food sources contain different amounts and (ISRCTN35739639). All participants provided written informed
proportions of SFAs, such as myristic, lauric, and stearic acids, consent according to a protocol approved by the institutional
which can have differential effects on blood lipids and in- review boards.
flammation, and therefore on CVD risk (9). Although the as-
sociations between SFA consumption from different food sources Dietary and other covariate assessment
and the incidence of CVD were investigated previously in the
Multi-Ethnic Study of Atherosclerosis (9), these associations At baseline and yearly during the follow-up, trained dietitians
have not been investigated in a population at high CVD risk with completed a semiquantitative food-frequency questionnaire (FFQ)
the use of repeated measurements of intake. Furthermore, the in a face-to-face interview with the participants. This questionnaire
relation of these different sources of SFAs with all-cause mor- has been validated in a population at high CVD risk from Spain;
tality has never been reported. however, the questionnaire was not compared with objective mea-
In the present prospective study, we hypothesized that MUFAs surements (12). Reproducibility and validity of the FFQ for total
and PUFAs are associated with a lower risk of CVD and death, dietary fat, SFAs, MUFAs, and PUFAs estimated by the Pearson
whereas SFAs and trans fat are associated with a higher risk. We correlation (r) were 0.61, 0.67, 0.59, and 0.63, respectively. The
examined the association between intakes of the major types of intraclass correlation coefficients for reproducibility and validity
were 0.75 and 0.63 for total fat, 0.81 and 0.68 for SFAs, 0.74 and
0.67 for MUFAs, and 0.77 and 0.60 for PUFAs. The FFQ
14
Abbreviations used: CVD, cardiovascular disease; EVOO, extra-virgin included 137 food items, and frequencies of consumption of
olive oil; FFQ, food-frequency questionnaire; MedDiet, Mediterranean diet; food items were reported on an incremental scale with 9
PREDIMED, PREvención con DIeta MEDiterránea; RCT, randomized levels (never or almost never; 1–3 times/mo; 1, 2–4, and 5–6
controlled trial. times/wk; and 1, 2–3, 4–6, and .6 times/d). We used Spanish
 DIETARY FAT INTAKE, CVD, AND ALL-CAUSE DEATH 1565
food-composition tables to estimate energy and nutrient in- subtypes of fat intake (classified in quintiles for the percentage of
take (13). At baseline and yearly during the follow-up, a ques- energy), CVD, and death during follow-up. HRs and their 95% CIs
tionnaire on lifestyle variables, educational achievement, history of were calculated by using the lowest quintile as the reference
illnesses, and medication use was administered. Physical activity category.
was assessed by using the validated Spanish version of the Multivariable model 1 for the total fat analysis was stratified by
Minnesota Leisure-Time Physical Activity questionnaire (14). recruitment center and adjusted for age, sex, intervention group,
Participants were considered to be diabetic, hypercholesterol- updated total energy (kcal/d), alcohol intake (continuous, adding
emic, or hypertensive if they had previously been diagnosed as a quadratic term), updated quintiles of protein, and dietary
such and/or if they were being treated with antidiabetic, cho- cholesterol. Model 1 for specific subtypes of fat also included as
lesterol-lowering, or antihypertensive agents, respectively. covariates the other subtypes of fat. Model 2 was further adjusted
Anthropometric and blood pressure measurements were taken for nondietary confounders and the following classical CVD risk
by trained personnel. We used calibrated scales and a wall- factors: BMI, smoking status (never, former, or current smoker),
mounted stadiometer to measure weight and height, respectively, educational level (primary education, secondary education, or
with participants wearing light clothing and no shoes; waist academic/graduate), leisure-time physical activity (metabolic
circumference was measured midway between the lowest rib equivalent task minutes/d), baseline diabetes (yes or no), hy-
and the iliac crest by using an anthropometric tape; and we used pertension (yes or no), hypercholesterolemia (yes or no), family

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

a validated oscillometer (Omron HEM705CP) to measure blood history of coronary heart disease (yes or no), use of antihyper-
pressure, in triplicate, with a 5-min interval between each tensive medication (yes or no), use of oral antidiabetic agents (yes
measurement, and we recorded the mean of these 3 values. or no), and use of lipid-lowering drugs (yes or no). The estimated
HRs can be interpreted as the estimated differences in risk of
a lower energy intake from carbohydrates and a concomitant higher
Ascertainment of CVD and death energy intake from total fat or subtypes of dietary fat.
For the present analysis, we used the following 3 different To test for linear trend across successive categories, we
endpoints: 1) the primary endpoint of the PREDIMED trial, replaced quintiles by the median intake for each quintile and
which is a composite of cardiovascular events (myocardial in- modeled these values as a continuous variable. HRs and 95% CIs
farction, stroke, or death from cardiovascular causes); 2) cardio- for CVD and all-cause death for a 5% energy increment for total
vascular death; and 3) all-cause death. The endpoint adjudication and subtypes of dietary fat were also calculated.
committee, whose members were blinded to treatment alloca- We also estimated the associations of isocaloric substitutions
tion, updated information on these endpoints once per year. The of carbohydrates, MUFAs, and PUFAs for SFAs and trans fat by
committee used different sources of information, as follows: including both continuous variables in the same multivariable
1) yearly questionnaires and examinations for all participants, model, which also contained the same covariates as in the pre-
2) family physicians, 3) comprehensive yearly review of medical vious models. We used the difference between regression co-
records of all participants, and 4) yearly consultation of the efficients and their variances and covariances to derive the HRs
National Death Index. Medical records of deceased participants and 95% CIs of the substitution analyses. We conducted these
were requested, and the endpoint adjudication committee de- analyses both for all-cause death and CVD as outcomes.
termined the cause of death and confirmed major events. This In addition, we evaluated the association between SFA intake
allowed us to assess mortality, regardless of attrition, and therefore from different food sources (vegetable SFAs, dairy SFAs, SFAs
mortality results were not affected by the drop-out rates. from pastries, meat and processed-meat SFAs, and fish SFAs) and
CVD and all-cause death using the same statistical models as in
previous analysis. The estimated HRs of these models can be
Statistical analysis interpreted as the estimated differences in risk of a lower energy
Participants with total energy intakes less or more than intake from a specific SFA food source and a concomitant higher
predefined limits (800 and 4000 kcal/d for men and 500 and energy intake from the rest of the SFA food sources, because SFA
3500 kcal/d for women) and those with missing information on intake was not included as a covariate in the model.
the FFQ were excluded from the present analysis. To take ad- Statistical tests were 2-sided, and P , 0.05 was considered to
vantage of the repeated measurements of the diet, we used yearly indicate significance. Analyses were performed by using Stata
updated measures of total dietary fat and subtypes of fat intake 12.1 (StataCorp).
using data from baseline to the last FFQ before the onset of
disease or death. Participants were categorized into 5 groups
according to quintiles of the percentage of energy obtained from RESULTS
total fat and each type of fat (MUFAs, PUFAs, SFAs, and trans Among the 7038 participants who were followed for a median
fat). Baseline characteristics are presented according to extreme of 6 y, we documented 336 cardiovascular events (myocardial
quintiles of total fat and subtypes of fat as means (6SDs) for infarction, stroke, or cardiovascular death), 102 cardiovascular
quantitative traits and n (%) for categorical variables. deaths, and 414 total deaths. Individuals were grouped into
Follow-up time was calculated as the interval between the date quintiles of yearly updated measures of total fat and specific types
of cardiovascular event, death, or end of follow-up (the date of the of fat intake (Table 1). The mean age of participants at baseline
last visit or the last recorded clinical event of participants while was 67 y. The mean intake of total fat (percentage of energy) in
still alive), whichever came first, and the date of random assign- the lowest quintile was 29.7% compared with 48.7% in the top
ment. Time-dependent Cox regression models were used to assess quintile of total fat intake. Those participants with a high intake
the associations between yearly updated measures of total fat and of total fat, SFAs, and trans fat were less physically active and
TABLE 1
Baseline characteristics according to quintile of updated measurements of specific types of fat and total fat intake1
1566

Total fat MUFAs PUFAs SFAs trans Fat

Total
population Q1 Q5 Q1 Q5 Q1 Q5 Q1 Q5 Q1 Q5

Participants, n 7038 1408 1407 1408 1407 1408 1407 1408 1407 1408 1407
Age, y 67 6 62 67 6 6 67 6 6 67 6 6 67 6 6 67 6 6 67 6 6 67 6 6 67 6 6 67 6 6 67 6 6*
Women, n (%) 4045 (57.5) 765 (54.3) 859 (61.0)* 792 (56.2) 872 (61.9)* 829 (58.9) 787 (55.9) 800 (56.8) 832 (59.1) 892 (63.3) 828 (58.8)*
BMI, kg/m2 29.9 6 3.8 29.7 6 3.7 30.1 6 4.0 29.9 6 3.7 30.2 6 3.9 30.1 6 3.7 29.7 6 3.8 29.6 6 3.6 30.1 6 3.8 29.6 6 3.7 30.2 6 3.9
Smoking habit, n (%)
Never 4324 (61.4) 859 (61.0) 887 (63.0) 856 (60.8) 889 (63.1) 881 (62.5) 861 (61.1) 875 (62.1) 850 (60.4) 940 (66.7) 846 (60.1)*
Former 1734 (24.4) 342 (24.3) 327 (23.2) 364 (25.8) 326 (23.1) 320 (22.7) 356 (25.3) 335 (23.8) 341 (24.4) 303 (21.5) 357 (25.3)*
Current 980 (13.9) 207 (14.7) 193 (13.7) 188 (13.5) 192 (13.6) 207 (14.7) 190 (13.5) 198 (14.0) 216 (15.3) 165 (11.7) 204 (14.5)*
Intervention group, n (%)
MedDiet + EVOO 2446 (34.7) 497 (35.3) 452 (32.1)* 477 (33.8) 496 (35.2)* 528 (37.5) 455 (32.3)* 496 (35.2) 486 (34.5) 494 (35.1) 488 (34.6)
MedDiet + nuts 2304 (32.7) 409 (29.0) 452 (32.1)* 419 (29.7) 455 (32.3)* 375 (26.6) 513 (34.4)* 442 (31.4) 456 (32.4) 456 (32.4) 470 (33.4)
Control group 2288 (32.5) 502 (35.6) 460 (32.7) 512 (36.3) 456 (32.4) 505 (35.8) 438 (31.2) 470 (33.4) 465 (33.0) 458 (32.5) 449 (31.9)
Education, n (%)
Primary 5466 (77.6) 1140 (80.1) 1062 (75.4)* 1144 (81.2) 1059 (75.2)* 1107 (78.6) 1122 (79.7) 1119 (79.4) 1072 (76.1) 1140 (80.9) 1079 (76.7)
Secondary 1314 (18.6) 215 (15.2) 292 (20.7)* 219 (15.5) 302 (21.4)* 246 (17.4) 243 (17.2) 236 (16.7) 278 (19.7) 221 (15.7) 277 (16.7)
University/graduate 258 (3.6) 53 (3.7) 53 (3.7) 45 (3.2) 46 (3.3) 55 (3.9) 42 (2.9) 53 (3.7) 57 (4.0) 47 (3.3) 51 (3.6)
Physical activity, MET-min/d 233 6 239 249 6 262 214 6 226* 249 6 267 213 6 204* 220 6 226 246 6 251* 247 6 259 225 6 244* 242 6 237 220 6 246*
Family history of CHD, n (%) 1731 (24.6) 355 (23.8) 329 (23.4)* 318 (22.6) 323 (22.9) 380 (26.9) 325 (23.1) 355 (25.2) 339 (24.0) 367 (26.0) 361 (25.6)
Diabetes, n (%) 3447 (48.9) 623 (44.2) 811 (57.6)* 745 (47.1) 787 (55.9)* 627 (44.5) 733 (52.1)* 581 (41.2) 806 (57.2)* 620 (44.0) 757 (53.8)*
Hypertension, n (%) 5815 (82.6) 1192 (84.6) 1128 (80.1)* 1189 (84.4) 1144 (81.3) 1196 (84.9) 1164 (82.7) 1185 (84.1) 1112 (79.0)* 1167 (82.8) 1139 (80.9)
Hypercholesterolemia, n (%) 5076 (72.2) 1101 (78.2) 935 (66.4)* 1067 (75.8) 938 (66.6)* 1020 (72.4) 1001 (71.4) 1122 (79.7) 930 (66.1)* 1079 (76.6) 943 (67.0)*
Oral antidiabetic 2269 (32.2) 405 (28.7) 559 (39.7)* 432 (30.7) 544 (38.6)* 413 (29.3) 486 (34.5) 372 (26.4) 554 (39.3)* 407 (28.9) 510 (36.2)*
agents, n (%)
Antihypertensive 5116 (72.7) 1030 (73.1) 1026 (72.9) 1032 (73.3) 1040 (73.9) 1045 (74.2) 1027 (72.9) 1047 (74.3) 970 (68.9)* 1034 (73.4) 1001 (71.1)
GUASCH-FERRÉ ET AL.

agents, n (%)
Lipid-lowering 3395 (48.2) 736 (52.2) 638 (45.3)* 721 (51.2) 641 (45.5)* 672 (47.7) 687 (48.8) 768 (54.5) 594 (42.2)* 748 (53.1) 626 (44.5)*
medication, n (%)
Energy and nutrient intake
Total energy intake, kcal/d 2238 6 544 2254 6 567 2161 6 507* 2282 6 557 2109 6 472* 2175 6 564 2314 6 543* 2219 6 563 2264 6 552* 2128 6 526* 2259 6 535*
Cholesterol, mg/d 361.3 6 123.3 327.1 6 110.4 371.1 6 131.7* 340.8 6 116.8 350.3 6 123.0* 340.9 6 117.6 365.5 6 125.5* 298.9 6 95.9 417.7 6 135.5* 300.3 6 98.8 408.3 6 139.5*
Carbohydrates, % 41.6 6 7.1 50.2 6 5.4 33.7 6 4.2* 49.4 6 5.8 34.6 6 4.7* 47.1 6 6.7 38.8 6 6.6* 48.1 6 6.5 36.5 6 5.8* 44.4 6 7.4 39.5 6 6.6*
of total energy
Protein, % of total energy 16.6 6 2.8 16.9 6 2.9 16.0 6 2.6* 17.0 6 2.9 15.7 6 2.4* 17.1 6 3.1 16.1 6 2.6* 16.3 6 2.9 16.9 6 2.8* 16.4 6 2.9 16.6 6 2.7*
Total fat, % of total energy 39.3 6 6.8 29.7 6 3.0 48.7 6 3.3* 30.8 6 4.4 47.9 6 3.9* 32.9 6 5.2 42.9 6 6.2* 32.4 6 5.4 44.6 6 5.5* 36.9 6 7.3 41.7 6 6.2*
MUFAs, % of total energy 19.6 6 4.5 14.0 6 2.3 25.4 6 3.2* 13.4 6 1.8 26.1 6 2.4* 16.4 6 3.3 19.7 6 4.8* 16.0 6 3.8 21.9 6 4.3* 18.9 6 4.9 20.1 6 4.4*
PUFAs, % of total energy 6.2 6 2.0 4.8 6 1.5 7.5 6 2.2* 5.8 6 2.5 6.8 6 1.7* 4.0 6 0.47 9.4 6 1.8* 5.8 6 2.0 6.4 6 1.9* 6.0 6 2.0 6.3 6 2.0*
SFAs, % of total energy 9.9 6 2.2 7.8 6 1.6 11.8 6 2.0* 8.3 6 2.0 11.3 6 2.0* 9.2 6 2.3 10.1 6 2.0* 7.0 6 0.8 13.2 6 1.4* 8.1 6 1.7 12.1 6 2.1*
trans Fat, % of total energy 0.22 6 0.14 0.17 6 0.11 0.27 6 0.16* 0.20 6 0.13 0.24 6 0.15* 0.21 6 0.15 0.23 6 0.15* 0.12 6 0.07 0.37 6 0.16* 0.07 6 0.02 0.46 6 0.12*
Dietary fiber, g/d 25.1 6 8.6 28.9 6 10.1 21.4 6 6.3* 29.2 6 10.10 21.4 6 6.39* 25.7 6 9.2 26.1 6 8.9* 29.4 6 10.7 21.9 6 7.1* 27.0 6 9.7 23.3 6 7.6*
Alcohol, g/d 8.4 6 14.3 10.8 6 18.3 5.1 6 9.0* 9.5 6 16.8 5.6 6 9.9* 9.5 6 16.8 7.4 6 12.4* 10.7 6 18.3 6.5 6 10.6* 7.4 6 13.8 6.9 6 11.0*
1
All quintiles were included in the analyses. *P value ,0.05 for comparisons between quintiles of dietary fat subtypes (Pearson’s chi-square test for categorical variables or 1-factor ANOVA for continuous
variables), as appropriate. CHD, coronary heart disease; EVOO, extra-virgin olive oil; MedDiet, Mediterranean Diet; MET-min, metabolic equivalent task minutes; Q, quintile.
2
Mean 6 SD (all such values).

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

 DIETARY FAT INTAKE, CVD, AND ALL-CAUSE DEATH 1567
TABLE 2
HRs (95% CIs) of cardiovascular disease and all-cause death according to quintile of updated measurements of total dietary fat and specific subtypes of
dietary fat intake1
Quintile

1 (lowest) 2 3 4 5 (highest) P-trend

Cardiovascular disease
Total fat
Cases, n (%) 1408 (4.2) 1408 (3.7) 1407 (4.3) 1408 (4.1) 1407 (3.3)
Median, % of energy 31.3 36.7 40.1 43.5 48.2
Multivariable model 1 1 (Ref) 0.95 (0.68, 1.34) 0.89 (0.62, 1.26) 0.81 (0.55, 1.17) 0.65 (0.44, 0.98) 0.02
Multivariable model 2 1 (Ref) 0.94 (0.67, 1.32) 0.86 (0.60, 1.22) 0.75 (0.52, 1.08) 0.58 (0.39, 0.86) ,0.01
MUFAs
Cases, n (%) 1408 (4.5) 1408 (3.9) 1408 (3.3) 1408 (4.4) 1407 (3.4)
Median, % of energy 14.9 18.3 20.7 23.0 26.1
Multivariable model 1 1 (Ref) 0.87 (0.61, 1.25) 0.63 (0.43, 0.94) 0.74 (0.50, 1.12) 0.52 (0.33, 0.84) ,0.01
Multivariable model 3 1 (Ref) 0.88 (0.61, 1.26) 0.66 (0.44, 0.98) 0.76 (0.50, 1.16) 0.50 (0.31, 0.81) ,0.01

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

PUFAs
Cases, n (%) 1408 (4.8) 1408 (3.8) 1407 (3.7) 1408 (3.7) 1407 (3.5)
Median, % of energy 4.2 5.3 6.1 7.2 9.0
Multivariable model 1 1 (Ref) 0.78 (0.56, 1.08) 0.71 (0.51, 1.01) 0.73 (0.51, 1.04) 0.70 (0.50, 0.99) 0.08
Multivariable model 3 1 (Ref) 0.78 (0.56, 1.09) 0.72 (0.51, 1.02) 0.71 (0.51, 1.01) 0.68 (0.48, 0.96) 0.04
SFAs
Cases, n (%) 1408 (2.9) 1408 (3.5) 1407 (4.2) 1408 (3.9) 1407 (4.9)
Median, % of energy 6.9 8.4 9.4 10.5 12.2
Multivariable model 1 1 (Ref) 1.46 (0.95, 2.24) 1.56 (0.98, 2.48) 1.49 (0.90, 2.47) 1.96 (0.90, 2.47) 0.03
Multivariable model 3 1 (Ref) 1.44 (0.94, 2.21) 1.48 (0.93, 2.35) 1.39 (0.84, 2.30) 1.81 (1.05, 3.13) 0.07
trans Fat
Cases, n (%) 1408 (2.9) 1408 (2.9) 1408 (4.1) 1408 (4.5) 1407 (5.2)
Median, % of energy 0.05 0.10 0.16 0.23 0.37
Multivariable model 1 1 (Ref) 1.05 (0.70, 1.58) 1.43 (0.96, 2.14) 1.55 (1.05, 2.30) 1.76 (1.15, 2.70) ,0.01
Multivariable model 3 1 (Ref) 1.03 (0.68, 1.55) 1.41 (0.94, 2.10) 1.48 (0.99, 2.21) 1.67 (1.09, 2.57) ,0.01
All-cause death
Total fat
Cases, n (%) 1408 (6.3) 1408 (6.9) 1407 (5.3) 1408 (6.0) 1407 (6.0)
Median, % of energy 31.3 36.7 40.2 43.5 48.2
Multivariable model 1 1 (Ref) 1.04 (0.77, 1.40) 0.73 (0.53, 1.01) 0.80 (0.58, 0.82) 0.57 (0.40, 0.82) ,0.01
Multivariable model 2 1 (Ref) 1.02 (0.75, 1.37) 0.71 (0.51, 0.98) 0.75 (0.54, 1.05) 0.53 (0.37, 0.76) ,0.01
MUFAs
Cases, n (%) 1408 (6.3) 1408 (6.1) 1407 (5.9) 1408 (5.4) 1407 (5.4)
Median, % of energy 14.7 17.9 20.5 22.8 26.0
Multivariable model 1 1 (Ref) 0.85 (0.62, 1.16) 0.78 (0.56, 1.08) 0.70 (0.49, 1.00) 0.64 (0.43, 0.95) 0.02
Multivariable model 2 1 (Ref) 0.86 (0.63, 1.17) 0.78 (0.56, 1.09) 0.69 (0.48, 0.99) 0.64 (0.43, 0.94) 0.01
PUFAs
Cases, n (%) 1408 (8.1) 1408 (5.9) 1407 (5.9) 1408 (4.9) 1407 (4.4)
Median, % of energy 4.2 5.3 6.2 7.2 9.0
Multivariable model 1 1 (Ref) 0.72 (0.53, 0.97) 0.71 (0.52, 0.97) 0.56 (0.40, 0.78) 0.50 (0.35, 0.71) ,0.01
Multivariable model 2 1 (Ref) 0.73 (0.54, 0.99) 0.72 (0.53, 0.99) 0.56 (0.39, 0.80) 0.50 (0.35, 0.73) ,0.01
SFAs
Cases, n (%) 1408 (5.0) 1408 (5.8) 1407 (5.7) 1408 (6.2) 1407 (6.6)
Median, % of energy 6.9 8.3 9.4 10.5 12.1
Multivariable model 1 1 (Ref) 1.20 (0.86, 1.68) 1.17 (0.82, 1.67) 1.22 (0.84, 1.77) 1.21 (0.81, 1.80) 0.47
Multivariable model 2 1 (Ref) 1.16 (0.83, 1.62) 1.12 (0.78, 1.61) 1.12 (0.78, 1.60) 1.08 (0.74, 1.58) 0.90
trans Fat
Cases, n (%) 1408 (4.8) 1408 (5.9) 1407 (4.7) 1408 (6.4) 1407 (7.5)
Median, % of energy 0.05 0.10 0.16 0.23 0.36
Multivariable model 1 1 (Ref) 1.15 (0.83, 1.60) 0.90 (0.64, 1.29) 1.22 (0.87, 1.72) 1.38 (1.00, 1.94) 0.03
Multivariable model 2 1 (Ref) 1.11 (0.80, 1.54) 0.86 (0.59, 1.24) 1.13 (0.78, 1.64) 1.29 (0.87, 1.90) 0.12
1
Time-dependent Cox regression models were used to assess the risk of cardiovascular disease and all-cause death by quintile of updated measurements
of total dietary fat and dietary fat subtype intake. Multivariable model 1 was adjusted for age (y), sex, intervention group, yearly updated total energy intake
(kcal/d), alcohol intake (continuous, adding a quadratic term), updated quintiles of fiber, protein intake, and dietary cholesterol for the total fat analysis. Model
1 for specific subtypes of fat also included as covariates the other subtypes of fat. Model 2 was further adjusted for nondietary variables and classical
cardiovascular disease risk factors: BMI (kg/m2), smoking status (never, former, or current smoker), educational level (primary education, secondary
education, or academic/graduate), leisure-time physical activity (metabolic equivalent task minutes/d), baseline diabetes (yes or no), hypertension (yes or
no), hypercholesterolemia (yes or no), family history of coronary heart disease (yes or no), use of antihypertensive medication (yes or no), use of oral
antidiabetic agents (yes or no), and use of lipid-lowering drugs (yes or no). Extremes of total energy intake (.4000 or ,800 kcal/d in men and .3500 or
,500 kcal/d in women) were excluded. A major event was a composite of myocardial infarction, stroke, or death from cardiovascular causes. Ref, reference.
1568 GUASCH-FERRÉ ET AL.

had a higher prevalence of diabetes. They also consumed more from SFAs was associated with a higher risk of CVD (HR: 1.35;
cholesterol and less dietary fiber and carbohydrates. The drop- 95% CI: 1.03, 1.73) but not death.
out rate was 4.2% (3.0% in the MedDiet groups and 6.9% in the The associations of various isocaloric dietary substitutions on
control group). The drop-out rates according to different quin- the risk of CVD and all-cause death are shown in Figure 2. The
tiles of total fat were 8.7% in the first, 4.0% in the second, 3.3% replacement of 5% of energy from SFAs with MUFAs or PUFAs
in the third, 2.6% in the fourth, and 2.8% in the highest quintile. was associated with 37% and 33% lower risks of CVD. The
The drop-out distribution for quintiles of specific fat subtypes replacement of SFAs with PUFAs was associated with a 39%
was similar to that of total fat. lower risk of all-cause death. The isocaloric dietary substitution
Table 2 shows the RRs of CVD according to updated mea- of SFAs by carbohydrates was nonsignificant. The replacement
surements of total dietary fat and specific subtypes of fat intake. of 1% of energy from trans fat with MUFAs was associated with
The HRs represented the replacement of energy from total 8% lower risk of CVD, and replacing trans fat with PUFAs was
carbohydrates with the same percentage of energy from total fat associated with 8% lower risk of all-cause death.
and each type of fat. Total fat intake was significantly associated When using the cumulative average of dietary fat subtypes and
with a lower risk of CVD. In the fully adjusted model, compared total fat, the results were consistent with those reported for re-
with the lowest quintile (reference), the highest quintile of total peated measurements of intake. In addition, the results that used
fat intake was associated with a 42% lower CVD risk (HR: 0.58; an intention-to-treat approach were also consistent with those

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

95% CI: 0.39, 0.86). For specific types of fat, MUFAs and from the primary analysis.
PUFAs were also significantly associated with a lower risk of
CVD (Table 2). On the contrary, higher intakes of SFA and trans
fat were significantly associated with a higher risk of CVD after DISCUSSION
the model was adjusted for potential confounders. In this prospective study in individuals at high CVD risk,
Total fat intake was significantly associated with a lower risk of a higher dietary intake of total fat was associated with a lower risk
total death (Table 2). In the fully adjusted model, the risk of all- of CVD, whereas a higher intake of SFAs and trans fat was
cause death was 47% lower in the top quintile compared with the associated with a greater risk. Higher intakes of MUFAs and
reference (HR: 0.53; 95% CI: 0.37, 0.76). Inverse associations PUFAs were both associated with a lower risk of CVD, car-
between the intake of MUFAs and PUFAs and total death were diovascular death, and all-cause death. The isocaloric sub-
found. In the fully adjusted model, compared with the reference stitution of SFAs or trans fat with MUFAs or PUFAs was
quintile, the HR (95% CI) for the top quintile of MUFAs was 0.64 associated with a lower risk of CVD and death. In addition, we
(0.43, 0.94) and was 0.50 (0.35, 0.73) for PUFAs (P-trend , 0.01). found that a higher intake of SFAs from pastries and processed
trans Fat intake was significantly associated with a 38% higher food was associated with a higher risk of CVD. A new finding to
risk of all-cause death in the multivariable model 1, but the as- highlight was that SFAs from vegetable and fish were associated
sociation was attenuated in the fully adjusted model. SFAs were with a lower risk of CVD. However, this last finding should be
not significantly associated with all-cause death. interpreted cautiously because the amount of SFAs from fish
PUFA and MUFA intakes were also inversely associated with consumed in our population was small.
cardiovascular death (data not shown). The HR (95% CI) was The observed benefit on CVD and death of replacing SFAs and
0.37 (0.18, 0.76) for participants in the highest quintile of trans fat with PUFAs or MUFAs found in our population is con-
PUFA intake compared with those in the reference quintile (P- sistent with a pooled analysis of 11 cohorts (15) and a meta-
trend , 0.01). For MUFAs, the corresponding HR was 0.78 analyses of .20 randomized controlled trials (RCTs) (16).
(0.34, 0.97). SFA intake was associated with a greater risk of PUFAs in place of SFAs were found to be associated with a
cardiovascular death in the age-adjusted and the multivariable lower risk of coronary heart disease and death in a pooled anal-
model 1 (HR: 1.73; 95% CI: 1.10, 4.80); however, in the fully ysis of 11 prospective cohort studies (15). Another recent pooled
adjusted model, the association was nonsignificant. The re- analysis of 8 RCTs concluded that the coronary heart disease
spective HR for trans fat for individuals in the highest quintile risk is lowered by 10% for every 5% intake of energy from
was 1.53 (1.07, 2.18) in the fully adjusted model. PUFAs replacing SFAs (17). Even though the intake of MUFAs
When we analyzed the association between SFAs from dif- has been related to beneficial effects on health (18), there is still
ferent food sources and CVD and all-cause death (Table 3), we no unanimous rationale for MUFA recommendations (4). Our
observed that updated measurements of vegetable SFA intake results showed inverse associations between MUFAs and CVD,
(from oils, nuts, vegetables, margarine) tended to be associated all-cause death, and cardiovascular death. Of note, our pop-
with a 47% lower risk of CVD (P-trend = 0.06). Fish SFAs also ulation had a high intake of nuts and olive oil, which are com-
tended to be inversely associated with CVD and death [HR (95% mon MUFA sources. A possible reduction in the risk of CVD
CI): 0.70 (0.46, 0.99) and 0.63 (0.39, 0.98), respectively], al- and death was also found for the replacement of SFAs with
though the intake of this nutrient was low. SFAs from pastries MUFAs in previous studies (15, 16, 19). These data are in line
and processed foods were associated with a higher risk of CVD with our results, in which we found that substituting SFA with
(HR: 1.46; 95% CI: 1.01, 2.13; P-trend = 0.04). No significant MUFA intake was associated with a 37% lower risk of CVD.
associations were found for SFAs from dairy products or from meat Dietary fats were an important part of the dietary intervention
and processed meat and CVD and all-cause death (Table 3). conducted in the trial. Therefore, we acknowledge that it is
Figure 1 shows the risk of all-cause death and CVD for a 5% difficult to disentangle whether the associations found are driven
energy increment from total and subtypes of dietary fat. A 5% by the specific intake of MUFAs and PUFAs or by the supple-
energy increment from MUFAs and PUFAs was inversely as- ment foods provided (olive oil or mixed nuts) and the MedDiet
sociated with all-cause death and CVD. A 5% energy increment intervention. Nevertheless, we adjusted the analysis for intervention
 DIETARY FAT INTAKE, CVD, AND ALL-CAUSE DEATH 1569
TABLE 3
HRs (95% CIs) of cardiovascular disease and all-cause death according to quintile of updated measurements of SFA subtypes1
Quintile

1 (lowest) 2 3 4 5 (highest) P-trend

Cardiovascular disease
Vegetable SFAs (oils, nuts, bread,
vegetables, rice, margarine, other)
Median, % of energy 2.32 3.11 3.63 4.08 4.78
Multivariable model 1 1 (Ref) 0.81 (0.56, 1.15) 0.68 (0.45, 1.03) 0.78 (0.49, 1.22) 0.58 (0.33, 0.99) 0.08
Multivariable model 2 1 (Ref) 0.78 (0.55, 1.12) 0.68 (0.45, 1.02) 0.76 (0.48, 1.19) 0.53 (0.30, 0.93) 0.06
Dairy SFAs
Median, % of energy 0.55 1.36 2.11 2.98 4.50
Multivariable model 1 1 (Ref) 0.84 (0.56, 1.27) 1.15 (0.78, 1.70) 0.98 (0.65, 1.49) 0.86 (0.54, 1.36) 0.64
Multivariable model 2 1 (Ref) 0.83 (0.56, 1.25) 1.15 (0.78, 1.71) 0.98 (0.65, 1.49) 0.83 (0.52, 1.31) 0.54
SFAs from pastries and other processed
foods (cookies, donuts, bakery,

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

sauces, pizza, other)
Median, % of energy 0.04 0.25 0.58 1.07 1.86
Multivariable model 1 1 (Ref) 1.16 (0.79, 1.70) 1.02 (0.69, 1.51) 1.34 (0.92, 1.95) 1.37 (0.94, 2.01) 0.07
Multivariable model 2 1 (Ref) 1.21 (0.83, 1.78) 1.09 (0.74, 1.63) 1.44 (0.99, 2.09) 1.46 (1.01, 2.13) 0.04
SFAs from meat, processed meats
(sausages, hamburger, other), and eggs
Median, % of energy 0.91 1.46 1.96 2.52 3.58
Multivariable model 1 1 (Ref) 0.98 (0.66, 1.44) 0.95 (0.64, 1.42) 1.22 (0.81, 1.82) 1.19 (0.77, 1.84) 0.27
Multivariable model 2 1 (Ref) 0.99 (0.67, 1.46) 0.95 (0.64, 1.42) 1.22 (0.82, 1.83) 1.19 (0.77, 1.83) 0.28
Fish SFAs (includes seafood)
Median, % of energy 0.20 0.32 0.43 0.62 0.91
Multivariable model 1 1 (Ref) 0.92 (0.66, 1.29) 0.98 (0.70, 1.39) 0.92 (0.64, 1.32) 0.68 (0.45, 0.97) 0.12
Multivariable model 2 1 (Ref) 0.89 (0.64, 1.25) 0.97 (0.69, 1.36) 0.95 (0.66, 1.36) 0.70 (0.46, 0.99) 0.11
All-cause death
Vegetable SFAs (oils, nuts, bread,
vegetables, rice, margarine, other)
Multivariable model 1 1 (Ref) 0.76 (0.48, 1.21) 0.65 (0.38, 1.11) 0.79 (0.44, 1.43) 0.67 (0.34, 1.31) 0.28
Multivariable model 2 1 (Ref) 0.75 (0.47, 1.20) 0.64 (0.37, 1.10) 0.81 (0.45, 1.43) 0.70 (0.35, 1.38) 0.34
Dairy SFAs
Multivariable model 1 1 (Ref) 0.95 (0.56, 1.59) 1.37 (0.83, 2.27) 1.18 (0.68, 2.02) 1.47 (0.82, 2.62) 0.18
Multivariable model 2 1 (Ref) 0.94 (0.56, 1.57) 1.36 (0.82, 2.25) 1.16 (0.67, 1.99) 1.46 (0.82, 2.61) 0.19
SFAs from pastries and other processed
foods (cookies, donuts, bakery,
sauces, pizza, other)
Multivariable model 1 1 (Ref) 0.80 (0.49, 1.29) 0.95 (0.59, 1.53) 1.04 (0.65, 1.66) 1.26 (0.80, 1.99) 0.13
Multivariable model 2 1 (Ref) 0.79 (0.49, 1.29) 0.96 (0.59, 1.53) 1.03 (0.64, 1.65) 1.22 (0.78, 1.93) 0.17
SFAs from meat, processed meats
(sausages, hamburger, other), and eggs
Multivariable model 1 1 (Ref) 0.92 (0.56, 1.52) 1.02 (0.61, 1.68) 1.32 (0.80, 2.19) 1.06 (0.61, 1.85) 0.67
Multivariable model 2 1 (Ref) 0.91 (0.55, 1.50) 1.00 (0.60, 1.65) 1.28 (0.77, 2.12) 1.02 (0.58, 1.78) 0.77
Fish SFAs (includes seafood)
Multivariable model 1 1 (Ref) 0.76 (0.51, 1.14) 0.49 (0.30, 0.78) 0.67 (0.43, 1.05) 0.62 (0.38, 1.00) 0.10
Multivariable model 2 1 (Ref) 0.78 (0.52, 1.16) 0.50 (0.31, 0.81) 0.70 (0.45, 1.10) 0.63 (0.39, 0.98) 0.11
1
Time-dependent Cox regression models were used to assess the risk of cardiovascular disease and all-cause death by quintile of updated measurements
of subtypes of SFA intake. Multivariable model 1 was adjusted for age (y), sex, intervention group, yearly updated total energy (kcal/d), alcohol intake
(continuous, adding a quadratic term), and quintiles of updated fiber, protein, carbohydrates, dietary cholesterol, and other subtypes of fat intake (MUFAs,
PUFAs, and trans fat). Model 2 was further adjusted for nondietary variables and classical cardiovascular disease risk factors: BMI (kg/m2), smoking status
(never, former, or current smoker), educational level (primary education, secondary education, or academic/graduate), leisure-time physical activity (metabolic
task equivalent minutes/d), baseline diabetes (yes or no), hypertension (yes or no), hypercholesterolemia (yes or no), family history of coronary heart disease
(yes or no), use of antihypertensive medication (yes or no), use of oral antidiabetic agents (yes or no), use of lipid-lowering drugs (yes or no). Extremes of total
energy intake (.4000 or ,800 kcal/d in men and .3500 or ,500 kcal/d in women) were excluded. A major event was a composite of myocardial infarction,
stroke, and death from cardiovascular causes.

group to control for the effect of the intervention. In addition, in Although dietary patterns and analysis of food items often tell us
a further analysis, when stratifying the sample by intervention more about lifestyles than do single nutrients, it is important to
group we found similar results in the 3 intervention groups, investigate which individual nutrients within dietary patterns
suggesting that the results are independent of the intervention. exert or amplify beneficial or detrimental effects on disease risk.
1570 GUASCH-FERRÉ ET AL.

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

FIGURE 1 HRs (95% CIs) of all-cause death and cardiovascular disease for a 5% energy increment of total and subtypes of dietary fat. Multivariable
time-dependent Cox regression models were adjusted for age (y), sex, intervention group, BMI (kg/m2), smoking status (never, former, or current smoker),
educational level (primary education, secondary education, or academic/graduate), leisure-time physical activity (metabolic equivalent task minutes/d),
baseline diabetes (yes or no), hypertension (yes or no), hypercholesterolemia (yes or no), family history of coronary heart disease (yes or no), use of
antihypertensive medication (yes or no), use of oral antidiabetic agents (yes or no), use of lipid-lowering drugs (yes or no), yearly updated total energy
(kcal/d), alcohol intake (continuous, adding a quadratic term), and yearly updated fiber, protein, and dietary cholesterol for the total fat analysis. For specific
subtypes of fat, the model also included as covariates the other subtypes of fat. All models were stratified by recruitment center. Extremes of total energy
intake were excluded (n = 7038). The estimated RRs can be interpreted as the estimated differences in risk of a 5% lower energy intake from carbohydrate and
a concomitant higher energy intake from total dietary fat and subtypes of dietary fat.

A higher intake of total fat in our population was associated 58,672 Japanese individuals (22), total fat intake was not asso-
with a lower risk of CVD and all-cause death. Of note, the greater ciated with cardiovascular death in the present study. In agree-
proportion of total fat consumed in this population came from ment with several other studies (7, 16, 19, 23), trans-fat intake
MUFAs and PUFAs, which have shown beneficial effects on was associated with a higher risk of CVD, and benefits of its
health (20). On the contrary, and in accordance with the results of replacement were also shown. Our results contribute to the
the Nurses’ Health Study (21) and a recent prospective study in convincing evidence on the benefits of decreasing the intake of

FIGURE 2 Estimated HRs (95% CIs) of all-cause death and cardiovascular disease associated with isocaloric substitutions of one dietary component for
another. Multivariable time-dependent Cox regression models were adjusted for age (y), sex and intervention group, BMI (kg/m2), smoking status (never,
former, or current smoker), educational level (primary education, secondary education, or academic/graduate), leisure-time physical activity (metabolic
equivalent task minutes/d), baseline diabetes (yes or no), hypertension (yes or no), hypercholesterolemia (yes or no), family history of coronary heart disease
(yes or no), use of antihypertensive medication (yes or no), use of oral antidiabetic agents (yes or no), use of lipid-lowering drugs (yes or no), yearly updated
total energy (kcal/d), alcohol intake (continuous, adding a quadratic term), and yearly updated fiber, protein, carbohydrates, dietary cholesterol, and other
subtypes of fat intake. All models were stratified by recruitment center. Extremes of total energy intake were excluded (n = 7038). The estimated RRs can be
interpreted as the estimated differences in risk of a 5% lower energy intake from SFAs or 1% lower energy from trans fat and a concomitant higher energy
intake from carbohydrates, MUFAs, and PUFAs, respectively. E, energy.
 DIETARY FAT INTAKE, CVD, AND ALL-CAUSE DEATH 1571
trans fat to a minimum for its known adverse effects on health, resistance, endothelial function (26), and glycemic control (27).
particularly industrial hydrogenated trans fat (24, 25). Second, PUFAs have been shown to exert antithrombotic and
Substantial accruing evidence from meta-analyses indicates antiarrhythmic effects (especially n–3 fatty acids) and improve
that the effects of the consumption of SFAs on CVD risk vary serum lipids, inflammation, blood pressure, endothelial func-
depending on the replaced nutrient (6, 15–17, 19). Most of the tion, and myocardial oxygen utilization (28–30), thus improving
meta-analyses failed to show significant associations between atherosclerosis pathways and consequently decreasing the risk
the intake of SFAs and risk of CVD, stroke, or death (6, 19, 23). of CVD. Third, evidence has suggested that SFAs increase LDL
However, they were unable to consider the effects of replacing cholesterol, and therefore they can increase the risk of CVD (28,
nutrients and the effects of different food sources of these SFAs 31). The replacement of SFAs with PUFAs or MUFAs has been
(7, 8). Our results showed that increased consumption of SFAs shown to decrease total and LDL cholesterol but had minimal
was associated with a higher risk of CVD. But no associations effect on HDL cholesterol (32, 33). SFA replacement with PUFA
with cardiovascular and total death were found. SFA intake was and/or high-quality carbohydrates can reduce as well the risk
not associated with all-cause mortality and CVD mortality in of coronary heart disease (34). SFAs may also reduce insulin
a recent meta-analysis of 6 prospective cohort studies (23). sensitivity, promote inflammation, and have adverse effects on
Contrary to our study, in that meta-analysis (23) the replacement vascular function (6), but this still remains to be elucidated.
of SFAs or trans fat with other nutrients was not evaluated. We Moreover, consistent scientific evidence has shown that trans fat,

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

found a consistent reduction in the risk of both CVD and death regardless of its source, increases the ratio of plasma LDL to HDL
when replacing the intake of SFAs with other dietary compo- cholesterol (24). Our results may help to highlight the importance
nents such as MUFAs or PUFAs. No significant associations of the quality of fat rather than total fat in the specific recom-
were found for the substitution of SFAs with total carbohydrates mendations for individuals at high CVD risk and provide epide-
and CVD and death, probably because it is important to dif- miologic evidence for the current guidelines. Our study is the first,
ferentiate between carbohydrate types, such as legumes, whole to our knowledge, to test the association between different SFA
grains, and refined sugars. These results are in accordance with food sources on CVD and death in individuals at high CVD risk
the Scientific Report of the 2015 Dietary Guidelines for Americans by using repeated measurements of dietary intake.
Advisory Committee (3). Limitations of the present study require discussion. First,
Because food sources contain different proportions of specific because the method used for dietary assessment was an FFQ,
SFAs, which may have a different influence on CVD, we eval- measurement errors in total fat and subtypes of fat intake were
uated the associations of SFAs from several food sources. MUFAs inevitable. However, the FFQ used was validated, and the re-
mainly come from olive oil, PUFAs derive from nuts, and the peated measurements of dietary fat analyzed may provide a more
amount of trans fatty acids consumed is very low in this elderly robust approach to test associations than with only one-time
Mediterranean population at high CVD risk. As expected, we assessment as an exposure. Although dropouts were different
found that SFAs from pastries and processed foods were asso- across categories of fat intake, it is unlikely that a differential
ciated with a higher risk of CVD. Our findings are consistent attrition bias may have occurred and might provide an alternative
with those from a large multiethnic cohort that evaluated the explanation to our findings. Differences in attrition rates among
association of SFA consumption from different food sources and categories of fat intake were therefore small in magnitude and not
the incidence of CVD (9). In that study they found that dairy very likely to induce a relevant differential misclassification bias
SFAs were associated with a lower CVD risk, but in our analysis in the outcomes. Furthermore, the potential bias, if any, would be
SFAs from dairy products were not significantly associated with in the direction of an underestimation of the number of outcomes
CVD and all-cause death. Perhaps the beneficial effects of dairy occurring in the lowest category of total fat intake, because there
products on CVD are mainly attributed to other nutrients, such were more dropouts in the first quintile and we may have missed
as calcium, potassium, peptides, and some vitamins, instead of
some cases of CVD in the dropouts. Finally, we estimated the
only the type of SFA. Vegetable SFAs, mainly from vegetable
hypothetical effect of isocaloric substitutions of dietary fat
oils, nuts, vegetables, and margarine, were inversely associated
for another dietary component under causality of assumption.
with CVD in our study. In the study by de Oliveira Otto et al.
Strengths of the present study include its prospective design, the
(9), no associations between plant SFAs and CVD were re-
use of repeated dietary measurements during follow-up, the
ported, but the SFA consumption of these food sources was quite
ability to control for potential confounders due to recording of
low compared with our population. In addition, that study (9)
comprehensive data, and the accurate and blind assessment of
comprised the US general population, whereas our study was
incident cases of CVD and death.
conducted in Mediterranean individuals at high CVD risk. In-
In summary, MUFAs and PUFAs were inversely associated
verse associations between fish SFAs and CVD were also found
with CVD and death, whereas SFAs and trans fat were associated
in our analysis. Because this approach is relatively new, further
with a higher risk of CVD in individuals at high CVD risk. Total
studies are needed to confirm these promising findings. It should
dietary fat was associated with a lower risk of CVD and all-
be acknowledged that other dietary components in the food
cause death. The replacement of SFAs with MUFAs or PUFAs
sources containing SFAs may have played a role in the observed
or trans fat with PUFAs was associated with a lower risk of
associations, such as refined carbohydrates in pastries or salt in
CVD. Finally, the consumption of SFAs from pastries and
processed foods.
Several biological mechanisms underlie the aforementioned processed food was associated with a greater risk in high-risk
associations. First, MUFAs have beneficial effects on CVD risk individuals.
factors. Scientific evidence has shown that MUFAs improve the The PREDIMED investigators are as follows—University of Navarra and
lipid profile (4), decrease blood pressure, and modulate insulin Osasunbidea (Servicio Navarro de Salud), Primary Care Centers, Pamplona,
1572 GUASCH-FERRÉ ET AL.
Spain: B Sanjulian, A Sanchez-Tainta, A Marti, P Buil-Cosiales, M Serrano- REFERENCES
Martínez, J Diez-Espino, A García-Arellano, J Basterra-Gortari, 1. Food and Agriculture Organization of the United Nations. Summary of
EH Martínez-Lapiscina, N Ortuno, N Berrade, V Extremera-Urabayen, conclusions and dietary recommendations on total fat and fatty acids in
C Arroyo-Azpa, L García -Perez, J Villanueva Telleria, F Cortes Ugalde, fats and fatty acids in human nutrition—report of an expert consulta-
T Sagredo Arce, MD García de la Noceda Montoy, MD Vigata López, tion. FAO/WHO, Geneva; 2010.
MT Arceiz Campo, A Urtasun Samper, MV Gueto Rubio, and B Churio 2. Hu FB, Manson JE, Willett WC. Types of dietary fat and risk of cor-
Beraza; Hospital Clinic, IDIBAPS, Barcelona, Spain: M Serra, A Perez- onary heart disease: a critical review. J Am Coll Nutr 2001;20:5–19.
3. USDA; US Department of Health and Human Services. Scientific re-
Heras, C Viñas, R Casas, L de Santamaria, S Romero, JM Baena, M
port of the 2015 Dietary Guidelines Advisory Committee. UDSA,
García, M Oller, J Amat, I Duaso, Y García, C Iglesias, C Simon, USA; 2015.
L Quinzavos, L Parra, M Liroz, J Benavent, J Clos, I Pla, M Amoros, 4. Michas G, Micha R, Zampelas A. Dietary fats and cardiovascular
MT Bonet, MT Martin, MS Sanchez, J Altirruba, E Manzano, A Altes, disease: putting together the pieces of a complicated puzzle. Athero-
M Cofan, C Valls-Pedret, A Sala-Vila, and M Domenech; University sclerosis 2014;234:320–8.
Rovira i Virgili, Reus, Spain: J Basora, R González, C Molina, 5. Estruch R, Ros E, Salas-Salvado J, Covas MI, Corella D, Aros F,
F Marquez, P Martínez, N Ibarrola, M Sorli, J García Roselló, A Castro, Gomez-Gracia E, Ruiz-Gutierrez V, Fiol M, Lapetra J, et al. Primary
F Martin, N Tort, A Isach, M Baldrich, JJ Cabre, G Mestres, F Paris, prevention of cardiovascular disease with a Mediterranean diet. N Engl
M Llauradó, R Pedret, J Basells, J Vizcaino, and J Fernández-Ballart; J Med 2013;368:1279–90.
6. Siri-Tarino PW, Sun Q, Hu FB, Krauss RM. Meta-analysis of pro-
Institute de Recerca Hospital del Mar, Barcelona, Spain: MI Covas,
spective cohort studies evaluating the association of saturated fat with
S Tello, J Vila, R de la Torre, D Munoz-Aguayo, R Elosua, J Marrugat,

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

cardiovascular disease. Am J Clin Nutr 2010;91:535–46.
and M Ferrer; University of Valencia, Valencia, Spain: P Carrasco, 7. Skeaff CM, Miller J. Dietary fat and coronary heart disease: summary
R Osma, M Guillen, P Guillem-Saiz, O Portoles, V Pascual, C Riera, of evidence from prospective cohort and randomised controlled trials.
J Valderrama, A Serrano, E Lazaro, A Sanmartin, A Girbes, V Santamaria, Ann Nutr Metab 2009;55:173–201.
C Sanchez, and Z Pla; University Hospital of Alava, Vitoria, Spain: I 8. Chowdhury R, Warnakula S, Kunutsor S, Crowe F, Ward HA, Johnson
Salaverria, T del Hierro, J Algorta, S Francisco, A Alonso-Gómez, E L, Franco OH, Butterworth AS, Forouhi NG, Thompson SG, et al.
Sanz, J Rekondo, MC Belló, and A Loma-Osorio; University of Malaga, Association of dietary, circulating, and supplement fatty acids with
Malaga, Spain: J Wärnberg, R Benitez Pont, M Bianchi Alba, R Gomez- coronary risk: a systematic review and meta-analysis. Ann Intern Med
2014;160:398–406.
Huelgas, J Martínez-González, V Velasco García, J de Diego Salas, A
9. de Oliveira Otto MC, Mozaffarian D, Kromhout D, Bertoni AG, Sibley
Baca Osorio, J Gil Zarzosa, JJ Sánchez Luque, and E Vargas López; CT, Jacobs DR, Nettleton JA. Dietary intake of saturated fat by food
Instituto de la Grasa, Consejo Superior de Investigaciones Científicas, source and incident cardiovascular disease: the Multi-Ethnic Study of
Seville, Spain: J Sánchez Perona, E Montero Romero, M García- Atherosclerosis. Am J Clin Nutr 2012;96:397–404.
García, and E Jurado-Ruiz; Institute of Health Sciences IUNICS, 10. Astrup A. Yogurt and dairy product consumption to prevent car-
University of Balearic Islands, and Hospital Son Espases, Palma de diometabolic diseases: epidemiologic and experimental studies. Am J
Mallorca, Spain: M García-Valdueza, M Moñino, A Proenza, R Prieto, Clin Nutr 2014;99(Suppl):1235S–42S.
G Frontera, M Ginard, F Fiol, A Jover, D Romaguera, and J García; 11. Martínez-González MÁ, Corella D, Salas-Salvado J, Ros E, Covas
Department of Family Medicine, Primary Care Division of Seville, Seville, MI, Fiol M, Warnberg J, Aros F, Ruiz-Gutierrez V, Lamuela-Raventos
RM, et al. Cohort profile: design and methods of the PREDIMED
Spain: J Lapetra, M Leal, E Martínez, M Ortega-Calvo, P Roman, F Jose
study. Int J Epidemiol 2012;41:377–85.
García, P Iglesias, Y Corchado, E Mayoral, L Mellado, L Miró, JM 12. Fernández-Ballart JD, Pinol JL, Zazpe I, Corella D, Carrasco P, Toledo
Lozano, and C Lama; School of Pharmacy, University of Barcelona, E, Perez-Bauer M, Martinez-Gonzalez MA, Salas-Salvado J, Martin-
Barcelona, Spain: MC López- Sabater, AI Castellote-Bargallo, A Medina- Moreno JM. Relative validity of a semi-quantitative food-frequency
Remon, and A Tresserra-Rimbau; University of Las Palmas de Gran questionnaire in an elderly Mediterranean population of Spain. Br J
Canaria, Las Palmas, Spain: J Alvarez-Perez, E Diez Benitez, I Bautista Nutr 2010;103:1808–16.
Castaño, I Maldonado Diaz, A Sanchez-Villegas, MJ Férnandez- 13. Mataix J. Tablas de composición de alimentos. [Food composition
Rodríguez, F Sarmiendo de la Fe, C Simon García, I Falcon Sanabria, tables.] 4th ed. Granada (Spain): Universidad de Granada; 2003
B Macias Gutierrez, and AJ Santana Santana; Hospital Universitario de (in Spanish).
14. Elosua R, Marrugat J, Molina L, Pons S, Pujol E; The MARATHOM
Bellvitge, Hospitalet de Llobregat, Barcelona, Spain: E de la Cruz, A
Investigators. Validation of the Minnesota Leisure Time Physical
Galera, Y Soler, F Trias, I Sarasa, E Padres, and E Corbella; and Activity Questionnaire in Spanish men Am J Epidemiol 1994;139:
Primary Care Division, Catalan Institute of Health, Barcelona, Spain: 1197–209.
C Cabezas, E Vinyoles, MA Rovira, L García, G Flores, JM Verdu, P 15. Jakobsen MU, O’Reilly EJ, Heitmann BL, Pereira MA, Bälter K,
Baby, A Ramos, L Mengual, P Roura, MC Yuste, A Guarner, A Rovira, Fraser GE, Goldbourt U, Hallmans G, Knekt P, Liu S, et al. Major
MI Santamaria, M Mata, C de Juan, and A Brau. Other investigators of types of dietary fat and risk of coronary heart disease: a pooled analysis
the PREDIMED network: JA Tur (University of Balearic Islands), MP of 11 cohort studies. Am J Clin Nutr 2009;89:1425–32.
Portillo (University of Basque Country), and G Saez (University of 16. Hooper L, Summerbell CD, Thompson R, Sills D, Roberts FG, Moore
Valencia). HJ, Davey Smith G. Reduced or modified dietary fat for preventing
cardiovascular disease. Cochrane Database Syst Rev 2012;5:CD002137.
The authors’ responsibilities were as follows—MAM-G, DC, ER, RE,
17. Mozaffarian D, Micha R, Wallace S. Effects on coronary heart disease
FA, EG-G, M Fiol, JMS-L, LS-M, M Fitó, and JS-S: designed the re- of increasing polyunsaturated fat in place of saturated fat: a systematic
search; MG-F, NB, MAM-G, DC, ER, SM-P, RE, FA, EG-G, review and meta-analysis of randomized controlled trials. PLoS Med
M Fiol, JMS-L, LS-M, MB, ET, RB, M Fitó, and JS-S: conducted 2010;7:e1000252.
the research; MG-F, NB, AG, and JS-S: analyzed the data; MG-F 18. Yu S, Derr J, Etherton TD, Kris-Etherton PM. Plasma cholesterol-
and JS-S: wrote the manuscript; MAM-G, DC, ER, RE, M Fiol, predictive equations demonstrate that stearic acid is neutral and
JMS-L, LS-M, and JS-S: were the coordinators of subject recruit- monounsaturated fatty acids are hypocholesterolemic. Am J Clin Nutr
ment at the outpatient clinics; MG-F and JS-S: had full access to all 1995;61:1129–39.
of the data in the study and take responsibility for the integrity of 19. Mente A, de Koning L, Shannon HS, Anand SS. A systematic review
of the evidence supporting a causal link between dietary factors and
the data and the accuracy of the data analysis; and all authors:
coronary heart disease. Arch Intern Med 2009;169:659–69.
revised the manuscript for important intellectual content and read 20. Schwab U, Lauritzen L, Tholstrup T, Haldorssoni T, Riserus U,
and approved the final manuscript. None of the funding sources played Uusitupa M, Becker W. Effect of the amount and type of dietary fat on
a role in the design, collection, analysis, or interpretation of the data or cardiometabolic risk factors and risk of developing type 2 diabetes,
in the decision to submit the manuscript for publication. The authors cardiovascular diseases, and cancer: a systematic review. Food Nutr
declared that they have no competing interests. Res 2014 Jul 10;58. DOI: 10.3402/fnr.v58.25145. eCollection 2014.
 DIETARY FAT INTAKE, CVD, AND ALL-CAUSE DEATH 1573
21. Oh K, Hu FB, Manson JE, Stampfer MJ, Willett WC. Dietary fat intake 27. Garg A. High-monounsaturated-fat diets for patients with diabetes
and risk of coronary heart disease in women: 20 years of follow-up of mellitus: a meta-analysis. Am J Clin Nutr 1998;67(Suppl):577S–82S.
the nurses’ health study. Am J Epidemiol 2005;161:672–9. 28. Mensink RP, Zock PL, Kester ADM, Katan MB. Effects of dietary fatty
22. Wakai K, Naito M, Date C, Iso H, Tamakoshi A. Dietary intakes of fat acids and carbohydrates on the ratio of serum total to HDL cholesterol
and total mortality among Japanese populations with a low fat intake: and on serum lipids and apolipoproteins: a meta-analysis of 60 con-
the Japan Collaborative Cohort (JACC) Study. Nutr Metab (Lond) trolled trials. Am J Clin Nutr 2003;77:1146–55.
2014;11:12. 29. Hall WL. Dietary saturated and unsaturated fats as determinants of
23. de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, blood pressure and vascular function. Nutr Res Rev 2009;22:18–38.
Uleryk E, Budylowski P, Schünemann H, Beyen J, et al. Intake of 30. Pepe S, McLennan PL. Cardiac membrane fatty acid composition
saturated and trans unsaturated fatty acids and risk of all cause modulates myocardial oxygen consumption and postischemic recovery
mortality, cardiovascular disease, and type 2 diabetes: systematic of contractile function. Circulation 2002;105:2303–8.
review and meta-analysis of observational studies. BMJ 2015;351: 31. Baum SJ, Kris-Etherton PM, Willett WC, Lichtenstein AH, Rudel LL,
h3978. Maki KC, Whelan J, Ramsden CE, Block RC. Fatty acids in cardio-
24. Brouwer IA, Wanders AJ, Katan MB. Effect of animal and industrial vascular health and disease: a comprehensive update. J Clin Lipidol
trans fatty acids on HDL and LDL cholesterol levels in humans— 2012;6:216–34.
a quantitative review. PLoS One 2010;5:e9434. 32. Siri-Tarino PW, Sun Q, Hu FB, Krauss RM. Saturated fat, carbohy-
25. Bendsen NT, Christensen R, Bartels EM, Astrup A. Consumption of drate, and cardiovascular disease. Am J Clin Nutr 2010;91:502–9.
industrial and ruminant trans fatty acids and risk of coronary heart 33. Micha R, Mozaffarian D. Saturated fat and cardiometabolic risk fac-
disease: a systematic review and meta-analysis of cohort studies. Eur J tors, coronary heart disease, stroke, and diabetes: a fresh look at the
Clin Nutr 2011;65:773–83. evidence. Lipids 2010;45:893–905.

Downloaded from ajcn.nutrition.org at UNIVERSITY OF NEBRASKA on December 1, 2015

26. Esposito K, Marfella R, Ciotola M, Di Palo C, Giugliano F, Giugliano 34. Li Y, Hruby A, Bernstein AM, Ley SH, Wang DD, Chiuve SE,
G, D’Armiento M, D’Andrea F, Giugliano D. Effect of a Mediterranean- Sampson L, Rexrode KM, Rimm EB, Willett WC, et al. Saturated fats
style diet on endothelial dysfunction and markers of vascular in- compared with unsaturated fats and sources of carbohydrates in re-
flammation in the metabolic syndrome: a randomized trial. JAMA lation to risk of coronary heart disease: a prospective cohort study.
2004;292:1440–6. J Am Coll Cardiol 2015;66:1538–48.