474_N01-268 8/4/2002 4:07 pm Page 474 (Black plate)

© International Epidemiological Association 2002 Printed in Great Britain International Journal of Epidemiology 2002;31:474–480

CARDIOVASCULAR DISEASE

Risk of first non-fatal myocardial infarction

negatively associated with olive oil
consumption: a case-control study in Spain
E Fernández-Jarne,a,b E Martínez-Losa,a,c M Prado-Santamaría,a,c C Brugarolas-Brufau,a,c
M Serrano-Martíneza,c and MA Martínez-Gonzáleza

Background Olive oil is the main source of dietary lipids in most Mediterranean countries
where mortality and incidence rates for coronary heart disease (CHD) are the
lowest in Europe. Although international comparisons and mechanistic reasons
support the hypothesis that a high olive oil intake may prevent CHD, limited data
from studies of individuals are available.
Methods A hospital-based case-control study was conducted in Pamplona (Spain) recruiting
171 patients (81% males, age ,80 years) who suffered their first acute myocardial
infarction and 171 age-, gender- and hospital-matched controls (admitted to minor
surgery, trauma or urology wards). A validated semi-quantitative food frequency
questionnaire (136 items) was used to appraise previous long-term dietary
exposures. The same physician conducted the face-to-face interview for each case
patient and his/her matched control. Conditional logistic regression modelling
was used to take into account potential dietary and non-dietary confounders.
Results The exposure to the upper quintile of energy-adjusted olive oil (median intake:
54 g/day) was associated with a statistically significant 82% relative reduction in
the risk of a first myocardial infarction (OR = 0.18; 95% CI : 0.06–0.63) after
adjustment for dietary and non-dietary confounders.
Conclusions Our data suggest that olive oil may reduce the risk of coronary disease. These
findings require confirmation in further observational studies and trials.
Keywords Diet, epidemiology, coronary disease, fatty acids
Accepted 10 October 2001

Dietary patterns found in olive-growing areas of the Mediter-
ranean region have been postulated as protective against
coronary heart disease (CHD).1 Olive oil, rich in monounsaturated
fatty acids (MUFA), is the main source of dietary lipids in most
Mediterranean countries.
The very low CHD mortality rates found in Mediterranean
countries, together with a wide array of mechanistic reasons,
have led to the idea that instead of recommending a low-fat diet
to prevent CHD, it would be worthwhile to give the population
a Department of Epidemiology and Public Health, University of Navarre,
Pamplona, Spain.
b Department of Cardiology. University Clinic of Navarre, Spain.
c Navarre Primary Care Health Services, Spain.
Correspondence: Prof. Miguel Ángel Martínez-González, Epidemiología y
Salud Pública, Facultad de Medicina, Universidad de Navarra, Irunlarrea 1,
E-31080 Pamplona, Spain. E-mail: mamartinez@unav.es
the message of augmenting the intake of olive oil, while avoid-
ing animal and trans-fats.2,3 A widespread recommendation
promoting olive oil consumption to replace saturated and trans-
unsaturated fat needs to be solidly based on epidemiological
findings conducted in populations where consumption levels
are high and heterogeneous.
Apart from international comparisons and ecological correla-
tions, and the outstanding findings of the Seven Countries Study,4
there is little direct evidence from analytical epidemiological
studies relating diet to CHD in Mediterranean countries. A small
randomized trial of corn oil and olive oil carried out almost
40 years ago by GA Rose found no benefit for olive oil and even
an adverse significant effect for corn oil in 80 coronary patients
after 2 years of follow-up.5 A case-control study in Italian
women (287 cases/649 controls) reported no significant benefit
for oil consumption (odds ratio [OR] = 0.7 for the second tertile
and 1.1 for the third).6 A case-control study in Greece did not
find any significant protection from MUFA intake.7

474
474_N01-268 8/4/2002 4:07 pm Page 475 (Black plate)
OLIVE OIL CONSUMPTION AND MYOCARDIAL INFARCTION
The same group8 found that an a priori defined Mediter-
ranean dietary pattern was associated with advantageous
survival in a cohort of elderly people. Three other small studies
have consistently reported similar results in Australia,9 Spain10
and Italy11 using analogous methodologies. However, the out-
come in all of these four studies included every cause of death
and no information about the specific role of olive oil on CHD
risk was reported.
A randomized secondary prevention trial conducted in
France12 showed an impressive protection provided by an
experimental Mediterranean diet on the risk of death and re-
infarction among survivors of a first acute myocardial infarction
(AMI). Nevertheless, as the major element of the assigned diet
was an experimental canola-oil based margarine and the diet
simultaneously included a high intake of alpha-linolenic acid,
fruit and vegetables, it was not possible to attribute its benefit to
a single factor. In addition, the nutritional factors associated
with primary and secondary prevention of CHD need not to be
the same. The aim of our study was to assess the potential role
of olive oil for the primary prevention of CHD and to quantify
the reduction in the risk of a first AMI that can be provided by
a high dietary olive oil intake.
Methods
Cases were defined as male or female subjects, aged under 80,
survivors of a first AMI (ICD code 410) admitted to one of the
three tertiary hospitals of Pamplona (Spain) within the periods
October 1999–June 2000 or October 2000–February 2001. They
had to fulfil the criteria13 for definite AMI of the MONICA pro-
ject (two or more ECG showing specific changes; ECG showing
probable changes plus abnormal cardiac enzymes; or typical
symptoms plus abnormal enzymes). A previous history of
angina pectoris, a previous diagnosis of CHD or other prior diag-
nosis of major cardiovascular disease were exclusion criteria.
Informed consent was obtained from the patients and the
project was approved by the Institutional Review Board of the
Medical School. We identified 180 eligible cases. Nine of them
refused to participate (participation = 95%).
A control subject of the same age (5-year bands), gender
and hospital was matched to each case. Eligible controls were
patients admitted to the surgical, trauma or urology wards of
the same hospital during the same month that matched cases
for diseases believed to be unrelated to diet. Eight eligible
controls refused to participate (participation: 96%) and each of
them was replaced by other patients of similar characteristics for
matching variables.
Cases and controls were interviewed in a standard way with
the same questionnaire. All interviews were conducted by four
physicians belonging to the research team (EFJ, EML, MP, CB).
The same physician who interviewed a case patient also inter-
viewed the respective matched control. The physicians had to
exclude cases with a previous history of angina or other cardio-
vascular symptoms. Therefore, they were not blinded to the
participant’s disease status. The physician approached the patients,
invited them to participate and provided them with the self-
administered questionnaire. It included a semi-quantitative
food-frequency questionnaire (118 food items), previously
validated in Spain,14 that was slightly expanded for this study
(136 items plus vitamin supplements). For each food item, a
475
commonly used portion size was specified, and participants
were asked how often they had consumed that unit on average
over the previous year. Emphasis was added to ensure that the
answers were related to long-term dietary exposures and not
to recent changes in diet. Nine options for frequency of con-
sumption were possible. The type of fat used in frying was
specifically assessed. A dietitian updated the nutrient data bank
using the latest available information included in the food com-
position tables for Spain. Total energy-adjusted intakes were
computed using the residuals method.15
The participants were asked to report their usual time spent
practising the following activities: walking, jogging, running,
athletics, cycling, swimming, racquet sports, soccer, team-sports
other than soccer, dancing, aerobics, hill-walking, climbing,
gardening, skiing, skating, fishing, martial arts, and watersports.
To quantify the volume and intensity of leisure-time physical
activity, we computed an activity metabolic equivalent (MET)
index by assigning a multiple of resting metabolic rate (MET
score) to each activity. It was multiplied by the weekly time
spent in each activity obtaining a value of overall weekly MET-
hours.16 This measurement represents both the amount and
relative intensity of physical exercise during a week for each
participant. Five cases did not personally answer the question-
naire, and we used the answers given by a relative. We used the
same procedure for matched controls.
The physician clarified any questions the patient may have
had in completing the questionnaire, and subsequently the
physician conducted a face-to-face interview about his or her
coronary risk factors (smoking, diabetes, high blood pressure,
high blood cholesterol, recent weight changes) and family
history of cardiovascular disease. The physician took systolic
and fifth-phase diastolic blood pressure readings and measured
weight and height according to a standardized protocol, with
the subject barefoot and dressed in light clothing. For each
participant we calculated the body mass index (BMI) as the
weight in kilograms divided by the squared height in metres
(kg/m2).
Most cases (166/171) were interviewed in the cardiology ward
once they had been discharged from the coronary unit. Two of
them were interviewed at their homes after being discharged
from the hospital. All control subjects were interviewed in hos-
pital wards, except one who was interviewed at home.
The association of olive oil consumption with myocardial
infarction was calculated through conditional logistic regression
using matched data of 171 case-control pairs. Quintiles of olive
oil intake defined according to the distribution among controls
were compared regarding several potential confounding
variables. First, we fitted models using crude olive oil intake
(unadjusted for total energy intake) as the independent variable.
Then we used energy-adjusted values of olive oil.15 Potential
confounders were introduced in two steps in both multivariable
models. First, we introduced non-dietary confounders. In a
second model we also added the dietary confounders. Quadratic
terms for some confounders, including ethanol, were used to
account for non-linear relationships. We selected confounders
by taking into account previous published literature about
coronary risk factors and avoided the reliance on P-values or
stepwise approaches. Tests for trend were done using the median
of each quintile as a continuous variable. Reported P-values are
two-tailed; values ,0.05 were considered significant.
474_N01-268 8/4/2002 4:07 pm Page 476 (Black plate)

476 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Results BMI, prior history of hypertension, diabetes or high blood

cholesterol. Also as expected, leisure-time physical activity was
The average daily intake of olive oil was 22.8 g (SD: 19.9) in higher among controls. Most case-control differences in the
women and 25.3 g (SD: 18.0) in men. The total energy-adjusted crude mean intake of nutrients were not statistically significant,
mean was 24.9 for both genders. Thus, the slightly higher and in general, they were small.
absolute intake among men was explained by their higher The cut-off points for the quintiles of energy-adjusted olive
energy intake. oil intake were 10.4, 17.6, 21.2 and 37.9 g/day in the whole
The distribution of socio-demographic variables was similar in sample and 11.4, 18.2, 24.6 and 40.7 g/day among controls.
cases and controls (Table 1). Nevertheless, a slightly higher pro- The mean paired difference in the intake of energy-adjusted
portion of cases than controls had a university degree, whereas olive oil between cases and controls was –1.4 g/day (higher
a higher proportion of controls than cases were working in intake among controls).
qualified non-manual jobs (i.e. ‘white collar‘). As expected, Table 2 shows the distribution of potential confounding vari-
cases were more likely to be current smokers, have a higher ables across quintiles of energy-adjusted olive oil intake among
control subjects. A higher proportion of married subjects, of those
Table 1 Characteristics of case and control participants with prior history of diabetes and of current smokers was found
among those with a higher energy-adjusted olive oil intake.
Cases Controls A higher mean ethanol intake was also associated with a higher
(n = 171) (n = 171) Pa
olive oil intake. The intake of some nutrients previously reported
Age (years, mean) 61.7 61.4 b
to be inversely associated with CHD, such as total dietary fibre,17
b
Gender (% men) 81 81 folic acid,18 vitamin B6,18 and vitamin C,19 was significantly
Educational level (%) 0.37 lower among those controls with greater energy-adjusted olive
,Primary 28 29 oil consumption. A lower glycaemic load20 was found in the
Primary 44 45 highest quintile of olive oil consumption, but the trend was not
Secondary 12 16 statistically significant.
University 16 10 When we used the quintiles of olive oil intake without
Occupational level (%) 0.63 energy-adjustment as the exposure variable (Table 3), the
White collar 21 25 point-estimates for the OR were lower than 1 in the three upper
Blue collar 18 16
quintiles of olive oil intake. Exposure to the upper quintile
of olive oil was associated with a relative risk reduction of 64%
Retired 44 46
(OR = 0.36, 95% CI : 0.12–1.08) with respect to the first quintile
Housewife 12 10
(median intake: 7.2 g/day). The linear trend test was in the limit
Other 5 3
of statistical significance when we adjusted for smoking (four
Marital status (% married) 79 76 0.52 categories), BMI (continuous variable, adding a quadratic term
Smoking (%) 0.001 to account for non-linearity), high blood pressure, high blood
Never 32 44 cholesterol, diabetes, leisure-time physical activity (METS-h/week,
Currently 40 23 continuous variable, adding a quadratic term), marital status,
Ex-smoker (,3 years) 12 6 occupation and educational level (four categories). Further
Ex-smoker (>3 years) 17 26 adjustment for other nutrients (saturated fat, trans fat and
Body mass index (kg/m2, mean) 27.7 27.3 0.37 total fibre intake as continuous variables) led to statistically
History of hypertension (%) 42 30 0.02 significant results with OR = 0.26 (95% CI : 0.08–0.85) for the
History of diabetes (%) 16 8 0.01 upper quintile and P = 0.02 for the linear trend test.
High blood cholesterol in last 5 years (%) 19 11 0.04
We also fitted conditional logistic regression models using
energy-adjusted intake of olive oil as the exposure variable
Leisure-time physical activity
(METSc-h/week, mean) 31.5 34.5 0.32 (Table 4). The risk reduction was then more apparent. We found
Total energy intake (kcal/day, mean) 2631 2578 0.55 point-estimates lower than 1 for the OR in the four upper
quintiles of energy-adjusted olive oil intake and a significant
% energy from fat (mean) 31.2 31.3 0.90
linear trend test either when we adjusted only for non-dietary
MUFA/SFAd intake (mean) 1.51 1.54 0.51
confounders (P = 0.03) or also for relevant nutrients (P = 0.03).
% energy from trans-fatty acids (mean) 0.17 0.19 0.97
The relative reduction in the risk of a first myocardial infarction
Total ethanol intake (g/day, mean) 19.0 18.2 0.75
was greater than 75% for the upper quintile, OR = 0.22 (95%
Glycaemic load (g/day, mean) 233 225 0.42 CI : 0.07–0.67) after adjusting for non-dietary confounders and
Vitamin B6 intake (mg/day, mean) 2.8 2.8 0.93 OR = 0.18 (95% CI : 0.05–0.63) after adjustment for dietary
Vitamin C intake (mg/day, mean) 268 257 0.53 and non-dietary confounders.
Vitamin E intake (mg/day, mean) 8.0 7.7 0.54 When we excluded diabetic subjects (28 cases and 13 con-
Folic acid intake (microg/day, mean) 419 428 0.67 trols) and fitted a multivariate unconditional logistic regression
a Means were compared using t-tests. Proportions were compared using model (adjusting for age and gender in addition to the variables
χ2 tests (or a linear trend test for educational level). shown in the first footnote of Table 4) the odds ratio for the
b Age and gender were matching variables. upper quintile of energy-adjusted olive oil intake was 0.45
c Metabolic equivalents. (95% CI : 0.23–1.00, P , 0.05). When we excluded case and
d Monounsaturated fatty acids/saturated fatty acids. control subjects with previous history of high blood cholesterol
474_N01-268 8/4/2002 4:07 pm Page 477 (Black plate)

OLIVE OIL CONSUMPTION AND MYOCARDIAL INFARCTION 477

Table 2 Distribution of potential confounding variables across quintiles of energy-adjusted olive oil intake among control subjects (n = 171)

Quintiles of energy-adjusted olive oil intake

1 2–4 5 P for trend
Energy-adjusted olive oil (g/day, mean) 6.3 22.5 54.1 –
% white collar 18 28 24 0.91
% educational level higher than primary 18 30 21 0.75
% married 56 80 85 0.03
% smokers 18 21 35 0.06
% high blood cholesterol 15 12 6 0.24
% high blood pressure 29 33 21 0.25
% diabetes 6 5 18 0.02
Body mass index (kg/m2, mean) 26.8 27.3 27.8 0.30
Ethanol intake (g/day, mean) 16.0 16.1 27.2 0.02
Leisure-time physical activity (METSa-h/week, mean) 30.5 36.3 33.8 0.90
Total energy intake (kcal/day, mean) 2882 2417 2778 0.49
% energy from fat (mean) 28.7 30.6 35.8 ,0.001
% energy from saturated fat (mean) 10.8 10.3 10.3 0.58
% energy from monounsaturated fat (mean) 12.4 15.1 20.4 ,0.001
MUFA/SFAb intake (mean) 1.18 1.50 2.02 ,0.001
% energy from trans-fatty acids (mean) 0.23 0.19 0.14 0.11
Glycaemic load (g/day, mean) 235 231 207 0.34
Total fibre intake (g/day, mean) 37.6 30.2 29.0 0.04
Folic acid intake (microg/day, mean) 513 409 395 0.03
Vitamin B6 intake (mg/day, mean) 3.3 2.6 2.5 0.02
Vitamin C intake (mg/day, mean) 320 262 233 0.06
Vitamin E intake (mg/day, mean) 8.7 7.4 7.5 0.54
a Metabolic equivalents.
b Monounsaturated fatty acids/saturated fatty acids.

Table 3 Odds ratios (OR) (95% CI) of a first myocardial infarction according to olive oil intake (unadjusted for total energy intake)

Controls/cases Median intake Multivariate adjusted ORa Multivariate adjusted ORb

Quintile (n) (g/day) (95% CI) (95% CI)
1 32/36 7.2 1 (ref.) 1 (ref.)
2 35/37 12.0 1.17 (0.46–3.02) 1.16 (0.46–2.95)
3 36/30 25.0 0.69 (0.28–1.67) 0.60 (0.24–1.49)
4 31/39 29.3 0.91 (0.38–2.18) 0.83 (0.34–2.01)
5 37/29 54.3 0.36 (0.12–1.08) 0.26 (0.08–0.85)
Trend test P-value 0.05 0.02
a Conditional logistic regression (age-, hospital- and gender-matched pairs), adjusted for smoking, body mass index, high blood pressure, high blood cholesterol,
diabetes, leisure-time physical activity (METS-hours/week), marital status, occupation and study level.
b Additionally adjusted for saturated fat, trans fat and total fibre intake.

(total cholesterol .240 mg/dl, 21 cases and 15 controls), the
adjusted OR for the highest quintile of energy-adjusted olive oil
intake was 0.12 (95% CI : 0.03–0.54).
Discussion
To our knowledge this is the first analytical epidemiological
study finding direct evidence supporting an important protect-
ive effect of olive oil against a first AMI. The protective role for
olive oil is firmly consistent with many international comparisons
of CHD mortality and incidence rates. Very low rates of CHD
have usually been found in countries where olive oil consump-
tion is higher.21 Our results can be also explained according
to several plausible biological mechanisms. In comparison with
saturated fatty acids, olive oil reduces low-density lipoprotein
(LDL) cholesterol,22 and compared with carbohydrates, it main-
tains or even increases the levels of high-density lipoprotein
(HDL) cholesterol.23 In addition, it is relatively resistant to
oxidation and contains a large amount of antioxidants relative
to its polyunsaturated fat content.24 Some polyphenol constitu-
ents of olive oil (hydroxytirosol and oleuropein) are potent
scavengers of superoxide radicals25 and inhibit LDL oxidation.26
Olive oil has induced a regression of atherosclerosis in animal
models27 and may slow the development of coronary athero-
sclerosis, being associated with a reduced DNA synthesis in
human coronary smooth muscle cells.28 A recent trial showed
474_N01-268 8/4/2002 4:07 pm Page 478 (Black plate)
478 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 4 Odds ratios (OR) (95% CI) of a first myocardial infarction according to energy-adjusted olive oil intake
Controls/cases Median intake
Quintile (n) (g/day)
1 28/40 6.1
2 38/31 13.6
3 38/30 21.0
4 29/40 30.9
5 38/30 52.2
Trend test P-value
a Conditional logistic regression (age-, hospital- and gender-matched pairs), adjusted for smoking, body mass index, high blood pressure, high blood cholesterol,
diabetes, leisure-time physical activity (METS-hours/week), marital status, occupation and study level.
b Additionally adjusted for % energy derived from saturated fat, % energy derived from trans fat, total fibre consumption, folic acid intake, vitamin C intake,
glycaemic load and ethanol intake (adding a quadratic term to account for non-linearity).
that olive oil significantly improved endothelial function in
hypercholesterolaemic men.29
Olive oil also favourably affects postprandial factor VII
activity, avoiding a prolonged thrombotic response to a high-fat
diet.30,31 A beneficial effect of MUFA on the endothelium and
von Willebrand factor,32 as well as other benefits of olive oil on
the haemostatic system33 have been recently suggested. In dia-
betic patients, olive oil improves the lipid profile34 and glycaemic
control.35 The potential cardiovascular benefits purportedly
attributed to olive oil in diabetics led us to include them as
eligible either as cases or controls in our study.
A randomized trial showed that olive oil markedly lowered
blood pressure and reduced daily antihypertensive dosage
requirement among hypertensive subjects.36 Therefore, many
roads may lead from olive oil intake to a lower risk of a first
AMI, increasing the likelihood of a causal association.
In addition to the contrast of our results with the negative
findings reported from previous studies,5–7 some other incon-
sistencies must also be acknowledged. For example, the
decreasing CHD mortality and the decreasing average per capita
consumption of olive oil in Spain (and other Mediterranean
countries) during the last three decades show secular trends
that are in sharp contrast with the protective hypothesis for olive
oil that our data support.37 However, mortality data depend
not only on incidence rates, but also on medical care of CHD
patients, which is very likely to have improved during these
decades in Spain and other Mediterranean countries.
Several limitations of our study which might be alternative
explanations to our findings must be acknowledged. Our sample
size was not very large. At the design stage, the study size was
calculated assuming an alpha error = 0.05, and 80% power
to detect OR >2.0 with a 20% probability of exposure among
controls and that the exposure correlation between case and
control subjects would be less than 0.45.38 If this correlation
were 0.3 (as was the actual case in our study), the power would
be 90%. Subsequently, we have used the comparison among
extreme quintiles, thus reducing the statistical power. We chose
to focus on this comparison between extreme quintiles because
it is very unlikely that it would lead to misclassification due to
measurement error.39 Although the magnitude of the relative
effect was large (relative risk reduction .75%), we were
approaching the limits of statistical significance. However, after
adjustment for total energy intake and several known con-
founders, the inverse association with olive oil intake was still
apparent.
Multivariate adjusted ORa Multivariate adjusted ORb
(95% CI) (95% CI)
1 (ref.) 1 (ref.)
0.39 (0.15–1.00) 0.45 (0.16–1.25)
0.40 (0.17–0.93) 0.44 (0.18–1.07)
0.59 (0.23–1.52) 0.70 (0.24–2.02)
0.22 (0.07–0.67) 0.18 (0.05–0.63)
0.03 0.03
Our design is susceptible to being affected by recall bias.
However, recall bias is more likely to happen when differential
over-reporting exists in cases, because they may be more aware
of the greater risk associated with publicly known determinants
of disease. This is unlikely to occur when assessing the intake of
protective factors (olive oil in our study). Moreover, when the
assessment of exposure is done via different items in a com-
prehensive questionnaire, such as in this case, it would be more
difficult for a patient to consistently underestimate his/her ex-
posure to olive oil. Although the in-hospital selection of controls
facilitates higher participation, it also imposes some caution in
the interpretation of findings because the exposure may be
related to the diseases causing the hospital admission of con-
trols. Olive oil has not been found to induce any trauma or
genitourinary disease or any common disease needing minor
surgery. Thus, it is very unlikely that our findings could be
alternatively explained by selection bias due to an association
between olive oil exposure and a higher probability of being
admitted to a hospital with these diagnoses.
Although the mean of energy expenditure was slightly higher
among controls than cases, we found no statistically significant
association of energy expenditure (METS-hours/weeks) with
the risk of myocardial infarction (P = 0.21 for the linear trend
test). In previous epidemiological studies a higher caloric
intake was usually protective. One of the most consistent
findings of prospective studies about diet and CHD is a strong
inverse association between total energy intake and CHD.40
However, as some of our control patients were bedridden and
had limited physical activity, it is not surprising that we missed
this association.
An inherent limitation of a case-control design is that some
fatal cases may be lost for the case series if they are fatal early
on. Thus, the observed results would be compatible with olive
oil having no effect on risk of myocardial infarction but an
effect on chance of survival. However, this potential bias would
require impossible assumptions to explain our results. For ex-
ample, an incredibly high adverse effect of olive oil on survival,
in the range of an early mortality OR .100 (fifth versus first
quintile) would be needed to explain the OR found for the fifth
quintile. Therefore, the alternative explanation that the observed
inverse association with non-fatal infarction was due to poorer
survival caused by olive oil seems extremely unlikely.
The possibility that people with first myocardial infarction
may have been aware of early symptoms and have made more
recent changes to their diet was explicitly addressed in the
474_N01-268 8/4/2002 4:07 pm Page 479 (Black plate)

OLIVE OIL CONSUMPTION AND MYOCARDIAL INFARCTION 479

design stage. To avoid it, we made the decision to exclude cases
with previous symptoms (angina or other cardiovascular mani-
festations) and used physicians as interviewers to carefully screen
case patients and exclude those with any previous symptom
predating the infarction. We also interviewed patients as soon as
possible after the acute event to prevent this bias. Moreover,
this bias would tend to diminish the association if people with
myocardial infarction had recently increased their consumption of
olive oil. The precautions we took to prevent this bias might help
to explain the divergence of our results with two previous case-
control studies conducted in Mediterranean countries which found
no association for MUFA intake5 or for oils consumption.7
Another potential concern is represented by the fact that the
physicians who carried out the interviews were not blind to
the case-control status of the patient. However, this fact is very
unlikely to have caused a serious bias in our estimates of the
effect of olive oil consumption on AMI risk because the food-
frequency questionnaire was self-administered. The interviewing
physician did not ask any questions of the patient regarding the
consumption of olive oil, but simply provided him/her with
the self-administered questionnaire, and the patient completed
the questionnaire by himself/herself.
Food-frequency questionnaires have become the primary
method for measuring dietary intake in epidemiological
studies.41 Although the food-frequency questionnaire used in
this study was specifically validated in Spain against dietary
records,14 and dietary records are likely to have the least cor-
related errors,39 we acknowledge that this instrument might
have misclassified participants at some level regarding their olive
oil consumption. Because olive oil is widely used in cooking and
on foods, it may be difficult to quantify accurately. A trained
dietitian (CF) with special expertise in nutritional epidemiology
reviewed the major available sources of information about oil
use in Spain in order to make the calculations and derive the
total intake of olive oil from the answers to the food-frequency
questionnaires. However, measurement error always exists
in nutritional epidemiology and it usually introduces non-
differential exposure misclassification.42 It is commonly
believed that non-differential misclassification of a exposure
predictably biases the OR towards the null value. Nevertheless,
sometimes, when the exposure has more than two categories,
this bias can be away from the null value.43 This possibility
might result in an alternative explanation to our results, but
would be unusual. Moreover, it has been shown that measure-
ment error is very unlike to misclassify individuals from one
extreme quintile into the other extreme39 because much of
the measurement error probably concentrates in the middle
categories.42 The strongest association we found was precisely
for the comparison between extreme quintiles.
The degree of between-subject variability in the intake of a
particular nutrient or food item in the population under study
is a strong determinant of the ability of a study to detect an
association between that nutrient and CHD. A higher level of
consumption of a food item is usually associated with higher
between-subject variability. This has been the case in our
findings with a wide contrast of consumption between extreme
quintiles (medians = 7 versus 54 g/day). Therefore, the very
high levels and the heterogeneity of olive oil consumption
found in Spain are advantages of our study and may explain why
we have found such a strong association. However, because of
the inconsistency with previous studies (and the inherent limita-
tions of our design), further epidemiological studies, preferably
following a cohort instead of a case-control design, and also trials
are needed in Mediterranean countries to confirm our findings.
Acknowledgement
Partially funded by the Department of Health (Navarre Regional
Government Project 24/99) and by another grant from Banco
Santander-Central-Hispano. We are specially indebted to Ms
Carmen de la Fuente, our dietitian, who updated the food com-
position data-bank according to current Spanish food composition
tables and worked in the calculation of nutrients taken into
account as potential confounders in our analyses. We thank the
following persons for technical assistance and support: Prof.
Jokin De Irala-Estévez, Prof. J Alfredo Martínez, Dr Isabel Coma
MD, Ms Almudena Sánchez-Villegas, Ms Jane Hoashi and Ms
Estefanía Ruiz-Gaona. We thank the Cardiology Chairmen of
the three hospitals that participated in this study: Hospital de
Navarra (Dr Enrique de los Arcos, MD), Hospital Virgen del
Camino (Dr Eugenio Torrano, MD), and University Clinic of
Navarre (Dr Joaquín Barba, MD).

KEY MESSAGES
• A wide variety of physiological reasons support that olive oil may exert a beneficial effect on the risk of coronary
disease.
• International comparisons, with the pioneering results of the Seven Countries Study, also support this hypothesis.
• In this hospital-based matched case-control study, conducted in Navarre (Spain), the relative risk reduction for
myocardial infarction was greater than 75% for participants in the highest quintile of olive oil consumption
(median = 54 g/day) versus the first quintile (median = 7 g/day), after multivariate adjustment for a wide array
of dietary and non-dietary confounders.
• Some inconsistencies remain: previous case-control studies in Italy and Greece did not find any association, and
a small randomized trial on coronary patients conducted by Rose almost 40 years ago was also negative for olive
oil, and even harmful for corn oil.
• There is a need for large, prospective cohort studies conducted in Mediterranean countries where high and
heterogeneous consumption of olive oil exists.
474_N01-268 8/4/2002 4:07 pm Page 480 (Black plate)
480 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
References
1 Willett WC, Sacks F, Trichopoulou A et al. Mediterranean diet
pyramid: a cultural model for healthy eating. Am J Clin Nutr 1995;
61(Suppl.):1402S–06S.
2 Ascherio A, Katan MB, Zock PL, Stampfer MJ, Willett WC. Trans fatty
acids and coronary heart disease. N Engl J Med 1999;340:1994–98.
3 Ascherio A, Hennekens CH, Buring JE, Master C, Stampfer MJ,
Willett WC. Trans-fatty acids intake and risk of myocardial infarction.
Circulation 1994;89:94–101.
4 Menotti A, Keys A, Aravanis C et al. Seven Countries Study. First 20-
year mortality data in 12 cohorts of six countries. Ann Med 1989;
21:175–79.
5 Rose GA, Thomson WB, Williams RT. Corn oil in treatment of
Ischaemic Heart Disease. BMJ 1965;i:1531–33.
6 Gramenzi A, Gentile A, Fasoli M, Negri E, Parazzini F, La Vecchia C.
Association between certain foods and risk of acute myocardial infarc-
tion in women. BMJ 1990;300:771–73.
7 Tzonou A, Kalandidi A, Trichopoulou A et al. Diet and coronary
heart disease: a case-control study in Athens, Greece. Epidemiology
1993;4:511–16.
8 Trichopoulou A, Kouris-Blazos A, Wahlqvist ML et al. Diet and overall
survival in elderly people. BMJ 1995;311:1457–60.
9 Kouris-Blazos A, Gnardellis C, Wahlqvist ML, Trichopoulos D, Lukito
W, Trichopoulou A. Are the advantages of the Mediterranean diet
transferable to other populations? A cohort study in Melbourne,
Australia. Br J Nutr 1999;82:57–61.
10 Lasheras C, Fernandez S, Patterson AM. Mediterranean diet and
age with respect to overall survival in institutionalized, nonsmoking
elderly people. Am J Clin Nutr 2000;71:987–92.
11 Fortes C, Forastiere F, Farchi S, Rapiti E, Pastori G, Perucci CA. Diet
and overall survival in a cohort of very elderly people. Epidemiology
2000;11:440–45.
12 De Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N.
Mediterranean diet, traditional risk factors and the rate of cardio-
vascular complications after myocardial infarction; final report of the
Lyon Diet Heart Study. Circulation 1999;99:779–85.
13 WHO MONICA Project. MONICA Manual. Geneva: WHO, 1990, Part
IV(a), pp.11–32.
14 Martin-Moreno JM, Boyle P, Gorgojo L et al. Development and
validation of a food frequency questionnaire in Spain. Int J Epidemiol
1993;22:512–19.
15 Willett W, Stampfer MJ. Implications of total energy intake for
epidemiologic analysis. In: Willett W (ed.). Nutritional Epidemiology,
2nd Edn. New York: Oxford University Press, 1998, pp.273–301.
16 Paffenbarger RS, Wing AL, Hyde RT. Physical activity as an index of
heart attack risk in college alumni. Am J Epidemiol 1978;108:161–75.
17 Rimm EB, Ascherio A, Giovannucci E, Spiegelman D, Stampfer MJ,
Willett WC. Vegetable, fruit, and cereal fiber intake and risk of coronary
heart disease among men. JAMA 1996;275:447–51.
18 Rimm EB, Willett WC, Hu FB et al. Folate and vitamin B6 from diet
and supplements in relation to risk of coronary heart disease among
women. JAMA 1998;279:359–64.
19 Khaw KT, Bingham S, Welch A et al. Relation between plasma
ascorbic acid and mortality in men and women in EPIC-Norfolk pros-
pective study: a prospective population study. European Prospective
Investigation into Cancer and Nutrition. Lancet 2001;357:657–63.
20 Liu S, Willett WC, Stampfer MJ et al. A prospective study of dietary
glycemic load, carbohydrate intake, and risk of coronary heart disease
in US women. Am J Clin Nutr 2000;71:1455–61.
21 Tunstall-Pedoe H, Kuulasmaa K, Mahonen M, Tolonen H, Ruokokoski
E, Amouyel P. Contribution of trends in survival and coronary-event
rates to changes in coronary heart disease mortality: 10-year result
from 37 WHO MONICA project populations. Monitoring trends and
determinants in cardiovascular disease. Lancet 1999;353:1547–57.
22 Roche HM, Zampelas A, Knapper JM et al. Effect of long-term olive oil
dietary intervention on postprandial triacylglycerol and factor VII
metabolism. Am J Clin Nutr 1998;68:552–60.
23 Katan MB, Grundy SM, Willett WC. Beyond low-fat diets. N Engl J
Med 1997;337:563–66.
24 Visioli F, Bellomo G, Montedoro G, Galli C. Low density lipoprotein
oxidation is inhibited in vitro by olive oil constituents. Atherosclerosis
1995;117:25–32.
25 Visioli F, Bellomo G, Galli C. Free radical-scavenging properties of
olive oil polyphenols. Biochem Biophys Res Commun 1998;247:60–64.
26 Fito M, Covas MI, Lamuela-Raventós RM et al. Protective effect of
olive oil and its phenolic compounds against low density lipoprotein
oxidation. Lipids 2000;35:633–38.
27 Mangiapane EH, McAteer MA, Benson GM, White DA, Salter AM. Modu-
lation of the regression of atherosclerosis in the hamster by dietary lipids:
comparison of coconut oil and olive oil. Br J Nutr 1999;82:401–09.
28 Mata P, Varela O, Alonso R, Lahoz C, de Oya M, Badimon L.
Monounsaturated and polyunsaturated n-6 fatty acid-enriched diets
modify LDL oxidation and decrease human coronary smooth muscle
cell DNA synthesis. Arterioscler Thromb Vasc Biol 1997;17:2088–95.
29 Fuentes F, López-Miranda J, Sánchez E et al. Mediterranean and low-
fat diets improve endothelial function in hypercholesterolemic men.
Ann Intern Med 2001;134:1115–19.
30 Roche HM, Gibney MJ. Postprandial coagulation factor VII activity:
the effect of monounsaturated fatty acids. Br J Nutr 1997;77:537–49.
31 Larsen LF, Jespersen J, Marckmann P. Are olive oil diets antithrom-
botic? Diets enriched with olive, rapeseed, or sunflower oil affect
postprandial factor VII differently. Am J Clin Nutr 1999;70:976–82.
32 Thomsen C, Rasmussen OW, Ingerslev J, Hermansen K. Plasma levels
of von Willebrand factor in non-insulin-dependent diabetes mellitus
are influenced by dietary monounsaturated fatty acids. Thromb Res
1995;77:347–56.
33 Junker R, Kratz M, Neufeld M et al. Effects of diets containing olive
oil, sunflower oil, or rapeseed oil on the hemostatic system. Thromb
Haemost 2001;85:280–86.
34 Madigan C, Ryan M, Owens D, Collins P, Tomkin GH. Dietary
unsaturated fatty acids in type 2 diabetes: higher levels of postprandial
lipoprotein on a linoleic acid-rich sunflower oil diet compared with an
oleic acid-rich olive oil diet. Diabetes Care 2000;23:1472–77.
35 Garg A. High-monounsaturated-fat diets for patients with diabetes
mellitus: a meta-analysis. Am J Clin Nutr 1998;67:577S–82S.
36 Ferrara LA, Raimondi AS, d’Episcopo L, Guida L, Dello Russo A,
Marotta T. Olive oil and reduced need for antihypertensive medica-
tions. Arch Intern Med 2000;160:837–42.
37 Serra-Majem L, Ribas L, Tresserras R, Ngo J, Salleras L. How could
changes in diet explain changes in coronary heart disease mortality in
Spain? The Spanish paradox. Am J Clin Nutr 1995;61:1351S–59S.
38 Arcus QuickStat software. London: Addison Wesley Longman, Ltd, 1997.
39 Willett W, Lenart E. Reproducibility and validity of food-frequency
questionnaires. In: Willett W (ed.). Nutritional Epidemiology, 2nd Edn.
New York: Oxford University Press, 1998, pp.101–47.
40 Willett W. Diet and coronary heart disease. In: Willett W (ed.).
Nutritional Epidemiology. 2nd Edn. New York: Oxford University Press,
1998, pp.414–66.
41 Willett W. Food-frequency methods. In: Willett W (ed.). Nutritional Epi-
demiology. 2nd Edn. New York: Oxford University Press, 1998, pp.74–100.
42 Michels KB. A renaissance for measurement error. Int J Epidemiol
2001;30:421–22.
43 Rothman KJ, Greenland S. Precision and validity in epidemiologic
studies. In: Rothman KJ, Greenland S (eds). Modern Epidemiology. 2nd
Edn. Philadelphia: Lippincott-Raven, pp.115–34.
474_N01-268 8/4/2002 4:07 pm Page 481 (Black plate)
© International Epidemiological Association 2002 Printed in Great Britain
Commentary: Is olive oil a key ingredient
in the Mediterranean recipe for health?
AR Ness
There is no single Mediterranean diet. The traditional diets of
the countries surrounding the Mediterranean, however, share a
number of common features. They are (or were) plant-based
with olive oil as the principal fat.1 These diets appear to be health-
ful. The Seven Countries study showed that Mediterranean
populations studied in the 1960s and 1970s had low rates of
death attributed to coronary heart disease and low rates of death
from all causes in middle age.2 Furthermore, the favourable
health experience of Greek migrants to Australia and the
declines in death rates attributed to coronary disease in many
Mediterranean countries in the face of a less traditional diet
suggest that it may even be possible to improve on this tried and
tested cultural recipe for health.3–5
The challenge is to tease out (if possible) the healthful feat-
ures of the Mediterranean diets.6 This is important for several
reasons. First, while some of us have sufficient resources to eat
more Mediterranean diets there are concerns that many who
are less privileged do not.7 Second, the widespread adoption of
Mediterranean diets (and how far this can be altered to take
account of what can be grown locally) has implications for agri-
culture8 and for the environment.9 Thus, to inform individual
choices and policy formulation, it is important to try and unpick
the Mediterranean advantage.
The study by Fernández-Jarne and colleagues represents
an attempt to do just this.10 They report a hospital-based case-
control study of 171 cases of first acute myocardial infarction
and age- and sex-matched controls they carried out in a Spanish
population to examine the association with olive oil intake.
There was marked heterogeneity of olive oil consumption
(assessed by food frequency questionnaire) in this population—
54 g per day in the top quintile versus 7 g per day in the bottom
quintile in the controls. While those controls in the top quintile
for olive oil consumption were more likely to be married, to
eat a more plant-based diet and consume more alcohol they
were also more likely to smoke and to be diabetic. After adjust-
ment for a number of confounders the odds ratio in the top
quintile was 0.18 (95% CI : 0.05–0.63). Though the test for
trend was statistically significant (P = 0.03) there was no clear
dose response. This study was small and these findings are not
consistent with the results of other studies in Mediterranean
populations. A case-control study in Italian women found no
association with oil consumption11 and a case-control study in
Greece found no association with monounsaturated fat intake.12
Nevertheless, it does raise the possibility that olive oil is a par-
ticularly important component of the Mediterranean diet. While
cohort studies such as the European Prospective Investigation
into Cancer (EPIC)13 may be able to confirm whether this
University of Bristol, Department of Social Medicine, Canynge Hall,
Whiteladies Road, Bristol BS8 2PR, UK.
481
International Journal of Epidemiology 2002;31:481–482
observational association exists in prospectively collected data,
ultimately trials will be required to establish whether an import-
ant causal relationship exists.
Though there have been no large trials of olive oil supple-
mentation specifically there are several other trials that have
tested various aspects of Mediterranean diets. One secondary
prevention trial of advice to eat a more Mediterranean diet was
carried out in France.14 In this trial 605 middle-aged people
with a recent myocardial infarction were randomized to receive
advice to eat a Mediterranean diet (more bread, more vegetables,
more fruit, more fish, and less meat), and to replace butter and
cream with rapeseed margarine. After 27 months the trial was
stopped prematurely because of better outcomes in the inter-
vention group. There were 20 deaths in the control group and
8 in the intervention group. The adjusted risk of death from all
causes was 0.30 (95% CI : 0.11–0.82). These findings have not
been replicated.
Traditional Mediterranean diets contain modest amounts
of fish.1 Two large trials of advice to eat more fish or fish oil
supplements (and thus more n-3 fatty acids)15,16 have reported
a reduced risk of all-cause mortality in people with existing
coronary heart disease. The pooled rate ratio for dietary fish or
fish oil on total mortality based largely on the results of these
two trials was 0.83 (95% CI : 0.73–0.94, with no significant hetero-
geneity).17 A long-term follow-up of the Diet and Re-infarction
Trial (DART1), however, failed to show any substantial long-
term survival benefit.18 Furthermore, the results of a large
randomized trial of fish advice in men with angina, the Diet and
Angina Randomised Trial (DART2), are not consistent with the
results from these previous trials.19
The Mediterranean diets being plant-based are rich in anti-
oxidants.6 The results of trials of antioxidant supplementation
(such as β-carotene, vitamin E and vitamin C) have, however,
been disappointingly null.20–23
In conclusion, a more profitable epidemiological approach to
exploring the Mediterranean advantage may be to move away
from constituent-based analyses and experiments and to carry out
food-based analyses (such as the one reported by Fernández-
Jarne and colleagues) in the first instance and to follow these up
with trials of whole food modification or dietary advice.
References
1 Nestle M. Mediterranean diets: science and policy implications. Am J
Clin Nutr 1995;61:1313S–427S.
2 Keys A, Menotti A, Karvonen MJ et al. The diet and 15-year death rate
in the seven countries study. Am J Epidemiol 1986;124:903–15.
3 Powles JW. Greek migrants in Australia: surviving well and helping their
hosts. In: Marks L, Worboys M (eds). Migrants, Minorities and Medicine:
Historical and Contemporary Perspectives. London: Routledge, 1994.
474_N01-268 8/4/2002 4:07 pm Page 482 (Black plate)
482 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
4 Powles JW, Day NE, Sanz MA, Bingham SA. Protective foods in winter
and spring: a key to lower vascular mortality? Lancet 1996;348:898–99.
5 Powles J. Commentary: Mediterranean paradoxes continue to
provoke. Int J Epidemiol 2001;30:1076–77.
6 James WPT, Duthie GG, Wahle KWJ. The Mediterranean diet:
protective or simply non-toxic? Eur J Clin Nutr 1989;43:31–41.
7 Morris JN, Donkin AJ, Wonderling D, Wilkinson P, Dowler EA. A
minimum income for healthy living. J Epidemiol Community Health
2000;54:885–89.
8 O’Brien P. Dietary shifts and implications for US agriculture. Am J Clin
Nutr 1995;61:1390S–96S.
9 Gussow JD. Mediterranean diets: are they environmentally responsible?
Am J Clin Nutr 1995;61:1383S–89S.
10 Fernández-Jarne E, Martinez-Losa E, Prado-Santamaria M et al. Risk
of first non-fatal myocardial infarction negatively associated with
olive oil consumption: a case-control study in Spain. Int J Epidemiol
2002;31:474–480.
11 Gramenzi A, Gentile A, Fasoli M, Negri E, Parazzini F, La Vecchia C.
Association between certain foods and risk of acute myocardial
infarction in women. BMJ 1990;300:771–73.
12 Tzonou A, Kalandidi A, Trichopolou A et al. Diet and coronary heart
disease: a case-control study in Athens, Greece. Epidemiology 1993;
4:511–16.
13 Riboli E. Nutrition and cancer: background and rationale of the
European prospective investigation into cancer and nutrition (EPIC).
Ann Oncol 1992;3:783–91.
The bark of a 600 year old olive tree. Photography: Mary Shaw
14 de Lorgeril M, Renaud S, Mamelle N et al. Mediterranean alpha-
linolenic acid-rich diet in secondary prevention of coronary heart
disease. Lancet 1994;343:1454–59.
15 Burr ML, Fehily AM, Gilbert JF et al. Effects of changes in fat, fish,
and fibre intakes on death and myocardial reinfarction: diet and
reinfarction trial (DART). Lancet 1989;ii:757–61.
16 GISSI-Prevenzione I. Dietary supplementation with n-3 poly-
unsaturated fatty acids and vitamin E after myocardial infarction:
results of the GISSI-Prevenzione trial. Lancet 1999;354:447–55.
17 Hooper L, Ness AR, Higgin JPT, Moore T, Ebrahim SB. GISSI-
Prevenzione trial. Lancet 1999;354:1557.
18 Ness AR, Hughes J, Elwood PC, Whitley E, Smith GD, Burr ML. The
long-term effect of dietary advice in men with coronary disease:
follow-up of the Diet and Reinfarction trial (DART). Eur J Clin Nutr
2002 (in press).
19 Burr ML, Ashfield-Watt PAL, Dunstan FDJ et al. Lack of benefit of
dietary advice to men with angina: results of a controlled trial. Eur J
Clin Nutr (Submitted).
20 Egger M, Schneider M, Davey Smith G. Spurious precision? Meta-
analysis of observational studies. BMJ 1998;316:140–44.
21 Hooper L, Ness AR, Smith GD. Antioxidant strategy for cardiovascular
diseases. Lancet 2001;357:1705.
22 Ness AR, Egger M, Davey Smith G. Meta-analysis seems to exclude
benefit of vitamin C supplementation. BMJ 1999;319:577.
23 Kmietowicz Z. Statins are the new aspirin, Oxford researchers say.
BMJ 2001;323:1145.