Ioi70171 2461 2468

ORIGINAL INVESTIGATION
Mediterranean Dietary Pattern and Prediction

of All-Cause Mortality in a US Population
Results From the NIH-AARP Diet and Health Study
Panagiota N. Mitrou, PhD; Victor Kipnis, PhD; Anne C. M. Thiébaut, PhD; Jill Reedy, PhD;
Amy F. Subar, PhD; Elisabet Wirfält, PhD; Andrew Flood, PhD; Traci Mouw, MPH;
Albert R. Hollenbeck, PhD; Michael F. Leitzmann, MD, DrPH; Arthur Schatzkin, MD, DrPH
Background: The Mediterranean diet has been sug- Results: The Mediterranean diet was associated with re-
gested to play a beneficial role for health and longevity. duced all-cause and cause-specific mortality. In men, the
However, to our knowledge, no prospective US study has multivariate HRs comparing high to low conformity for
investigated the Mediterranean dietary pattern in rela- all-cause, CVD, and cancer mortality were 0.79 (95% CI,
tion to mortality. 0.76-0.83), 0.78 (95% CI, 0.69-0.87), and 0.83 (95% CI,
0.76-0.91), respectively. In women, an inverse associa-
Methods: Study participants included 214 284 men and tion was seen with high conformity with this pattern: de-
166 012 women in the National Institutes of Health creased risks that ranged from 12% for cancer mortality
(NIH)-AARP (formerly known as the American Associa- to 20% for all-cause mortality (P =.04 and P ⬍.001, re-
tion of Retired Persons) Diet and Health Study. During spectively, for the trend). When we restricted our analy-
follow-up for all-cause mortality (1995-2005), 27 799 ses to never smokers, associations were virtually un-
deaths were documented. In the first 5 years of follow-
changed.
up, 5985 cancer deaths and 3451 cardiovascular disease
(CVD) deaths were reported. We used a 9-point score
Conclusion: These results provide strong evidence for a
to assess conformity with the Mediterranean dietary
pattern (components included vegetables, legumes, beneficial effect of higher conformity with the Mediterra-
fruits, nuts, whole grains, fish, monounsaturated fat– nean dietary pattern on risk of death from all causes, includ-
saturated fat ratio, alcohol, and meat). We calculated haz- ing deaths due to CVD and cancer, in a US population.
ard ratios (HRs) and 95% confidence intervals (CIs) using
age- and multivariate-adjusted Cox models. Arch Intern Med. 2007;167(22):2461-2468
T
HE M EDITERRANEAN DIET the cohort allowed us to address poten-
may play a beneficial role in tial smoking status confounding by re-
Author Affiliations: Nutritional
health and longevity.1 Its key stricting the analysis to participants who
Epidemiology Branch, Division components have been op- had never smoked and, in addition, to
of Cancer Epidemiology and erationalized as a dietary pat- evaluate potential effect modification by
Genetics (Drs Mitrou, Thiébaut, tern using a 9-point score.2-5 The Mediter- lifestyle factors such as smoking status and
Leitzmann, and Schatzkin and ranean dietary pattern has been associated body mass index (BMI).
Ms Mouw), Biometry Research with a decrease in overall mortality in a
Group, Division of Cancer number of fairly small studies,3,4,6-11 and re- METHODS
Prevention (Dr Kipnis), and cently in the European Prospective Inves-
Applied Research Branch,
tigation into Cancer and Nutrition (EPIC),
Division of Cancer Control and STUDY POPULATION
Population Sciences (Drs Reedy
a cohort study of over half a million people
and Subar), National Cancer from 10 countries across Europe.5 The NIH-AARP Diet and Health Study12 was
Institute, Bethesda, Maryland; To date, no US study has investigated begun in 1995-1996, when questionnaires elic-
Department of Clinical Sciences, the effect of the Mediterranean dietary pat- iting information on demographic and health-
Lund University, Malmö, tern on mortality. We examined the rela- related behaviors were mailed to 3.5 million
Sweden (Dr Wirfält); Division of tion of the Mediterranean dietary pattern AARP members aged 50 to 71 years who re-
Epidemiology and Community to all-cause and cause-specific mortality sided in 6 US states (California, Florida, Loui-
Health, University of Minnesota, in the National Institutes of Health (NIH)- siana, New Jersey, North Carolina, and Penn-
Minneapolis (Dr Flood); and sylvania) and 2 metropolitan areas (Atlanta,
AARP (formerly known as the American
AARP, Washington, DC Georgia; and Detroit, Michigan). Of the 617 119
(Dr Hollenbeck). Dr Mitrou is
Association of Retired Persons) Diet and persons who returned the questionnaires,
now with the Department of Health Study.12 We hypothesized that 566 407 respondents filled out the survey in sat-
Public Health and Primary Care, greater conformity with the Mediterra- isfactory detail and consented to participate in
University of Cambridge, nean dietary pattern is associated with a the study. Additional study details have been
Cambridge, England. reduction in mortality. The large size of described previously.12
(REPRINTED) ARCH INTERN MED/ VOL 167 (NO. 22), DEC 10/24, 2007 WWW.ARCHINTERNMED.COM
2461
©2007 American Medical Association. All rights reserved.
Downloaded From: https://jamanetwork.com/ on 05/30/2023

We excluded proxy respondents (n=15 760) and subjects centile for the components considered healthy and unhealthy,
with a self-reported history of cancer (n=51 207), end-stage re- respectively, while maintaining the cut points for alcohol.
nal disease (n=997), heart disease (n=68 663), stroke (n=6435),
emphysema (n = 9481), or diabetes (n = 30 181). We also ex- STATISTICAL ANALYSIS
cluded individuals who were outside 2 interquartile ranges of
the 25th to 75th percentile interval of the normalized distri- Associations between nonnutritional covariates and the Medi-
bution for energy intake (n=3387). The resulting cohort in- terranean diet scores were examined by cross-tabulations. The
cluded 380 296 participants (214 284 men and 166 012 wom- median, 25th, and 75th percentiles of intake of food groups and
en) with no history of chronic disease. macronutrients were determined for men and women sepa-
rately at baseline. Spearman rank correlation coefficients re-
COHORT FOLLOW-UP AND MORTALITY lating the scores to food group and macronutrient intakes were
ASCERTAINMENT calculated.
We used Cox proportional hazards regression,18 with person-
Addresses were updated annually by matching the cohort da- years of follow-up as the time variable, to assess the associa-
tabase to the National Change of Address maintained by the tion between aMED and tMED and all-cause mortality and mor-
US Postal Service.13 Follow-up time for all-cause mortality was tality from CVD, cancer, and other causes with follow-up
between 1995-1996 and December 31, 2005. Vital status was through 2001, separately for men and women. We verified that
ascertained by the Social Security Administration Death Mas- the proportional hazards assumption was not violated for our
ter File, which documented 27 799 deaths during 10 years of main exposure and other fixed covariates by including cross-
follow-up. Deaths occurring in the 1995-2001 period were linked product terms of the Mediterranean diet score and person-
to the National Death Index to ascertain underlying cause of years and testing that all coefficients equaled 0.
death. We investigated cause-specific mortality, including car- All multivariate models were adjusted for potential risk fac-
diovascular disease (CVD) (n=3451), cancer (n=5985), and tors for mortality, including age (continuous), total energy (con-
other-cause (n=2669) mortality using the Surveillance Epide- tinuous), smoking (never smokers, former smokers of ⱕ1 pack
miology and End Results coding system.14 The NIH-AARP Diet per day, former smokers of ⬎1 pack per day, current smokers
and Health Study12 was approved by the special studies insti- of ⱕ1 pack per day, and current smokers of ⬎1 pack per day),
tutional review board of the US National Cancer Institute. education (less than high school, high school, some college,
and completion of college and postgraduate college), BMI (cal-
culated as weight in kilograms divided by height in meters
EXPOSURE ASSESSMENT squared) (⬍18.5, 18.5 to ⬍25.0, 25.0 to ⬍30.0, 30.0 to ⬍35.0,
35.0 to ⬍40.0, and ⱖ40.0), physical activity (never, rarely, 1-3
Study participants completed an early version of the Diet His- times per month, 1-2 times per week, 3-4 times per week, and
tory Questionaire,15 a 124-item food frequency questionnaire ⱖ5 times per week), race (white, black, and other), marital sta-
(FFQ) that was validated against two 24-hour recalls.16 Pyra- tus (married and not currently married), and in women, meno-
mid servings for food groups were generated by linking the FFQ pausal hormone therapy (never, past, and current). For finer
data with the Pyramid Servings Database, version 2.0.17 The FFQ control of smoking status, we used a multilevel variable that
provided estimates for intake of alcohol, monounsaturated fat, integrated combinations of smoking status (never, former, and
polyunsaturated fat, and saturated fat (grams). To adjust for current), time since quitting for former smokers (ⱖ10 years,
energy intake, we divided daily intakes by an individual’s re- 5-9 years, 1-4 years, and within the last year), and smoking dose
ported total energy intake and multiplied by 2000 calories in for both former and current smokers (1-10, 11-20, 21-30, 31-
women and 2500 calories in men.7-9,11 Information about de- 40, 41-60, and ⱖ61 cigarettes per day). Additional adjustment
mographics, smoking, physical activity, and menopausal hor- for other dietary variables (quintiles of polyunsaturated fatty ac-
mone therapy were also reported at baseline. ids, eggs, and potatoes) did not appreciably change the risk es-
timates, and these nutritional factors were not considered fur-
MEDITERRANEAN DIET SCORE ther. In a subanalysis we excluded smokers to evaluate the
potential for residual confounding by smoking. Missing values
Conformity with the Mediterranean dietary pattern was as- for adjusting covariates were included as dummy variables in the
sessed by the traditional Mediterranean diet score (tMED), simi- models. Using the alternative Horvitz-Thompson method19 for
lar to that constructed by Trichopoulou et al,4,11 and the alter- missing data did not change the results appreciably.
nate Mediterranean diet score (aMED) created by Fung et al.2 Interactions between tMED and aMED and lifestyle factors
The tMED includes 9 components and takes values from 0 to 9 such as smoking and BMI were tested using cross-product terms
points (minimum to maximum conformity). One point each is of the Mediterranean diet score and lifestyle factor and likeli-
given for intake at or above the sex-specific median intake for hood ratio test statistics comparing models with and without
components considered to be healthy (vegetables [excluding po- a cross-product term. We further conducted stratified analy-
tatoes], fruit and nuts, legumes, grains, fish, and monounsatu- ses of all-cause mortality by examining the Mediterranean diet
rated fat–saturated fat ratio), and 1 point is given for intake less score in relation to mortality among 6 subgroups of smoking
than the median for those components considered to be un- status (never/ever) and BMI (18.5-25.0, ⬎25.0 to ⬍30.0, ⱖ30.0).
healthy (dairy and meat). In addition, 1 point is given for alco- Age-adjusted incidence rates were calculated20 with 5-year age
hol consumption within a specified range (5-25 g/d for women; bands standardized to the entire NIH-AARP population. All sta-
10-50 g/d for men). The aMED differs from the tMED in that tistical analyses were carried out using SAS statistical soft-
the aMED separates fruit and nuts into 2 groups, eliminates dairy, ware, version 9.1 (SAS Institute, Cary, North Carolina). A 2-sided
includes only whole grains, only red and processed meat (beef, P value of less than .05 was considered statistically significant.
pork, luncheon meats, and organ meats), and uses the same al-
cohol range for both men and women (5-25 g/d).
RESULTS
The aMED and tMED scores were divided into 3 groups: 0
to 3, 4 and 5, and 6 to 9 points. In sensitivity analyses, we con-
sidered different sex-specific cut points by assigning points to During 5 years of follow-up, 12 105 deaths were docu-
intakes at or above the 75th percentile or below the 25th per- mented (including 5985 cancer deaths and 3451 CVD
2462

Table 1. Distribution of Participants by aMED Score in Conjunction With Nonnutritional Variables a
aMED Score
Men Women
Characteristic 0-3 4-5 6-9 0-3 4-5 6-9

Age ⬎62 y b 46.94 50.31 52.89 47.35 49.35 51.07
BMI
⬍18.5 0.65 0.64 0.68 1.61 1.41 1.54
18.5 to ⬍25.0 26.44 28.62 35.04 39.28 43.73 51.62
25.0 to ⬍30.0 49.05 49.98 48.26 31.94 32.31 30.03
30.0 to ⬍35.0 17.13 15.13 11.93 14.90 13.04 10.15
35.0 to ⬍40.0 3.71 2.96 1.98 5.69 4.38 2.76
ⱖ40.0 1.08 0.88 0.53 3.22 2.23 1.22
Race
White 93.14 92.10 92.74 90.70 89.38 90.05
Black 2.39 2.76 2.73 4.61 5.32 5.23
Other 3.21 3.99 3.55 3.12 3.78 3.40
Education
⬍High school 6.77 4.80 3.23 7.24 5.08 3.17
High school 19.24 14.17 10.10 29.70 24.18 18.17
Some college 33.49 30.37 27.52 34.91 35.67 36.41
College/postgraduate college 37.70 48.12 57.06 24.72 31.94 40.50
Smoking status, pack/d
Never 27.59 33.08 36.88 43.73 46.54 46.79
Past, ⱕ1 25.64 29.85 32.72 22.44 26.89 31.10
Past, ⬎1 25.83 24.56 22.58 10.45 10.97 11.30
Current, ⱕ1 8.99 5.46 3.25 14.01 9.61 6.33
Current, ⬎1 8.09 3.32 1.24 6.12 2.88 1.35
Marital status
Married 83.99 85.50 86.01 44.68 45.33 46.34
Not currently married 15.25 13.84 13.54 54.21 53.72 52.86
Menopausal hormone therapy
Never NA NA NA 49.65 44.32 40.32
Past NA NA NA 8.53 8.83 8.92
Current NA NA NA 41.58 46.57 50.56
Physical activity
Never 4.17 2.44 1.20 7.45 4.26 2.29
Rarely 14.39 9.82 6.17 20.95 14.53 9.40
1-3 Times/mo 16.29 13.31 9.97 16.83 14.54 10.85
1-2 Times/wk 23.69 23.25 21.35 21.26 21.86 20.79
3-4 Times/wk 23.58 28.88 33.65 19.77 26.46 33.10
Greater than or equal to 5 times/wk 16.83 21.37 27.05 12.31 17.11 22.74
Abbreviations: aMED, alternate Mediterranean diet; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); NA, not
applicable.
a All data are reported as percentages, which might not total 100 because of rounding or missing information.
b Median age was 62 years.
deaths). Five more years of follow-up resulted in a total correlations with the total score (aMED). Energy intake
of 27 799 deaths (18 126 men and 9673 women). The 2 was weakly negatively associated with the score. The me-
Mediterranean scoring systems showed similar distribu- dian intakes for each of the components were similar in
tions among men and women (the scores were posi- men and women, with the exception of alcohol and total
tively correlated; r ⬎0.70). Because the results from the energy, both of which were higher in men than in women.
2 scores were similar, we present results for the aMED Higher conformity with the Mediterranean dietary pat-
only because this score was modified for use in a US popu- tern was associated with a statistically significant reduc-
lation.2 In men, conformity with the Mediterranean diet tion in all-cause mortality, including mortality due to can-
tended to be higher among older people with the follow- cer and CVD in both men and women (Table 3 and
ing characteristics: BMIs between 18.5 and less than 30.0, Table 4). In men, the multivariate hazard ratios (HRs)
higher education, higher physical activity, married, and of mortality from any cause, cancer, and CVD for the
current nonsmokers or former smokers who had smoked top vs the bottom level of conformity with the dietary
1 pack per day or less (Table 1). A similar pattern was pattern was 0.77 (95% confidence interval [CI], 0.74-
seen in women, with the exception of marital status; in 0.80), 0.79 (95% CI, 0.73-0.87), and 0.76 (95% CI,
addition, higher conformity in women was associated with 0.68-0.85), respectively. In women, higher conformity
current use of menopausal hormone therapy (Table 1). with aMED was associated with a 22%, 14%, and 21%
Table 2 reports the median, 25th, and 75th percen- decreased risk of all-cause (P⬍.001 for trend), cancer
tile values of intake of the score components and their (P=.01 for trend), and CVD (P=.01 for trend) mortality,
2463

Table 2. Daily Dietary Intakes of Food Groups and Nutrients and Spearman Correlation Coefficients (r) With aMED a
Men Women
25th 75th 25th 75th

Diet Score Components Percentile Median Percentile rb Percentile Median Percentile rb
Vegetables (no potatoes) 2.39 3.49 4.99 0.50 2.48 3.65 5.29 0.47
Legumes 0.09 0.17 0.33 0.41 0.05 0.12 0.23 0.39
Fruit and nuts 2.16 3.53 5.37 0.45 2.26 3.57 5.26 0.38
Total fruit 1.88 3.24 5.10 0.43 2.08 3.40 5.10 0.37
Nuts 0.06 0.14 0.31 0.21 0.04 0.09 0.19 0.18
Whole grain 0.80 1.42 2.19 0.41 0.72 1.18 1.79 0.36
Total grain 5.51 6.98 8.55 0.25 4.48 5.70 7.03 0.17
Total meat 2.85 4.02 5.40 −0.21 1.89 2.80 3.88 −0.20
Red and processed meat 1.86 2.84 3.97 −0.35 1.01 1.66 2.46 −0.36
Dairy products 0.84 1.42 2.30 −0.02 0.75 1.32 2.14 −0.04
Monounsaturated fat–saturated fat ratio 1.10 1.23 1.37 0.37 1.07 1.22 1.37 0.33
Fish 0.33 0.58 1.01 0.35 0.26 0.47 0.84 0.32
Alcohol, g 0.90 6.09 24.08 −0.02 0.04 1.44 7.35 0.09
Energy, cal 1455 1889 2440 −0.13 1119 1455 1881 −0.07
Abbreviations: aMED, alternate Mediterranean diet score; r, Spearman correlation coefficient.

a Unless otherwise indicated data are reported as servings per day adjusted to 2000 calories in women and 2500 calories in men.
b All correlations are statistically significant from zero (P⬍ .001).
Table 3. Results of Statistical Analysis for the Mediterranean Dietary Pattern (aMED) and All-Cause Mortality a
aMED Score aMED Score

P Value P Value
Characteristic 0-3 4-5 6-9 for Trend 0-3 4-5 6-9 for Trend
All Subjects
Men (n=214 284) Women (n = 166 012)
Cases, No. 7616 6903 3607 4073 3891 1709

Age-adjusted rates b 1167.6 870.5 658.5 782.1 583.2 462.1
Age-adjusted HR 1 [Reference] 0.74 (0.72-0.77) 0.56 (0.54-0.59) ⬍.001 1 [Reference] 0.75 (0.71-0.78) 0.59 (0.56-0.63) ⬍.001
Multivariate HR c 1 [Reference] 0.89 (0.87-0.93) 0.77 (0.74-0.80) ⬍.001 1 [Reference] 0.88 (0.84-0.92) 0.78 (0.74-0.83) ⬍.001
Multivariate HR d 1 [Reference] 0.91 (0.88-0.94) 0.79 (0.76-0.83) ⬍.001 1 [Reference] 0.89 (0.85-0.93) 0.80 (0.75-0.85) ⬍.001
Never Smokers
Men (n=68 971) Women (n = 75 740)
Cases, No. 1236 1578 996 1106 1269 608

Multivariate HR e 1 [Reference] 0.95 (0.88-1.02) 0.83 (0.76-0.90) ⬍.001 1 [Reference] 0.89 (0.82-0.97) 0.80 (0.73-0.89) ⬍.001
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CI, confidence interval; HR, hazard ratio;
MHT, menopausal hormone therapy; NIH-AARP, National Institutes of Health–AARP (formerly American Association of Retired Persons) Diet and Health Study.12
a Unless otherwise indicated, data are reported as HR (95% CI).
b Incidence rates per 100 000 person-years standardized to the age distribution of the entire baseline NIH-AARP population.
c Adjusted for age (continuous), race (white, black, other, and data missing), total energy (continuous), BMI (⬍ 18.5, 18.5 to ⬍ 25, 25 to ⬍ 30, 30 to ⬍35, 35 to
⬍ 40, ⱖ 40, and data missing), education (less than high school, high school, some college, college/postgraduate college, and data missing), marital status
(married, not currently married, and data missing), physical activity (never, rarely, 1-3 times per month, 1-2 times per week, 3-4 times per week, ⱖ 5 times per
week, and data missing), MHT (never, past, current, unknown, and data missing) in women only, and for 6-level smoking variable (never smokers, former
smokers ⱕ 1 pack per day, former smokers ⬎ 1 pack per day, current smokers ⱕ 1 pack per day, current smokers ⬎ 1 pack per day, and data missing).
d Adjusted for age (continuous), race (white, black, other, and data missing), total energy (continuous), BMI (⬍ 18.5, 18.5 to ⬍ 25, 25 to ⬍ 30, 30 to ⬍35, 35 to
⬍40, ⱖ 40, and data missing), education (less than high school, high school, some college, college/postgraduate college, and data missing), marital status
week, and data missing), MHT (never, past, current, unknown, and data missing) in women only, and for 32-level smoking variable (integrating smoking status
[never, former, and current]), years since quitting for former smokers [ⱖ 10, 5-9, 1-4, and within the last year], smoking dose for both former and current
smokers [1-10, 11-20, 21-30, 31-40, 41-60, and ⱖ61 cigarettes per day], and data missing for smoking).
e Adjusted for age (continuous), race (white, black, other, and data missing), total energy (continuous), BMI (⬍ 18.5, 18.5 to ⬍ 25, 25 to ⬍ 30, 30 to ⬍35, 35 to
⬍ 40, ⱖ 40, and data missing), education (less than high school, high school, some college, college/postgraduate college, and data missing), marital status
week, and data missing), and MHT (never, past, current, unknown, and data missing) in women only.
2464

Table 4. Results of Statistical Analysis for the Mediterranean Dietary Pattern (aMED) and Cause-Specific Mortality a
aMED Score aMED Score
Men (n=214 284) Women (n = 166 012)

P Value P Value
Characteristic 0-3 4-5 6-9 for Trend 0-3 4-5 6-9 for Trend
Cancer
Cases, No. 1608 1346 763 919 921 428
Multivariate HR c 1 [Reference] 0.84 (0.78-0.91) 0.79 (0.73-0.87) ⬍.001 1 [Reference] 0.92 (0.83-1.01) 0.86 (0.76-0.97) .01
Multivariate HR d 1 [Reference] 0.86 (0.80-0.93) 0.83 (0.76-0.91) ⬍.001 1 [Reference] 0.93 (0.85-1.02) 0.88 (0.78-1.00) .04
Cardiovascular Disease
Cases, No. 1012 952 461 446 400 180
Multivariate HR c 1 [Reference] 0.94 (0.86-1.03) 0.76 (0.68-0.85) ⬍.001 1 [Reference] 0.85 (0.74-0.97) 0.79 (0.66-0.95) .01
Multivariate HR d 1 [Reference] 0.95 (0.86-1.04) 0.78 (0.69-0.87) ⬍.001 1 [Reference] 0.85 (0.74-0.98) 0.81 (0.68-0.97) .01
Other Causes
Cases, No. 761 657 329 421 354 147
Multivariate HR c 1 [Reference] 0.88 (0.79-0.98) 0.74 (0.65-0.85) ⬍.001 1 [Reference] 0.81 (0.70-0.94) 0.70 (0.58-0.86) ⬍.001
Multivariate HR d 1 [Reference] 0.90 (0.81-1.00) 0.77 (0.70-0.88) ⬍.001 1 [Reference] 0.82 (0.71-0.95) 0.72 (0.59-0.87) ⬍.001
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); CI, confidence interval; HR, hazard ratio;
MHT, menopausal hormone therapy; NIH-AARP, National Institutes of Health–AARP (formerly American Association of Retired Persons) Diet and Health Study.12
a Unless otherwise indicated, data are reported as HR (95% CI).
c
Adjusted for age (continuous), race (white, black, other, and data missing), total energy (continuous), BMI (⬍18.5, 18.5 to ⬍25, 25 to ⬍30, 30 to ⬍35, 35 to
⬍40, ⱖ40, and data missing), education (less than high school, high school, some college, college/postgraduate college, and data missing), marital status
(married, not currently married, and data missing), physical activity (never, rarely, 1-3 times per month, 1-2 times per week, 3-4 times per week, ⱖ5 times per
smokers ⱕ1 pack per day, former smokers ⬎1 pack per day, current smokers ⱕ1 pack per day, current smokers ⬎1 pack per day, and data missing).
d
Adjusted for age (continuous), race (white, black, other, and data missing), total energy (continuous), BMI (⬍18.5, 18.5 to ⬍25, 25 to ⬍30, 30 to ⬍35, 35 to
week, and data missing), MHT (never, past, current, unknown, and data missing) in women only, and for 32-level smoking variable (integrating smoking status
[never, former, and current]), years since quitting for former smokers [ⱖ10, 5-9, 1-4, and within the last year], smoking dose for both former and current
smokers [1-10, 11-20, 21-30, 31-40, 41-60, and ⱖ61 cigarettes per day], and data missing for smoking).
respectively. Analysis for 1-point increments in the toward lower conformity (0-2), with only 1% of the popu-
scores showed a 5% decrease (P ⬍ .001) in all-cause lation having a score of 9 points for aMED. Higher con-
mortality for both men and women. Finer adjustment formity to this modified score was also associated with a
for smoking showed similar results. significantly lower risk of all-cause mortality (for men,
Similar results were seen with the tMED score: the mul- n=408; 95% CI, 0.71-0.87; for women, n=144; 95% CI,
tivariate HRs for high vs low conformity for all-cause mor- 0.66-0.92) (data not shown).
tality in men and women were 0.79 (95% CI, 0.76-0.82) We conducted stratified analysis of all-cause mortal-
and 0.84 (95% CI, 0.79-0.89), respectively. Higher con- ity examining subgroups of smoking and BMI (Table 5).
formity with the tMED score also conferred decreased The results in all subgroups, with the exception of those
risks for cause-specific mortality. In men, the multivar- among never smokers with a BMI of 30.0 or higher,
iate HRs for high vs low conformity for cancer and CVD showed a statistically significant inverse association be-
mortality were 0.79 (95% CI, 0.72-0.87) and 0.76 (95% tween the Mediterranean dietary pattern and all-cause
CI, 0.68-0.85), respectively. In women, a 20% de- mortality. This inverse association appeared to be more
creased risk of CVD mortality was seen with tMED pronounced among smokers and especially among smok-
(P⬍.01) and a weaker relation for cancer mortality (HR, ers with normal BMI (18.5-25.0). The multivariate HR
0.89; 95% CI, 0.79-1.01) (data not shown). in the latter subgroup comparing high vs low level of ad-
In separate analyses excluding 1, 2, 3, and 5 years of herence was 0.54 (95% CI, 0.50-0.59) in men and 0.59
follow-up, the association between the Mediterranean di- (95% CI, 0.53-0.66) in women. In high vs low confor-
etary pattern and all-cause mortality remained in both mity, the pattern also conferred a decreased risk of death
men and women (data not shown). Also, when smokers among smokers with high BMI. When we compared par-
were excluded from the analyses, the associations be- ticipants with contrasting diet and lifestyle behavior such
tween aMED and all-cause mortality remained statisti- as those who were obese and ever smokers with low con-
cally significant (Table 3). In sensitivity analysis using formity with the Mediterranean pattern vs those who were
alternate sex-specific cut points, the distribution shifted lean and never smokers with high conformity with the
2465

Table 5. Results of Statistical Analysis Between the Mediterranean Dietary Pattern and All-Cause Mortality Within Smoking Status
and BMI Strata a
aMED Score
P Value
Smoking Status/BMI Variable 0-3 4-5 6-9 for Trend
Men
Never/18.5-25.0 Cases, No. 307 462 396
Age-adjusted rate b 640.8 531.1 480.1
Age-adjusted HR 1 [Reference] 0.83 (0.72-0.96) 0.75 (0.64-0.87) ⬍.001
Multivariate HR c 1 [Reference] 0.87 (0.75-1.01) 0.82 (0.70-0.96) .02
Never/⬎25.0 to ⬍30.0 Cases, No. 576 721 409
Multivariate HR c 1 [Reference] 0.96 (0.86-1.07) 0.77 (0.68-0.88) ⬍.001
Never/ⱖ30.0 Cases, No. 321 348 158
Age-adjusted HR 1 [Reference] 0.97 (0.83-1.13) 0.84 (0.69-1.02) .09
Ever/18.5-25.0 Cases, No. 1909 1402 728
Ever/⬎25.0 to ⬍30.0 Cases, No. 2623 2358 1165
Ever/ⱖ30.0 Cases, No. 1294 1113 493
Women
Never/18.5-25.0 Cases, No. 368 475 285
Never/⬎25.0 to ⬍30.0 Cases, No. 311 414 173
Never/ⱖ30.0 Cases, No. 349 310 120
Ever/18.5-25.0 Cases, No. 1228 1050 498
Ever/⬎25.0 to ⬍30.0 Cases, No. 760 719 305
Ever/ⱖ30.0 Cases, No. 557 538 163
Abbreviations: aMED, alternate Mediterranean diet; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared);
CI, confidence interval; HR, hazard ratio; MHT, menopausal hormone therapy; NIH-AARP, National Institutes of Health–AARP (formerly American Association of
Retired Persons) Diet and Health Study.12
a Unless otherwise indicated, data are reported as HR (95% CI). Stratified analysis was restricted to participants with no missing information for smoking and
BMI and with BMI of at least 18.5.
c Adjusted for age (continuous), race (white, black, other, and data missing), total energy (continuous), BMI (⬍18.5, 18.5 to ⬍25, 25 to ⬍30, 30 to ⬍35, 35 to
smokers ⱕ1 pack per day, former smokers ⬎1 pack per day, current smokers ⱕ1 pack per day, current smokers ⬎1 pack per day, and data missing).
2466

diet, among the latter there was a 56% (95% CI, 0.39- The beneficial effect of the Mediterranean diet on mor-
0.49) decreased risk of death in men and a 61% (95% CI, tality may be mediated by a number of mechanisms, in-
0.34-0.45) decreased risk in women, and the test for including oxidative stress and chronic inflammation. Con-
teraction was statistically significant (P⬍.001 for inter- formity with the Mediterranean dietary pattern is
action) (data not shown). associated with high antioxidant capacity and low con-
centrations of oxidized low-density lipoprotein choles-
terol,27 suggesting that this dietary pattern could be cap-
COMMENT
turing the combined effect of dietary antioxidants, which
could, partially, explain a lower risk of mortality. The
In this large, prospective US study, conformity with the Mediterranean diet also includes other important di-
Mediterranean dietary pattern was associated with lower etary constituents such as fiber and a low omega-6–
mortality. The inverse association in men and women re- omega-3 fatty acid ratio, both of which potentially pre-
mained statistically significant after restricting the analy- vent cancer initiation and progression.28,29
sis to nonsmokers. We also found similar results using Additionally, chronic inflammation has been associ-
different aMED cut points. The beneficial effect of the ated with a higher risk of CVD and possibly cancer.30,31
Mediterranean pattern was more pronounced in smok- Conformity with the Mediterranean dietary pattern has
ers, especially among smokers with a BMI of 18.5 to 25.0. also been associated with lower levels of C-reactive pro-
To our knowledge, 9 cohort studies (8 in Eu- tein, interleukin-6, homocysteine, and fibrinogen and with
rope3-7,9-11 and 1 in Australia8) have used variations of the lower white blood cell counts.32 In 1 US study,2 it was
original score11 to assess the Mediterranean diet and mor- associated with lower concentrations of biomarkers of
tality relation. Despite these modifications, all studies re- inflammation and endothelial dysfunction.
ported inverse associations between the Mediterranean To our knowledge, the present study is the first and
dietary pattern and mortality.21 largest US cohort to evaluate the Mediterranean dietary
Although our primary analysis was based on aMED, pattern and mortality. Although using such a score in an
we also examined tMED, which includes dairy and total American population might not adequately represent con-
meat products and does not distinguish between total formity with the traditional Mediterranean diet, it does
grains and whole grains.4,11 Recent evidence suggests that include the key features of this diet and, as we have shown,
high glycemic load has adverse effects on blood lipid lev- may have a substantial beneficial impact on mortality in
els and is associated with an increased risk of CVD but the United States.
that whole grains are beneficial, and therefore, the ef- Because of concern that the median intake of some di-
fects of grain products may depend on the degree of pro- etary components might be lower in the US population than
cessing.22,23 In addition, the range of alcohol consump- in a Mediterranean population, we considered alternate cut
tion in tMED is wider than that in most US studies. points. Using this approach, we assigned points to only the
Although a higher intake of alcohol has been shown to highest intakes, and the diets consumed were more com-
be protective against heart disease,24 it also increases the patible with the traditional Mediterranean diet. This sen-
risk of several types of cancer.25 In our study, higher con- sitivity analysis showed that higher conformity with the diet
formity with tMED was also inversely associated with all- was also associated with a significantly lower risk of mor-
cause and cause-specific mortality in both men and tality, confirming that it was reasonable to use the median
women, indicating that both scores showed a beneficial of intake as the cut point for constructing the score.
effect on mortality. The Mediterranean dietary pattern has the potential
In our study, the beneficial effect of the pattern was to minimize confounding by including nutritional con-
more pronounced among smokers with high confor- founders in the score and capturing effect modification
mity with the diet. Similarly, another study found that among the nutritional variables.33 Smoking was associ-
risk was lower among smokers with the highest confor- ated with decreased conformity with the diet and may
mity with the Mediterranean diet.6 Furthermore, the ap- therefore confound the association between the Medi-
parent benefit of high conformity with the Mediterra- terranean dietary pattern and mortality. Although we care-
nean dietary pattern was more pronounced among lean fully adjusted for smoking, this may not be sufficient to
or overweight individuals, particularly among smokers. address potential confounding by smoking. We had
This held true for both men and women. Possibly, con- adequate statistical power to restrict the analysis to par-
formity with the Mediterranean diet plays a particularly ticipants who had never smoked (approximately 69 000
important role among smokers, who are characterized men and 76 000 women). When we did so, the relation
by high levels of oxidative stress and an adverse blood of dietary pattern to mortality remained unchanged.
lipid profile.26 On the other hand, among obese individu- Food frequency questionnaires are known to contain
als who do not smoke, conformity with the Mediterra- a certain degree of measurement error, as suggested by
nean diet might not play so much of a role because the studies using recovery biomarkers of energy and pro-
increased risk of death due to obesity may mask the in- tein intakes.34 Therefore, results that depend on FFQs to
verse association of the Mediterranean diet with all- assess diet and risk of chronic disease might be affected
cause mortality. However, to our knowledge, this is the by this measurement error.35 In the present study, all in-
first study of mortality that examines the effect of the dividual components of the score were adjusted for total
Mediterranean dietary pattern within both smoking and energy intake (density method) and standardized to 2000
BMI strata, and therefore, additional studies should fur- calories in women and 2500 calories in men. Total en-
ther investigate this finding. ergy intake was also included as a covariate in the model
2467

to achieve the equivalent of an isocaloric diet and to re- 6. Haveman-Nies A, de Groot LP, Burema J, Cruz JA, Osler M, van Staveren WA.
Dietary quality and lifestyle factors in relation to 10-year mortality in older Eu-
duce measurement error in the score.34
ropeans: the SENECA study. Am J Epidemiol. 2002;156(10):962-968.
In conclusion, this large prospective study demon- 7. Knoops KT, de Groot LC, Kromhout D, et al. Mediterranean diet, lifestyle factors,
strates that higher conformity with the Mediterranean di- and 10-year mortality in elderly European men and women: the HALE project.
etary pattern is associated with lower all-cause and cause- JAMA. 2004;292(12):1433-1439.
specific mortality. These findings confirm results from 8. Kouris-Blazos A, Gnardellis C, Wahlqvist ML, Trichopoulos D, Lukito W, Tricho-
poulou A. Are the advantages of the Mediterranean diet transferable to other popu-
previous studies and suggest that the Mediterranean di-
lations? a cohort study in Melbourne, Australia. Br J Nutr. 1999;82(1):57-61.
etary score is a useful tool for evaluating diet and mor- 9. Lasheras C, Fernandez S, Patterson AM. Mediterranean diet and age with re-
tality in a non-Mediterranean US population. spect to overall survival in institutionalized, nonsmoking elderly people. Am J
Clin Nutr. 2000;71(4):987-992.
Accepted for Publication: July 20, 2007. 10. Osler M, Schroll M. Diet and mortality in a cohort of elderly people in a north
Correspondence: Panagiota N. Mitrou, PhD, Depart- European community. Int J Epidemiol. 1997;26(1):155-159.
11. Trichopoulou A, Kouris-Blazos A, Wahlqvist ML, et al. Diet and overall survival
ment of Public Health and Primary Care, University of in elderly people. BMJ. 1995;311(7018):1457-1460.
Cambridge, Cambridge CB1 8RN, England (pm277 12. Schatzkin A, Subar AF, Thompson FE, et al. Design and serendipity in establish-
@medschl.cam.ac.uk). ing a large cohort with wide dietary intake distributions. Am J Epidemiol. 2001;
Author Contributions: Dr Mitrou had full access to all 154(12):1119-1125.
of the data in the study and takes full responsibility for 13. Michaud D, Midthune D, Hermansen S, et al. Comparison of cancer registry case
ascertainment with SEER estimates and self-reporting in a subset of the NIH-
the integrity of the data and the accuracy of the data analy- AARP Diet and Health Study. J Registry Manage. 2005;32:70-75.
sis. Study concept and design: Mitrou, Kipnis, Reedy, Subar, 14. National Cancer Institute. SEER Cause of Death Recode 1969⫹ (9/17/2004). http:
Mouw, Hollenbeck, Leitzmann, and Schatzkin. Acquisi- //seer.cancer.gov/codrecode/1969+_d09172004/index.html. Accessed May 28,
tion of data: Mitrou, Subar, and Schatzkin. Analysis and 2007.
interpretation of data: Mitrou, Kipnis, Thiébaut, Reedy, 15. National Cancer Institute. Diet History Questionnaire. http://riskfactor.cancer.gov
/DHQ/. Accessed May 28, 2007.
Subar, Wirfält, Flood, Leitzmann, and Schatzkin. Draft- 16. Thompson FE, Kipnis V, Midthune D, et al. Performance of a food frequency ques-
ing of the manuscript: Mitrou, Reedy, Subar, Flood, and tionnaire in the US NIH-AARP (National Institutes of Health-American Associa-
Leitzmann. Critical revision of the manuscript for impor- tion of Retired Persons) Diet and Health Study [published online ahead of print
tant intellectual content: Mitrou, Kipnis, Thiébaut, Reedy, July 5, 2007]. Public Health Nutr.
Subar, Wirfält, Flood, Mouw, Hollenbeck, Leitzmann, and 17. Millen AE, Midthune D, Thompson FE, Kipnis V, Subar AF. The National Cancer
Institute diet history questionnaire. Am J Epidemiol. 2006;163(3):279-288.
Schatzkin. Statistical analysis: Mitrou, Kipnis, Thiébaut, 18. Cox DR. Regression models and life tables. J Royal Stat Soc. 1972;B34:187-220.
Reedy, Subar, Leitzmann, and Schatzkin. Obtained fund- 19. Horvitz DG, Thompson DJ. A generalization of sampling without replacement from
ing: Schatzkin. Administrative, technical, and material sup- a finite universe. J Am Stat Assoc. 1952;47:663-685.
port: Mouw, Hollenbeck, and Schatzkin. Study supervi- 20. Breslow N, Day N. Statistical Methods in Cancer Research. Lyon, France: Inter-
sion: Kipnis, Subar, Flood, Leitzmann, and Schatzkin. national Agency for Research on Cancer; 1987.
21. Serra-Majem L, Trichopoulou A, Ngo de la Cruz J, et al. Does the definition of the
Financial Disclosure: None reported. Mediterranean diet need to be updated? Public Health Nutr. 2004;7(7):927-929.
Funding/Support: This research was supported by the 22. Edge MS, Jones JM, Marquart L. A new life for whole grains. J Am Diet Assoc.
Intramural Research Program of the National Institutes 2005;105(12):1856-1860.
of Health, National Cancer Institute, Division of Cancer 23. Slavin J, Jacobs D, Marquart L. Whole-grain consumption and chronic disease:
Epidemiology and Genetics. protective mechanisms. Nutr Cancer. 1997;27(1):14-21.
24. Grønbaek M. Factors influencing the relation between alcohol and cardiovascu-
Role of the Sponsors: The funding organization had no lar disease. Curr Opin Lipidol. 2006;17(1):17-21.
role in the design or conduct of the study; collection, man- 25. Boffetta P, Hashibe M. Alcohol and cancer. Lancet Oncol. 2006;7(2):149-156.
agement, analysis, or interpretation of the data; or prepa- 26. Yanbaeva DG, Dentener MA, Creutzberg EC, Wesseling G, Wouters EF. Sys-
ration, review, or approval of the manuscript. temic effects of smoking. Chest. 2007;131(5):1557-1566.
Additional Contributions: The participants in the NIH- 27. Pitsavos C, Panagiotakos D, Trichopoulou A, et al. Interaction between Mediter-
ranean diet and methylenetetrahydrofolate reductase C677T mutation on oxi-
AARP Diet and Health Study provided outstanding co-
dized low density lipoprotein concentrations: the ATTICA study. Nutr Metab Car-
operation; Neal Freedman, PhD, and Douglas Midthune, diovasc Dis. 2006;16(2):91-99.
MA, provided statistical advice; and Tawanda Roy, BA, 28. Klurfeld DM. Dietary fibre-mediated mechanisms in carcinogenesis. Cancer Res.
provided research assistance. 1992;52(7)(suppl):2055s-2059s.
29. Simopoulos AP. Evolutionary aspects of diet, the omega-6/omega-3 ratio and
genetic variation. Biomed Pharmacother. 2006;60(9):502-507.
REFERENCES 30. Danesh J, Collins R, Appleby P, Peto R. Association of fibrinogen, C-reactive pro-
tein, albumin, or leukocyte count with coronary heart disease: meta-analyses of
1. Trichopoulou A. Traditional Mediterranean diet and longevity in the elderly: a review. prospective studies. JAMA. 1998;279(18):1477-1482.
Public Health Nutr. 2004;7(7):943-947. 31. Lawrence T. Inflammation and cancer: a failure of resolution? Trends Pharma-
2. Fung TT, McCullough ML, Newby PK, et al. Diet-quality scores and plasma con- col Sci. 2007;28(4):162-165.
centrations of markers of inflammation and endothelial dysfunction. Am J Clin 32. Chrysohoou C, Panagiotakos DB, Pitsavos C, Das UN, Stefanadis C. Adherence
Nutr. 2005;82(1):163-173. to the Mediterranean diet attenuates inflammation and coagulation process in
3. Lagiou P, Trichopoulos D, Sandin S, et al. Mediterranean dietary pattern and mor- healthy adults: the ATTICA study. J Am Coll Cardiol. 2004;44(1):152-158.
tality among young women. Br J Nutr. 2006;96(2):384-392. 33. Trichopoulou A, Critselis E. Mediterranean diet and longevity. Eur J Cancer Prev.
4. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a Mediter- 2004;13(5):453-456.
ranean diet and survival in a Greek population. N Engl J Med. 2003;348(26): 34. Kipnis V, Subar AF, Midthune D, et al. Structure of dietary measurement error:
2599-2608. results of the OPEN biomarker study. Am J Epidemiol. 2003;158(1):14-26.
5. Trichopoulou A, Orfanos P, Norat T, et al. Modified Mediterranean diet and sur- 35. Kristal AR, Peters U, Potter JD. Is it time to abandon the food frequency
vival: EPIC-elderly prospective cohort study. BMJ. 2005;330(7498):991. questionnaire? Cancer Epidemiol Biomarkers Prev. 2005;14(12):2826-2828.
2468

Ioi70171 2461 2468

Uploaded by

Copyright:

Available Formats

Ioi70171 2461 2468

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ioi70171 2461 2468

Uploaded by

Copyright:

Available Formats

ORIGINAL INVESTIGATION

Mediterranean Dietary Pattern and Prediction

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

Characteristic 0-3 4-5 6-9 0-3 4-5 6-9

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

25th 75th 25th 75th

Abbreviations: aMED, alternate Mediterranean diet score; r, Spearman correlation coefficient.

aMED Score aMED Score

Cases, No. 7616 6903 3607 4073 3891 1709

Cases, No. 1236 1578 996 1106 1269 608

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

aMED Score aMED Score

Men (n=214 284) Women (n = 166 012)

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

©2007 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ on 05/30/2023

You might also like