Food Group Diet Scores and Coronary Heart Disease

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

International food group–based diet quality and risk of coronary heart

disease in men and women


Teresa T Fung,1,2 Sheila Isanaka,2 Frank B Hu,2,3 and Walter C Willett2,3

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


1 Department of Nutrition, Simmons College, Boston, MA; 2 Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA; and 3 Channing
Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital, Boston, MA

ABSTRACT hunger, achieving food security, and promoting nutrition. In order


Background: Standard diet quality assessment tools, which mea- to monitor progress in nutrition, a standard diet quality assess-
sure micronutrient sufficiency and food consumption related to dis- ment, which holistically measures sufficiency of micronutrients
ease and applicable to different populations, are needed to track and consumption of foods related to noncommunicable disease
progress in meeting the Sustainable Development Goals related to and can be applied to different populations, is needed. These as-
hunger, food security, and nutrition. Diet quality scores have been sessments should be quick to administer, applicable internation-
constructed for high- and low-income countries, but none are simple ally, able to be used for population monitoring and clinical use,
to administer or applicable internationally. and related to health outcomes. Often, diet quality is measured by
Objective: We prospectively examined the association between the nutrient adequacy and whether or not an upper limit of unhealthy
Food Group Index (FGI), the Minimal Diet Diversity Score for dietary components, such as added sugar and sodium, is exceeded
Women (MDDW), and a new Prime Diet Quality Score (PDQS), and (2–4).
the risk of ischemic heart disease (IHD) in 3 US cohorts. A number of diet quality scores or indexes have been devel-
Design: In total, 75,045 women (baseline age 43–63 y), 43,966 men oped based on recommendations put forth by government or pro-
(aged 40–75 y), and 93,131 younger women (aged 27–44 y) without fessional clinical groups or on empirical evidence. For example,
a history of cardiovascular disease were followed up to 28 y. Diet was the Healthy Eating Index was developed by the US Department of
assessed multiple times using food frequency questionnaires and the Agriculture and is composed of 12 food and nutrient groups that
3 diet quality scores were computed for each individual. The associ- measure adherence to the 2010 Dietary Guidelines for Americans
ation with IHD was modeled with Cox proportional hazard models, (3). Global cancer prevention recommendations by the World
controlling for potential confounders. Cancer Research Fund and the American Institute for Cancer
Results: During follow-up, we ascertained 2908 incident IHD cases
Research were used to develop the WCRF/AICR score, consist-
in the Nurses’ Health Study, 3722 in the Health Professionals Follow-
ing of 6–7 lifestyle and dietary components (5). In addition, sev-
up Study, and 505 in the Nurses’ Health Study II. The FGI was not
eral versions of the Mediterranean diet score have been devel-
associated with total IHD in any cohort. The PDQS was significantly
oped to measure adherence to a traditional Mediterranean diet
associated with IHD in all 3 cohorts separately and the pooled RR
(6). Although these dietary quality scores have been associated
for each SD increase was 0.89 (95% CI: 0.87, 0.91). This was sig-
with lower risk of noncommunicable disease and mortality, they
nificantly different than the pooled RR of 0.93 for MDDW (95% CI:
were primarily evaluated in high-income countries (4, 7–9). In
0.90, 0.96) and the RR of 0.98 for the FGI (95% CI: 0.95, 1.01). The
association did not appear to differ by age.
Conclusion: We found that the PDQS with the most detailed differ- Supported by NIH research grants: UM1 CA186107, R01 HL034594,
entiation of healthy and unhealthy foods was associated with a lower UM1 CA176726, UM1 CA167552, and R01 HL35464.
risk of IHD in a high-income country. On the other hand, diet quality Supplemental Tables 1–6 are available from the “Supplementary data” link
in the online posting of the article and from the same link in the online table
scores that do not account for unhealthy foods had a limited associ-
of contents at https://academic.oup.com/ajcn/.
ation with IHD. Am J Clin Nutr 2018;107:120–129.
Address correspondence to TTF (e-mail: fung@simmons.edu).
Abbreviations: FANTA, Food and Nutrition Technical Assistance project;
Keywords: men, women, diet, nutrition, diet quality, diet scores, FFQ, food-frequency questionnaire; FGI, Food Group Index; HPFS, Health
coronary heart disease, cardiovascular Professionals’ Follow-up Study; IHD, ischemic heart disease; MDDW, Min-
imal Diet Diversity score for Women; MI, myocardial infarction; NHS,
INTRODUCTION Nurses’ Health Study; NHS II, Nurses’ Health Study II; PDQS, Prime Diet
Quality Score.
In 2015, the United Nations announced 17 Sustainable De- Received February 7, 2017. Accepted for publication November 1, 2017.
velopment Goals for the next 15 y (1). These goals are applica- First published online January 26, 2018; doi: https://doi.org/10.1093/
ble internationally. Specifically, the second goal involves ending ajcn/nqx015.

120 Am J Clin Nutr 2018;107:120–129. Printed in USA. © 2018 American Society for Nutrition. All rights reserved.
FOOD GROUP DIET SCORES AND CORONARY HEART DISEASE 121
addition, they require software to compute the nutrient composi- the FFQ was administered 6 times between 1986 and 2006, and
tion of the diet, and the scoring algorithm was based on the intake was administered 5 times between 1991 and 2007 in NHS II.
level of the population. Therefore, they are not simple to admin- Each FFQ contained ∼135 items and each item had 9 frequency
ister and are difficult to apply across different countries. choices, ranging from <1 time/mo to ≥6 times/d. A standard por-
Currently, 2 simple food-based diet quality scores have been tion size was also provided.
developed for use internationally. The WHO and the Pan Amer- Three diet quality scores were computed, which were based
ican Health Organization developed a Food Group Index (FGI) entirely on food groups with predetermined intake levels for score
consisting of 8 broad food groups to measure diet quality in in- assignments (Supplemental Table 1). The FGI score was de-
fants and young children in low-income countries (10). A higher veloped by the Food and Nutrition Technical Assistance project
FGI score was associated with better micronutrient adequacy. In (FANTA) and is based on 8 food groups considered to predict
addition, the Minimal Diet Diversity Score for Women (MDDW) nutrient adequacy in infants and children in 10 developing coun-
was developed for women of reproductive age in low-income tries (17). These food groups are fruits and vegetables, legumes
countries and consists of 10 food groups (11). This score has also and nuts, animal flesh (red, processed, poultry, fish), eggs, dairy,
been shown to associate with better micronutrient adequacy (12). grains and roots, vegetables with >130 RE/100 g, and added fats
These 2 indexes are based on food groups only and are easy to and oils. A scoring algorithm was designed for data collected by

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


administer, but they have not been examined for use in higher- the 24-h recall method, with 1 point assigned to each food group
income countries or beyond predicting nutrient adequacy. if it was consumed within the last 24 h and 0 if not. Because each
It would be ideal to have a universal diet quality assessment food group in this score covers a broad range of food, the high diet
tool that is associated with the risk of a range of poor health out- diversity in our cohorts would result in many participants achiev-
comes in both higher- and lower-income countries. Therefore, in ing high scores. To adhere to the characteristics of the FGI score
this analysis, we examined the association between the FGI and but also adapt the scoring algorithm for use with the FFQ, we as-
MDDW and the risk of ischemic heart disease (IHD) in 3 US co- signed 1 point for consumption of ≥3 times/wk and 0 points for
horts to test the utility of these 2 scores in a higher-income coun- less. Points for all food groups were summed to arrive at the total
try. We also developed a 21-food group Prime Diet Quality Score score. The FGI ranges from 0 to 8 points.
(PDQS), which aims to combine simplicity in assessing diet qual- The MDDW was developed by FANTA and the Food and Agri-
ity with the ability to differentiate healthy foods from unhealthy culture Organization of the WHO. It is based on 10 food groups
ones, and examined its association with the risk of IHD. that predicted micronutrient intake adequacy in women of re-
productive age from low-income countries (18). This score has
METHODS been proposed for global use to measure diet quality of women
(19). The food groups included are grains and starchy vegeta-
Study population bles, pulses, nuts and seeds, dairy, animal flesh, eggs, dark green
Participants from 3 ongoing long-standing cohorts in the leafy vegetables, vitamin A–rich vegetables and fruits, other veg-
United States were included in this study: the Nurses’ Health etables, other fruits. The original scoring criteria was 1 point for
Study (NHS), the Health Professionals’ Follow-up Study each food group when intake was ≥1 serving/d and 0 for less. Ad-
(HPFS), and the Nurses’ Health Study II (NHS II). The NHS co- ditionally, 5 total points was used as the minimum cutoff value for
hort began in 1976 and is comprised of women nurses who were achieving diet diversity. Because of this predetermined cutoff, as
35–55 y old at the time (13). The HPFS cohort began in 1986 well as the detailed groupings of plant foods which reflect diet
with 51,529 male health professionals aged 40–75 y (14). The quality more precisely than the FGI, we adhered to the original
NHS II began in 1989 with 116,671 women nurses aged 25–42 scoring criteria using 1 serving/d as cutoff when computing the
y (15). Participants of each cohort receive a questionnaire every MDDW for our cohorts. The MDDW total score ranges from 0
2 y to assess health outcomes, lifestyle, and medication use. A to 10 points.
validated food frequency questionnaire (FFQ) is sent every 4 y The PDQS was based on the Prime Screen questionnaire, a
to assess usual dietary intake, including vitamin and mineral sup- short diet assessment tool developed for clinical use to quickly
plements (16). assess diet quality (20). Foods were classified as healthy and
In this analysis, we used the year 1984 as the baseline for NHS unhealthy based on 2 major considerations: 1) data from the
because that was when a detailed FFQ was first administered. literature on the direction of association with the risk of non-
For HPFS, 1986 was used as the baseline and 1991 was used communicable diseases; and 2) nutrient contribution in the world-
for NHS II. We included participants who returned the FFQ in wide setting. The food groups included that were considered
1984 (NHS), 1986 (HPFS), and 1991 (NHS II), whose energy in- healthy were dark leafy green vegetables, cruciferous vegetables,
take was within a plausible range (500–3500 kcal/d for women, carrots, other vegetables, whole citrus fruits, other whole fruits,
800–4000 kcal/d for men). At baseline, we excluded those who legumes, nuts, poultry, fish, eggs, whole grains, and liquid veg-
reported a history of coronary heart disease or stroke. A total of etable oils. The unhealthy food groups included in the score were
75,045 NHS, 43,966 HPFS, and 93,131 NHS II participants were red meat, potatoes, processed meat, whole milk dairy, refined
included. This study was approved by the Institutional Review grains and baked goods, sugar sweetened beverages, fried foods
Board at Brigham and Women’s Hospital, Boston, MA. obtained away from home, and desserts and ice cream. Points
were assigned according to the following criteria: 0–1 serving/wk
(0 point) compared with 2–3 servings/wk (1 point) compared with
Diet assessment and computation of diet quality scores ≥4 servings/wk (2 points) for the healthy food groups. Scoring
In NHS, the FFQ was administered in 1984 and 1986, and was reversed and points deducted for the unhealthy food groups.
every 4 y thereafter for a total of 8 times until 2010. In HPFS, Points for each food group were then summed to give an overall
122 FUNG ET AL.

score. The PDQS has 21 food groups and ranges from 0 to 42 scores, therefore it was categorized as <5.0, 5.0 to <6.0, 6.0 to
total points. <7.0, 7.0 to <8.0, and 8 points (full score) for NHS and HPFS. In
NHS II, the number of cases were fewer, therefore, scores were
classified into 4 categories: <5.0, 5.0 to <6.0, 6.0 to <7.0, and
Assessment of outcomes ≥7.0 points. For the MDDW, all 3 cohorts were categorized into
IHD was defined as non-fatal myocardial infarction (MI) and either 0 to <3.0 points, 3.0 to <4.0 points, 4.0 to <5.0 points, 5.0
fatal coronary disease in this study. When a participant self- to <6.0 points, or ≥6 points (full score was 10 points). The PDQS
reported a IHD event in each biennial questionnaire during the has a wide score range, and therefore we ranked individuals into
follow-up period, permission from the participants or their next of quintiles.
kin (for deceased participants) was obtained to allow study physi- Spearman correlation coefficients were computed for the 3 diet
cians (who were unaware of the participant’s dietary intake) to quality scores. We used time-dependent Cox proportional haz-
verify the event by reviewing the medical record. Non-fatal MI ard models conditioning on age and follow-up cycle to examine
was verified using WHO criteria which includes clinical symp- the association between the diet score and IHD, using the low-
toms and changes on electrocardiogram or cardiac enzymes (21). est score category as reference. Models were fitted separately for
each cohort. Multivariable models were adjusted for age (contin-

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


Our analysis also included “probable” cases, in which the self-
report was affirmed by interview or letter but medical records uous), energy intake (quintiles), family history of MI (yes or no),
were not released (or available). physical activity (5 categories), BMI (9 categories), alcohol in-
All deaths, including fatal cardiovascular disease, were identi- take (5 categories), smoking (5 categories), and menopausal sta-
fied by searching state vital records or the National Death Index, tus and postmenopausal hormone use (4 categories). Except for
or reported by participants’ next of kin or the postal system. Fa- family history, all of the variables were time-dependent variables
tal cardiovascular disease was considered “definite” if confirmed updated every 2 y. To compare the magnitude of the association
by review of hospital or autopsy records, or when the participant between various diet quality indexes and IHD, we first assigned
had a history of IHD and the death certificate listed IHD as the z scores to each dietary quality score and modeled 1 SD increase
cause of death. Fatal cases confirmed by death certificate only in score for total IHD. We then pooled the β coefficients for each
were classified as “presumed” cases. diet score from all 3 cohorts using the method of DerSimonian
and Laird (23). Finally, we tested for difference in the β coef-
ficients 2 diet quality scores at a time using the Wald test. In
Assessment of lifestyle characteristics this analysis, the number of individuals with missing values was
generally <5%, except for physical activity where it was ∼12%.
Participants report lifestyle characteristics in the biennial ques- Missing values were assigned to a missing indicator variable.
tionnaires, which include current smoking status and the number For the sensitivity analysis, we created an alternative version
of cigarettes per day. Women also reported menopausal status and of the FGI using 1 serving/d as the cutoff (as originally desig-
menopausal hormone use, including the specific type and dura- nated by FANTA and WHO), as well as an alternative version
tion. We computed updated BMI using weight reported in each of the PDQS also with a 1 serving/d cutoff to mimic the same
2-y period and height reported at baseline. Average duration of scoring algorithm as the alternative FGI and MDDW. To exam-
leisure-time physical activity over the past year was assessed us- ine if the association was consistent among different age groups,
ing 10 questions that included the most common types of leisure we also stratified our analysis into <65.0 y, 65.0 to <75.0 y, and
time activities. Family history of MI was assessed in 1976, 1984, ≥75.0 y for the NHS and the HPFS. Because women in NHS
and 1996 in the NHS; in 1986 in the HPFS; and 1989, 1993, 1997, II were younger, the age groups were stratified into <55.0
2001, and 2005 in the NHS II. and ≥55.0 y. We also examined the association of individual
food groups, adjusting for other food groups and potential con-
founders, to understand if certain food groups may influence the
Statistical analysis direction of association of the diet quality scores. Analysis was
We calculated each participant’s person-years from the date of conducted using SAS version 9.4 (SAS Institute).
return of the baseline FFQ to the date of diagnosis of IHD, return
of the last questionnaire, death, or the end of follow-up (30 June
2012 for NHS, 31 January 2010 for HPFS, and 30 June 2011 for RESULTS
NHS II). In ≤28 y of follow-up, we ascertained 2908 incidents of IHD
We computed cumulative averages of each diet score to reduce in NHS, 3722 in HPFS, and 531 in NHS II. The Spearman corre-
intraperson variation and to better reflect long-term diet qual- lations between the diet quality scores were generally the highest
ity (22). However, we stopped updating dietary intake when a between the PDQS and MDDW (Spearman r = 0.62 in NHS,
participant reported a diagnosis of hypertension, hyperlipidemia, 0.59 in HPFS, and 0.66 in NHS II, all P < 0.001) and the lowest
angina, or diabetes, as individuals might improve their diet in between the FGI and the PDQS (Spearman r = 0.49 in NHS, 0.34
response to these diagnoses but diet might have limited impact in HPFS, and 0.50 in NHS II, all P < 0.001). Men and women
on IHD outcome at that stage. We categorized the FGI and the with a higher FGI score tended to have a slightly higher BMI,
MDDW scores into 5 categories. The cutoffs were determined to whereas those with a higher MDDW score and PDQS tended to
avoid having a small number of cases in each score category while have a somewhat lower BMI (Tables 1–3). On the other hand,
allowing for as many categories as possible to detect a change in individuals who scored higher in any of the food quality scores
risk with a change in score, especially in the upper range of the were less likely to be current smokers, had higher levels of phys-
score. For the FGI, many individuals in our cohorts achieved high ical activity, and had higher energy intake. Intake of specific food
TABLE 1
Age-standardized baseline characteristics of NHS, HPFS, and NHS II for the bottom, middle, and top categories of the FGI score1

NHS HPFS NHS II

Points 0 to <5.0 6.0 to <7.0 8.0 0 to <5.0 6.0 to <7.0 8.0 0 to < 5.0 6.0 to <7.0 ≥7.0
n 3500 22,377 13,340 2344 13,967 6218 6921 32,005 25,041
BMI, kg/m2 24.5 ± 4.5 25.0 ± 4.8 25.3 ± 4.8 24.7 ± 4.7 24.9 ± 4.9 25.2 ± 5.2 24.1 ± 5.2 24.7 ± 5.3 24.8 ± 5.4
Current smokers, % 33 25 19 9 10 10 16 12 10
Physical activity (METs) 12 ± 17 13 ± 20 16 ± 23 20 ± 28 21 ± 29 23 ± 33 20 ± 29 21 ± 27 23 ± 28
MDDW 2.7 ± 1.1 4.2 ± 1.2 5.5 ± 1.1 3.1 ± 1.3 4.3 ± 1.3 5.5 ± 1.2 2.4 ± 1.3 4.4 ± 1.3 5.1 ± 1.2
PDQS 16.0 ± 4.2 19.3 ± 4.2 24.1 ± 4.0 18.6 ± 5.1 20.2 ± 4.8 23.8 ± 3.8 16.8 ± 4.6 21.1 ± 4.4 23.9 ± 4.2
Energy intake, kcal/d 1170 ± 384 1652 ± 459 2093 ± 512 1407 ± 435 1896 ± 537 2443 ± 618 1233 ± 407 1830 ± 485 2144 ± 518
Fiber, g/d 14.7 ± 5.5 15.6 ± 4.8 18.5 ± 4.2 20.5 ± 9.3 20.5 ± 7.2 22.4 ± 5.4 17.4 ± 6.5 18.4 ± 5.4 20.0 ± 5.1
Alcohol intake, g/d 7.0 ± 12.5 7.0 ± 11.5 6.8 ± 10.7 10.8 ± 16.2 11.4 ± 15.4 11.7 ± 15.0 2.8 ± 5.9 3.1 ± 6.1 3.2 ± 6.1
Fruit, servings/d 1.3 ± 1.2 1.9 ± 1.3 2.9 ± 1.5 1.9 ± 1.6 2.2 ± 1.6 2.9 ± 1.6 1.2 ± 1.1 1.9 ± 1.3 2.5 ± 1.5
Vegetables, servings/d 2.1 ± 1.3 3.3 ± 1.6 5.4 ± 2.1 2.3 ± 1.7 2.8 ± 1.7 4.4 ± 1.9 1.1 ± 0.8 1.7 ± 1.1 2.2 ± 1.3
Legumes, servings/d 0.1 ± 0.1 0.2 ± 0.1 0.3 ± 0.2 0.2 ± 0.2 0.2 ± 0.3 0.4 ± 0.3 0.1 ± 0.2 0.2 ± 0.2 0.4 ± 0.3
Nuts, servings/d 0.1 ± 0.2 0.3 ± 0.4 0.5 ± 0.5 0.7 ± 1.5 1.5 ± 1.8 2.1 ± 1.7 0.1 ± 0.2 0.3 ± 0.3 0.4 ± 0.4
Grains and tubers, servings/d 3.8 ± 1.8 4.9 ± 2.1 5.6 ± 2.2 2.4 ± 1.7 3.3 ± 1.8 4.1 ± 1.9 2.6 ± 1.4 3.9 ± 1.7 4.6 ± 1.8
Eggs, servings/d 0.1 ± 0.2 0.3 ± 0.3 0.5 ± 0.3 0.1 ± 0.2 0.3 ± 0.4 0.6 ± 0.4 0.1 ± 0.1 0.2 ± 0.2 0.3 ± 0.2
Dairy, servings/d 0.4 ± 0.7 1.4 ± 1.0 1.9 ± 1.1 0.6 ± 1.0 1.4 ± 1.1 1.9 ± 1.3 0.7 ± 0.9 1.8 ± 1.2 2.2 ± 1.3
Animal flesh, servings/d 0.9 ± 0.5 1.3 ± 0.6 1.6 ± 0.6 1.2 ± 0.7 1.6 ± 0.7 2.1 ± 0.8 1.0 ± 0.6 1.5 ± 0.7 1.8 ± 0.7
Sweets and desserts, servings/d 0.8 ± 1.0 1.1 ± 1.2 1.3 ± 1.2 1.0 ± 1.2 1.4 ± 1.4 1.7 ± 1.5 0.8 ± 0.9 1.0 ± 1.0 1.1 ± 1.0
Sugar sweetened beverages, servings/d 0.3 ± 0.7 0.3 ± 0.6 0.3 ± 0.5 0.3 ± 0.7 0.4 ± 0.6 0.4 ± 0.5 0.5 ± 1.0 0.5 ± 0.8 0.5 ± 0.8
1 Values are means ± SDs unless otherwise indicated. NHS and HPFS: 5 categories total, categories 1, 3, 5 are shown; NHS II: 4 categories total, categories 1, 3, 4 are shown. FGI, Food Group Index;
FOOD GROUP DIET SCORES AND CORONARY HEART DISEASE

HPFS, Health Professionals’ Follow-up Study; MDDW, Minimal Diet Diversity score for Women; MET, metabolic equivalent task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; PDQS, Prime
Diet Quality Score.
123

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


124

TABLE 2
Age-standardized baseline characteristics of NHS, HPFS, and NHS II for the bottom, middle, and top categories of the MDDW score1

NHS HPFS NHS II

Points 0 to <3.0 4.0 to <5.0 >6.0 0 to <3.0 4.0 to <5.0 >6.0 0 to <3.0 4.0 to <5.0 >6.0
n 6746 17,628 18,010 3771 11,306 10,298 12,956 22,857 16,576
BMI, kg/m2 24.6 ± 4.6 24.9 ± 4.7 25.2 ± 4.8 24.9 ± 5.1 24.9 ± 5.1 24.9 ± 5.0 24.4 ± 5.4 24.6 ± 5.3 24.8 ± 5.2
Current smokers, % 35 25 19 15 10 8 17 12 9
Physical activity (METs) 10 ± 18 13 ± 19 18 ± 25 16 ± 24 20 ± 27 26 ± 35 16 ± 24 20 ± 26 28 ± 32
FGI 5.1 ± 1.0 6.2 ± 1.0 7.1 ± 0.8 5.3 ± 1.0 6.2 ± 1.0 7.0 ± 0.9 4.9 ± 0.9 5.8 ± 0.9 6.6 ± 0.9
PDQS 15.2 ± 3.8 19.4 ± 3.9 24.1 ± 3.9 16.1 ± 4.3 20.0 ± 4.2 24.4 ± 4.0 15.8 ± 4.1 20.1 ± 4.0 25.4 ± 3.6
Fiber, g/d 13.2 ± 4.1 15.8 ± 4.5 18.6 ± 4.9 16.5 ± 6.0 20.1 ± 6.3 24.1 ± 7.2 15.5 ± 4.8 17.9 ± 5.3 21.7 ± 5.6
Energy intake, kcal/d 1219 ± 386 1652 ± 256 2072 ± 509 1373 ± 404 1874 ± 507 2438 ± 606 1230 ± 379 1760 ± 447 2210 ± 510
Alcohol intake, g/d 6.6 ± 12.0 6.8 ± 6.4 7.1 ± 10.6 11.5 ± 16.3 11.3 ± 15.5 11.5 ± 15.0 2.8 ± 6.1 3.1 ± 6.1 3.5 ± 6.1
Fruit, servings/d 0.9 ± 0.7 1.9 ± 1.2 3.0 ± 1.5 1.1 ± 1.2 2.2 ± 1.4 3.3 ± 1.9 0.8 ± 0.6 1.8 ± 1.1 2.9 ± 1.6
FUNG ET AL.

Vegetables, servings/d 1.7 ± 0.7 3.2 ± 1.3 5.6 ± 2.3 1.4 ± 0.7 2.6 ± 1.2 4.8 ± 2.2 0.7 ± 0.5 1.5 ± 0.8 2.8 ± 1.5
Legumes, servings/d 0.1 ± 0.1 0.2 ± 0.2 0.3 ± 0.2 0.1 ± 0.1 0.2 ± 0.2 0.4 ± 0.4 0.1 ± 0.1 0.2 ± 0.2 0.3 ± 0.4
Nuts, servings/d 0.2 ± 0.2 0.3 ± 0.4 0.5 ± 0.7 1.1 ± 1.5 1.4 ± 1.7 2.1 ± 2.0 0.1 ± 0.2 0.2 ± 0.3 0.4 ± 0.5
Grains and tubers, servings/d 4.1 ± 1.9 4.9 ± 2.1 5.5 ± 2.2 2.3 ± 1.5 3.3 ± 1.7 4.2 ± 2.0 2.6 ± 1.4 3.8 ± 1.6 4.7 ± 1.8
Eggs, servings/d 0.2 ± 0.2 0.3 ± 0.3 0.5 ± 0.4 0.2 ± 0.3 0.3 ± 0.4 0.5 ± 0.6 0.1 ± 0.1 0.2 ± 0.2 0.2 ± 0.3
Dairy, servings/d 0.6 ± 0.6 1.3 ± 1.0 2.1 ± 1.1 0.7 ± 0.7 1.3 ± 1.1 2.0 ± 1.3 0.8 ± 0.8 1.7 ± 1.2 2.3 ± 1.2
Animal flesh, servings/d 0.9 ± 0.4 1.3 ± 0.6 1.6 ± 0.6 1.2 ± 0.6 1.7 ± 0.7 2.0 ± 0.9 1.0 ± 0.5 1.5 ± 0.7 1.8 ± 0.7
Sweets and desserts, servings/d 1.0 ± 1.1 1.1 ± 1.1 1.2 ± 1.2 1.2 ± 1.3 1.5 ± 1.4 1.6 ± 1.5 0.9 ± 1.0 1.0 ± 1.0 1.0 ± 1.0
Sugar sweetened beverages, servings/d 0.4 ± 0.7 0.3 ± 0.6 0.2 ± 0.5 0.4 ± 0.7 0.4 ± 0.6 0.3 ± 0.6 0.5 ± 1.0 0.5 ± 0.9 0.4 ± 0.7
1 Values are means ± SDs unless otherwise indicated. Five categories total, categories 1, 3, 5 are shown. FGI, Food Group Index; HPFS, Health Professionals’ Follow-up Study; MDDW, Minimal Diet

Diversity score for Women; MET, metabolic equivalent task; NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; PDQS, Prime Diet Quality Score.

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


TABLE 3
Age-standardized baseline characteristics of NHS, HPFS, and NHS II for quintiles of the PDQS1

NHS HPFS NHS II

Q1 Q3 Q5 Q1 Q3 Q5 Q1 Q3 Q5
BMI, kg/m2 25.1 ± 5.1 24.9 ± 4.6 24.9 ± 4.5 25.1 ± 5.2 25.0 ± 5.0 24.5 ± 5.0 25.1 ± 6.0 24.6 ± 5.3 24.2 ± 4.8
Current smokers, % 35 23 16 17 9 5 19 11 8
Physical activity (METs) 10 ± 17 14 ± 18 20 ± 29 15 ± 23 21 ± 29 29 ± 37 14 ± 20 20 ± 25 30 ± 34
FGI 5.6 ± 1.0 6.4 ± 1.1 7.1 ± 0.9 5.8 ± 1.0 6.4 ± 1.1 6.7 ± 1.0 5.2 ± 0.9 5.9 ± 1.0 6.6 ± 0.9
MDDW 3.3 ± 1.2 4.5 ± 1.2 5.6 ± 1.1 3.3 ± 1.2 4.5 ± 1.2 5.6 ± 1.3 2.8 ± 1.2 4.2 ± 1.2 5.5 ± 1.1
PDQS 13.8 ± 2.2 20.5 ± 0.5 27.0 ± 2.0 13.8 ± 2.2 21.0 ± 0.8 28.0 ± 1.9 13.6 ± 2.3 21.0 ± 0.8 27.8 ± 1.8
Fiber, g/d 12.8 ± 3.3 16.2 ± 3.9 20.2 ± 5.1 15.4 ± 4.3 20.5 ± 5.5 27.3 ± 7.5 14.0 ± 3.4 18.2 ± 4.4 23.2 ± 5.9
Energy intake, kcal/d 1657 ± 525 1736 ± 535 1852 ± 526 1878 ± 591 2011 ± 631 2091 ± 618 1638 ± 537 1784 ± 541 1987 ± 527
Alcohol intake, g/d 6.5 ± 12.0 7.0 ± 11.2 6.8 ± 10.4 11.9 ± 17.2 11.6 ± 15.6 10.4 ± 13.7 2.5 ± 6.0 3.2 ± 6.0 3.7 ± 6.4
Fruit, servings/d 1.2 ± 1.0 2.1 ± 1.3 3.1 ± 1.5 1.3 ± 1.1 2.4 ± 1.4 3.6 ± 2.0 0.9 ± 0.8 1.9 ± 1.2 2.9 ± 1.6
Vegetables, servings/d 2.3 ± 1.2 3.6 ± 1.5 5.5 ± 2.4 1.7 ± 0.9 3.0 ± 1.4 5.0 ± 2.3 0.9 ± 0.6 1.6 ± 0.9 2.6 ± 1.4
Legumes, servings/d 0.2 ± 0.1 0.2 ± 0.2 0.3 ± 0.2 0.2 ± 0.2 0.3 ± 0.2 0.4 ± 0.4 0.1 ± 0.1 0.2 ± 0.2 0.4 ± 0.4
Nuts, servings/d 0.2 ± 0.3 0.3 ± 0.4 0.5 ± 0.6 1.3 ± 1.7 1.6 ± 1.8 1.8 ± 1.8 0.2 ± 0.3 0.2 ± 0.3 0.4 ± 0.4
Grains and tubers, servings/d 5.2 ± 2.1 5.1 ± 2.2 4.8 ± 2.1 3.2 ± 1.8 3.5 ± 1.9 3.7 ± 2.0 3.3 ± 1.6 3.8 ± 1.7 4.3 ± 1.8
Eggs, servings/d 0.3 ± 0.3 0.4 ± 0.3 0.4 ± 0.3 0.3 ± 0.4 0.3 ± 0.4 0.3 ± 0.4 0.2 ± 0.2 0.2 ± 0.2 0.2 ± 0.2
Dairy, servings/d 1.0 ± 0.9 1.5 ± 1.1 2.0 ± 1.2 1.2 ± 1.1 1.5 ± 1.2 1.7 ± 1.2 1.2 ± 1.1 1.8 ± 1.2 2.2 ± 1.3
Animal flesh, servings/d 1.3 ± 0.6 1.3 ± 0.6 1.5 ± 0.7 1.8 ± 0.9 1.7 ± 0.8 1.6 ± 0.8 1.5 ± 0.7 1.5 ± 0.7 1.5 ± 0.7
Sweets and desserts, servings/d 1.3 ± 1.2 1.2 ± 1.1 0.9 ± 1.0 1.8 ± 1.6 1.6 ± 1.4 1.0 ± 1.2 1.2 ± 1.1 1.0 ± 1.0 0.8 ± 0.9
Sugar sweetened beverages, servings/d 0.6 ± 0.9 0.3 ± 0.5 0.1 ± 0.3 0.6 ± 0.8 0.3 ± 0.5 0.1 ± 0.3 0.9 ± 1.2 0.4 ± 0.7 0.2 ± 0.4
1 Values are means ± SDs unless otherwise indicated. FGI, Food Group Index; HPFS, Health Professionals’ Follow-up Study; MDDW, Minimal Diet Diversity Score for Women; MET, metabolic task
FOOD GROUP DIET SCORES AND CORONARY HEART DISEASE

equivalent; NHS, Nurses’ Health Study; NHS II: Nurses’ Health Study II; PDQS, Prime Diet Quality Score; Q, quintile.
125

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


126 FUNG ET AL.
TABLE 4
RRs (95% CIs) for diet quality scores and total IHD1

Score category

1 2 3 4 5 Point increase
FGI2
NHS
Cases, n 153 636 1163 785 171
Age and energy adjusted 1 0.89 (0.74, 1.06) 0.90 (0.75, 1.08) 0.84 (0.70, 1.02) 1.00 (0.79, 1.26) 0.93 (0.89, 0.98)
Multivariable adjusted3 1 0.94 (0.79, 1.13) 0.97 (0.81, 1.16) 0.90 (0.75, 1.09) 0.94 (0.74, 1.19) 0.96 (0.92, 1.01)
HPFS
Cases, n 208 744 1404 1067 299
Age and energy adjusted 1 0.89 (0.76, 1.04) 0.90 (0.77, 1.05) 0.95 (0.81, 1.12) 1.14 (0.94, 1.38) 1.02 (0.98, 1.06)
Multivariable adjusted3 1 0.91 (0.78, 1.06) 0.91 (0.78, 1.06) 0.93 (0.79, 1.10) 1.04 (0.86, 1.27) 1.01 (0.97, 1.05)
NHS II
Cases, n 53 214 185 79 —

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


Age and energy adjusted 1 1.04 (0.76, 1.42) 0.84 (0.60, 1.18) 0.78 (0.52, 1.15) — 0.85 (0.76, 0.96)
Multivariable adjusted3 1 1.06 (0.78, 1.45) 0.91 (0.65, 1.27) 0.83 (0.56, 1.24) — 0.91 (0.81, 1.02)
MDDW4
NHS
Cases, n 252 519 936 821 380
Age and energy adjusted 1 0.88 (0.76, 1.03) 0.89 (0.76, 1.03) 0.78 (0.66, 0.91) 0.73 (0.61, 0.88) 0.90 (0.87, 0.94)
Multivariable adjusted3 1 0.97 (0.83, 1.13) 1.01 (0.87, 1.18) 0.90 (0.76, 1.05) 0.84 (0.70, 1.01) 0.95 (0.91, 0.99)
HPFS
Cases, n 274 679 1070 1022 677
Age and energy adjusted 1 1.04 (0.90, 1.20) 0.93 (0.81, 1.08) 0.92 (0.79, 1.06) 0.87 (0.74, 1.02) 0.94 (0.91, 0.97)
Multivariable adjusted3 1 1.07 (0.93, 1.24) 0.98 (0.85, 1.13) 0.97 (0.83, 1.12) 0.92 (0.78, 1.08) 0.96 (0.93, 0.99)
NHS II
Cases, n 82 115 165 120 49
Age and energy adjusted 1 0.85 (0.63, 1.14) 0.83, 0.61, 1.12) 0.69, 0.49, 0.96) 0.58 (0.38, 0.88) 0.84 (0.77, 0.91)
Multivariable adjusted3 1 0.92 (0.69, 1.24) 1.01 (0.75, 1.37) 0.91 (0.65, 1.27) 0.79 (0.52, 1.20) 0.93 (0.85, 1.01)
PDQS5
NHS
Cases, n 684 599 627 525 473
Age and energy adjusted 1 0.77 (0.69, 0.86) 0.73 (0.65, 0.81) 0.59 (0.53, 0.66) 0.51 (0.45, 0.57) 0.76 (0.73, 0.79)
Multivariable adjusted3 1 0.88 (0.79, 0.98) 0.89 (0.79, 0.99) 0.77 (0.68, 0.87) 0.70 (0.61, 0.79) 0.87 (0.83, 0.91)
HPFS
Cases, n 723 775 795 779 650
Age and energy adjusted 1 0.90 (0.81, 0.99) 0.84 (0.76, 0.93) 0.79 (0.71, 0.87) 0.67 (0.60, 0.75) 0.85 (0.82, 0.88)
Multivariable adjusted3 1 0.96 (0.87, 1.06) 0.93 (0.84, 1.03) 0.91 (0.82, 1.01) 0.79 (0.70, 0.88) 0.90 (0.87, 0.94)
NHS II
Cases, n 142 121 98 99 71
Age and energy adjusted 1 0.79 (0.62, 1.00) 0.61 (0.47, 0.79) 0.57 (0.44, 0.74) 0.39 (0.29, 0.53) 0.70 (0.63, 0.77)
Multivariable adjusted3 1 0.95 (0.74, 1.21) 0.80 (0.61, 1.04) 0.83 (0.63, 1.09) 0.63 (0.46, 0.86) 0.85 (0.76, 0.94)
1 FGI, Food Group Index; HPFS, Health Professionals’ Follow-up Study; IHD, ischemic heart disease; MDDW, Minimal Diet Diversity Score for Women;

NHS, Nurses’ Health Study; NHS II, Nurses’ Health Study II; PDQS: Prime Diet Quality Score; Q, quintile.
2 Score categories for NHS and HPFS: 1, 0 to <5.0; 2, 5.0 to <6.0; 3, 6.0 to <7.0; 4, 7.0 to <8.0; 5, 8.0. Score categories for NHS II: 1, 0 to <5.0; 2, 5.0

to <6.0; 3, 6.0 to <7.0; 4, ≥7.0; 5, none. Point increase: 1.


3 Adjusted for age; energy intake; smoking status; alcohol intake; physical activity; BMI; family history of myocardial infarction; multivitamin use; aspirin

use; baseline history of diabetes, hypertension, or hyperlipidemia; menopausal status, and postmenopausal hormone use (NHS only); and oral contraceptive
use (NHS II only). Data analyzed with Cox proportional hazard models.
4 Score categories: 1, 0 to <3.0; 2, 3.0 to <4.0; 3, 4.0 to <5.0; 4, 5.0 to <6.0; 5, ≥6.0. Point increase: 1.
5 Score categories: quintiles (Q1–Q5).

groups increased in the higher categories of any of the food qual- sitivity analysis, we created an alternative version of the FGI
ity scores except for the PDQS, in which the increase of animal in which points were awarded for consumption of 1 time/d for
flesh intake was not consistently associated with the food quality each food group instead of 3 times/wk. This version of the FGI
score. was also not associated with total IHD in any of the cohorts
In the age- and energy-adjusted model, the FGI was not asso- (Supplemental Table 4).
ciated with total, nonfatal or fatal IHD in either men or women The MDDW showed a modest inverse association with to-
(Table 4, Supplemental Tables 2 and 3). After adjusting for po- tal IHD and nonfatal IHD in NHS and HPFS (total IHD:
tential confounders, the association remained null. In the sen- RR for 1-point increase = 0.95 for NHS and 0.96 for HPFS,
FOOD GROUP DIET SCORES AND CORONARY HEART DISEASE 127
95% CI: 0.91, 0.99 for NHS, 0.93, 0.99 for HPFS). On the other DISCUSSION
hand, we observed a clear inverse association between the PDQS In this analysis, the FGI, which was constructed from 8 broad
and total, nonfatal, and fatal IHD in all 3 cohorts. For total IHD, food groups, showed no association with IHD in both young
RRs comparing top to bottom quintiles were 0.70 (95% CI: 0.61, and older men and women. The MDDW, constructed from 10
0.79) for NHS, 0.79 (95% CI: 0.70, 0.88) for HPFS, and 0.63 food groups with more detailed differentiation of plant foods,
(95% CI: 0.46, 0.86) for NHS II. RRs for nonfatal and fatal IHD was modestly associated with a lower risk of IHD. The PDQS,
were similar. composed of 21 food groups that account for both healthy and
After pooling data from all 3 cohorts, each SD increase in the unhealthy food consumption, was clearly associated with lower
PDQS was associated with an RR of 0.89 (95% CI: 0.87, 0.91). risk of IHD.
This was significantly different than the pooled RR of 0.93 for The FGI and MDDW consist of a small number of nonspecific
MDDQ (95% CI: 0.90, 0.96) and the RR of 0.98 for the FGI (95% food groups, and scoring was based on presence or absence of
CI: 0.95, 1.01) (P difference = 0.02 for PDQS compared with consumption in the previous 24 h. While the scores achieve sim-
MDDQ, and <0.0001 compared with FGI). The pooled RR for 1 plicity, speed in administration, and broad applicability in differ-
SD increase of MDDQ was also significantly different from that ent countries with these characteristics, it would be more difficult
of the FGI (P difference = 0.02).

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


to detect variability in diet quality in countries with more varied
When we stratified the analysis by age to explore the hetero- dietary patterns. We are not surprised by their limited associa-
geneity of these scores by different age groups, we observed sim- tion with IHD in our cohorts. As shown in our examination of
ilar results that showed the FGI was not associated with total IHD components of FGI and MDDW, these diet quality scores con-
in any age groups (Supplemental Table 5). The RRs for MDDW tain food groups that showed a direct and inverse association
were similar across age groups and similar to the main analysis, with IHD. Therefore, while easy to administer, they do not ap-
but did not reach statistical significance. The PDQS was inversely pear to be sufficiently sensitive to quantify diet quality in popula-
associated with total IHD to a similar extent in all age groups. The tions where overconsumption is a significant concern. The PDQS
test for interaction was not significant in any of the age-stratified maintained a reasonable level of simplicity, exemplified by the
analyses. FGI and MDDW, but expanded the number of food groups to 21.
When examining individual food groups, animal flesh was as- In addition, the food groups included both healthy and unhealthy
sociated with a higher risk of total IHD in men (RR for each components, and the scoring algorithm was designed to encour-
serving: 1.03; 95% CI: 1.00, 1.06) and older women (RR: 1.14; age an overall healthy diet. The PDQS also has greater gradation
95% CI: 1.05, 1.22), but the association did not reach statis- of scoring in order to better categorize the healthfulness of the
tical significance in younger women (Table 5, Supplemental diet yet keep the administration procedure manageable. Our re-
Table 6). On the other hand, an inverse association was observed sults showed that the PDQS is better able to predict IHD risk than
with nuts in NHS (RR: 0.87; 95% CI: 0.77, 0.97), and leafy veg- the FGI and MDDW in both young and older men and women.
etables in HPFS (RR: 0.88; 95% CI: 0.82, 0.95). A number of indexes created specifically to measure diet qual-
ity have long been available, and many were associated with

TABLE 5
Multivariable RR (95% CI) of 1 serving of score components of FGI and MDDW scores for total IHD1

Score components NHS HPFS NHS II


Food groups common to both FGI and MDDW
Grains and tubers 1.00 (0.97, 1.02) 0.98 (0.96, 1.00) 0.92 (0.85, 1.00)
Animal flesh 1.13 (1.05, 1.22) 1.03 (1.00, 1.06) 1.11 (0.94, 1.32)
Dairy 1.01 (0.96, 1.03) 0.99 (0.96, 1.03) 0.97 (0.88, 1.07)
Eggs 1.09 (0.95, 1.27) 1.05 (0.96, 1.14) 0.84 (0.54, 1.31)
High–β carotene fruits and vegetables 0.87 (0.77, 0.98) 0.96 (0.87, 1.06) 1.11 (0.87, 1 42)
Food groups unique to FGI
Legume and nuts 0.94 (0.86, 1.02) 0.98 (0.94, 1.04) 0.87 (0.68, 1.12)
Other fruits and vegetables2 1.00 (0.97, 1.02) 0.99 (0.97, 1.01) 0.95 (0.90, 1.00)
Added fat 1.00 (0.96, 1.03) 1.01 (0.98, 1.05) 1.07 (0.98, 1.16)
Food groups unique to MDDW
Legumes 1.12 (0.90, 1.40) 0.92 (0.82, 1.03) 0.89 (0.55, 1.43)
Nuts 0.87 (0.77, 0.97) 0.99 (0.94, 1.05) 0.81 (0.59, 1.12)
Leafy vegetables 0.92 (0.84, 1.00) 0.88 (0.82, 0.95) 0.91 (0.73, 1.13)
Other vegetables2 1.03 (0.98, 1.08) 1.05 (1.01, 1.10) 0.96 (0.86, 1.08)
Other fruits2 1.02 (0.98, 1.05) 0.99 (0.97, 1.03) 0.94 (0.85, 1.04)
1 Adjusted for age; energy intake; smoking status; alcohol intake; physical activity; BMI; family history of myocardial infarction; multivitamin use; aspirin

use; baseline history of diabetes, hypertension, or hyperlipidemia; menopausal status and postmenopausal hormone use (NHS only); and oral contraceptive use
(NHS II only). Food groups in the same diet quality scores are mutually adjusted for each other. Data analyzed with Cox proportional hazard models. FGI, Food
Group Index; HPFS, Health Professionals’ Follow-up Study; IHD, ischemic heart disease; MDDW, Minimal Diet Diversity score for Women; NHS, Nurses’
Health Study; NHS II, Nurses’ Health Study II.
2 Other than high–β carotene fruits and vegetable.
128 FUNG ET AL.

disease risk in higher-income countries. However, many were with a lower risk of IHD in a high-income country. On the other
not created with ease of administration or scoring in mind. The hand, crude diet quality scores that include a smaller number of
Healthy Nordic Diet score consists of 6 components that reflect food groups and do not account for unhealthy foods had limited
healthy food in a traditional Nordic diet (24). Although the num- (MDDW) or no (FGI) association with IHD.
ber of components is manageable and the score is associated with
The authors’ contributions are as follows—TTF: conducted the statistical
a lower risk of mortality, assigning points for each food group analysis and wrote the manuscript, and is fully responsible for the results and
depends on population-specific intake level, thus making appli- writing of the manuscript; and all authors: contributed to the design of the
cation in different countries complicated. In addition, the food statistical analysis, critiqued and revised the manuscript, had full access to
groups are specific to those consumed in northern Europe only. the data, and read and approved the manuscript. None of the authors reported
The Elderly Dietary Index consists of a slightly larger number of a conflict of interest related to the study.
food groups (10 groups) and uses 4 levels of intake for scoring
(25). This score has been associated with a lower risk of mortal-
ity in elderly British men. While the scoring algorithm allows for REFERENCES
1. United Nations. Internet: http://www.un.org/sustainabledevelopment/
differentiation between low and high intake of each food group,
sustainable-development-goals/ (accessed 10 June 2016).
whole compared with refined grains, and red or processed meats

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021


2. Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang M,
compared with poultry were not separately assessed. Other diet Stampfer MJ, Willett WC. Alternative dietary indices both strongly
quality scores, such as the Healthy Eating Index and the Alternate predict risk of chronic disease. J Nutr 2012;142:1009–18.
3. Guenther PM, Casavale KO, Reedy J, Kirkpatrick SI, Hiza HAB,
Healthy Eating Index, require the assessment of both foods and
Kuczynski KJ, HKahle LL, Krebs-Smith SM. Healthy eating index.
nutrients (2, 3). The additional complexity in the scoring proce- Center for Nutrition Policy and Promotion, USDA; 2013.
dure thus limits their use in settings where speed and a low tech- 4. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to
nology burden are essential. a Mediterranean diet and survival in a Greek population. N Engl J Med
2003;348:2599–608.
Given the characteristics of current dietary quality scores that
5. Romaguera D, Vergnaud A-C, Peeters PH, van Gils CH, Chan DS,
are predictive of chronic diseases, a measurement tool that is sim- Ferrari P, Romieu I, Jenab M, Slimani N, Clavel-Chapelon F, et al. Is
pler and applicable across different countries is needed. Future concordance with World Cancer Research Fund/American Institute for
studies should investigate if the PDQS and other simple diet qual- Cancer Research guidelines for cancer prevention related to subsequent
risk of cancer? Results from the EPIC study. Am J Clin Nutr
ity scores are associated with a wide range of health outcomes rel-
2012;96:150–63.
evant to diverse populations in both high- and low-income coun- 6. Sofi F, Macchi C, Abbate R, Gensini GF, Casini A. Mediterranean
tries. This would then provide an efficient field tool without the diet and health status: an updated meta-analysis and a proposal for a
need for elaborate diet assessment methods that require extensive literature-based adherence score. Pub Health Nutr 2014;17:2769–82.
7. Buckland G, Agudo A, Luján L, Jakszyn P, Bueno-de-Mesquita HB,
development or data analysis such as the FFQ and 24-h recall.
Palli D, Boeing H, Carneiro F, Krogh V, Sacerdote C, et al. Adherence
A major strength of this analysis is the availability of repeated to a Mediterranean diet and risk of gastric adenocarcinoma within the
measurements of diet. This allowed us to take into account di- European Prospective Investigation into Cancer and Nutrition (EPIC)
etary changes during the long follow-up period. However, for cohort study. Am J Clin Nutr 2010;91:381–90.
8. Fung TT, Rexrode KM, Mantzoros CS, Manson JE, Willett WC, Hu
individuals who developed intermediate endpoints such as hy-
FB. Mediterranean diet and incidence of and mortality from coronary
perlipidemia and hypertension, we stopped updating the dietary heart disease and stroke in women. Circulation 2009;119:1093–
data because they may change their diet upon these diagnoses. 100.
Another strength is the availability of detailed and updated in- 9. McNaughton SA, Bates CJ, Mishra GD. Diet Quality Is Associated
with all-cause mortality in adults aged 65 years and older. J Nutr
formation on lifestyle and health information, which we used to
2012;142:320–5.
finely adjust for potential confounders. Nonetheless, since this 10. FANTA. Developing and validating simple indicators of dietary quality
is an observational study, residual confounding cannot be com- and energy intake of infants and young children in developing countries:
pletely ruled out. In addition, because health and dietary infor- summary of findings from analysis of 10 data sets. Washington (DC);
2006.
mation was self-reported, some degree of misreporting cannot be
11. FAO, 360 F. Minimum dietary diversity for women: a guide for
avoided. However, we have documented reliable reporting from measurement. Rome: FAO; 2016.
the validated FFQ compared to multiple-week diet records (16). 12. Martin-Prevel Y, Allemand P, Wiesmann D, Arimond M, Ballard T,
For the construction of the PDQS, we assigned equal weight to Deitchler M, Dop Mc, Kennedy G, Lee WTK, Moursi M. Moving
forward on choosing a standard operational indicator of women’s
each food group, though it is unlikely that each of them make
dietary diversity. Rome: FAO; 2015.
an equal contribution to the development of IHD. However, our 13. Colditz GA, Martin P, Stampfer MJ, Willett WC, Sampson L, Rosner
goal was to evaluate an assessment score that would be applicable B, Hennekens CH, Speizer FE. Validation of questionnaire information
to other noncommunicable diseases as well, for which different on risk factors and disease outcomes in a prospective cohort study of
women. Am J Epidemiol 1986;123:894–900.
food groups would have different amounts of influence. There-
14. Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB,
fore, to maintain simplicity in score construction, we did not at- Willett WC. Reproducibility and validity of an expanded self-
tempt to assign different weights to food groups. All 3 cohorts in administered semiquantitative food frequency questionnaire among
this analysis are health professionals, therefore our results need male health professionals. Am J Epidemiol 1992;135):1114–26.
15. Schulze MB, Manson JE, Willett WC, Hu FB. Processed meat intake
to be confirmed in other populations, especially those outside the
and incidence of Type 2 diabetes in younger and middle-aged women.
United States and lower-income countries. In addition, further Diabetologia 2003;46:1465–73.
examination of the PDQS and other health outcomes is necessary 16. Salvini S, Hunter DJ, Sampson L, Stampfer MJ, Colditz GA, Rosner
to evaluate the potential of this diet quality score to be used as an B, Willett WC. Food-based validation of a dietary questionnaire: the
effects of week-to-week variation in food consumption. Int J Epidemiol
assessment tool for the UN Sustainable Development Goals.
1989;18:858–67.
In conclusion, we found that the PDQS with most detailed 17. Indicators. WGoIaYCF. Developing and validating simple indicators
differentiation of healthy and unhealthy foods was associated of dietary quality and energy intake of infants and young children in
FOOD GROUP DIET SCORES AND CORONARY HEART DISEASE 129
developing countries: summary of findings from analysis of 10 data sets. 21. Rose D, Meershoek S, Ismael C, McEwan M. Evaluation of a rapid field
Washington (DC): Food and Nutrition Technical Assistance Project tool for assessing household diet quality in Mozambique. Food Nutr
(FANTA); 2006. Bull 2002;23:181–9.
18. Arimond M, Wiesmann D, Becquey E, Carriquiry A, Daniels MC, 22. Hu FB, Stampfer MJ, Rimm E, Ascherio A, Rosner BA, Spiegelman
Deitchler M, Fanou-Fogny N, Joseph ML, Kennedy G, Martin-Prevel D, Willett WC. Dietary fat and coronary heart disease: a comparison of
Y, et al. Simple Food Group Diversity Indicators Predict Micronutrient approaches for adjusting for total energy intake and modeling repeated
Adequacy of Women’s Diets in 5 Diverse, Resource-Poor Settings. J dietary measurements. Am J Epidemiol 1999;149:531–40.
Nutr 2010;140:2059S–69S. 23. DerSimonian R, Laird N. Meta-analysis in clinical trials. Controlled
19. Food and Nutrition Technical Assistance III Project (FANTA). Clin Trials 1986;7:177–88.
Measuring the quality of women’s diet: Consensus on a global 24. Olsen A, Egeberg R, Halkjær J, Christensen J, Overvad K, Tjønneland
indicator for women’s dietary. Washington D.C., 2005. no. FHI A. Healthy aspects of the Nordic diet are related to lower total mortality.
360/FANTA. J Nutr 2011;141:639–44.
20. Rifas-Shiman SL, Willett WC, Lobb R, Kotch J, Dart C, Gillman 25. Atkins JL, Whincup PH, Morris RW, Lennon LT, Papacosta O,
MW. PrimeScreen, a brief dietary screening tool: reproducibility and Wannamethee SG. High diet quality is associated with a lower risk
comparability with both a longer food frequency questionnaire and of cardiovascular disease and all-cause mortality in older men. J Nutr
biomarkers. Pub Health Nutr 2001;4:249–54. 2014;144:673–80.

Downloaded from https://academic.oup.com/ajcn/article/107/1/120/4825209 by guest on 19 April 2021

You might also like