Adapted dietary inflammatory index and its association with a summary
score for low-grade inflammation and markers of glucose metabolism:
the Cohort study on Diabetes and Atherosclerosis Maastricht (CODAM)
and the Hoorn study1–4
Geertruida J van Woudenbergh, Despoina Theofylaktopoulou, Anneleen Kuijsten, Isabel Ferreira,
Marleen M van Greevenbroek, Carla J van der Kallen, Casper G Schalkwijk, Coen DA Stehouwer, Marga C Ocke´,
Giel Nijpels, Jacqueline M Dekker, Ellen E Blaak, and Edith JM Feskens
ABSTRACT
Background: Diet may be associated with the development of type
2 diabetes through its effects on low-grade inflammation.
Objectives: We investigated whether an adapted dietary inflamma-
tory index (ADII) is associated with a summary score for low-grade
inflammation and markers of glucose metabolism. In addition, we
investigated the mediating role of inflammation in the association
between ADII and markers of glucose metabolism.
Design: We performed cross-sectional analyses of 2 Dutch cohort
studies (n= 1024). An ADII was obtained by multiplying standard-
ized energy-adjusted intakes of dietary components by literature-
based dietary inflammatory weights that reflected the inflammatory
potential of components. Subsequently, these multiplications were
summed. Six biomarkers of inflammation were compiled in a sum-
mary score. Associations of the ADII (expressed per SD) with the
summary score for inflammation and markers of glucose metabo-
lism were investigated by using multiple linear regression models.
Inflammation was considered a potential mediator in the analysis
with markers of glucose metabolism.
Results: A higher ADII was associated with a higher summary
score for inflammation [b-adjusted = 0.04 per SD (95% CI: 0.01,
0.07 per SD)]. The ADII was also adversely associated with insulin
resistance [homeostatic model assessment of insulin resistance
(HOMA-IR): b-adjusted = 3.5% per SD (95% CI: 0.6%, 6.3% per
SD)]. This association was attenuated after the inclusion of the
summary score for inflammation [b-adjusted+inflammation =
2.2% (95% CI: 20.6%, 5.0%)]. The ADII was also adversely as-
sociated with fasting glucose and postload glucose but not with
glycated hemoglobin.
Conclusion: The significant mediating role of low-grade inflammation
in the association between the ADII and HOMA-IR suggests that in-
flammation might be one of the pathways through which diet affects
insulin resistance. Am J Clin Nutr 2013;98:1533–42.
INTRODUCTION
Chronic low-grade inflammation is characterized by slightly
elevated concentrations of circulating proinflammatory markers,
including C-reactive protein (CRP)5, IL-6, and TNF-a. These
markers have been associated with higher risk of type 2 diabetes
in observational studies (1, 2) and can be induced by risk factors
for type 2 diabetes, such as overweight (3), physical inactivity (4),
Am J Clin Nutr 2013;98:1533–42. Printed in USA. Ó 2013 American Society for Nutrition
and diet (3). Nutrients assumed to have an antiinflammatory ef-
fect, eg, fiber and moderate amounts of ethanol (3), may be as-
sociated with lower risk of diabetes (5, 6). In contrast, nutrients
Downloaded from ajcn.nutrition.org by guest on June 14, 2017
assumed to have a proinflammatory effect, eg, SFAs and trans
fatty acids (3), may be associated with higher risk of diabetes (7).
Therefore, it could be hypothesized that certain nutrients may
influence markers of glucose metabolism through their effect on
chronic low-grade inflammation. However, nutrients are not con-
sumed as individual components but with others present within
1
From the Division of Human Nutrition, Wageningen University, Wage-
ningen, Netherlands (GJvW, DT, AK, and EJMF); the CARIM School for
Cardiovascular Diseases (IF, MMvG, CJvdK, CGS, and CDAS), the Depart-
ment of Human Biology (EEB) and the NUTRIM School for Toxicology,
Metabolism, and Nutrition (EEB), Maastricht University, Maastricht Nether-
lands; the Departments of Internal Medicine (IF, MMvG, CJvdK, CGS, and
CDAS) and Clinical Epidemiology and Medical Technology Assessment
(IF), Maastricht University Medical Center, Maastricht, Netherlands; the
National Institute for Public Health and the Environment, Bilthoven, Nether-
lands (MCO); and the Departments of General Practice (GN) and Epidemiology
and Biostatistics (JMD), EMGO Institute for Health and Care Research, Vrije
Universiteit University Medical Center, Amsterdam, Netherlands.
2
Funding sources were not involved in the conduct of this study or writing
of the manuscript.
3
The Hoorn study was supported by grants from the Dutch Diabetes Re-
search Foundation, the Dutch Organization for Scientific Research, the Neth-
erlands Heart Foundation, and the Health Research and Development
Council of the Netherlands. The Cohort study on Diabetes and Atheroscle-
rosis Maastricht was supported by grants from the Netherlands Organization
for Scientific Research (940-35-034) and the Dutch Diabetes Foundation (98.
901). IF is supported by the Dutch Heart Foundation (senior postdoc research
grant 2006T06).
4
Address reprint requests and correspondence to GJ van Woudenbergh,
Division of Human Nutrition, Wageningen University, PO Box 8129, 6700
EV Wageningen, Netherlands. E-mail: truus.vanwoudenbergh@wur.nl.
5
Abbreviations used: ADII, adapted dietary inflammatory index; AHEI, al-
ternative healthy eating index; CODAM, Cohort study on Diabetes and Ath-
erosclerosis Maastricht; CRP, C-reactive protein; DII, dietary inflammatory
index; FFQ, food-frequency questionnaire; Hb A1c, glycated hemoglobin;
MDS, multiarray detection system; MEDI, alternate Mediterranean diet index;
SAA, serum amyloid A; sICAM, soluble intercellular adhesion molecule-1.
Received December 12, 2012. Accepted for publication September 30, 2013.
First published online October 23, 2013; doi: 10.3945/ajcn.112.056333.
1533
Supplemental Material can be found at:
http://ajcn.nutrition.org/content/suppl/2013/11/14/ajcn.112.0
56333.DCSupplemental.html
1534 VAN WOUDENBERGH ET AL
a certain food product. Therefore, besides studying effects of nu-
trients on inflammation and markers of glucose metabolism, it is
also important to study whether the overall diet is associated with
markers of glucose metabolism through its effects on low-grade
inflammation. A dietary index that reflects the quality of the diet,
the alternative healthy eating index (AHEI), was associated with
markers of low-grade inflammation and risk of type 2 diabetes (8,
9). However, this index was not designed to reflect the in-
flammatory potential of the diet. To design an index that does
reflect the inflammatory potential of the diet, Cavicchia et al (10)
developed dietary inflammatory weights for a number of dietary
components on the basis of a systematic review of available lit-
erature on diet and inflammation. For instance, ethanol had an
antiinflammatory weight of 20.53, whereas SFA had a proin-
flammatory weight of 0.25. The weights can be used to obtain
a dietary inflammatory index (DII) that reflects the inflammatory
potential of the diet. To our knowledge, whether a DII is associated
with any inflammation-related health outcomes, such as markers
of glucose metabolism, has not been previously investigated.
Aims of this study were to investigate whether a DII is as-
sociated with 1) a summary score for low-grade inflammation
and 2) markers of glucose metabolism in a Dutch population. In
addition, we investigated whether the association between the
DII and markers of glucose metabolism is mediated by chronic
low-grade inflammation.
SUBJECTS AND METHODS
Study population
The study population consisted of participants from the fol-
lowing 2 Dutch cohorts: the Cohort study on Diabetes and
Atherosclerosis Maastricht (CODAM) and the Hoorn study.
Briefly, the CODAM started in 1999 and is an ongoing cohort
study. It was designed to investigate the effects of disturbed glucose
metabolism, obesity, blood lipids concentrations, lifestyle factors,
and genetic factors on cardiovascular disease and mortality (11–15).
The CODAM comprises 574 participants who were selected on the
basis of elevated risk of type 2 diabetes and cardiovascular diseases
from a large population-based cohort (n . 20,000) and had un-
dergone a glucose metabolism screening test (n = 2715) (11, 16).
Briefly, the Hoorn study started in 1989 with a sample of the
general population of Hoorn, Netherlands (n = 2484) (17). The
Hoorn study is a population-based cohort study designed to
investigate the effect of disturbed glucose metabolism on car-
diovascular disease risk factors and complications (17). In 2000–
2001, 822 participants were examined again (18, 19) [ie, 648
surviving participants in the Hoorn study and an additional group
of 174 participants with type 2 diabetes from the Hoorn screening
study (20)].
Both studies obtained written informed consent from all
participants and were approved by the local Ethics Committees
(CODAM: Medical Ethical Review Committee of the Maastricht
University Medical Centre; Hoorn study: Ethical Review Com-
mittee of the VU University Medical Center Amsterdam).
Population for analysis
For the current investigation, data from both the baseline
examination of the CODAM and the follow-up examination of
the Hoorn study were used. We combined these studies because
they had followed a similar data-collection research protocol and
had been used as a combined cohort in previous investigations
(21–23). Of the 1397 participants with reliable measures of food
intake obtained from a food-frequency questionnaire (FFQ) (518
subjects from the CODAM; 879 subjects from the Hoorn study),
we excluded, in consecutive order, 138 participants with missing
information on glucose metabolism status or inflammation
markers, 105 participants with known diabetes, 22 participants
with missing information on covariates used in the analysis, and
108 participants with a CRP concentration .10 mg/L. Partici-
pants with CRP concentrations .10 mg/L were excluded be-
cause these higher concentrations reflect acute rather than
chronic inflammation (24). Finally, the population for analysis
comprised 1024 participants (420 subjects from the CODAM;
604 subjects from the Hoorn study).
Assessment of dietary intake
In both cohorts, dietary intake was assessed by using a self-
administered semiquantitative FFQ, which had been validated in
Downloaded from ajcn.nutrition.org by guest on June 14, 2017
a Dutch population (25). Intakes of all food items were converted
into intakes of energy and nutrients by using an extended version
of the Dutch Food Composition table 2001. The caffeine content
of food items was not included in the Dutch Food Composition
table. Therefore, the caffeine content was estimated to be 68 mg/
100 mL coffee (26), 20 mg/100 mL tea (27), and 8 mg/100 mL
cola soft drink (27).
Calculation of adapted dietary inflammatory index
Cavicchia et al (10) has developed literature-based dietary in-
flammatory weights that reflect the inflammatory potential of en-
ergy, 32 nutrients, 4 food products, 4 spices, and caffeine (Table 1).
In line with Cavicchia et al (10), we obtained a DII by multiplying
the dietary inflammatory weights of the dietary components by the
daily intake. Subsequently, these multiplications were summed.
Details about the calculation of the DII are shown in Table 1.
On the basis of a nutritional rationale, we also obtained an
adapted dietary inflammatory index (ADII) by multiplying the
dietary inflammatory weight of 26 nutrients, one food product,
one spice, and caffeine by the standardized energy-adjusted in-
take. Subsequently, these multiplications were summed (Table 1).
We explain 1) why an energy-adjusted intake was used, 2) why
the intake was standardized, and 3) why some dietary compo-
nents were excluded. Other details are shown in Table 1.
Energy-adjusted intake
We adjusted all dietary components for energy by using the
residual method to reduce the between-person variation in dietary
intake resulting from differences in physical activity, body size,
and metabolic efficiency (29). Therefore, the ADII was used as
a measure of diet quality.
Standardized intake
To avoid that the variation in the ADII was solely driven by
a few dietary components with a large range in intake, we
standardized intake of all components. Standardization was done
by subtracting the mean intake of the population from the in-
dividual intake and dividing the difference by the SD of the study
 ADII, INFLAMMATION, AND GLUCOSE METABOLISM 1535
TABLE 1
Dietary components included in the DII and ADII1
Components IW2 Included in DII Included in ADII

Energy (kcal/d) 0.230 Included Not included3

Protein (g/d) 20.050 Included Included
Total fat (g/d) 0.323 Included Not included3
SFAs (g/d) 0.250 Included Included
MUFAs (g/d) 0.050 Included Included
trans Fatty acids (g/d) 0.260 Not included4 Included
n23 PUFAs (g/d) 20.384 Included Included
n26 PUFAs (g/d) 0.016 Included Included
Cholesterol (mg/d) 0.210 Included Included
Carbohydrate (g/d) 0.346 Included Included
Fiber (g/d) 20.520 Included Included
Ethanol (g/d) 20.534 Included Included5
Wine (g/d) 20.480 Included Not included3
Beer (g/d) 20.200 Included Not included3
Liquor (g/d) 20.100 Included Not included3
Caffeine (g/d) 20.035 Included Included
Vitamin A (mg/d) 20.580 Included Included
b-Carotene (mg/d) 20.725 Included Included
Thiamin (mg/d) 20.050 Included Included

Downloaded from ajcn.nutrition.org by guest on June 14, 2017

Riboflavin (mg/d) 20.160 Included Included
Niacin (mg/d) 20.260 Included Included
Vitamin B-6 (mg/d) 20.286 Included Included
Folate (mg/d) 20.214 Included Included
Vitamin B-12 (mg/d) 0.090 Included Included
Vitamin C (mg/d) 20.367 Included Included
Vitamin D (mg/d) 20.342 Included Included
Vitamin E (mg/d) 20.401 Included Included
Iron (mg/d) 20.029 Included Included
Magnesium (mg/d) 20.905 Included Included
Selenium (mg/d) 20.021 Included Included
Zinc (mg/d) 20.316 Included Included
Tea (g/d) 20.552 Included Included6
Quercetin (mg/d) 20.490 Included Included
Genistein (mg/d) 20.680 Not included7 Not included7
Epicatechin (mg/d) 20.120 Not included7 Not included7
Luteolin (mg/d) 20.430 Not included7 Not included7
Daidzein (mg/d) 20.170 Not included7 Not included7
Cyanidin (mg/d) 20.130 Not included7 Not included7
Garlic (g/d) 20.270 Included Included
Ginger (g/d) 20.180 Not included7 Not included7
Saffron (g/d) 20.180 Not included7 Not included7
Turmeric (g/d) 20.774 Not included7 Not included7
Total DII for a participant8 — + (intakei 3 IWi) O 100 —
Total ADII for a participant — — + (energy-adjusted standardized intakei 3 IWi)
1
ADII, adapted dietary inflammatory index; DII, dietary inflammatory index; IW, inflammatory weight.
2
Dietary components with a positive IW were considered proinflammatory. Dietary components with a negative IW were considered antiinflammatory.
3
Energy was excluded in the ADII because all macronutrients were already included. Total fat was excluded in the ADII because all fatty acids were
already included. Alcoholic beverages beer, wine, and liquor were excluded in the ADII because the intake of ethanol was already included.
4
trans Fatty acids were not included in the previously published DII because the intake of trans fatty acids could not be calculated in the study by
Cavicchia et al (10).
5
Dietary IW for ethanol was assumed to be zero when the intake of ethanol was .40 g/d because the intake of ethanol is not likely to be antiin-
flammatory when an intake is .40 g/d (29).
6
Intake of tea was still included because the intake of epicatechin could not be calculated from our food-frequency questionnaire.
7
These dietary components were not taken into account in our DII and ADII calculations because intakes of these components could not be calculated
from our food-frequency questionnaire.
8
Intake of n23 and n26 PUFAs were multiplied by 10 because intakes were low and expressed as grams per day. Intakes of vitamin A and b-carotene
were divided by 100 to equilibrate the range of intake to other micronutrients according to Cavicchia et al (10).

population (z score) to equilibrate the intake of all nutrients to data-dependent 10 as Cavicchia et al (10) did (Table 1). We also
the same unit. Therefore, it was not necessary to divide intakes did not divide the overall ADII by 100 because division did not
of vitamin A and b-carotene by the arbitrary, data-dependent improve the interpretation of the results as it improved the in-
100 and multiply n23 and n26 fatty acids with the arbitrary, terpretation of the previously published DII (10).
1536 VAN WOUDENBERGH ET AL
Exclusion components
We excluded several components when we calculated the ADII
to avoid an overestimation of inflammatory effects of ethanol, fat,
and energy. To reduce the impact of ethanol on the ADII, separate
antiinflammatory effects of the alcoholic beverages beer, wine,
and liquor were not taken into account. Antiinflammatory effects
of these beverages are likely to be attributable to ethanol (3).
Energy was excluded because it is likely that the inflammatory
effect of energy is the sum of the inflammatory effects of all
energy-providing macronutrients. Total fat was also excluded
because it was assumed that the inflammatory effect of total fat is
the sum of the inflammatory effects of all separate fatty acids.
Markers of glucose metabolism
Venous blood samples were drawn from all participants at the
research center after an overnight fast (.10 h) to be able to
measure, eg, the fasting glucose concentration, fasting insulin
concentration, and glycated hemoglobin (Hb A1c). A 2-h post-
load glucose concentration was determined after a standard 75-g
oral glucose tolerance test except in participants with established
diabetes or a very high fasting plasma glucose concentration
(CODAM: .10 mmol/L; Hoorn study: .8.0 mmol/L). Fasting
and 2-h postload glucose concentrations were measured in plasma
by using glucose hexokinase methods [CODAM: ABX Diagnostics
Glucose HK125 (Horiba-ABX); Hoorn study: Roche Diagnostics].
Hb A1c was analyzed by ion-exchange HPLC (CODAM and
Hoorn study: Bio-rad). The insulin concentration was measured
in plasma by a 2-site immunoradiometric assay by using paired
monoclonal antibodies (CODAM and Hoorn study: Medgenix
Diagnostics). Insulin resistance was estimated from fasting
plasma glucose and plasma insulin concentrations by using the
homeostasis model assessment calculator (HOMA2) (30).
Markers of low-grade inflammation
In both cohorts, concentrations of 6 biomarkers of low-grade
inflammation [ie, CRP, IL-6, IL-8, TNF-a, serum amyloid A
(SAA), and soluble intercellular adhesion molecule-1 (sICAM)]
were measured in plasma by using a multiarray detection system
(MDS) on the basis of electro-chemiluminescence detection
(SECTOR Imager 2400; MesoScaleDiscovery). All measure-
ments were performed at the Research Laboratory, Department
of Internal Medicine, Maastricht University Medical Centre
(head: CGS). In the CODAM, CRP was also measured in serum
by using a high-sensitivity immunoturbidimetry assay (Latex;
Roche Diagnostics Netherlands BV), and IL-6, SAA, sICAM
were also measured in EDTA plasma by using an ELISA (IL-6:
R&D Systems; SAA and sICAM: Biosource; Invitrogen). These
measurements were done at the Laboratory of Toxicology, Ge-
netics and Pathology of the National Institute for Public Health
and the Environment, Bilthoven, Netherlands. Values obtained
by using the immunoturbidimetry assay or ELISA were cali-
brated on the values obtained by using the MDS in the CODAM.
Subsequently, the calibrated and MDS values were averaged and
used for CODAM participants in the current analysis. Intraassay
CVs ranged from 0.6% to 6.4%, and interassay CVs ranged from
1.9% to 17.5%. More information about measurements has been
shown elsewhere (12, 13, 31).
Calculation summary score for low-grade inflammation
A summary score for low-grade inflammation was calculated
to cluster conceptually related markers of low-grade inflam-
mation and improve statistical efficiency. To obtain this summary
score, a z score for each marker of low-grade inflammation was
calculated because the markers of low-grade inflammation were
expressed on different scale units. Subsequently, these z scores
were averaged to obtain a summary score for low-grade in-
flammation for each participant as follows:
Summary score ¼ ½z scoreðloge CRPÞ þ z scoreðloge IL  6Þ
þ z scoreðloge IL  8Þ þ z scoreðTNF  aÞ
þ z scoreðloge SAAÞ þ z scoreðsICAMÞO6 ð1Þ
This summary score for low-grade inflammation had been used
in previous investigations (14, 31, 32).
Covariates
Downloaded from ajcn.nutrition.org by guest on June 14, 2017
In both cohorts, the participant completed a self-administered
questionnaire that, among other things, included questions about
age, sex, smoking behavior, family history of diabetes in first-
degree relatives, and use of medications (eg, antihypertensive,
lipid-lowering, glucose-lowering medications). On the basis of
the questions about smoking behavior, the participant was cat-
egorized as a never, former, or current smoker. Family history of
diabetes was defined as a parent, a sibling, or both with diagnosed
diabetes. Height was measured to the nearest centimeter, and
weight was measured to the nearest 100 g by trained personnel.
The participant was weighed in a standing position wearing light
indoor cloths and no shoes. BMI (in kg/m2) was calculated as
weight divided by height squared. Waist circumference (cm)
was obtained at levels halfway between the lateral lower rib
margin and the spina iliaca anterior superior. Habitual physical
activity was assessed by using a validated short physical activity
questionnaire (short questionnaire to assess health-enhancing
physical activity), which measured the duration and intensity of
different activities (min/wk 3 intensity) (33).
Statistical analysis
Summary statistics were used to describe population charac-
teristics by tertiles of the ADII. To get more insight into the
contribution of the individual dietary components to the total
ADII, the contribution of the different dietary components to the
variation between participants in the ADII was assessed by using
forward linear regression. Before additional analyses, 7 skewed
variables were loge transformed to improve their distribution
toward normal (CRP, IL-6, IL-8, SAA, fasting plasma glucose,
postload glucose, and HOMA-IR).
First, the association between the ADII and markers of low-
grade inflammation was investigated by using linear regression.
Model 1 included, as the main independent variable, the ADII
(expressed per SD) and the covariates age (y), sex, cohort
(CODAM or Hoorn), smoking (never, former, or current),
physical activity (min/wk 3 intensity, family history of diabetes
(yes, no, or missing), use of lipid-lowering medication (yes or
no), having hypertension (yes or no), and the intake of energy
(kcal/d). Except for cohort, these covariates were included
 ADII, INFLAMMATION, AND GLUCOSE METABOLISM
because of their association with inflammation and diabetes
observed in the literature. In model 2, BMI was added to model
1 because we were also interested in the effect of the ADII on
inflammation independent of BMI. Waist circumference was not
included as an additional covariate because its inclusion did not
change the conclusions, and waist circumference was missing
for 8 participants. Effect-measure modification by sex was in-
vestigated by adding an interaction term between the ADII and
sex to model 2 when the summary score for low-grade in-
flammation was studied as a dependent variable.
Second, to investigate whether this study also confirmed the
well-known adverse associations between low-grade inflammation
and markers of glucose metabolism, the association between the
summary score for low-grade inflammation and the 4 markers of
glucose metabolism was studied. Model 1 of the linear regression
model included age (y), sex, cohort (CODAM or Hoorn), smoking
(never, former, or current), physical activity (min/wk 3 intensity),
family history of diabetes (yes, no, or missing), use of lipid-lowering
medication (yes or no), and having hypertension (yes or no). Model
2 included BMI in addition to model 1.
1537
subjects were men, 26% of subjects had a normal weight, 18%
of subjects were current smokers, and 51% of subjects had
normal glucose metabolism. The Hoorn study, compared with
the CODAM, included participants with an older age (68 6 7
compared with 58 6 7 y, respectively), more women (49%
compared with 37%, respectively), and fewer current smokers
(16% compared with 20%), respectively. The mean BMI was
comparable (27 6 4 in the Hoorn study; 28 6 4 in the CODAM).
ADII and its components
The ADII ranged from 212.0 to 15.7 (range: 27.7 units)
(Table 2). Participants with a high ADII smoked more and were
more often men than were participants with a low ADII (Table
2). Intakes of SFAs, MUFAs, and trans fatty acids were higher in
participants with a high than low ADII. Intakes of protein, n23
PUFA, and n26 PUFA were lower in participants with a high
than low ADII (see Supplemental Table 1 under “Supplemental
data” in the online issue). The Spearman’s correlation between
the ADII and energy was low (r = 0.10, P = 0.02). The intake of

Downloaded from ajcn.nutrition.org by guest on June 14, 2017

Third, the association between the ADII and 4 markers of
glucose metabolism (ie, fasting plasma glucose, postload glucose,
HOMA-IR, and Hb A1c) was investigated by using linear re-
gression. Model 1 included the ADII and other covariates as in
model 1. To investigate the mediating role of low-grade inflam-
mation, the summary score for low-grade inflammation was in-
cluded in addition to covariates included in model 1 (model 1 +
inflammation). To investigate whether the association of the ADII
with markers of glucose metabolism was attributed to inflammation
independent of BMI, the summary score for low-grade inflammation
and BMI were simultaneously added to model 1. For this purpose,
we also used the multiple-mediation analysis as described by
Preacher and Hayes (34). This mediation analysis provides an effi-
cient way to quantify the independent mediating effects of low-grade
inflammation and BMI (Figure 1).
All analyses were performed with the SAS statistical software
package (version 9.2; SAS Institute Inc). P # 0.05 was con-
sidered statistically significant.
RESULTS
The mean (6SD) age of the population of analysis was 646 9 y,
59% of subjects were participants from the Hoorn study, 55% of
magnesium explained most of the variation (34%) between
participants in the ADII, followed by intakes of folate (25%),
quercetin (16%), and n23 PUFA (7%) (Table 3). Regarding
the DII, which ranged from 220.7 to 6.3 (range: 27.0 units)
(see Supplemental Table 2 under “Supplemental data” in the
online issue), the intake of tea explained most of the variation
(55%) between participants in the DII, followed by intakes
of SFA (17%), beer (13%), energy (6%), and wine (5%)
(Table 3).
ADII and low-grade inflammation
An increment of 1 SD in the ADII (ie, 2.88 units) was as-
sociated with a 0.04-unit (95% CI 0.01, 0.07-unit) higher sum-
mary score for low-grade inflammation in model 2 (P = 0.01)
(Table 4). This association was mainly driven by 4 of 6 markers
of inflammation (ie, CRP, IL-6, TNF-a, and sICAM). The
original DII was not associated with the summary score for
low-grade inflammation [model 2: b = 20.002 (95% CI: 20.03,
0.03)] (Table 4). The association between the ADII and sum-
mary score for low-grade inflammation did not differ between
men and women (P-interaction = 0.80).

FIGURE 1. Model used in the multiple mediation analysis of the association between the adapted dietary inflammatory index and markers of glucose
metabolism, including fasting glucose, 24-h glucose, HOMA-IR, and glycated hemoglobin [adapted from Preacher and Hayes (34)]. Paths a represent the regression
coefficient of the association between the adapted dietary inflammatory index and the summary score for low-grade inflammation (path a1) or BMI (path a2). Paths
b represent the regression coefficient of the association between the summary score for low-grade inflammation (path b1) or BMI (path b2) and markers of glucose
metabolism. In addition to other covariates, path b1 was adjusted for BMI and the adapted dietary inflammatory index, whereas path b2 was adjusted for the
summary score for low-grade inflammation and the adapted dietary inflammatory index. The product of regression coefficients of paths a and b represents the
mediated effect of inflammation (path a1 3 path b1) or BMI (path a2 3 path b2). Path c’ represents that part of the association that was not explained by low-grade
inflammation or BMI. This path is referred to as the direct association between the adapted dietary inflammatory index and markers of glucose metabolism.
1538 VAN WOUDENBERGH ET AL
TABLE 2
Characteristics of the study population by tertiles of the ADII (n = 1024)1
ADII

Tertile 1 Tertile 2 Tertile 3 P2

Range 212.0 to less than 21.27 21.27 to less than 1.15 1.15 to less than 15.7 —
Median 22.51 20.03 2.72 —
n 341 342 341 —
Age (y) 65.0 6 8.13 63.9 6 8.4 64.1 6 9.3 0.23
Sex (M) (%) 50.7 53.8 61.6 0.01
Study (% from the Hoorn study) 61.6 59.4 56.0 0.33
Smoking (current) (%) 12.3 18.1 23.5 0.01
Any physical activity (h/d) 4.2 6 2.8 4.2 6 2.9 3.8 6 3.0 0.20
BMI (kg/m2) 27.3 6 3.7 27.9 6 3.8 27.7 6 4.1 0.09
Waist circumference (cm)4 94.8 6 10.8 96.4 6 11.4 97.7 6 12.4 0.01
Diabetes category (% with NGM) 53.7 50.0 49.9 0.79
Family history of diabetes (%)5 27.9 27.5 27.0 0.61
Lipid-lowering medication (%) 15.3 18.7 15.0 0.34
Antihypertensive medication (%) 34.3 33.9 35.5 0.90
1
ADII, adapted dietary inflammatory index; NGM, normal glucose metabolism.
2
Chi-square test was used for categorical variables; ANOVA was used for continuous variables.
3
Mean 6 SD (all such values).

Downloaded from ajcn.nutrition.org by guest on June 14, 2017

4
Missing for 8 participants.
5
Missing for 122 participants.

Low-grade inflammation and markers of glucose baseline examination. Of the 592 participants without type 2
metabolism diabetes at baseline and with follow-up data, 99 subjects had
The summary score for low-grade inflammation was associ- type 2 diabetes at follow-up. The ADII was not associated
ated with adverse concentrations of all markers of glucose me- with the incidence of type 2 diabetes [incidence proportion
tabolism (Table 5). An increment of 1 unit in the summary score ratiomodel 1= 1.00 (95% CI: 0.91, 1.09)].
for low-grade inflammation was associated with, on average,
a 4% (95% CI: 2%, 6%) higher fasting glucose concentration, a DISCUSSION
9% (95% CI: 4%, 14%) higher postload glucose concentration, The aims of this study were to investigate whether the in-
a 16% (95% CI: 11%, 22%) higher HOMA-IR, and a 0.21% flammatory potential of the diet as assessed with the ADII is
(95% CI: 0.13%, 0.29%) higher Hb A1c concentration (model 2). associated with 1) the summary score for low-grade in-
flammation and 2) markers of glucose metabolism. We observed
an adverse association between the ADII and summary score
ADII and markers of glucose metabolism
An increment of 1 SD in the ADII (ie, 2.88 units) was as- TABLE 3
sociated with, on average, a 0.9% (95% CI: 0.1%, 1.7%) higher Explained interindividual variance in the ADII and DII by dietary
fasting glucose concentration, a 2.3% (95% CI: 0.0%, 4.6%) components included in the index calculation (n = 1024)1
higher postload glucose concentration, and a 3.5% (95% CI: Components R2 Model R2
0.6%, 6.3%) higher HOMA-IR (Table 6, model 1). The ADII
was not associated with Hb A1c. After inclusion of the summary ADII
Magnesium (mg/d) 0.34 0.34
score for low-grade inflammation, all associations attenuated
Folate (mg/d) 0.25 0.60
[eg, HOMA-IR: b-model 1+inflammation= 2.2% (95% CI: Quercetin (mg/d) 0.16 0.76
20.6%, 5.0%)]. Additional adjustment of BMI attenuated the n23 PUFAs (g/d) 0.07 0.82
association further [eg, HOMA-IR: b-model 1+inflammation b-Carotene (mg/d) 0.04 0.86
+BMI (c#) = 1.4% (95% CI: 21.1, 3.9)] (Figure 1, Table 6). Ethanol (g/d) 0.04 0.89
When the summary score for low-grade inflammation and BMI Vitamin D (mg/d) 0.02 0.91
were simultaneously added to model 1, the summary score for Other components 0.09 1.00
low-grade inflammation, but not BMI, explained a significant DII
Tea (g/d) 0.55 0.55
proportion of the association between the ADII and HOMA-IR
SFAs (g/d) 0.17 0.72
(path a1 3 path b1 = 0.7% higher per SD through inflammation Beer (g/d) 0.13 0.85
independent of BMI) and between ADII and postload glucose Energy (kcal/d) 0.06 0.91
(path a1 3 path b1 = 0.5% higher per SD through inflammation Wine (g/d) 0.05 0.96
independent of BMI) (Figure 1, Table 6). ADII had no direct Magnesium (mg/d) 0.02 0.98
association (c#) with the 4 markers of glucose metabolism Other components 0.02 1.00
(Figure 1, Table 6). 1
Forward linear regression was used to calculate the R2 and model R2.
Part of our population for analysis (n = 720; 70%) was retested Components that explained .1% of the interindividual variation are shown.
with an oral glucose tolerance test, on average, 7.2 y after the ADII, adapted dietary inflammatory index; DII, dietary inflammatory index.
 ADII, INFLAMMATION, AND GLUCOSE METABOLISM 1539
TABLE 4
b-Coefficients (95% CIs) of the association between dietary inflammatory indexes and markers of inflammation (n = 1024)1
Crude P Model 1 P Model 2 P

ADII (per SD of 2.88)

Inflammation score2 0.04 (0.01, 0.08) 0.01 0.04 (0.02, 0.07) ,0.01 0.04 (0.01, 0.07) 0.01
CRP3 0.07 (0.02, 0.13) 0.01 0.06 (0.01, 0.12) 0.02 0.05 (20.01, 0.10) 0.08
IL-63 0.06 (0.03, 0.10) ,0.01 0.05 (0.01, 0.08) 0.01 0.04 (0.01, 0.08) 0.02
IL-83 20.03 (20.08, 0.01) 0.16 20.01 (20.03, 0.02) 0.64 20.01 (20.03, 0.02) 0.63
TNF-a 0.15 (20.03, 0.32) 0.10 0.17 (20.00, 0.33) 0.05 0.16 (20.01, 0.33) 0.07
Serum amyloid A3 0.00 (20.04, 0.04) 0.97 0.02 (20.03, 0.06) 0.44 0.01 (20.03, 0.05) 0.67
sICAM 5.28 (1.81, 8.75), 0.01 4.57 (1.32, 7.81) 0.01 3.96 (0.76, 7.15) 0.02
DII (per SD of 2.07)
Inflammation score2 20.05 (20.08, 20.01) 0.01 0.02 (20.01, 0.05) 0.28 20.002 (20.03, 0.03) 0.90
CRP3 0.05 (20.00, 0.11) 0.06 0.08 (0.02, 0.13) 0.01 0.03 (20.02, 0.09) 0.21
IL-63 0.02 (20.02, 0.06) 0.29 0.01 (20.03, 0.05) 0.63 20.003 (20.04, 0.03) 0.90
IL-83 20.16 (20.21, 20.12) 0.01 -0.03 (20.06, 0.00) 0.06 20.03 (20.06, 0.00) 0.05
TNF-a 20.23 (20.40, 20.05) 0.01 20.01 (20.18, 0.17) 0.93 20.03 (20.20, 0.15) 0.75
Serum amyloid A3 20.02 (20.07, 0.02) 0.30 0.03 (20.02. 0.07) 0.21 0.01 (20.03, 0.05) 0.68
sICAM 21.62 (25.10, 1.86) 0.36 1.61 (21.77, 5.00) 0.35 20.06 (23.30, 3.42) 0.97
1
Median (25th–75th percentiles) or mean (6SD) of the inflammation markers were as follows: inflammation score, 20.1 6 0.5; CRP (mg/L), 1.7 (0.97–
3.4) ; IL-6 (ng/L), 1.4 (1.1–2.9); IL-8 (ng/L), 2.2 6 0.7; TNF-a (ng/L), 7.9 6 2.9; serum amyloid A (mg/L), 1.5 (0.97–2.4); and sICAM (mg/L), 239.5 6 56.7.
b-Coefficients (95% CIs) were obtained by using linear regression. Model 1 included age, sex, cohort, physical activity, smoking, family history of diabetes,

Downloaded from ajcn.nutrition.org by guest on June 14, 2017

use of lipid-lowering medication, and hypertension as covariates. The ADII was also adjusted for the intake of energy. In addition to model 1, model 2
also included BMI as a covariate. ADII, adapted dietary inflammatory index; CRP, C-reactive protein; DII, dietary inflammatory index; sICAM, soluble
intercellular adhesion molecule.
2
Summary score for low-grade inflammation was obtained by the following formula: [z score(logeCRP) + z score(logeIL-6) + z score(logeIL-8) + z score
(TNF-a) + z score(logeserum amyloid A) + z score(sICAM)] O 6.
3
For the analysis, these markers of inflammation were loge transformed to improve their distribution toward normal. Therefore, when the DII or ADII was
1 SD higher, these markers of low-grade inflammation were, on average, b 3 100% higher or lower.

for low-grade inflammation, which suggested that the in-
flammatory potential of the diet affects markers of inflammation.
The adverse association between the ADII and HOMA-IR
suggested that the inflammatory potential of the diet affects
insulin resistance. This effect was supported by the mediating
role of low-grade inflammation in this analysis on insulin re-
sistance.
On the basis of our results, it is likely that the adaptations in the
DII calculation improved the estimation of the inflammatory
potential of the diet. First, the variation in the ADII was not solely
driven by components with a large range in intake, in contrast to
the previously published DII. The intake of tea explained most of
the variation in the original DII in our study because the intake of
tea ranged from 0 to 1500 mL/d. With the use of standardized
intakes, the ADII was less dependent on the intake range of
components in the study under investigation. Therefore, it is
likely that the results from the ADII will be more comparable
between populations. Second, the ADII also avoided an over-
estimation of the inflammatory effect of certain nutrients by
excluding alcoholic beverages, total fat, and energy. Third, the
ADII was associated with the summary score for low-grade
inflammation, whereas the original DII was not associated with
the summary score for low-grade inflammation in our study. The
previously published DII was not associated with CRP on a
continuous scale, although it was concluded that diet can affect
low-grade inflammation on the basis of the observed adverse
association between the DII and elevated CRP concentration
(.3 mg/L) (10).

TABLE 5
b-coefficients (95% CIs) of the association between the summary score for low-grade inflammation and markers of glucose metabolism1
Summary score for low-grade inflammation

n Crude P Model 1 P Model 2 P

Fasting glucose (mmol/L) 2

1022 0.07 (0.05, 0.09) ,0.01 0.05 (0.04, 0.07) ,0.01 0.04 (0.02, 0.06) ,0.01
Postload glucose (mmol/L)2 907 0.12 (0.08, 0.16) ,0.01 0.12 (0.07, 0.17) ,0.01 0.09 (0.04, 0.14) ,0.01
HOMA-IR2 1003 0.21 (0.15, 0.26) ,0.01 0.28 (0.22, 0.34) ,0.01 0.16 (0.11, 0.22) ,0.01
Hb A1c (%) 1008 0.27 (0.21, 0.34) ,0.01 0.23 (0.16, 0.31) ,0.01 0.21 (0.13, 0.29) ,0.01
1
Median (25th–75th percentiles) or mean (6SD) of markers of glucose metabolism were as follows: fasting glucose (mmol/L), 5.7 (5.3–6.4); postload
glucose (mmol/L), 6.6 (5.3–8.6); HOMA-IR, 1.1 (0.8–1.6); and Hb A1c (%), 5.9 6 0.6. b-coefficients (95% CIs) were obtained by using linear regression. The
summary score for low-grade inflammation was obtained by using the following formula: [z score(logeC-reactive protein) + z score(logeIL-6) + z score
(logeIL-8) + z score(TNF-a) + z score(logeserum amyloid A) + z score(soluble intercellular adhesion molecule)] O 6. Model 1 included age, sex, cohort,
physical activity, smoking, family history of diabetes, use of lipid-lowering medication, and hypertension as covariates. In addition to model 1, model 2 also
included BMI. Hb A1c, glycated hemoglobin.
2
For the analysis, these markers of glucose metabolism were loge transformed to improve their distribution toward normal. Therefore, when the summary
score for low-grade inflammation was 1 unit higher, these markers of glucose metabolism were, on average, b 3 100% higher.
1540 VAN WOUDENBERGH ET AL
TABLE 6
b-coefficients (95% CIs) of the association between the ADII and markers of glucose metabolism1
ADII (per SD of 2.88) P
2
Fasting glucose (n = 1022) (mmol/L)
Crude 0.008 (20.002, 0.017) 0.10
Total effect (model 1)3 0.009 (0.001, 0.017) 0.03
Model 1 + BMI 0.007 (20.001, 0.015) 0.08
Model 1 + inflammation 0.007 (20.002, 0.0156) 0.11
Direct effect [model 1+ BMI and inflammation (c#)] 0.006 (20.002, 0.014) 0.16
Indirect effect through BMI (a2 3 b2)4 0.001 (0.000, 0.003) —5
Indirect effect through inflammation (a1 3 b1)6 0.002 (0.001, 0.004) —5
Postload glucose (n = 907) (mmol/L)2
Crude 0.019 (20.004, 0.042) 0.10
Total effect (model 1)3 0.023 (0.000, 0.046) 0.05
Model 1 + BMI 0.019 (20.003, 0.041) 0.09
Model 1 + inflammation 0.017 (20.006, 0.039) 0.14
Direct effect [model 1 + BMI and inflammation (c#)] 0.015 (20.007, 0.037) 0.18
Indirect effect through BMI (a2 3 b2)4 0.003 (20.000, 0.008) —5
Indirect effect through inflammation (a1 3 b1)6 0.005 (0.002, 0.010) —5
HOMA-IR (n =1003)2
Crude 0.037 (0.007, 0.067) 0.02
Total effect (model 1)3 0.035 (0.006, 0.063) 0.02

Downloaded from ajcn.nutrition.org by guest on June 14, 2017

Model 1 + BMI 0.020 (20.005, 0.045) 0.12
Model 1 + inflammation 0.022 (20.006, 0.050) 0.12
Direct effect [model 1 + BMI and inflammation (c#)] 0.014 (20.011, 0.039) 0.28
Indirect effect through BMI (a2 3 b2)4 0.014 (20.001, 0.028) —5
Indirect effect through inflammation (a1 3 b1)6 0.007 (0.003, 0.014) —5
Hb A1c (n = 1008) (%)
Crude 0.010 (20.027, 0.048) 0.05
Total effect (model 1)3 0.011 (20.025, 0.046) 0.13
Model 1 + BMI 0.007 (20.029, 0.042) 0.06
Model 1 + inflammation 0.000 (20.035, 0.036) 0.05
Direct effect [model 1+ BMI and inflammation (c#)] 20.001 (20.037, 0.034) 0.10
Indirect effect through BMI (a2 3 b2)4 0.002 (0.000, 0.008) —5
Indirect effect through inflammation (a1 3 b1)6 0.010 (0.003, 0.019) —5
1
Median (25th–75th percentiles) or mean (6SD) of markers of glucose metabolism were as follows: fasting glucose
(mmol/L), 5.7 (5.3–6.4); postload glucose (mmol/L), 6.6 (5.3–8.6); HOMA-IR, 1.1 (0.8–1.6); and Hb A1c (%), 5.9 6 0.6.
b-coefficients (95% CIs) were obtained by using linear regression. See Figure 1 for the interpretation of c#, a1, a2, b1, and b2.
ADII, adapted dietary inflammatory index; Hb A1c, glycated hemoglobin.
2
For the analysis, these markers of glucose metabolism were loge transformed to improve their distribution toward
normal. Therefore, when the ADII was 1 SD (SD: 2.88) higher, these markers of glucose metabolism were, on average, b 3
100% higher or lower.
3
Model 1 was adjusted for age, sex, cohort, physical activity, smoking, family history of diabetes, use of lipid-
lowering medication, hypertension, and intake of energy.
4
When the ADII was 1 SD higher, the marker of glucose metabolism was b 3 100% higher or lower through the effect
of the ADII on BMI.
5
Multiple mediation analysis described by Preacher and Hayes did not provide P values (34).
6
When the ADII was 1 SD higher, the marker of glucose metabolism was b 3 100% higher or lower through the
effect of the ADII on inflammation.

Other diet-quality scores have been shown to be associated
with chronic low-grade inflammation. The AHEI (8), the alter-
nate Mediterranean diet index (MEDI) (8), and the Mediterranean
diet score (35) were inversely associated with CRP, IL-6, and
sICAM. If examined closely, these scores have some similarities
with the ADII. A low intake of cereal fiber and a high intake of
trans fat are considered unhealthy in the AHEI, which is in line
with the dietary inflammatory weights for total fiber (20.52) and
trans fatty acids (0.26) in the ADII. Furthermore, the AHEI
gives a preference to PUFAs and MEDI to MUFAs over SFAs.
In the ADII, n23 PUFAs and MUFAs are considered antiin-
flammatory, whereas SFAs are considered pro-inflammatory. In
addition, the intake of ethanol is considered healthy in the
AHEI, MEDI, and Mediterranean diet score, which is in line
with the dietary antiinflammatory weights for ethanol (20.53) in
the ADII. Even though the purpose of these diet quality scores
was not to assess the inflammatory potential of diet, these
studies provided evidence that diet as a whole may play a role in
chronic low-grade inflammation.
The summary score for low-grade inflammation explained
a significant proportion of the association between the ADII and
HOMA-IR, even independent of BMI. This result supported the
hypothesis that low-grade inflammation mediates, at least in part,
the association between diet and insulin resistance. As part of an
inflammatory environment, a more proinflammatory diet could
lead to an impaired action of insulin (7). This hypothesis was not
further confirmed by an association between ADII and incidence
of type 2 diabetes in our cohorts. However, the analysis was
 ADII, INFLAMMATION, AND GLUCOSE METABOLISM
limited because information about the exact time of diagnosis
was not available, and the loss to follow-up was 30%. Therefore,
the validity should be confirmed by results of prospective studies,
preferably also with a larger number of incident cases before
a clinical application of the ADII is considered.
There were a number of strengths of this study to consider.
First, in addition to CRP, 5 other markers of inflammation were
examined, which provided a more-thorough assessment of low-
grade inflammation. Second, a validated FFQ was used to assess
intake. This FFQ has been shown to be appropriate for ranking
participants for 10 of the nutrients included in the ADII (25).
Third, the ADII was strengthened by its theory and literature-
based instead of data-driven nature. However, despite the use of
a systematic approach for constructing the literature-based di-
etary inflammatory weights, subjective decisions were made by
Cavicchia et al (10).
Besides these strengths, the study had limitations as well. First,
our results were limited by the cross-sectional nature of our study,
which did not allow conclusions to be made about causality. We
tried to limit the possibility of reverse causation by excluding
participants with known diabetes who may have changed their
diet recently. Second, the external validity of our study might
have been low because all participants were white, and the
CODAM included participants with high risk of impaired glucose
metabolism. However, the inclusion of a high-risk population
increased the variation in markers of glucose metabolism. Third,
intakes of some dietary components, which were included in the
previously published DII, such as of ginger and saffron, could not
be calculated from our FFQ (Table 1). However, the variation in
intakes of those specific dietary components was expected to be
low in a mostly nonvegetarian Dutch population. Fourth, al-
though extensive information about potential confounders was
available, residual confounding might have remained because
potential confounders could have been measured with error.
In conclusion, adverse associations between the inflammatory
potential of the diet, as assessed with the ADII, with low-grade
inflammation and HOMA-IR suggest that low-grade inflam-
mation might be one of the pathways through which diet affects
insulin resistance. More research is needed to verify whether the
ADII is associated with low-grade inflammation in other pop-
ulations and to investigate whether low-grade inflammation
mediates the association between diet and the development of
type 2 diabetes.
We thank each staff member who contributed to the data collection of the
CODAM and the Hoorn study. We also acknowledge the contributions of all
participants in the studies.
The authors’ responsibilities were as follows—GJvW: prepared data for
analyses, performed the analysis, and drafted the manuscript; DT and EJMF:
contributed to the interpretation of data; IF, MMvG, CJvdK, CGS, CDAS,
EEB, and EJMF: participated in the design and coordination of the CODAM;
CDAS, GN, and JMD: participated in the design and coordination of the
Hoorn study; and DT, AK, IF, MMvG, CJvdK, CGS, CDAS, MCO, GN,
JMD, EEB, and EJMF: critically revised the manuscript and agreed to be
listed as authors. None of the authors had a conflict of interest.
REFERENCES
1. Lee CC, Adler AI, Sandhu MS, Sharp SJ, Forouhi NG, Erqou S, Luben
R, Bingham S, Khaw KT, Wareham NJ. Association of C-reactive
protein with type 2 diabetes: prospective analysis and meta-analysis.
Diabetologia 2009;52:1040–7.
1541
2. Kolb H, Mandrup-Poulsen T. An immune origin of type 2 diabetes?
Diabetologia 2005;48:1038–50.
3. Calder PC, Ahluwalia N, Brouns F, Buetler T, Clement K, Cunningham
K, Esposito K, Jonsson LS, Kolb H, Lansink M, et al. Dietary factors
and low-grade inflammation in relation to overweight and obesity. Br J
Nutr 2011;106(suppl 3):S5–78.
4. Petersen AM, Pedersen BK. The anti-inflammatory effect of exercise.
J Appl Physiol 2005;98:1154–62.
5. Schulze MB, Schulz M, Heidemann C, Schienkiewitz A, Hoffmann K,
Boeing H. Fiber and magnesium intake and incidence of type 2 di-
abetes: a prospective study and meta-analysis. Arch Intern Med 2007;
167:956–65.
6. Baliunas DO, Taylor BJ, Irving H, Roerecke M, Patra J, Mohapatra S,
Rehm J. Alcohol as a risk factor for type 2 diabetes: a systematic re-
view and meta-analysis. Diabetes Care 2009;32:2123–32.
7. Kolb H, Mandrup-Poulsen T. The global diabetes epidemic as a con-
sequence of lifestyle-induced low-grade inflammation. Diabetologia
2010;53:10–20.
8. Fung TT, McCullough ML, Newby PK, Manson JE, Meigs JB, Rifai N,
Willett WC, Hu FB. Diet-quality scores and plasma concentrations of
markers of inflammation and endothelial dysfunction. Am J Clin Nutr
2005;82:163–73.
9. Fung TT, McCullough M, van Dam RM, Hu FB. A prospective study
of overall diet quality and risk of type 2 diabetes in women. Diabetes
Care 2007;30:1753–7.
Downloaded from ajcn.nutrition.org by guest on June 14, 2017
10. Cavicchia PP, Steck SE, Hurley TG, Hussey JR, Ma Y, Ockene IS,
Hebert JR. A new dietary inflammatory index predicts interval changes
in serum high-sensitivity C-reactive protein. J Nutr 2009;139:2365–72.
11. Kruijshoop M, Feskens EJ, Blaak EE, De Bruin TW. Validation of
capillary glucose measurements to detect glucose intolerance or type 2
diabetes mellitus in the general population. Clin Chim Acta 2004;341:
33–40.
12. Jacobs M, Van Greevenbroek MM, Van der Kallen CJ, Ferreira I,
Blaak EE, Feskens EJ, Jansen EH, Schalkwijk CG, Stehouwer CD.
Low-grade inflammation can partly explain the association between the
metabolic syndrome and either coronary artery disease or severity of
peripheral arterial disease: the CODAM study. Eur J Clin Invest 2009;
39:437–44.
13. Thewissen MM, Damoiseaux JG, Duijvestijn AM, Van Greevenbroek
MM, Van der Kallen CJ, Feskens EJ, Blaak EE, Schalkwijk CG,
Stehouwer CD, Cohen Tervaert JW, et al. Abdominal fat mass is as-
sociated with adaptive immune activation: the CODAM Study. Obesity
(Silver Spring) 2011;19:1690–8.
14. Wlazlo N, Van Greevenbroek MM, Ferreira I, Jansen EJ, Feskens EJ,
Van der Kallen CJ, Schalkwijk CG, Bravenboer B, Stehouwer CD.
Low-grade inflammation and insulin resistance independently explain
substantial parts of the association between body fat and serum C3: the
CODAM study. Metabolism 2012;61:1787–96.
15. Van Woudenbergh GJ, Kuijsten A, Van der Kallen CJ, Van Greevenbroek
MM, Stehouwer CD, Blaak EE, Feskens EJ. Comparison of fatty acid
proportions in serum cholesteryl esters among people with different
glucose tolerance status: the CoDAM study. Nutr Metab Cardiovasc Dis
2012;22:133–40.
16. Mensink M, Corpeleijn E, Feskens EJ, Kruijshoop M, Saris WH, de
Bruin TW, Blaak EE. Study on lifestyle-intervention and impaired
glucose tolerance Maastricht (SLIM): design and screening results.
Diabetes Res Clin Pract 2003;61:49–58.
17. Mooy JM, Grootenhuis PA, de Vries H, Valkenburg HA, Bouter LM,
Kostense PJ, Heine RJ. Prevalence and determinants of glucose in-
tolerance in a Dutch caucasian population. The Hoorn Study. Diabetes
Care 1995;18:1270–3.
18. Snijder MB, Dekker JM, Visser M, Bouter LM, Stehouwer CD,
Yudkin JS, Heine RJ, Nijpels G, Seidell JC; Hoorn study. Trunk fat
and leg fat have independent and opposite associations with fasting
and postload glucose levels: the Hoorn study. Diabetes Care 2004;
27:372–7.
19. van Dam RM, Snijder MB, Dekker JM, Stehouwer CD, Bouter LM,
Heine RJ, Lips P. Potentially modifiable determinants of vitamin D
status in an older population in the Netherlands: the Hoorn Study. Am J
Clin Nutr 2007;85:755–61.
20. Spijkerman AM, Adriaanse MC, Dekker JM, Nijpels G, Stehouwer
CD, Bouter LM, Heine RJ. Diabetic patients detected by population-
based stepwise screening already have a diabetic cardiovascular risk
profile. Diabetes Care 2002;25:1784–9.
1542 VAN WOUDENBERGH ET AL
21. Engelen L, Ferreira I, Gaens KH, Henry RM, Dekker JM, Nijpels G,
Heine RJ, ’t Hart LM, van Greevenbroek MM, van der Kallen CJ, et al.
The association between the -374T/A polymorphism of the receptor for
advanced glycation endproducts gene and blood pressure and arterial
stiffness is modified by glucose metabolism status: the Hoorn and
CoDAM studies. J Hypertens 2010;28:285–93.
22. Du H, Van der A D, Van Bakel MM, Van der Kallen CJ, Blaak EE,
Van Greevenbroek MM, Jansen EH, Nijpels G, Stehouwer CD, Dekker
JM, et al. Glycemic index and glycemic load in relation to food and
nutrient intake and metabolic risk factors in a Dutch population. Am J
Clin Nutr 2008;87:655–61.
23. DuH, van der A DL, van Bakel MM, van der Kallen CJ, Blaak EE, van
Greevenbroek MM, Jansen EH, Nijpels G, Stehouwer CD, Dekker JM,
et al. Association of polymorphism in the receptor for advanced gly-
cation end products (RAGE) gene with circulating RAGE levels. J Clin
Endocrinol Metab 2009;94:5174–80.
24. Pearson TA, Mensah GA, Alexander RW, Anderson JL, Cannon RO
3rd, Criqui M, Fadl YY, Fortmann SP, Hong Y, Myers GL, et al.
Markers of inflammation and cardiovascular disease: application to
clinical and public health practice: a statement for healthcare pro-
fessionals from the Centers for Disease Control and Prevention and the
American Heart Association. Circulation 2003;107:499–511.
25. Ocké MC, Bueno-de-Mesquita HB, Pols MA, Smit HA, van Staveren
WA, Kromhout D. The Dutch EPIC food frequency questionnaire. II.
Relative validity and reproducibility for nutrients. Int J Epidemiol
1997;26(suppl 1):S49–58.
26. Schel J. Koffie en gezondheid. [Coffee and health.] Haarlem, Nether-
lands: Voorlichtingsbureau voor Koffie en Gezondheid, 2005 (in
Dutch).
27. USDA. National nutrient database for standard reference database.
Available from: http://ndbnalusdagov (cited 22 May 2013).
28. Imhof A, Woodward M, Doering A, Helbecque N, Loewel H, Amouyel
P, Lowe GD, Koenig W. Overall alcohol intake, beer, wine, and sys-
temic markers of inflammation in western Europe: results from three
MONICA samples (Augsburg, Glasgow, Lille). Eur Heart J 2004;25:
2092–100.
29. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake
in epidemiologic studies. Am J Clin Nutr 1997;65(4 suppl):1220S–8S;
discussion 1229S–31S.
30. Levy JC, Matthews DR, Hermans MP. Correct homeostasis model
assessment (HOMA) evaluation uses the computer program. Diabetes
Care 1998;21:2191–2.
31. van Bussel BC, Henry RM, Schalkwijk CG, Dekker JM, Nijpels G,
Stehouwer CD. Low-grade inflammation, but not endothelial dys-
function, is associated with greater carotid stiffness in the elderly: the
Hoorn Study. J Hypertens 2012;30:744–52.
32. van Greevenbroek MM, Jacobs M, van der Kallen CJ, Vermeulen VM,
Jansen EH, Schalkwijk CG, Ferreira I, Feskens EJ, Stehouwer CD. The
cross-sectional association between insulin resistance and circulating
complement C3 is partly explained by plasma alanine aminotransfer-
ase, independent of central obesity and general inflammation (the
CODAM study). Eur J Clin Invest 2011;41:372–9.
33. Wendel-Vos GC, Schuit AJ, Saris WH, Kromhout D. Reproducibility
and relative validity of the short questionnaire to assess health-
enhancing physical activity. J Clin Epidemiol 2003;56:1163–9.
34. Preacher KJ, Hayes AF. Asymptotic and resampling strategies for as-
Downloaded from ajcn.nutrition.org by guest on June 14, 2017
sessing and comparing indirect effects in multiple mediator models.
Behav Res Methods 2008;40:879–91.
35. Chrysohoou C, Panagiotakos DB, Pitsavos C, Das UN, Stefanadis C.
Adherence to the Mediterranean diet attenuates inflammation and co-
agulation process in healthy adults: The ATTICA Study. J Am Coll
Cardiol 2004;44:152–8.