Journal of Diabetes Research

Effects of Lifestyle Modifications and Dietary Habits on Prevention of Diabetes and Cardiovascular Disease

Guest Editors: Girardin Jean‑Louis, Samy McFarlane, and Charles Agyemang

Effects of Lifestyle Modificationsand Dietary Habits on Prevention ofDiabetes and Cardiovascular Disease

Journal of Diabetes Research

Effects of Lifestyle Modificationsand Dietary Habits on Prevention ofDiabetes and Cardiovascular Disease

Lead Guest Editor: Girardin Jean-LouisGuest Editors: Samy McFarlane and Charles Agyemang

Copyright © 2017 Hindawi. All rights reserved.

This is a special issue published in “Journal of Diabetes Research.” All articles are open access articles distributed under the CreativeCommons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the originalwork is properly cited.

Editorial Board

Steven F. Abcouwer, USAReza Abdi, USAAbdelaziz Amrani, CanadaFabrizio Barbetti, ItalyMichaelangela Barbieri, ItalySimona Bo, ItalyVirginia Boccardi, ItalyAntonio Brunetti, ItalyMarco Bugliani, ItalyMonica Bullo, SpainRiccardo Calafiore, ItalyStefania Camastra, ItalyNorman Cameron, UKIlaria Campesi, ItalyRiccardo Candido, ItalySergiu Catrina, SwedenSubrata Chakrabarti, CanadaMunmun Chattopadhyay, USAEusebio Chiefari, ItalySecundino Cigarran, SpainKim Connelly, CanadaLaurent Crenier, BelgiumStephane Dalle, FranceChristophe De Block, BelgiumKhalid M. Elased, USAUlf J. Eriksson, SwedenRosa Fernandes, PortugalPaolo Fiorina, USAAndrea Flex, ItalyDaniela Foti, ItalyGeorgia Fousteri, ItalyMaria Pia Francescato, ItalyPedro M. Geraldes, CanadaSanjoy Ghosh, CanadaEric Hajduch, France

Erifili Hatziagelaki, GreeceThomas J. Hawke, CanadaOle Kristian Hejlesen, DenmarkSeok Jong Hong, USAGianluca Iacobellis, USAAkihiro Kakehashi, JapanYasphal Kanwar, USARenu A. Kowluru, USADaisuke Koya, JapanShinji Kume, JapanVaia Lambadiari, GreeceFrida Leonetti, ItalyAfshan Malik, UKRoberto Mallone, FranceChiara Mameli, ItalyRaffaele Marfella, ItalyMichel Marre, FranceCarlos Martinez Salgado, SpainLucy Marzban, CanadaTakayuki Masaki, JapanRaffaella Mastrocola, ItalyWilliam Mayhan, USADavid Meyre, CanadaMaria G. Montez, USAJiro Nakamura, JapanPratibha V. Nerurkar, USAMonika A. Niewczas, USAFrancisco Javier Nóvoa, SpainCraig S. Nunemaker, USAHiroshi Okamoto, JapanIke S. Okosun, USAFernando Ovalle, USACesare Patrone, SwedenSubramaniam Pennathur, USAJonathan M. Peterson, USA

Bernard Portha, FranceGiuseppe Pugliese, ItalyEd Randell, CanadaJordi Lluis Reverter, SpainMaria Rosaria Rizzo, ItalyUlrike Rothe, GermanyChristoph H. Saely, AustriaYoshifumi Saisho, JapanPonnusamy Saravanan, UKToshiyasu Sasaoka, JapanFerdinando Carlo Sasso, ItalyFerdinando Carlo Sasso, ItalyAndrea Scaramuzza, ItalyYael Segev, IsraelSuat Simsek, NetherlandsMarco Songini, ItalyJanet H. Southerland, USAVincenza Spallone, ItalyDavid Strain, UKBernd Stratmann, GermanyAkira Sugawara, JapanKiyoshi Suzuma, JapanPatrizio Tatti, ItalyFarook Thameem, USAMichael J. Theodorakis, UKPeter Thule, USAAndrea Tura, ItalyRuben Varela-Calvino, SpainChristian Wadsack, AustriaKazuya Yamagata, JapanMark Yorek, USALiping Yu, USADavid Zangen, IsraelDan Ziegler, Germany

Contents

Differential and Combined Effects of Physical Activity Profiles and Prohealth Behaviors on DiabetesPrevalence among Blacks andWhites in the US Population: A Novel Bayesian Belief Network MachineLearning AnalysisAzizi A. Seixas, Dwayne A. Henclewood, Aisha T. Langford, Samy I. McFarlane, Ferdinand Zizi,and Girardin Jean-LouisVolume 2017, Article ID 5906034, 10 pages

Rs46522 in the Ubiquitin-Conjugating Enzyme E2Z Gene Is Associated with the Risk of CoronaryArtery Disease in Individuals of Chinese Han Population with Type 2 DiabetesDifei Lu, Jia Huang, Xiaowei Ma, Nan Gu, Junqing Zhang, Hong Zhang, and Xiaohui GuoVolume 2017, Article ID 4501794, 5 pages

The Predictive Role of Tooth Extractions, Oral Infections, and hs-C-Reactive Protein for Mortality inIndividuals with and without Diabetes: A Prospective Cohort Study of a 12 1/2-Year Follow-UpLise Lund Håheim, Kjersti S. Rønningen, Morten Enersen, and Ingar OlsenVolume 2017, Article ID 9590740, 9 pages

Association between the Time of Length since Smoking Cessation and Insulin Resistance inAsymptomatic KoreanMale Ex-SmokersKo-Woon Kim, Sung-Goo Kang, Sang-Wook Song, Na-Rae Kim, Jun-Seung Rho, and Yun-Ah LeeVolume 2017, Article ID 6074760, 7 pages

The Effectiveness of Lifestyle Adaptation for the Prevention of Prediabetes in Adults: A SystematicReviewGeorge Kerrison, Richard B. Gillis, Shahwar I. Jiwani, Qushmua Alzahrani, Samil Kok, Stephen E. Harding,Ian Shaw, and Gary G. AdamsVolume 2017, Article ID 8493145, 20 pages

Reducing Caloric Intake Prevents Ischemic Injury andMyocardial Dysfunction and Affects AnestheticCardioprotection in Type 2 Diabetic RatsCharissa E. van den Brom, Christa Boer, Rob F. P. van den Akker, Stephan A. Loer, and R. Arthur BouwmanVolume 2017, Article ID 4126820, 10 pages

Case Finding andMedical Treatment of Type 2 Diabetes among Different Ethnic Minority Groups: TheHELIUS StudyMarieke B. Snijder, Charles Agyemang, Ron J. Peters, Karien Stronks,Joanna K. Ujcic-Voortman, and Irene G. M. van ValkengoedVolume 2017, Article ID 9896849, 8 pages

Research ArticleDifferential and Combined Effects of Physical Activity Profiles andProhealth Behaviors on Diabetes Prevalence among Blacks andWhites in the US Population: A Novel Bayesian Belief NetworkMachine Learning Analysis

Azizi A. Seixas,1 Dwayne A. Henclewood,2 Aisha T. Langford,1 Samy I. McFarlane,3

Ferdinand Zizi,1 and Girardin Jean-Louis1

1Department of Population Health, Center for Healthful Behavior Change, NYU School of Medicine, 227 East 30th Street, New York,NY 10016, USA2Booz Allen Hamilton Inc., Boston, MA, USA3Division of Endocrinology, Department of Medicine, State University of New York (SUNY) Downstate Medical Center, Brooklyn,NY 11203, USA

Correspondence should be addressed to Azizi A. Seixas; azizi.seixas@nyumc.org

Received 19 June 2017; Accepted 17 July 2017; Published 27 August 2017

Academic Editor: Daniela Foti

Copyright © 2017 Azizi A. Seixas et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The current study assessed the prevalence of diabetes across four different physical activity lifestyles and infer through machinelearning which combinations of physical activity, sleep, stress, and body mass index yield the lowest prevalence of diabetes inBlacks and Whites. Data were extracted from the National Health Interview Survey (NHIS) dataset from 2004–2013 containingdemographics, chronic diseases, and sleep duration (N = 288,888). Of the total sample, 9.34% reported diabetes (where theprevalence of diabetes was 12.92% in Blacks/African Americans and 8.68% in Whites). Over half of the sample reportedsedentary lifestyles (Blacks were more sedentary than Whites), approximately 20% reported moderately active lifestyles (Whitesmore than Blacks), approximately 15% reported active lifestyles (Whites more than Blacks), and approximately 6% reportedvery active lifestyles (Whites more than Blacks). Across four different physical activity lifestyles, Blacks consistently had a higherdiabetes prevalence compared to their White counterparts. Physical activity combined with healthy sleep, low stress, and averagebody weight reduced the prevalence of diabetes, especially in Blacks. Our study highlights the need to provide alternative andpersonalized behavioral/lifestyle recommendations to generic national physical activity recommendations, specifically amongBlacks, to reduce diabetes and narrow diabetes disparities between Blacks and Whites.

1. Introduction

The prevalence of dysglycemia, prediabetes, and diabetesamong blacks is almost double the rate among whites (13.2versus 7.6% in Blacks and Whites, resp.). Furthermore,Blacks have poorer glycemic control and higher diabetes-related mortality and morbidity such as amputations, renaldisease, and diabetes-related blindness [1–7]. Biology, com-munity/neighborhood, poor or limited healthcare, andbehavior/lifestyle are all potential contributors to this dispar-ity [8]. Of the possible causes for this disparity, behavioral/

lifestyle factors (e.g., diet, exercise/physical activity, inade-quate podiatry care, and monitoring blood glucose levels)are considered the most proximal and easily modifiable.Specifically, exercise/physical activity might provide the mosteffective population health benefits for diabetes [9] as theCenter for Disease Control and Prevention (CDC) estimatesthat 15–30% of individuals with prediabetes will developT2D within five years if they do not engage in regularmoderate-vigorous physical activity and lose weight.

Despite the universal health benefits of exercise/physicalactivity in diabetes prevention and management, Blacks

HindawiJournal of Diabetes ResearchVolume 2017, Article ID 5906034, 10 pageshttps://doi.org/10.1155/2017/5906034

https://doi.org/10.1155/2017/5906034

compared to Whites are less likely to adhere to the UnitedStates Department of Health & Human Services (USDHHS)recommendations of at least 150 minutes of moderate-vigorous physical activity per week for optimal healthbenefits [10], and as a result, they may not receive the samehealth benefits, therefore further widening diabetes dispar-ities between Blacks and Whites [11, 12]. Poor adherence tocrude physical activity recommendations, such as theUSDHHS recommendations, may be partly due to the lackof personalized and precise interventions. It is thereforelikely that differences in exercise/physical activity contrib-ute to differences in diabetes prevalence between Blacksand Whites. However, little is known about which physicalactivity lifestyles—combinations of quality and intensity(moderate or vigorous), quantity (duration of physicalactivity), and frequency—could reduce diabetes disparitybetween Blacks and Whites. Additionally, little is knownabout the additive effects other behaviors and lifestylefactors (sleep, stress, and body mass index) may have ondiabetes prevalence when combined with physical activity,especially among Blacks and Whites.

To address these central questions, we developed amachine learning model, specifically a Bayesian BeliefNetwork machine learning model that utilizes advancedmathematical algorithms that iteratively learn complexrelationships and deep insights from big data, using a pub-licly available national dataset (National Health InterviewSurvey) to (1) estimate the prevalence of sedentary, moder-ately active, active, and very active lifestyles; (2) determinethe prevalence of diabetes across the different activitylifestyles; (3) determine the prevalence of diabetes acrossdifferent activity lifestyles between Blacks and Whites; and(4) simulate which physical activity profiles, as well as combi-nation of behaviors and lifestyle factors (sleep, stress, andbody mass index), among Blacks would yield a similar diabe-tes prevalence as Whites. Altogether, we anticipate that ourfindings will provide more nuanced psychical activity profilesoutside of the WHO’s guidelines that are associated with lowand high diabetes prevalence in Blacks and Whites.

2. Methods

2.1. Sample. Analysis was based on the data from the2004–2013 National Health Interview Survey (NHIS) dataset.The NHIS dataset is a nationally representative population-based study of noninstitutionalized adults from 50 US states.Sociodemographic, behavioral, and health conditions andphysician-diagnosed chronic disease data were obtainedthrough face-to-face interviews using computer-assistedpersonal interviewing (CAPI). Interviewers enter partici-pants’ responses in CAPI via a laptop computer, whichmade administration of surveys efficient and improveddata quality.

2.2. Variables. Variables of interest were derived from athree-step system. First, using PubMed, we reviewed theliterature on potential T2D and cardiometabolic risk factors.Second, we compiled a list of T2D and cardiometaboliccondition risk factors based on whether data was available

for that variable for the years 2004–2013 in the NHIS. Third,the list was winnowed to thirty-four variables, whichincluded (1) target variables—moderate and vigorous physi-cal activity (e.g., leisurely walking/bicycling, slow swimming/dancing, running, lifting weights, and simple gardeningactivities); (2) sleep duration (short, average, and long); and(3) covariates (sociodemographic, general health, andchronic health conditions, health risk behaviors, and historyof physician-diagnosed T2D) (see Table 1).

We acknowledge that our list of T2D risk factors and cor-relates are not exhaustive, which is partly due to the limitedselection of clinical factors and biomarkers in the NHIS.Once the data were retrieved, NHIS sample weights wereapplied. Additionally, we adjusted the effects of clustering,stratifying, and oversampling for specific populationsubgroups such as non-Hispanic blacks and individualswho are ±65 years old. After selecting the available variablesfrom the NHIS, the final data set contained 288,888 cases.

2.2.1. History of Diabetes. Individuals were asked whetherthey ever had a physician who diagnosed them with diabetes.

2.2.2. Moderate Physical Activity. These are activities that“cause only light sweating or a slight-to-moderate increasein breathing or heart rate.” The Field Representative’sManuals for the 1997 NHIS provide examples of light ormoderate activities. Examples of moderate physical activityinclude leisure walking or bicycling, slow swimming ordancing, and light gardening. Participants were also askedabout the duration and frequency of their activity.

2.2.3. Vigorous Physical Activity. These are activities that“cause heavy sweating or large increases in breathing or heartrate.” The Field Representative’s Manuals for the 1997 NHISprovide examples of vigorous activities. Examples of vigorousphysical activity include fast walking, fast bicycling, jogging,strenuous swimming or sports play, vigorous aerobic dance,and strenuous gardening. Participants were also asked aboutthe duration and frequency of their activity.

2.2.4. Sedentary and Active Lifestyles. A sedentary lifestylewas defined as engaging in less than 10 minutes of moderateor vigorous physical activity less than two days per week.While an active lifestyle was defined as engaging in morethan 30 minutes of moderate or vigorous physical activityfor three or more days per week.

2.2.5. Sleep Duration. Participants were asked “On average,how many hours of sleep do you get in a 24-hour period.”Hours of sleep were entered in whole numbers. Sleepduration reports greater than or equal to thirty minutes (halfhour) were rounded up to the next whole number, and thoseless than or equal to twenty-nine minutes were dropped. Forexample, total sleep time of 6 hours and 45 minutes wasrounded up to a total sleep time of 7 hours, and a total sleeptime of 6 hours and 5 minutes was rounded down to a totalsleep time of 6 hours.

2.3. Machine Learning Analysis. Since traditional regression-based models are sensitive to the effect of confounders,

2 Journal of Diabetes Research

especially in highly complex models, we decided to use BBNmachine learning modeling because it is robust to the effectsof confounders. BBN machine learning models allow us toinvestigate highly complex and dimensional networkedmodels and relationships without the interference of con-founders and determine hidden insights in big data.

BBN machine learning modeling, unlike traditionalregression analysis models, is better at (1) investigatingdynamic relationships in big data (data with high levels ofvolume, velocity, or variety); (2) predictive modeling indetermining Y in Y= f(X), especially in dynamic modelingand big data; and (3) providing a more accurate explanatorymodel that captures a model structure that best fits the data,specifically causal inference, as they provide one of the mostefficient methods of optimizing data. Its benefits can berevolutionary for the field of medicine as it can yield moreaccurate and dynamic relationships among multiple factors,a priori and a posteriori [13]. We used the BayesiaLabVersion 6.07 statistical software package to construct thecurrent BBN model.

BayesiaLab offers a suite of advanced mathematicaland analytical applications predicated on sophisticatedalgorithms and artificial intelligence that generate structuraldirected acyclic graphs (DAGs). BayesiaLab can utilize simu-lation modeling or machine learning modeling of data basedon unsupervised and supervised machine learning algo-rithms that may be employed to derive the conditionalprobabilities of highly dimensional relationships. The resul-tant of applying the most appropriate learning algorithm isa directed acyclic graph (DAG), with nodes representingvariables and arcs indicative of conditional relationshipsbetween nodes. In the resulting DAG, not all nodes may bea part of the learned network. The selected algorithm, thedata at hand, and the calculated correlation among thevariables will determine whether or not a given node isincluded in the network. For the current model, we utilizedthe Tree Augmented Naïve Bayes model algorithm becauseof its ability to accurately and precisely quantify joint proba-bility distributions and omnidirectional relationships amongdiabetes correlates and diabetes.

2.3.1. Probability Graphical Models (PGMs) and ConditionalProbability Joint Distribution. The BBN machine learningmodel is a probabilistic graphical model that uses graphicalrepresentations to capture the rich independencies andjoint distribution of a highly dimensional and complexnetworked model in space. The independencies and rela-tionships among variables are conditional, and makingobservational inferences is based on applying conditionalevidence to a given node/variable which propagatesinformation throughout the model and updates informationon nodes that are conditionally linked—the conditionalprobability joint distribution.

2.4. Preparing the Data. Two of the core phases of data inges-tion or preparing data include (1) preprocessing the data tofocus on the variables of interest as well as accuratelyrepresenting valid and missing data entries and (2) binningthe data in their most appropriate bins to support the taskat hand while remaining in line with the standard practice(see Table 1 for bins and percentage prevalence for eachbin). Having the data in the form necessary to supportproposed analysis, the data was broken into two, a learningmodel and test model. The learning model which consistedof 231,111 cases was used as the training data set, while the

Table 1: Differences in sociodemographic, health behavior, andchronic health conditions among Blacks and Whites (N = 288,888).

VariablesTotal sample

(%)Blacks(%)

Whites(%)

Gender

Female 55.77 55.62 55.80

Male 44.23 44.38 44.20

Age

≤45 yrs 48.10 46.69 48.3746–65 yrs 32.94 33.47 32.84

>65 yrs 18.96 19.84 18.80Income

$75K 24.11 23.84 24.19Education (college) 44.76 44.93 44.64

Usual medical care 83.11 83.59 83.02

Region

Northeast 16.67 16.22 16.82

Northcentral/midwest 22.00 17.55 23.70

South 36.84 56.29 34.19

West 24.49 9.98 25.29

Weight

Overweight 37.03 36.92 37.05

Obese 25.67 26.63 25.49

Alcohol

Moderate 14.19 13.89 14.25

Heavy 4.99 4.87 5.01

CKD 2.00 2.24 1.96

Hypertension 30.84 32.54 30.52

Sedentary lifestyle

Moderate PA (≤10 minutes) 51.89 52.30 51.81Vigorous PA (≤10 minutes) 62.42 63.07 62.28

Moderately active lifestyle

Moderate PA (11–30 minutes) 26.81 26.62 26.85

Vigorous PA (11–30 minutes) 14.27 14.08 14.31

Active lifestyle

Moderate PA (31–60 minutes) 15.03 14.88 15.06

Vigorous PA (31–60 minutes) 16.34 16.02 16.41

Very active lifestyle

Moderate PA (>60 minutes) 6.27 6.21 6.28Vigorous PA (>60 minutes) 6.97 6.83 7.00

Diabetes 9.34 12.92 8.68

Note. Alcohol =moderate-to-heavy alcohol drinker; CKD= physician-diagnosed chronic kidney disease; hypertension = physician-diagnosedhypertension; overweight = 25 to 29.9 kg/m2; obese = ≥30 kg/m2.

3Journal of Diabetes Research

test model which consisted of 57,777 was used to evaluatehow well the BBN model is able to represent data beyondthe developed model. After data ingestion, 35 variables/nodes(34 factors and diabetes) were used to build a BBN to inves-tigate the effects of sleep duration and physical activity ondiabetes, in the context of demographic, socioeconomic,and general health conditions and health behavior factors(see Figure 1).

2.5. Validating the Model and Missing Data. BayesiaLabanalytical package was selected for the current study becauseof its ability to accurately treat missing values through multi-level imputation. Missing values are not uncommon in largepopulation-based data sets like the NHIS. While missingvalues do represent a lack of information from an incompleterecord, retaining incomplete records for subsequent analysesis valuable, and as such, the means of appropriately treatingmissing values is critical [13].

Once a network has been defined to best represent thedata at hand, a series of calibration exercises are thenperformed to examine the model to ensure that relationshipsidentified by the learning algorithm are valid and intuitiveand ensure that the network is not over- or underfitting thedata. Once calibrated, the model must then be validated todetermine the validity/level of confidence, in order to makeinferences from the model [13].

The BBN is an effective modeling tool as it provides asnapshot and high-level overview of relationships underlyingthe data. Its value lies in its ability to interact with the modeland observe changing conditional probabilities, reflecting theimpact of the observation of given variable states (bins).Interacting with the model is an effective approach by which

to derive observational inference—making inferences basedon observation in the collected data. While this level ofinference does provide insights into the data and imbeddedrelationships, such networks lend themselves to furtheranalyses to derive predictive inferences [13].

3. Results

3.1. Descriptive Statistics. The mean age of the sample was47.83± 18.04 years, and 48.1% were ≤45 years. Of the sample,77.4% were White; 15.9%, Black/African American; and6.7%, “other”; 44.3% were male and 55.7% were female;45.3% reported an annual income less than $35,000 and24.0% reported an annual family income of $75,000 andhigher; 44.76% did not go to college; 83.11% had a usualplace where they sought medical care; and more than athird of the sample lived in southern states of the US(36.8%). Regarding health behaviors and conditions,62.7% were overweight/obese (BMI≥ 25 kg/m2), 19.18%were moderate-to-heavy alcohol drinker, 43.6% smokedmore than 100 cigarettes in their lifetime, 2% reportedphysician-diagnosed chronic kidney disease, and 30.84%reported physician-diagnosed hypertension.

Of the total sample, 51.89% reported less than or equalto 10 minutes of moderate physical activity, 62.42% reportedless than or equal to 10 minutes of vigorous physical activity(sedentary lifestyle), 26.81% reported 11–30 minutes ofmoderate physical activity, 14.27% reported 11–30 minutesof vigorous physical activity (moderately active lifestyle),15.03% reported 31–60 minutes of moderate physicalactivity, 16.34% reported 31–60minutes of vigorousphysical activity (active lifestyle), 6.27% reported more

Joint probability : 100.00%Log - likehood : 0Cases : 231, 111Total value : 441.472Mean value : 15,223

WEIGHTBMICALC

LIVERCONYRHEIGHT

HEALTHKIDNEYWK

ANGIPECEVCHEARTDIVE

HEARTATTEV

HEARTATTEV

STROKEVAGE

HYPERTENEV

USUALPLMARSTAT

POORYN

INCFAM97ON2

EDUCREC1

VIG10DMIN

MOD10DMIN

RACEA

VIG10FWK

MOD10FWK

WORKEVHRSLEEP

EMPSTATWKTR

DIABETICEV

EMPHYSEMEV

EMODISTRESS

ALCSTAT1

ALCSTAT2SMOKEV

HISPETHREGION

SEX

DIABETICEV

Mean: 1.119 Dev: 0.361Value: 1.119

89.39% 19.34% 2

1.27%3

REGION

Mean: 2.692 Dev: 1.018Value: 2.692

16.67% 1

22.00% 2

36.84% 3

24.49% 4

AGE

Mean: 47.829 Dev: 18.042

Value: 47.829 48.10%

than 60 minutes of moderate physical activity, and 6.97%reported more than 60 minutes of vigorous physicalactivity (very active lifestyle).

Regarding differences between Blacks and Whites, thegreatest concentration of Blacks and Whites was in thesouthern region of the US, which was the only region whereBlacks outnumbered Whites. Overall, Blacks (32.54%) had ahigher prevalence of hypertension compared to Whites(30.52%). Although the overall diabetes prevalence was9.34%, Blacks (12.92%) had a higher prevalence than Whites(8.68%) (see Table 1).

4. Primary Findings

4.1. Sedentary Lifestyle and Diabetes

4.1.1. Sedentary Lifestyle Comparison (Moderate andVigorous PA). We investigated the prevalence of diabetesamong sedentary Blacks and Whites and found that theprevalence of diabetes amongWhites who reported ≤10minsof vigorous physical activity ≤2 times per week was 11.58%,compared to that among their Black counterparts who hada prevalence of 16.96%.

We observed a similar pattern regarding sedentaryindividuals who reported moderate physical activity.Among Whites who reported ≤10mins of moderate physicalactivity ≤2 times per week, the prevalence of diabetes was10.10%, while for their Black counterparts, it was 14.90%(see Table 2).

4.1.2. Observational Inference Comparing Sedentary Whitesand Blacks. To determine whether increasing the frequencyof physical activity per week would significantly lower theprevalence of diabetes among Blacks relative to the averageWhite sedentary individual, we conducted a series ofconditional observational inferences where we tested theprevalence of diabetes if an individual reported either 0, 1,2, 3, 4, 5, 6, or 7 days of moderate or vigorous physicalactivity. We found that in order for the prevalence of diabetesamong sedentary Blacks to be comparable to that amongtheir White counterparts (11.58%), they need to engage invigorous physical activity 6 days per week (12.32%). Forsedentary Blacks who reported some moderate physicalactivity, in order for them to have a similar diabetes preva-lence as Whites (10.10%), they have to exercise 4 days perweek (13.51%). Regardless of moderate or vigorous physicalactivity, the prevalence of diabetes among sedentary Whitesis lower than that among their Black counterparts and evenamong Blacks who exercise more frequently (3, 4, 5, 6, or 7days per week).

4.2. Moderately Active Lifestyle and Diabetes. For moderatelyactive individuals (10–30 minutes of physical activity 4times/week) who reported some moderate physical activity,the overall prevalence of diabetes was 6.52%, 9.12% formoderately active Blacks and 6.05% for moderately activeWhites. For moderately active individuals who reportedsome vigorous physical activity, the overall prevalence ofdiabetes was 4.19%, 5.94% for Blacks and 3.88% forWhites. Similar to the results on sedentary individuals,

increasing the frequency of physical activity (how manydays the individual engaged in moderate physical activity)among Blacks did not lower the prevalence of diabetes tothe level of an average moderately active White individual.

4.3. Active Lifestyle and Diabetes. The overall prevalence ofdiabetes among individuals with an active lifestyle (31–60mins of moderate or vigorous physical activity) was5.87% (for those who reported moderate physical activity)and 3.11% (for those who reported vigorous physicalactivity). Active Whites who reported vigorous physicalactivity 6 days per week had the lowest diabetes prevalenceof 2.87%, and for active Blacks, it was 4.42%. Regardingmoderate physical activity, active Whites (those whoreported 31–60mins of moderate physical activity 4 daysper week) had the lowest diabetes prevalence of 5.44%,while for their Black counterparts, the prevalence was8.25%. We also found that active Blacks who engaged ingreater frequencies of physical activity did not have similardiabetes prevalence as active Whites.

4.4. Very Active Lifestyle and Diabetes. For very activeindividuals who reported moderate physical activity (>60minutes of physical activity 4 times/week), the overallprevalence of diabetes was 5.99%, 8.41% for very activeBlacks and 5.56% for very active Whites. For very activeindividuals who reported vigorous physical activity, theoverall prevalence of diabetes was 3.15%, 4.48% for Blacksand 2.91% for Whites. Similar to previous physically activelifestyles, increasing the frequency of physical activity (howmany days the individual reported moderate physicalactivity) among Blacks did not lower the prevalence ofdiabetes to the level of the average moderately activeWhite individual.

4.5. Physical Activity Profiles Where Blacks and Whites HaveComparable Diabetes Prevalence. Since within-class analyses(only comparing Blacks and Whites with the same physicalactivity lifestyle profiles) demonstrated that Blacks andWhites had different diabetes prevalence, we tested scenariosin which Whites and Blacks would have similar diabetesprevalence. To do so, we tested across the four differentactivity lifestyle profiles. We found that in order for Blacksto have a comparable diabetes prevalence to sedentaryWhites (

Table 2: Observational inference of physical activity lifestyles and the combined effects of sleep, stress, and body mass index on diabetes inBlacks and Whites.

ProfilesDiabetestotal (%)

Blacks with diabetes (%)Whites withdiabetes (%)

Sedentary lifestyle

Moderate PA (≤10mins of moderate physicalactivity ≤2 times per week) 10.84 14.90 10.10

Vigorous PA (≤10mins of vigorous physicalactivity ≤2 times per week) 12.41 16.96 11.58

Observational inference of combined effects of sleep,stress, and body mass index

Moderate PA (≤10mins of moderate physicalactivity 2 days per week for Blacks)

10.84

14.90 (PA 2 times/week)

10.10

13.51 (PA 4 times/week)

11.74 (PA 4 times/week & 7-8 hrs of sleep)

13.08 (PA 4 times/week & low stress)

6.48 (PA 4 times/week & BMI = 18.4 to

very active) and prohealth behavioral/lifestyle factors(body mass index, sleep duration, and stress) reduceddiabetes prevalence among Blacks and differences betweenBlacks and Whites. We found that across all physicalactivity lifestyle profiles, having normal body mass index(18.5–24.9 kg/m2) lowered the prevalence of diabetes themost, followed by sleeping 7-8 hours, then reducing stresslevels, and then increasing the frequency of physicalactivity (see Table 2).

5. Discussion

The current study provides new insights into the effectsdifferent physical activity lifestyle profiles (sedentary,moderately active, active, and very active) and prohealthbehaviors among Blacks and Whites have on diabetes in theUnited States. First, to our knowledge, our paper is one ofthe first to pluralize physical activity recommendations—visa vis the four different physical activity lifestyles—anddescribe the prevalence of diabetes across these lifestylesamong Blacks and Whites. Such an approach decentralizesuniversal one-size-fits-all health behavior and lifestyle rec-ommendations and offers alternative recommendations forindividuals to follow and successfully adhere to. Second,our study attempts to undercut intractable diabetes dispar-ities between Blacks and Whites by offering personalizedphysical activity profiles that would equally benefit Blacksand Whites, where the two groups would have similarlylow diabetes prevalence. Through our findings, it was evi-dent that Whites compared to Blacks had greater diabeteshealth benefits across the different physical activity lifestyleprofiles—sedentary, moderately active, active, and veryactive—which we discuss below. Third, our study indicatesthat physical activity alone cannot reduce diabetes dispar-ities between Blacks and Whites based on observationalinferences, an inferential approach in Bayesian BeliefNetwork machine learning modeling. We learned that forBlacks, other behavioral strategies, such as healthy sleep,

low stress, and normal body mass index (average weight18.4 to 60 minutes, 4 times/week) 5.99 8.41 5.56Vigorous PA (>60 minutes) 3.15 4.48 2.91Observational inference of combined effects of sleep,stress, and body mass index

Moderate PA

8.41 (PA 4 times/week)

7.25 (PA 4 times/week & 7-8 hrs of sleep)

8.13 (PA 4 times/week & low stress)

3.91 (PA 4 times/week & BMI = 18.4 to

Our attempt to pluralize and assess alternative physicalactivity lifestyle recommendations emanated from thegrowing need to solve the high level of physical activitynonadherence in the United States (52.5% of US individualsdo not adhere to Health People 2010 or United StatesDepartment of Health & Human Services recommendations)[14]. Although previous studies suggest that behavioral,clinical- and health-related, cognitive and psychological,demographic, environmental, and social factors mightcontribute to poor adherence to physical activity recommen-dations, to our knowledge, there has been little attentiongiven to how lifestyle factors compromise an individual’sability to engage in recommended physical activity. Of thelittle done in this area, there is promising data indicating thatracial/ethnic minority women tend to engage in less physicalactivity due to lifestyle factors such as family needs and childcare [15]. Minimizing lifestyle factors that impede adherenceto physical activity recommendations might only partiallysolve the issue and is limited in scope because there arecertain lifestyle factors that are not easily modifiable, suchas family and work responsibilities. Therefore, providingphysical activity recommendations that are accommodatingto the heterogeneity of individuals’ lifestyles will likelyincrease overall activity which will decrease diabetes riskand better regulate and manage clinical indicators, such asglucose levels.

Our findings also highlight a possibly inherent racial/ethnic bias in the health benefits of physical activity ondiabetes. Some studies have attributed biological, environ-mental, and lifestyle explanations to the differential benefitsof physical activity between Blacks and Whites. Previousstudies indicate that Blacks tend to have lower metabolicrates and fat oxidation rates compared to Whites. However,evidence indicating that lifestyle factors such as higher caloricintake and more sedentary lifestyle which are greater inBlacks compared to Whites renders the biology argumentinconclusive and highlights the importance of lifestylefactors. Environment also plays an integral role in therelationship between physical activity and diabetes amongBlacks and Whites. Rural versus urban settings may affectthe impact physical activity has on the prevalence of diabetes[16]. Goedecke and Ojuka argue that the growing urbaniza-tion of populations forces individuals to adapt lifestylebehaviors that increase risk of diabetes (such as poor eatinghabits and physical inactivity) [17]. Additionally, individualsfrom rural settings are more likely to be physically inactive[18, 19]. Since the majority of Blacks in the NHIS datasetare from rural and urban settings in the United States, it islikely that environment might partially explain why Blacksappear to reap the same benefits as Whites. Therefore,personalized- and precision-based behavioral healthapproaches—those that deliver targeted and individualizedhealth solutions based on an individual’s biology, environ-ment, and behavior/lifestyle—are needed to address racial/ethnic diabetes disparities.

We found that combining other healthful behavioral andlifestyle factors with physical activity may provide additionalhealth benefits for Blacks. Across all physical activity lifestyleprofiles, we found that increasing frequency to 4 times per

week for moderate physical activity and 6 times per weekfor vigorous physical activity as well as having low stresswas linked to lower diabetes prevalence among Blacks,although not to the prevalence of Whites. Additionally,sleeping 7-8 hours per day on average greatly reduceddiabetes prevalence among Blacks than low stress andincreasing frequency of physical activity. Overall, havingan average body mass index combined with physicalactivity provided the most health benefits for diabetes,where the prevalence of diabetes was reduced by morethan 50% among Blacks across all physical activity lifestyleprofiles. Additionally, Blacks who had an average BMIregardless of physical activity profile had a lower diabetesprevalence compared to Whites who had similar physicalactivity profiles.

The fourth contribution our paper makes is the use ofprobability-based machine learning analytical and modelingtools (Bayesian Belief Network modeling) to make obser-vational inferences about population health which couldbe further used to make inferences about individuals andtypes of individuals. This was evident in our current paperwhere we were able to determine prevalence of diabetesamong different physical activity lifestyles, as well as to testthe impact several combinations of behaviors (duration,frequency, and intensity of physical activity, sleep duration,stress levels, and body mass index) have on diabetes anddetermine which combinations of behavioral/lifestyle healthfactors lowered the prevalence of diabetes among andbetween Blacks andWhites. These types of multidimensionaland multifactorial observational inferences cannot be done intraditional regression-based models, but through BBN, weare able to circumvent these challenges that fraught tradi-tional regression models. Ultimately, this method brings uscloser to determining “what” combination of behaviors maywork for “whom”—physical activity lifestyles.

One major limitation with our findings is the inherentbias in physical activity among racial/ethnic minorities.Previous studies found that self-report physical activityvaries by race/ethnicity where Blacks/African Americanscompared to Whites consistently report lower prevalenceof moderate-vigorous physical activity. It is believed thatself-report questions about physical activity are culturallyand linguistically biased, where Blacks/African Americansconsider work and daily activities forms of exercise, andprivilege certain organized physical activities that are lesscommon in racially ethnic groups [20]. To avoid theseissues, we argue that future studies should utilize bothself-report and objective physical activity instruments toaccurately determine quantity and quality of physicalactivity. Although our study utilized advanced mathemati-cal machine learning modeling to test the omnidirectionaleffects of variables, we were unable to fully parse theconfounding effects medical comorbidities have on physicalactivity and inactivity. Previous studies indicate a dose-response relationship between chronic medical conditionsand physical inactivity, and therefore, it is likely that greaterlevels of medical comorbidity are linked to higher levels ofinactivity which increases risk for other medical conditions.In general, distribution of data may affect results; however,

8 Journal of Diabetes Research

we did apply sample weights which we believe may mitigateinferential issues with restricted sample size. However, wedo acknowledge that applying multiple evidence to ourmodel in observational inference might reduce the samplesize and limit inference, but our large sample size of288,888 may protect against this.

6. Conclusion

Unlike previous studies that emphasized a one-size-fits-allphysical activity standard (150 minutes of moderate physicalactivity), the current study investigated the prevalence ofdiabetes across four physical activity profiles/phenotypes(sedentary, moderate, active, and very active) among Blacksand Whites in a large population-based dataset in the UnitedStates. We argue that such an approach can potentially leadto targeted approaches to physical activity recommendationsand audience segmentation. Our secondary findings are thatsleep, stress, and body mass index may provide additionalhealth benefits when paired with physical activity for certainprofiles of individuals and that this varies by race/ethnicity.Therefore, we highly recommend that diabetes counsellingshould take a personalized and “bundled approach” where avariety of prohealth behaviors are recommended but person-alized to the individual’s profile. Moving forward, ubiquitoussensing wearables may be a solution to more accuratelyassess multiple behaviors without unduly burdening indi-viduals with self-tracking. Data from these wearables canbe potentially integrated into electronic medical healthrecords to improve patient-provider communication andshared decision-making around diabetes care. Future studiesshould assess the acceptability, accuracy, and adherence towearables by race/ethnicity, physical activity level and profile,and diabetes severity and chronicity (e.g., are wearables moreused by newly diagnosed patients or those who are sicker as aresult of their diabetes). Lastly, our use of Bayesian BeliefNetwork modeling is novel, and to our knowledge, we areone of the first to use this highly dimensional and robustartificial intelligence machine learning tool, which allowedus to test our primary and secondary questions underdifferent scenarios while taking into consideration theomnidirectional relationships of target variables andconfounders in a joint network.

Conflicts of Interest

The authors have no conflict of interest to report.

Acknowledgments

This work was supported by funding from the NHLBI(K01HL13452), the NIMHD (R01MD007716), the NINDS(U54NS081765), and the NHLBI (K24HL111315).

References

[1] Centers for Disease Control and Prevention,National DiabetesStatistics Report: Estimates of Diabetes and Its Burden in theUnited States, US Department of Health and Human Services,2014, Atlanta, GA, 2014.

[2] W. H. Herman and R. M. Cohen, “Racial and ethnicdifferences in the relationship between HbA1c and bloodglucose: implications for the diagnosis of diabetes,” TheJournal of Clinical Endocrinology & Metabolism, vol. 97,no. 4, pp. 1067–1072, 2012.

[3] J. K. Kirk, R. B. D'Agostino Jr., R. A. Bell et al., “Disparities inHbA1c levels between African-American and non-Hispanicwhite adults with diabetes: a meta-analysis,” Diabetes Care,vol. 29, no. 9, pp. 2130–2136, 2006.

[4] E. K. Spanakis and S. H. Golden, “Race/ethnic difference indiabetes and diabetic complications,” Current DiabetesReports, vol. 13, no. 6, pp. 814–823, 2013.

[5] L. C. Lanting, I. M. A. Joung, J. P. Mackenbach, S. W. J.Lamberts, and A. H. Bootsma, “Ethnic differences inmortality, end-stage complications, and quality of care amongdiabetic patients: a review,” Diabetes Care, vol. 28, no. 9,pp. 2280–2288, 2005.

[6] X. Zhang, J. B. Saaddine, C. F. Chou et al., “Prevalence ofdiabetic retinopathy in the United States, 2005-2008,” Journalof the American Medical Association, vol. 304, no. 6, pp. 649–656, 2010.

[7] B. A. Young, C. Maynard, G. Reiber, and E. J. Boyko, “Effectsof ethnicity and nephropathy on lower-extremity amputationrisk among diabetic veterans,” Diabetes Care, vol. 26, no. 2,pp. 495–501, 2003.

[8] L. S. Mayberry, E. M. Bergner, R. J. Chakkalakal, T. A. Elasy,and C. Y. Osborn, “Self-care disparities among adults withtype 2 diabetes in the USA,” Current Diabetes Reports,vol. 16, no. 11, p. 113, 2016.

[9] R. J. Sigal, G. P. Kenny, D. H. Wasserman, C. Castaneda-Sceppa, and R. D. White, “Physical activity/exercise and type2 diabetes: a consensus statement from the American DiabetesAssociation,” Diabetes Care, vol. 29, pp. 1433–1438, 2006.

[10] Physical Activity Guidelines Advisory Committee, PhysicalActivity Guidelines Advisory Committee Report, vol. 67, no. 2,p. 683, Washington, DC, USA, 2008.

[11] B. R. Shah, M. Mamdani, L. Jaakkimainen, and J. E. Hux, “Riskmodification for diabetic patients are other risk factors treatedas diligently as glycemia?,” Canadian Journal of ClinicalPharmacology, vol. 11, no. 2, pp. e239–e244, 2004, https://www.scopus.com/inward/record.uri?eid=2-s2.0-33645508538&partnerID=40&md5=ca2151266db4e226e8ead47d62205f5b.

[12] C. J. Crespo, E. Smit, R. E. Andersen, O. Carter-Pokras, andB. E. Ainsworth, “Race/ethnicity, social class and their relationto physical activity during leisure time: results from the ThirdNational Health and Nutrition Examination Survey, 1988–1994,” American Journal of Preventive Medicine, vol. 18,no. 1, pp. 46–53, 2000.

[13] S. Conrady and L. Jouffe, Bayesian Networks and Bayesia Lab:A Practical Introduction for Researchers, Bayesia, USA, 2015.

[14] C. E. Matthews, S. M. George, S. C. Moore et al., “Amount oftime spent in sedentary behaviors and cause-specific mortalityin US adults,” The American Journal of Clinical Nutrition,vol. 95, no. 2, pp. 437–445, 2012.

[15] C. A. Schoenborn and M. Stommel, “Adherence to the 2008adult physical activity guidelines andmortality risk,”AmericanJournal of Preventive Medicine, vol. 40, no. 5, pp. 514–521,2011.

[16] D. L. Richter, S. Wilcox, M. L. Greaney, K. A. Henderson, andB. E. Ainsworth, “Environmental, policy, and cultural factorsrelated to physical activity in African American women,”

9Journal of Diabetes Research

https://www.scopus.com/inward/record.uri?eid=2-s2.0-33645508538&partnerID=40&md5=ca2151266db4e226e8ead47d62205f5bhttps://www.scopus.com/inward/record.uri?eid=2-s2.0-33645508538&partnerID=40&md5=ca2151266db4e226e8ead47d62205f5bhttps://www.scopus.com/inward/record.uri?eid=2-s2.0-33645508538&partnerID=40&md5=ca2151266db4e226e8ead47d62205f5b

Women &Health, vol. 36, no. 2, pp. 91–109, 2002, http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=12487143.

[17] J. H. Goedecke and E. O. Ojuka,Diabetes and Physical Activity,vol. 60, 2014.

[18] J. H. Liu, K. J. Bennett, N. Harun et al., “Overweight and phys-ical inactivity among rural children aged 10-17: a national andstate portrait,” 2007, Retrieved at: http://rhr.sph.sc.edu/report/SCRHRC_ObesityChartbook_Exec_Sum_10.15.07.pdf.

[19] P. D. Patterson, C. G. Moore, J. C. Probst, and J. A. Shinogle,“Obesity and physical activity in rural America,” Journal ofRural Health (Spring), vol. 20, no. 2, pp. 151–159, 2004.

[20] K. Y. Wolin, C. Fagin, N. Ufere, H. Tuchman, and G. G.Bennett, “Physical activity in US Blacks: a systematic reviewand critical examination of self-report instruments,” TheInternational Journal of Behavioral Nutrition and PhysicalActivity, vol. 7, p. 73, 2010.

10 Journal of Diabetes Research

http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=12487143http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=12487143http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=12487143http://rhr.sph.sc.edu/report/SCRHRC_ObesityChartbook_Exec_Sum_10.15.07.pdfhttp://rhr.sph.sc.edu/report/SCRHRC_ObesityChartbook_Exec_Sum_10.15.07.pdf

Research ArticleRs46522 in the Ubiquitin-Conjugating Enzyme E2Z Gene IsAssociated with the Risk of Coronary Artery Disease inIndividuals of Chinese Han Population with Type 2 Diabetes

Difei Lu, Jia Huang, Xiaowei Ma, Nan Gu, Junqing Zhang, Hong Zhang, and Xiaohui Guo

Endocrinology Department, Peking University First Hospital, No. 8, Xishiku Street, Beijing 100034, China

Correspondence should be addressed to Xiaowei Ma; mxiaowei1967@sina.com

Received 17 March 2017; Revised 25 May 2017; Accepted 6 July 2017; Published 3 August 2017

Academic Editor: Samy McFarlane

Copyright © 2017 Difei Lu et al. This is an open access article distributed under the Creative Commons Attribution License, whichpermits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Aims. We investigated the association between ubiquitin-conjugating enzyme E2Z (UBE2Z) gene SNP rs46522 and the risk ofCAD in a Chinese Han population with type 2 diabetes and explored a possible interactive effect with environmental riskfactors of CAD. Methods. 665 patients with T2D were enrolled; 390 were CAD patients and 275 were non-CAD patients.Genotype analysis of rs46522 (T>C) was performed using PCR-RFLP. Results. The SNP rs47522 was associated with therisk of CAD supposing recessive inheritance model (TT versus CC+CT, OR′ = 1 277, 95%CI′ 1.039–1.570, p′ = 0 020) andcodominant model (TT versus CT, OR′ = 1 673, 95%CI′ 1.088–2.570, p′ = 0 019) after adjustment for confounders of CAD.A synergistic effect of rs46522 and BMI was discovered (β = 0 012, p for interreaction= 0.028). In subgroup analysis, minorallele T was significantly associated with CAD in overweight and obesity subgroup (p = 0 034), and the association was alsoproved in recessive model (OR = 1 537, 95%CI 1.075–2.196, p = 0 018). Smokers with genotype TT had threefold risk ofCAD in comparison to nonsmokers with genotype TC or CC (p < 0 001). Conclusions. The SNP rs46522 in UBE2Z gene isassociated with the risk of CAD in the individuals of Chinese Han descent with type 2 diabetes and is of synergistic effectwith BMI.

1. Introduction

Diabetes mellitus is one of the most harmful metabolic dis-eases worldwide, and a major challenge to public health,especially in China [1]. Coronary artery disease (CAD), as amain complication and comorbidity of T2D, is a leadingcause of mortality. Both environmental and genetic factorsplay a critical role in the development of CAD [2]. It isestimated that heritable factors account for 30–60% of theinterindividual variation in the pathogenesis of coronaryheart disease [3]. To date, genome-wide associated studies(GWAS) have identified more than 60 CAD risk loci,and a series of following studies proved their susceptibilityto CAD in different ethnics [4]. In a recent large-scaleassociation analysis, rs46522 on chromosome 17q21.32,in the intronic region of ubiquitin-conjugating enzymeE2Z (UBE2Z) encoding gene, was associated with risk ofCAD in European descent [5]. Meanwhile, the susceptibility

is also proved in an Iranian population [6]. However, a con-tradicted conclusion was conducted in a Japanese popula-tion, indicating the association of rs46522 and CAD mightbe ethnic-specific [7]. UBE2Z is physically located near GIPgene along chromosome 17, while it is known that GIP playsa protective role in atherosclerosis and obesity. In this study,our aim was to evaluate whether or not rs46522 close to GIPwas associated with CAD risk in the Chinese subjects withtype 2 diabetes.

2. Materials and Methods

2.1. Ethics Statement. The study was designed in accordancewith the Helsinki Declaration. The research protocol andinformed consent forms were approved by the research ethicscommittee of Peking University First Hospital. Beforerecruited, all subjects gave informed written consent form.

HindawiJournal of Diabetes ResearchVolume 2017, Article ID 4501794, 5 pageshttps://doi.org/10.1155/2017/4501794

https://doi.org/10.1155/2017/4501794

2.2. Patient Selection. Diagnosis of type 2 diabetes was basedon the 1999 WHO diagnostic criteria, which is defined asfasting blood glucose (FBG) of 7.0mmol/L (126mg/dL) orover and/or 2 hours postprandial blood glucose (PBG) of11.1mmol/L (200mg/dL)) or over. Subjects were selectedfrom those patients who underwent coronary angiographyor coronary artery CT scan during their hospitalization inPeking University First Hospital. 665 unrelated ChineseHan individuals with T2D were enrolled, 390 of whom wereCAD-positive and 275 were CAD-negative. Subjects withCAD were defined as a stenosis size of 50% or above in atleast one of major coronary arteries or their main brancheson cardioangiography at Peking University First Hospital.Non-CAD subjects were confirmed with cardioangiographyor coronary artery CT scan with the stenosis less than 50%.Once recruited, complete medical records written by resi-dents of these patients were collected including the followingdata: age, gender, ethnicity, family, social and medication his-tory, vital signs, body weight, body height, and smoking sta-tus (“ever”or “never,” “ever”defined as having smoked morethan 1 cigarette per day for more than 1 year).

2.3. Single Nucleotide Polymorphism Genotyping. Peripheralblood was withdrawn and stored in EDTA-containing tubes.Genomic DNA was isolated with the salting-out methodfrom peripheral blood leukocytes [8]. The extracted DNAwas stored at −20°C until analyzed.

The SNP rs46522 (T>C) at an intron of UBE2Z gene islocated on chromosome 17q21.32. It was genotyped usingpolymerase chain reaction following restriction fragmentlength polymorphism (PCR-RFLP). Primers were designedas 5′-CATAGTGAGGTGGAGCAGCA and 5′-ACGCAACCAGTAAGGACAGG using Primer 3.0 software. A 25μLPCR system consisted of 1μL DNA template of each subject,1μL primers of both forward and reverse ones (5 pmol/μL),12.5μL 2×PCR Mix of Taq polymerase and dNTP, and10.5μL double distilled water. Annealing temperature wasset at 61°C for 35 cycles. The reaction system of RFLP

was constituted of 7.5μL PCR product, 0.5μL restrictionendonuclease (FokI, NEB corporation), 2μL 10× buffer(NEB corporation), and 10μL double distilled water. After2 hours of water bath heating (65°C), RFLP product wasanalyzed using agarose gel electrophoresis. Furthermore,direct sequencing method was performed in 5% of ran-domized selected samples in order to minimize bias forthe quality control.

2.4. Statistical Analysis. The statistical analyses were carriedout using the Statistical Package for the Social Sciences forWindows (SPSS), version 16.0 (IBM, Chicago, IL, USA). Sta-tistical significance was defined as p value of

rate of genotyping success was greater than 98% for the SNPrs46522. The genotypes were consistent with the results fromdirect sequencing for the randomly chosen samples, and theconsistent rate was 95%. There were no significant differ-ences between CAD and non-CAD groups in the distributionof allele frequencies and genotype frequencies.

Assumed in recessive inheritance model, genotype TTwas associated with a significantly higher risk of CAD com-pared to genotypes TC and CC (p = 0 044), and the signifi-cance remained after adjustment for the confoundersincluding age, gender, history of hypertension and diabe-tes, smoking status, BMI, fasting glucose, and lipid levels(p = 0 020, Table 2). Power was calculated as 80.7%, indicat-ing that the possibility of type II error was less than 20%.

In codominant inheritance model, the carriers of geno-type TT had a higher risk of CAD compared to those of geno-type TC after adjustment for the above covariates (p′ = 0 019,Table 3). No significant difference was found in the twogroups in a dominant inheritance model.

3.3. Subgroup Analysis. Both genetic variations and envi-ronmental factors contributed to the pathogenesis of CAD.While taken as a successive covariate, BMI exhibited asynergistic effect with rs46522 (β = 0 012, p for interreac-tion=0.028) on the CAD risk.

The subjects were further stratified according to theenvironmental risk factors, BMI and smoking habits forsubgroup analyses. In overweight and obesity subgroup,defined as BMI of 24 kg/m2 or above, the frequency of

allele T demonstrated a higher risk of CAD than allele C(OR = 1 341, 95%CI 1.023–1.757, p = 0 034). Also, the car-riers of genotype TT had a higher risk of CAD than carriersof genotypes TC and CC in overweight and obesity subgroup,assumed in recessive model (OR = 1 537, 95%CI 1.075–2.196, p = 0 018, Table 4).

Smoking is a conventional risk factor of CAD. Thesmokers with genotype TT had a risk of CAD even twofoldhigher compared to nonsmokers with genotypes TC andCC (OR = 3 001, 95%CI 1.883–4.785, p < 0 001, Table 5).However, this significance no longer remained after beingadjusted for the conventional confounding factors for CAD.

4. Discussion

Genetic variants including SNPs are involved in multipleaspects of the pathogenesis of human diseases. In recentyears, the functional research of SNPs has increased rapidly,but the genetic background for certain diseases is even morebeyond our knowledge and yet to be explained.

So far, researches on genetic factors of CAD have beenperformed in general populations and revealed new loci asso-ciated with CAD. Meanwhile, a few convincing studies sug-gested that some genes affect both atherosclerosis anddiabetes [9]. Interestingly, a number of the studies includingsome of ours have shown that CAD risk may be partially dif-ferent in diabetic individuals than in general population interms of genetic susceptibility [10].

Table 2: Genotype distribution of rs46522 between CAD and non-CAD groups in recessive inheritance model.

Genotype frequencyCAD group

n (%)Non-CAD group

n (%)OR (95%CI) p value OR′ (95%CI′) p′ value

T/T 194 (49.7) 115 (41.8)

C/T+C/C 196 (50.3) 160 (58.2) 1.377 (1.009–1.880) 0.044 1.277 (1.039–1.570) 0.020

CAD: coronary artery disease. OR′, 95%CI′, and p′ were after adjustment with age, gender, history of hypertension and diabetes, smoking status, BMI, fastingglucose, and lipid levels.

Table 3: Genotype distribution of rs46522 between CAD and non-CAD groups in codominant inheritance model.

Genotype frequencyCAD group

n (%)Non-CAD group

n (%)OR (95%CI) p value OR′ (95%CI′) p′ value

T/T 194 (49.7) 115 (41.8)

C/T 165 (42.3) 135 (49.1) 1.380 (0.998–1.908) 0.051 1.673 (1.088–2.570) 0.019

C/C 31 (7.9) 25 (9.1) 1.360 (0.765–2.418) 0.293 1.182 (0.799–1.749) 0.402

CAD: coronary artery disease. OR′, 95%CI′, and p′ were after adjustment with age, gender, history of hypertension and diabetes, smoking status, BMI, fastingglucose, and lipid levels.

Table 4: Genotype distribution of rs46522 between CAD and non-CAD groups in recessive model in overweight and obesity subgroup.

Genotype frequencyCAD group

n (%)Non-CAD group

n (%)OR (95%CI) p value OR′ (95%CI′) p′ value

T/T 156 (52.0) 86 (41.3) 1.537 (1.075–2.196) 0.018 1.202 (0.950–1.521) 0.125

C/T+C/C 144 (48.0) 122 (58.7)

CAD: coronary artery disease. OR′, 95%CI′, and p′were after adjustment with age, gender, history of hypertension and diabetes, smoking status, fasting glucose,and lipid levels.

3Journal of Diabetes Research

UBZ2E functions as a ubiquitin-conjugating enzyme toactivate ubiquitin-like modifier (UBL), which is transferredonto the target protein with an isopeptide bond formation[11]. As of date, GWAS discovered that rs46522 in UBE2Zgene contributed to CAD susceptibility mainly in Caucasians[5]. Several population-based studies demonstrated that theassociations between rs46522 and CAD varied across differ-ent descents [6, 7]. In the study, SNP rs46522 in UBE2Zwas found contributing to the development of CAD inChinese Han population with type 2 diabetes. Carriers ofgenotype TT had significantly higher risk of CAD in compar-ison with those of genotypes TC and CC, before and evenafter adjustment for conventional confounders of CAD.Our results are consistent with the observations in Europeansand a case-control study in Iranian population [5, 6]. How-ever, a conflicted conclusion arose from a cohort study ingeneral Japanese population [7]. We postulated that the dif-ferent results occurred primarily due to varied populations.The study subjects in our research were type 2 diabetespatients of Chinese Hans, while the Japanese study mainlyfocused on the elderly of Japan.

The potential mechanism by which the genetic variationsin UBE2Z gene contribute to CAD risk was not yet clearlyexplicated. As the SNP rs46522 is located in an intronicregion of UBE2Z gene, one possible explanation is thatrs46522 is in strong linkage disequilibrium with the causalSNPs of gastric inhibitory polypeptide (GIP) gene, whilethe encoding protein GIP could modulate glucose and lipidmetabolism. GIP gene is located on chromosome 17q21.32and contains 6 exons. It was proved that functional variantSNPs in GIP gene were associated with cardiovascular dis-ease in a small-scale case-control study [12]. Interestingly,the leading CAD-associated SNPs in GIP gene from previousstudies were found to be in high linkage disequilibrium with2 functional variants in GIP gene: Ser103Gly (rs2291725) anda slice site variant (rs2291726) in a recent study [13]. Furtherstudy on GIP gene variants and CAD risk could be conductedin Chinese population with or without type 2 diabetes toprove the hypothesis.

Meanwhile, efforts were made to explore if there was apossible association between GIP gene variants and therisk of type 2 diabetes. In a case-control analysis in aSouth Indian population, no significant association wasfound between common variants in GIP (rs2291725 andrs2291726) and diabetes [14]. If the association of GIPgene and CAD was proved, a tentative inference on thisconclusion is that the mechanism of GIP gene influencing

the risk of CAD might be independent of the developmentof diabetes.

As a coronary artery disease-related variant, rs46522 atthe UBE2Z locus yielded significant association with type2 diabetes in a genome-wide association study [15]. Explo-ration could be made to discover the mechanism ofUBE2Z locus affecting both CAD and type 2 diabetes ina molecular level.

The potential interactive effect of rs46522 was investi-gated with traditional environmental confounders in CADrisk. BMI, as one of the known CAD risk factors, displayeda possible effect of interaction with the SNP rs46522 in thestudy (β = 0 012, p for interreaction=0.028). In other words,every 1 kg/m2 increase in BMI resulted in additional 1.2%higher CAD risk in the given genetic group. To date, accu-mulating evidence indicated that BMI and waist circumfer-ence were significantly related to insulin resistance andchronic inflammatory state [16], which might lead to cardio-vascular diseases in conjunction with genetic factors.

As a main risk factor of CAD, smoking was also takeninto consideration in the study. The study showed that thesmokers with genotype TT at rs46522 had significant twofoldincrease in CAD risk compared to nonsmokers with geno-type TC/CC, although the difference did not exist anymoreafter adjustment.

In general, the study can be interpreted that rs46522 isassociated with the CAD risk in patients with type 2 dia-betes, and increased BMI and smoking could increaseCAD risk, especially in the individuals with higher geneticsusceptibility. Clinically, the study provides strong evi-dence for diabetes education that body weight manage-ment and smoking cessation can significantly benefit thepatients from CAD.

However, the following limitations should be acknowl-edged in the study. The number of enrolled individuals wasrelatively small, especially the sample size of non-CADgroup. Also, clinical values of CAD and non-CAD groupswere not perfectly matched, which might bias the results eventhough adjustments were performed. Functional study andperspective study should further be conducted on the geneticvariants in UBE2Z and GIP gene to discover the underlyingpathway and confirm the predicting value for CAD in Chi-nese population, respectively.

In conclusion, the study showed that rs46522 in UBE2Zgene contributed to the development of CAD in individualsof Chinese Han descent with type 2 diabetes and had aninteractive effect with BMI on CAD risk. Genotype TT at

Table 5: The subgroup analysis in smoking status and genotypes of rs46522.

CAD groupn (%)

Non-CAD groupn (%)

OR (95%CI) p value OR′ (95%CI′) p′ value

Nonsmoking +T/C or C/C 102 (26.2) 103 (37.5) 1 1 1 1

Nonsmoking +T/T 87 (22.3) 79 (28.7) 0.899 (0.597–1.354) 0.611 0.969 (0.747–1.257) 0.812

Smoking +T/C or C/C 94 (24.1) 57 (20.7) 1.665 (1.085–2.555) 0.019 1.076 (0.746–1.552) 0.693

Smoking +T/T 107 (27.4) 36 (13.1) 3.001 (1.883–4.785)

rs46522 is a risk factor for CAD, especially in patients withsmoking habits.

Conflicts of Interest

The authors declare that there are no conflicts of interest.

Authors’ Contributions

Xiaowei Ma designed the experiments. Jia Huang performedthe experiments and statistical analysis. Difei Lu wrote thepaper. Nan Gu, Hong Zhang, and Junqing Zhang contributedto collecting information. Xiaohui Guo and Xiaowei Marevised the paper. Difei Lu and Jia Huang equally contributedto the study.

Acknowledgments

The authors are grateful to their colleagues from the Dept. ofCardiology in Peking University First Hospital for their helpin data collection and to all the patients involved in thisstudy. The work was supported by the National 973 project(2006CB503903, 2006CB503908) in China.

References

[1] American Diabetes Association, “Diagnosis and classificationof diabetes mellitus,” Diabetes Care, vol. 27, Supplement 1,pp. S5–S10, 2004.

[2] F. Wang, C. Q. Xu, Q. He et al., “Genome-wide associationidentifies a susceptibility locus for coronary artery disease inthe Chinese Han population,” Nature Genetics, vol. 43,pp. 345–349, 2011.

[3] M. E. Marenberg, N. Risch, L. F. Berkman, B. Floderus, andU. Fairede, “Genetic susceptibility to death from coronaryheart disease in a study of twins,” The New England Journalof Medicine, vol. 330, pp. 1041–1046, 1994.

[4] J. Y. Lee, B. S. Lee, D. J. Shin et al., “A genome-wide associationstudy of a coronary artery disease risk variant,” Journal ofHuman Genetics, vol. 58, pp. 120–126, 2013.

[5] H. Schunkert, I. R. Konig, S. Kathiresan et al., “Large-scaleassociation analysis identifies 13 new susceptibility loci forcoronary artery disease,” Nature Genetics, vol. 43, no. 4,pp. 333–338, 2011.

[6] M. Bastami, S. M. Ghaderian, M. D. Omrani et al., “Evaluatingthe association of common UBE2Z variants with coronaryartery disease in an Iranian population,” Cellular and Molecu-lar Bialogy (Noisy-le-grand), vol. 61, no. 7, pp. 50–54, 2015.

[7] S. Dechamethakun, S. Ikeda, T. Arai, N. Sato, M. Sawabe,and M. Muramatsu, “Association between the CDKN2A/B,ADTRP and PDGFD polymorphisms and the developmentof coronary atherosclerosis in Japanese patients,” Journalof Atherosclerosis and Thrombosis, vol. 21, no. 7, pp. 680–690, 2014.

[8] S. A. Miller, D. D. Dykes, and H. F. Polesky, “A simplesalting-out procedure for extracting DNA from humannucleated cells,” Nucleic Acids Research, vol. 16, no. 3,pp. 1215–1218, 1988.

[9] S. Senemar, M. E. Edraki, and S. Toosi, “Association betweentype 2 diabetes mellitus, biochemical factors and UCSNP-43 polymorphisms of CALPIN-10 gene in patients with

atherosclerosis of coronary artery disease in Southern Iranpopulation,” Journal of Cardiovascular and Thoracic Research,vol. 8, no. 1, pp. 13–19, 2016.

[10] X.Wei, X. Ma, R. Lu et al., “Genetic variants in PCSK1 gene areassociated with the risk of coronary artery disease in type 2diabetes in a Chinese Han population: a case control study,”PloS One, vol. 9, no. 1, article e87168, 2014.

[11] J. Schelpe, D. Monte, F. Dewitte, T. K. Sixma, and P. Rucktooa,“Structure of UBE2Z enzyme provides functional insight intospecificity in the FAT10 protein conjugation machinery,”The Journal of Biological Chemistry, vol. 291, no. 2, pp. 630–639, 2016.

[12] I. Nitz, E. Fisher, C. Weikert et al., “Association analyses ofGIP and GIPR polymorphisms with traits of the metabolicsyndrome,” Molecular Nutrition & Food Research, vol. 51,no. 8, pp. 1046–1052, 2007.

[13] B. S. Winsvold, C. P. Nelson, R. Malik et al., “Geneticanalysis for a shared biological basis between migraine andcoronary artery disease,” Neurology. Genetics, vol. 1, no. 1,article e10, 2015.

[14] D. Sugunan, A. K. Nair, H. Kumar, and A. Gopalakrishnapillai,“A case-control analysis of common variants in GIP withtype 2 diabetes and related biochemical parameters in aSouth Indian population,” BMC Medical Genetics, vol. 11,p. 118, 2010.

[15] M. E. Johnson, J. Zhao, J. Schug et al., “Two novel type 2diabetes loci revealed through integration of TCF7L2 DNAoccupancy and SNP association data,” BMJ Open DiabetesResearch & Care, vol. 2, no. 1, article e000052, 2014.

[16] D. A. Garca-Estévez, D. Araújo-Vilar, A. Saavedra-González,G. Fiestras-Janeiro, and J. Cabezas-Cerrato, “Analysis of therelationship between body mass index, insulin resistance, andbeta-cell function: a cross-sectional study using the minimalmodel,” Metabolism, vol. 53, no. 11, pp. 1462–1466, 2004.

5Journal of Diabetes Research

Research ArticleThe Predictive Role of Tooth Extractions, OralInfections, and hs-C-Reactive Protein for Mortality inIndividuals with and without Diabetes: A ProspectiveCohort Study of a 12 1/2-Year Follow-Up

Lise Lund Håheim,1 Kjersti S. Rønningen,2 Morten Enersen,1 and Ingar Olsen1

1Department of Oral Biology, Faculty of Dentistry, University of Oslo, Oslo, Norway2Department of Paediatric Research, Division for Women and Children, Oslo University Hospital, Rikshospitalet, Oslo, Norway

Correspondence should be addressed to Lise Lund Håheim; a.l.haheim@odont.uio.no

Received 30 December 2016; Revised 19 April 2017; Accepted 9 May 2017; Published 21 June 2017

Academic Editor: Girardin Jean-Louis

Copyright © 2017 Lise Lund Håheim et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

The predictive role of high-sensitivity C-reactive protein (hs-CRP), number of tooth extractions, and oral infections formortality in people with and without diabetes is unclear. This prospective cohort study is a 12 1/2-year follow-up ofthe Oslo II study, a health survey in 2000. In all, 12,764 men were invited. Health information was retrieved from6434 elderly men through questionnaire information, serum measurements, and anthropometric and blood pressuremeasurements. Diabetes was reported by 425 men. Distinct differences were observed in baseline characteristics inindividuals with and without diabetes. In the diabetes group, age and hs-CRP were statistically significant whereas inthe nondiabetes group, age, hs-CRP, number of tooth extractions, tooth extractions for infections and oral infectionscombined, nonfasting glucose, systolic blood pressure, total cholesterol, regular alcohol drinking, daily smoking, andlevel of education were independent risk factors. The number of tooth extractions

progress to the periodontal tissues causing destruction oftooth attachment and eventually tooth loss [9]. Tooth extrac-tions which most often are due to infections such as marginalor apical periodontitis have been found to predict mortalityin individuals with diabetes [10, 11]. In this respect, diabe-tes and periodontal infections have been identified asindependent risk factors for all-cause mortality becausesystematic treatment of these infections improves glycae-mic control in diabetes patients [12–15]. The consensusreport of the Joint EFP/AAP Workshop on Periodontitisand Systemic Diseases provides evidence for the beneficialeffect of blood glucose control by systematic treatment ofperiodontal infections [16].

hs-CRP is a general indicator of inflammation and infec-tion which also includes oral infections. It is recognized as arisk factor for CVD, a major cause for mortality in elderlymen. Therefore, it is relevant to assess how hs-CRP relatesto specific infections such as tooth extractions and oral infec-tions including periodontitis.

This study explores whether tooth extractions, oral infec-tions, or hs-CRP affects mortality in people with regard todiabetes status in a 12 1/2-year follow-up of men from theOslo II study.

2. Material and Methods

2.1. Patient Data and Exposure Variables. Data on diabetesand oral infections were extracted from the database of theOslo II study [17]. The participants were men previouslyinvited to the health screening of the Oslo study in 1972/73(Figure 1) [17, 18]. Women were not invited in the 1972/73screening and as a consequence not in 2000 as a major aimwas to study changes in risk factors in these men. In addition,funding for including women was not available. The infor-mation comprised medical history on specific conditionsincluding diabetes (type 1 or 2 was not specified) and dentaldiseases. The latter included information on tooth extractiondue to periodontal infection, infection of single teeth (pulpi-tis), trauma, orthodontic treatment, and other causes.Number of tooth extractions for any cause was stratified intogroups as follows: 1 = 0-1, 2 = 2–4, 3 = 5–8, 4 = 9–31, 5 = 32.The extraction categories were chosen to accommodate forthe skewed distribution of the number of tooth extractions.Oral infections included marginal periodontitis, apical peri-odontitis, and nondental infections. The term oral infectionsin these analyses is the combined extractions due to toothinfections and any other oral infection. The other risk factorsrecorded and used in these analyses were smoking, alcoholdrinking, insulin injections during the last 12 months, tabletsto reduce blood glucose, cholesterol reducing drugs, and anti-hypertensiva. Blood pressure, height, weight, and waist andhip measures were recorded at the screening. Blood sampleswere taken and analysed for total cholesterol, HDL choles-terol, triglycerides, and nonfasting glucose. hs-CRP was mea-sured for 5323 persons who participated in both the 1972/73and 2000 health surveys (hence fewer numbers of fatal casesand survivors in the statistical analyses). All serum analyseswere performed at the Oslo University Hospital, Ullevål,

Oslo, Norway [19]. The total number of men available foranalysis was 6434.

2.2. Endpoint Ascertainment of Total Mortality. The partici-pants were followed with regard to cause-specific mortalityfrom the screening in 2000 to the 31st of December 2012.Information on mortality was provided by StatisticsNorway after acquiring permission from the NorwegianData Inspectorate and the Regional Committee on Ethicsin Medical Research. The study followed the Helsinkideclaration. Informed consent was given by all participantsthat guided the future use of the screening data and thestored biological material.

2.3. Statistical Analyses. Descriptive statistics of includedvariables used in the analyses were mean with standarddeviation (SD) or number and percent. hs-CRP was alsoanalysed in ln-transformation as it normally has a skeweddistribution. As no marked changes on the level of signif-icance were observed, the variable values were not suffi-ciently skewed. Cox proportional hazards regressionanalyses were used in prediction analyses on mortality ascrude, age adjusted, and in multivariate analyses. The lat-ter analyses were adjusted for known confounders for oralinfections and total mortality, namely, age, daily smoking,and level of education which indicate social status. TheCox regression plot on the hazard displays the cumulativehazard function on a linear scale (Figure 2). A p value

13.6% without did not attend the dentist. Four or morevisits to the general practitioner were common in menwith diabetes, 53.1% versus 22.2%, and no visit wasreported by 8.6% of men with diabetes versus 20.1% in menwithout diabetes.

3.2. Predictors for Mortality. Cox proportional hazards anal-yses (Tables 2-3) showed distinct differences according todiabetes status. Age was as expected a significant predictorfor mortality in the whole cohort. In age-adjusted analyses,hs-CRP (continuous scale) predicted mortality in bothgroups, whilst the number of tooth extractions (continuousscale), oral infections, nonfasting glucose, systolic blood pres-sure, total cholesterol (inversely), hs-CRP (continuous scale),alcohol-drinking pattern, daily smoking, and level of educa-tion (inversely) predicted mortality in people not having dia-betes (Table 2). Total number of teeth extracted, that is,

Although not statistically significant, the hazard rate for thetotal number of tooth extractions were similar in diabetesand nondiabetes.

The plot on the risk of mortality over 12 1/2 years(Figure 2) shows that the cumulative hazard of having diabe-tes was higher than not having diabetes and increased in acurvilinear fashion with time.

4. Discussion

Risk factors for mortality in this 12 1/2-year follow-up studyof men were distinctly different in individuals with diabetescompared to individuals without. This confirmed prior

knowledge of the diabetes risk profile. The general inflamma-tion and infection parameter hs-CRP was a significant pre-dictor for total mortality in individuals with and withoutdiabetes. Oral health infection factors such as the total num-ber of extractions and number of extracted teeth for infectionand oral infections combined were predictors for total mor-tality in individuals without diabetes. However, the hazardrate for the number of extracted teeth was similar for bothgroups. In a former cross-sectional analysis of this study,we found that hs-CRP was elevated in concomitant diseasesand can be viewed as an indicator of the total burden ofinflammations and infections in elderly persons [21]. Formeranalyses (ELISA) of the Oslo II study of antibodies towards

Table 1: Baseline characteristics among persons according to diabetes status. A 12 1/2-year follow-up of Oslo men of the Oslo II study of2000.

Risk factorDiabetesn = 425

Nondiabetesn = 6009

Mean (SD) or n (%) Mean (SD) or n (%)

Oral health

Total number of extractions, n mean (SD) 5.46 (8.36) 4.11 (6.78)∗

Total number of tooth extractions by group, n, test of trend

1 = 0-1 179 (42.1) 2747 (45.7)∗

2 = 2–4 111 (26.1) 1815 (30.2)

3 = 5–8 58 (13.6) 731 (12.2)

4 = 9–31 60 (14.1) 572 (9.5)

5 = 32 17 (4.0) 144 (2.4)

Tooth extraction by infection (TE), n (%) 267 (62.8) 3350∗ (55.7)

Periodontitis (PI), n (%) 49 (12.5) 342∗ (6.0)

TE and PI combined (OI), n (%) 280 (65.9) 3510∗ (58.4)

Medication

Antihypertensive drugs, n (%) 196 (54.3) 1507∗∗∗ (28.0)

Cholesterol-reducing drugs, n (%) 113 (33.0) 899∗∗∗ (17.0)

Take insulin, n (%) 342 (80.5) —

Glucose-reducing drugs, n (%) 222 (52.2) —

General measurements

Age in 2000, years mean (SD) 70.4 (4.7) 69.2∗ (6.3)

Glucose, nonfasting, mmol/l mean (SD) 9.72 (4.0) 5.6∗ (1.2)

Systolic blood pressure, mmHg mean (SD) 149.7 (22.6) 144.0∗ (20.2)

BMI, Kg/mm2 mean (SD) 27.8 (3.8) 26.2∗ (3.3)

Total cholesterol, mmol/l mean (SD) 5.4 (1.1) 6.0∗ (1.1)

Triglycerides, mmol/l mean (SD) 2.3 (1.3) 1.9∗ (1.1)

HDL, mmol/l mean (SD) 1.3 (0.4) 1.4∗ (0.4)

hs-CRP, mmol/l mean (SD) n diabetes 331, nondiabetes 4662 4.5 (9.3) 3.5∗ (9.0)

Alcohol drinking, 4–7 times per week, n (%) 309 (72.7) 3792 ∗ (63.1)

Daily smoking, n (%) 77 (18.1) 1247 (20.8)

Level of education, 1–12 mean (SD) 11.7 (3.6) 12.2∗ (3.4)

Health service use

Visit to the dentist in the last 12 months

None 68 (17.1) 796 (13.6)∗∗

1–3 times 290 (72.9) 4630 (79.4)

4 or more times 40 (10.1) 406 (7.0)

Visit to a general practitioner in the last 12 months

None 35 (8.6) 1175 (20.1) ∗∗∗

1–3 times 155 (38.3) 3389 (57.9)

4 or more times 215 (53.1) 1286 (22.2)∗p < 0 05; ∗∗p < 0 01; ∗∗∗p < 0 001.

4 Journal of Diabetes Research

antigens against oral bacteria in persons with a history ofmyocardial infarction versus those without showed hs-CRPnot to be a predictor when adjusting for the oral infectionindicating a collinearity between the two predictors, hs-CRP and oral infections [22].

Current evidence in this study indicates missing teeth as apredictor for mortality. In a 13-year follow-up of the nationalFINNRISK study of 1997, it was found that ≥9 missing teethwere significantly associated with mortality from CVD, dia-betes, and death of any cause but not stroke [10]. In theSwedish National Study on Aging and Care in Blekinge(SNACBlekinge), 870 dentate persons (age range 60–96)

were followed for 6 years [11]. The surviving dentate individ-uals had more teeth (mean 19.3, SD 7.9) than those who haddied (mean 15.9, SD 7.3; p = 0 001), but periodontitis was notshown to predict mortality. Thus, there are diverging resultspossibly due to different modes of registration of the level ofperiodontitis and oral infections.

This study shows hs-CRP to be a dominant risk factorgiving evidence for vulnerability in individuals with andwithout diabetes when exposed to infection and inflamma-tion. hs-CRP is elevated in infectious diseases and inflamma-tion for a number of diseases including oral infections andhas been shown to be a strong independent predictor of death

Table 2: Age-adjusted univariate Cox analyses for total mortality—individuals with or without diabetes. A 12 1/2-year follow-up of the OsloII study of 2000.

Risk factors

Total mortalityDiabetesn = 425

n fatal cases = 241

Nondiabetesn = 6009

n fatal cases = 2078HR 95% CI p value HR 95% CI p value

Oral health

Total number of extractions, n 1.01 0.99–1.02 0.245 1.01 1.01–1.02

Table3:Multivariateanalyses

∗forprediction

oftotalm

ortalityby

oralinfectionvariablesandhs-C

RPin

diabetesandno

ndiabetes.A12

1/2-year

follow-upof

Oslomen

oftheOsloIIstud

yof

2000.

Tooth

extraction

anycause

Tooth

extraction

anycause,grou

p∗Tooth

extraction

infection(TE)∗

Periodo

ntitis(PI)∗

Oralinfection

s(O

I)∗

hs-C

RPmmol/l

hs-C

RPmmol/l,

quartiles∗

HR∗(95%

CI)∗

pvalue

HR(95%

CI)

pvalue

HR(95%

CI)

pvalue

HR(95%

CI)

pvalue

HR(95%

CI)

pvalue

HR(95%

CI)

pvalue

HR(95%

CI)

pvalue

Diabetes

Oralinfection

variableor

hs-C

RP

1.010

1.048

0.923

1.092

0.941

1.022

1.236

(0.995–1.025)

(0.940–1.169)

(0.684–1.245)

(0.717–1.664)

(0.693–1.277)

(1.011

–1.034)

(1.061–1

.439)

0.210

0.399

0.600

0.682

0.695

in people with diabetes [21, 23–27]. The current study indi-cates that oral infections add to the risk for total mortalityboth in individuals with or without diabetes. It was noticedthat some of the risk factors had similar hazard ratios butwere not significantly different among individuals with dia-betes in contrast to individuals without diabetes. This couldbe due to statistical properties related to the large differencein numbers of participants in the two groups.

Metabolic syndrome that is closely linked to obesity is amajor etiologic factor in type II diabetes and increased levelof hs-CRP is known to be associated with obesity [28–30].The individuals with diabetes had high blood glucose thathas also been shown to increase the hs-CRP level [31]. As aconsequence, the strength of hs-CRP as a risk factor formortality in this study can be due to inflammation as wellas infection. The oral microbiota has for some time beenstudied in relation to antibodies to oral bacteria indicating asystemic effect [9, 32]. Noncultivated bacteria identified bybacterial DNA have expanded the means for studying thesystemic effect of oral infections [9]. The specific infectionsmeasured in this study are characterized by a varying butprogressing disease status that can be characterized as low-grade infections with acute exacerbations [33]. The immunebarrier at the gingival margin has been extensively studied[34]. Periodontal infections have also been found to contrib-ute to an increased level of systemic infections [35, 36].

Gender differences cannot be explored in this study forreasons mentioned earlier. In this study, we wanted to eluci-date the association between oral health and diabetes onmortality as we are concerned about diabetes as a complica-tion to oral health infection and conversely oral healthinfections as a comorbidity in diabetes. In either cases, oralinfection control is of importance to all, but in particular topersons with diabetes. Our results show that persons withdiabetes have experienced significantly more tooth extrac-tions. Judging by our results, persons with diabetes use thehealth service and get necessary medication that assist inthe reduction of comorbidities. We have chosen to adjustfor the level of education, as it is a good proximate for socio-economic status that is considered to be related to oral health,type II diabetes, and mortality. Exploring further the associ-ation between education and these outcomes was not anaim of these analyses; however, it is of interest. Further-more, poor health choices in general may give possibleexplanations for the overlap between diabetes and oralcare. To achieve optimal oral health requires knowledgeand means to follow-up oral health status and otherhealth-promoting habits as physical activity and diet mayplay a part in diabetes.

The self-reported information on diabetes status in thisstudy is confirmed by the basic characteristics reported bythe participants. This reduces any potential selection andinformation bias. This is an important prerequisite for givingtrust to the results on risk factors for total mortality aspresented. Other previous studies have estimated the validityof specific self-reported information in health studies [37, 38].Self-reported diabetes was over 92% reliable over time in theAtherosclerosis Risk in Communities Study [37]. The self-reported information is, however, a limitation of the current

study as clinical oral measurements were not recorded. How-ever, Buhlin et al. concluded from their study that the ques-tionnaire information of the number of remaining teeth anduse of removable dentures are reliable [38]. Less reliable isinformation regarding specific periodontal variables. Buhlinet al. considered, however, such data to be a valuable toolfor epidemiological studies of periodontal health. To assessthe direction of any reporting bias causing nondifferentialmisclassification of exposure can be difficult. There was alsoincomplete recording of oral health indicators among all par-ticipants. The strength of the current study is the reliable andcomplete follow-up regarding mortality by Statistics Norwaythat is possible due to the person identification number ofNorwegian citizens.

5. Concluding Remarks

The results indicate that infection and inflammation areassociated with mortality in both individuals with diabetesor those without. Baseline characteristics of persons with dia-betes were distinctly different from person without diabetes.The general indicator for inflammation and infection, hs-CRP, and the infection variables such as the total numberof tooth extractions and tooth extraction and other oralinfection combined independently predicted increased riskfor mortality among elderly men without diabetes. Furtherresearch is needed to get an improved understanding of thepathophysiology of tooth extractions, oral infections, andhs-CRP in diabetes and nondiabetes on total mortality.

Conflicts of Inte