2-3 page

RESEARC

H ARTICLE

& 2017 American Journal of Preventive Medicine. Pu
reserved.

From the 1D
Medical Cen
Medicine, Ba
Hypertension
Boston, Mass
Atlanta, Geo
Wellness, M
Washington
Public Healt
Winston-Sale
Center, Jack
Feinberg Sch

Address
Medicine, T
McCampbell
[email protected]

0749-3797
https://do

blished by Elsevier Inc. All r

Modifiable Lifestyle Risk Factors and Incident Diabetes
in African Americans

Joshua J. Joseph, MD,1,2 Justin B. Echouffo-Tcheugui, MD, PhD,3,4

Sameera A. Talegawkar, PhD,5 Valery S. Effoe, MD,6 Victoria Okhomina, MPH,7

Mercedes R. Carnethon, PhD,8 Willa A. Hsueh, MD,1 Sherita H. Golden, MD2

Introduction: The associations of modifiable lifestyle risk factors with incident diabetes are not well
investigated in African Americans (AAs). This study investigated the association of modifiable
lifestyle risk factors (exercise, diet, smoking, TV watching, and sleep-disordered breathing burden)
with incident diabetes among AAs.

Methods: Modifiable lifestyle risk factors were characterized among 3,252 AAs in the Jackson
Heart Study who were free of diabetes at baseline (2000–2004) using baseline questionnaires and
combined into risk factor categories: poor (0–3 points), average (4–7 points), and optimal (8–11
points). Incidence rate ratios (IRR) for diabetes (fasting glucose ≥126 mg/dL, physician diagnosis,
use of diabetes drugs, or glycosylated hemoglobin A1c ≥6.5%) were estimated using Poisson
regression modeling adjusting for age, sex, education, occupation, systolic blood pressure, and BMI.
Outcomes were collected 2005–2012 and data analyzed in 2016.

Results: Over 7.6 years, there were 560 incident diabetes cases (mean age¼53.3 years, 64% female).
An average or optimal compared to poor risk factor categorization was associated with a 21%
(IRR¼0.79, 95% CI¼0.62, 0.99) and 31% (IRR¼0.69, 95% CI¼0.48, 1.01) lower risk of diabetes.
Among participants with BMI o30, IRRs for average or optimal compared to poor categorization
were 0.60 (95% CI¼0.40, 0.91) and 0.53 (95% CI¼0.29, 0.97) versus 0.90 (95% CI¼0.67, 1.21) and
0.83 (95% CI¼0.51, 1.34) among participants with BMI ≥30.

Conclusions: A combination of modifiable lifestyle factors are associated with a lower risk of
diabetes among AAs, particularly among those without obesity.
Am J Prev Med 2017;53(5):e165–e174. & 2017 American Journal of Preventive Medicine. Published by
Elsevier Inc. All rights reserved.

INTRODUCTION

epartment of Medicine, The Ohio State University, Wexner
ter, Columbus, Ohio; 2Johns Hopkins University, School of
ltimore, Maryland; 3Division of Endocrinology, Diabetes and
, Brigham and Women’s Hospital, Harvard Medical School,
achusetts; 4Rollins School of Public Health, Emory University,
rgia; 5Sumner M. Redstone Global Center for Prevention and
ilken Institute School of Public Health at the George
University, Washington, District of Columbia; 6Division of
h Sciences, Wake Forest University, School of Medicine,
m, North Carolina; 7University of Mississippi Medical
son, Mississippi; and 8Department of Preventive Medicine,
ool of Medicine, Northwestern University, Chicago, Illinois
correspondence to: Joshua J. Joseph, MD, Department of
he Ohio State University, Wexner Medical Center, 566
Hall, 1581 Dodd Drive, Columbus OH 43210. E-mail:
su.edu.
/$36.00
i.org/10.1016/j.amepre.2017.06.018

Type 2 diabetes mellitus (diabetes) is more prev-
alent among African Americans (AAs) compared
to non-Hispanic whites (NHWs).1,2 Recent

trends indicate that diabetes incidence has plateaued
among NHWs, but continues to rise among AAs.1

Modifiable diabetes risk factors such as physical activity,
sedentary behaviors, and smoking are well described
in NHWs;3–5 however, the association of other modifi-
able lifestyle risk factors with diabetes, including sleep
parameters, are less well known. In the U.S.-based
Cardiovascular Health Study,6 among older adults (age
465 years), the low-risk lifestyle group defined by
more favorable physical activity, diet, smoking, alcohol

ights Am J Prev Med 2017;53(5):e165–e174 e165

Joseph et al / Am J Prev Med 2017;53(5):e165–e174e166

consumption, BMI, and waist circumference, had an 89%
lower diabetes risk compared to the high-risk group. In
the American Association of Retired Persons Diet and
Health Study,7 self-reported favorable levels of lifestyle
factors, including BMI, diet, smoking, alcohol consump-
tion, and physical activity were associated with a 72% and
84% lower risk of diabetes among men and women,
respectively. Although these prior studies are consistent
in their findings, they were primarily conducted among
NHWs. Evidence on the role of modifiable risk factors
among AAs is lacking. Data from the Multi-Ethnic Study
of Atherosclerosis suggest that increasing attainment of
ideal cardiovascular health components (including total
cholesterol, blood pressure [BP], dietary intake, tobacco
use, physical activity, and BMI) is associated with lower
risk of incident diabetes among AAs.8 A key limitation of
prior investigations was the inclusion of BMI in com-
bined modifiable lifestyle risk factor metrics, which is
counterintuitive as: (1) obesity may be a transient state in
the pathway to diabetes and (2) the relationship between
adiposity and diabetes may vary by race/ethnicity, as
evidenced by stronger relationships between BMI and
diabetes among NHWs versus AAs.9–11 Thus, this study
investigates the association of lifestyle factors, including
dietary intake, physical activity, sedentary behavior,
sleep-disordered breathing burden (SDBB), and smoking
with incident diabetes in AAs, as well as the modifying
effect of baseline adiposity and glycemia. The hypothesis
is that a combination of higher amounts of healthy
dietary intake and physical activity and lower amounts of
sedentary behavior, SDBB, and smoking will be associ-
ated with lower risk of incident diabetes.

METHODS
Study Sample
The Jackson Heart Study (JHS) is a prospective study of the
development and progression of cardiovascular disease in a cohort
of 5,301 AA adults, aged 21–94 years from the tri-county area of
metropolitan Jackson, Mississippi. Enrollment and baseline exami-
nations were performed from 2000–2004 with two subsequent in-
person follow-up examinations from 2005–2008 and 2009–2013.
Details about the study design, recruitment, and methods used
have been described elsewhere.12 The JHS was approved by the
University of Mississippi Medical Center IRB, and the participants
gave written informed consent.

Measures
Baseline information was obtained using standardized question-
naires including: demographics, occupation (management/profes-
sional versus not), level of education (Bachelor’s degree or higher
versus less than Bachelor’s degree), alcohol use, and current
prescription medication usage. Calibrated devices were used by
certified technicians and nurses to measure participants’ weight,
waist circumference (average of two measurements around the

umbilicus), and height. BMI was calculated as weight (kilograms)/
height (meters)2. Resting seated BP was measured twice at
5-minute intervals using an appropriately sized cuff with standard
Hawksley random-zero instruments and measurements were
averaged. Fasting blood samples were processed and stored using
a standardized protocol.12,13 Fasting glucose and insulin concen-
trations were measured on a Vitros 950 or 250, Ortho-Clinical
Diagnostics analyzer using standard procedures that met the
College of American Pathologists accreditation requirements.13

Insulin resistance was estimated using homeostatic model assess-
ment for insulin resistance (fasting plasma glucose [mg/dL]×fast-
ing plasma insulin [mU/mL])÷405.14 A high-performance liquid
chromatography system was used to measure glycosylated hemo-
globin A1c (HbA1c) concentrations. Serum concentrations of total
adiponectin were measured by an enzyme-linked immunosorbent
assay system with interassay coefficient of variation of 8.8%.15

Modifiable lifestyle risk factors were measured at the baseline
exam (2000–2004), as described in the following. Physical activity
was assessed using the validated JHS Physical Activity Cohort
survey,16,17 and defined according to the American Heart Associ-
ation categorization.18 Physical activity was considered optimal if
participant achieved ≥150 minutes/week moderate intensity or
≥75 minutes/week vigorous intensity physical activity or ≥150
minutes/week moderate/vigorous physical activity; average if
participant performed 1–149 minutes/week moderate intensity
or 1–74 minutes/week vigorous intensity physical activity or 1–149
minutes/week of moderate/vigorous intensity physical activity;
and poor if less than these levels. Time spent watching TV in the
last year was measured in hours/day assessed using the JHS
Physical Activity Cohort survey.16,17 Potential responses were ≥4
hours/day, 2–4 hours/day, 1–2 hours/day, 1–7 hours/week, and
o1 hour/week. TV watching was categorized as optimal (o1
hour/day), average (1–3.99 hours/day), or poor (≥4 hours/day).19

Dietary intake was assessed using a culturally appropriate,
validated 158-item food frequency questionnaire administered
in-person by trained AA interviewers.20,21 Diet quality was
operationalized using the American Heart Association categoriza-
tion with slight modifications.18 Components (based on 2,000
kcal/day intake) included: fruits and vegetables ≥4.5 cups/day; fish
43.5 ounces, twice per week (non-fried); sodiumo1,500 mg/day;
sugar-sweetened beverages o450 kcal/week; and whole grains ≥3
servings/day. Participants were assigned 1 point per ideal dietary
component, for a total score ranging from 0 to 5. Dietary intake
was classified as optimal (4–5 dietary components), average (2–3
dietary components), or poor (0–1 dietary components).

Smoking status was classified as optimal (never smoking or quit
≥12 months ago), average (quit o12 months ago), or poor
(current smoking).8

SDBB was assessed using an analytic method created by Fülop et
al.22 Prevalent sleep symptoms were defined as a positive response
to a limited set of five questions adapted from the Berlin Sleep
Questionnaire.23 SDBB was quantified by first coding the
responses to the sleep symptom questions (never, seldom, some-
times, often, or almost always) from 0 for never to 4 for almost
always and then summing the individual scores, resulting in a sleep
burden score that ranged from 0 to 20. Sleep burden was classified
as “none” (score: 0), “mild” (score: 1–5), “moderate” (score: 6–10),
or “severe” (score: ≥11).22

A modifiable lifestyle risk factor score was created, as in prior
analyses,6–8,24 using the five baseline modifiable lifestyle factors

www.ajpmonline.org

Table 1. Definitions of Modifiable Lifestyle Risk Factors and Total Score

Modifiable lifestyle risk factor

Points

0 1 2 3

Current smoking, months Yes Former, ≤12 Never or quit, ≥12 —

TV watching, hours/day 44 1–4 o1 —

AHA physical activity, minutes/week MVPAa o1 1–149 4150 —

AHA healthy diet score, componentsb 0–1 2–3 4–5 —

SDBBc Severe Moderate Mild None

Note: Modifiable lifestyle risk factor total point groupings and associated scores are as follows: 0–3 ¼ Poor; 4–7 ¼ Moderate; 8–11 ¼ Optimal.
aAHA physical activity: poor health: (1) 0 minutes of moderate physical activity and (2) 0 minutes of vigorous physical activity; average health: (1) 0
o minutes of moderate physical activityo150 or (2) 0 o minutes of vigorous physical activityo75 or (3) 0 o minutes of combined MVPAo150;
and optimal health: (1) minutes of moderate physical activity ≥150 or (2) minutes of vigorous physical activity ≥75 or (3) minutes of combined MVPA
≥150.

bAdapted for Jackson Heart Study: fruits and vegetables: ≥4.5 cups/day (1.08 L); fish: 43.5 ounces (98 grams), twice per week, sodium: o1,500
mg/day, sugar-sweetened beverages: o450 kcal/week, and whole grains: ≥3 servings/day.

cAssessed using an analytic method created by Fülöp et al.22

AHA, American Heart Association; MVPA, moderate to vigorous physical activity; SDBB, sleep-disordered breathing burden.

Joseph et al / Am J Prev Med 2017;53(5):e165–e174 e167

selected a priori: physical activity, TV watching (a proxy of
sedentary behaviors), diet, smoking, and SDBB. Each baseline
metric was given 0 points for poor status, 1 point for average
status, and 2 points for optimal status, except for sleep, which was
assigned points for severity of SDBB: 0 points for severe, 1 point
for moderate, 2 points for mild, and 3 points for none. The
modifiable lifestyle risk factor score was classified into three levels:
poor (0–3 points), average (4–7 points), or optimal (8–11 points)
modifiable lifestyle risk factor status (Table 1).
Diabetes was defined as HbA1c ≥6.5%, fasting blood glucose

≥126 mg/dL, taking diabetes medications, or with a self-reported
physician diagnosis.25 Persons without diabetes at baseline and
meeting criteria for diabetes at one of the two subsequent exams
were considered to have incident diabetes.

Statistical Analysis
Participants with diabetes at baseline (n¼1,152), missing diabetes
status at baseline (n¼61), missing diabetes data at follow-up
(n¼689), or missing data on baseline covariates (n¼147) were
excluded. The 897 excluded participants without known diabetes
at baseline were predominantly male, with lower educational
status, occupational status, BMI, and physical activity, and higher
systolic BP and current smoking (po0.01 for all comparisons,
Appendix Table 1, available online). Baseline characteristics of all
participants were compared using appropriate parametric or non-
parametric tests for continuous variables and the chi-square test
for categorical variables. Spearman’s correlations were compared
between individual modifiable lifestyle risk factors (Appendix
Table 2, available online). The association of modifiable lifestyle
risk factors or risk factor score with incident diabetes was
examined by comparing participants with average or optimal
versus poor status (reference group). Unadjusted diabetes inci-
dence rates for modifiable risk factor scores were calculated using
person−time analysis assuming a Poisson distribution. Poisson
regression modeling was utilized to estimate incident rate ratios
(IRR) for diabetes. Sequential modeling was performed as follows:
Model 1: age, sex; Model 2: Model 1 and education, current
occupation status, alcohol use, and systolic blood pressure; Model
3: Model 2 and BMI; Model 4: Model 2 and waist circumference.
Statistical significance was defined as two-sided αo0.05.

November 2017

Associations of modifiable lifestyle risk factors with incident
diabetes may differ by age, sex, BMI, waist circumference, and
glycemic status, multiplicative interaction testing with application
of the likelihood ratio test was performed with a p-value o0.10
considered statistically significant (Appendix Table 3, available
online). Results for BMI, waist circumference, and glycemic status
were significant and sensitivity analyses were performed with
stratification by: (1) baseline normoglycemia (fasting plasma
glucose o100 mg/dL and HbA1c o5.7%) versus prediabetes
(fasting blood glucose 100–125 mg/dL or HbA1c 5.7%−6.4%);26

(2) central obesity (waist circumference ≥35 inches in women and
≥40 inches in men) versus normal waist circumference;27 and (3)
BMIo30 versus BMI≥30. Analyses were performed in 2016 using
Stata, version 13.1.

RESULTS
The baseline characteristics of the 3,252 participants
stratified by modifiable risk factor categories (poor,
average, and optimal) are presented in Table 2. Partic-
ipants with a more favorable risk factor profile had higher
education and adiponectin, and lower waist circum-
ference, BMI, systolic BP, diastolic BP, homeostatic
model assessment for insulin resistance, but no difference
in HbA1c. Participants in the optimal category of
modifiable risk factor status had higher baseline levels
of factors potentially associated with lower risk of
diabetes, including physical activity and optimal dietary
intake, and lower baseline levels of factors perceived to
increase diabetes risk, including SDBB, TV watching, and
smoking (Table 2).
During a median follow-up of 7.5 years, 560 partic-

ipants developed diabetes (incidence rate 22.9 per 1,000
person years) (Table 2). The unadjusted incident rates
decreased in a monotonic fashion with a rate ratio of 0.93
(95% CI¼0.88, 0.98) per 1 unit in increase in score
(Appendix Figure 1, available online). Diabetes incidence

Table 2. Baseline Characteristics of Participants by Modifiable Diabetes Lifestyle Risk Factor Scorea

Characteristic
All

(n¼3,252)
Poor

(n¼365)
Average
(n¼2,544)

Optimal
(n¼343) p-valueb

Age, years, M (SD) 53.3 (12.5) 52.3 (10.7) 53.5 (12.8) 53.2 (11.5) 0.27
Female, sex, n (%) 2,066 (64) 199 (55) 1,643 (65) 224 (65) 0.001
Education, Bachelor’s degree
or higher, n (%)

1,217 (37) 77 (21) 943 (37) 197 (57) o0.001

Occupation, management/
professional, n (%)

1,287 (40) 92 (25) 1,000 (39) 195 (57) o0.001

Alcohol use, n (%) 1,616 (50) 222 (61) 1,216 (48) 178 (52) o0.001
Systolic BP, mmHg, M (SD) 125 (17) 127 (18) 125 (17) 122 (17) o0.001
Diastolic BP, mmHg, M (SD) 79 (10) 81 (11) 79 (10) 78 (10) 0.004
Waist circumference, cm, M (SD) 98.6 (15.7) 101.1 (17.5) 98.8 (15.6) 94.3(12.9) o0.001
BMI, M (SD) 31.2 (7.0) 31.4 (7.8) 31.4 (7.1) 29.6 (5.4) o0.001
Fasting plasma glucose,c M (SD) o0.001
mmol/L 5.0 (0.5) 5.1 (0.5) 5.0 (0.5) 4.9 (0.4)
mg/dL 90 (9) 91 (9) 90 (9) 89 (8)

Hemoglobin A1c, % (n¼3,176),d

M (SD)
5.5 (0.5) 5.5 (0.5) 5.5 (0.5) 5.5 (0.4) 0.53

HOMA-IR (n¼3,125), M (SD) 3.6 (2.3) 3.8 (2.5) 3.6 (2.2) 3.1 (10.7) o0.001
Adiponectin, ng/mL (n¼3,193),
M (SD)

5,304 (3,866) 4,546 (3,220) 5,327 (3,753) 5,948 (5,027) o0.001

Current smoking, n (%) 383 (12) 227 (62) 154 (6) 2 (1) o0.001
TV watching o1 hour/day, n (%) 468 (14) 8 (2) 298 (12) 162 (47) o0.001
SDBB, none, n (%) 351 (11) 8 (2) 233 (9) 110 (32) o0.001
Ideal AHA physical activity, n (%) 697 (21) 3 (1) 452 (18) 242 (71) o0.001
Ideal AHA dietary intake, n (%) 24 (1) 0.0 (0) 8 (0.3) 16 (5) o0.001
Incident diabetes, rate per
1,000 person years (95% CI)

22.9 (21.1, 24.9) 28.7 (23.0, 35.8) 22.9 (20.8, 25.1) 16.9 (12.6, 22.8) 0.001

Note: Boldface indicates statistical significance (po0.05).
aModifiable diabetes lifestyle risk factor score: poor (score 0–3), average (score 4–7), optimal (score 8–11). Smoking: current smoker (0 points), former
≤12 months (1 point), never or quit ≥12 months (2 points); TV watching:44 hours/day (0 points), 1–4 hours/day (1 point),o1 hour/day (2 points);
AHA physical activity: poor (0 points), intermediate (1 point), ideal (2 points); AHA healthy diet: poor (0 points), intermediate (1 point), ideal (2 points);
sleep-disordered breathing burden (assessed using an analytic method created by Fülöp et al.22): severe (0 points), moderate (1 point), mild
(2 points), none (3 points).

bThe p-values are calculated using chi-square (categorical variables), ANOVA (parametric continuous variables), Kruskal−Wallis test (non-parametric
continuous variables), and Mantel−Cox for incident rate comparisons.

cConversion formula for fasting plasma glucose from mg/dL to mmol/L¼mg/dL×0.0555.
dConversion formula for hemoglobin A1c from % to mmol/mol¼10.93×%−23.5.
AHA, American Heart Association; BP, blood pressure; HOMA-IR, homeostatic model assessment of insulin resistance; SDBB, sleep-disordered
breathing burden.

Joseph et al / Am J Prev Med 2017;53(5):e165–e174e168

rates per 1,000 person years among participants in
categories of poor, average, or optimal modifiable risk
were 28.7 (95% CI¼23.0, 35.8), 22.9 (95% CI¼20.8,
25.1), and 16.9 (95% CI¼12.6, 22.8), respectively, with
a rate ratio of incident diabetes per category of 0.79
(95% CI¼0.66, 0.94) (Table 2). Participants who devel-
oped diabetes had higher baseline BMI (33.6 versus 30.7),
waist circumference (105.0 versus 97.2 cm), systolic BP
(128 versus 124 mmHg), fasting plasma glucose (97
versus 89 mg/dL), and HbA1c (5.9% versus 5.4%) (p for
comparisons o0.001, Appendix Table 4, available
online).
The adjusted IRRs for incident diabetes associated

with baseline modifiable lifestyle risk factors are pre-
sented in Table 3. After adjustment for covariates

including BMI, the direction of the association of
individual risk factors with incident diabetes was as
expected, but non-significant. For the combined modifi-
able risk factors scores, in adjusted analysis without a
measure of adiposity (Model 2), the IRRs for average or
optimal compared to poor categories were 0.79 (95%
CI¼0.62, 1.00) and 0.66 (95% CI¼0.45, 0.96), respec-
tively. After adjustment for covariates including BMI, the
IRRs for average or optimal compared to poor categories
were 0.79 (95% CI¼0.62, 0.99) and 0.69 (95% CI¼0.48,
1.01), respectively. A modifiable risk factor category
increase (poor to average or average to optimal) was
associated with an 18% lower risk of incident diabetes
(p¼0.03). Similar results were seen with adjustment for
waist circumference instead of BMI.

www.ajpmonline.org

Table 3. Associations of Modifiable Diabetes Risk Factors With Incident Diabetes Over 8 Years

Exposure

Incident Diabetes

n/Diabetes
cases

Poisson regression model, IRR (95% CI) (n¼3,252 with 560 cases)

Model 1a Model 2b Model 3c Model 4d

Smoking
Current 383/68 ref ref ref ref
Former ≤12 months 30/5 0.65 (0.26, 1.63) 0.66 (0.27, 1.65) 0.72 (0.29, 1.79) 0.72 (0.29, 1.77)
Never or quit ≥12
months

2,839/487 0.90 (0.70, 1.15) 0.90 (0.70, 1.16) 0.83 (0.64, 1.06) 0.83 (0.65, 1.07)

Continuouse — 0.95 (0.84, 1.08) 0.96 (0.84, 1.08) 0.91 (0.80, 1.03) 0.91 (0.81, 1.04)
TV watching
≥4 hours/day 1,117/211 ref ref ref ref
1–3.9 hours/day 1,667/281 0.95 (0.79, 1.13) 0.96 (0.80, 1.15) 0.97 (0.81, 1.17) 0.99 (0.82, 1.18)
o1 hour/day 468/68 0.88 (0.67, 1.15) 0.91 (0.69, 1.19) 0.95 (0.72, 1.25) 0.95 (0.72, 1.25)
Continuouse — 0.94 (0.83, 1.07) 0.95 (0.84, 1.08) 0.97 (0.86, 1.11) 0.98 (0.86, 1.11)

Physical activity
Poor 1,468/279 ref ref ref ref
Intermediate 1,087/181 0.97 (0.80, 1.16) 0.99 (0.82, 1.20) 1.00 (0.83, 1.21) 1.01 (0.84, 1.22)
Ideal 697/100 0.84 (0.67, 1.06) 0.88 (0.70, 1.11) 0.91 (0.72, 1.15) 0.92 (0.72, 1.16)
Continuouse — 0.93 (0.83, 1.03) 0.95 (0.85, 1.06) 0.96 (0.86, 1.07) 0.97 (0.86, 1.08)

Healthy diet score
Poor 2,063/354 ref ref ref ref
Intermediate 1,165/203 0.92 (0.78, 1.10) 0.95 (0.79, 1.13) 0.93 (0.78, 1.11) 0.94 (0.78, 1.11)
Ideal 24/3 0.64 (0.20, 1.99) 0.66 (0.21, 2.06) 0.74 (0.24, 2.33) 0.77 (0.24, 2.40)
Continuouse — 0.91 (0.77, 1.08) 0.94 (0.79, 1.11) 0.92 (0.78, 1.10) 0.93 (0.78, 1.10)

SDBBf

Severe 113/23 ref ref ref ref
Moderate 880/182 0.95 (0.64, 1.42) 0.97 (0.65, 1.45) 0.95 (0.64, 1.40) 0.96 (0.65, 1.41)
Mild 1,908/299 0.73 (0.49, 1.08) 0.74 (0.50, 1.10) 0.77 (0.53, 1.13) 0.78 (0.54, 1.14)
None 351/56 0.73 (0.46, 1.15) 0.75 (0.47, 1.18) 0.83 (0.53, 1.31) 0.84 (0.53, 1.31)
Continuouse — 0.85 (0.75, 0.95) 0.85 (0.76, 0.96) 0.89 (0.79, 1.01) 0.90 (0.80, 1.01)

Combined modifiable lifestyle risk factors scoreg

Poor 365/79 ref ref ref ref
Average 2,544/437 0.78 (0.61, 0.98) 0.79 (0.62, 1.00) 0.79 (0.62, 0.99) 0.80 (0.63, 1.01)
Optimal 343/44 0.63 (0.43, 0.90) 0.66 (0.45, 0.96) 0.69 (0.48, 1.01) 0.71 (0.49, 1.03)
Continuouse — 0.79 (0.66, 0.94) 0.81 (0.67, 0.97) 0.82 (0.68, 0.98) 0.83 (0.69, 1.00)

Note: Boldface indicates statistical significance (po0.05).
aModel 1: age, sex.
bModel 2: Model 1+education, current occupation status, alcohol use, systolic blood pressure.
cModel 3: Model 2+BMI.
dModel 4: Model 2+waist circumference.
eContinuous per categorical increase in modifiable lifestyle risk factor category.
fAssessed using an analytic method created by Fülöp et al.22
gModifiable lifestyle risk factor score: poor (0–3), average (4–7), optimal (8–11). Smoking: current smoker (0 points), former ≤12 months (1 point),
never or quit ≥12 months (2 points); TV watching: 44 hours/day (0 points), 1–4 hours/day (1 point), o1 hour/day (2 points); AHA physical activity:
poor (0 points), intermediate (1 point), ideal (2 points); AHA healthy diet: poor (0 points), intermediate (1 point), ideal (2 points); SDBB: severe (0
points), moderate (1 point), mild (2 points), none (3 points).
AHA, American Heart Association; IRR, incident rate ratio; SDBB, sleep-disordered breathing burden.

Joseph et al / Am J Prev Med 2017;53(5):e165–e174 e169

For participants with BMI o30, after adjustment for
covariates including BMI, the IRRs for average or
optimal compared to poor categories were 0.60 (95%
CI¼0.40, 0.91) and 0.53 (95% CI¼0.29, 0.97), respec-
tively, compared to 0.90 (95% CI¼0.67, 1.21) and 0.83

November 2017

(95% CI¼0.51, 1.34) among participants with BMI ≥30.
For participants with normal waist circumference, after
adjustment for covariates including BMI, the IRRs for
average or optimal compared to poor categories were
0.55 (95% CI¼0.33, 0.91) and 0.56 (95% CI¼0.26, 1.21),

Joseph et al / Am J Prev Med 2017;53(5):e165–e174e170

respectively, compared to 0.89 (95% CI¼0.68, 1.17) and
0.80 (95% CI¼0.52, 1.22), among participants with
central obesity. For participants with baseline normogly-
cemia, in multivariable adjusted models prior to adjust-
ment for measures of adiposity, the IRRs for average or
optimal compared to poor categories were 0.64 (95%
CI¼0.43, 0.96) and 0.57 (95% CI¼0.31, 1.04) among
participants, respectively, compared to 0.90 (95%
CI¼0.69, 1.19) and 0.80 (95% CI¼0.52, 1.23) among
participants with prediabetes. The findings among nor-
moglycemic participants were attenuated and became
non-significant after adjustment for BMI or waist cir-
cumference (Table 4).

DISCUSSION
In this large, contemporary, prospective cohort study, AAs
with average and optimal modifiable risk factor scores had
a 21% and 31% lower risk of incident diabetes, respec-
tively, compared to participants with a poor modifiable
risk factor score. Compared with previous studies exam-
ining the combined effects of multiple risk factors on the
incidence of diabetes that included AAs,8 this study used a
different approach. First, factors above and beyond
physical activity and dietary intake were accounted for,
namely sedentary behaviors, smoking, and sleep disorders,
that may influence insulin sensitivity. Second, BMI was
not included as a modifiable risk factor, as has been done
previously,6–8,24 but instead, findings were adjusted or
stratified by BMI. The rationale is that obesity is a known
precursor of diabetes9 and weight reduction is proven to
reduce diabetes risk.28 However, it remains difficult to
achieve in large populations of AAs using current
approaches, including community-based translations of
the Diabetes Prevention Program.29,30 Thus, the results
indicate that a cumulative modification of several risk
factors above and beyond physical activity and diet may
lower risk of diabetes independent of adiposity.
Analyses of modifiable lifestyle risk factors with

incident diabetes are limited among AAs. In the Multi-
Ethnic Study of Atherosclerosis, improved levels (ideal
versus poor cardiovascular health) of a combination of
total cholesterol, blood pressure, dietary intake, tobacco
use, physical activity, and BMI were associated with 66%
lower risk of diabetes among AAs (n=1,293).8 Individ-
ually, improved levels of smoking, physical activity, or
dietary intake were not associated with lower risk of
diabetes among AAs, whereas more TV watching was
associated with increased risk of diabetes, consistent with
data from the Black Women’s Health Study.8,19,31 Con-
sistent with the SDBB findings, obstructive sleep apnea
has been linked with incident diabetes in a study
including AAs.32 In this study, the individual effect of

each modifiable lifestyle risk factor on diabetes incidence
was non-significant but in the expected direction, though
this was attenuated after accounting for BMI. These
results differ from larger analyses of NHWs in which
ideal dietary intake,33 higher physical activity,19

SDBB,34,35 and smoking36 were individually associated
with a lower risk of developing diabetes. Results from this
study suggest that the combination of adjusting modifi-
able risk factors to optimal levels is likely to result in the
greatest benefit for lowering diabetes risk among AAs. A
key finding is that the associations varied by BMI, waist
circumference, and baseline glycemic status with the
greater magnitude of associations observed among par-
ticipants with BMI o30, normal waist circumference,
and normoglycemia at baseline. Therefore, AAs at the
lower end of the diabetes risk spectrum may derive
greater long-term benefit from prevention strategies
focused on the outlined modifiable lifestyle risk factors.
A combination of clinical practice guidelines that empha-
size a healthy lifestyle in metabolically normal AAs and
public health policies that focus on primordial preven-
tion by directing resources to increase physical activity
and healthy diet while reducing sedentary activities and
smoking are likely necessary to reduce the burden of
diabetes in AA communities.2,37–40

Limitations
Strengths of the study include the large, socioeconomically
diverse, AA cohort with over a decade of follow-up,
validated questionnaires, and comprehensive ascertain-
ment of diabetes. Additionally, adiposity was accounted
for using BMI and waist circumference to show the
robustness of the findings. Despite these strengths, there
are several limitations. First, the JHS participants are from
one geographic area in the southeastern U.S. and may not
be representative of all AAs. Second, the JHS does not
include other racial/ethnic groups to allow for racial/ethnic
comparisons. Third, although validated,17,21 self-reported
measures of physical activity, dietary intake, and SDBB
were used. Thus, there was a potential for misclassification
and residual confounding by these variables due to lack of
precision compared to objective measures. The 4-point
scoring system for SDBB versus 3 points for the other
components gives a slightly greater “weight” for this
component. Longitudinal tracking of risk factors that
would account for changes over time was not included,
which may have minimized misclassification. Smoking was
considered a modifiable lifestyle risk factor, consistent with
prior studies,6–9 but may also be considered an addiction.41

Lastly, the relationship of modifiable lifestyle risk factors
with incident diabetes may have been underestimated, as
individuals with impaired glucose tolerance, may have
remained undetected.

www.ajpmonline.org

Table 4. Stratified Associations of Modifiable Diabetes Risk Factors With Incident Diabetes Over 8 Years

Exposure

Incident diabetes

n/Diabetes cases

Poisson regression model, IRR (95% CI)

Model 1a Model 2b Model 3c Model 4d

Incident diabetes stratified by WHO BMI categories
Modifiable lifestyle risk factor score, BMI o30 (n¼1,634)

Poor 177/27 ref ref ref ref
Average 1,246/138 0.63 (0.42, 0.93) 0.64 (0.43, 0.96) 0.60 (0.40, 0.91) 0.62 (0.42, 0.92)
Optimal 211/18 0.53 (0.30, 0.96) 0.57 (0.31, 1.04) 0.53 (0.29, 0.97) 0.55 (0.30, 1.00)
Continuouse — 0.71 (0.52, 0.96) 0.73 (0.54, 1.00) 0.70 (0.51, 0.96) 0.71 (0.52, 0.97)

Modifiable lifestyle score, BMI ≥30 (n¼1,618)
Poor 188/52 ref ref ref ref
Average 1,298/299 0.90 (0.67, 1.20) 0.90 (0.67, 1.21) 0.90 (0.67, 1.21) 0.90 (0.67, 1.21)
Optimal 132/26 0.81 (0.51, 1.30) 0.83 (0.51, 1.34) 0.83 (0.52, 1.34) 0.84 (0.52, 1.36)
Continuouse — 0.90 (0.72, 1.13) 0.91 (0.72, 1.14) 0.91 (0.72, 1.14) 0.91 (0.73, 1.15)

Normal waist circumference versus central obesity
Modifiable lifestyle risk factor score, normal waist circumference (n¼1,126)
Poor 125/19 ref ref ref ref
Average 856/70 0.57 (0.35, 0.92) 0.56 (0.34, 0.92) 0.55 (0.33, 0.91) 0.55 (0.34, 0.91)
Optimal 145/11 0.54 (0.26, 1.13) 0.58 (0.27, 1.24) 0.56 (0.26, 1.21) 0.57 (0.27, 1.23)
Continuouse — 0.69 (0.47, 1.01) 0.70 (0.47, 1.05) 0.69 (0.46, 1.04) 0.69 (0.46, 1.04)

Modifiable lifestyle risk factor score, central obesity (n¼2,126)
Poor 240/60 ref ref ref ref
Average 1,688/367 0.88 (0.67, 1.16) 0.89 (0.68, 1.17) 0.89 (0.68, 1.17) 0.89 (0.68, 1.17)
Optimal 198/33 0.77 (0.50, 1.17) 0.77 (0.51, 1.20) 0.80 (0.52, 1.22) 0.80 (0.52, 1.22)
Continuouse — 0.88 (0.72, 1.08) 0.88 (0.72, 1.09) 0.89 (0.73, 1.10) 0.89 (0.73, 1.10)

Normal versus prediabetes
Modifiable lifestyle risk factor score,f normoglycemia at baseline (n¼1,836)

Poor 195/20 ref ref ref ref
Average 1,426/78 0.56 (0.34, 0.91) 0.58 (0.36, 0.96) 0.63 (0.38, 1.02) 0.65 (0.40, 1.07)
Optimal 215/10 0.48 (0.23, 1.03) 0.53 (0.24, 1.15) 0.63 (0.29, 1.37) 0.65 (0.30, 1.44)
Continuouse — 0.65 (0.44, 0.97) 0.68 (0.46, 1.02) 0.74 (0.49, 1.10) 0.76 (0.51, 1.14)

Modifiable lifestyle risk factor score,f prediabetes at baseline (n¼1,416)
Poor 170/59 ref ref ref ref
Average 1,118/359 0.91 (0.69, 1.19) 0.90 (0.69, 1.19) 0.89 (0.68, 1.18) 0.90 (0.68, 1.18)

(continued on next page)

Joseph
et
al/

A
m

J
Prev

M
ed

2017;53(5):e165
–e174

e171

N
ovem

ber
2017

Ta
bl
e
4
.S

tr
at
ifi
ed

As
so
ci
at
io
ns

of
M
od

ifi
ab

le
D
ia
be

te
s
R
is
k
Fa

ct
or
s
W
ith

In
ci
de

nt
D
ia
be

te
s
O
ve
r
8
Ye

ar
s
(c
on

tin
ue

d)

Ex
po

su
re

In
ci
de

nt
di
ab

et
es

n/
D
ia
be

te
s
ca

se
s

P
oi
ss
on

re
gr
es
si
on

m
od

el
,I
R
R
(9
5%

C
I)

M
od

el
1
a

M
od

el
2
b

M
od

el
3c

M
od

el
4
d

O
pt
im

al
1
2
8
/3

4
0
.8
0
(0
.5
2
,1

.2
1
)

0
.8
0
(0
.5
2
,1

.2
3
)

0
.8
1
(0
.5
3
,1

.2
3
)

0
.8
1
(0
.5
3
,1

.2
4
)

C
on

tin
uo

us
e

—
0
.9
0
(0
.7
3
,1

.1
0
)

0
.9
0
(0
.7
3
,1

.1
0
)

0
.9
0
(0
.7
3
,1

.1
0
)

0
.9
0
(0
.7
3
,1

.1
0
)

N
ot
e:

B
ol
df
ac
e
in
di
ca
te
s
st
at
is
tic
al

si
gn

ifi
ca
nc
e
(p
o
0
.0
5
).

a
M
od

el
1
:a

ge
,s
ex
.

b
M
od

el
2
:M

od
el

1
+
ed

uc
at
io
n,

cu
rr
en

t
oc
cu
pa

tio
n
st
at
us
,a

lc
oh

ol
us
e,

sy
st
ol
ic
bl
oo

d
pr
es
su
re
.

c
M
od

el
3
:M

od
el

2
+
B
M
I.

d
M
od

el
4
:M

od
el

2
+
w
ai
st

ci
rc
um

fe
re
nc
e.

e
C
on

tin
uo

us
pe

r
ca
te
go
ric

al
in
cr
ea

se
in

m
od

ifi
ab

le
lif
es
ty
le

ris
k
fa
ct
or

ca
te
go
ry
.

f M
od

ifi
ab

le
lif
es
ty
le
ris

k
fa
ct
or

sc
or
e:
po

or
sc
or
e
(0
–
3
),
av
er
ag

e
(4
–
7
),
op

tim
al
(8
–
1
1
).
Sm

ok
in
g:
cu
rr
en

ts
m
ok
er

(0
po

in
ts
),
fo
rm

er
≤
1
2
m
on

th
s
(1

po
in
t),

ne
ve
ro

rq
ui
t≥

1
2
m
on

th
s
(2

po
in
ts
);
TV

w
at
ch
in
g:

4
4
ho

ur
s/
da

y
(0

po
in
ts
),
1
–
4
ho

ur
s/
da

y
(1

po
in
t),

o
1
ho

ur
/d
ay

(2
po

in
ts
);
Am

er
ic
an

H
ea

rt
As
so
ci
at
io
n
ph

ys
ic
al
ac
tiv
ity
:p

oo
r(
0
po

in
ts
),
in
te
rm

ed
ia
te

(1
po

in
t),

id
ea

l(
2
po

in
ts
);
ah

a
he

al
th
y
di
et
:p

oo
r
(0

po
in
ts
),
in
te
rm

ed
ia
te

(1
po

in
t),

id
ea

l(
2
po

in
ts
);
sl
ee

p-
di
so
rd
er
ed

br
ea

th
in
g
bu

rd
en

:s
ev
er
e
(0

po
in
ts
),
m
od

er
at
e
(1

po
in
t),

m
ild

(2
po

in
ts
),
no

ne
(3

po
in
ts
).

IR
R
,i
nc
id
en

t
ra
te

ra
tio

.

Joseph et al / Am J Prev Mee172

CONCLUSIONS
The findings underscore the importance of combining
both primordial prevention and primary prevention
approaches to curb the toll of diabetes among AAs.
Lifestyle interventions to reduce obesity have focused on
individuals with high BMI, prediabetes (high-risk
approach), or both. This study suggests that a comple-
mentary approach that includes AAs at earlier stages in
the continuum of risk may improve results for diabetes
prevention. This indicates a need for a societal approach
for primordial and primary preventive interventions
targeting a combination of modifiable risk factors in
those traditionally considered to be at low risk, especially
among AAs.

d 2017;53(5):e165–e174

ACKNOWLEDGMENTS
The authors thank the other investigators, the data collection
staff, and the participants of the Jackson Heart Study for their
valuable contributions. The authors thank Susan Langan
and Dr. Michael Griswold for assistance with the data analysis.
The Jackson Heart Study is supported by contracts
HHSN268201300046C, HHSN268201300047C, HHSN268
201300048C, HHSN268201300049C, and HHSN26820
1300050C from the National Heart, Lung, and Blood Institute
and the National Institute on Minority Health and Health
Disparities. Joshua Joseph was supported by an institutional
training grant from the National Institute of Diabetes, Digestive,
and Kidney Diseases (T32 DK062707). The funding sources
had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; prepa-
ration, review, or approval of the manuscript; or decision to
submit the manuscript for publication. The views expressed in
this manuscript are those of the authors and do not necessarily
represent the views of the National Heart, Lung, and Blood
Institute, NIH, or DHHS. The authors declare that there is no
conflict of interest associated with this manuscript.

No financial disclosures were reported by the authors of this
paper. All authors fulfill the contribution requirements for
authorship credit, including, for each author listed: substantial
contributions to conception and design, acquisition of data, or
analysis and interpretation of data; drafting the article or
revising it critically for important intellectual content; and final
approval of the version to be published. JJJ is the guarantor of
this work.

SUPPLEMENTAL MATERIAL
Supplemental materials associated with this article can be
found in the online version at https://doi.org/10.1016/
j.amepre.2017.06.018.

REFERENCES
1. Geiss LS, Wang J, Cheng YJ, et al. Prevalence and incidence trends for

diagnosed diabetes among adults aged 20 to 79 years, United States,

www.ajpmonline.org

Joseph et al / Am J Prev Med 2017;53(5):e165–e174 e173

1980–2012. JAMA. 2014;312(12):1218–1226. https://doi.org/10.1001/
jama.2014.11494.

2. Golden SH, Brown A, Cauley JA, et al. Health disparities in endocrine
disorders: biological, clinical, and nonclinical factors—an Endocrine
Society scientific statement. J Clin Endocrinol Metab. 2012;97(9):
E1579–E1639. https://doi.org/10.1210/jc.2012-2043.

3. Helmrich SP, Ragland DR, Leung RW, Paffenbarger RS. Physical
activity and reduced occurrence of non-insulin-dependent diabetes
mellitus. N Engl J Med. 1991;325(3):147–152. https://doi.org/10.1056/
NEJM199107183250302.

4. Jeon CY, Lokken RP, Hu FB, van Dam RM. Physical activity of
moderate intensity and risk of type 2 diabetes: a systematic review.
Diabetes Care. 2007;30(3):744–752. https://doi.org/10.2337/dc06-1842.

5. Mikhailidis DP, Papadakis JA, Ganotakis ES. Smoking, diabetes and
hyperlipidaemia. J R Soc Health. 1998;118(2):91–93. https://doi.org/
10.1177/146642409811800209.

6. Mozaffarian D, Kamineni A, Carnethon M, Djoussé L, Mukamal KJ,
Siscovick D. Lifestyle risk factors and new-onset diabetes mellitus in
older adults: the cardiovascular health study. Arch Intern Med.
2009;169(8):798–807. https://doi.org/10.1001/archinternmed.2009.21.

7. Reis JP, Loria CM, Sorlie PD, Park Y, Hollenbeck A, Schatzkin A.
Lifestyle factors and risk for new-onset diabetes: a population-based
cohort study. Ann Intern Med. 2011;155(5):292–299. https://doi.org/
10.7326/0003-4819-155-5-201109060-00006.

8. Joseph JJ, Echouffo-Tcheugui JB, Carnethon MR, et al. The association
of ideal cardiovascular health with incident type 2 diabetes mellitus:
the Multi-Ethnic Study of Atherosclerosis. Diabetologia. 2016;59(9):
1893–1903. https://doi.org/10.1007/s00125-016-4003-7.

9. Hu FB, Manson JE, Stampfer MJ, et al. Diet, lifestyle, and the risk of
type 2 diabetes mellitus in women. N Engl J Med. 2001;345(11):
790–797. https://doi.org/10.1056/NEJMoa010492.

10. Taylor HA, Coady SA, Levy D, et al. Relationships of BMI to
cardiovascular risk factors differ by ethnicity. Obesity (Silver Spring).
2010;18(8):1638–1645. https://doi.org/10.1038/oby.2009.407.

11. Lipton RB, Uao Y, Cao G, Cooper RS, McGee D. Determinants of
incident non-insulin-dependent diabetes mellitus among blacks and
whites in a national sample. The NHANES I epidemiologic follow-up
study. Am J Epidemiol. 1993;138(10):826–839. https://doi.org/10.1093/
oxfordjournals.aje.a116786.

12. Taylor HA Jr, Wilson JG, Jones DW, et al. Toward resolution of
cardiovascular health disparities in African Americans: design and
methods of the JacksonHeart Study. EthnDis. 2005;15(4 Suppl 6): S6-4-17.

13. Carpenter MA, Crow R, Steffes M, et al. Laboratory, reading center,
and coordinating center data management methods in the Jackson
Heart Study. Am J Med Sci. 2004;328(3):131–144. https://doi.org/
10.1097/00000441-200409000-00001.

14. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF,
Turner RC. Homeostasis model assessment: insulin resistance and β-cell
function from fasting plasma glucose and insulin concentrations in man.
Diabetologia. 1985;28(7):412–419. https://doi.org/10.1007/BF00280883.

15. Bidulescu A, Liu J, Musani SK, et al. Association of adiponectin with
left ventricular mass in blacks. The Jackson Heart Study. Circ Heart
Fail. 2011;4(6):747–753. https://doi.org/10.1161/CIRCHEARTFAILURE.
110.959742.

16. Dubbert PM, Carithers T, Ainsworth BE, Taylor HA, Wilson G, Wyatt
SB. Physical activity assessment methods in the Jackson Heart Study.
Ethn Dis. 2005;15(4 suppl 6):S6–S56 61.

17. Smitherman TA, Dubbert PM, Grothe KB, et al. Validation of the
Jackson Heart Study physical activity survey in African Americans. J Phys
Act Health. 2009;6(1):S124. https://doi.org/10.1123/jpah.6.s1.s124.

18. Lloyd-Jones DM, Hong Y, Labarthe D, et al. Defining and setting
national goals for cardiovascular health promotion and disease
reduction: the American Heart Association’s strategic impact goal
through 2020 and beyond. Circulation. 2010;121(4):586–613. https:
//doi.org/10.1161/CIRCULATIONAHA.109.192703.

November 2017

19. Joseph JJ, Echouffo-Tcheugui JB, Golden SH, et al. Physical activity,
sedentary behaviors and the incidence of type 2 diabetes mellitus: the
Multi-Ethnic Study of Atherosclerosis (MESA). BMJ Open Diabetes Res
Care. 2016;4:e000185. https://doi.org/10.1136/bmjdrc-2015-000185.

20. Carithers T, Dubbert PM, Crook E, et al. Dietary assessment in African
Americans: methods used in the Jackson Heart Study. Ethn Dis.
2005;15(4 suppl 6):S6–S49-55.

21. Carithers TC, Talegawkar SA, Rowser ML, et al. Validity and
calibration of food frequency questionnaires used with African-
American adults in the Jackson Heart Study. J Am Diet Assoc.
2009;109(7):1184–1193. https://doi.org/10.1016/j.jada.2009.04.005.

22. Fülöp T, Hickson DA, Wyatt SB, et al. Sleep-disordered breathing
symptoms among African-Americans in the Jackson Heart Study. Sleep
Med. 2012;13(8):1039–1049. https://doi.org/10.1016/j.sleep.2012.06.005.

23. Netzer NC. Using the Berlin Questionnaire to identify patients at
risk for the sleep apnea syndrome. Ann Intern Med. 1999;131(7):485.
https://doi.org/10.7326/0003-4819-131-7-199910050-00002.

24. Fretts AM, Howard BV,McKnight B, et al. Life’s simple 7 and incidence of
diabetes among American Indians: the Strong Heart Family study.
Diabetes Care. 2014;37(8):2240–2245. https://doi.org/10.2337/dc13-2267.

25. American Diabetes Association. Diagnosis and classification of dia-
betes mellitus. Diabetes Care. 2010;33(suppl 1):S62–S69. https://doi.
org/10.2337/dc10-S062.

26. American Diabetes Association. Standards of medical care in diabetes
−2015.Diabetes Care. 2015;38(suppl 1):S1–S2. https://doi.org/10.2337/
dc15-S001.

27. Grundy S. Executive Summary of the Third Report of the National
Cholesterol Education Program (NCEP) Expert Panel on Detection,
Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult
Treatment Panel III). JAMA. 2001;285(19):2486–2497. https://doi.org/
10.1001/jama.285.19.2486.

28. The Diabetes Prevention Program. Reduction in the incidence of type 2
diabetes with lifestyle intervention or metformin. N Engl J Med.
2002;346(6):393–403. https://doi.org/10.1056/NEJMoa012512.

29. Kahn R, Davidson MB. The reality of type 2 diabetes prevention.
Diabetes Care. 2014;37(4):943–949. https://doi.org/10.2337/dc13-1954.

30. Samuel-Hodge CD, Johnson CM, Braxton DF, Lackey M. Effectiveness
of diabetes prevention program translations among African Ameri-
cans: African Americans and DPP translations. Obes Rev. 2014;15
(suppl 4):107–124. https://doi.org/10.1111/obr.12211.

31. Krishnan S, Rosenberg L, Palmer JR. Physical activity and television
watching in relation to risk of type 2 diabetes: the Black Women’s
Health Study. Am J Epidemiol. 2008;169(4):428–434. https://doi.org/
10.1093/aje/kwn344.

32. Nagayoshi M, Punjabi NM, Selvin E, et al. Obstructive sleep apnea and
incident type 2 diabetes. Sleep Med. 2016;25:156–161. https://doi.org/
10.1016/j.sleep.2016.05.009.

33. Esposito K, Kastorini C-M, Panagiotakos DB, Giugliano D. Prevention
of type 2 diabetes by dietary patterns: a systematic review of
prospective studies and meta-analysis. Metab Syndr Relat Disord.
2010;8(6):471–476. https://doi.org/10.1089/met.2010.0009.

34. Botros N, Concato J, Mohsenin V, Selim B, Doctor K, Yaggi HK.
Obstructive sleep apnea as a risk factor for type 2 diabetes. Am J
Med. 2009;122(12):1122–1127. https://doi.org/10.1016/j.amjmed.2009.04.026.

35. Cappuccio FP, D’Elia L, Strazzullo P, Miller MA. Quantity and quality of
sleep and incidence of type 2 diabetes: a systematic review andmeta-analysis.
Diabetes Care. 2009;33(2):414–420. https://doi.org/10.2337/dc09-1124.

36. Willi C, Bodenmann P, Ghali WA, Faris PD, Cornuz J. Active smoking and
the risk of type 2 diabetes: a systematic review and meta-analysis. JAMA.
2007;298(22):2654–2664. https://doi.org/10.1001/jama.298.22.2654.

37. Colten HR, Altevogt BM. National Academy of Medicine (U.S.),
Committee on Sleep Medicine and Research. Sleep Disorders and Sleep
Deprivation: An Unmet Public Health Problem. Washington, DC:
National Academies Press, 2006. www.ncbi.nlm.nih.gov/bookshelf/br.
fcgi?book=nap11617. Accessed October 10, 2013.

Joseph et al / Am J Prev Med 2017;53(5):e165–e174e174

38. Auchincloss AH, Diez Roux AV, Mujahid MS, Shen M, Bertoni AG,
Carnethon MR. Neighborhood resources for physical activity
and healthy foods and incidence of type 2 diabetes mellitus:
the Multi-Ethnic study of Atherosclerosis. Arch Intern Med. 2009;169
(18):1698–1704. https://doi.org/10.1001/archinternmed.2009.302.

39. Hill JO, Galloway JM, Goley A, et al. Scientific statement: socio-
ecological determinants of prediabetes and type 2 diabetes. Diabetes
Care. 2013;36(8):2430–2439. https://doi.org/10.2337/dc13-1161.

40. Christine PJ, Auchincloss AH, Bertoni AG, et al. Longitudinal
associations between neighborhood physical and social environments
and incident type 2 diabetes mellitus: The Multi-Ethnic Study of
Atherosclerosis (MESA). JAMA Intern Med. 2015;175(8):1311–1320.
https://doi.org/10.1001/jamainternmed.2015.2691.

41. Benowitz NL. Neurobiology of nicotine addiction: implications for
smoking cessation treatment. Am J Med. 2008;121(4 suppl 1):S3–S10.
https://doi.org/10.1016/j.amjmed.2008.01.015.

www.ajpmonline.org