Longitudinal Association Between Depressive Symptoms and Incident Type 2 Diabetes Mellitus in Older Adults
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
ORIGINAL INVESTIGATION
Longitudinal Association Between Depressive
Symptoms and Incident Type 2 Diabetes Mellitus
in Older Adults
The Cardiovascular Health Study
Mercedes R. Carnethon, PhD; Mary L. Biggs, MPH; Joshua I. Barzilay, MD; Nicholas L. Smith, PhD, MPH;
Viola Vaccarino, MD, PhD; Alain G. Bertoni, MD, MPH; Alice Arnold, PhD; David Siscovick, MD
Background: Prospective studies indicate that a single Results: The mean CES-D score at baseline was 4.5 (SD,
self-report of high depressive symptoms is associated 4.5). The incidence rate of diabetes was 4.4 per 1000 per-
with an increased risk of developing type 2 diabetes son-years. Following adjustment for baseline demo-
mellitus. graphic characteristics and measures of physical activ-
ity, smoking, alcohol intake, body mass index, and
Methods: We tested whether a single report of high de- C-reactive protein during follow-up, each measure of de-
pressive symptoms, an increase in depressive symp- pressive symptoms was significantly associated with in-
toms, or persistently high depressive symptoms over time cident diabetes (high baseline CES-D score: hazard ra-
were associated with the development of diabetes in adults tio, 1.6 [95% confidence interval, 1.1-2.3]; CES-D score
65 years and older. Participants from the Cardiovascu- increase: hazard ratio, 1.5 [95% confidence interval, 1.1-
lar Health Study completed the 10-item Center for Epi- 2.2]; and persistently high symptoms: hazard ratio, 1.5
demiological Studies–Depression Scale (CES-D) annu- [95% confidence interval, 1.1-2.3]).
ally from 1989 to 1999. A single report of high depressive
symptoms (CES-D score, ⱖ8), an increase in symptoms Conclusion: Older adults who reported higher depressive
during follow-up (ⱖ5 from baseline), and persistently symptoms were more likely to develop diabetes than their
high symptoms (2 consecutive scores ⱖ8) were each stud- counterparts; this association was not fully explained by risk
ied in relation to incident diabetes, defined by initiation factors for diabetes.
of diabetes control medications among participants who
were free from diabetes at baseline (n = 4681). Arch Intern Med. 2007;167:802-807
M
OST PREVIOUS STUD - mass index [BMI]) taken at a single point
ies1-8 indicate that the (typically at baseline) do not permit an ac-
relative risk of develop- curate assessment of their role in the devel-
ing type 2 diabetes opment of diabetes.
mellitus is elevated in In the present study, we tested the
persons who report high depressive symp- hypothesis that high depressive symp-
toms or clinical depression compared with toms were associated with the develop-
those with fewer symptoms or without a ment of diabetes in older adults (those
ⱖ65 years). Our approach differed from
CME course available at prior studies in 3 ways. First, using
www.archinternmed.com depressive symptoms characterized annu-
ally, we studied the relationship of a
clinical diagnosis. With the exception of 2 single high report of depressive symp-
studies3,6 that included a measure of clini- toms, an increase in depressive symptoms
cal depression, high depressive symptoms during follow-up, and persistently high
are typically defined based on a single self- symptom scores with the development of
reported survey. Given the episodic na- diabetes. Second, we tested whether any
ture of depression and depressive symp- observed association was independent of
toms, a single self-report of symptoms may known correlates of depression and dia-
not fully characterize the association be- betes by statistically adjusting for health
tween depressive symptoms and diabetes. behaviors, BMI, and inflammatory mark-
Similarly, measurements of correlates of de- ers measured repeatedly during follow-
Author Affiliations are listed at pressive symptoms and diabetes (eg, ciga- up. Third, to our knowledge, this is the
the end of this article. rette smoking, physical inactivity, and body only study of a population of older adults
(REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM
802
Downloaded from www.archinternmed.com on September 25, 2010
©2007 American Medical Association. All rights reserved.who, according to national surveys, have a high preva- increase. In accordance with previous research in the CHS,18,19 we
lence of diabetes9 and depression.10 defined an increase in depressive symptoms over time as a CES-D
score of 5 or higher from baseline, and persistent depressive symp-
toms as 2 consecutive CES-D scores of 8 or higher. Participants
METHODS using medications classified as antidepressants (selective seroto-
nin reuptake inhibitors, tricyclic agents, and monoamine oxidase
STUDY POPULATION inhibitors) were identified during the medication inventory.
The Cardiovascular Health Study (CHS) is a prospective, popu- DIABETES ASCERTAINMENT
lation-based, cohort study of cardiovascular disease in adults 65
years and older. In 1989 and 1990, 5201 men and women were We classified participants as having new-onset diabetes based on
recruited from a random sample of Medicare-eligible residents the initiation of insulin or oral hypoglycemic therapy ascertained
in 4 US communities.11 A supplemental cohort of 687 predomi- by annual medication inventory. In an effort to identify undiag-
nantly African American men and women was recruited during nosed diabetes, we created a second incident diabetes definition
1992 and 1993 from 3 of the same communities using the same thatincludedmedicationuseandelevatedfastingglucoselevel(ⱖ126
sampling and recruitment methods. Details of the study design, mg/dL [ⱖ7.0 mmol/L]), measured during 1992-1993 and 1996-
sampling, and recruitment have been published previously.11,12 1997. If a participant’s diabetes medication use is missing at a given
examination, but available at the prior examination, the partici-
DATA COLLECTION pant was assumed to have the same medication use as the previ-
ously reported year. We excluded observations from participants
The CHS participants were examined yearly from baseline through missing medication use at 2 consecutive examinations.
1999. Further details about data collection instruments and ex-
amination schedules were published.13 Data collected by stan- EXCLUSIONS
dardized interview included sociodemographic characteristics (ie,
age, race, sex, marital status, and educational attainment) and To identify a cohort of persons free from diabetes at baseline,
health behaviors (ie, physical activity, smoking, and alcohol in- we excluded 919 participants with prevalent diabetes and 104
take). Medication use was assessed at baseline and annually by a about whom we were unable to determine diabetes status. Also,
medication inventory.14 Clinical measures included height (mea- 8 participants who were missing baseline CES-D scores and 176
sured in 1989-1990, 1992-1993, and 1996-1997), weight (mea- who did not participate in the second and third clinic visits were
sured annually), and waist circumference (measured in 1989- excluded, leaving 4681 participants for this analysis.
1990, 1992-1993, 1996-1997, and 1998-1999). Body mass index
was calculated as weight in kilograms divided by the height in STATISTICAL ANALYSIS
meters squared.2 Venipuncture was conducted following an over-
night fast. Plasma and serum samples were frozen at −70°C and Baseline characteristics were calculated for all CHS participants and
shipped to the CHS Central Laboratory (University of Vermont, stratified by CES-D score. Kaplan-Meier curves were used to de-
Burlington) for analysis. Serum glucose was assayed according scribe the association between quartiles of baseline CES-D score
to standard methods.15 C-reactive protein level was measured using and incident diabetes. No effect modification was detected between
an ultrasensitive enzyme-linked immunosorbent assay.15 Fast- depressive symptoms and selected covariates of interest (ie, sex,
ing serum glucose level was measured during the annual exami- race, educational age, and marital status), so we built a series of
nations in 1989-1990, 1992-1993, and 1996-1997. multivariate Cox proportional hazards models to calculate hazard
ratios (HRs) between baseline CES-D scores and incident diabe-
MEASUREMENT OF DEPRESSIVE SYMPTOMS tes. The time to event was calculated as the interval between en-
rollment date and the earliest of the following: (1) date of clinic
Depressive symptoms were evaluated annually using the 10- examination when new diabetes was ascertained, (2) date of the
item version of the Center for Epidemiological Studies– last clinic visit during which diabetes status could be ascertained
Depression Scale (CES-D).16 The CES-D is a self-reported mea- (a description of missing data criteria is given in the “Diabetes As-
sure of depressive symptoms experienced during the previous certainment” subsection of this section), (3) date of death, or (4)
week. Previous research17 has compared the validity of the 10- date of last follow-up. Logistic regression analysis was used to cal-
with the 20-item CES-D and found that the 10-item version culateoddsratiosfortheassociationofincreasesindepressivesymp-
shows good predictive accuracy ( = 0.97, P⬍.001) in an el- toms or persistently high symptoms with incident diabetes. We
derly population and repeatability comparable to other sur- included covariates from multiple follow-up examinations and im-
veys (r=0.71). Questions focus on mood (5 items), irritability puted values for missing data using methods for longitudinal data
(1 item), calories (energy) (2 items), concentration (1 item), previously validated in the CHS cohort.20 The proportion of val-
and sleep (1 item). Items are coded on a scale of 0 (rarely or ues imputed ranged between 6% and 17% for data that were col-
none of the time [⬍1 day]) to 3 (most or all of the time [5-7 lected yearly (CES-D scores, weight, alcohol intake, and smoking
days]) points, for a maximum of 30 points. status), and was 71% for physical activity, which was measured
The CES-D is not a diagnostic instrument for clinical depres- only during 1989-1990, 1992-1993, and 1996-1997. Statistical sig-
sion. Higher scores on the CES-D indicate a higher burden of de- nificance was set at P⬍.05. All analyses were conducted using com-
pressive symptoms. Scores above both 8 and 10 on the 10-item scale puter software (Stata, version 9.0; Stata Corp, College Station, Tex).
have been used to indicate high depressive symptoms.17 We ap-
plied a cut point of CES-D scores of 8 and higher based on reports RESULTS
from the baseline clinic examination to indicate high depressive
symptoms, hereafter referred to as “depressive symptoms.” We at-
tempted to verify our findings using a CES-D cut point of 10 or Baseline demographic and clinical characteristics in this
higher. To evaluate dose-response for incident diabetes, baseline sample of participants are reported in Table 1. Notably,
CES-D scores were categorized into quartiles (ⱕ1, 2-3, 4-6, and the proportion of participants who were overweight or obese
ⱖ7) and continuous scores were studied per standard deviation was similar across depressive symptom categories. Figure 1
(REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM
803
Downloaded from www.archinternmed.com on September 25, 2010
©2007 American Medical Association. All rights reserved.Table 1. Baseline Characteristics of the Population*
Depressive Symptom Score (CES-D Score)
Characteristic Total Sample High (ⱖ8) Low (⬍8) P Value
Age, y† 72.7 (5.5) 73.0 (5.7) 72.6 (5.5) .07
African American race 13.0 18.2 11.6 ⬍.001
Female sex 59.2 70.1 56.4 ⬍.001
⬍High school education 27.0 34.5 25.1 ⬍.001
Marital status
Single 8.7 10.0 8.4 ⬍.001
Married 67.5 58.6 69.7
Widowed 23.8 31.4 21.9
Life events score† 1.1 (1.2) 1.5 (1.4) 1.0 (1.1) ⬍.001
Physical activity, total kcal/wk† 1783.7 (2038.9) 1443.7 (1803.9) 1868.3 (2085.0) ⬍.001
Smoking status‡
Current 12.1 14.9 11.4 .002
Former 41.3 37.4 42.2
Never 46.7 47.8 46.4
Total alcohol intake per week, No. of beverages† 2.7 (7.3) 1.8 (4.9) 2.9 (7.8) ⬍.001
Antidepressant medication use 3.6 7.3 2.7 ⬍.001
BMI† 26.3 (4.5) 26.4 (2.9) 26.3 (4.4) .43
Weight status
Normal (BMI, ⱕ24.9) 41.4 41.9 41.2 .15
Overweight (BMI, 25.0-29.9) 41.6 39.3 42.2
Obese (BMI, ⱖ30.0) 17.0 18.8 16.6
Waist circumference, cm† 93.3 (12.8) 93.4 (13.8) 93.3 (13.8) .84
Log of C-reactive protein level, mg/L† 0.87 (1.00) 0.98 (1.06) 0.84 (1.00) ⬍.001
Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by the height in meters squared); CES-D, Center for Epidemiological
Studies–Depression Scale.
*Data are given as percentage of each group unless otherwise indicated.
†Data are given as mean (SD) unless otherwise indicated.
‡Percentages do not total 100 because of rounding.
800 0.06
Cumulative Incidence of Diabetes Mellitus
Quartile 1
Quartile 2
Quartile 3
600 Quartile 4
0.04
Frequency
400
0.02
200
0.00
0 2 4 6 8
Follow-up, y
0 8 10 20 30
CES-D Score
Figure 2. Kaplan-Meier plot of the association between quartiles of
Figure 1. Distribution of depressive symptom scores at baseline. The depressive symptom score and incidence of diabetes mellitus. Quartile 1
median score was 3 (interquartile range, 1-7); the mean score was 4.5 indicates a Center for Epidemiological Studies–Depression Scale (CES-D)
(SD, 4.5). CES-D indicates Center for Epidemiological Studies–Depression score of 0 to 1; 2, a CES-D score of greater than 1 to 3; 3, a CES-D score of
Scale. The vertical line at CES-D score of 8 signifies the reference marker. greater than 3 to 7; and 4, a CES-D score of greater than 7.
displays the distribution of depressive symptom scores at develop diabetes than those in the lowest quartile; there
baseline: 20.0% of participants had CES-D scores of 8 or was no association between each of the middle quartiles
higher and 13.2% had scores of 10 or higher. During follow- of depressive symptoms and the lowest quartile. Higher de-
up, the CES-D scores increased by at least 5 U in 47.2% of pressive symptoms at baseline were associated with inci-
participants; 37.7% had 2 consecutive CES-D scores of 8 dent diabetes (Table 2), and this association persisted with
or higher. There was little evidence of dose response in the statistical adjustment for baseline demographic character-
association between baseline CES-D score and incident dia- istics and correlates of depression and diabetes measured
betes, as displayed using Kaplan-Meier curves (Figure 2). during follow-up. The magnitude of association in fully ad-
Participants in the uppermost quartile were more likely to justed models was identical when CES-D scores of greater
(REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM
804
Downloaded from www.archinternmed.com on September 25, 2010
©2007 American Medical Association. All rights reserved.Table 2. Association Between Depressive Symptoms at Baseline and the Development of Diabetes Mellitus
CES-D Score
Variable ⬍8 ⱖ8 Per 4.5 Points Higher
No. of patients 3745 936 4681
No. of events 108 39 147
Rate per 1000 person-years 3.9 6.2 4.4
Model*
1 1 [Reference] 1.59 (1.10-2.30) 1.18 (1.02-1.37)
2 1 [Reference] 1.63 (1.12-2.36) 1.19 (1.03-1.39)
3† 1 [Reference] 1.56 (1.07-2.28) 1.16 (0.99-1.35)
4 1 [Reference] 1.57 (1.07-2.29) 1.17 (1.00-1.36)
Abbreviation: See Table 1.
*Data are given as hazard ratio (95% confidence interval). The models are as follows: 1, the crude model; 2, adjusted for age, race, and sex; 3, adjusted for the
variables in model 2 plus educational attainment, marital status, physical activity, smoking, alcohol intake, and body mass index; 4, adjusted for the variables in
model 3 plus C-reactive protein level.
†Statistical adjustment includes covariates from multiple follow-up examinations.
than 10 were used (HR, 1.57; 95% confidence interval, 1.01-
2.42). There was no association between antidepressant use Table 3. Association Between Depressive Symptom Scores
During Follow-up and the Development of Diabetes Mellitus
at baseline and incident diabetes (HR, 1.20; 95% confi-
dence interval, 0.62-2.34).
Increase in CES-D Score 2 Consecutive CES-D
Participants whose depressive symptom scores increased Model* by ⱖ5 Points† Scores ⱖ8†
by at least 5 during follow-up were significantly more likely
1 1.57 (1.13-2.20) 1.83 (1.31-2.56)
to develop diabetes than participants whose scores did not 2 1.70 (1.19-2.43) 1.91 (1.35-2.71)
increase (Table 3). Similarly, 2 consecutive CES-D scores 3 1.59 (1.10-2.30) 1.64 (1.14-2.36)
of 8 or higher were associated with a greater likelihood of 4 1.54 (1.06-2.24) 1.54 (1.06-2.26)
developing diabetes. Statistical adjustment for confounders 5 1.63 (1.12-2.37) 1.41 (0.91-2.19)
attenuated the strength of these associations, but they re-
tained statistical significance. Adjusting for baseline CES-D Abbreviation: See Table 1.
*The models are as follows: 1, the crude model; 2, adjusted for age, race,
score (⬍8 or ⱖ8) did not attenuate the association between and sex; 3, adjusted for the variables in model 2 plus educational attainment,
an increase in CES-D scores and incident diabetes; however, marital status, physical activity, smoking, alcohol intake, and body mass
adjustmentforbaselinedidattenuatetheassociationbetween index; 4, adjusted for the variables in model 3 plus C-reactive protein level;
2 consecutive high scores and incident diabetes to marginal and 5, adjusted for the variables in model 4 plus baseline CES-D score
(⬍8 or ⱖ8).
statistical significance. When we stratified the analyses by †Data are given as odds ratio (95% confidence interval).
categories of baseline CES-D score, an increase in CES-D
scores was only statistically significant in those with low
CES-D scores (⬍8) at baseline (HR, 1.73; 95% confidence sociation between increasing scores and incident diabetes
interval, 1.12-2.67); by contrast, the association was not sig- was strongest among those with initially low baseline scores
nificant in those with baseline CES-D scores of 8 or higher (CES-D score, ⬍8). These findings were present across
(HR, 1.26; 95% confidence interval, 0.54-2.94). Two con- demographic strata and persisted with statistical adjust-
secutive high CES-D scores were not associated with inci- ment for known correlates of depression and diabetes, such
dent diabetes in either stratum of baseline CES-D scores. as BMI, physical activity, cigarette smoking, alcohol in-
Repeating all analyses using a definition of incident dia- take, and C-reactive protein level.
betes that included elevated measured fasting glucose level Because inflammatory markers are associated with the
in addition to medications yielded generally similar results. development of diabetes21,22 and with depressive symp-
During follow-up, 234 participants (7.9/1000 person-years) toms,23-26 inflammation is often proposed as a mechanism
developed diabetes; rates were higher among persons with in the association between depressive symptoms and in-
CES-D scores of 8 or more vs less than 8 (9.4 vs 7.5/1000 cident diabetes. However, our findings demonstrating no
person-years). The direction of multivariate HRs was simi- attenuation of the association following adjustment for
lar,butthefindingswerenotstatisticallysignificant(P⬎.05). C-reactive protein level suggest that other biological mecha-
nisms previously proposed, such as hypothalamic-pituitary-
COMMENT adrenal axis dysregulation and sympathetic nervous sys-
tem stimulation,2,7,27 may be more salient. Depression is
In this sample of older adults, a single report of high de- associated with adverse autonomic nervous system func-
pressive symptoms, an increase in symptoms with time, and tioning,12 and prospective studies28,29 demonstrate that au-
persistently high symptoms over time are each associated tonomic nervous system dysfunction can be detected prior
with an excess incidence of diabetes. Furthermore, increas- to the development of diabetes. Under conditions of short-
ing symptoms with time are associated with incident dia- and long-term sympathetic activation, pancreatic -cell
betes beyond initial high depressive symptoms and the as- functioning is suppressed and insulin secretion declines.
(REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM
805
Downloaded from www.archinternmed.com on September 25, 2010
©2007 American Medical Association. All rights reserved.At the same time, sympathetic activation causes declines be interpreted in light of some limitations. Incident diabe-
in muscle insulin sensitivity, hepatic glucogenesis, and rate tes in the CHS has commonly been identified as the initia-
of glucose uptake, all of which could result in the devel- tion of hypoglycemic agents or insulin therapy because the
opment of a clinically elevated glucose level.30,31 Future medication inventory is conducted annually, whereas fast-
mechanistic studies of the association between depressive ing glucose level was measured at a few follow-up visits.11
symptoms and diabetes incidence should investigate the An important consequence of this ascertainment strategy
role of the autonomic nervous system in depth. is that some cases of diabetes are misclassified. Diabetes is
Our primary findings have been described in most,1-5 typically present 4 to 7 years before clinical diagnosis,35 and
but not all,8 prior studies. However, even those studies6,8 approximately one third of diabetes is undiagnosed.9
reporting a statistically null association between depres- It is difficult to determine the impact of this misclassi-
sive symptoms and incident diabetes trend toward a posi- fication on our findings, because it is not clear whether de-
tive association. Often, the association became nonsignifi- pressive symptoms predispose persons to less or more fre-
cant following adjustment for factors such as overweight quent health care visits in order to be diagnosed. In our
or physical inactivity, which may fall along the causal path- secondary analysis, we defined incident diabetes with el-
way toward diabetes development.8 Depressive symp- evated fasting glucose level in addition to medications, which
toms are associated with risk factors for diabetes, such as resulted in a slightly weaker association. Thus, it is plau-
physical inactivity or excess calorie intake, leading to in- sible that persons with higher depressive symptoms were
creased BMI or disturbed sleeping patterns.32-34 We still ob- less likely to be under the regular care of a physician and,
served an association following adjustment for measures therefore, less likely to be diagnosed and treated with medi-
of BMI and physical activity collected during study follow- cations. If so, our primary results may have overestimated
up, suggesting that other mechanisms associated with de- the strength of the association, despite effect sizes repre-
pressive symptoms contribute to diabetes risk. However, senting an approximately 50% higher risk of diabetes among
our measure of physical activity was not available at all time persons with higher depressive symptoms, which is con-
points during follow-up, so we relied on imputation for our sistent with prior studies.1-5,27
longitudinal analyses, which may have introduced error. In summary, high depressive symptoms may be related
Population-based studies such as this one commonly rely to the development of diabetes in older adults, and this as-
on self-reported symptom surveys as opposed to diagnos- sociation may not be attributable solely to the adoption of
tic interviews. An important consequence of the differen- adverse health behaviors or weight gain. The pathophysi-
tial assessment of depression across studies is the wide vari- ologic mechanism for this association remains unclear. Our
ability in the presence and strength of association between findings in this population of older adults are of particular
depressive symptoms and incident diabetes. While we are public health importance because there are 35 million US
unable to identify clinical depression using the 10-item CES- adults older than 65 years.36 An estimated 2 million older
D, we do report consistent findings using multiple classi- adults may have a depressive illness,10 which is the second
fications taken from a range of time periods. In addition, we highest prevalence of depression across the age range.37 The
observedthatconsistentlyhighdepressivesymptomsareonly highest prevalence of diabetes in the population is among
marginallyandnonsignificantlyassociatedwithincidentdia- those 65 years or older (15.3%),38 and nearly 39% of diabe-
betes after adjustment for baseline CES-D category. It is pos- tes cases in older adults were diagnosed after the age of 65
sible that increases in depressive symptoms have the great- years. Thus, findings from this study of a novel and highly
est influence on the incidence of diabetes, as evidenced by prevalent risk factor for diabetes have important implica-
the statistically significant association between increasing tions for a substantial subset of our population.
symptoms during follow-up and incident diabetes, even af-
ter adjustment for baseline symptom score. Further support Accepted for Publication: December 21, 2006.
in favor of this hypothesis was found in our observation that Author Affiliations: Department of Preventive Medicine,
increasing scores were most strongly associated with inci- Feinberg School of Medicine, Northwestern University, Chi-
dent diabetes among those with initially low (CES-D score, cago, Ill (Dr Carnethon); Departments of Biostatistics
⬍8) symptom scores at baseline. (Ms Biggs and Dr Arnold) and Epidemiology (Dr Smith),
Consistent with previous research studies,2,4,7 we do not School of Public Health and Community Medicine, Car-
find a graded association between depressive symptoms and diovascular Health Research Unit (Drs Smith and Siscovick)
diabetes incidence. Rather, there seems to be a threshold and Collaborative Studies Coordinating Center
of symptoms associated with diabetes risk. Our finding that (Dr Siscovick), University of Washington, Seattle; Kaiser
antidepressant medication use was not associated with the Permanente of Georgia (Dr Barzilay), and Divisions of En-
development of diabetes was unexpected because this as- docrinology (Dr Barzilay) and Cardiology (Dr Vaccarino),
sociation has been described in another study.3 However, Department of Medicine, Emory University School of Medi-
few older adults in this cohort reported using antidepres- cine, Atlanta; and Departments of Epidemiology and In-
sant medications; thus, we may have been underpowered ternal Medicine, Wake Forest University School of Medi-
to detect an association. Alternatively, antidepressant medi- cine, Winston-Salem, NC (Dr Bertoni).
cations may have successfully controlled the somatic and Correspondence: Mercedes R. Carnethon, PhD, Depart-
behavioral symptoms of depression. ment of Preventive Medicine, Feinberg School of Medi-
Despite the novel strengths of our study, which include cine, Northwestern University, 680 N Lake Shore Dr, Suite
investigation in a cohort of older adults, multiple measure- 1102, Chicago, IL 60611 (carnethon@northwestern.edu).
ments of depressive symptoms during follow-up, and the Author Contributions: Dr Carnethon had full access to all
abilitytoadjustfortime-varyingcovariates,ourfindingsmust the data in the study and takes responsibility for the integ-
(REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM
806
Downloaded from www.archinternmed.com on September 25, 2010
©2007 American Medical Association. All rights reserved.rity of the data and the accuracy of the data analysis. Study 11. Fried LP, Borhani NO, Enright P, et al. The Cardiovascular Health Study: design
and rationale. Ann Epidemiol. 1991;1:263-276.
concept and design: Carnethon and Siscovick. Acquisition of 12. Carney RM, Blumenthal JA, Stein PK, et al. Depression, heart rate variability, and
data: Arnold and Siscovick. Analysis and interpretation of acute myocardial infarction. Circulation. 2001;104:2024-2028.
data: Carnethon, Biggs, Barzilay, Smith, Vaccarino, Bertoni, 13. Ives DG, Fitzpatrick AL, Bild DE, et al. Surveillance and ascertainment of cardiovas-
and Siscovick. Drafting of the manuscript: Carnethon. Criti- cular events: the Cardiovascular Health Study. Ann Epidemiol. 1995;5:278-285.
cal revision of the manuscript for important intellectual con- 14. Psaty BM, Lee M, Savage PJ, Rutan GH, German PS, Lyles M; Cardiovascular
Health Study Collaborative Research Group. Assessing the use of medications
tent: Carnethon, Biggs, Barzilay, Smith, Vaccarino, Bertoni, in the elderly: methods and initial experience in the Cardiovascular Health Study.
Arnold, and Siscovick. Statistical analysis: Carnethon, Biggs, J Clin Epidemiol. 1992;45:683-692.
Arnold, and Siscovick. Obtained funding: Siscovick. Admin- 15. Cushman M, Cornell ES, Howard PR, Bovill EG, Tracy RP. Laboratory methods
istrative, technical, and material support: Barzilay and Arnold. and quality assurance in the Cardiovascular Health Study. Clin Chem. 1995;
41:264-270.
Study supervision: Barzilay and Vaccarino. 16. Radloff LS. The CES-D scale: a self-report depression scale for research in the
Financial Disclosure: None reported. general population. Appl Psychol Meas. 1977;1:385-401.
Funding/Support: This study was supported by con- 17. Andresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in
tracts N01-HC-85079 through N01-HC-85086, N01-HC- well older adults: evaluation of a short form of the CES-D (Center for Epidemio-
35129, N01-HC-15103, N01-HC-55222, and U01-HL- logic Studies Depression Scale). Am J Prev Med. 1994;10:77-84.
18. Steffens DC, Helms MJ, Krishnan KR, Burke GL. Cerebrovascular disease and
080295 from the National Heart, Lung, and Blood depression symptoms in the Cardiovascular Health Study. Stroke. 1999;30:
Institute, National Institutes of Health, with additional 2159-2166.
contribution from the National Institute of Neurologi- 19. Steffens DC, Krishnan KRR, Crump C, Burke GL. Cerebrovascular disease and
cal Disorders and Stroke; in part by career development evolution of depressive symptoms in the Cardiovascular Health Study. Stroke.
2002;33:1636-1644.
award 5-K01-HL-73249-03 from the National Heart, Lung, 20. Engels JM, Diehr P. Imputation of missing longitudinal data: a comparison of
and Blood Institute (Dr Carnethon); and by an unre- methods. J Clin Epidemiol. 2003;56:968-976.
stricted grant from Amgen Inc (for data analysis). 21. Barzilay JI, Abraham L, Heckbert SR, et al. The relation of markers of inflamma-
Role of the Sponsor: The CHS publications and presen- tion to the development of glucose disorders in the elderly: the Cardiovascular
tations committee and the National Heart, Lung, and Health Study. Diabetes. 2001;50:2384-2389.
22. Schmidt MI, Duncan BB, Sharrett AR, et al. Markers of inflammation and pre-
Blood Institute reviewed and approved this manuscript. diction of diabetes mellitus in adults (Atherosclerosis Risk in Communities study):
The funding bodies had no role in data extraction and a cohort study. Lancet. 1999;353:1649-1652.
analyses, in the writing of the manuscript, or in the de- 23. Toker S, Shirom A, Shapira I, Berliner S, Melamed S. The association between burn-
cision to submit the manuscript for publication. out, depression, anxiety, and inflammation biomarkers: C-reactive protein and fibrino-
gen in men and women. J Occup Health Psychol. 2005;10:344-362.
Previous Presentation: This study was presented at the 24. Douglas KM, Taylor AJ, O’Malley PG. Relationship between depression and C-reactive
46th Annual Meeting of the American Heart Associa- protein in a screening population. Psychosom Med. 2004;66:679-683.
tion Council on Epidemiology and Prevention; March 3, 25. Miller HW. Plan and operation of the health and nutrition examination survey:
2006; Phoenix, Ariz. United States–1971-1973. Vital Health Stat 1. 1973;(10a):1-46.
Additional Information: A full list of participating CHS 26. Raison CL, Capuron L, Miller AH. Cytokines sing the blues: inflammation and
the pathogenesis of depression. Trends Immunol. 2006;27:24-31.
investigators and institutions can be found at http://www 27. Everson-Rose SA, Meyer PM, Powell LH, et al. Depressive symptoms, insulin
.chs-nhlbi.org. resistance, and risk of diabetes in women at midlife. Diabetes Care. 2004;27:
2856-2862.
28. Carnethon MR, Jacobs DR Jr, Sidney S, Liu K. Influence of autonomic nervous
REFERENCES system dysfunction on the development of type 2 diabetes: the CARDIA study.
Diabetes Care. 2003;26:3035-3041.
1. Arroyo C, Hu FB, Ryan LM, et al. Depressive symptoms and risk of type 2 dia- 29. Carnethon MR, Golden SH, Folsom AR, Haskell W, Liao D. Prospective investi-
betes in women. Diabetes Care. 2004;27:129-133. gation of autonomic nervous system function and the development of type 2 dia-
2. Carnethon MR, Kinder LS, Fair JM, Stafford RS, Fortmann SP. Symptoms of de- betes: the Atherosclerosis Risk in Communities Study, 1987-1998. Circulation.
pression as a risk factor for incident diabetes: findings from the National Health 2003;107:2190-2195.
and Nutrition Examination Epidemiologic Follow-up Study, 1971-1992. Am J 30. Lautt WW, Macedo MP, Sadri P, Takayama S, Duarte Ramos F, Legare DJ.
Epidemiol. 2003;158:416-423. Hepatic parasympathetic (HISS) control of insulin sensitivity determined by feed-
3. Brown LC, Majumdar SR, Newman SC, Johnson JA. History of depression increases ing and fasting. Am J Physiol Gastrointest Liver Physiol. 2001;281:G29-G36.
risk of type 2 diabetes in younger adults. Diabetes Care. 2005;28:1063-1067. 31. Moore MC, Satake S, Baranowski B, Hsieh PS, Neal DW, Cherrington AD. Effect
4. Kawakami N, Takatsuka N, Shimizu H, Ishibashi H. Depressive symptoms and of hepatic denervation on peripheral insulin sensitivity in conscious dogs. Am J
occurrence of type 2 diabetes among Japanese men. Diabetes Care. 1999; Physiol Endocrinol Metab. 2002;282:E286-E296.
22:1071-1076. 32. Ayas NT, White DP, Al-Delaimy WK, et al. A prospective study of self-reported sleep
5. Palinkas LA, Lee PP, Barrett-Connor E. A prospective study of type 2 diabetes duration and incident diabetes in women. Diabetes Care. 2003;26:380-384.
and depressive symptoms in the elderly: the Rancho Bernardo Study. Diabet Med. 33. Mallon L, Broman J-E, Hetta J. High incidence of diabetes in men with sleep com-
2004;21:1185-1191. plaints or short sleep duration: a 12-year follow-up study of a middle-aged
6. Eaton WW, Armenian H, Gallo J, Pratt L, Ford DE. Depression and risk for onset population. Diabetes Care. 2005;28:2762-2767.
of type II diabetes: a prospective population-based study. Diabetes Care. 1996; 34. Nilsson PM, Roost M, Engstrom G, Hedblad B, Berglund G. Incidence of diabe-
19:1097-1102. tes in middle-aged men is related to sleep disturbances. Diabetes Care. 2004;
7. Golden SH, Williams JE, Ford DE, et al. Depressive symptoms and the risk of 27:2464-2469.
type 2 diabetes: the Atherosclerosis Risk in Communities study. Diabetes Care. 35. Harris MI, Klein R, Welborn TA, Knuiman MW. Onset of NIDDM occurs at least
2004;27:429-435. 4-7 years before clinical diagnosis. Diabetes Care. 1992;15:815-819.
8. Saydah SH, Brancati FL, Golden SH, Fradkin J, Harris MI. Depressive symptoms 36. He W, Sengupta M, Velkoff VA, DeBarros KA. 65⫹ in the United States: 2005.
and the risk of type 2 diabetes mellitus in a US sample. Diabetes Metab Res Rev. In: Current Population Reports. Washington, DC: US Government Printing Of-
2003;19:202-208. fice; 2005:1. US Census Bureau report P23-209.
9. Cowie CC, Rust KF, Byrd-Holt DD, et al. Prevalence of diabetes and impaired fast- 37. Alexopoulos G. Mood disorders: psychiatric disorders in late life. In: Sadock BJ,
ing glucose in adults in the US population: National Health and Nutrition Exami- Sadock VA, eds. Comprehensive Textbook of Psychiatry. Vol 2. 7th ed. Balti-
nation Survey 1999-2002. Diabetes Care. 2006;29:1263-1268. more, Md: Lippincott Williams & Wilkins; 2000:3060-3068.
10. National Institute of Mental Health. Older Adults: Depression and Suicide Facts. 38. Selvin E, Coresh J, Brancati FL. The burden and treatment of diabetes in elderly
Bethesda, Md: National Institutes of Health; 2003. Publication 03-4593. individuals in the US. Diabetes Care. 2006;29:2415-2419.
(REPRINTED) ARCH INTERN MED/ VOL 167, APR 23, 2007 WWW.ARCHINTERNMED.COM
807
Downloaded from www.archinternmed.com on September 25, 2010
©2007 American Medical Association. All rights reserved.You can also read
