Integrated Management of Type 2

Diabetes Mellitus and Depression Treat-

ment to Improve Medication Adherence:
A Randomized Controlled Trial
Hillary R. Bogner, MD, MSCE1,2 ABSTRACT
Knashawn H. Morales, ScD2 PURPOSE Depression commonly accompanies diabetes, resulting in reduced
Heather F. de Vries, MSPH1,2 adherence to medications and increased risk for morbidity and mortality.
The objective of this study was to examine whether a simple, brief integrated
Anne R. Cappola, MD, ScM3 approach to depression and type 2 diabetes mellitus (type 2 diabetes) treatment
1
Department of Family Medicine and Com- improved adherence to oral hypoglycemic agents and antidepressant medica-
munity Health, University of Pennsylvania, tions, glycemic control, and depression among primary care patients.
Philadelphia, Pennsylvania
METHODS We undertook a randomized controlled trial conducted from April
2
Center for Clinical Epidemiology and 2010 through April 2011 of 180 patients prescribed pharmacotherapy for type 2
Biostatistics, University of Pennsylvania, diabetes and depression in primary care. Patients were randomly assigned to an
Philadelphia, Pennsylvania
integrated care intervention or usual care. Integrated care managers collaborated
3
Division of Endocrinology, Diabetes, and with physicians to offer education and guideline-based treatment recommenda-
Metabolism, Department of Medicine, tions and to monitor adherence and clinical status. Adherence was assessed using
University of Pennsylvania, Philadelphia, the Medication Event Monitoring System (MEMS). We used glycated hemoglobin
Pennsylvania (HbA1c) assays to measure glycemic control and the 9-item Patient Health Ques-
tionnaire (PHQ-9) to assess depression.

RESULTS Intervention and usual care groups did not differ statistically on base-
line measures. Patients who received the intervention were more likely to achieve
HbA1c levels of less than 7% (intervention 60.9% vs usual care 35.7%; P <.001)
and remission of depression (PHQ-9 score of less than 5: intervention 58.7% vs
usual care 30.7%; P <.001) in comparison with patients in the usual care group
at 12 weeks.

CONCLUSIONS A randomized controlled trial of a simple, brief intervention inte-

grating treatment of type 2 diabetes and depression was successful in improving
outcomes in primary care. An integrated approach to depression and type 2
Annals Journal Club selection; diabetes treatment may facilitate its deployment in real-world practices with com-
see inside back cover or http://www. peting demands for limited resources.
annfammed.org/AJC/.
Ann Fam Med 2012;10:15-22. doi:10.1370/afm.1344.

Conflicts of interest: authors report none.

INTRODUCTION

A
bidirectional association has been found between depression and
CORRESPONDING AUTHOR
diabetes mellitus.1 Depression is a risk factor for diabetes,2 and
Hillary R. Bogner, MD, MSCE
Department of Family Medicine and diabetes increases risk for the onset of depression.3 Not only is
Community Health depression common in patients with diabetes, it also contributes to poor
Center for Clinical Epidemiology and adherence to medication and dietary regimens, physical inactivity, poor
Biostatistics glycemic control, reduced quality of life, disability, and increased health
Perelman School of Medicine care expenditures.4-9
The University of Pennsylvania
9 Blockley Hall, 423 Guardian Dr The purpose of this study was to carry out a randomized controlled
Philadelphia, PA 19104 trial to test the effectiveness of integrated care management of type 2 dia-
hillary.bogner@uphs.upenn.edu betes mellitus (type 2 diabetes) and depression in comparison with usual

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

15
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

care services in primary care. Several studies have cated hemoglobin (HbA1c) levels, (3) a greater propor-
shown that a variety of primary care interventions can tion of patients who had 80% or greater adherence
improve diabetes10 and depression outcomes.11 Few of to an antidepressant, and (4) a greater proportion of
these interventions are being implemented in practice, patients who had 80% or greater adherence to an oral
however.12,13 Integrated care is needed to enhance qual- hypoglycemic agent.
ity of care, quality of life, consumer satisfaction, and
system efficiency for patients with complex, long-term
problems cutting across multiple services, clinicians, METHODS
and settings.14 Recruitment Procedures
We chose an adherence-based approach because, Patients were recruited from 3 primary care practices in
although efficacious pharmacotherapy for many Philadelphia, Pennsylvania. The protocol was approved
chronic medical conditions exists, many patients by the University of Pennsylvania Institutional Review
are not adherent to treatment and therefore are at Board. From April 2010 to April 2011, patients who had
increased risk for a variety of complications.15 Poor a diagnosis of type 2 diabetes mellitus, a prescription
adherence to treatment remains a major impediment for an oral hypoglycemic agent within the past year,
to improving care, particularly among patients with and a prescription for an oral antidepressant within
comorbid diabetes and depression.16 Compared with the past year were identified by means of an electronic
patients who are not depressed, depressed patients health record. Identified patients with an upcoming
who have diabetes are more likely to be nonadher- appointment were approached for further screen-
ent to medication regimens17,18 and exhibit worsening ing. The inclusion criteria were (1) aged 30 years and
diabetes management.19,20 The management of comor- older, (2) a diagnosis of type 2 diabetes and a current
bid depression and diabetes should be integrated and prescription for an oral hypoglycemic agent, and (3) a
tailored for preference, tolerance, and simplicity to current prescription for an antidepressant. We chose to
enhance adherence to prescribed medical regimens.4 include patients with a range of depressive symptoms
A review of the literature found only 2 randomized reflecting the relapsing, remitting nature of depression
controlled trials integrating care for the management in primary care.23 The age cutoff was chosen because
of depression with diabetes.21 Katon and colleagues at of its importance in the detection, screening, and inter-
Group Health Cooperative, a nonprofit health mainte- vention for diabetic patients.24 Exclusion criteria were
nance organization in Seattle, Washington, tested an (1) inability to give informed consent, (2) cognitive
intensive intervention for adults with major depression impairment at baseline (Mini-Mental State Examina-
and poorly controlled diabetes and/or coronary heart tion [MMSE] less than 21),25 (3) residence in a care
disease carried out by an advanced-practice nurse; they facility that provides medications on schedule, and (4)
found the intervention significantly improved control unwillingness or inability to use the Medication Event
of medical disease and depression.22 Partners Health- Monitoring System (MEMS), a system in which micro-
care, a nonprofit integrated health care system in Bos- electronic monitors on pill bottles provide the precise
ton, Massachusetts, is testing an intensive intervention date and time of container opening.
for adults with poorly controlled diabetes and major
depression or dysthymia carried out by a master’s Study Design
trained therapist, but the data have yet to be published. This trial consisted of 2 phases: a run-in phase and a
In contrast to the Group Health Cooperative and Part- randomized controlled trial phase. The purpose of the
ners Healthcare studies, our study was conducted in 2-week run-in phase was to collect preintervention
community-based primary care practices and assessed adherence rates for all patients. During this phase data
a brief, simple intervention with a focus on improving were also collected on demographic characteristics,
adherence that was specifically developed for patients blood pressure, low-density lipoprotein (LDL) choles-
with type 2 diabetes mellitus. Our study involves an terol levels, body mass index (BMI), depressive symp-
interventionist who acts as an intermediary or liaison toms, and HbA1c levels. No intervention was performed
between the depressed patient with type 2 diabetes during this phase. Following completion of the 2-week
and the physician in promoting type 2 diabetes and run-in phase, patients entering phase 2 of the study
depression treatment and patient adherence. We were randomized within each practice by flip of a coin
hypothesized that in a sample of primary care patients to either the integrated care intervention or usual care.
with depression and type 2 diabetes, patients who were Physicians were told which patients were enrolled in
randomized to receive the intervention compared with the integrated care intervention to allow for collabora-
usual care would show the following after a 3-month tion with the integrated care manager, but they were
period: (1) fewer depressive symptoms, (2) lower gly- blinded to enrollment in the usual care group.

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

16
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

Intervention Usual Care

We carried out an integrated care intervention in At baseline, 6, and 12 weeks, patients in the usual care
which the integrated care manager collaborated with group underwent the same assessments as patients in
physicians to offer education and guideline-based the integrated care intervention. Assessments were
treatment recommendations to patients and to monitor conducted in person (as were assessments in the inter-
adherence and clinical status. The key components of vention group). Research assistants who conducted all
this intervention were the provision of an individual- assessments were blinded to patient’s randomization
ized program to improve adherence to antidepressants status.
and oral hypoglycemic agents that recognizes patients’
social and cultural context, and the integration of Measurement Strategy
depression treatment with type 2 diabetes manage- Potential study patients were screened for cognitive
ment. The integrated care manager worked individu- impairment using the MMSE, a short standardized
ally with patients to address the factors involved in mental status examination widely used for clinical and
adherence presented in our conceptual model, adapted research purposes.30 Patients were asked whether they
from Cooper and colleagues,26 and previously pub- resided in a care facility that provided medications on
lished.27,28 In our conceptual model, the patient-level schedule and whether they were unwilling or unable
factors resulting in nonadherence included depres- to use MEMS. At baseline sociodemographic charac-
sion, chronic medical conditions, function, cognition, teristics were assessed using standard questions. Blood
social support, cost of medications, side effects, and pressure was assessed in accordance with American
past experiences with medications. Patient-level fac- Heart Association Guidelines.31 BMI was calculated as
tors were addressed through a variety of activities that weight in kilograms divided by the square of height in
included in-person sessions, telephone contacts, and meters.32 LDL cholesterol was obtained using standard
collaborating with the physician. Through in-person laboratory techniques in accordance with National
sessions and telephone conversations, the integrated Heart Lung and Blood Institute Expert Panel Guide-
care manager provided education about depression lines.33 Functional status was measured using the Medi-
and type 2 diabetes, emphasizing the importance of cal Outcomes Study Short Form (SF-36).34 Adherence
controlling depression to manage diabetes; help to to antidepressants and oral hypoglycemic agents was
identify target symptoms; explanations for the ratio- measured during the 2-week run-in phase and at 6 and
nale for antidepressant and oral hypoglycemic agent 12 weeks using electronic-monitoring data obtained
use; assessment for side-effects and assistance in their from MEMS caps.
management; assessment for progress (eg, reduction in At baseline and 12 weeks blood glycemic control
depressive symptoms and improvement in finger stick was assessed in accordance with American Diabetes
results); assistance with referrals; and monitoring and Association Guidelines.35 HbA1c assays were performed
response to life-threatening symptoms (eg, chest pain, using the in2it A1C Analyzer (Bio-Rad Laboratories,
suicidal thoughts and actions). Hercules, California). Point-of-care testing using this
The intervention was presented to patients as a device has acceptable precision and agreement in com-
supplement to, rather than a replacement for, existing parison with laboratory services.36 Depressive symp-
primary care treatment. We chose this multifaceted toms were measured using the 9-item Patient Health
approach because education alone has not been found Questionnaire (PHQ-9) at baseline, 6, and 12 weeks.
to be effective for improving adherence.29 The inter-
vention consisted of 3, 30-minute in-person sessions Analytic Strategy
(at baseline, 6 weeks, and 12 weeks) and 2, 15-min- We compared baseline characteristics of patients in the
ute telephone-monitoring contacts over a 3-month integrated care intervention with baseline usual care
period. Two research coordinators (1 master’s level patient characteristics using the t test and Fisher’s exact
and 1 bachelor’s level) were trained as integrated care test (for continuous or categorical variables as appro-
managers and administered all intervention activities. priate). In addition, characteristics of patients who
Before trial initiation, the integrated care managers participated and who refused were compared using
received training on pharmacotherapy for depres- t tests and Fisher’s exact tests. Analysis proceeded at
sion and type 2 diabetes management during weekly the patient level, and data were analyzed according
clinical sessions with the principal investigator. (A to the treatment to which patients were randomized
summary of the intervention is available in the Supple- (intent-to-treat). Practice site was included in the mod-
mental Appendix, at http://www.annfammed.org/ els to account for potential clustering by practice. The
content/10/1/15/suppl/DC1.) We conducted a pilot models were adjusted for baseline measures (HbA1c
study among 58 participants.28 or PHQ-9). We used repeated measures linear models

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

17
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

in which mean response (eg, HbA1c and depressive approached; 75 refused to participate, and 190 were
symptoms) depends on the covariates of interest (treat- enrolled (71.7% participation rate). Patients who par-
ment assignment and time since randomization) and ticipated and patients who refused were similar in age,
an unstructured variance-covariance matrix to account sex, and ethnicity. Consent was followed by a 2-week
for the extra correlation within individual patients. run-in phase in which adherence to medications was
The parameter of interest was the time by treatment assessed. After randomization at the 2-week meeting,
interaction, which represents the relative difference in 2 patients in the integrated care intervention were
change over time among the patients assigned to the lost to follow-up, but the remaining 180 patients com-
intervention group compared with patients assigned pleted the final study visit.
to the usual care group. We contrasted the expected
value of the outcome in each treatment group at 12 Sample Characteristics
weeks with the value at baseline, the time of random- Baseline characteristics of the 180 patients random-
ization. The intervention effect was measured as the ized the intervention or usual care are displayed in
difference between the 12-week effect in the treatment Table 1. Baseline characteristics of patients in the
group and the 12-week effect in the
usual care group.
We also considered categorical Figure 1. Study flow diagram.
versions of the type 2 diabetes and
depression outcomes. As recommended 715 Patients assessed for eligibility who were
identified through an electronic medical record
by clinical guidelines, we calculated with a diagnosis of type 2 diabetes mellitus and
whether a patient achieved an HbA1c a prescription for an oral hypoglycmic agent
35 within the past year and a prescription for an
level of less than 7% at 12 weeks. oral antidepressant within the past year
Depression remission was defined by a
PHQ-9 score of less than 5 at follow-
450 patients
up.37 We used logistic regression to
86 No current prescription for
model the categorical diabetes outcome an oral hypoglycemic agent
and repeated measures logistic regres- 240 No current prescription for
sion to model depression remission. For an antidepressant

both models, we report the odds ratio 124 No current prescription for
an oral hypoglycemic agent
and 95% confidence interval comparing or an antidepressant
the intervention group with the usual 75 Patients refused participation
care group.
We defined adherence as the per-
190 Enrolled in 2-week
centage of prescribed doses taken, run-in phase
which we calculated as the number of
doses taken divided by the number of
doses prescribed during the observa- 5 Physicians discontinued the
antidepressant
tion period times 100%. Adherence was
1 Physician discontinued the oral
dichotomized at a threshold of 80% hypoglycemic agent
because the proportion of pills taken 2 Lost to follow-up
was highly skewed and failed normal-
ity assumptions. The 80% cut point
182 Randomized
has been used as a threshold to assess
adherence to medication regimens.38
Analyses were conducted using SAS
9.2 (SAS Institute, Inc, Cary, North
88 To usual care 94 To integrated
Carolina). care intervention

RESULTS 2 Lost to follow-up

The flow of patients through the

trial is depicted in Figure 1. Of 715 88 Completed final 92 Completed final
patients identified by electronic medi- assessment at 12 weeks assessment at 12 weeks
cal records, 265 were eligible and were

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

18
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

integrated care intervention did not differ significantly Clinical Outcomes

from those of patients in the usual care group, includ- Clinical outcomes at 12 weeks are shown in Table 2.
ing adherence rates at baseline measured during the Patients randomized to the integrated care intervention
2-week run-in period. were more likely to achieve a HbA1c levels of less than
7% compared with patients in the usual care group at
Adherence Outcomes 12 weeks (intervention 60.9% vs usual care 35.7%; P
Figure 2 depicts proportions with 80% or greater <.001). Patients in the integrated care intervention also
adherence to oral hypoglycemic agents and antidepres- had a significantly improved mean change in HbA1c
sants over time according to treatment assignment. At levels from baseline compared with patients in the
6 and 12 weeks, a significant improvement in adher- usual care group at 12 weeks (intervention –0.70 vs
ence to oral hypoglycemic agents (P <.001) and anti- usual care 0.50; P <.001). Patients randomized to the
depressants (P <.001) was seen in the intervention in integrated care intervention were more likely to achieve
comparison with usual care. remission of depression compared with patients in the
usual care group at 12 weeks
(intervention 58.7% vs usual care
Table 1. Baseline Characteristics of Study Population 30.7%; P <.001). Patients in the
integrated care intervention also
Usual Care Intervention
 Characteristic (n = 88) (n = 92) P Value had significantly improved mean
change in PHQ-9 scores from
Sociodemographic      
Age, mean (SD) y 57.1 (9.6) 57.8 (9.4) .63
baseline compared with patients
African American, n (%) 48 (54.5) 54 (58.7) .30 in the usual care group at 12
White, n (%) 36 (40.9) 29 (31.5) weeks (intervention –2.42 vs usual
Hispanic, n (%) 3 (3.4) 4 (4.3) care –0.29; P = .007).
Other, n (%) 1 (1.1) 5 (5.4)
Sex, women n (%) 58 (65.9) 64 (69.6) .64
Less than high school education, n (%) 15 (17.0) 14 (15.2) .84 DISCUSSION
Type 2 diabetes mellitus       The primary goal of this study
Years of diabetes, mean (SD) 12.0 (11.8) 10.5 (10.2) .35
was to test the effectiveness of an
HbA1c, mean (SD), % 7.0 (1.9) 7.2 (1.8) .51
integrated care intervention for
Depression      
patients with type 2 diabetes and
PHQ-9 score, mean (SD)a 9.9 (7.2) 10.6 (7.9) .54
Medications      
depression that focused on adher-
Number of medications, mean (SD) 10.1 (5.1) 9.8 (4.5) .66 ence compared with usual care
≥ 80% adherent to oral hypoglycemic 37 (42.0) 33 (35.9) .45 for primary care patients. Our
agent, n (%) principal finding was that pri-
≥ 80% adherent to antidepressant, n (%) 34 (39.0) 28 (30.4) .27 mary care patients randomized to
Functional status (SF-36) scoreb      
the integrated care intervention
Physical function, mean (SD) 53.6 (31.7) 50.8 (32.6) .56
showed higher rates of adher-
Social function, mean (SD) 67.7 (39.9) 76.6 (36.9) .12
Role physical, mean (SD) 49.4 (46.7) 59.5 (46.6) .15
ence to oral hypoglycemics and
Role emotional, mean (SD) 65.9 (46.0) 67.8 (44.6) .79
antidepressants, as well as greater
Bodily pain, mean (SD) 42.3 (31.4) 50.9 (31.7) .07 glucose control and fewer depres-
Cognitive status       sive symptoms, at the final study
MMSE score, mean (SD)c 28.2 (2.3) 28.2 (2.3) .99 visit. Our results show the useful-
Cardiovascular disease risk factors ness of a simple, brief, integrated
Systolic blood pressure, mean (SD), mm Hg 132.3 (21.4) 133.9 (20.4) .62 care management intervention for
Diastolic blood pressure, mean (SD), mm Hg 77.0 (11.4) 76.6 (11.1) .85 primary care patients with type 2
Body mass index, mean (SD) 33.8 (8.3) 34.5 (10.8) .62 diabetes and depression.
LDL cholesterol,d mean (SD), mg/dL 96.9 (27.4) 95.6 (33.4) .77
Before discussing implications
HbA1c = glycated hemoglobin; LDL = low-density lipoprotein; MMSE = Mini-Mental State Examination; of our findings, the limitations
PHQ-9 = 9-item Patient Health Questionnaire; SF-36 = Medical Outcomes Study Short Form.
of our study require discussion.
Note: P Values represent comparisons according to the Fisher exact test and t tests for categorical or continuous
data, respectively. First, our results were obtained
a
Scored on a range from 0 to 27, where lower scores represent fewer depressive symptoms.
from patients who receive care at
b
Scored on a range from 0 to 100, where lower scores represent functional impairment. 3 primary care sites, which might
c
Scored on a range from 0 to 30, where lower scores represent cognitive impairment. not be representative of most pri-
d
n = 175 because of missing values.
mary care practices. The 3 prac-

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

19
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

tices, however, were diverse and varied in size,

Figure 2. Patient outcome of ≥80% adherence to oral
and they were probably similar to other primary
hypoglycemic agent and to antidepressant medication.
care practices in the region. Second, all meth-
Usual Care Intervention ods for assessing adherence have limitations.
100 We chose to use MEMS caps as our primary
90 measure of adherence because MEMS caps have
80 a low failure rate38 and may be more sensitive
Oral Hypoglycemic, %

P<.001 P<.001
≥80% Adherence to

than other adherence measures.39 Any effect of

}
70
60
50
P=.45 } MEMS caps on medication adherence would
be experienced equally in both groups. Third,
}

40 while the 80% threshold for adherence has been

30 assessed in some clinical research (eg, George et
20 al38), the clinical relevance of this threshold has
10 not been tested for many medications. Fourth,
0 patients in the usual care group did not have the
A Baseline 6 Weeks 12 Weeks same number of in-person contacts as those in
the integrated care intervention to control for
Usual Care Intervention
the effects of attention.
100
We adopted a comprehensive set of guide-
90
lines for the management of depression and type
80
2 diabetes into an integrated care intervention
to Antidepressant, % 

70
P<.001 P<.001 and applied them in the context of a primary
≥80% Adherence

60
care setting in which most adults receive their
P=.45 medical care. Approximately 90% of all persons
50
}

40 with diabetes receive continuous care from pri-

40
30 mary care physicians. No more than 20% of
20 people with diabetes ever see an endocrinolo-
10 gist, and there are not enough endocrinologists
0 to handle the increasing number of people with
B Baseline 6 Weeks 12 Weeks diabetes.41 The American Diabetes Associa-
Note: Assessed with the Medication Event Monitoring System at baseline (preinterven-
tion has modified its guidelines to recommend
tion) and at 6 and 12 weeks postintervention, according to treatment assignment. routine screening for depression for diabetic
patients, particularly those
with poor adherence.35 Yet,
Table 2. Clinical Outcomes of Glycemic Control and Depression Symptoms many primary care patients
in Usual Care and in the Integrated Intervention at 12 Weeks with type 2 diabetes receive
Estimated
inadequate treatment
Unadjusted Estimate Between-Group because of poor adherence to
Odds Ratio oral hypoglycemic agents. At
Outcomes for Type 2 Usual Care Intervention or Difference P
Diabetes Mellitus (n = 88) (n = 92) (95% CI) Value
the same time, the treatment
of type 2 diabetes needs to
Glucose control
be better integrated with the
Achieved HbA1c <7%, n (%) 25 (35.7) 67 (60.9) 8.48 (3.24 to 22.2)a <.001
HbA1c, change from
depression that often co-
0.50 (1.11) –0.70 (1.32) –1.20 (–1.56 to 0.84)b <.001 occurs. Improved manage-
baseline, mean (SD)
Depression         ment of both type 2 diabetes
Achieved remission and depression could have
27 (30.7) 54 (58.7) 6.15 (2.93 to 12.92) d
<.001
(PHQ-9 <5c), n (%)
PHQ-9 scorec change from
an important public health
–0.29 (5.74) –2.42 (4.75) –2.13 (-3.68 to –0.59) .007
baseline, mean (SD) impact on patient functional
HbA = glycated hemoglobin; PHQ-9 = 9-item Patient Health Questionnaire; SD = standard deviation.
1c
status and mortality.42
Estimates, 95% confidence intervals, and P values from the statistical models. Our intervention builds
a
Odds ratio (95% CI) from a logistic regression model. on the work of Katon et al,22
b
Mean group difference (95% CI) from a repeated measures linear regression model. because our intervention
c
PHQ-9 scored on a range from 0 to 27, where lower scores represent fewer depressive symptoms.
is brief, does not require a
d
Odds ratio (95% CI) from a repeated measures logistic regression model.
high level of expertise, and is

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

20
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

adaptable to community-based primary care settings, 3. Nouwen A, Winkley K, Twisk J, et al; European Depression in Dia-
betes (EDID) Research Consortium. Type 2 diabetes mellitus as a risk
where more than one-half of patients seen may be from factor for the onset of depression: a systematic review and meta-
underrepresented minority groups at high risk for type analysis. Diabetologia. 2010;53(12):2480-2486.
2 diabetes and poor outcomes.43-45 A recent review of 4. Lustman PJ, Clouse RE. Depression in diabetic patients: the relation-
diabetes self-management interventions noted that an ship between mood and glycemic control. J Diabetes Complications.
2005;19(2):113-122.
assessment of the feasibility of many interventions is
limited by failure to report overall contact time with 5. Simon GE, Katon WJ, Lin EH, et al. Cost-effectiveness of systematic
depression treatment among people with diabetes mellitus. Arch
study patients.12 The total contact time for our inter- Gen Psychiatry. 2007;64(1):65-72.
vention was 2 hours (3, 30-minute in-person meetings 6. Gonzalez JS, Peyrot M, McCarl LA, et al. Depression and diabetes
and 2, 15-minute telephone contacts). Compared with treatment nonadherence: a meta-analysis. Diabetes Care. 2008;31
several systematic reviews,10,12,46 we had fewer study (12):2398-2403.

visits and substantially less total contact time than 7. Von Korff M, Katon W, Lin EH, et al. Potentially modifiable factors
associated with disability among people with diabetes. Psychosom
most interventions targeting diabetes management. Med. 2005;67(2):233-240.
Our study provides a sustainable solution that can 8. Schram MT, Baan CA, Pouwer F. Depression and quality of life
be implemented in primary care or other settings for in patients with diabetes: a systematic review from the European
patients managing multiple medical conditions and vary- depression in diabetes (EDID) research consortium. Curr Diabetes
Rev. 2009;5(2):112-119.
ing degrees of complexity in pharmacotherapeutic regi-
mens. Given the low rates of adherence during the first 9. Koopmans B, Pouwer F, de Bie RA, van Rooij ES, Leusink GL, Pop
VJ. Depressive symptoms are associated with physical inactivity in
2 weeks before randomization in our study and the cor- patients with type 2 diabetes. The DIAZOB Primary Care Diabetes
responding difficulty patients experience in adhering to study. Fam Pract. 2009;26(3):171-173.
physician recommendations for the treatment of depres- 10. Renders CM, Valk GD, Griffin SJ, Wagner EH, Eijk Van JT, Assendelft
sion and type 2 diabetes, interventions that allow for WJ. Interventions to improve the management of diabetes in pri-
mary care, outpatient, and community settings: a systematic review.
tailoring content and providing tools to match the indi- Diabetes Care. 2001;24(10):1821-1833.
vidualized needs of patients are needed. Ancillary health 11. Williams JW Jr, Gerrity M, Holsinger T, Dobscha S, Gaynes B, Diet-
personnel who are already working in primary care rich A. Systematic review of multifaceted interventions to improve
practices could be trained to carry out the interven- depression care. Gen Hosp Psychiatry. 2007;29(2):91-116.
tion. Our results call for greater emphasis within health 12. Leeman J. Interventions to improve diabetes self-management: util-
ity and relevance for practice. Diabetes Educ. 2006;32(4):571-583.
care systems and policy organizations on the develop-
13. Olfson M, Marcus SC, Tedeschi M, Wan GJ. Continuity of antide-
ment and promotion of clinical programs to enhance
pressant treatment for adults with depression in the United States.
medication adherence, particularly among patients with Am J Psychiatry. 2006;163(1):101-108.
chronic medical conditions and depression. 14. Kodner DL, Spreeuwenberg C. Integrated care: meaning, logic,
applications, and implications—a discussion paper. Int J Integr Care.
To read or post commentaries in response to this article, see it
2002;2:e12.
online at http://www.annfammed.org/content/10/1/15.
15. Balkrishnan R, Rajagopalan R, Camacho FT, Huston SA, Murray FT,
Key words: Medication adherence; type 2 diabetes mellitus; depres- Anderson RT. Predictors of medication adherence and associated
health care costs in an older population with type 2 diabetes mel-
sion; comorbidity; chronic disease; primary health care; randomized
litus: a longitudinal cohort study. Clin Ther. 2003;25(11):2958-2971.
controlled trial
16. DiMatteo MR, Lepper HS, Croghan TW. Depression is a risk factor
Submitted July 19, 2011; submitted, revised, October 12, 2011; accepted for noncompliance with medical treatment: meta-analysis of the
effects of anxiety and depression on patient adherence. Arch Intern
October 25, 2011.
Med. 2000;160(14):2101-2107.

Funding support: This work was supported by American Diabetes Asso- 17. Dirmaier J, Watzke B, Koch U, et al. Diabetes in primary care: pro-
spective associations between depression, nonadherence and glyce-
ciation Clinical Research Award 1-09-CR-07. Dr Bogner was supported by
mic control. Psychother Psychosom. 2010;79(3):172-178.
NIMH grant MH082799 and MH047447. Dr Morales was supported by
a NIMH-mentored Career Development Award (MH073903). 18. Ciechanowski PS, Katon WJ, Russo JE. Depression and diabetes:
impact of depressive symptoms on adherence, function, and costs.
Arch Intern Med. 2000;160(21):3278-3285.
Clinical Trial Registration: Integrating Depression Services Into
DM Management, NCT01098253, http://clinicaltrials.gov/show/ 19. Katon WJ, Von Korff M, Lin EH, et al. The Pathways Study: a ran-
NCT01098253. domized trial of collaborative care in patients with diabetes and
depression. Arch Gen Psychiatry. 2004;61(10):1042-1049.
20. McKellar JD, Humphreys K, Piette JD. Depression increases diabetes
References symptoms by complicating patients’ self-care adherence. Diabetes
Educ. 2004;30(3):485-492.
1. Golden SH, Lazo M, Carnethon M, et al. Examining a bidirectional
association between depressive symptoms and diabetes. JAMA. 21. Petrak F, Herpertz S. Treatment of depression in diabetes:
2008;299(23):2751-2759. an update. Curr Opin Psychiatry. 2009;22(2):211-217.
2. Knol MJ, Twisk JW, Beekman AT, Heine RJ, Snoek FJ, Pouwer F. 22. Katon WJ, Lin EH, Von Korff M, et al. Collaborative care for
Depression as a risk factor for the onset of type 2 diabetes mellitus. patients with depression and chronic illnesses. N Engl J Med.
A meta-analysis. Diabetologia. 2006;49(5):837-845. 2010;363(27):2611-2620.

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

21
 M A N AG E M EN T O F D I A B E T E S A N D D EP R E S S I O N

23. Angst J. Clinical course of affective disorders. In: Helgason T, Daly 35. American Diabetes Association. Clinical practice recommendations.
R, eds. Depression Illness: Prediction of Course and Outcome. Berlin, Diabetes Care. 2010;33(Suppl 1):S1-S100.
Germany: Springer-Verlag; 1988:1-47.
36. Moridani MY, Verjee Z, Allen LC. Analytical evaluation of hemoglo-
24. Kahn R, Alperin P, Eddy D, et al. Age at initiation and frequency of bin A(1c) dual kit assay on Bio-Rad Variant II: an automated HPLC
screening to detect type 2 diabetes: a cost-effectiveness analysis. hemoglobin analyzer for the management of diabetic patients. Clin
Lancet. 2010;375(9723):1365-1374. Biochem. 2003;36(4):317-320.
25. Crum RM, Anthony JC, Bassett SS, Folstein MF. Population-based 37. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief
norms for the Mini-Mental State Examination by age and educa- depression severity measure. J Gen Intern Med. 2001;16(9):606-613.
tional level. JAMA. 1993;269(18):2386-2391.
38. George CF, Peveler RC, Heiliger S, Thompson C. Compliance with
26. Cooper LA, Gonzales JJ, Gallo JJ, et al. The acceptability of treat- tricyclic antidepressants: the value of four different methods of
ment for depression among African-American, Hispanic, and white assessment. Br J Clin Pharmacol. 2000;50(2):166-171.
primary care patients. Med Care. Apr 2003;41(4):479-489.
39. Farmer KC. Methods for measuring and monitoring medication
27. Bogner HR, de Vries HF: Integration of depression and hyperten- regimen adherence in clinical trials and clinical practice. Clin Ther.
sion treatment: a pilot, randomized controlled trial. Ann Fam Med. 1999;21(6):1074-1090, discussion 1073.
2008;6(4): 295-301. 
40. Smith DM. Toward common ground. Diabetes Care. 1997;20(4):
28. Bogner HR, de Vries HF. Integrating type 2 diabetes mellitus and 467-468.
depression treatment among African Americans: a randomized con-
41. Saudek CD. The role of primary care professionals in managing dia-
trolled pilot trial. Diabetes Educ. 2010;36(2):284-292.
betes. Clin Diabetes. 2002;20(2):65-66.
29. Mundt JC, Clarke GN, Burroughs D, Brenneman DO, Griest JH.
42. Bogner HR, Morales KH, Post EP, Bruce ML. Diabetes, depression,
Effectiveness of antidepressant pharmacotherapy: the impact of
and death: a randomized controlled trial of a depression treatment
medication compliance and patient education. Depress Anxiety.
program for older adults based in primary care (PROSPECT). Diabe-
2001;13(1):1-10.
tes Care. 2007;30(12):3005-3010.
30. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practi-
43. Wisdom K, Fryzek JP, Havstad SL, Anderson RM, Dreiling MC, Til-
cal method for grading the cognitive state of patients for the clini-
ley BC. Comparison of laboratory test frequency and test results
cian. J Psychiatr Res. 1975;12(3):189-198.
between African-Americans and Caucasians with diabetes: opportu-
31. American Heart Association. Blood pressure testing and assess- nity for improvement. Findings from a large urban health mainte-
ment. http://www.americanheart.org. Accessed Dec 14, 2010. nance organization. Diabetes Care. 1997;20(6):971-977.
32. Carnethon MR, Kinder LS, Fair JM, Stafford RS, Fortmann SP. Symp- 44. Harris MI. Racial and ethnic differences in health care access and
toms of depression as a risk factor for incident diabetes: findings health outcomes for adults with type 2 diabetes. Diabetes Care.
from the National Health and Nutrition Examination Epidemiologic 2001;24(3):454-459.
Follow-up Study, 1971-1992. Am J Epidemiol. 2003;158(5):416-423.
45. Kirk JK, D’Agostino RB Jr, Bell RA, et al. Disparities in HbA1c levels
33. National Cholesterol Education Program (NCEP) Expert Panel on between African-American and non-Hispanic white adults with dia-
Detection, Evaluation, and Treatment of High Blood Cholesterol in betes: a meta-analysis. Diabetes Care. 2006;29(9):2130-2136.
Adults (Adult Treatment Panel III). Third Report of the National Cho-
46. Harkness E, Macdonald W, Valderas J, Coventry P, Gask L, Bower P.
lesterol Education Program (NCEP) Expert Panel on Detection, Eval-
Identifying psychosocial interventions that improve both physical
uation, and Treatment of High Blood Cholesterol in Adults (Adult
and mental health in patients with diabetes: a systematic review
Treatment Panel III) final report. Circulation. 2002;106(25):3143.
and meta-analysis. Diabetes Care. 2010;33(4):926-930.
34. Stewart AL, Hays RD, Ware JE Jr. The MOS short-form general
health survey. Reliability and validity in a patient population. Med
Care. 1988;26(7):724-735.

ANNALS O F FAMILY MEDICINE ✦ WWW.ANNFA MME D.O R G ✦ VO L. 10, N O. 1 ✦ JA N UA RY/FE BRUA RY 2012

22