Guo et al.

Risk of Diabetes Mellitus Associated With

Atypical Antipsychotic Use Among Patients With
Bipolar Disorder: A Retrospective, Population-Based,
Case-Control Study
Jeff J. Guo, Ph.D.; Paul E. Keck, Jr., M.D.; Patricia K. Corey-Lisle, Ph.D.;
Hong Li, Ph.D.; Dongming Jiang, M.S.; Raymond Jang, Ph.D.;
and Gilbert J. L’Italien, Sc.D.

Received Aug. 31, 2005; accepted Jan. 17, 2006. From the College
of Pharmacy, University of Cincinnati Medical Center, Cincinnati, Ohio
(Drs. Guo and Jang and Mr. Jiang); the Institute for the Study of Health,
University of Cincinnati, Cincinnati, Ohio (Dr. Guo); the Department of
Background: Drug-induced diabetes onset Psychiatry, University of Cincinnati College of Medicine, and the Mental
has not been adequately quantified in patients with Health Care Line and General Clinical Research Center, Cincinnati
bipolar disorder, although atypical antipsychotics Veterans Affairs Medical Center, Cincinnati, Ohio (Dr. Keck); and
have been widely used as new mood stabilizers. Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford,
Objectives: To quantify the association between Conn. (Drs. Corey-Lisle, Li, and L’Italien).
atypical antipsychotics and diabetes mellitus. This project was conducted with research grant support provided
by the Bristol-Myers Squibb Pharmaceutical Research Institute.
Method: A retrospective, population-based, This study was presented at the European College of
case-control study was conducted using the medical Neuropsychopharmacology (ECNP) 18th Conference, Amsterdam,
claims database from U.S. managed care organiza- Netherlands, October 22–26, 2005; and the International Conference
tions from January 1, 1998, to December 31, 2002. of Pharmacoepidemiology (ICPE), Nashville, Tenn., August 21–24, 2005.
Nine hundred twenty incident cases of diabetes Dr. Keck is a consultant to or member of the scientific advisory
boards of Abbott, AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline,
were matched with 5258 controls by age, sex, Janssen, Eli Lilly, Ortho-McNeil, Pfizer, and Shire, and is a principal or
and bipolar index month and year. Diabetes cases co-investigator on research studies sponsored by Abbott, the American
were identified by either diagnosis of ICD-9 codes Diabetes Association, AstraZeneca, Bristol-Myers Squibb,
or diabetic medications. Patients with diabetes had GlaxoSmithKline, Eli Lilly, Janssen, Merck, the National Institute
a minimum 3-month exposure to any medications of Mental Health, the National Institute of Drug Abuse, Organon,
or at least 3 prescriptions for their bipolar or Ortho-McNeil, Pfizer, the Stanley Medical Research Institute, and UCB
Pharma. Drs. Corey-Lisle, Li, and L’Italien are employees of Bristol-
comorbidity treatment. Cox proportional hazard Myers Squibb. Drs. Guo and Jang and Mr. Jiang report no additional
regression was conducted to assess the risk of financial or other relationships relevant to the subject of this article.
diabetes associated with antipsychotic use. The opinions and conclusions expressed in this manuscript are solely
Results: Of 920 cases, 41% received atypical those of the authors.
antipsychotics (e.g., olanzapine, risperidone, que- Corresponding author and reprints: Jeff J. Guo, Ph.D., Associate
tiapine, ziprasidone, clozapine) and 34% received Professor of Pharmacoepidemiology and Pharmacoeconomics,
University of Cincinnati Medical Center, 3225 Eden Ave.,
conventional antipsychotics. Compared to patients Cincinnati, OH 45267-0004 (e-mail: [email protected]).
receiving conventional antipsychotics, the risk
of diabetes was greatest among patients taking
clozapine (hazard ratio [HR] = 7.0, 95% confidence
interval [CI] = 1.7 to 28.9), risperidone (HR = 3.4,
95% CI = 2.8 to 4.2), olanzapine (HR = 3.2, 95%
CI = 2.7 to 3.8), and quetiapine (HR = 1.8, 95%
CI =1.4 to 2.4), with controlling covariates of age;
sex; duration of follow-up; use of lithium, anticon-
M ood stabilizers like lithium, divalproex, and car-
bamazepine are traditionally used for bipolar
treatment. Antiepileptic drugs (lamotrigine) and atypical
vulsants, antidepressants, or concomitant drugs; antipsychotics (aripiprazole, clozapine, olanzapine, que-
and psychiatric and medical comorbidities. tiapine, risperidone, ziprasidone) are emergent therapies
Conclusion: Development or exacerbation of for bipolar disorder.1,2 Atypical antipsychotic agents with
diabetes mellitus is associated with antipsychotic different mechanisms of action from conventional anti-
use in bipolar patients. Metabolic complications
are a major issue in patients receiving antipsychotic
psychotics have been widely adopted in the treatment of
therapy. Thus, the propensity of an antipsychotic bipolar disorder since the mid-1990s.3 Although atypical
to induce diabetes should be a consideration when antipsychotics reduce extrapyramidal side effects, they
selecting an agent for patients with bipolar disorder. have a different spectrum of side effects, including weight
(J Clin Psychiatry 2006;67:1055–1061) gain, alterations in glucose metabolism, increased con-
centrations of blood cholesterol and lipids, myocarditis,
and cardiomyopathy.4–9

1056
1055 © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE
J Clin PRESS
Psychiatry 67:7, July ,2006
INC.
 Diabetes Risk in Bipolar Patients

Evidence has shown an association between some Figure 1. Selection of Incident Cases of Diabetesa and
antipsychotics and diabetes in patients with schizophre- Controls From Patients With Bipolar Disorder in a
nia.7,10–15 Recently, some cases of diabetic ketoacidosis Large Managed Care Population, 1998–2002
and diabetes associated with antipsychotics were also re-
Managed Care Population With
ported in adult16–18 and pediatric19,20 bipolar patients. Al- Affective Disorder or Cyclothymia
though most of the articles were case reports documenting (N = 709,865)

the incidence of diabetes or hyperglycemia with use of 44,095 Medicaid Patients Were Excluded
atypical antipsychotics, some studies reported that pa-
tients with schizophrenia exposed to clozapine, olanza- 665,770 Non-Medicaid
Patients With 296.xx or
pine, and risperidone were significantly associated with 301.13 Diagnosis
an increased risk of glucose intolerance ranging from a
hazard ratio (HR) of 1.2 based on the Veterans Affairs da- 15,199 Patients With Schizophrenia (295.xx)
and 527,279 Patients With Only Depression
tabase,21,22 to HRs of 4.7 and 5.8 based on the United (296.2x or 296.3x) Were Excluded
Kingdom General Practice Research (GPRD) database,7,23
to an HR of 10.22 based on the World Health Organization 123,292 Study Patients
With Bipolar Disorder
adverse drug reaction database.24 Very few case reports
exist for quetiapine or ziprasidone despite these drugs
having similar pharmacotherapy characteristics.
Diabetes is a known and infrequent adverse effect 920 Patients Selected 5258 Patients Selected
of olanzapine and risperidone. Drug-induced diabetes on- as Cases of Diabetesb as Controlsc

set has not been adequately quantified in patients with a

Incident cases of diabetes were identified by either earliest diagnosis
bipolar disorder, although atypical antipsychotics are be- of ICD-9 code 250.xx or treatment for diabetes.
b
ing increasingly used in the treatment of bipolar disorder. Patients were selected if they had at least a minimum of 3 month’s
exposure to medications or at least 3 prescriptions during the study
Published reports indicated some drugs are known to af- period.
c
fect the risk of developing diabetes or hyperglycemia, in- Each case was matched with 6 controls by age, sex, and bipolar index
month and year. Eighty-two case subjects with fewer than 6 matched
cluding α-adrenergic blockers (e.g., doxazosin, prazosin, controls were included in the analysis.
terazosin), β-adrenergic blockers (e.g., atenolol, betaxo-
lol, bisoprolol), thiazide diuretics (e.g., chlorothiazide,
chlorthalidone, polythiazide), corticosteroids (e.g., meth- managed care claims databases to conduct pharmacoepi-
ylprednisolone, hydrocortisone), phenytoin, oral contra- demiologic studies has been well documented.28–30
ceptives containing norgestrel, and valproic acid.8,25,26 We To protect patient confidentiality, patient names, insur-
used medical claims data from U.S. managed-care organi- ance plan identification numbers, and other patient iden-
zations to quantify the risk of diabetes associated with tifiers were deleted from the claims database. Random-
antipsychotics, especially atypical antipsychotics, in pa- ized patient numbers and patient birth years were used
tients with bipolar disorder. for identification and calculation of age, respectively. The
research project was approved by the University of
PATIENTS AND METHOD Cincinnati Medical Center Institutional Review Board.
A retrospective, population-based, case-control (nested
Study Design and Population case-control) study was conducted. From 1998 to 2002,
The primary data source was a multi-state managed a total of 709,865 patients, including 6.2% Medicaid en-
care claims database (PharMetrics) covering January 1, rollees, had at least 1 diagnosis of an affective disorder or
1998, to December 31, 2002 (5 calendar years). The data- cyclothymia (Figure 1). Due to different socioeconomic
base included all pharmacy, medical, and institutional characteristics of the Medicaid population, we selected
claims. Each medical claim was recorded with accom- a cohort of 123,292 non-Medicaid patients who had a bi-
panying diagnostic codes (International Classification of polar diagnosis indicated by any of the following ICD-9
Diseases, Ninth Revision [ICD-9]) that justified the med- codes: 296.0, 296.1, 296.4–296.8. Patients with a diagno-
ical service. The database includes over 45 million lives sis of depression only (ICD-9 code = 296.2x or 296.3x)
enrolled in managed care organizations with 70 health or schizophrenia (ICD-9 code = 295.xx) during the study
plans, including managed care Medicaid programs, in 4 period were excluded from this population. Because num-
U.S. regions: mid-west (34.1%), east (15.6%), south bers of patients with cyclothymia were less than 0.1%, pa-
(23.9%), and west (26.4%). Population distributions are tients with cyclothymia were not categorized separately.
similar to the U.S. population distribution by age and gen-
der distributions (PharMetrics, 2004).27 This geographi- Patient Selection
cally diversified claims database provides a large pop- Because published reports show that drug-induced
ulation perspective of health information. The use of diabetes usually occurs with recent or current use of anti-

J©Clin
COPYRIGHT 2006
Psychiatry PHYSICIANS
67:7, July 2006 POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. 1057
1056
Guo et al.

psychotic drugs,10–23 we selected a cohort of patients who Table 1. Characteristics for Study Population, Incident Cases
had at least a minimum of 3 month’s exposure to any of Diabetes, and Controls
medications or at least 3 prescriptions for their bipolar or Study Population Cases Controls
comorbidity treatment during the study period. Incident (N = 123,292), (N = 920), (N = 5258),
Characteristic N (%) N (%) N (%)
cases of diabetes were identified by either earliest diag-
Age, y
nosis of ICD-9 code 250.xx or treatment for diabetes. The ≤ 12 5515 (4.47) 19 (2.07) 101 (1.92)
date for the first diabetes diagnosis or use of diabetic 13–17 12,006 (9.74) 39 (4.24) 234 (4.45)
medication was defined as the diabetes index date. To en- 18–34 35,916 (29.13) 144 (15.65) 854 (16.24)
35–49 45,191 (36.65) 413 (44.89) 2477 (47.11)
sure that the patients with diabetes were incident cases, 50–64 21,754 (17.64) 263 (28.59) 1504 (28.60)
we checked the medical and prescription claim records 65+ 2910 (2.36) 42 (4.57) 88 (1.67)
for any diagnosis of or treatment for diabetes before the Sex
Female 74,786 (60.66) 601 (65.33) 3473 (66.05)
diabetes index date. Patients identified as cases should Male 48,506 (39.34) 319 (34.67) 1785 (33.95)
not have had a prescription for oral antidiabetic agents Use of medicationsa
before the diabetes index date. A total of 78 patients who Lithium 13,014 (10.56) 177 (19.24) 666 (12.67)
Anticonvulsants 30,313 (24.59) 395 (42.93) 1355 (25.77)
had received insulin and/or oral antidiabetic agents be- Atypical antipsychotics 13,560 (11.00) 378 (41.09) 592 (11.26)
fore the diabetes index date were excluded in order to Olanzapine 6020 (4.88) 186 (20.22) 258 (4.91)
eliminate potential patients with preexisting diabetes. Quetiapine 3228 (2.62) 79 (8.59) 166 (3.16)
Risperidone 4566 (3.70) 130 (14.13) 186 (3.54)
The oral antidiabetic agents included sulfonylurea drugs Ziprasidone 472 (0.38) 9 (0.98) 11 (0.21)
(e.g., acetohexamide, glipizide, glyburide), biguanide Clozapine 30 (0.02) 2 (0.22) 3 (0.06)
(metformin), glitazones (e.g., pioglitazone, rosiglita- Switched atypicals 627 (0.51) 20 (2.17) 29 (0.55)
Antidepressants 40,521 (32.87) 436 (47.39) 1912 (36.36)
zone), α-glucosidase inhibitors (e.g., miglitol, acarbose), Conventional 20,042 (16.26) 314 (34.13) 1005 (19.11)
and other new drugs like repaglinide and nateglinide. antipsychotics
For each case, we matched 6 controls with age at in- a
Use of different medications was not mutually exclusive for one
dex date (standard deviation of 5 years), sex, and bipolar patient.
diagnosis index month and year. Controls that met the
matching criteria were selected at random with SAS ver-
sion 8.0 software (SAS Institute, Cary, N.C.). Controls
were selected from patients who had been diagnosed as roids, thiazide diuretics, phenytoin, oral contraceptives,
having bipolar disorder but had not been diagnosed as or valproic acid.
having diabetes and were not treated for diabetes at any
time during the study period. Because bipolar diagnosis Statistical Analysis
index month and year were part of matching criteria, the The age of each patient was calculated as the number
calendar time distributions of the bipolar index date were of years between the index date of bipolar diagnosis and
the same for both cases and controls. birth year. The index date of bipolar diagnosis was the
first date of diagnosis indicated by defined ICD-9 codes
Drug Use for bipolar during the study period. Age categories were
We classified antipsychotics as conventionals and ≤ 12, 13–17, 18–34, 35–49, 50–64, and 65 years or
atypicals. Atypical antipsychotics included olanzapine, older.
risperidone, quetiapine, ziprasidone, and clozapine. Ari- We conducted all analyses with SAS version 8.0. We
piprazole was not included for this analysis as it was not conducted the Cox proportional hazard regression to as-
available during the study period. Patients might switch sess the risk of development diabetes associated with
from one atypical antipsychotic to another during the de- antipsychotic use due to the consideration of time-to-
fined study period. Conventional antipsychotics included event with censoring and covariates. We used 2 different
haloperidol, chlorpromazine, fluphenazine, loxapine, referent groups to compare the risk of diabetes develop-
molindone, perphenazine, thioridazine, trifluoperazine, ing among patients receiving different antipsychotics.
thiothixene, and pimozide. The first group included all patients except those receiv-
For both cases and controls, we abstracted all pre- ing the specific atypical antipsychotic drug of interest.
scription drug claims dispensed and reimbursed for the The second group included patients taking conventional
treatment of bipolar disorder and diabetes between the antipsychotics.
start of the study period and the index date of diabetes, In addition to matching variables, we adjusted the
the end of the study period, or the end of enrollment, analysis for use of other drugs known to affect the risk
whichever came first. We used dichotomous variables to of diabetes, psychiatric comorbidities (alcohol abuse,
indicate whether a patient had received concomitant substance abuse disorder, personality disorder, anxiety
drugs that have known association with diabetes or hy- disorder, and impulse-control disorder), and medical
perglycemia, that is, α-blockers, β-blockers, corticoste- comorbidities (hypertension, obesity, arthritis, cerebral

1058
1057 © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE
J Clin PRESS
Psychiatry 67:7, July ,2006
INC.
 Diabetes Risk in Bipolar Patients

Table 2. Exposure Hazard Ratios and 95% Confidence Intervals (CIs) for Development of Diabetes in Patients Using Different
Antipsychoticsa
Use of Antipsychotics Unadjusted Hazard Ratiob (95% CI) p Value Adjusted Hazard Ratioc (95% CI) p Value
Atypical antipsychotics
Olanzapine 5.378 (4.556 to 6.348) < .0001 4.045 (3.384 to 4.834) < .0001
Quetiapine 3.588 (2.833 to 4.545) < .0001 2.300 (1.799 to 2.943) < .0001
Risperidone 4.868 (4.025 to 5.888) < .0001 3.484 (2.842 to 4.270) < .0001
Ziprasidone 6.643 (3.423 to 12.891) < .0001 4.642 (2.383 to 9.042) < .0001
Clozapine 7.289 (1.811 to 29.335) .0052 6.872 (1.702 to 27.746) < .0001
Switched atypicals 3.896 (2.490 to 6.095) < .0001 2.293 (1.452 to 3.621) < .0001
Conventional antipsychotics 2.127 (1.849 to 2.447) < .0001 1.495 (1.263 to 1.770) < .0001
a
For each Cox proportional hazard regression, the referent group involved all patients except those receiving the drug of interest.
b
Unadjusted model includes age, sex, and bipolar follow-up months.
c
Adjusted for age, sex, bipolar follow-up months, and use of medication (lithium, anticonvulsants, antidepressants, α-blockers, β-blockers,
corticosteroids, thiazide diuretics, phenytoin, valproic acid, or oral contraceptives).

vascular disease [CVD], chronic obstructive pulmonary models for age, sex, duration of bipolar follow-up, use of
disease [COPD], dyslipidemia, and coronary heart disease medications, and concomitant drugs.
[CHD]). Compared to patients receiving conventional antipsy-
chotics, the risk of diabetes was also greatest among pa-
RESULTS tients taking clozapine (HR = 7.0, 95% CI = 1.7 to 28.9),
olanzapine (HR = 3.2, 95% CI = 2.7 to 3.8), risperidone
For study patients with bipolar disorder, females were (HR = 3.4, 95% CI = 2.8 to 4.2), and quetiapine (HR =
more frequent than males (see Table 1). During the study 1.8, 95% CI = 1.4 to 2.4), with controlling covariates of
period from 1998 to 2002, 13,560 study patients (11%) age; sex; duration of follow-up; use of lithium, anticon-
had at least 1 prescription for atypical antipsychotics, vulsants, antidepressants, or concomitant drugs; and psy-
20,042 patients (16%) had at least 1 prescription for con- chiatric and medical comorbidities (see Table 3).
ventional antipsychotics, 13,014 patients (11%) had at
least 1 prescription for lithium, 30,313 patients (25%) had DISCUSSION
at least 1 prescription for anticonvulsants, and 40,521 pa-
tients (33%) had at least 1 prescription for antidepressants. This is a multi-state, population-based, case-control
Based on the study inclusion and exclusion criteria, study examining the risk of developing diabetes associ-
920 cases of diabetes were identified and matched with ated with antipsychotics in patients with bipolar disorder.
5258 controls. Eighty-two cases that had fewer than 6 After controlling for personal risk factors and concomi-
controls per case were kept for the analysis. The majority tant drug use, we found that patients receiving conven-
of those cases were older patients who had a range of tional or atypical antipsychotics for bipolar disorder have
matched controls from 2 to 4 patients. The age and sex an increased risk of diabetes. It is unclear how much dia-
of these cases and controls were similar. Compared to betes mellitus in the study population might be due to the
controls, the cases more frequently used atypical anti- use of antipsychotics compared to the underlying disease
psychotics and conventional antipsychotics, as well as of bipolar disorder, poorer overall physical health, less
lithium, anticonvulsants, and antidepressants (see Table healthy lifestyles, or poorer access to health care services.
1). Of 920 cases, 41% received atypical antipsychotics, Atypical and conventional antipsychotics are often
including 20% olanzapine, 14% risperidone, 9% quetia- distinguished by their adverse effects. Atypical antipsy-
pine, and 1% ziprasidone. About 2% of patients in the chotics are generally regarded as having low potential for
case group switched from one atypical antipsychotic to causing extrapyramidal symptoms and a high serotonin-
another. to-dopamine receptor affinity.9,31 Literature indicates
Table 2 summarizes the Cox proportional hazard re- that clozapine and olanzapine are more likely to be as-
gression analyses. The risk of developing diabetes was sociated with diabetes mellitus (indicated by diabetic
greatest among clozapine users (HR = 6.9, 95% CI = 1.7 ketoacidosis and atherogenic lipid profile) than other
to 27.7), ziprasidone users (HR = 4.6, 95% CI = 2.4 to atypical agents.7,21,22,32,33 One possible mechanism for hy-
9.0), olanzapine users (HR = 4.0, 95% CI = 3.4 to 4.8), perglycemia is impairment of insulin resistance, which
risperidone users (HR = 3.5, 95% CI = 2.8 to 4.3), quetia- may occur because of weight gain or a change in body fat
pine users (HR = 2.3, 95% CI = 1.8 to 2.9), patients re- distribution or by a direct effect on insulin-sensitive target
ceiving switched atypical antipsychotics (HR = 2.3, 95% tissues.7,34
CI = 1.5 to 3.6), and patients receiving conventional anti- Compared to published pharmacoepidemiologic stud-
psychotics (HR = 1.5, 95% CI = 1.3 to 1.8), with adjusted ies of patients with schizophrenia,7,21–24 our findings from

J©Clin
COPYRIGHT 2006
Psychiatry PHYSICIANS
67:7, July 2006 POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. 1059
1058
Guo et al.

Table 3. Exposure Hazard Ratios (HRs) for Development of Diabetes in Patients Receiving Atypical Antipsychotics Compared
With Patients Receiving Conventional Antipsychotics
Cases Controls
(N = 920), (N = 5258), Model 1,a Model 2,b Model 3,c
Variable N (%) N (%) HR (95% CI) HR (95% CI) HR (95% CI)
Use of medication
Atypical antipsychotics
Olanzapine 186 (20.22) 258 (4.91) 4.032 (3.363 to 4.834) 3.889 (3.238 to 4.670) 3.188 (2.650 to 3.834)
Quetiapine 79 (8.59) 166 (3.16) 2.197 (1.703 to 2.836) 2.121 (1.641 to 2.741) 1.824 (1.413 to 2.357)
Risperidone 130 (14.13) 186 (3.54) 3.524 (2.864 to 4.337) 3.409 (2.767 to 4.201) 3.403 (2.757 to 4.199)
Ziprasidone 9 (0.98) 11 (0.21) 1.237 (0.614 to 2.491) 1.279 (0.636 to 2.571) 1.685 (0.844 to 3.365)
Clozapine 2 (0.22) 3 (0.06) 6.217 (1.525 to 25.338) 5.313 (1.285 to 21.967) 7.003 (1.698 to 28.877)
Lithium 177 (19.24) 666 (12.67) 1.034 (0.867 to 1.233) 1.077 (0.902 to 1.287) 1.077 (0.900 to 1.287)
Anticonvulsants 395 (42.93) 1355 (25.77) 1.414 (1.192 to 1.677) 1.399 (1.176 to 1.664) 1.359 (1.139 to 1.621)
Antidepressants 436 (47.39) 1912 (36.36) 0.832 (0.707 to 0.978) 0.80 (0.681 to 0.948) 0.820 (0.694 to 0.969)
Conventional antipsychoticsd 314 (34.13) 1005 (19.11) 1.000 1.000 1.000
Concomitant drugs
β-Blocker 128 (13.91) 408 (7.76) 1.339 (1.098 to 1.634) 1.327 (1.088 to 1.620) 1.025 (0.839 to 1.252)
α-Blocker 29 (3.15) 45 (0.86) 1.760 (1.175 to 2.634) 1.785 (1.192 to 2.674) 1.012 (0.678 to 1.511)
Corticosteroid 149 (16.20) 593 (11.28) 1.120 (0.932 to 1.345) 1.093 (0.910 to 1.314) 0.941 (0.778 to 1.139)
Thiazide diuretic 67 (7.28) 134 (2.55) 1.877 (1.444 to 2.440) 1.886 (1.449 to 2.454) 1.249 (0.959 to 1.627)
Oral contraceptive 17 (1.85) 101 (1.92) 0.707 (0.426 to 1.174) 0.677 (0.406 to 1.130) 0.750 (0.451 to 1.248)
Valproic acid 7 (0.76) 29 (0.55) 1.181 (0.557 to 2.501) 1.179 (0.557 to 2.497) 1.172 (0.554 to 2.482)
Phenytoin 4 (0.43) 24 (0.46) 0.373 (0.137 to 1.013) 0.364 (0.133 to 1.001) 0.345 (0.126 to 0.946)
Psychiatric comorbidities
Alcohol abuse 81 (8.80) 325 (6.18) … 1.180 (0.922 to 1.510) 1.258 (0.984 to 1.609)
Substance abuse disorder 58 (6.30) 240 (4.56) … 1.082 (0.808 to 1.449) 1.112 (0.831 to 1.489)
Anxiety disorder 415 (45.11) 1916 (36.44) … 1.211 (1.057 to 1.387) 1.050 (0.914 to 1.206)
Impulse-control disorder 26 (2.83) 65 (1.24) … 1.744 (1.153 to 2.638) 1.634 (1.080 to 2.470)
Personality disorder 66 (7.17) 215 (4.09) … 1.261 (0.971 to 1.637) 1.200 (0.925 to 1.557)
Medical comorbidities
Hypertension 451 (49.02) 1009 (19.19) … … 2.741 (2.343 to 3.217)
Obesity 203 (22.07) 331 (6.30) … … 2.244 (1.897 to 2.656)
Arthritis 48 (5.22) 152 (2.89) … … 1.155 (0.851 to 1.568)
COPD 76 (8.26) 182 (3.46) … … 1.201 (0.933 to 1.546)
CVD 65 (7.07) 124 (2.36) … … 1.467 (1.118 to 1.925)
CHD 21 (2.28) 19 (0.36) … … 2.558 (1.616 to 4.048)
Dyslipidemia 28 (3.04) 58 (1.10) … … 2.703 (1.825 to 4.005)
a
Model for age, sex, bipolar follow-up months, and use of medications; χ2 = 620.90, p < .0001.
b
Model for age, sex, bipolar follow-up months, use of medications, and psychiatric comorbidities; χ2 = 643.82, p < .0001.
c
Model for age, sex, bipolar follow-up months, use of medications, and psychiatric and medical comorbidities; χ2 = 987.54, p < .0001.
d
HR = 1.000, because use of conventional antipsychotics was considered as the reference group.
Abbreviations: CHD = coronary heart disease, CI = confidence interval, COPD = chronic obstructive pulmonary disease, CVD = cerebral vascular
disease.

the present study of bipolar patients are similar or com- In addition to antipsychotic use, the present study in-
parable. For example, patients with schizophrenia had dicates that the risk of developing diabetes is also as-
the risk of developing diabetes associated with clozapine sociated with a patient’s comorbidity, especially obesity
(HR = 7.4–8.4),24,35–37 olanzapine (HR = 1.2–5.8),7,21–23 (HR = 2.2, 95% CI = 1.9 to 2.7), hypertension (HR = 2.7,
and risperidone (HR = 1.1–2.2),7,21–24 compared to the risk 95% CI = 2.3 to 3.2), CVD (HR = 1.5, 95% CI = 1.1 to
among bipolar patients for clozapine (HR = 7.0), olanza- 1.9), CHD (HR = 2.6, 95% CI = 1.6 to 4.0), and dyslipi-
pine (HR = 3.2), and risperidone (HR = 3.4) reported in demia (HR = 2.7, 95% CI = 1.8 to 4.0) (Table 3). As
Table 3. Our results indicated the risk of developing the literature indicates, some antipsychotics like olan-
diabetes is statistically significant for bipolar patients zapine, clozapine, and risperidone are associated with
taking clozapine, olanzapine, risperidone, and quetiapine weight gain,5,38,39 hyperlipidemia, and hypertriglyceride-
antipsychotics after controlling for comorbidities, per- mia, which are independent risk factors for heart dis-
sonal risk factors, and concomitant drug use. The hazard ease.7,8,40,41 It is likely that incident diabetes was associ-
ratio associated with ziprasidone was large (HR = 4.6) ated with metabolic syndrome, as indicated by higher
without controlling for comorbidities; then it became HRs for obesity, hypertension, CVD, CHD, and dyslipi-
smaller (HR = 1.7) and not statistically significant after demia in this study. This study also suggested that patients
controlling for comorbidities. This indicated that comor- with impulse-control disorder or anxiety disorder had
bidities are critical covariates for assessing the risk of higher risk for diabetes. It is possible that patients with
drug-induced diabetes. impulse-control disorder or anxiety disorder might have

1060
1059 © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE
J Clin PRESS
Psychiatry 67:7, July ,2006
INC.
 Diabetes Risk in Bipolar Patients

less healthy lifestyles, less medication compliance, or doxazosin (Cardura and others), fluphenazine (Prolixin and others),
glipizide (Glucotrol and others), glyburide (Diabeta, Micronase,
poorer access to health care services.42,43 and others), hydrocortisone (Hydrocortone, Cortef, and others), lamo-
Our study has several limitations. Drug use was in- trigine (Lamictal), lithium (Lithobid, Eskalith, and others), loxapine
ferred from automated pharmacy claims data. Because of (Loxitane and others), metformin (Riomet, Fortamet, and others),
methylprednisolone (Medrol, A-Methapred, and others), miglitol
the retrospective nature of a claims database review, it (Glyset), molindone (Moban), nateglinide (Starlix), olanzapine
is not possible to review the direct information on the (Zyprexa), phenytoin (Dilantin, Phenytek, and others), pimozide
severity of bipolar disorder, socioeconomic class, lipid (Orap), pioglitazone (Actos), polythiazide (Renese), prazosin
(Minipress and others), quetiapine (Seroquel), repaglinide (Prandin),
profiles, fasting glucose, or body mass index related to risperidone (Risperdal), rosiglitazone (Avandia), terazosin (Hytrin and
weight gain. We were unable to adjust the patients’ ethnic- others), thiothixene (Navane and others), trifluoperazine (Stelazine and
ity because the variable was missing when PharMetrics others), valproic acid (Depakene and others), ziprasidone (Geodon).
(data vendor) collected the medical claim data from dif-
ferent managed care organizations. It is unclear whether REFERENCES
different medications prescribed before the study period
1. Keck PE Jr, McElroy SL. Treatment of bipolar disorder. In: Schatzberg
might be partially limited to the increased risk of diabetes. AF, Nemeroff CB, eds. American Psychiatric Publishing Textbook of
Because clinicians may have prescribed one drug over the Psychopharmacology, 3rd edition. Washington, DC: American Psychiatric
other based on the different moods of bipolar patients, we Publishing, Inc; 2004:627–638
2. Keck PE Jr, Marcus R, Tourkodimitris S, et al. A placebo-controlled,
attempted to reduce this potentially confounding bias by double-blind study of the efficacy and safety of aripiprazole in patients
adjusting for known concomitant drugs and comorbidi- with acute bipolar mania. Am J Psychiatry 2003;160:1651–1658
ties. We also included comorbidities of dyslipidemia and 3. Guo JJ, Keck PE Jr, Li H, et al. Evolution of mood stabilizer utilization
among patients with bipolar disorder in a managed care Medicaid popula-
CHD as a rough proxy for controlling high risk patients tion [abstract]. Presented at the annual meeting of the American College
for diabetes. It is possible that this study underestimated of Clinical Pharmacists; October 24–26, 2004; Dallas, Tex
the prevalence of diabetes due to the limited time window 4. Allison DE, Mentore JL, Hero M, et al. Antipsychotic-induced weight
gain: a comprehensive research synthesis. Am J Psychiatry 1999;156:
and changes of managed care enrollment and other mental 1686–1696
services not billed to patients’ managed care organiza- 5. Keck PE Jr, McElroy SL. Bipolar disorder, obesity, and pharmaco-
tions. Comorbid conditions were identified by diagnostic therapy-associated weight gain. J Clin Psychiatry 2003;64:1426–1435
6. Meyer JM. A retrospective comparison of lipid, glucose, and weight
codes without considering the combination of medica- changes at one year between olanzapine- and risperidone-treated inpa-
tions for obesity, hypertension, CVD, and other diseases. tients. Biol Psychiatry 2001;49:536
Despite the above limitations, the present study is a 7. Koro CE, Fedder DO, L’Italien GJ, et al. Assessment of independent
effect of olanzapine and risperidone on risk of diabetes among patients
contribution to the limited literature about diabetes risk in with schizophrenia: population based nested case-control study. BMJ
bipolar patients and provides useful information for dis- 2002;325:243–248
8. Dunner DL. Safety and tolerability of emerging pharmacological treat-
ease management strategies in terms of selection of mood ments for bipolar disorder. Bipolar Disord 2005;7:307–325
stabilizers and consideration of relevant comorbidities for 9. Newcomer JW. Second-generation (atypical) antipsychotics and meta-
patients with bipolar disorder. Atypical antipsychotics bolic effects: a comprehensive literature review. CNS Drugs 2005;
19(suppl 1):1–93
provide great benefit to a wide variety of people with psy- 10. Fertig MK, Brooks VG, Shelton PS, et al. Hyperglycemia associated
chiatric disorders and have one constellation of adverse with olanzapine. J Clin Psychiatry 1998;59:687–689
effects related to increased risk of obesity, diabetes, and 11. Lindenmayer JP, Patel R. Olanzapine-induced ketoacidosis with diabetes
mellitus [letter]. Am J Psychiatry 1999;156:1471
dyslipidemia.9,34 12. Rigalleau V, Gatta B, Bonnaud S, et al. Diabetes as a result of atypical
In conclusion, some atypical antipsychotics like cloza- antipsychotic drugs: a report of three cases. Diabet Med 2000;17:484–486
pine, olanzapine, risperidone, and quetiapine are consis- 13. Biswas PN, Wilton LV, Pearce GL, et al. The pharmacovigilance of olan-
zapine: results of a post-marketing surveillance study on 8858 patients
tently associated with a clinically important increased risk in England. J Psychopharmacol 2001;15:265–271
of diabetes mellitus in bipolar patients after adjustment 14. Roefaro J, Mukherjee SM. Olanzapine-induced hyperglycemic
for relevant risk factors. Metabolic complications are a nonketonic coma. Ann Pharmacother 2001;35:300–302
15. Colli A, Cocciolo M, Francobandiera F, et al. Diabetic ketoacidosis
major issue for patients receiving antipsychotic therapy. associated with clozapine treatment. Diabetes Care 1999;22:176–177
The choice of atypical antipsychotics for a specific bi- 16. Mithat B, Alpaslan T, Bulent C, et al. Risperidone-associated transient
polar patient should consider the risk-benefit of antipsy- diabetic ketoacidosis and diabetes mellitus type 1 in a patient treated with
valproate and lithium. Pharmacopsychiatry 2005;38:105–106
chotics and depends on relevant high-risk comorbid con- 17. Laghate VD, Gupta SB. Acute pancreatitis and diabetic ketoacidosis in
ditions. Thus, the propensity of an antipsychotic to induce non-diabetic person while on treatment with sodium valproate, chlorpro-
diabetes is a critical consideration when selecting an mazine and haloperidol. J Assoc Physicians India 2004;52:257–258
18. Tavakoli SA, Arguisola MS. Diabetic ketoacidosis in a patient treated
agent for patients with bipolar disorder. with olanzapine, valproic acid, and venlafaxine. South Med J 2003;96:
729–730
Drug names: acarbose (Precose), aripiprazole (Abilify), atenolol 19. Courvoisie HE, Cooke DW, Riddle MA. Olanzapine-induced diabetes
(Tenormin and others), betaxolol (Kerlone, Betoptic, and others), in a seven-year-old boy. J Child Adolesc Psychopharmacol 2004;14:
bisoprolol (Zebeta and others), carbamazepine (Carbatrol, Equetro, 612–616
and others), chlorothiazide (Diuril and others), chlorpromazine (Thor- 20. Domon SE, Cargile CS. Quetiapine-associated hyperglycemia and hyper-
azine, Sonazine, and others), chlorthalidone (Thalitone and others), triglyceridemia [letter]. J Am Acad Child Adolesc Psychiatry 2002;41:
clozapine (Clozaril, FazaClo, and others), divalproex (Depakote), 495–496

J©Clin
COPYRIGHT 2006
Psychiatry PHYSICIANS
67:7, July 2006 POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. 1061
1060
Guo et al.

21. Miller EA, Leslie DL, Rosenheck RA. Incidence of new-onset diabetes 33. Chae BJ, Kang BJ. The effect of clozapine on blood glucose metabolism.
mellitus among patients receiving atypical neuroleptics in the treatment Hum Psychopharmacol 2001;16:265–271
of mental illness: evidence from a privately insured population. J Nerv 34. American Diabetes Association (ADA) and American Psychiatric
Ment Dis 2005;193:387–395 Association. Consensus development conference on antipsychotic
22. Leslie DL, Rosenheck RA. Incidence of newly diagnosed diabetes drugs and obesity and diabetes. Diabetes Care 2004;27:596–601
attributable to atypical antipsychotics medications. Am J Psychiatry 35. Giantrancesco FD, Grogg AL, Mahmoud RA, et al. Differential effects
2004;161:1709–1711 of risperidone, olanzapine, clozapine, and conventional antipsychotics
23. Kornegay CJ, Vasilakis-Scaramozza C, Jick H. Incident diabetes associ- on type 2 diabetes: findings from a large health plan database. J Clin
ated with antipsychotic use in the United Kingdom General Practice Psychiatry 2002;63:920–930
Research database. J Clin Psychiatry 2002;63:758–762 36. Lambert B, Chou CH, Chang KY, et al. Assessing the risk of antipsy-
24. Hedenmalm K, Hagg S, Stahl M, et al. Glucose intolerance with atypical chotic-induced type II diabetes among schizophrenics: a matched
antipsychotics. Drug Saf 2002;25:1107–1116 case-control study [abstract]. Eur Neuropsychopharmacol 2002;
25. Wynn V. Effects of duration of low-dose contraceptive administration 12(suppl 3):307–308
on carbohydrate metabolism. Am J Obstet Gynecol 1982;142:739–746 37. Citrome L, Jaffe A, Levine J, et al. Antipsychotic medication treatment
26. Isojarvi JI, Rattya J, Myllyla VV, et al. Valproate, lamotrigine, and and new prescriptions for insulin and oral hypoglycemics [abstract].
insulin-mediated risks in women with epilepsy. Ann Neurol 1998;43: Eur Neuropsychopharmacol 2003;13(suppl 4):S306
446–451 38. McIntyre RS, Mancini DA, Basile VS, et al. Antipsychotic-induced
27. PharMetrics. PharMetrics Patient-Centric Database: a large weight gain: bipolar disorder and leptin. J Clin Psychopharmacol 2003;
medical claims data source in US. Watertown, Mass. Available at: 23:323–327
http://www.pharmetrics.com. Accessed Aug 1, 2004 39. Sachs GS, Guille C. Weight gain associated with use of psychotropic
28. Carson JL, Strom BL, Soper KA, et al. The association of nonsteroidal medications. J Clin Psychiatry 1999;60(suppl 21):16–19
anti-inflammatory drugs with upper gastrointestinal tract bleeding. Arch 40. Koller EA, Cross JT, Doraiswamy PM, et al. Risperidone-associated
Intern Med 1987;147:85–88 diabetes mellitus: a pharmacovigilance study. Pharmacotherapy 2003;
29. Guo JJ, Curkendall S, Jones JK, et al. Impact of cisapride label change 23:735–744
on codispensing of contraindicated medications. J Pharmacoepidemiol 41. Jeppesen J, Hein HO, Suadicani P, et al. Triglyceride concentration and
Drug Saf 2003;12:295–301 ischemic heart disease: an eight-year follow-up in the Copenhagen male
30. Guo JJ, Jang R, Louder A, et al. Acute pancreatitis associated with study. Circulation 1998;97:1029–1036
different combination therapies in patients infected with human immuno- 42. Fleck DE, Keck PE Jr, Corey KB, et al. Factors associated with medica-
deficiency virus. Pharmacotherapy 2005;25:1044–1054 tion adherence in African American and white patients with bipolar
31. Taylor DM, McAskill R. Atypical antipsychotics and weight gain: disorder. J Clin Psychiatry 2005;66:646–652
a systematic review. Acta Psychiatr Scand 2000;101:416–432 43. Keck PE Jr, McElroy SL, Strakowski SM, et al. Factors associated with
32. Wirshing DA, Boyd JA, Meng LR, et al. The effects of novel antipsy- maintenance antipsychotic treatment of patients with bipolar disorder.
chotics on glucose and lipid levels. J Clin Psychiatry 2002;63:856–865 J Clin Psychiatry 1996;57:147–151

The CME Institute of Physicians Postgraduate Press, Inc. is accredited by the Accreditation Council for Continuing Medical
Education to provide continuing medical education for physicians. This activity was independently developed by the
CME Institute of Physicians Postgraduate Press, Inc. and the American Society of Clinical Psychopharmacology.

1062
1061 © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE PRESS, INC. © COPYRIGHT 2006 PHYSICIANS POSTGRADUATE
J Clin PRESS
Psychiatry 67:7, July ,2006
INC.