Frequency of new-onset diabetes mellitus and use of antipsychotic drugs among Central Texas veterans.

Abstract

To determine whether the frequency of new-onset diabetes mellitus differs between patients taking atypical antipsychotic agents and those taking typical agents, whether the frequency of new-onset diabetes differs among those taking the atypical antipsychotic agents, and what clinical and demographic factors influence the occurrence of new-onset diabetes.
Retrospective analysis.
Central Texas Veterans Health Care System.
Continuously enrolled adult (> or = 18 yrs) patients with no previous (6 mo) antipsychotic use and no history (previous 1 yr) of diabetes.
Data from the Central Texas Veterans Health Care System were extracted from September 1995-November 2002. Clinical and demographic factors used in the analysis were antipsychotic agent taken, body mass index, diabetes-related risk factors, type of mental health comorbidity, age, sex, and race. Among those who met the inclusion criteria (3469 patients), chi2 analyses revealed no significant difference in the frequency of diabetes between the typical and atypical groups (p=0.5553) or among those taking atypical agents (p=0.6520). Multivariate logistic regression (1587 patients) revealed that increasing age (odds ratio [OR] 1.213, 95% confidence interval [CI] 1.016-1.447, p=0.0324), nonwhite race (OR 1.761, 95% CI 1.174-2.640, p=0.0062), and hyperlipidemia (OR 1.606, 95% CI 1.064-2.425, p=0.0242) were significantly related to new-onset diabetes.
Among veterans taking antipsychotic agents, no difference was noted in the frequency of diabetes between patients who took typical agents and those who took atypical agents. After controlling for demographic and clinical variables, still no significant difference was noted among the agents. The main factors (increasing age, nonwhite race, and hyperlipidemia) related to new-onset diabetes were those that are typically associated with the disease.

Full text

Page 1

Frequency of New-Onset Diabetes Mellitus and Use of
Antipsychotic Drugs Among Central Texas Veterans
Jamie C. Barner, Ph.D., Jason Worchel, M.D., and Min Yang, M.S.
Study Objectives. To determine whether the frequency of new-onset diabetes
mellitus differs between patients taking atypical antipsychotic agents and
those taking typical agents, whether the frequency of new-onset diabetes
differs among those taking the atypical antipsychotic agents, and what
clinical and demographic factors influence the occurrence of new-onset
diabetes.
Design. Retrospective analysis.
Setting. Central Texas Veterans Health Care System.
Patients. Continuously enrolled adult (≥ 18 yrs) patients with no previous (6
mo) antipsychotic use and no history (previous 1 yr) of diabetes.
Measurements and Main Results. Data from the Central Texas Veterans
Health Care System were extracted from September 1995–November 2002.
Clinical and demographic factors used in the analysis were antipsychotic
agent taken, body mass index, diabetes-related risk factors, type of mental
health comorbidity, age, sex, and race. Among those who met the inclusion
criteria (3469 patients), ?2analyses revealed no significant difference in the
frequency of diabetes between the typical and atypical groups (p=0.5553)
or among those taking atypical agents (p=0.6520). Multivariate logistic
regression (1587 patients) revealed that increasing age (odds ratio [OR]
1.213, 95% confidence interval [CI] 1.016–1.447, p=0.0324), nonwhite
race (OR 1.761, 95% CI 1.174–2.640, p=0.0062), and hyperlipidemia (OR
1.606, 95% CI 1.064–2.425, p=0.0242) were significantly related to new-
onset diabetes.
Conclusions. Among veterans taking antipsychotic agents, no difference was
noted in the frequency of diabetes between patients who took typical agents
and those who took atypical agents. After controlling for demographic and
clinical variables, still no significant difference was noted among the agents.
The main factors (increasing age, nonwhite race, and hyperlipidemia)
related to new-onset diabetes were those that are typically associated with
the disease.
Key Words: antipsychotics, diabetes mellitus, risk factors.
(Pharmacotherapy 2004;24(11):1529–1538)
Atypical antipsychotic agents have been well
received because of their increased efficacy and
decreased rate of extrapyramidal symptoms
compared with those of typical antipsychotics.
Although the atypical agents have some distinct
advantages, they also have some disadvantages.
Some of the most recently noted issues involve
weight gain,1–6elevation in triglyceride and
cholesterol levels,7–10and new-onset diabetes
mellitus and diabetes-related complications.11–34
The literature surrounding the issue of new-
onset diabetes and its association with anti-
psychotic therapy is primarily populated with
case reports11–30that date back to 1994 and a few
small clinical trials.31–34Of the atypical anti-
psychotic agents, olanzapine and clozapine have

Page 2

PHARMACOTHERAPY Volume 24, Number 11, 2004
been mentioned more times regarding new-onset
diabetes or diabetes-related problems compared
with risperidone and quetiapine. Most of the
case reports were in men and in people of
African-American descent. Most of the case
reports did not indicate a family or personal
history of diabetes, although it is not clear if
these variables were validly assessed. Some
reports involved diabetic ketoacidosis, and other
cases reported that discontinuation of the
atypical antipsychotic resulted in normalization
of blood glucose levels.
A number of researchers have conducted
several large-scale studies in the last few years to
substantiate the claims of treatment-emergent
diabetes or exacerbation of preexistent diabetes
in patients newly prescribed a particular
antipsychotic.35–46However, the results from
these large-scale studies are far from conclusive.
The mechanism for the potential link between
antipsychotic agents and diabetes is not well
understood. It has been proposed that the
increase in weight gain through stimulation of
serotonin, histamine, dopamine, prolactin, and
leptin receptors could potentiate glucose
dysregulation and subsequently promote new-
onset diabetes mellitus.5, 30Another mechanism
could involve the relationship among triglyceride
levels, antipsychotic agents, and diabetes.32, 33
In one study, the authors assessed whether
patients who switched to ziprasidone experienced
significant changes in body mass index (BMI)
and glucose, cholesterol, and triglyceride levels.31
The study showed no significant changes in BMI
or glucose level, but a significant improvement in
cholesterol and triglyceride levels.
Another group conducted a retrospective study
to assess differences in weight, glucose level,
cholesterol level, and blood pressure in patients
From the Pharmacy Administration Division and Center
for Pharmacoeconomic Studies, University of Texas at
Austin (Dr. Barner and Ms. Yang), and the Veterans
Administration, Central Texas Veterans Health Care System
(Dr. Worchel), Austin, Texas.
Supported by Eli Lilly and Company, Indianapolis,
Indiana.
Presented in part at the American Psychiatric Association
55th Institute on Psychiatric Services, Boston,
Massachusetts, October 31, 2003, and at the annual
meetings of the American College of Clinical Pharmacy,
Atlanta, Georgia, November 4, 2003, and the American
Society of Health-System Pharmacists, New Orleans,
Louisiana, December 11, 2003.
Address reprint requests to Jamie C. Barner, Ph.D.,
Pharmacy Administration Division and Center for
Pharmacoeconomic Studies, University of Texas at Austin,
PHAR-PHARMACY ADMIN, 1 University Station A1930,
Austin, TX 78712-0127; e-mail: jbarner@mail.utexas.edu.
who received olanzapine and haloperidol.32
Significant weight increases were found in
patients receiving olanzapine compared with
those receiving haloperidol. Although the
olanzapine group had significantly higher
glucose and cholesterol levels (when compared
with those of the haloperidol group), no
significant differences were noted in the
frequency of increased glucose level, cholesterol
level, or blood pressure.
Another group of investigators assessed the
frequency of new-onset diabetes among patients
treated with clozapine.33Diabetes mellitus was
diagnosed in more than one third (36.6%) of the
patients. The development of diabetes was
significantly associated with increased
triglyceride levels. The study also found that
there was a significant increase in weight and that
the weight gain was significantly associated with
increased serum cholesterol and triglyceride
levels.
Results of two of the above studies32, 33suggest
that weight gain may not be the direct link to
new-onset diabetes. In a Swedish study, the
authors assessed the prevalence of diabetes and
impaired glucose tolerance in patients taking
clozapine versus those taking depot neuroleptics.34
Although the study found that the prevalence of
diabetes mellitus or impaired glucose tolerance
was not statistically significant between the two
groups, the percentages of patients who
developed diabetes mellitus and impaired glucose
tolerance were higher in the clozapine group
(12% clozapine vs 6% depot neuroleptics and
10% clozapine vs 3% depot neuroleptics,
respectively).
A brief review of the literature shows that the
issue of new-onset diabetes among patients
taking antipsychotic agents is not well understood.
Interpretation of the numerous case reports is
difficult because of the lack of a rigorous method
to systematically combine the data from the
cases. Many of the clinical trials were conducted
with small sample sizes and primarily focused on
one agent or comparisons of two agents. Most of
the large database studies did not control for
known factors related to diabetes such as weight
gain, hypertension, or hyperlipidemia. In our
study, we incorporated these covariates, as well as
others, in a multi-variate analysis to determine
the relationship between antipsychotic therapy
and new-onset diabetes.
The following three objectives were the focus
of this study: to determine whether the frequency
of new-onset diabetes differed between those
1530

Page 3

ANTIPSYCHOTIC THERAPY AND INCIDENT DIABETES MELLITUS Barner et al
taking atypical agents and those taking typical
agents; to determine whether the frequency of
new-onset diabetes differed among those taking
the atypical antipsychotic agents; and to
determine what factors (e.g., antipsychotic agent
used, BMI, diabetes-related risk factors, and
demographics) influenced the occurrence of
new-onset diabetes.
Methods
Data Source
We used data from the Central Texas Veterans
Health Care System (CTVHCS). One distinct
aspect of CTVHCS is the computerized patient
record system, which electronically captures
most patient and clinical information from the
medical chart.
Inclusion Criteria
Individual patient level claims records were
extracted and analyzed for patients who were
aged 18 years or older; had not received a pre-
scription for an atypical or typical antipsychotic
agent 6 months before the dispensing of an
atypical or typical antipsychotic agent; had no
previous use of an antidiabetic drug or diagnosis
of diabetes for 1 year before a prescription for a
typical or atypical antipsychotic; and were
continuously enrolled for 12 months before and
after the date of receiving an atypical or typical
antipsychotic agent.
Study Variables
The dependent variable for all analyses was
whether or not the subject developed diabetes.
This was operationalized by a diagnosis of
diabetes (International Classification of Diseases,
Ninth Revision [ICD-9] code 250.xx), blood
glucose levels greater than 200 mg/dl, and/or use
of an antidiabetic drug between 8 and 365 days
after the index date (i.e., the date of the first
prescription for an antipsychotic agent).
The primary independent variable was the type
of antipsychotic agent that the patient was taking
initially—typical agents or the individual atypical
agents. The atypical agents were olanzapine,
quetiapine, risperidone, clozapine, and ziprasidone.
The control independent variables (and
possible answers) were the following: diabetes-
related risk factors (yes or no), which included
change to higher BMI category, previous
hyperlipidemia, and change in hypertension
status; persistence (total number of days without
a 15-day gap); type of mental health comorbidity
(yes or no), which included bipolar disorder,
depression, schizophrenia, and substance abuse;
age group (18–39 yrs, 40–49 yrs, 50–59 yrs,
60–69 yrs, or 70 yrs or older); sex (male or
female); and race (white or nonwhite).
The BMI was calculated for the weight closest
(within a 6-mo time frame) to the first use of the
antipsychotic agent and for the weight closest
(within a 6-mo time frame) to the last use of an
antipsychotic agent. Based on their BMI, subjects
were assigned to one of six BMI categories
established by the National Heart, Lung, and
Blood Institute.47A dichotomous variable was
used to indicate whether or not the subject
moved to a higher BMI category status from the
first to the last use of the antipsychotic agent.
Previous hyperlipidemia was defined as a
cholesterol level of 200 mg/dl or above, a low-
density lipoprotein cholesterol level of 130 mg/dl
or above, or a triglyceride level of 150 mg/dl or
above within 6 months before the first
antipsychotic used. Because of limited data
within the study time frame, lipid level changes
could not be calculated.
Blood pressure measurements were extracted
closest to the first and last use of the anti-
psychotic within a 6-month window. Subjects
were categorized into one of six blood pressure
groups established by the National Heart, Lung,
and Blood Institute.48A dichotomous variable
was used to indicate whether or not the subject
moved to a higher blood pressure category.
Persistence was calculated by summing the
total number of continuous days the patient took
an antipsychotic agent without a gap (i.e., a 15-
day lapse in therapy).
The study received institutional review board
approval from both the CTVHCS and the
University of Texas.
Data Collection and Analysis
Data from the CTVHCS record system was
extracted for the time frame of September 30,
1995–November 1, 2002. The follow-up time
period was 1 year after the index date. To
compare the frequency of diabetes between
patients who received atypical agents and those
who received typical agents, ?2analyses were
used. To compare the frequency of new-onset
diabetes among patients receiving atypical
antipsychotic agents while controlling for clinical
and patient-related variables, a logistic regression
1531

Page 4

PHARMACOTHERAPY Volume 24, Number 11, 2004
analysis was used. An a priori significance level
of 0.05 was used for all analyses.
Results
A total of 6735 patients were identified as
taking antipsychotics in the CTVHCS database.
One was excluded for being younger than 18
years, 1999 were excluded because of previous
antipsychotic use, 819 were excluded because of
previous diabetes, and 447 were excluded owing
to a less than 12-month enrollment period after
antipsychotic use. This resulted in a total of
3469 subjects meeting the inclusion criteria.
Most of the sample was male (94.3%) and white
(69.9%). Mean ± SD age was 59.4 ± 14.5 years,
with subjects aged 50–59 years and those aged 70
years or older constituting 34% and 28% of the
sample, respectively.
By using any of the three criteria mentioned in
Methods for defining the dependent variable
new-onset diabetes, the frequency of new-onset
diabetes was 7.1%. When using single categories
of elevated blood glucose levels, ICD-9 diagnosis,
or antidiabetic drug as the criterion for diabetes,
the frequency rates were 4.1%, 3.0%, and 2.1%,
respectively. The time to diabetes onset was
approximately 5 months (mean ± SD 151.9 ±
105.6 days).
More than 40% (44.3%) of the subjects were
taking atypical agents, with olanzapine (23.0%)
being prescribed most often, followed by
risperidone (16.2%). Among the typical agents,
haloperidol (20.0%) was most often prescribed.
For the atypical agents, the mean number of
persistent days ranged from 117–167 days
(3.9–5.6 mo), and for the typical agents the range
was 141–220 days (4.7–7.3 mo).
The subjects had various mental health and
other comorbidities. Of the 3469 subjects, 1461
(42.1%) did not have ICD-9 data. Of the 2008
subjects with documented ICD-9 data, the
diagnoses were substance abuse in 841 (41.9%)
patients, depression in 715 (35.6%), and bipolar
disorder in 689 (34.3%). Schizophrenia was
diagnosed in 681 (33.9%) subjects. On average,
the subjects had three (mean ± SD 2.5 ± 1.4)
mental health comorbidities.
Subjects were categorized into one of six BMI
groups established by the National Heart, Lung,
and Blood Institute.47Changes in BMI categories
from the first to the last use of an antipsychotic
agent were assessed. Approximately 36% and
34% (first BMI and last BMI, respectively) of the
subjects were considered to be of normal weight,
and approximately 38% (both first BMI and last
BMI) were categorized as overweight. Approxi-
mately 23% and 26% (first BMI and last BMI,
respectively) of the patients were categorized as
obese. Therefore, 61% and 64% (first BMI and
last BMI, respectively) of the patients were either
overweight or obese. Most subjects (88%) did
not change to a higher BMI category from first to
last antipsychotic use.
Thirty percent of the subjects met the criteria
for previous hyperlipidemia (cholesterol level
≥ 200 mg/dl, low-density lipoprotein cholesterol
≥ 130 mg/dl, or triglyceride level ≥ 150 mg/dl 6
mo before taking the first antipsychotic) when
using any of the three lipid level categories.49
Subjects were categorized into one of six hyper-
tension groups established by the National Heart,
Lung, and Blood Institute.48According to the
guidelines, 36% and 34% (first blood pressure
and last blood pressure, respectively) of the
subjects had hypertension. In most subjects
(72%), the hypertension status category did not
change.
Study Objectives
Tables 1–3 address our first objective, which
was to determine whether the frequency of new-
onset diabetes differed between those taking
typical agents and those taking atypical agents.
1532
Table 1. Frequency of New-Onset Diabetes Mellitus in the
Atypical and Typical Agent Groups
No. (%) of Patients
Atypical
Group
Variable(n=1537)
Diabetes 105 (6.8)a
No diabetes1432 (93.2)
Intent-to-treat methodology was used.
a?2=0.3479, p=0.5553.
Typical
Group
(n=1932)
142 (7.3)a
1790 (92.7)
Table 2. Frequency of New-Onset Diabetes Mellitus
Among the Atypical, Typical, and Both Agents Groups
No. (%) of Patients
Typical
Group
(n=992)
69 (7.0)b
923 (93.0)
Atypical
Group
(n=1390)
94 (6.8)b
1296 (93.2)
Botha
Groups
(n=1087)
84 (7.7)b
1003 (92.3)
Variable
Diabetes
No diabetes
aIndicates concomitant use of both atypical and typical agents or
switching between the two.
b?2=0.9160, p=0.6325.

Page 5

ANTIPSYCHOTIC THERAPY AND INCIDENT DIABETES MELLITUS Barner et al
The results show that the frequency of new-onset
diabetes was not significantly different
(p=0.5553) between the atypical group (6.8%)
and the typical group (7.3%).
Two additional analyses were performed to
further investigate this issue. Because of the
concomitant use of atypical and typical agents,
the analysis was run separating the data into
three groups: atypical, typical, and both. Table 2
shows that the results were similar to those of the
previous analysis: no significant difference in
new-onset diabetes among the three groups. The
third analysis (Table 3) collapses the atypical and
both categories into one group (atypical+). Once
again, the results show that the frequency of
new-onset diabetes was not significantly different
between the typical and atypical groups.
Table 4 addresses our second objective, which
was to determine whether the frequency of new-
onset diabetes differed among those taking the
atypical antipsychotic agents (note that clozapine
and ziprasidone were dropped from the analyses
due to small sample sizes). The ?2results show
that there was no significant difference in
frequency of new-onset diabetes among the
atypical agents.
Tables 5 and 6 address our third objective,
which was to determine what factors—anti-
psychotic agent used, BMI category increase,
previous hyperlipidemia, increase in hyper-
tension status category, persistence, type of
mental health comorbidity, age, sex, and race—
influence the occurrence of new-onset diabetes.
Table 5 shows that increasing age, minority race,
and previous hyperlipidemia were the only
variables significantly related to new-onset
diabetes. To increase power, another logistic
regression analysis was run to include only
1533
Table 3. Frequency of New-Onset Diabetes Mellitus
Among the Atypical+ and Typical Agent Groups
No. (%) of Patients
Atypical+a
Group
(n=2477)
178 (7.2)b
2299 (92.8)
Typical
Group
(n=992)
69 (7.0)b
923 (93.0)
Variable
Diabetes
No diabetes
aIncludes patients taking atypical agents only, as well as those
taking both atypical and typical agents.
b?2=0.0569, p=0.8115.
Table 4. Frequency of New-Onset Diabetes Mellitus
Among Patients Taking Atypical Agents
No. (%) of Patients
Diabetes
42 (7.5)b
9 (5.8)b
51 (6.4)b
Atypical Agenta
Risperidone
Quetiapine
Olanzapine
aZiprasidone (2 patients) and clozapine (21 patients) were not
included because of the small sample sizes.
b?2=0.8554, p=0.6520.
No Diabetes
520 (92.5)
147 (94.2)
745 (93.6)
Table 5. Logistic Regression Analysis of Factors Related to New-Onset Diabetes Mellitus in
1587 Patients
95% Confidence
Variablea
Odds Ratio
Olanzapine0.976 0.594–1.605
Quetiapine1.149 0.531–2.485
Risperidone0.9260.544–1.579
Increasing age 1.2131.016–1.447
Nonwhite 1.7611.174–2.640
Female 0.7180.277–1.857
Persistent days 1.001 0.999–1.002
Comorbidity
Depression1.305 0.850–2.002
Substance abuse0.869 0.561–1.345
Bipolar disorder1.1920.772–1.840
Schizophrenia1.117 0.678–1.570
Body mass index 1.0320.477–1.581
Hypertension 0.7590.415–1.388
Previous hyperlipidemia 1.6061.064–2.425
aReference categories for the atypical agents, nonwhite, and female variables were the typical agents, white, and
male, respectively.
bModel ?2=20.00, p=0.1302.
cSignificance at p<0.05.
Wald
?2Valueb
0.0411
0.2330
0.2283
4.5773
7.4881
0.4679
1.4273
Interval p Value
0.8394
0.6293
0.6328
0.0324c
0.0062c
0.4940
0.2322
1.4837
0.3983
0.6262
0.0215
0.2137
0.7994
5.0804
0.2232
0.5280
0.4288
0.8834
0.6439
0.3713
0.0242c

Page 6

PHARMACOTHERAPY Volume 24, Number 11, 2004
demographics (age, race, and sex) and mental
health comorbidities. This analysis increased the
sample size from 1587 to 3170 subjects. Table 6
shows that once again, increasing age and
minority race were significantly related to new-
onset diabetes.
Discussion
Our study showed that among central Texas
veterans who used antipsychotics, the overall
frequency of diabetes was 7.1%. These results
compare with those of another study that found
that the frequency of diabetes was 6.3% among
Ohio veterans.38Our study may have had a
higher rate because we used a more inclusive
definition of diabetes, that is to include not only
ICD-9 diagnoses and antidiabetic drugs, but also
elevated glucose levels. In addition, our sample
population was composed of approximately 30%
minorities, which may be higher than some of
the other populations studied. Diabetes is more
prevalent among minorities than nonminorities.
Another group found that the diabetes
incidence rate for all of the patients with
schizophrenia who were taking antipsychotics
was 4.4/1000 person-years.42In another study,
the incidence rates in the general population, in
patients taking typical antipsychotics, and in
patients taking atypical antipsychotics were 15.7,
84, and 67/1000 patient-years, respectively.35
Our study showed no difference in the
frequency of diabetes between the atypical and
typical agent groups or among those taking
atypical agents in both the unadjusted and
adjusted analyses. One group of authorsreported
that the overall diabetes rates among patients
with schizophrenia who were treated with
clozapine versus typical antipsychotics were 4.0%
and 3.4%, respectively, which were not
statistically significantly different.43Another
group found no increased risk of diabetes when
comparing clozapine with typical agents in a
Medicaid population.45In another study, the
authors found no difference in incident diabetes
between the atypical and typical antipsychotic
cohorts, but both groups were significantly
associated with increased risk of diabetes when
compared with the general population.35
Another group found no difference in the
frequency of diabetes between atypical and
typical agents.46
In contrast to these study results, other large
database studies36, 39, 40, 42, 44have found significant
differences in the frequency of new-onset
diabetes between atypical agents and typical
agents and/or among atypical agents. However,
one study found atypical agents to have a
significant increased risk of diabetes compared
with typical agents, but no difference was noted
in new-onset diabetes among the atypical
agents.37In other studies comparing the atypical
agents, olanzapine was most often associated
with the increased risk of diabetes,36, 39, 40, 42, 44
whereas risperidone was associated with new-
onset diabetes in two studies.35, 46In one of those
studies, the authors compared haloperidol use to
individual atypical antipsychotics and found
risperidone to have a significantly increased risk
of diabetes.35In the other study, the authors
compared the occurrence of diabetes among
1534
Table 6. Logistic Regression Analysis of Factors Related to New-Onset Diabetes Mellitus
That Included Only Demographics and Comorbidities in 3170 Patients
95% Confidence
Variablea
Odds Ratio
Olanzapine0.941 0.639–1.384
Quetiapine1.034 0.508–2.106
Risperidone0.9630.641–1.447
Increasing age 1.2741.130–1.435
Nonwhite 1.689 1.274–2.239
Female 0.797 0.395–1.605
Comorbidity
Depression 1.3020.895–1.893
Substance abuse0.929 0.641–1.346
Bipolar disorder1.321 0.907–1.925
Schizophrenia0.953 0.670–1.355
aReference categories for the atypical agents, nonwhite, and female variables were the typical agents, white,
and male, respectively.
bModel ?2=29.57, p=.0010.
cSignificance at p<0.05.
Wald
?2Valueb
0.0901
0.0370
0.0190
15.7806
13.2741
0.4025
Intervalp Value
0.7641
0.8474
0.8903
<0.0001c
0.0003c
0.5258
1.9060
0.1505
2.1033
0.0722
0.1674
0.5981
0.1470
0.7881

Page 7

ANTIPSYCHOTIC THERAPY AND INCIDENT DIABETES MELLITUS Barner et al
patients with schizophrenia and found those
taking risperidone to have a higher risk compared
with those taking typical agents or olanzapine.46
Other authors found that both olanzapine and
clozapine, as well as selected typical antipsychotic
agents, were associated with an increased risk of
1535
Table 7. Comparison of Methodologies of Retrospective Database Studies Examining Antipsychotic Use and New-Onset
Diabetes Mellitus
Setting orStudy SampleStudy Sample
Database Time FramePopulationDesignSize
Managed care35
1998–2000 All AP usersRC 58,751
and general(AP users)
population 5,816,473
(general
population)
Quebec public1997–1999 All risperidoneRC 33,946
health plan36
and olanzapine
users
Demographic
Covariates
Age, sex
Drugs
All APs
Clinical Covariates
AP exposure
duration,
AP dosage
Risperidone,
olanzapine
AP exposure duration,
concomitant
haloperidol,
psychiatric
diagnosis
Use of drugs
associated with
glucose
intolerance
AP exposure duration,
psychiatric diagnoses,
use of lithium and
valproic acid,
concomitant AP use
AP exposure duration,
concurrent AP use,
AP dosage,
psychotropic use,
psychiatric
diagnoses
AP exposure duration,
concurrent AP use,
?-blocker use,
prior weight gain
treatment,
psychotropic use,
psychiatric diagnoses
AP exposure duration,
concurrent AP use,
body mass index,
smoking, use of drugs
associated with
glucose intolerance,
alcoholism,
cardiovascular history,
psychiatric diagnoses
AP exposure duration,
use of drugs
associated with
glucose intolerance
Age, sex
VA37
1999–2001 Schizophrenic
patients
RC,
CC
12,235 All APsAge, sex, race,
VA facility,
marital status
Ohio VA
(men only)38
1997–2000All AP usersRC 5837 Risperidone,
olanzapine,
haloperidol,
fluphenazine
Age, race
Managed care39
1996–1997Psychiatric
patients
RC7933All APs Age, sex,
health care
coverage
Managed care40
1997–2000 Psychiatric
patients
RC 10,296 All APsAge, sex,
health care
coverage
United
Kingdom
General
Practice
Research41
1994–1998All AP users CC 1946
(424 cases,
1522
controls)
All APsAge, sex,
practice
setting
United
Kingdom
General
Practice
Research42
Iowa
Medicaid43
VA44
1987–2000 Schizophrenic
AP users
CC 3147
(451 cases,
2696
controls)
All APsAge, sex
1990–1994 Schizophrenic
patients
Schizophrenic
AP users
RC 3013Clozapine,
typical APs
All APs
AP exposure durationAge, sex
1998–1999RPC 38,632Previous
hospitalization,
psychiatric diagnoses,
comorbidities
Age, sex,
race, income,
distance to
hospital, VA
compensation

Page 8

PHARMACOTHERAPY Volume 24, Number 11, 2004
diabetes when compared with no treatment for
psychoses.39In a later study by the same authors,
only olanzapine was associated with an increased
risk of diabetes.40Another group compared
olanzapine and risperidone and found that
olanzapine was significantly associated with
incident diabetes.36In a study that compared
typical with atypical agents, the authors found
that atypicals (with the exception of risperidone)
were associated with increased diabetes in a large
Veterans Affairs study.44In the younger age
groups (< 40 yrs), all atypical agents were
associated with increased diabetes. This
discussion shows that the literature is far from
conclusive regarding the relationship between
antipsychotic agents and incident diabetes.
Several factors could explain differences in the
results. Table 7 shows a comparison of the study
methodologies. Several of the large database
studies were conducted in various settings: in
veterans37, 38, 44; in Medicaid populations43, 45; at
managed care organizations35, 39, 40, 46; and outside
the United States.36, 41, 42Comparison of results
across each of the settings may have inherent
biases. Also, the study designs differed in terms
of inclusion criteria: all patients taking
antipsychotic agents35, 36, 38, 41versus persons with
psychiatric diagnoses.37, 39, 40, 42–46The studies
differed in terms of comparison groups (typical
vs atypicals)35, 37, 38, 41–46; among atypicals36, 38, 46;
antipsychotic use versus no antipsychotic use
among persons with psychoses39, 40; and anti-
psychotic use versus no antipsychotic use among
the general population.35
Another factor that differed across the studies
was the covariates used. Most studies controlled
for demographics such as age35–46and sex,35–37,
39–46whereas fewer studies controlled for race.37,
38, 44, 45As mentioned previously, race is an
important risk factor for diabetes, with minorities
more likely to develop diabetes compared with
nonminorities. Regarding clinical variables,
several studies controlled for psychoses treatment–
specific issues such as other antipsychotic use,36,
38–41, 45psychiatric diagnoses,36, 38–41, 44–46and
treatment exposure duration,35, 36, 38–43, 45, 46and
three studies controlled for dosage.35, 39, 45Several
studies controlled for use of other drugs that may
cause diabetes, such as steroids, ?-blockers,
anticonvulsants,37, 38, 40–42, 45whereas only a few
captured other factors associated with diabetes
such as BMI41and weight gain,40hypertension,41,
45, 46and dyslipidemia.46
One distinct aspect of our study is the use of
elevated glucose levels as a proxy for incident
diabetes, in addition to drug therapies and
diagnoses. Another distinct aspect of our study
was the inclusion of clinical covariates in the
analysis. Blood pressure changes, previous
hyperlipidemia, and BMI changes were incor-
porated into the multivariate analysis to control
for known factors related to diabetes. The results
showed that even after controlling for these
comorbid conditions, increasing age and
minority race were consistently related to new-
onset diabetes. In a separate multivariate
analysis, previous hyperlipidemia was also
associated with new-onset diabetes. Other
studies have also found age to be a significant
factor in incident diabetes. One group43found
occurrence to be more prevalent in the younger
age group (20–34 yrs), whereas other studies
1536
Table 7. Comparison of Methodologies of Retrospective Database Studies Examining Antipsychotic Use and New-Onset
Diabetes Mellitus (continued)
Setting orStudy Sample StudySample
Database Time FramePopulationDesign Size
New Jersey1990–1995PsychiatricCC 14,007
Medicaid andpatients(7227 cases,
Medicare,6780
Pharmaceutical controls)
Assistance to
Aged and
Disabled45
Demographic
Covariates
Age, sex,
race, socio-
economic
status
Drugs
Clozapine vs
nonclozapine psychotropic use,
clozapine duration,
clozapine dosage,
comorbidity score,
use of drugs
associated with
glucose intolerance
All APsAP exposure duration,
general health
comorbidities,
psychiatric diagnoses
Clinical Covariates
Psychiatric diagnoses,
Managed care46
1996-1998 Schizophrenic
AP users
RC 815Age, sex
region,
enrollment
status
AP = antipsychotics; RC = retrospective cohort; VA = Veterans Administration; CC = case-control; RPC = retrospective (prevalent cases).

Page 9

ANTIPSYCHOTIC THERAPY AND INCIDENT DIABETES MELLITUS Barner et al
found increasing age to be related to incident
diabetes.35, 39, 40, 46
In September 2003, the U.S. Food and Drug
Administration issued a request for manufacturers
of atypical antipsychotic agents to modify their
labeling to include the risk of glucose abnormalities.
They also recommended regular monitoring for
hyperglycemia in patients with diabetes and
those at risk for diabetes. Although our study
results show no difference in new-onset diabetes
between patients taking typical agents and those
taking atypical agents, the overall frequency in
this population was 7.1%, which is higher than
the 6.3% prevalence in the general population.50
Thus, it is important that all patients taking
antipsychotic agents be monitored for symptoms
of diabetes.
Limitations
Our results should be interpreted with caution.
Although retrospective database studies can
capture effectiveness among a large patient
group, causality cannot be established. The
database used in this study involved veterans in
central Texas and more than 90% of the study
subjects were male; thus, differences in outcomes
regarding sex and other regions cannot be fully
assessed. It is possible that subjects could have
obtained drugs from outside the Veterans Affairs
system; however, internal resources indicate that
veterans tended to use the Veterans Affairs
resources exclusively since they were free during
the time of this study. We used an intent-to-treat
analysis; this does not account for switching and
concomitant use of antipsychotics, which
routinely occur in practice.38However, in our
unadjusted analyses comparing typical and
atypical antipsychotic agents, we tried to
overcome this limitation by performing analyses
to incorporate switching and concomitant use.
We found no differences in the frequency of
diabetes between the typical and atypical groups
when using this method. Studies have shown
that primary care, in terms of routine screening
for diabetes, for the mentally ill may be
suboptimal and thus, new cases of diabetes may
go undetected.51Although we used a very
inclusive definition (ICD-9, antidiabetic drugs,
and elevated glucose levels) to identify new-onset
diabetes, it is likely that this may have been
underestimated since the American Diabetes
Association reports that in nearly one third of
persons with diabetes is undetected.50Also, we
did not control for antipsychotic dosage or other
nonantipsychotic drugs (e.g., lithium, steroids,
thiazide diuretics) that may have been related to
new-onset diabetes.
Conclusion
This study found that the frequency of new-
onset diabetes mellitus among a population of
veterans in central Texas was 7.1%. No signifi-
cant difference was noted in the frequency of
new-onset diabetes between patients taking typical
agents and those taking atypical antipsychotic
agents or among those taking atypical antipsychotic
agents. In addition, a multivariate logistic
regression analysis revealed that when controlling
for demographic and clinical variables, no
significant difference was noted among the
antipsychotics. The analysis revealed that new-
onset diabetes was significantly related to
increasing age and minority race. Nevertheless,
patients who are taking antipsychotic agents and
have diabetes or are at risk for diabetes should be
monitored for any adverse effects related to
diabetes.
Acknowledgments
The authors would like to thank Eli Lilly and
Company for financial support and the Central Texas
Veterans Health Care System for their time and
expertise in obtaining the data.
References
1. Allison DB, Mentore JL, Moonseong H, et al. Antipsychotic-
induced weight gain: a comprehensive research synthesis. Am J
Psychiatry 1999;156:1686–96.
2. Wetterling T. Bodyweight gain with atypical antipsychotics: a
comparative review. Drug Saf 2001;24:59–73.
3. Green AI, Patel JK, Goisman RM, Allison DB, Blackburn G.
Weight gain from novel antipsychotic drugs: need for action.
Gen Hosp Psychiatry 2000;22:224–35.
4. Allison DB, Casey DE. Antipsychotic-induced weight gain: a
review of the literature. J Clin Psychiatry 2001;62(suppl
7):22–31.
5. Casey DE, Zorn SH. The pharmacology of weight gain with
antipsychotics. J Clin Psychiatry 2001;62(suppl 7):4–10.
6. Kurzthaler I, Fleischhacker WW. The clinical implications of
weight gain and schizophrenia. J Clin Psychiatry 2001;62(suppl
7):32–7.
7. Sheitman BB, Bird PM, Binz W, Akinli L, Sanchez C.
Olanzapine-induced elevation of plasma triglyceride levels
[letter]. Am J Psychiatry 1999;156:1471–2.
8. Gaulin BD, Markowitz JS, Caley CF, Nesbitt LA, Dufresne RL.
Clozapine-associated elevation in serum triglycerides. Am J
Psychiatry 1999;156:1270–2.
9. Osser DN, Najarian DM, Dufresne RL. Olanzapine increases
weight and serum triglyceride levels. J Clin Psychiatry
1999;60:767–70.
10. Ghaeli P, Dufresne RL. Serum triglyceride levels in patients
treated with clozapine. Am J Health-Syst Pharm 1996;53:
2079–81.
11. Kamran A, Doraiswamy PM, Jane JL, Hammett EB, Dunn L.
Severe hyperglycemia associated with high doses of clozapine
1537

Page 10

PHARMACOTHERAPY Volume 24, Number 11, 2004
[letter]. Am J Psychiatry 1994;151:1395.
12. Koren W, Kreis Y, Duchowiczny K, et al. Lactic acidosis and
fatal myocardial failure due to clozapine. Ann Pharmacother
1997;31:168–70.
13. Kostakoglu AE, Yazici KM, Erbas T, Guvener N. Ketoacidosis
as a side-effect of clozapine: a case report. Acta Psychiat Scand
1996;93:217–18.
14. Koval MS, Rames LJ, Christie S. Diabetic ketoacidosis
associated with clozapine treatment. Am J Psychiatry
1994;151:1520–1.
15. Smith H, Kenney-Herbert J, Kowles L. Clozapine-induced
diabetic ketoacidosis. Aust N Z J Psychiatry 1999;33:121–2.
16. Peterson GA, Byrd SL. Diabetic ketoacidosis from clozapine
and lithium cotreatment. Am J Psychiatry 1994;153:737–8.
17. Popli AP, Konicki PE, Jurjus GJ, Fuller MA, Jaskiw GE.
Clozapine and associated diabetes mellitus. J Clin Psychiatry
1997;58:108–11.
18. Wirshing DA, Spellberg BJ, Erhart SM, Marder SR, Wirshing
WC. Novel antipsychotics and new onset diabetes. Biol
Psychiatry 1998;44:778–83.
19. Goldstein LE, Sporn J, Brown S, et al. New-onset diabetes
mellitus and diabetic ketoacidosis associated with olanzapine
treatment. Psychosomatics 1999;40:438–43.
20. Lindenmayer JP, Patel R. Olanzapine-induced ketoacidosis
with diabetes mellitus [letter]. Am J Psychiatry 1999;156:1471.
21. Sobel M, Jaggers ED, Franz MA. New-onset diabetes mellitus
associated with the initiation of quetiapine treatment. J Clin
Psychiatry 1999;60:556–7.
22. Ai D, Roper TA, Riley JA. Diabetic ketoacidosis and clozapine.
Postgrad Med J 1998;74:493–4.
23. Fertig MK, Brooks VG, Shelton PS, English CW.
Hyperglycemia associated with olanzapine. J Clin Psychiatry
1998;59:687–9.
24. Wirshing DA, Pierre JM, Eyeler J, Weinbach J, Wirshing WC.
Risperidone-associated new-onset diabetes. Biol Psychiatry
2001;50:148–9.
25. Colli A, Cocciolo M, Francobandiera G, Rogantin F, Cattalini
N. Diabetic ketoacidosis associated with clozapine treatment.
Diabetes Care 1999;22:176–7.
26. Wehring H, Alexander B, Perry PJ. Diabetes mellitus
associated with clozapine therapy. Pharmacotherapy
2000;20:844–7.
27. Bettinger TL, Mendelson SC, Dorson PG, Crismon ML.
Olanzapine-induced glucose dysregulation. Ann Pharmacother
2000;34:865–7.
28. Roefaro J, Mukherjee SM. Olanzapine-induced hyperglycemic
nonketonic coma. Ann Pharmacother 2001;35:300–2.
29. Bonanno DG, Davydov L, Botts SR. Olanzapine-induced
diabetes mellitus. Ann Pharmacother 2001;35:563–5.
30. McIntyre RS, McCann SM, Kennedy SH. Antipsychotic
metabolic effects: weight gain, diabetes mellitus, and lipid
abnormalities. Can J Psychiatry 2001;46:273–81.
31. Kingsbury SJ, Fayek M, Trufasiu D, Zada J, Simpson GM. The
apparent effects of ziprasidone on plasma lipids and glucose. J
Clin Psychiatry 2001;62:347–9.
32. Kinon BJ, Basson BR, Gilmore JA, Tollefson GD. Long-term
olanzapine treatment: weight change and weight-related health
factors in schizophrenia. J Clin Psychiatry 2001;62:92–100.
33. Henderson DC, Cagliero E, Gray C, et al. Clozapine, diabetes
mellitus, weight gain, and lipid abnormalities: a five-year
naturalistic study. Am J Psychiatry 2000;157:975–81.
34. Hagg S, Joelsson L, Mjorndal T, Spigset O, Oja G, Dahlqvist
R. Prevalence of diabetes and impaired glucose tolerance in
patients treated with clozapine compared with patients treated
with conventional depot neuroleptic medications. J Clin
Psychiatry 1998;59:294–9.
35. Buse JB, Cavazzoni P, Hornbuckle K, Hutchins D, Breier A,
Jovanovic L. A retrospective cohort study of diabetes mellitus
and antipsychotic treatment in the United States. J Clin
Epidemiol 2003;56:164–70.
36. Caro JJ, Ward A, Levinton C, Robinson K. The risk of diabetes
during olanzapine use compared with risperidone use: a
retrospective database analysis. J Clin Psychiatry
2002;63:1135–9.
37. Cunningham F, Lambert B, Miller R, Daluck G, Kim JB, Hur
K. Antipsychotic induced diabetes in veteran schizophrenic
patients [abstr]. Pharmacoepidemiol Drug Saf 2003;12:305.
38. Fuller MA, Shermock KM, Secic M, Grogg AL. Comparative
study of the development of diabetes mellitus in patients taking
risperidone and olanzapine.
2003;23:1037–43.
39. Gianfrancesco FD, Grogg AL, Mahmoud RA, Wang RH,
Nasrallah HA. Differential effects of risperidone, olanzapine,
clozapine, and conventional antipsychotics on type 2 diabetes:
findings from a large health plan database. J Clin Psychiatry
2002;63:920–30.
40. Gianfrancesco FD, White R, Wang RH, Nasrallah HA.
Antipsychotic-induced type 2 diabetes: evidence from a large
health plan database. J Clin Psychopharmacol 2003;23:328–35.
41. Kornegay CJ, Vasilakis-Scaramozza C, Jick H. Incident
diabetes associated with antipsychotic use in the United
Kingdom general practice research database. J Clin Psychiatry
2002;63:758–62.
42. Koro CE, Fedder DO, L’Italien GJ, et al. Assessment of
independent effect of olanzapine and risperidone on risk of
diabetes among patients with schizophrenia: population based
nested case-control study. BMJ 2002;325:243–7.
43. Lund BC, Perry PJ, Brooks JM, Arndt S. Clozapine use in
patients with schizophrenia and the risk of diabetes,
hyperlipidemia, and hypertension: a claims-based approach.
Arch Gen Psychiatry 2001;58:1172–6.
44. Sernyak MJ, Douglas LL, Alarcon RD, Losonczy MF,
Rosenheck R. Association of diabetes mellitus with use of
atypical neuroleptics in the treatment of schizophrenia. Am J
Psychiatry 2002;159:561–6.
45. Wang PS, Glynn RJ, Ganz DA, Schneeweiss S, Levin R, Avorn
J. Clozapine use and risk of diabetes mellitus. J Clin
Psychopharmacol 2002;22:236–43.
46. Zhao A, Tunis SL, Loosbrock DL. Risk of diabetes for
individuals treated with antipsychotics. J Appl Res
2003;3:287–95.
47. National Institutes of Health, National Heart, Lung, and
Blood Institute, National Institute of Diabetes and Digestive
and Kidney Disease. Clinical guidelines on the detection,
evaluation, and treatment of overweight and obesity in adults:
the evidence report. NIH publication no. 98-4083, September
1998, table IV-1. Available from http://www.nhlbi.nih.gov/
guidelines/obesity/ob_gdlns. pdf. Accessed January 4, 2003.
48. National Institutes of Health, National Heart, Lung, and
Blood Institute, National High Blood Pressure Education
Program. The sixth report of the Joint National Committee on
prevention, detection, evaluation, and treatment of high blood
pressure. NIH publication no. 98-4080, November 1997, table
2. Available from http://www.nhlbi.nih.gov/guidelines/
hypertension/jnc6.pdf. Accessed January 4, 2003.
49. National Cholesterol Education Program, National Heart,
Lung, and Blood Institute, National Institutes of Health.
Detection, evaluation, and treatment of high blood cholesterol
in adults (Adult Treatment Panel III): final report. NIH
publication no. 02-5215, September 2002:II-5–10. Available
from http://www.nhlbi.nih.gov/guidelines/cholesterol/atp3full.
pdf. Accessed January 4, 2003.
50. American Diabetes Association. All about diabetes. Available
from http://www.diabetes.org/about-diabetes.jsp. Accessed May
12, 2004.
51. Cradock-O’Leary J, Young AS, Yano EM, et al. Use of general
medical services by VA patients with psychiatric disorders.
Psychiatr Serv 2002;53:574–8.
Pharmacotherapy
1538